From the Department of Medicine, Solna

Clinical Epidemiology Division

Karolinska Institutet, Stockholm, Sweden

THE IMPACT OF AUTOIMMUNE THYROID

DISEASE (AITD) ON RHEUMATOID
ARTHRITIS (RA) AND THE IMPACT OF RA
ON AITD

Kristin Waldenlind

Stockholm 2022
All previously published papers were reproduced with permission from the publisher.
Published by Karolinska Institutet.
Printed by Universitetsservice US-AB, 2022
©Kristin Waldenlind, 2022
ISBN 978-91-8016-439-9
Cover illustration: The painting “Flowers and Fruit” by Pierre-Auguste Renoir (1841-1919).
Renoir suffered from severe rheumatoid arthritis which seriously impacted his mobility the
last decades of his life.
The impact of Autoimmune Thyroid Disease (AITD)
on Rheumatoid Arthritis (RA) and the impact of RA
on AITD

THESIS FOR DOCTORAL DEGREE (Ph.D.)

By

Kristin Waldenlind

AKADEMISK AVHANDLING
som för avläggande av medicine doktorsexamen vid Karolinska Institutet offentligen försvaras i
lokal Lars Klareskog, NB:U102, N-huset, Karolinska Universitetssjukhuset, Solna

Fredagen den 8 april, 2022, kl 09.00

.
Principal Supervisor: Opponent:
Professor Johan Askling Docent Aladdin Mohammad
Karolinska Institutet Lunds University
Department of Medicine, Solna Department of Clinical Sciences/Rheumatology
Clinical Epidemiology Division University of Cambridge
Department of medicine
Co-supervisors:
PhD Bénédicte Delcoigne Examination Board:
Karolinska Institutet Professor Ingrid Lundberg
Department of Medicine, Solna Karolinska Institutet
Clinical Epidemiology Division Department of Medicine, Solna

PhD Saedis Saevarsdottir Docent Hannah Brooke

Karolinska Institutet Uppsala University
Department of Medicine, Solna Department of Surgical Sciences, Medical
Clinical Epidemiology Division Epidemiology
University of Iceland, Reykjavik
Faculty of Medicine, School of Health Sciences Docent Sophie Bensing
Karolinska Institutet
Department of Molecular Medicine and Surgery
Division of Experimental and Clinical
Neuroendocrinology
Dedicated to Carl, Sigrid, Axel & Rickard
POPULAR SCIENCE SUMMARY OF THE THESIS
In epidemiological research, factors that may affect the development of diseases and the
distribution of diseases in the population are investigated. The Swedish nationwide health
care registers, with high coverage of the population, offer unique possibilities for
epidemiological studies. Based on these registers, it is possible to conduct observational
studies to investigate the development of diseases during a long period of time. By using
epidemiological methods, the studies in this thesis have investigated the relationship between
two autoimmune diseases, autoimmune thyroid disease (AITD) and rheumatoid arthritis
(RA). Based on the health care registers used for the different substudies, it was possible to
compare patients with these two diseases in relation to the general population in Sweden. In
this way we could compare a large number of RA patients and controls in our studies, e.g. 13
000 and 63 000 controls in study IV. By using other disease related information, we could
adjust for possible factors that may disturb the assessment of the association between AITD
and RA.

RA

RA is the most common chronic inflammatory disease in the population with a prevalence of
approximately 0,5-1%. RA is approximately twice as common among women and the mean
age of RA onset is around 55 years. The chronic inflammation in RA is due to a disturbed
immune response resulting in an “attack” directed against normal components in an
individual, a so-called autoimmune process. The mechanisms behind the autoimmune process
in RA is most likely due to an interplay between environmental and genetic factors in the
individual. The chronic inflammation in the joints in patients with RA leads to swollen and
tender joints, morning stiffness, elevated inflammatory laboratory tests and often systemic
symptoms. In the 90’s new antirheumatic treatments were introduced that were targeted more
specifically against different components of the immune response, called biologic disease-
modifying antirheumatic drugs (bDMARDs). These treatments revolutionized the treatment
and prognosis of RA. Still, there are patients with RA not responding to the antirheumatic
treatment as expected. One field of interest in RA research is to find factors that can predict
the treatment response to a specific treatment which could be a help in tailoring the therapy
for the specific patient. RA patients have an increased risk of developing other diseases, for
example several autoimmune conditions, that may also potentially affect the treatment
response.

AITD

AITD, including hypothyroidism and hyperthyroidism, is one of the most common

autoimmune diseases in the population, with a prevalence of approximately 5% in the general
population. The autoimmune process in the thyroid gland leads to disturbed levels of thyroid
hormones, affecting all cells, and can cause a wide range of symptoms. After the initial
therapy for hyperthyroidism most patients having both hypothyroidism and hyperthyroidism
are treated with thyroid hormones. It has been shown that AITD and RA share genetic
background. For decades research studies have reported that AITD is one of the most
common comorbid condition in patients with RA, approximately two times more frequent
than in the general population.

Study I: Do patients who are registered with a RA diagnosis in the health care registers, have
RA in clinical practice?

The aim of study I was to verify if two of the registers, The National Patient Register (NPR)
and the Swedish Rheumatology Quality Register (SRQ) have a validity high enough to be
used in different research studies to identify patients with RA. This validation was performed
by comparing if patients registered with a RA diagnosis in the NPR and the SRQ would also
be assessed as having a “true” RA in clinical practice. Based on the medical files from
patients coded as RA in the registers we found that the validity of the RA diagnosis was high
for both patients with long-standing and new onset RA. We concluded that these registers can
be used in epidemiological studies to correctly identify patients with RA.

Study II: When does the increased risk of AITD in RA patients occur?

Although previous studies have shown that AITD is more common in patients with RA
compared to the general population, these studies have not disentangled when the risk of
AITD develops in relation to time before and after RA diagnosis. The aim in study II was to
assess the risk of developing AITD up to 7 years before and 8 years after RA diagnosis,
compared to matched controls from the general population. We found that at RA diagnosis,
approximately 10% of the RA patients also suffered from AITD, compared to 7% in the
general population. The risk of developing AITD increased during a 5-year period before RA
diagnosis. On the contrary, following RA diagnosis the risk of new-onset AITD gradually
decreased. This temporal change of AITD risk in relation to RA diagnosis raised the question
if antirheumatic treatments may have a protective effect on developing AITD.

Study III: Does AITD affect the disease activity or response to treatment in RA patients?

Studies have concluded that AITD is one of the most common comorbidities in patients with
RA, but it is not well understood if AITD might affect the disease activity or the response to
antirheumatic treatments. The aims of study III were to compare disease activity and
treatment response to the most common treatment, methotrexate, in RA patients with AITD
compared to those without this comorbidity. The disease activity and treatment response at
the first visit, when the patient was diagnosed with RA, was compared with the follow-up
visits after 3 and 6 months. We found that RA patients with AITD scored worse on the
patient-reported disease activity measures such as pain and global health at diagnosis, but not
on the objective disease activity measures recorded by the physician. Concomitant AITD did
not impact the overall chance of responding to treatment, with the exception for the younger
RA patients who were less likely to achieve a good treatment response. In clinical practice,
when treating and evaluating patients with RA who have a concomitant AITD, it is important
to evaluate both subjective- and objective disease activity measures.
Study IV: Does antirheumatic treatment lower the risk of AITD in RA patients?

Findings from study II demonstrated a decreased risk of developing AITD following the
diagnosis of RA. Based on these results, we investigated if antirheumatic treatments with
bDMARDs could reduce the risk of developing AITD. The risk of new-onset AITD up to 14
years from RA diagnosis, was assessed and compared to the general population controls, and
in relation to treatment with bDMARDs. The results demonstrated that following the
diagnosis of RA, patients treated with bDMARDs, have a decreased risk of developing
AITD. These findings indicate that these therapies could have a protective effect of
developing AITD and may open for studies exploring new indications for bDMARDs.
ABSTRACT
Study I

Background: The Swedish National Patient Register (NPR) is a unique source for the
identification of large cohorts of rheumatoid arthritis (RA) patients. The general validity of
this register has been reported to be high.
Aims: The aims were to specifically validate the RA diagnosis from the outpatient specialist
care in this register, and to assess the proportion of patients identified through specific
algorithms to define incident RA, who in clinical practice would be assessed as new-onset
disease.
Methods: Diagnosis-related parameters were extracted from the medical records of
approximately 200 patients coded with a RA diagnosis in the NPR, of whom the majority
was also included in the Swedish Rheumatology Quality Register (SRQ). The patients were
assessed if they fulfilled the 1987 ACR- and 2010 ACR/EULAR classification criteria for
RA. Furthermore, it was determined if clinical diagnosis correlated with disease onset as
defined through register-based algorithms.
Results: The prevalent and strictly incident patients with a RA diagnosis fulfilled to a large
extent (>90%) the classification criteria or clinical diagnosis for RA and the patients coded as
incident RA represented new-onset disease.
Conclusion: The validity of the RA diagnosis was high and specific algorithms adequately
identified new-onset RA, which allows the NPR to be used to identify RA populations with
high validity for epidemiological studies.

Study II

Background: Autoimmune thyroid disease (AITD), including hypothyroidism and

hyperthyroidism, is the most common organ-specific autoimmune disorder in the general
population. AITD and RA share genetic background and the prevalence of AITD is also
increased in patients with RA. Studies on when the increased risk of AITD develops in
relation to time before and after RA diagnosis are missing.
Aims: The aim was to assess the risk of thyroxine treated AITD at different points in time
before and after RA diagnosis.
Methods: By using the SRQ and other nationwide registers approximately 8000 early RA
patients and 80 000 controls were identified. A combined case-control and cohort design was
used to investigate the prevalence and relative risk of incident AITD up to 7 years before and
8 years after RA diagnosis. AITD was defined as thyroxine prescription in the Swedish
Prescribed Drug Register (PDR), with exclusion of non-autoimmune indications.
Results: The prevalence of AITD was approximately 10% at time of RA diagnosis. The risk
of new-onset AITD increased during a 5-year period prior to RA diagnosis. By contrast,
following the diagnosis of RA, the risk of developing AITD decreased below the expected
rate.
Conclusion: The temporal pattern of AITD risk in relation to RA diagnosis raised the
questions of a potential temporal change in diagnostic intensity, a bilateral influence between
the two conditions or a potential protective effect of antirheumatic therapies.

Study III

Background: Although AITD is one of the most common comorbidities in RA it is less well
understood if AITD might affect RA disease activity or response to anti-rheumatic therapies.
Aims: The aims were to compare disease activity and treatment response to the first-line
therapy with methotrexate in early RA patients with AITD versus those without this
comorbidity.
Methods: The SRQ and other nationwide registers were used to identify approximately 9000
RA patients and comorbidity with AITD to compare the disease activity at the 3- and 6-
month follow-up visits, based on DAS 28 and its individual components, and EULAR
response criteria.
Results: AITD was associated with worse subjective- but not objective measures of disease
activity among RA patients. AITD did not impact the overall treatment response, but
a subgroup of younger patients with both RA and AITD were less likely to achieve a good
treatment response to methotrexate.
Conclusions: Concomitant AITD impacts the RA disease activity measures used in clinical
practice through worse subjective disease activity components and thus the composite disease
activity measures. It is of importance to analyze all individual disease activity components
when treating and evaluating patients with early RA and concomitant AITD.

Study IV

Background/Aims: Based on the result from study II, we investigated if biologic disease-
modifying antirheumatic drugs (bDMARDs) used for the treatment of RA might lower the
risk of developing AITD.
Methods: The SRQ and other nationwide registers were used to perform a cohort study
including approximately 13 000 RA patients and 63 000 population controls to estimate the
relative risk of AITD following RA diagnosis, compared to the general population controls,
and in relation to treatment with bDMARDs.
Results: The risk of developing AITD following RA diagnosis was decreased in RA patients
compared to controls and this was most pronounced in RA patients treated with bDMARDs.
Conclusion: Treatment with bDMARDs may have and preventive effect on AITD.
LIST OF SCIENTIFIC PAPERS

I. Validation of the Rheumatoid Arthritis diagnosis in the Swedish National

Patient Register: a cohort study from Stockholm County

Waldenlind K, Eriksson JK, Grewin B, Askling J

BMC Musculoskeletal Disorders, 2014, Dec,15:432.

II. Risk of Thyroxine-Treated Autoimmune Thyroid Disease Associated

With Disease Onset in Patients With Rheumatoid Arthritis

Waldenlind K, Saevarsdottir S, Bengtsson C, Askling J

JAMA Network Open. 2018; Oct, 1(6):e183567

III. Does autoimmune thyroid disease affect RA disease activity or response

to methotrexate?

Waldenlind K, Delcoigne B, Saevarsdottir S, Askling J

RMD open, 2020 Jul; 6(2):e001282

IV. Does treatment with disease-modifying antirheumatic drugs (DMARDs)

lower the risk of Autoimmune Thyroid Disease in Patients with RA?

Waldenlind K, Delcoigne B, Saevarsdottir S, Askling J

Manuscript
CONTENTS
INTRODUCTION........................................................................................................... 1
LITERATURE REVIEW ............................................................................................... 3
2.1 RHEUMATOID ARTHRITIS (RA) .................................................................... 3
2.1.1 Clinical presentation and epidemiology ................................................... 3
2.1.2 Etiology ..................................................................................................... 3
2.1.3 Classification criteria ................................................................................ 4
2.1.4 Treatment in RA ........................................................................................ 7
2.1.5 Measures of disease activity ..................................................................... 8
2.2 AUTOIMMUNE THYROID DISEASE (AITD) ................................................ 9
2.2.1 Pathogenesis and etiology ......................................................................... 9
2.2.2 Epidemiology and clinical picture of AITD subgroups .........................10
2.2.3 The definition of AITD, management and treatment .............................11
2.3 AITD and RA ......................................................................................................11
2.3.1 AITD in RA .............................................................................................11
2.3.2 RA in AITD .............................................................................................14
2.3.3 Thyroid autoantibodies in RA patients ...................................................14
2.3.4 Shared susceptibility genes .....................................................................14
2.3.5 Timing of onset of AITD in relation to RA............................................15
2.3.6 The effect of AITD on RA disease presentation and treatment
response ...................................................................................................15
2.3.7 bDMARDs and AITD .............................................................................16
2.3.8 Organ and systemic autoimmunity .........................................................16
2.4 Validity of the RA diagnosis in the National Patient Register (NPR) and
the Swedish Rheumatology Quality register (SRQ)...........................................17
OBJECTIVES ...............................................................................................................19
3.1 Overall objectives ................................................................................................19
3.2 Specific aims........................................................................................................19
MATERIALS AND METHODS .................................................................................21
4.1 Setting ..................................................................................................................21
4.2 Data sources .........................................................................................................21
4.3 Study design and study population .....................................................................23
4.3.1 Overview .................................................................................................23
4.3.2 Study I......................................................................................................23
4.3.3 Study II ....................................................................................................24
4.3.4 Study III ...................................................................................................25
4.3.5 Study IV...................................................................................................27
4.4 Statistical analyses ...............................................................................................27
4.4.1 Statistical concepts ..................................................................................28
4.4.2 Statistical analyses in the specific studies ..............................................29
4.5 Ethical considerations..........................................................................................33
MAIN RESULTS ..........................................................................................................35
5.1 Study I .................................................................................................................. 35
5.2 Study II ................................................................................................................ 36
5.3 Study III ............................................................................................................... 38
5.4 Study IV ............................................................................................................... 41
DISCUSSION ............................................................................................................... 43
6.1 Methodological considerations ........................................................................... 43
6.1.1 Study design and outcome measures ...................................................... 43
6.1.2 Internal validity ....................................................................................... 44
6.1.3 Precision .................................................................................................. 47
6.1.4 External validity ...................................................................................... 47
6.2 Findings ............................................................................................................... 48
6.2.1 Validity of the RA diagnosis .................................................................. 48
6.2.2 AITD in relation to RA diagnosis........................................................... 49
6.2.3 The effect of AITD on RA disease activity and response to
antirheumatic treatment .......................................................................... 50
6.2.4 The impact of antirheumatic treatments on the development of
AITD........................................................................................................ 51
CONCLUSIONS ........................................................................................................... 53
POINTS OF PERSPECTIVE ....................................................................................... 55
ACKNOWLEDGEMENTS.......................................................................................... 57
REFERENCES .............................................................................................................. 61
LIST OF ABBREVIATIONS
ACPA Antibodies against citrullinated peptides
ACR American college of rheumatology
AITD Autoimmune thyroid disease
ARTIS The Swedish biologics register
ATC Anatomical therapeutic chemical classification system
bDMARDs Biologic disease-modifying antirheumatic drugs
csDMARDs Conventional synthetical DMARDs
CI Confidence Interval
CRP C-reactive protein
DAS 28 RA disease activity score of 28 joints
ESR Erythrocyte sedimentation rate
EULAR European league against rheumatism
HAQ Health assessment questionnaire
HR Hazard ratio
ICD International classification of diseases
NPR National Patient Register
NSAID Non-steroidal anti-inflammatory drug
OR Odds ratio
PDR Prescribed Drug Register
PROMs Patient-reported outcomes
RA Rheumatoid arthritis
RCT Randomized controlled trial
RF Rheumatoid factor
SE Shared epitope
SJC Swollen joint count
SRQ Swedish Rheumatology Quality Register
THs Thyroid hormones
TJC Tender joint count
TPOAb Anti-thyroid peroxidase antibodies
TgAb Thyroglobulin antibodies
TRAb TSH-receptor antibodies
TNF Tumor necrosis factor
VAS Visual analogue scale
 INTRODUCTION
When I started as a resident doctor in rheumatology in 2011, I realized how much the last
decades’ paradigm shift towards new antirheumatic drugs had impacted everyday life for
patients with rheumatic diseases. With the introduction of tumor necrosis factor inhibitors
(TNF-inhibitors), a class of targeted biologic disease modifying antirheumatic drug
(bDMARD), in the beginning of the 21 st century, the prognosis for patients with rheumatoid
arthritis (RA) and other rheumatologic diseases became much more benign. When meeting
patients with RA in clinical practice, I also realized the increased burden of autoimmune
comorbidities, like type 1 diabetes and Hashimoto’s thyroiditis, in this group of patients. This
raised the question if their autoimmune comorbidity had an impact on their rheumatic
disease, or vice versa. The topic of my upcoming PhD-project, taking shape in 2014,
therefore seemed very clinically relevant when presented to me by my supervisor. Since I
became a rheumatology consultant in 2019, I have had the possibility to deepen my
knowledge in RA, in particular while working with patients in the “early arthritis program”
(Tidig reumatoid artrit -TRA) at the clinic. For instance, I noticed that although RA is the
most common chronic inflammatory joint disease in the population, and autoimmune thyroid
disease (AITD) is one of the most common comorbidities in RA, surprisingly little is known
on their association and its impact. This raised the question on how co-existing AITD impacts
RA, and, conversely, how RA impacts AITD.

The studies in this thesis explore the association between AITD and RA. In the first study we
investigated the validity of the registers used to capture patients with RA, necessary for
further studies. In the three other studies we investigated the temporal association between
AITD and RA, if concurrent AITD had an impact on the disease activity or treatment
response in RA patients and if antirheumatic therapies may have a protective effect on
developing AITD.

1
 LITERATURE REVIEW
2.1 RHEUMATOID ARTHRITIS (RA)

2.1.1 Clinical presentation and epidemiology

RA is the most common chronic inflammatory joint disease in the population, with a
prevalence of 0,5-1% in western countries (1, 2). RA is approximately twice as common
among women and the onset usually between 45-60 years of age. The reported incidence in
western countries vary between 20-50 cases/100 000 depending on studies (3). The
nationwide incidence has been reported to be 40/100 000 in Sweden (4). The typical patient
with RA presents with symmetrical swollen and tender joints, morning stiffness, and elevated
inflammatory laboratory tests, erythrocyte sedimentation rate (ESR) and C-Reactive Protein
(CRP). Although the main symptoms are related to inflammation in the joints, RA is
considered to be a systemic disease. There are often symptoms of systemic inflammation
such as malaise and weight loss. RA can also be associated with extra-articular
manifestations like serositis, interstitial lung disease and cutaneous vasculitis (5). There are
two main subgroups of RA, seropositive and seronegative RA, based on the presence of
autoantibodies against immunoglobulin G (rheumatoid factor, RF) and autoantibodies against
citrullinated peptides (ACPAs). The majority of patients have a seropositive RA
(approximately 50% in early disease and 80% in established disease) (6). Seronegative RA
may be associated with higher inflammatory parameters and a more active disease than
seropositive RA, initially. On the other hand, seropositive RA is associated with a more
erosive disease and worse prognosis over time (7). Patients with RA have an increased
burden of comorbidities including cardiovascular diseases, infections and malignancies
leading to a decreased life expectancy (8-10).

2.1.2 Etiology
The pathogenesis in RA is not completely understood. Several environmental risk factors and
risk determinants have been identified and associated with the development of RA, such as
smoking, low socioeconomic status and educational level, hormonal factors and work
exposures (textile- and silica dust) (1, 11-16). During the past years there has also been an
increasing interest in investigating the potential impact of the microbiome in the risk of
developing RA (17). Genetic risk factors as the HLA-system (mainly through the HLA-
DRB1 gene) and its specific peptide binding is involved in the pathogenesis. The most known
genetic risk factor is the disease-associated allele, the shared epitope (SE, a specific amino
acid sequence in the peptide binding grove), localized in HLA-DRB1 (18). The strongest
environmental risk factor identified is cigarette smoking (19-21), especially for developing
ACPA-positive RA (22) and for RF-positive RA in individuals carrying the SE (23, 24).
These findings imply that there is a gene-environment interaction between smoking and SE-
alleles triggering a immunological response (25). Previously reported protective factors for
RA have been alcohol intake (26) and high intake of fish oil in the diet (27).

3
A family history of RA among first degree relatives is a significant risk factor for developing
RA (28, 29). The heritability of RA has been estimated to approximately 50% in ACPA
positive RA and to 20% in ACPA negative RA (30). A suggested pathway for the
development of RA is that environmental factors affecting genetically predisposed
individuals (perhaps via epigenetic modifications) leads to post-translational protein
modifications such as altered citrullination of proteins. This results in a loss of tolerance. In
the synovium, CD 4+ T-cells are activated by antigen presenting cells and differentiate to
subgroups of T-cells (Th1 and Th17), leading to activation of B-cells, fibroblasts and
macrophages. In turn, pro-inflammatory cytokines (such as interleukins and tumor necrosis
factor α, TNF-α) are produced and specific B-cells start to produce autoantibodies resulting in
an inflammatory response, and synovitis (6). Both RF and ACPAs can often be detected years
before the clinical onset of RA and are also predictors of disease risk (31-33). It has been
suggested that ACPAs are potentially directly pathogenic through the formation of immune-
complex and macrophage activation (6). Osteoclasts activated by the cytokine release
subsequently leads to bone destruction in the joint (34).

2.1.3 Classification criteria

The RA diagnosis is based on the physician’s assessment in combination with classification
criteria. The classification criteria for RA have primarily been developed for research rather
than clinical practice, to identify homogenous cohorts of patients who share the same
condition. Since there is no “gold standard” to diagnose RA the criteria are not considered
diagnostic criteria. Nevertheless (for lack of a better option) they are often used as such.
Historically, the 1987 American College of Rheumatology (ACR) classification criteria for
RA have been used (35) (see Table 2.1). However, in order to classify RA at an earlier stage,
to take both autoantibodies into account and to identify patients in need of antirheumatic
treatment, the 2010 ACR/ European League Against Rheumatism (EULAR) classification
criteria for RA were introduced in 2010 (36) (see Table 2.2). These criteria should be applied
to patients with at least one synovitis on examination which is not better explained by another
disease. Patients with typically erosive changes in the joints can also be classified as having
RA although not fulfilling the 2010 ACR/EULAR classification criteria (37). Both
classification criteria are applied in the studies included in this thesis.

4
Table 2.1 The 1987 ACR classification criteria for RA

5
Table 2.2 The 2010 ACR-EULAR classification criteria for RA

6
2.1.4 Treatment in RA
Reducing the level of inflammation is the main target of the antirheumatic treatments, since it
is closely linked to the level of symptoms, articular destruction, prognosis and comorbidities
(38). The current approach is to start treatment at an early phase and treat to remission, to
prevent structural damage. Non-steroidal anti-inflammatory drugs (NSAIDs), glucocorticoids
and disease modifying antirheumatic drugs (DMARDs) is the standard initial
pharmacological treatment combination for patients with RA. By definition, DMARDs have
a disease modifying action and reduce disease progression. They can be grouped as
synthetical (sDMARDs) and biological (bDMARDs). sDMARDs can be further subgrouped
into conventional synthetical (csDMARDs) (including agents such as methotrexate,
sulfasalazine, leflunomide and hydroxychloroquine), where the mode of action is to a large
extent unknown, and the recently developed targeted synthetical (tsDMARDs), modulating a
specific target in the inflammatory process (e.g. the Janus kinase inhibitors) (6). The golden
standard for all patients diagnosed with RA, is to initiate treatment with methotrexate and
low-dose glucocorticoids. The dose of methotrexate is increased up to 20 mg per week. In
Sweden, patients are followed-up according to an early RA treatment guideline with a visit to
a rheumatologist every 3-6 month, the first 2 years. If the treatment with methotrexate fails to
get the patient into remission, another sDMARD or bDMARD (if high disease activity or
predictors of severe disease) should be added as additional treatment (39-41). The treatments
grouped as bDMARDs, include primarily four different modes of actions, TNF-inhibition, B-
cell depletion, T-cell costimulation blockade and interleukin 6 inhibition (see table 2.3) (42).
The most common approach when adding a bDMARD is to start with a TNF inhibitor. The
introduction of TNF-inhibitors in the end of the 1990´s led to a major improvement in the
treatment of RA thanks to their potent effect in reducing symptoms, prevent radiographic
progression and joint damage. The goal is to treat the patient into sustained remission and
then consider to taper the antirheumatic treatment over time (39).

7
 Table 2.3 Antirheumatic treatments and mode of action discussed in this thesis
Name Mode of action
csDMARDs
Methotrexate Purine metabolism inhibitor
Leflunomide Pyrimidine synthesis inhibitor
Hydroxychloroquine Decrease chemotaxis and phagocytosis, induce apoptosis
of inflammatory cells, decrease TNF-alpha and IFN-
gamma
Sulfasalazine Induce apoptosis of inflammatory cells, anti-prostaglandin
effect, decrease TNF-α and IL-1
bDMARDs
Adalimumab Anti TNF-α
Certolizumab pegol Anti TNF-α
Etanercept Anti TNF-α
Golimumab Anti TNF-α
Infliximab Anti TNF-α
Abatacept CTLA4-Ig, T-cell costimulation blockade
Tocilizumab Anti-IL 6 receptor
Rituximab Anti-CD 20, B-cell depletion
Anakinra Anti-IL-1
tsDMARDs
Tofacitinib, JAK-inhibitors
Baricitinib,
Upadacitinib,
Filgotinib

Table 2.3 Antirheumatic treatments discussed in this thesis

2.1.5 Measures of disease activity

The follow-up of antirheumatic treatments is based on evaluation of the patients’ disease
activity. Several composite measures of disease activity are used, where DAS 28 is the most
frequently used in clinical practice in Sweden. The disease activity score DAS 28 is based on
the number of a 28-swollen joint count (28-SJC) and a 28-tender joint count (28-TJC),
inflammatory markers (erythrocyte sedimentation rate, ESR or C-reactive protein, CRP) and
the patients’ global assessment of disease activity (patient global, PtGA) (43, 44).
Historically, DAS 28-ESR has been the measure most frequently used, but both DAS 28-ESR
and DAS 28-CRP are used in clinical practice and correlate (45, 46). The range of DAS 28 is
0-9.4 and can be classified as low (DAS 28 <3.2), moderate (DAS 28, 3.2-5.1) and high
(DAS 28 >5.1) disease activity. DAS 28 <2.6 is defined as remission. There are also
response- and remission criteria based on DAS 28 to evaluate antirheumatic treatment (46).
The Boolean remission criteria, requires SJC, TJC, CRP and PtGA to be ≤1. Other measures
of disease activity are the Simplified Disease Activity Index (SDAI) (including 5 items: 28-
SJC, 28-TJC, PtGA, the provider global assessment of disease activity (PrGA) and CRP) and
the Clinical Disease Activity Index (CDAI) (including 4 items: 28-SJC, 28-TJC, PtGA and
PrGA). No calculator is needed to calculate the CDAI or SDAI-score (in contrast to DAS28).

8
The health assessment questionnaire (HAQ), including questions about daily life activities, is
the most frequently used instrument to evaluate physical function in RA patients (47, 48).

2.2 AUTOIMMUNE THYROID DISEASE (AITD)

2.2.1 Pathogenesis and etiology

The thyroid hormones (THs), T4 and T3, are regulators of numerous metabolic processes in
the body. The levels of THs are regulated by the hypothalamus-pituitary-thyroid axis. In this
hormone regulation system, the hypothalamus secretes the thyrotropin-releasing hormone
(TRH) which in turn stimulates the pituitary-gland to secrete the thyroid-stimulating hormone
(TSH). The release of TSH activates the TSH receptor in the thyroid gland to stimulate the
synthesis and secretion of THs (49). Autoimmune thyroid disease (AITD) encompasses a
group of thyroid disorders that leads to infiltration of T-lymphocytes in the thyroid gland,
thyroid autoantibody production and subsequently altered function of the thyroid gland (50).

The main subgroups of AITD are autoimmune hypothyroidism, the end stage of Hashimoto’s
thyroiditis, and Graves’ disease, the most common form of hyperthyroidism. In Hashimoto´s
thyroiditis an inflammation in the thyroid gland driven by CD4+ T-cells leads to activation of
B-cells producing thyroid specific antibodies (anti-thyroid peroxidase antibodies, TPOAb and
thyroglobulin antibodies, TgAb) and activation of CD8+ T-cells. This results in a destruction
of the thyroid parenchyma and a deficient production of THs. Hashimoto´s thyroiditis is
associated with the development of goiter. Decreasing levels of THs and increasing levels of
TSH, due to a negative feedback mechanism, are often seen, but sometimes the levels are
normal (euthyroidism). In approximately 90% of cases the titers of TPOab and 60% of the
TgAb are high (51). The diagnosis of Hashimoto’s thyroiditis is based on the development of
goiter in association with high levels of TPOab. Sometimes fine-needle aspirate and a
cytologic evaluation of the thyroid gland are performed to differentiate from other conditions.
A subclass of Hashimoto’s thyroiditis based on elevated IgG4 levels has been described in
the literature (52). Chronic autoimmune thyroiditis without the development of goiter is
called atrophic autoimmune thyroiditis. The vast majority of all cases with hypothyroidism
are due to a disturbed production of THs in the thyroid gland, primary hypothyroidism. Other
rare causes are central hypothyroidism due to the underproduction of TSH or TRH
(secondary and tertiary hypothyroidism) (51).

In Graves’ disease the activation of CD4+ T-cells lead to the activation of B-cells and the
production of activating TSH-receptor antibodies (TRAbs). The TRAbs increase the
production of THs, the level of THs increases and the level of TSH decreases (50). Typical
symptoms of hyperthyroidism, goiter, increased levels of THs and TRAbs and typical
findings on radiographic examination of the thyroid gland, support the diagnosis of Graves’
disease.

The etiology of AITD is not completely clarified but genetic susceptibility (HLA-DR alleles

9
and CTLA-4 polymorphisms among others) in combination with environmental factors might
induce and propagate thyroid autoimmunity (53). More than 90% of patients with AITD have
been reported to have positive TGAb and/or TPOAb. These autoantibodies have also been
found in healthy individuals, with a prevalence of up to 12% in the general population (54).
Findings on smoking as a risk factor for AITD has not been consistent in the literature. There
has been evidence that smoking may increase the risk of Graves' disease, but on the contrary
decrease the risk for Hashimoto's thyroiditis (55).

2.2.2 Epidemiology and clinical picture of AITD subgroups

AITD is one of the most common autoimmune diseases in the population and the most
common cause of goiters in non-iodine deficient countries, with a reported prevalence in the
general population of approximately 5% (56).

2.2.2.1 Prevalence and incidence of hypothyroidism and hyperthyroidism respectively

The levels of previously reported prevalence figures vary with the geographical region, study
population, definition of hypothyroidism/hyperthyroidism and iodine intake (57). Iodine is
essential for the production of thyroid hormones and may also have a potential antigenic
effect on the thyroid gland (58). Hypothyroidism displays a U-shaped relationship with
iodine, with an increased prevalence in populations with both iodine deficiency and high
intake (58). In areas with iodine sufficiency, Hashimoto’s disease is the main cause of
hypothyroidism (~90%) and Graves’ disease the most common cause of hyperthyroidism (53,
54, 59). Previously reported prevalence of overt hypothyroidism in the general populations
has ranged between 0.2-5.3% (60-64). In a meta-analysis, including seven European studies,
the reported mean prevalence of thyroid dysfunction overall was 6.7 % (95% CI, 6.5-6.9%);
4.9% (95% CI, 4.8-5.1%) for hypothyroidism and 1.7% (95% CI, 1.7-1.9%) for
hyperthyroidism, respectively (61). The definition of thyroid dysfunction was based on the
levels of TSH, THs and for some studies the presence thyroid autoantibodies (TPOAb and
TgAb). The same meta-analysis reported a mean incidence rate of thyroid dysfunction of 259
(95% CI, 254-264) per 100 000; 226 (95% CI, 222-230) and 51 (95% CI, 49-53) per 100 000
for hypothyroidism and hyperthyroidism, respectively. The incidence rate was approximately
5 times higher in women compared to men for both hypothyroidism and hyperthyroidism in
this study (61). Other studies have reported an incidence of hyperthyroidism ranging between
20-40 per 100 000 per year (65). The prevalence of hypothyroidism increases with age, with
a peak incidence between the 30 and 50 years of age (66, 67). The peak age-specific
incidence of hyperthyroidism (Graves' disease) varies between studies. A peak between 20
and 49 years has been reported in two studies, but other studies have reported an increased
incidence with age in Iceland and a peak at 60–69 years in Malmö, Sweden (54). Both
hypothyroidism and hyperthyroidism are 5-10 time more common in women than in men
(65-67).

The disturbed levels of thyroid hormones in hypo- and hyperthyroidism, affecting all cells,
can cause a wide range of symptoms. Frequently reported symptoms of hypothyroidism are

10
weight gain, fatigue, slow thinking, constipation, dry skin, myalgia and arthralgia while the
overproduction of thyroid hormones in hyperthyroidism is associated with weight loss,
tachycardia, tremor, a hypermetabolic state and ophthalmopathy (58, 65, 68).
Musculoskeletal manifestations resembling the symptoms of rheumatic disorders such as
morning stiffness, shoulder girdle pain and weakness and periarthritis have been described in
the literature, for both hypothyroidism and hyperthyroidism (69-71).

2.2.3 The definition of AITD, management and treatment

Most patients with primary hypothyroidism are diagnosed and followed in primary care in
Sweden (72). Hypothyroidism has been treated with chronic thyroxine substitution, an
exogenous form of T4, for more than 60 years. Thyroxine substitution therapy
(levothyroxine) is often initiated with a starting dose of 25-50 µg daily and then increased
every 4-6 week until normalisation of TSH and regress of hypothyroid related symptoms
(72). Thyroxine in monotherapy is the standard treatment, although in some cases, additional
treatment with T3 (liothyronine) may be indicated (73). The majority of patients with
hyperthyroidism in Sweden are diagnosed and initially treated in an endocrinology outpatient
clinic (74). After treatment of hyperthyroidism with either radioactive iodine, anti-thyroid
medications, or thyroidectomy, most patients are also treated with chronic thyroxine
substitution. Other reasons for thyroxine substitution are after the removal of the thyroid
gland due to cancer or because of adverse effects of iodine-containing drugs (lithium,
amiodarone, interferon -alpha) (68, 75, 76). There are no standardized international
classification criteria for AITD, but after excluding reasons for non-autoimmune cause of
thyroxine treatment, the majority of patients treated with thyroxine substitution have to a
large extent an underlying AITD (77-79).

2.3 AITD AND RA

2.3.1 AITD in RA
For overview of published studies of AITD in RA, see Table 2.4. Previous studies have
reported a higher prevalence of AITD in several rheumatic diseases such as Sjögren’s
syndrome, systemic lupus erythematosus and psoriatic arthritis (80-85), and to some extent
also in polymyalgia rheumatica and giant cell arteritis (86, 87). AITD is also one of the most
common autoimmune comorbidities among patients with RA. A large body of evidence,
since the beginning of the 1960s, indicates that the reported prevalence of AITD is increased
in patients with RA compared to the general population. The prevalence of AITD among RA-
patients has varied broadly between studies, with a range of 3-30%, depending on the study
population, geographical region, study design and definition of AITD (78, 88-93). One of the
largest previous studies examining thyroid dysfunction in RA patients by Mahagna et al. (94)
found that the prevalence of hypothyroidism and hyperthyroidism was higher in RA patients
compared to controls (16% vs 11.7% and 2.3% vs 1.8%, respectively). RA was associated
with both hypothyroidism (OR, 1.42; 95% CI, 1.34-1.50) and hyperthyroidism (OR, 1.26;
95% CI, 1.10-1.45). However, some previous studies have reported a lack of, or only a weak

11
association, between AITD and RA (95-97) (see table 2.4). A study by Chan et al. (89)
suggests that RA patients develop AITD at an earlier age than the general population due to
an “autoimmune locus”. They also suggest that concurrent RA may accelerate the
progression from mild thyroiditis to clinical overt hypothyroidism. Another study reported a
stronger link between AITD and RA among younger in relation to older patients (the OR for
hypothyroidism in RA patients, 18-44 years: 2.12; 95% CI, 1.73-2.58, 45-69 years: 1.42; 95%
CI, 1.31-1.54, and 70+ years: 1.37; 95% CI, 1.26-1.49) (94). Higher BMI and socioeconomic
status have been reported to be independent predictors for both hypothyroidism and
hyperthyroidism in RA patients, while smoking was a modest risk factor for hypothyroidism
only (OR, 1.09; 95% CI, 1.04-1.15) (94).

12
Setting/author study design study population Results
Israel, Mahagna, 2018 cross-sectional 11,782 RA patients and The prevalence of hypothyroidism
57,973 controls and hyperthyroidism was higher in RA
patients compared to controls (16%
vs 11.7% and 2.3% vs 1.8%,
respectively). RA was associated with
both hypothyroidism (OR=1.42, 95%
CI 1.34-1.50) and hyperthyroidism
(OR=1.26, 95% CI 1.10-1.45)
Netherlands, Raterman, cohort 358 RA patients Three-fold increased prevalence of
2008 hypothyroidism in RA patients
compared to the general population
(6.8% vs 2.7%)
Canada; Shiroky, 1993 cohort 91 RA patients and 93 30% of RA patients compared to 11%
controls of controls had thyroid dysfunction
(hypothyroidism or hyperthyroidism)
(OR=3.5, 95% CI 1.6-7.5)
Austria, Pongratz, 2000 cross-sectional 383 RA patients and 409 The prevalence of circulating thyroid
controls with osteoarthritis antibodies was significantly higher
among RA patients than those with
osteoarthritis (9.1% versus 3.7%, p =
0.0016)
UK, Somers, 2001 cohort 22,888 RA patients AITD more prevalent among RA
patients compared to general
population controls (standardized
incidence ratio=161.5, 95% CI 145-
180)
Slovakia, Lazurova, 2009 cross-sectional 80 patients with AITD, 80 The prevalence of AITD (TPO-ab) in
patients with RA/SLE, 34 the SLE/RA-group was significantly
healthy controls higher than in the control subjects
(24% vs. 8%, P < 0.05).
Colombia, Roldan, 2012 cross-sectional 800 RA patients The prevalence of AITD was 9.8%
among RA patients (presence of
thyroid antibodies was 37.8% for
TPOab and 20.8% for Tgab)
UK, Silman, 1996 cross-sectional 58 multicase RA families and 8.6% of the RA patients had
504 family members concurrent clinical thyroid disease
(12.1% had Tgab)
USA, McCoy, 2012 cohort 650 RA patients and 650 No significant difference in the
controls prevalence of hypothyroidism (clinical
and subclinical), Graves' disease or
levothyroxine use between groups at
RA diagnosis or at follow-up (the
prevalence of hypothyroidism was
14-16% in both groups)
Sweden, Bengtsson, 2014 case-control 2000 RA patients and 2000 Thyroxin substitution was associated
controls with ACPA-positive (OR=1.9, 95% CI
1.4 to 2.6) and ACPA-negative RA
(OR=2.1, 95% CI 1.5 to 3.1)
Finland, Kerola, 2014 cohort 7209 incident RA patients The standardized rate ratio for
levothyroxine-treated
hypothyroidism preceding RA was
1.51 (95% CI 1.35 to 1.67).

Table 2.4 Previously published studies on AITD and thyroid dysfunction in patients with RA

13
2.3.2 RA in AITD
Previous studies have suggested an increased prevalence of RA in patients with AITD (98,
99). One study reported a prevalence of co-existing autoimmune disease in Graves’ disease
and Hashimoto’s thyroiditis of approximately 10% and 14% respectively (100). In this study
RA was the most frequent co-existing autoimmune disorder with a prevalence of
approximately 3% in Graves’ disease and 4% in Hashimoto’s thyroiditis. Another study
investigating if individuals with one autoimmune disease had a higher risk of a second
autoimmune disorder, found an increased occurrence of RA in patients with AITD
(standardized incidence ratio 131.8; 95% CI, 117-148) (93). Similarly, a review article by
Fallahi et al. (84) of approximately 3000 patients with AITD and 2000 controls found a
slightly increased prevalence of RA of 2.4% of AITD patients with AITD.

In a study by Bengtsson et al. (78), thyroxine treated AITD was associated with a doubled
risk of developing both ACPA-positive (OR, 1.9; 95% CI, 1.4-2.6) and ACPA-negative RA
(OR, 2.1; 95% CI, 1.5-3.1). The risk was severalfold increased (OR, 11.8; 95% CI, 6.9-20.0)
in individuals carrying the shared epitope.

2.3.3 Thyroid autoantibodies in RA patients

Some studies have indicated an increased prevalence of thyroid autoantibodies in patients
with RA, while others have showed somewhat contradictory results. In several studies
examining the levels of thyroid autoantibodies, the results indicate higher levels in patients
with RA compared to the general population (101-104). One study found that the prevalence
of AITD was 9.8% among RA patients while the presence of thyroid antibodies was 37.8%
for TPOAb and 20.8% for TGAb (105). Similarly, another study found approximately the
same levels of TPOAb and TGAb in RA patients, while the prevalence of thyroid hormone
alterations was considerably lower (106). In a meta-analysis by Pan et al. (107), the risk of
having TgAb and TPOAb was approximately three times higher in RA patients compared to
healthy controls (TgAb OR 3.2, 95 % CI 2.2-4.5; TPOAb OR 2.3, 95 % CI 1.2-4.4). By
contrast, in another study there was no significant difference in prevalence of TPOAb in RA
patients compared to healthy controls (108) and other findings indicate that thyroid
autoantibodies can become negative over time (109).

2.3.4 Shared susceptibility genes

The specific pathogenic mechanisms behind the co-existence of AITD and RA are not
completely understood, but these diseases share several susceptibility genes, and the numbers
of identified susceptibility genes have increased rapidly in recent years. Several sequence
variants pointing to genes involved in the regulation of T-cell responses have been identified,
including PTPN22 (protein tyrosine phosphatase non-receptor type 22, involved in
preventing mechanisms of spontaneous T cell activation), CTLA4 (cytotoxic T-lymfocyte-
associated leukocyte antigen, encodes for a negative regulation of the T cell activation) and
the HLA gene complex (human leukocyte antigen), which play an important role (56, 110). It
has been hypothesized that the Th1-driven autoimmune response seen in both AITD and RA

14
reflects a common or at least overlapping immune-pathogenesis (99). A genome-wide
association meta-analysis study, including approximately 30 000 cases and 725 000 controls,
from Iceland and the UK Biobank, identified 99 variants that associate with AITD risk. Of
those, a variant in the tyrosine kinase receptor FLT3 (rs76428106-C) has the largest effect on
the risk of developing AITD (OR 1.46, P = 2.37 × 10−24) and was also associated with
developing RF and/or anti-CCP-positive RA (OR 1.41, P = 4.31 × 10−4) (110). The identified
mutation in FLT3 results in a dysregulation of hematopoietic progenitor and dendritic cells
and subsequently a disturbed immune system and increased disease risk (110).

2.3.5 Timing of onset of AITD in relation to RA

The majority of previous studies describing the co-existence of AITD and RA have used a
cross-sectional design and report an increased prevalence of AITD among patients with RA,
an increased prevalence of RA among patients with AITD or increased levels of thyroid
autoantibodies among patients with RA. However, the temporal relationship between the two
conditions have rarely been approached and these studies do not provide any evidence as to
when the increased co-occurrence comes about, and how the risk of developing these
conditions changes over time. Two previous studies on patients with incident RA describe an
increased prevalence of AITD already at the time of RA diagnosis, raising the question when
the risk starts to increase (78, 111). Data are limited on the risk of developing AITD after the
diagnosis of RA, which is relevant question both from a clinical and etiological perspective.

2.3.6 The effect of AITD on RA disease presentation and treatment response

Several comorbidities, such as diabetes and cardiovascular disease, have been reported to be
associated with a higher disease activity and a decreased likelihood of treatment response in
RA patients (112-115). However, in contrast to the many reports of the co-existence of AITD
and RA, considerably less is known about the impact of AITD on the phenotypic presentation
of RA, or the impact of AITD of the treatment response to antirheumatic drugs. Two review
articles describing the impact of thyroid hormones on the innate and adaptive immune
responses, reported an association between hypothyroidism and higher macrophage
inflammatory response with release of pro-inflammatory cytokines (Il-1, Il-6 and TNF-α)
(116, 117). The fact that AITD may affect the inflammatory response raises the question if
AITD may impact disease activity and treatment response in RA patients. One small cross-
sectional study of prevalent RA patients reported a correlation between hypothyroidism and
higher disease activity (118). Another study of Danish RA patients reported, somewhat
contradictory, no significant difference in RA disease activity at RA diagnosis but a reduced
response to antirheumatic drugs at 4 months follow-up from treatment initiation (119). These
findings call for more profound studies on how AITD impacts RA disease presentation and
disease activity over time, as measured by both patient- and physician-related disease activity
metrics.

15
2.3.7 bDMARDs and AITD
Findings indicate that both AITD and RA may partly have a common immune-pathogenesis.
However, studies investigating whether antirheumatic therapies may have an effect on AITD
occurrence, are limited. TNF-alpha is a proinflammatory cytokine suggested to have a
pathogenic role in both RA and thyroid dysfunction. A previous study found increased levels
of TNF-alpha in both hypothyroid and hyperthyroid patients and a normalization of TNF-
alpha levels was seen after the treatment of the hyperthyroid patients (120). One preclinical
study found that mice with induced autoimmune thyroiditis, treated with anti-TNF-alpha,
experience a reduction of proinflammatory cytokines and less inflammation in the thyroid
gland (121). Another study, using a mouse model, found that exposure to proinflammatory
cytokines (e.g. TNF, IL-6 and interferon) leads to an immune-mediated inhibition of thyroid
function, development of hypothyroidism due to the chronic cytokine exposure and
subsequently clinical signs of hypothyroidism (122). These findings indicate that a subgroup
of individuals with inflammation-mediated AITD may in fact have a reversible condition.

One small pilot study by Raterman et al. (123), of hypothyroid and TPOab-positive RA
patients, found that anti-TNF treatment improves the thyroid function in this group of
patients. A significantly larger decrease in TSH levels after anti-TNF treatment was seen in
the hypothyroid- compared to the euthyroid group. One review article by Bliddal et al. (124),
including six studies, investigated the thyroid autoimmunity and function in a total of 311
patients treated with anti-TNF treatment or the monoclonal CD 20-antibody rituximab. They
conclude a non-significant tendency towards slight improvement of the levels of
autoantibodies and thyroid function in hypothyroid patients. Two of the included studies
report decreased levels of TPOab or TgAb after bDMARD therapy (6 months of adalimumab
treatment and 24-40 months of infliximab or rituximab treatment, respectively) (123, 125).
However, unchanged or increased levels of thyroid antibodies is reported in one of the
included studies after 12 months of infliximab or etanercept treatment (126).

In some studies, rituximab has shown beneficial results when treating Graves’
ophthalmopathy (127, 128). Studies on the effect of csDMARDs, such as methotrexate, on
the development of AITD are limited. However, in a case report the results indicate reduced
inflammatory activity in thyroid related opthalmopathy after methotrexate treatment (129).

2.3.8 Organ and systemic autoimmunity

Autoimmune diseases can be categorized according to different criteria. Based on the target
organ of the autoimmune attack, autoimmune diseases can be grouped into either systemic or
organ-specific. Typically for autoimmune systemic diseases, their autoantigens are found in a
large number of cells in the body, resulting in a damage of different structures and organs. In
organ-specific autoimmune diseases, a specific tissue or organ is the target of the
autoimmune process (130). RA is considered to be a systemic autoimmune disease with non-
organ specific autoantibodies and both joint involvement and extra-articular manifestations
(6). Historically, AITD has been considered an organ-specific disorder with autoantibodies

16
targeted against thyroid-specific components (56). However, some studies argue that Graves´
disease is considered to be a systemic autoimmune disease based on the presence of both
thyroidal and extrathyroidal manifestations (131).

2.4 Validity of the RA diagnosis in the National Patient Register (NPR) and
the Swedish Rheumatology Quality register (SRQ)
In Swedish register-based studies, both the NPR and the SRQ have been used to identify and
include large populations of RA patients. In several studies on RA patients, the descriptive
epidemiology of RA (1, 4, 132-134) and its disease course, antirheumatic treatments and RA-
related outcomes (135-139) have been studied by using the NPR. In parallel to its extended
coverage, the SRQ has increasingly been used to identify large cohorts of RA patients to
study the descriptive epidemiology and related outcomes (1, 133, 140-142).

By using discharge notes from hospitalizations listing RA, the validity of the diagnosis codes
for RA from inpatient care have previously been investigated within a study examining the
risk of lymphoma in RA (143). The medical records were reviewed to confirm the RA
diagnosis. Among the participants, 9% were excluded because of missing information or not
fulfilling the criteria of RA. However, nowadays the majority of patients with RA are only
treated in outpatient care. Previous validations of the diagnosis codes for RA from outpatient
care have also only been based on the included patients within the study. One study found
that 90% of the assigned RA diagnoses are correct when validating the RA diagnosis in a
subset of patients within the study (n=70) (144). To study the validity or the positive
predictive value of the RA diagnosis in the outpatient part of the NPR and the SRQ, is thus
highly relevant as a means to assess whether these registers can be used with high accuracy
for the planned doctoral projects, as well as in other research projects. In this case the positive
predictive value can be expressed as the probability of an individual coded with an RA
diagnosis truly having the disease.

Since the health care system and registers differ between countries, previous validation
studies from Sweden are not directly comparable with corresponding international studies.
However, the validity of an RA diagnosis among RA-coded individuals from a primary care
database in the UK -population (the General Practice Research Database) seemed high for
patients with specific characteristics. Using a data-derived diagnostic algorithm with a  > 80%
sensitivity and specificity, 61% of RA-coded patients fulfilled the diagnostic algorithm (145).
Another validation study, using the medical records of rheumatologists in Ontario, Canada,
show a high level of accuracy where 84% of the patients also have an RA diagnosis
registered in the administrative data (146). A chart abstraction study from the same region
report a positive predictive value ranging from 51-83% amongst the primary care sample, and
ranging from 55-80% in the rheumatology sample, depending on the applied diagnostic
algorithm (147). One study examining the validity of the RA diagnosis in the Danish Danbio
clinical register and the Danish National Patient Registry, have found a proportion of true RA
cases of 96% and 79%, respectively (148).

17
In addition, it is of clinical interest to study if the validity of the RA diagnosis varies between
the classification criteria that have been used historically (1987 ACR), compared to the newer
classification criteria for RA (2010 ACR/EULAR) that are now being used in clinical
practice (35, 36). Previous studies have validated the RA-diagnosis primarily by using a
cross-sectional approach. Since RA is a chronic condition it should typically be possible to
verify the RA diagnosis over time. There is thus a need for assessing the RA diagnosis
longitudinally and also to take the temporal validity of the RA diagnosis into account.

Register-based studies have often investigated morbidity in relation to RA onset. Through

different register-based algorithms, previous studies have used algorithms to identify incident
RA (149, 150). Little is known if the patients who are registered as incident RA in the
registers also have a true new-onset disease. To validate if the RA patients identified as
incident from our registers also correspond to new-onset disease is therefore also an
important task for research.

18
 OBJECTIVES
3.1 OVERALL OBJECTIVES
The overall aim of this thesis was to address etiological and clinical issues related to the
impact of one of the most common comorbidities in RA patients, AITD, and to better
understand the impact of RA on AITD.

3.2 SPECIFIC AIMS

For overview of the objectives in relation to time from RA diagnosis see Figure 3.1.

The specific aims in the four sub-studies of this thesis were the following:

1) To assess the validity of the RA diagnoses in the registers used to identify patients
with RA in our studies (Study 1)
2) To investigate when, in time-wise relation to the clinical diagnosis of RA, the
increased risk of AITD in RA occurs (Study 2)
3) To examine whether AITD affects RA disease activity, or response to antirheumatic
treatment with methotrexate (Study 3)
4) To investigate whether antirheumatic treatments in RA patients may prevent the
development of AITD (Study 4)

Figure 3.1 Overview objectives in relation to time from RA diagnosis

19
 MATERIALS AND METHODS
4.1 SETTING
Since the start of using church books to register birth and death records in the 1600’s, there
has been a tradition in collecting information on the inhabitants in Sweden. In the 20th
century many health care registers were initiated. This opened the way for the nationwide
healthcare and demographic registers that we use in research today. The Swedish personal
identification number assigned to every inhabitant since 1947, allows linkage between
different health care registers with almost complete coverage and provides an excellent
setting for epidemiological studies (151). The Swedish health care system is publicly funded
and the majority of patients with RA are followed in hospital-based specialist care. Through
specific algorithms, patients with RA can be identified from the health care registers and
information about comorbidities and treatment can be collected. In addition, these individuals
can be matched to healthy controls in the population to compare for instance background risk
for comorbidities.

4.2 DATA SOURCES

The data sources used in the different sub-studies of this thesis were gathered from both
medical charts, a quality register and national registers. Illustration of included registers see
Figure 4.1.

4.2.1.1 The Swedish Rheumatology Quality Register

The Swedish Rheumatology Quality Register (SRQ), maintained by the Swedish Society of
Rheumatology, is a clinically integrated nationwide register. The SRQ started 1996 with the
inclusion of patients with incident RA. The register extended to include treatments with
bDMARDs in 1999 (the Swedish Biologics Register, ARTIS) and subsequently other
rheumatic diseases and prevalent RA cases. The SRQ is used to follow the patients’ disease
course longitudinally. Patients with RA represent the largest group and more than 54 000
individuals with this diagnosis are registered in the SRQ (152). Health related parameters,
patients’ and physicians’ assessment of disease activity, and antirheumatic treatments, are
registered in the SRQ. The antirheumatic treatments are registered with a start date, dose and
a stop date, if the patient has ended the treatment. The National Board of Health and Welfare
has reported an overall high coverage of the SRQ (85.8%) (153). By using the SRQ, patients
with RA were identified for study I-IV, information on disease activity and patient-related
parameters were extracted for study III, and antirheumatic treatments identified for study III
and IV.

4.2.1.2 The Total Population Register

The Total Population Register, run by Statistics Sweden since 1968, contains demographic
information such as date of birth and death, sex, emigration and residential area (154).

21
Information from the Total Population Register was used to identify general population
comparators and to extract demographic information used in study II and IV.

4.2.1.3 The National Patient Register

The Swedish National Patient Register (NPR) started in 1964 and is administered by the
National Board of Health and Welfare. The NPR includes primary and contributory diagnosis
codes from inpatient care, and since 2001, also from outpatient visits from non-primary
outpatient care (155). The diagnoses codes are registered according to the International
Classification of Diseases (ICD) system. In addition, hospital department, admission- and
discharge dates are registered. The NPR is used in study I to identify RA patients and study
II-IV to identify comorbidities.

4.2.1.4 The Swedish Prescribed Drug Register

The Swedish Prescribed Drug Register (PDR) includes information on all Swedish
dispensations of prescribed drugs since 2005. Every drug is registered according to the
Anatomical Therapeutic Chemical Classification System (ATC), including date, dose and
information on the prescriber and patient (156). Drugs administered to patients in hospitals,
for example intravenous bDMARDs (rituximab and infliximab) are usually not registered in
the PDR. In study I, the PDR was used to identify antirheumatic treatments, and in study II-
IV to identify AITD-related medications.

4.2.1.5 The Swedish Cancer Register

The Swedish Cancer Register run by the National Board of Health and Welfare, includes all
diagnoses codes of cancer since 1958. It is mandatory for all clinicians and pathologists to
report to this register, resulting in a very high coverage (157). In study II-IV, the Swedish
Cancer Register was used to identify thyroid cancer in patients.

Figure 4.1 Illustration of included registers

22
4.3 STUDY DESIGN AND STUDY POPULATION

4.3.1 Overview
All studies in this thesis are register-based. A cohort design was used in study II-IV, although
a case-control design was also used in study II. The RA diagnosis in study I was based on an
ICD code for RA from a rheumatology clinic in the NPR. In study II-IV the RA diagnosis
was based on the inclusion with a RA diagnosis in the SRQ. The diagnostic algorithm to
identify RA patients in the NPR was validated in study I. The definition of RA in the SRQ,
based on the inclusion with a RA diagnosis reported by a rheumatologist, was also validated
in study I. Through linkage with previous mentioned registers, the study populations,
exposures, outcomes and covariates could be identified. SAS statistical software (version 9.4)
was used in all studies.

4.3.2 Study I

4.3.2.1 Study design

Study I is a validation study, validating the proportion of patients with register-based RA
diagnosis in the NPR, who also fulfilled formal classification criteria or clinical diagnosis of
RA, 2005-2011.

4.3.2.2 Study Population

Through the NPR we randomly extracted and included 211 patients, of whom the majority
was also included in the SRQ. The prevalent group consisted of 100 patients with two or
more visits to a rheumatologist between 2005 and 2008, listing an ICD diagnosis code for RA
(M05.9 or M06.0). The diagnosis was evaluated at the second consecutive visit.

The incident group consisted of 111 individuals defined as patients with their first ever visit
to a rheumatologist listing a RA diagnosis in 2008 (base-case definition). In addition, a
stricter definition of incident RA was validated. This definition included two additional
criteria: a) a second visit within one year after the first visit, and b) no DMARD treatment > 6
months before the first visit with RA diagnosis. For the incident group the diagnosis was
validated at the “incident time point”, the date of the first visit listing a diagnosis code for
RA.

The register-based diagnoses were validated against both the 1987 ACR- and the 2010
ACR/EULAR classification criteria as well as the clinical diagnoses of RA, by extracting
diagnosis-related parameters from the medical files. A prespecified 20-item abstraction form
was used to include the parameters of interest. We determined if the patient would clinically
be regarded as having an incident disease, and if the RA diagnosis remained two years after
the first visit for the incident patients. For the patients not fulfilling criteria or qualifying for
the clinical diagnosis of RA, the alternative diagnoses were charted. The proportion of

23
patients developing radiographic changes up to the end of the study period, three years later,
was also explored.

4.3.3 Study II

4.3.3.1 Study design

Study II is a combined case-control and cohort study to examine the risk of AITD, 2006-
2013, among patients with RA at different time points before and after the diagnosis of RA
compared to the general population (see Figure 4.2).

Figure 4.2 Illustration of the study design in study II

4.3.3.2 Study Population

In the case-control part of the study, we identified patients with new-onset RA (symptom
duration < 12 months) included in the SRQ, 2006-2013 (n=8090). By linkage to the Total
Population Register, each RA patient was individually matched to ten randomly selected
comparators subjects from the general population, based on age, sex and residential area
(n=80 782) using risk-set sampling. The comparator subject had to be alive at the date of RA
diagnosis of their individual RA patient. The selection and matching criteria were the same
for the cohort part of the study. Participants with a history of AITD at RA diagnosis, a first
thyroxine prescription before 2006 (see definition of incident AITD that follows) and an
index date later than December 31, were excluded. In the cohort part of the study, 6515 RA
patients and 58 228 controls were included and the outcomes of interest were analyzed.

Risk-set sampling

In risk-set sampling the control is selected at the time-point when the corresponding case is
identified. The controls are selected from the source population of individuals without the
outcome at the specific time-point, but who are eligible to have the outcome and become a
case during follow-up (158).

24
4.3.3.3 Exposure and outcomes
AITD

The main outcome measures were the prevalence and relative risk of incident AITD before
and after RA diagnosis compared to the population comparators. AITD was defined as
having received a prescription of thyroxine after excluding non-autoimmune indications. The
prescriptions were based on the ATC codes for hormone substitution therapy, levothyroxine
or liothyronine (H03AA01= levothyroxine, T4 (levaxin+euthyrox), H03AA02= liothyronin,
T3 (liothyronin)). By linkage to the PDR, individuals treated with thyroxine substitution
between 2005 and 2013 were identified.

Individuals with a history of thyroid cancer, identified through linkage to the Swedish Cancer
Register (icd7=194, icd9=193, icd10=C73), were excluded. Also, individuals with a
prescription in the PDR of iodine-containing drugs, lithium, amiodarone, or interferon-alfa
(ATC-code: C01BD01 (Amiodarone), N05AN01 (Litium), L03AB01/L03AB04/L03AB05
(Interferon-alfa/alfa 2a/alfa 2b)), were excluded.

Prevalent and incident AITD

Prevalent AITD was defined as a history of ≥1 thyroxine prescription in the PDR from 2005
through 2013. A prescription in the PDR in 2005 could be either a renewed prescription or a
first-ever prescription. Therefore, incident AITD was defined as the first thyroxine
prescription in 2006 or later. Participants with a prescription in 2005 were excluded when
analyzing incident AITD.

Subgroups of AITD

Through the ICD codes in the NPR, AITD was subgrouped into hyperthyroidism or
hypothyroidism. Hyperthyroidism was defined as having received a prescription for
thyroxine and an ICD-9 or ICD-10 codes for Graves’ disease or thyrotoxicosis (E050, E051,
E052, E053, E058, E059, 242, 242A, 242B, 242D, 242E)

The majority of cases with hypothyroidism are treated in primary care and are not registered
in the NPR. Hypothyroidism was therefore defined as a prescription for thyroxine and an
existing ICD-9 or ICD-10 code for hypothyroidism (E038, E039, E063, 244x, 245c) or the
absence of an ICD code for hyperthyroidism.

4.3.4 Study III

4.3.4.1 Study design

In this cohort study we examined if AITD impacts disease activity or response to
methotrexate in patients with incident RA, 2006-2016 (see Figure 4.3).

25
Figure 4.3 Illustration of the study design in study III

4.3.4.2 Study population

By using data from the SRQ we assembled a cohort of patients with new-onset RA who
started methotrexate in monotherapy at diagnosis and who had disease activity measures
registered both at start and at follow-up visits (n=9004). Comorbidity with AITD before RA
diagnosis, was identified by linkage to the PDR and the cancer register, as in study II.

4.3.4.3 Follow-up and outcomes

Disease activity

The disease activity index DAS 28-ESR was used to define the RA disease activity and the
change in RA disease activity (ΔDAS28-ESR) between the treatment start and the 3-month or
6-month visit. DAS 28-ESR is referred to as DAS 28 below.

Treatment response

The EULAR DAS 28 response criteria, based on the level and change in DAS 28 at follow-
up, was used to define non-, moderate and good responders (46). A patient was also defined
as a non-responder if switching from methotrexate to another DMARD, before or at the
follow-up visit.

Patient-reported disease activity variables

The patient-reported outcome measures (PROMs) extracted from the SRQ were the
following: Patient Global VAS (patient assessment of disease activity on a 100 mm visual
analogue scale, (0 = best, 100 = worst)), VAS pain and Health Assessment Questionnaire
Disability Index (HAQ). We compared RA disease activity (DAS 28), its components,
PROMs, and EULAR-response, between RA patients with and without AITD.

26
4.3.5 Study IV

4.3.5.1 Study design

In this cohort study we investigated if antirheumatic therapies lower the risk of developing
AITD in RA patients, during the study period 2006-2019.

4.3.5.2 Study population

From the SRQ we identified patients with new-onset RA, included in the SRQ between 2006
and 2018 (n=15 558). Each RA patient was individually matched on age, sex and residential
area with five population comparators (n=77 685), through linkage to the Total Population
Register, according to risk-set sampling. Participants with a history of AITD at RA diagnosis
were excluded. Overall, 13 731 RA patients and 63 201 matched controls were included in
the study.

4.3.5.3 Exposure
DMARD treatments

By using information from the SRQ/ARTIS, we identified treatment with DMARDs and their
corresponding start and stop dates, during the study period. The DMARDs were grouped as
bDMARDs (primarily the four main modes of action TNF-inhibitors, B-cell
depletion/rituximab, T-cell co-stimulation blockade and interleukin 6 inhibition),
methotrexate and other csDMARDs.

4.3.5.4 Outcome
Incident AITD

The outcome measure was the relative risk of incident AITD after RA diagnosis in RA
patients (bDMARD and non bDMARD treated) and compared to the population controls.
AITD was defined as a prescription for thyroid hormone substitution therapy 2005-2019,
with exclusion of non-autoimmune indications as in study II-III, by linkage to the PDR and
the Swedish Cancer Register.

4.4 STATISTICAL ANALYSES

The aim in many research studies is to investigate the causality between an exposure (e.g.
RA) and an outcome (e.g. AITD). In a randomized control trial (RCT) it is possible to explore
causality if the pre-exposure variables are equally distributed between the exposure groups.
However, the statistical methods used in epidemiological research only allow to describe an
association between an exposure and an outcome rather than a causal relationship.
Conducting a RCT to address the research questions in this thesis would undoubtedly be
difficult to apply in practice and the included observational studies offers other possibilities
when it comes to the size of the study population, follow-up time and feasibility.

27
4.4.1 Statistical concepts

4.4.1.1 Linear regression

Regression analysis can be used to study the effect of one variable (the independent variable,
exposure x) on another variable (the dependent variable, outcome y). The most common
regression models in epidemiological research are linear regression, logistic regression and
Cox regression. Linear regression is based on the equation of the line (y=a+bx) and is used to
predict the value of the dependent variable y given the value(s) of the other variable(s), x (a
represent the intercept and b the coefficient/slope). Linear regression is applied when the
dependent variable is continuous.

4.4.1.2 Logistic regression

Logistic regression can be applied when the dependent variable is binary (e.g. RA yes/no).
The independent variable can be either binary, continuous or categorical. The logistic
regression determines the relationship between the independent (explanatory) variable and
the dependent variable and can be used to predict the binary outcome when the independent
variable is present. Logistic regression is often applied in case-control studies. The effect
measure in the logistic regression is odds ratios (ORs). The OR can be explained by the ratio
of an event occurring in the exposed group compared to the ratio of an event occurring in the
unexposed group (e.g. ratio of a ratio). OR is not equal to the relative risk (RR, ratio of a risk)
but the OR can approximate the RR if the “rare disease assumption” holds. This is the case if
the prevalence of the disease is < 10%. In case-control studies where risk-set sampling has
been applied, the OR can also approximate the risk ratio, irrespective of the outcome being
rare or not (158). Logistic regression can be subgrouped into conditional and unconditional
logistic regression. In a study design where the case is matched to a control, conditional
logistic regression is applied, and the matched cluster is compared to each other. Otherwise
unconditional logistic regression is used, where all cases are compared to all controls.

4.4.1.3 Cox proportional hazard regression

The cox proportional hazards regression is applied in cohort studies when the outcome is time
to an event, to estimate the effect of one or several exposures. This time-to-event analysis
estimates the (relative) risk of an event during a time-period where the incidence rates of the
event can vary. Different time-scales can be used such as follow-up time or attained age
depending on what is appropriate in the specific study. In Cox regression individuals are
censored if they are lost to follow-up, for example due to emigration. Censoring refers to that
the follow-up time of an individual is stopped without having had the outcome. The effect
measure in cox regression is hazard ratio (HR). The hazard ratio describes the relation
between two survival curves (e.g. hazard rate for the exposed/hazard rate for the unexposed).
The Cox regression is based on the proportional hazards assumption, meaning that the hazard
ratio of the two compared groups must stay constant over time. The proportional hazards
assumption can be tested in several ways, for example by visual inspection of the survival
curves or by the introduction of an interaction term in the statistical model.

28
A p-value <0.05 was considered significant in study II-IV.

4.4.2 Statistical analyses in the specific studies

4.4.2.1 Study I
For the register linkage to identify the RA patients and for the descriptive statistics, mean
(SD), and median (25th-75th) of the study population, SAS statistical software (version 9.4)
was used in study I. Diagnosis related parameters where extracted from the medical files, by
using a 20-item abstraction form, to evaluate the RA diagnosis against the classification
criteria and clinical diagnosis of RA. The data from the abstraction form from each individual
was compiled and summarized for the prevalent, base-case incident and strict incident RA
patients separately.

4.4.2.2 Study II
In study II a matched case-control design was used to assess the prevalence and relative risk
of prevalent AITD at RA diagnosis. To assess the relative risk of incident AITD before the
diagnosis of RA, a matched case-control design was also applied with the RA patients as
cases, the reference participants as controls and AITD as the exposure. The relative risk
(estimated by OR) for AITD in different strata of time (<3 months, 3 to <12 months, 12 to
<24 months, 24 to <60 months, and ≥60 months) before RA diagnosis, was calculated, by
using conditional logistic regression with adjustment for the matching factors.

To assess the relative risk of incident AITD after RA diagnosis (the start of follow-up) a
matched cohort design was used. Here the RA patients were defined as the exposed cohort,
the reference comparators as the unexposed cohort and AITD the outcome. Individuals with a
history of AITD at the start of each follow-up interval were excluded. By using Cox
proportional hazards regression with adjustment for the matching factors, the relative risk
(estimated by HR) of AITD was calculated overall and in strata based on time since RA
diagnosis (0 to <3 months, 3 to <12 months, 12 to <24 months, 24 to <60 months, and ≥60
months). The participants contributed with time (person-years) from start of follow-up until
onset of AITD, death, migration or end of study (December 31, 2013).

The analysis of incident AITD was stratified by age, sex and RA serostatus and further
subgrouped on hypothyroidism and hyperthyroidism. In a second step, we also adjusted for
comorbidities possibly associated with the detection of AITD and the number of physicians
visits in the NPR.

4.4.2.3 Study III

The statistical analysis in study III was performed in three steps. First, we investigated the
association between prevalent AITD and RA disease activity at diagnosis by using linear
regression. The disease activity was measured as DAS 28, its components, HAQ and pain.

29
The disease activity metrics were the dependent variables and AITD the independent variable
in the linear regression, with adjustment for sex and age (see Figure 4.4).

Figure 4.4 Illustration of the statistical concept, step one, for study III

In a second step, by using linear regression we examined the association between AITD and
the change (delta, Δ) of the disease activity parameter between the RA diagnosis and the 3-
and 6-month follow-up visits (see Figure 4.5).

Figure 4.5 Illustration of the statistical concept, step two, for study III

In a third step, by using logistic regression we investigated the association between AITD and
response to methotrexate 3 and 6 months from treatment start. EULAR DAS 28 good
responders (cases) were compared with moderate/non-responders (controls) and AITD was
the exposure. Additional adjustment was made for HAQ, smoking and the use of oral steroids

30
in additional models (see Figure 4.6).

Figure 4.6 Illustration of the statistical concept, step three, for study III

Multiple imputation

Since approximately 20% of the visits had missing values on the DAS 28 variables, two
separate analyses were performed. The first including all RA patients with a 3-month and/or
6-month visit. The second, limited to the patients with complete information on the DAS 28
variables. For those patients with missing information on EULAR DAS 28 response, the
values were imputed by using multiple imputation. In addition, we imputed information on
HAQ, the use of oral steroids and smoking, if there was a missing value. In the multiple
imputation, fifty imputed datasets were created and analyzed by logistic regression applying
the generalized-estimating-equation method.

4.4.2.4 Study IV
In study IV we investigated the risk of incident AITD following RA diagnosis in three
different models. In a first step, we used a matched cohort design where the RA patients
formed the exposed cohort, the population comparators formed the unexposed cohort and
incident AITD was the outcome (excluding prevalent AITD). Participants were followed
from RA diagnosis/index date until either onset of AITD, death, migration or end of study
(December 31, 2019). Cox proportional hazards regression was used to calculate relative
risks (HR) overall and by time since RA diagnosis (0 to<1, 1 to <2, 2 to <5, 5 to < 10, and 10
to <14 years). The corresponding incidence rates for AITD was calculated for the same
periods of time.

In a second step, the relative risks of AITD in RA patients, during bDMARD and non

31
bDMARD treatment, were compared vs. the population controls. In this analysis, bDMARD
treatment was treated as a time-varying exposure, based on all treatment episodes during
follow-up of the study. Accordingly, patients could contribute with person time to both the
bDMARD exposed and to the non-bDMARD exposed cohort.

In a third step, the relative risks of AITD were compared between bDMARD vs. non
bDMARD episodes among patients with RA. Treatment with different bDMARDs was
counted as one treatment episode if it belonged to the same treatment subgroup and the time
between treatments was within a specified time interval (<180 days for Rituximab and <90
days for other bDMARDs). The first bDMARD treatment was included in the main analysis.
In addition to the main analysis, sensitivity analyses were performed including several
bDMARD treatment episodes.

All analyses were adjusted for age, sex and calendar year. Additional analyses were
performed based on subsets of RA patients and csDMARD co-medication. Furthermore,
several sensitivity analyses were performed for different bDMARD treatments (TNF-
inhibitors/rituximab/other bDMARDs), additional adjustment for treatment with methotrexate
and for different time-scales in the Cox regression.

32
4.5 ETHICAL CONSIDERATIONS
The ethical concern in all research, including register-based studies, is the protection of data
privacy of the participants. The Declaration of Helsinki was developed by the World Medical
Association to promote ethical principles for research involving human subjects (159). These
guideline states that the ethical considerations in a planned research project should be
examined by an ethical committee before conducting the research. In Sweden, six Regional
Ethical Committees of the Ethics Review Authority, examine applications for ethics review
of research, in accordance with the Act of Ethical review (160).

In general, research on human subjects postulates informed consent from the study
participant. According to Swedish legislation, informed consent is not a requirement in large
scale register-based studies due to the large number of participants, since this would not be
feasible. However, the register-based study must be approved by an ethical committee and
the research must be considered to benefit patients with the same condition and contribute
with knowledge presumably unattainable in another type of research settings.

All studies in this thesis were approved by the Stockholm Regional Ethical Committee. In
study I, information was collected from the medical files. In study II-IV the National Board
of Health and Welfare replaced the personal identification number with an anonymous
participant number before transferring data from the registers. The data at the Clinical
Epidemiology Division are stored in secure servers and the access is limited to study specific
researchers. We believe that the studies in this thesis are clinically relevant and contributes
with knowledge on RA and AITD that may be beneficial to patients with the same conditions.

33
 MAIN RESULTS
5.1 STUDY I
The results showed that among the register-based prevalent RA patients from the NPR, the
chart review indicated that 91% fulfilled the classification criteria or clinical diagnosis of RA
(see Table 5.1). Among the patients who did not have RA, all but one were identified with
another inflammatory rheumatic disease. Of the patients verified with RA, 79% fulfilled both
classification criteria. Although six patients did not fulfill the classification criteria at the
time-point of evaluation, all the available data in the medical file supported the RA diagnosis.
For the majority of these patients, the classification criteria were fulfilled later on.

Table 5.1 Summary of medical records of prevalent patients registered with RA in the
Swedish National Patient Register

Total number of subjects 100

Verified RA 91(91%)
1987 ACR and 2010 ACR/EULAR criteria 72/91(79%)
1987 ACR criteria only 9/91(9.9%)
2010 ACR/EULAR criteria only 4/91(4.4%)

RA diagnosis not substantiated 9(9%)

Granulomatosis with polyangiitis 2
Reactive arthritis 1
Polyarthritis 4
Unspecified arthralgia 1
Psoriatic arthritis 1

n, values given as number of patients and (%) given as the equivalent percentage

Of the patients identified with incident RA according to the “base-case” definition, 83%
fulfilled the RA diagnosis at the date of evaluation and after the two-year follow-up period. In
the subgroup defined as “strict” incident, the RA diagnosis was confirmed in 91% of whom
92% also had true new-onset disease (see Table 5.2). Among these patients 87% fulfilled
both classification criteria for RA.

35
Table 5.2. Summary of medical records of incident patients in the strict definition of
register-based incident RA in the Swedish National Patient Register

Total number of subjects 82

Verified RA 75(91%)
thereof incident 69(92%)
1987 ACR criteria and 2010 ACR/EULAR criteria 60/69(87%)
1987 criteria only 6/69(8.7%)
2010 ACR/EULAR criteria only 1/69(1.4%)

RA diagnosis not substained 7(8.5%)

Psoriatic arthritis 3
Unspecified polyarthritis 4
n, values given as number of patients and (%) given as the equivalent
percentage.

n, values given as number of patients and (%) given as the equivalent percentage.

Validation of the patients who were also registered in the SRQ showed that the corresponding
number was 96% for the prevalent patients and 91% for the incident patients of whom 94%
could also be verified with new-onset disease.

5.2 STUDY II
At the time of RA diagnosis, the prevalence of AITD was 10.3% among RA patients (n=832)
compared to 7.1% among the controls (n=5725) (OR, 1.5; 95% CI, 1.4-1.7).

The relative risk of incident RA was assessed up to 7 years before and 8 years after RA
diagnosis. During this period of time 374 RA patients and 2587 controls developed incident
AITD. The overall risk of incident AITD was increased among RA patients compared to
controls before RA diagnosis (OR 1.7; 95% CI 1.5-2.0) (see Table 5.3). The increased risk
seen among RA patients developed during a 5-year period before RA diagnosis (OR, 1.5;
95% CI, 1.2-1.8) and had a peak at the time of RA diagnosis (0-3 months before) (OR 5.3;
95% CI, 3.7-7.6). Following RA diagnosis, the risk of developing AITD gradually decreased
below the expected rate (2-5 years after RA diagnosis, HR, 0.7; 95% CI, 0.5-1.0) (see Table
5.3 and Figure 5.1).

Analyses stratified by age and subtype of RA displayed higher ORs in the younger age group
(16-49 years: OR, 2.8; 95% CI, 2.2-3.5) and in seropositive RA (OR, 1.8; 95% CI, 1.5-2.1).
Additional adjustment for comorbidities and the number of physicians’ visits had little impact
on the risk estimates.

36
Table 5.3. Relative risk of incident AITD before, and then after, the diagnosis of RA in
7489 patients with RAa in the SRQ compared with 70 965 matched population controlsa

Characteristics Patients with RA, No. Population controls, No. OR (95% CI)b
Overall before RA diagnosis 255 1442 1.7 (1.5-2.0)
Sex
women 219 1226 1.7 (1.5-2.0)
men 36 216 1.6 (1.1-2.3)
RF status among cases
Positive 164 875 1.8 (1.5-2.1)
Negative 81 501 1.5 (1.2-1.9)
NA 10 66 1.5 (0.7-2.9)
Time between AITD and RA diagnosis
0-<3 months 47 84 5.3 (3.7-7.6)
3-<12 months 49 233 2.0 (1.5-2.7)
12-<24 months 42 284 1.4 (1.0-1.9)
24-<60 months 96 627 1.5 (1.2-1.8)
>60 months 21 214 0.9 (0.6-1.4)
Age at inclusion in SRQ
16-49 years 86 315 2.8 (2.2-3.5)
50-74 years 135 893 1.4 (1.2-1.7)
>74 years 34 234 1.3 (0.9-1.9)
HR (95% CI)b
Overall after RA diagnosis 119 1145 0.9 (0.8-1.1)
Time since RA diagnosis
0-<3 months 15 62 2.1 (1.2-3.7)
3-<12 months 26 205 1.1 (0.7-1.7)
12-<24 months 29 249 1.1 (0.7-1.6)
24-<60 month 40 485 0.7 (0.5-1.0)
>60 months 9 144 0.6 (0.3-1.1)
a
Participants with thyroxin treatment and treatment with iodine-containing drugs or a history of thyroid cancer
excluded; b Adjusted for matching factors; age, sex, residential area

37
Figure 5.1 Relative Risk of incident AITD before and after the diagnosis of RA

Analyses restricted to hypothyroidism (94.7% and 90.2% of all incident AITD for RA
patients and population controls, respectively), demonstrated similar pattern of relative risks
as for the whole AITD group. For hyperthyroidism, there was no overall increased risk for
hyperthyroidism at RA diagnosis in RA patients compared to population controls (OR, 1.1;
95% CI, 0.7-1.8), but an overall decreased risk following RA diagnosis (HR, 0.2; 95% CI,
0.05-0.8). Please see tables in paper II for the complete results in AITD subgroups.

5.3 STUDY III

The prevalence of AITD was 11% (n=1003) among the 9004 included RA patients. The
proportion of women was 88% in the AITD positive- compared to 65% in the AITD-negative
group. At the time of diagnosis, the RA patients with AITD, compared to those without, had
significantly higher PROMs; Patient Global VAS 57 vs. 52, HAQ 1.2 vs. 1.0 and VAS pain
58 vs 53 (p< 0.05 for all PROMs) (see Table 5.4). This resulted in a slightly higher DAS 28
(5.2 vs 5.1). The objective disease activity measures (ESR and swollen joint count) were not
statistically higher in patients with vs. without AITD.

38
Table 5.4 DAS 28 parameters at baseline and at the 3-month visit for RA patients in the
SRQ

For each of the seven variables named in the first column and for each time point, the estimate (β) was provided
by a linear regression model for which the variable in the first column was the dependent variable, AITD the
independent variable, and age and sex used as adjustment. The estimate β gives the mean difference (given the
linear model) between the two types of patients, a positive value meaning that the AITD+ patients have on
average a higher value than the AITD− patients.

When comparing the change in RA disease activity between baseline and the 3- and 6-month
follow-up visit, patients with AITD decreased significantly more in Patient Global VAS at
the 3-month visit (see Figure 5.2). For the other disease activity parameters under study, there
were only small numerical differences for those with vs. without AITD between visits.

Our result demonstrated that overall AITD had little impact on EULAR-response to
methotrexate at three months (OR of non/moderate response: 0.95; 95% CI, 0.8-1.1), nor at
six months follow-up visit. The results were similar when restricting the analysis to non-
imputed data on response status. However, in separate analyses stratified by age, there was a
trend indicating a lower chance of being a good responder with concomitant AITD at the 3
months visit among younger patients (<45 years) (OR of non/moderate response: 1.44; 95%
CI, 0.76-2.76). This lack of response was more pronounced at the 6-month follow-up visit for
this subgroup of patients (OR of non/moderate response: 2.75; 95% CI, 1.04-7.28). Please see
supplementary tables in paper III for the complete results.

39
Figure 5.2 Change in DAS28 Components between month 0 and 3 months, and EULAR
DAS28-ESR Response at 3 months

40
5.4 STUDY IV
Counting from RA diagnosis, with a mean follow up of 6.3 years (RA patients) and 6.4 years
(controls), 2.3% (n=321) of the RA patients and 2.9% (n=1838) of the controls developed
AITD. This corresponded to an incidence of 3.7 vs. 4.6 per 1000 person-years, and a
decreased risk of AITD in RA patients vs. their population controls, HR, 0.81; 95% CI, 0.72-
0.91. The incidence and relative risk of AITD gradually decreased over time from RA
diagnosis (e.g. 10-14 years after RA diagnosis: 2.9 vs. 4.5 events per 1000 person-years, HR,
0.64; 95% CI, 0.37-1.09). Please see the supplementary information in paper IV for the
complete results.

Furthermore, when we assessed the risks separately for RA patients based on bDMARD
treatment status, compared to the general population, the risk of incident AITD was
significantly decreased in RA patients not receiving bDMARD treatment (HR, 0.84; 95% CI,
0.74-0.96). The decreased risk was even more pronounced in the group of patients receiving
bDMARD treatment (HR, 0.54; 95% CI, 0.39-0.76) (see Figure 5.3).

Figure 5.3 Relative risk of incident AITD in early RA patients, during bDMARD
treatment and non bDMARD treatment, compared to the general population

41
In analyses restricted to the comparison between RA patients, the risk of incident AITD was
decreased during bDMARD treatment compared to non bDMARD treatment (HR, 0.65; 95%
CI, 0.45-0.93).

When including all the treatment episodes during follow-up, the protective association was
most pronounced for TNF-inhibitors (vs. no TNF inhibitor treatment), HR, 0.67; 95% CI,
0.47-0.96 (see Table 5.5). Subset analyses based on sex and rheumatoid factor status provided
similar results as the main analysis. Likewise, including several bDMARD treatment
episodes also yielded similar results compared to the main analysis. Additional adjustment for
co-medication with methotrexate had little impact on the risk estimates seen in the main
analysis.

Table 5.5 Relative risk of incident AITD in early RA patients receiving bDMARD
treatment compared to non bDMARD treatment, overall and stratified by bDMARD
mode of action

Relative riska (HR) confidence interval (95% CI)

non bDMARD treated ref
bDMARD treated
all bDMARD treatment episodes b 0.75 0.56-1.02
TNF-inhibitors 0.67 0.47-0.96
rituximab 1.12 0.57-2.19
other bDMARD 0.90 0.49-1.66

a
) adjusted for sex, age at diagnosis and calendar year b) time scale: attained age, counting all treatment
episodes, first untreated episode

42
 DISCUSSION
This thesis aimed to explore the etiological and clinical impact of AITD in patients with RA
and therewith the impact of RA on the development of AITD. The main results from the
studies in this thesis is discussed based on the overall methodological considerations followed
by findings related to the specific studies.

6.1 METHODOLOGICAL CONSIDERATIONS

6.1.1 Study design and outcome measures

There is a hierarchy in analytic studies where the experimental study design RCT is
considered to be the golden standard. In the RCT the participants are randomized to e.g. a
new treatment or the standard treatment to minimize the risk of confounding. However, in
order to investigate many research questions an experimental study design may be unethical
or not feasible for other reasons. Analytical observational studies, such as the case-control
and cohort study design are two of the most commonly used study designs in epidemiological
research. There are some limitations with observational studies which need to be considered
when conducting and interpreting the result from this type of studies. The two study designs
used for the studies in this thesis, were the case-control and cohort study design.

Case-control study design

The case-control study design is by definition backward-directed and the cases are chosen
based on having the outcome of interest (e.g. RA). The controls are to be sampled from the
same population (source population) as the cases (161). There are several methods of
sampling the controls such as case-cohort and risk-set sampling. The exposure of interest is
compared between the cases and controls to investigate a potential association. A limitation
of the case-control study may be that the sampling of the controls, or cases, can create a bias
of the assessment of the exposure, for example due to recall bias. This is why the level of
evidence in a case-control study is generally described as weaker compared to the cohort
study design. One way of avoiding this type of bias is to use prospectively recorded data on
the exposure independently of the case/control status. There can also be difficulties in
assessing the temporality between the exposure and the outcome, to state which of the two
conditions that occurred first. The case-control design is suitable for studying rare diseases
and the association with several exposures can be investigated. One strength of this study
design is that it is cost-effective due to that the study population can be kept smaller
compared with other study designs.

Cohort study design

The cohort study design is forward-directed and the participants are sampled based on the
exposure status and then followed over a time period where the outcome potentially can
occur (162). The outcome among the exposed and unexposed individuals are then to be
compared. The cohort design can be either prospective or retrospective based on when the

43
data collection occurs in relation to the start of study. The retrospective cohort study is still
considered forward-directed but the outcome may have occurred when starting the study.
Also, in the retrospective cohort study design, the exposure may have been recorded
prospectively and independently of the outcome and thus allowing to minimize information
bias. This approach is used in the cohort studies included in this thesis. Advantages of the
cohort design is the possibility to include a large number of participants that can be followed
during a long period of time which may increase the power. In addition, this study design
allows to study diseases in relation to rare exposures, time-varying exposures and several
exposures and outcomes.

Applied study designs in the specific studies

In study II we applied a case-control design to assess the risk of AITD before RA diagnosis.
The definition of the exposure AITD and the outcome RA used in this study allowed to assess
the temporality between the two conditions. The case-control design enabled us to control for
potential confounders related to time before RA diagnosis. To include the combination of a
case-control and cohort design in the same study allowed to estimate the risk of AITD at
different points in time before and after RA diagnosis. The long follow-up time in study II
and IV enabled assessment of incident AITD in comparison with the general population in a
longer perspective and to investigate the effect of a longitudinal exposure to bDMARDs on
AITD risk. In study II-IV where a cohort design approach was used, the information on the
exposure (RA) was recorded independently of the outcome (AITD). By using the Total
Population Register large numbers of individuals could be identified from the same source
population as the RA patients and used as controls in study II and as the unexposed cohorts in
study II and IV.

Disease activity measures

The DAS 28 score is a widely used disease activity score in clinical practice, but there are
some limitations. DAS 28 does not include joint counts of the feet, why the disease activity
may potentially be underestimated. A previous study found a discrepancy between the
physicians’ assessment of the disease activity and the level of DAS 28 (163). The Patient
Global VAS and the ESR may also be high due to other medical reasons than RA. There can
be a discrepancy between the patients’ objective and subjective disease activity measures. In
study II both objective disease activity measures and patient reported disease activity
measures (PROMs) were used to capture all aspects of disease activity in RA patients.

6.1.2 Internal validity

High internal validity may be achieved in a research study by minimizing the systematic
errors (bias). The main categories of systematic errors are information bias/misclassification,
selection bias and confounding.

44
Information bias

Information bias or misclassification is a systematic error in measuring the exposure, or the

outcome. It can be either a non-differential or differential misclassification (164). The non-
differential misclassification is random and unrelated to the exposure or the outcome. This
may potentially dilute an association between an exposure and an outcome and bias towards
the null. The differential misclassification is a measurement error of the exposure or outcome
that may bias the relative risk either away or towards the null.

One way to minimize a potential misclassification of the RA diagnosis is to validate the

diagnosis against the medical charts as we did in study I. Based on algorithms used in
previous studies, we found a high validity of RA diagnosis both in the NPR and the SRQ.
The completeness of registration in the medical records by the rheumatologist significantly
impacted our validation results and some patients were excluded due to incomplete medical
records (n=9).

The definition of AITD was based on prescription of thyroxine from the PDR after excluding
non-autoimmune indications. The PDR has a very high coverage, close to 100%, of
dispensed drugs and in combination with using ICD-codes from the NPR considered a valid
data source to minimize misclassification of AITD in study II-IV. Although, it cannot be
excluded that some individuals could still have a non-autoimmune indication for thyroxine
substitution, a possible misclassification of AITD should not be dependent on RA status and
therefore not affecting the associations seen in our studies.

Surveillance bias is one type of differential misclassification where the likelihood of

detecting an exposure is higher in individuals having the outcome than for those not having
the outcome (or vice versa). This type of bias may occur for example in RA patients who may
potentially have more health care contacts than their controls, leading to an increased
surveillance. This could partly explain the increased risk of AITD close to RA diagnosis seen
in study II. Since we included RA patients with symptoms less than 12 months before
diagnosis the increased surveillance and diagnostic intensity should be limited to this period
of time. However, we detected an increased incidence of AITD up to five years earlier. We
performed additional adjustment for the number of physician visits prior to RA diagnosis to
control for a potential surveillance bias, which had little impact on the results. A surveillance
bias would presumably also remain sometime after RA diagnosis, however we saw a
decreasing incidence of AITD among RA patients during that time period both in study II and
IV.

Another type of differential misclassification is recall bias. This type of bias refers to when
cases remember the exposure of interest differently than their controls. By using
prospectively recorded exposure data with no relation to the outcome in the case-control part
of study II, recall bias was avoided.

45
Selection bias

Selection bias refers to a systematic error in how individuals are chosen or participates in a
research study (164). This type of bias can occur if there is a different association between
exposure and outcome among the study participant compared to the controls or exposure
status impacts the selection of cases and controls in a study. Selection bias may both
underestimate and overestimate an association.

In study II-IV we identified the RA cohorts from the SRQ. The likelihood of a patient being
included in a health care quality register such as the SRQ may possibly be affected by the
patient characteristics or health status. Patients with a high disease burden and shorter life
expectancy may be less likely to be included in a quality register. A patient that is lacking
basic computer skills or do not speak Swedish is unable to register PROMs in the SRQ, and
therefor potentially less likely to be included in the register. This could lead to a selection of
healthier RA cohorts and subsequently a possible underestimation of the risk of comorbidities
such AITD, compared to the general population in these studies. Since the RA patients in
study III were conditioned on having a registration of DAS 28 parameters in the SRQ at
several visits some level of selection bias cannot be excluded.

By using a population based-register, the Total Population Register, in study II and IV we

decreased the risk of a potential selection bias when identifying the general population
controls.

Confounding

A confounder is defined as a factor associated with both the exposure and outcome but not a
causal effect of the exposure (164). The confounder may bias the relative risk if it is unevenly
distributed between cases/controls or exposed/non-exposed. Some of the methods used to
control for confounding are randomization, matching, restriction, adjustment and
stratification. The information on potential confounders is often limited in register-based
studies compared to for instance studies using extensive questionnaires to collect information
on the specific individual. In the statistical models in study II-IV we stratified on age, sex and
RA serostatus. We adjusted for age, sex and also matching factors when applicable.
Additional adjustment for other comorbidities possibly associated with detecting AITD and
the number of health care visits were performed in study II. Moreover, we adjusted for
potential confounders linked to both RA disease activity and AITD; HAQ, the use of oral
steroids and smoking, in study III.

Confounding by indication

A specific type of confounding is confounding by indication (channeling bias), meaning that

patients receiving one type of treatment differ from patients receiving another type of
treatment. One clinical example could be that individuals with less comorbidities are more
likely to undergo a specific surgical intervention compared to individuals with a higher

46
disease burden, which may lead to an overestimation of the effect of the intervention. A
history of a previous malignancy and severe heart failure are contraindications for treatment
with some bDMARDs. Historically, rituximab has been recommended instead of treatment
with TNF-inhibitors in patients with a previous malignancy, according to treatment
guidelines for RA patients (165). This treatment approach reflects in the lower prevalence of
previous cancer among TNF-inhibitors treated compared in rituximab treated seen in
previous studies (166, 167). Since concomitant AITD is not a contraindication for treatment
with any of the bDMARDs, confounding by indication was not considered a major issue in
study IV.

Reverse causation

A situation where the outcome causes the exposure rather than vice versa is referred to as
reversed causation. Potentially the onset of RA symptoms may mimic the onset of AITD. A
way of controlling for reverse causation is to have a long follow-up time and to stratify the
risk in different strata of time as in study II and IV.

6.1.3 Precision
The accuracy of a study is dependent on the validity and the precision. The precision can be
referred to as the reproducibility of a study. High precision corresponds to a low grade of
random errors. By using a large study population, the risk of random errors decreases and the
precision increases. Both the p-value and the confidence interval quantify the precision. The
confidence interval was set to 95% in our studies. Using large cohorts in study II-IV enabled
a high precision. However, in some subset analyses the number of observations was limited
resulting in wider confidence intervals and lack of statistical power. This was seen for the
hyperthyroidism subgroup in study II and the younger age group in study III indicating less
robust results. The stronger association seen in the younger age group of RA patients
(compared to older) and AITD were consistent over study II-IV, though.

6.1.4 External validity

In study I we included RA patients with an outpatient visit in the NPR. Since the NPR is
population based and covers all specialist outpatient visits the risk of selecting a specific
phenotype of RA patients is reduced. Patients with RA followed only in primary care, not
covered by the NPR, were not included. However, this is most likely a very limited group of
patients. Since all RA patients in study I were followed at the Karolinska University Hospital
our findings may not be generalizable to all parts of Sweden. Nevertheless, our findings on
the validity is still of importance since the study participants were identified from the
country’s largest catchment area. By contrast, the results from study II-IV based on large
sample sizes, population-based registers and a clinically relevant definition of AITD are
considered to be generalizable to early RA patients also in an international setting.

47
6.2 FINDINGS

6.2.1 Validity of the RA diagnosis

Previous investigations of the validity of the RA diagnosis in the NPR have been limited to
hospitalizations or study-specific validations. Since most RA patients are followed and
treated by rheumatologists in outpatient care, it was of importance to validate the outpatient
part of the NPR which was performed in study I. We found that the validity or the positive
predictive value of the RA diagnosis was around 90% for prevalent RA. Among patients
defined as strict incident RA the corresponding figure was also around 90%. A large majority
fulfilled both the 1987 ACR and 2010 ACR/EULAR classification criteria for RA. Around
10% of the patients classified as RA fulfilled only one of the classification criteria. Most of
the patients fulfilling the 1987 classification criteria only, had a seronegative RA with a low
number of joints involved. The common denominator for the majority of patients only
fulfilling the 2010 ACR/EULAR criteria was high levels of ACPA or RF and elevated ESR.
When reviewing the medical charts, a more structured routine for documentation of joint
involvement was observed after the introduction of the 2010 ACR/EULAR criteria which
might have affected the results. Our study showed that of the patients incorrectly coded with
RA all but one had another defined inflammatory rheumatic diagnosis. To verify true RA
cases an additional assessment of the RA diagnosis was performed approximately two years
after the first for the incident patients. The requirement that the patients needed to retain their
RA diagnosis after two years of follow-up may possibly have underestimated the diagnostic
accuracy, though. Our results correspond to a previous study-specific validation from Sweden
where more than 90% of all RA or polyarthritis diagnoses were correct (144). Regarding
previous international studies, the validity of the RA diagnosis ranges between moderate to
high, 51-84% (145-147, 168) (expressed as the sensitivity or positive predictive value). A
discrepancy was seen between studies regarding the structure of the health care systems and
the coverage of the databases in the specific countries.

When conducting study I, a separate validation was performed restricted to the patients
registered in the SRQ (n=92 of the prevalent patients and n=84 of the incident patients). Of
these the validity or the of the RA diagnosis was 96% for the prevalent patients and 91% for
the incident patients (of whom 94% was also classified as incident). The results were not
published in the final manuscript since the main focus of the study was the validity in the
NPR.

A recently published study, based on the Danish clinical quality register (DANBIO) and the
Danish National Patient Registry (DNPR), found similar numbers on the validity as in our
study (96% for DANBIO and 79% for DNPR) (148).

48
6.2.2 AITD in relation to RA diagnosis
In study II our results showed that the increased prevalence of AITD in RA patients to a large
extent developed during the last few years prior to RA diagnosis. Following RA diagnosis,
the incidence of AITD gradually decreased. Our findings of a prevalence of AITD in RA
patients at diagnosis of approximately 10% are supported by previous studies (78, 101, 105).
We also found similar prevalence figures in study II and III. The timing of risk development
found in our study is also in line with previous findings (78). The pathogenic mechanisms
behind the co-existence of AITD and RA may be related to several factors. Shared
susceptibility genes have previously been suggested (55, 110, 169). An additional explanation
behind the temporal risk development of AITD in relation to RA diagnosis may be lead time
bias. An earlier detection of AITD in RA-patients due to more health care contacts and an
increased surveillance would result in an increased incidence and thereafter a reciprocal
decrease. As discussed above, a surveillance bias would not explain the increase and decrease
seen in the longer time spans in our study, though. Also, adjustment for the number of health
care contacts only slightly attenuated the risk estimates seen. The peak in risk of AITD close
to RA diagnosis could perhaps reflect a critical causal link between AITD and RA. This
could be a possible explanation for the somewhat higher incidence of AITD seen in
seropositive compared to seronegative subgroup of RA patients. The persistent decrease of
AITD up to 8 years following RA diagnosis compared to the general population may
potentially be due to a protective effect of antirheumatic therapies. Based on the results from
study II we suggested that the potential impact of antirheumatic treatments on the
development of AITD needed to be further investigated, which subsequently led to the
initiation of study IV. Since we aimed to investigate the temporal pattern between the two
conditions we used the date of RA diagnosis as a proxy for clinical onset of RA and the date
of first thyroxine prescription as a clinical for the onset of AITD. The pathogenic mechanisms
for both conditions, as described in the background, are likely to have started up to several
years earlier, though. When adjusting for comorbid conditions, we saw that the results were
slightly attenuated for type 1 diabetes, malaise and fatigue, osteoporosis and pregnancy-
related outcomes, while adjustment for depression, dementia and hyperparathyroidism
slightly enhanced the results. We lacked data on other potential confounders such as BMI and
smoking status which would of course have been valuable information. Adjustment for
smoking did not explain the increase prevalence of AITD seen in RA patients in a previous
study, though (78). In a clinical context our results imply that patients with AITD that suffer
from arthralgia should promptly be evaluated for a potential new onset RA. On the other
hand, the RA patient could be reassured that the risk of developing a concomitant AITD is
not increased.

49
6.2.3 The effect of AITD on RA disease activity and response to
antirheumatic treatment
Our result from study III suggest that AITD did impact RA disease activity at time of
diagnosis mainly through patient-reported disease activity measures. However, this difference
did not impact the overall chance of obtaining a good treatment response to the first line
antirheumatic treatment with methotrexate. Our results underline the importance of
systematic evaluation of all disease activity parameters and not just composite measures like
DAS 28. Contrary to the overall findings, we saw a different pattern of disease activity and a
negative effect on the treatment response among younger patients (< 45 years). This is in line
with previous findings indicating a stronger association between AITD and RA in younger
patients (94).

There is no clear unanimity in the results from previous studies investigating the association
between AITD and the RA disease activity. Small study populations often examined with a
cross-sectional design is a limitation in several studies. One previous study addressing this
research topic found an increased disease activity (based on DAS 28 and ESR) in prevalent
RA patients with thyroid dysfunction (118). A review article (170), has described a possible
association between hypothyroidism and high RA disease activity based on DAS 28 or some
of the individual DAS 28 components and suggests screening for AITD, since detection and
treatment of this condition could be beneficial for the RA disease activity. The results from
our study is partly in line with a Danish study (119), which have found a lower treatment
response (level and change of DAS 28 CRP) at 4 months from treatment initiation in RA
patients with thyroid disorders. But the individual components of DAS 28 were not
evaluated. Our results extend those findings by showing that the lower treatment response is
primarily explained by higher levels of patient-related disease activity parameters at baseline.
Our findings of potential lack of treatment response in younger patients, may reflect a
stronger effect of shared susceptibility genes, but needs further investigation. An individual
who has developed two autoimmune conditions already at a young age has perhaps a stronger
shared inheritance.

We did not have information on the levels of thyroid antibodies. However, as mentioned
previously, the prevalence of thyroid antibodies has been reported to be significantly higher
than clinical overt AITD in the population (54). We aimed at investigating the effect of
clinical overt AITD on RA disease activity and treatment response and therefor found the
applied definition of AITD adequate for our study.

Based on my own clinical experience, although the patient is treated to remission, the
persisting symptoms with chronic pain, fatigue and a reduced overall capacity may still be a
major problem. The patients often consider the problems related to these subjective
symptoms overarching the objective and specific joint symptoms.

To achieve an early disease control after the initiation of antirheumatic treatment has

50
previously been demonstrated to correlate with the long-term outcome (171). Consequently,
to identify predictors of treatment response is of great interest when aiming to optimize the
disease control. Since AITD is such a prevalent comorbidity and the symptoms may be
similar to RA symptoms it is of special interest to investigate its impact of the RA disease
activity and treatment response. It is reassuring for the treating physician that overall
concomitant AITD is not linked to worse objective RA disease activity measures. On the
other hand, AITD has an impact on the phenotypic presentation of RA through worsening of
subjective disease activity measures. These measures may also reflect the symptoms from
both conditions. It is known that the physicians’ assessment of disease activity may not
correlate with the patients’ assessment of disease activity and that a proportion of patients
score a higher level of disease activity than their physician (172, 173). PROMs are registered
by patients, without the impact of their physician, and can serve as additional markers in the
assessment of treatment response (174). The impact of our results in clinical practice would
be to stress the importance of not just paying attention to the overall composite disease
activity measure but to unpick the specific components.

6.2.4 The impact of antirheumatic treatments on the development of AITD

In study IV, we extended our previous findings from study II of incident AITD gradually
decreasing after RA diagnosis and showed that the decreased risk persisted into the second
decade of follow-up. The risk decrease was most pronounced in patients receiving bDMARD
treatment with the strongest protective association for the TNF inhibitors. These finding were
similar across sex and serostatus groups and number of bDMARD treatment episodes.

To our knowledge, no study has investigated the risk of incident AITD after the initiation of
treatment with bDMARDs or methotrexate in patients with early RA. However, our results
point somewhat in the same direction as previous small studies, using other definitions of
thyroid autoimmunity and prevalent disease rather than incident disease. Two of these studies
(123, 125) have reported decreased levels of thyroid autoantibodies after the treatment with
bDMARDs, TNF-inhibitors and rituximab in RA patients. On the other hand, another study
reports rather unchanged or increased levels of thyroid antibodies in RA patients following
treatment with TNF-inhibitors (126). We could not find comparable references in the
literature regarding a potential protective effect of methotrexate in combination with
bDMARD on AITD, as was seen in our study. However, a previous case-report demonstrated
reduced inflammation in thyroid related orbitopathy after the treatment with methotrexate
(129).

The complex regulation of T and B-cells and the related expression of numerous cytokines
are the targets for treatment with bDMARDs. The most frequently used bDMARD treatment
in RA patients are the TNF-inhibitors. Previous findings indicate that the levels of TNF-alpha

51
are elevated among patients with both hypothyroidism and hyperthyroidism. Also, treatment
of hyperthyroidism can lead to a normalization of the TNF-alpha levels (120).

Apart from a decreasing risk of AITD as an effect of treatment with bDMARDs, there may
be other explanations. As previous discussed, one is a reciprocal decrease as a result of an
increased diagnostic intensity. Since this study demonstrated a persisting decrease up to 14
years of follow up, this is most likely not the main explanation to our findings.

Previous findings advocating an effect of immunomodulatory treatment on dampening the

level of inflammation in AITD (121) are supportive of our results where the risk of AITD
was lower during bDMARD compared to non bDMARD treatment. This observation was
most pronounced for treatment with the TNF-inhibitors. A protective effect was also seen in
bDMARD and methotrexate combination therapy. This could reflect a synergetic preventive
effect of the two treatments related to the broad immunosuppressive effect of methotrexate
(175). The sample size in our study, approximately hundred times larger than previous
mentioned studies, allowed us to investigate the risk of incident AITD in subset of RA
patients and in different subgroups of bDMARD treatment.

52
 CONCLUSIONS
In this thesis we explored several clinical and etiological research questions related to the
bilateral impact between AITD and RA. Our findings provide knowledge that may be of
importance in clinical practice and the starting point for future research studies. The main
conclusions from our studies were the following:

• Based on the medical files from patients coded as RA in the NPR, the validity of the
RA diagnosis was high for both prevalent patients and for patients identified with
incident RA. The NPR and specific data algorithms to identify new-onset RA can be
used to identify RA populations to allow for high-quality epidemiological studies.

• The increased prevalence of AITD among RA patients, present in approximately 10%

at RA diagnosis, is a result of AITD developing a few years prior to the diagnosis of
RA. On the contrary, following RA diagnosis the risk of developing AITD gradually
decreased, raising the question if antirheumatic therapies may prevent or postpone
new-onset AITD.

• RA patients with concomitant AITD score worse on patient-reported disease activity

measures but not on objective disease activity measures at diagnosis.

• AITD did not impact the overall chance of achieving a EULAR good response to
treatment with methotrexate at the 3- and 6-month follow-up visits, with the
exception of younger RA patients who were less likely to obtain a good treatment
response.

• Following RA diagnosis, patients treated with bDMARDs, TNF inhibitors in

particular, have a decreased risk of developing AITD. These findings indicate that
bDMARDs could have a preventive effect of developing AITD and may open for
drug repurposing studies.

53
 POINTS OF PERSPECTIVE
Based on the findings from the studies in this thesis new questions have been raised that may
serve as suggestions for future studies:

Study I

In recent years, the SRQ has extended to include the majority of the inflammatory rheumatic
diseases and may serve as a unique source for identification of patients with these conditions.
To conduct further validation studies of these diseases, also outside the study specific setting,
may help to develop algorithms for identification of larger patient populations with high
validity to be included in epidemiological studies.

Study II

One way to better understand the mechanisms behind the risk of developing AITD in relation
to RA onset would be to abstract information on the specific phenotypes of AITD and RA.
Such a study, would in an ideal setting, include sampling both from sera, thyroid tissue and
synovial fluid. Based on these samples, assessment of autoantibodies and histopathological,
immunological and genetic investigations can be performed, which could lead to a better
understanding of the pathogenic mechanisms behind the co-existence between AITD and RA.

The findings in study II raises the question if the temporal pattern of risk seen for AITD is
restricted only to RA patients, or if this pattern also is to be found in other rheumatic
conditions. This question may be addressed by using a combined case-control design and
cohort design with the inclusion of other rheumatic diseases treated and evaluated with a
similar routine as for patients with RA.

Study III

In study III, a discrepancy was found between subjective- and objective disease activity
measures in RA patients with concomitant AITD. The use of a register as the SRQ allows for
the abstraction of these disease activity measures. To include PROMs also in future studies
investigating disease activity and treatment response in rheumatic conditions is of major
interest since it adds valuable additional information on the disease activity.

Study IV

In study IV, RA patients treated with bDMARDs, TNF inbitors in particular, were at
decreased risk of developing AITD. These findings are indicative of a potential protective
effect of bDMARDs on AITD. One way to further approach this research question would be
to conduct drug repurposing studies that seek to identify if hypothyroidism and
hyperthyroidism may be new targets for bDMARD treatment. Moreover, to investigate if RA
patients treated with bDMARDs are at decreased risk of developing other autoimmune
conditions would be of great interest.

55
56
 ACKNOWLEDGEMENTS
I’m very grateful to all people who have supported me during this research journey as a PhD-
student. I would especially like to thank:

Johan Askling, my main supervisor, for giving me the opportunity to become a PhD-student
and introducing me to the field of epidemiological research. Thank you for the support and
guidance throughout the years and for always responding immediately to my questions by
turning mail regardless of the time of day. Your scientific and linguistic skills are a true
inspiration! Thank you for creating such a dynamic and inspiring research environment at
KEP, I´m very glad to have been a part of that.

Bénédicte Delcoigne, my co-supervisor, for sharing your expertise in biostatistics. Merci for
your patience when we have discussed SAS coding and for your valuable input when
planning and conducting the studies in this thesis. I´m so glad that you joined as co-
supervisor along the way.

Saedis Saevarsdottir, my co-supervisor, for sharing your deep knowledge in rheumatology

and epidemiology, for help with improving my manuscripts and for the encouraging words
during these years. I’m grateful that you could continue as my co-supervisor although you
moved to Iceland.

Camilla Bengtsson, my former co-supervisor, for sharing knowledge in epidemiology and

for the valuable comments on texts and manuscript.

Karin Hellgren, my mentor, for your invaluable support, pep talks and for sharing your
research experience throughout the years. For your ”golden” advices and for coaching me in
how to balance clinical work, research work and the life outside work.

Members in the “Epi-inflammation” group at KEP past and present: Thomas Frisell,
Martin Neovius, Jonas Eriksson, Gustaf Bruze, Helga Westerlind, Daniela Di Giuseppe,
Hannah Bower, Elizabeth Arkema and Marie Holmqvist for creating such a great research
environment and for all the interesting discussions.

All the members of the Askling research group for all the great and inspiring research
work you perform.

My fellow-PhD students, past and present: Ängla, Hjalmar, Katarina, Andrei, Viktor W,
Matilda, Peter A, Renata, Anton, Marina and all others for support.

My fellow-students at the Research school for Clinicians, all co-workers and researcher
at KEP. Michael Fored for guidance during the application process for dissertation.

All my co-workers and colleagues at the Rheumatology unit for the support throughout
the years. Thank you for taking care of my patients when I have not been in the clinic.

57
Lars Klareskog, Ingrid Lundberg, Ingiäld Hafström, Elisabet Svenungsson, Iva Gunnarsson,
Johan Askling and Per-Johan Jakobsson for creating a unique research environment in the
field of rheumatology at Karolinska.

Former heads of the Rheumatology Unit, Johan Bratt, Cecilia Carlens and present Jon Lampa
and my former chiefs throughout my residency and as a specialist doctor, Ola Börjesson,
Louise Ekholm and present Christina Dorph for giving me the opportunity to combine
research with clinical work.

Inga-Lill Engvall, my former clinical supervisor, for always taking the time to share your
medical experience. Thank you for all the good advices and for always being so kind and
supportive.

Sofia Ernestam, my former clinical supervisor, for encouraging me to start with research in
the first place.

Marie Holmqvist, for creating such a stimulating research environment at the clinic. Thank
you for all the pep talks, good advices and for the great joy of having you as a room-mate.

The “Reuma KEPers”- Liselotte Tidblad, Viktor Molander, Karin Gunnarsson and Viking
Huss for your support in different ways. Nancy Vivar Pomiano, for the luck of teaming up
with you during demanding clinical “randningar”.

My friends and family

Bokklubbsgänget- för alla middagar, bokdiskussioner och kollegiala samtal.

Hille, för vänskap sen födseln och för alla fjällsemestrar med familjerna. Åsa, för alla
gemensamma minnen sen vi blev som ler och långhalm på det där konfirmationslägret.

Pinglorna- Moa, Hanna, Emma, Sofia, Ylva och Erika för vår vänskap sedan skoltiden. För
alla pingelmiddagar, pingelresor och allt pingelstöd. Ni sätter guldkant på mitt liv.

Uppsalagänget- Sofia, Linda och Rouzbeh. För alla upptåg, studiecirklar och fester
tillsammans. Åren på läkarprogrammet blev fantastiska tack vare er.

Henrik och Fia, Stefan och Lovisa- jag är så glad att vi tillhör samma storfamilj och för
barnens fina kusiner.

Svärmor Elisabet och svärfar Lennart, för ert otroliga stöd genom åren. Tack för att ni alltid
tar emot oss med öppna armar och för att ert hem är som barnens andra hem.

Min moster Marie, för alla minnen och kulturella upplevelser vi syskon har fått med dig
under uppväxten och fortfarande får. Marie, Dag och Carl för alla semesterminnen på
”Panget” i Fjällnäs och i Oskarshamn.

58
Karin och Petter, för att ni är mina syskon. Tack för alla fantastiska barndomsminnen och att
ni alltid finns där, i medgång och i motgång. För era fina familjer, Ylva, Magda och barnens
kusiner.

Mamma Lotta och pappa Bosse för det villkorslösa stödet och kärleken under uppväxten och
i vuxenlivet. För att ni alltid har trott på mig och för allt ni gör för min familj och barnen.
Tack doktor pappa för att du inspirerade mig till det fantastiska yrkesvalet att bli läkare. Tack
professor mamma för att du inspirerade mig till att börja forska, ditt stöd under
doktorandtiden och din ovärderliga feed-back på kappan.

Carl, Sigrid och Axel-mina älskade barn, ord kan inte uttrycka vad ni betyder för mig. Ni är
mitt liv. Jag är så lyckligt lottad att få vara er mamma!

Rickard, min älskade man, du är mitt livs lyckokast! Jag är så tacksam och lycklig över vår
familj och vårt liv tillsammans. Ditt stöd har varit outtröttligt under hela den här resan och det
hade aldrig gått utan dig.

59
 REFERENCES

1. Neovius M, Simard JF, Askling J. Nationwide prevalence of rheumatoid

arthritis and penetration of disease-modifying drugs in Sweden. Annals of the rheumatic
diseases. 2011;70(4):624-9.
2. Symmons DP. Epidemiology of rheumatoid arthritis: determinants of onset,
persistence and outcome. Best practice & research Clinical rheumatology. 2002;16(5):707-
22.
3. Alamanos Y, Voulgari PV, Drosos AA. Incidence and prevalence of
rheumatoid arthritis, based on the 1987 American College of Rheumatology criteria: A
systematic review. Semin Arthritis Rheu. 2006;36(3):182-8.
4. Eriksson JK, Neovius M, Ernestam S, Lindblad S, Simard JF, Askling J.
Incidence of rheumatoid arthritis in Sweden: a nationwide population-based assessment of
incidence, its determinants, and treatment penetration. Arthritis care & research.
2013;65(6):870-8.
5. Turesson C, Jacobsson LTH. Epidemiology of extra-articular manifestations in
rheumatoid arthritis. Scandinavian journal of rheumatology. 2004;33(2):65-72.
6. Smolen JS, Aletaha D, McInnes IB. Rheumatoid arthritis. Lancet (London,
England). 2016;388(10055):2023-38.
7. Kroot EJ, de Jong BA, van Leeuwen MA, Swinkels H, van den Hoogen FH,
van't Hof M, et al. The prognostic value of anti-cyclic citrullinated peptide antibody in
patients with recent-onset rheumatoid arthritis. Arthritis and rheumatism. 2000;43(8):1831-5.
8. Kitas GD, Gabriel SE. Cardiovascular disease in rheumatoid arthritis: state of
the art and future perspectives. Annals of the rheumatic diseases. 2011;70(1):8-14.
9. Michaud K, Wolfe F. Comorbidities in rheumatoid arthritis. Best Pract Res Cl
Rh. 2007;21(5):885-906.
10. Widdifield J, Bernatsky S, Paterson JM, Tomlinson G, Tu K, Kuriya B, et al.
Trends in Excess Mortality Among Patients With Rheumatoid Arthritis in Ontario, Canada.
Arthritis care & research. 2015;67(8):1047-53.
11. Silman AJ, Newman J, MacGregor AJ. Cigarette smoking increases the risk of
rheumatoid arthritis. Results from a nationwide study of disease-discordant twins. Arthritis
and rheumatism. 1996;39(5):732-5.
12. Millar K, Lloyd SM, McLean JS, Batty GD, Burns H, Cavanagh J, et al.
Personality, socio-economic status and inflammation: cross-sectional, population-based
study. PloS one. 2013;8(3):e58256.
13. Doran MF, Crowson CS, O'Fallon WM, Gabriel SE. The effect of oral
contraceptives and estrogen replacement therapy on the risk of rheumatoid arthritis: a
population based study. The Journal of rheumatology. 2004;31(2):207-13.
14. Orellana C, Wedrén S, Källberg H, Holmqvist M, Karlson EW, Alfredsson L,
et al. Parity and the risk of developing rheumatoid arthritis: results from the Swedish
Epidemiological Investigation of Rheumatoid Arthritis study. Annals of the rheumatic
diseases. 2014;73(4):752-5.

61
15. Stolt P, Källberg H, Lundberg I, Sjögren B, Klareskog L, Alfredsson L. Silica
exposure is associated with increased risk of developing rheumatoid arthritis: results from the
Swedish EIRA study. Annals of the rheumatic diseases. 2005;64(4):582-6.
16. Too CL, Muhamad NA, Ilar A, Padyukov L, Alfredsson L, Klareskog L, et al.
Occupational exposure to textile dust increases the risk of rheumatoid arthritis: results from a
Malaysian population-based case-control study. Annals of the rheumatic diseases.
2016;75(6):997-1002.
17. Scher JU, Littman DR, Abramson SB. Microbiome in Inflammatory Arthritis
and Human Rheumatic Diseases. Arthritis & rheumatology (Hoboken, NJ). 2016;68(1):35-
45.
18. Gregersen PK, Silver J, Winchester RJ. The shared epitope hypothesis. An
approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis.
Arthritis and rheumatism. 1987;30(11):1205-13.
19. Symmons D, Harrison B. Early inflammatory polyarthritis: results from the
Norfolk Arthritis Register with a review of the literature. I. Risk factors for the development
of inflammatory polyarthritis and rheumatoid arthritis. Rheumatology. 2000;39(8):835-43.
20. Carlens C, Hergens MP, Grunewald J, Ekbom A, Eklund A, Höglund CO, et al.
Smoking, use of moist snuff, and risk of chronic inflammatory diseases. American journal of
respiratory and critical care medicine. 2010;181(11):1217-22.
21. Stolt P, Bengtsson C, Nordmark B, Lindblad S, Lundberg I, Klareskog L, et al.
Quantification of the influence of cigarette smoking on rheumatoid arthritis: results from a
population based case-control study, using incident cases. Annals of the rheumatic diseases.
2003;62(9):835-41.
22. Yahya A, Bengtsson C, Lai TC, Larsson PT, Mustafa AN, Abdullah NA, et al.
Smoking is associated with an increased risk of developing ACPA-positive but not ACPA-
negative rheumatoid arthritis in Asian populations: evidence from the Malaysian MyEIRA
case-control study. Modern rheumatology. 2012;22(4):524-31.
23. Padyukov L, Silva C, Stolt P, Alfredsson L, Klareskog L. A gene-environment
interaction between smoking and shared epitope genes in HLA-DR provides a high risk of
seropositive rheumatoid arthritis. Arthritis and rheumatism. 2004;50(10):3085-92.
24. Mattey DL, Dawes PT, Clarke S, Fisher J, Brownfield A, Thomson W, et al.
Relationship among the HLA-DRB1 shared epitope, smoking, and rheumatoid factor
production in rheumatoid arthritis. Arthritis and rheumatism. 2002;47(4):403-7.
25. Klein K, Gay S. Epigenetics in rheumatoid arthritis. Current opinion in
rheumatology. 2015;27(1):76-82.
26. Källberg H, Jacobsen S, Bengtsson C, Pedersen M, Padyukov L, Garred P, et
al. Alcohol consumption is associated with decreased risk of rheumatoid arthritis: results
from two Scandinavian case-control studies. Annals of the rheumatic diseases.
2009;68(2):222-7.
27. Rosell M, Wesley AM, Rydin K, Klareskog L, Alfredsson L. Dietary fish and
fish oil and the risk of rheumatoid arthritis. Epidemiology (Cambridge, Mass).
2009;20(6):896-901.
28. Jiang X, Frisell T, Askling J, Karlson EW, Klareskog L, Alfredsson L, et al. To
what extent is the familial risk of rheumatoid arthritis explained by established rheumatoid
arthritis risk factors? Arthritis & rheumatology (Hoboken, NJ). 2015;67(2):352-62.

62
29. Frisell T, Hellgren K, Alfredsson L, Raychaudhuri S, Klareskog L, Askling J.
Familial aggregation of arthritis-related diseases in seropositive and seronegative rheumatoid
arthritis: a register-based case-control study in Sweden. Annals of the rheumatic diseases.
2016;75(1):183-9.
30. Frisell T, Holmqvist M, Källberg H, Klareskog L, Alfredsson L, Askling J.
Familial risks and heritability of rheumatoid arthritis: role of rheumatoid factor/anti-
citrullinated protein antibody status, number and type of affected relatives, sex, and age.
Arthritis and rheumatism. 2013;65(11):2773-82.
31. Nielen MM, van Schaardenburg D, Reesink HW, van de Stadt RJ, van der
Horst-Bruinsma IE, de Koning MH, et al. Specific autoantibodies precede the symptoms of
rheumatoid arthritis: a study of serial measurements in blood donors. Arthritis and
rheumatism. 2004;50(2):380-6.
32. Nielsen SF, Bojesen SE, Schnohr P, Nordestgaard BG. Elevated rheumatoid
factor and long term risk of rheumatoid arthritis: a prospective cohort study. BMJ (Clinical
research ed). 2012;345:e5244.
33. Rantapää-Dahlqvist S, de Jong BA, Berglin E, Hallmans G, Wadell G,
Stenlund H, et al. Antibodies against cyclic citrullinated peptide and IgA rheumatoid factor
predict the development of rheumatoid arthritis. Arthritis and rheumatism. 2003;48(10):2741-
9.
34. McInnes IB, Schett G. The pathogenesis of rheumatoid arthritis. The New
England journal of medicine. 2011;365(23):2205-19.
35. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al.
The American Rheumatism Association 1987 revised criteria for the classification of
rheumatoid arthritis. Arthritis and rheumatism. 1988;31(3):315-24.
36. Aletaha D, Neogi T, Silman AJ, Funovits J, Felson DT, Bingham CO, 3rd, et al.
2010 rheumatoid arthritis classification criteria: an American College of
Rheumatology/European League Against Rheumatism collaborative initiative. Annals of the
rheumatic diseases. 2010;69(9):1580-8.
37. van der Heijde D, van der Helm-van Mil AH, Aletaha D, Bingham CO,
Burmester GR, Dougados M, et al. EULAR definition of erosive disease in light of the 2010
ACR/EULAR rheumatoid arthritis classification criteria. Annals of the rheumatic diseases.
2013;72(4):479-81.
38. Smolen JS, Aletaha D, Koeller M, Weisman MH, Emery P. New therapies for
treatment of rheumatoid arthritis. Lancet (London, England). 2007;370(9602):1861-74.
39. 2021 SRF. Riktlinjer för läkmedelsbehandling vid reumatoid artrit [Available
from: https://svenskreumatologi.se/srfs-riktlinjer/Downloads/riktlinjer-for-
lakemedelsbehandling-vid-ra-2021%20(1).pdf].
40. Smolen JS, Landewe RBM, Bijlsma JWJ, Burmester GR, Dougados M,
Kerschbaumer A, et al. EULAR recommendations for the management of rheumatoid
arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2019 update.
Annals of the rheumatic diseases. 2020;79(6):685-99.
41. Smolen JS, Landewé R, Breedveld FC, Dougados M, Emery P, Gaujoux-Viala
C, et al. EULAR recommendations for the management of rheumatoid arthritis with synthetic
and biological disease-modifying antirheumatic drugs. Annals of the rheumatic diseases.
2010;69(6):964-75.

63
42. Nam JL, Ramiro S, Gaujoux-Viala C, Takase K, Leon-Garcia M, Emery P, et
al. Efficacy of biological disease-modifying antirheumatic drugs: a systematic literature
review informing the 2013 update of the EULAR recommendations for the management of
rheumatoid arthritis. Annals of the rheumatic diseases. 2014;73(3):516-28.
43. van Riel PL. The development of the disease activity score (DAS) and the
disease activity score using 28 joint counts (DAS28). Clinical and experimental
rheumatology. 2014;32(5 Suppl 85):S-65-74.
44. Prevoo ML, van 't Hof MA, Kuper HH, van Leeuwen MA, van de Putte LB,
van Riel PL. Modified disease activity scores that include twenty-eight-joint counts.
Development and validation in a prospective longitudinal study of patients with rheumatoid
arthritis. Arthritis and rheumatism. 1995;38(1):44-8.
45. Inoue E, Yamanaka H, Hara M, Tomatsu T, Kamatani N. Comparison of
Disease Activity Score (DAS)28- erythrocyte sedimentation rate and DAS28- C-reactive
protein threshold values. Annals of the rheumatic diseases. 2007;66(3):407-9.
46. Wells G, Becker JC, Teng J, Dougados M, Schiff M, Smolen J, et al. Validation
of the 28-joint Disease Activity Score (DAS28) and European League Against Rheumatism
response criteria based on C-reactive protein against disease progression in patients with
rheumatoid arthritis, and comparison with the DAS28 based on erythrocyte sedimentation
rate. Annals of the rheumatic diseases. 2009;68(6):954-60.
47. Fries JF, Spitz P, Kraines RG, Holman HR. Measurement of patient outcome in
arthritis. Arthritis and rheumatism. 1980;23(2):137-45.
48. Fries JF. The hierarchy of quality-of-life assessment, the Health Assessment
Questionnaire (HAQ), and issues mandating development of a toxicity index. Controlled
clinical trials. 1991;12(4 Suppl):106s-17s.
49. Zoeller RT, Tan SW, Tyl RW. General background on the hypothalamic-
pituitary-thyroid (HPT) axis. Critical reviews in toxicology. 2007;37(1-2):11-53.
50. Weetman AP. Autoimmune thyroid disease. Autoimmunity. 2004;37(4):337-
40.
51. Werner S. Endokrinologi 2015 (3:e upplagan). 110-42 p.
52. Weetman A. Autoimmune thyroid disease. Endocrine. 2020;68(2):258-60.
53. Weetman A. The Autoimmune Diseases, Thyroid disease: 4th ed. Cambridge,
MA: AcademicPress; 2006.
54. Vanderpump MP. The epidemiology of thyroid disease. British medical
bulletin. 2011;99:39-51.
55. Effraimidis G, Wiersinga WM. Mechanisms in endocrinology: autoimmune
thyroid disease: old and new players. Eur J Endocrinol. 2014;170(6):R241-52.
56. Lazurova I, Jochmanova I, Benhatchi K, Sotak S. Autoimmune thyroid disease
and rheumatoid arthritis: relationship and the role of genetics. Immunologic research.
2014;60(2-3):193-200.
57. Canaris GJ, Manowitz NR, Mayor G, Ridgway EC. The Colorado thyroid
disease prevalence study. Arch Intern Med. 2000;160(4):526-34.
58. Chaker L, Bianco AC, Jonklaas J, Peeters RP. Hypothyroidism. Lancet
(London, England). 2017;390(10101):1550-62.

64
59. Carlé A, Laurberg P, Pedersen IB, Knudsen N, Perrild H, Ovesen L, et al.
Epidemiology of subtypes of hypothyroidism in Denmark. Eur J Endocrinol. 2006;154(1):21-
8.
60. Asvold BO, Vatten LJ, Bjoro T. Changes in the prevalence of hypothyroidism:
the HUNT Study in Norway. Eur J Endocrinol. 2013;169(5):613-20.
61. Garmendia Madariaga A, Santos Palacios S, Guillen-Grima F, Galofre JC. The
incidence and prevalence of thyroid dysfunction in Europe: a meta-analysis. The Journal of
clinical endocrinology and metabolism. 2014;99(3):923-31.
62. Tunbridge WM, Evered DC, Hall R, Appleton D, Brewis M, Clark F, et al. The
spectrum of thyroid disease in a community: the Whickham survey. Clinical endocrinology.
1977;7(6):481-93.
63. Knudsen N, Bulow I, Jorgensen T, Laurberg P, Ovesen L, Perrild H.
Comparative study of thyroid function and types of thyroid dysfunction in two areas in
Denmark with slightly different iodine status. Eur J Endocrinol. 2000;143(4):485-91.
64. Knudsen N, Jorgensen T, Rasmussen S, Christiansen E, Perrild H. The
prevalence of thyroid dysfunction in a population with borderline iodine deficiency. Clinical
endocrinology. 1999;51(3):361-7.
65. Davies TF, Andersen S, Latif R, Nagayama Y, Barbesino G, Brito M, et al.
Graves' disease. Nat Rev Dis Primers. 2020;6(1):52.
66. Chiovato L, Magri F, Carle A. Hypothyroidism in Context: Where We've Been
and Where We're Going. Advances in therapy. 2019;36:47-58.
67. Aoki Y, Belin RM, Clickner R, Jeffries R, Phillips L, Mahaffey KR. Serum
TSH and total T4 in the United States population and their association with participant
characteristics: National Health and Nutrition Examination Survey (NHANES 1999-2002).
Thyroid : official journal of the American Thyroid Association. 2007;17(12):1211-23.
68. Gaitonde DY, Rowley KD, Sweeney LB. Hypothyroidism: an update. Am Fam
Physician. 2012;86(3):244-51.
69. Delamere JP, Scott DL, Felix-Davies DD. Thyroid dysfunction and rheumatic
diseases. Journal of the Royal Society of Medicine. 1982;75(2):102-6.
70. Jervis W, Shah N, Mongolu SK, Sathyapalan T. Severe proximal myopathy
secondary to Hashimoto's thyroiditis. BMJ case reports. 2019;12(7).
71. Bland JH, Frymoyer JW, Newberg AH, Revers R, Norman RJ. Rheumatic
syndromes in endocrine disease. Semin Arthritis Rheum. 1979;9(1):23-65.
72. Palmér M, Tovi J. https://www.viss.nu/kunskapsstod/vardprogram/hypotyreos
2021 [
73. Wiersinga WM, Duntas L, Fadeyev V, Nygaard B, Vanderpump MP. 2012
ETA Guidelines: The Use of L-T4 + L-T3 in the Treatment of Hypothyroidism. Eur Thyroid
J. 2012;1(2):55-71.
74. Rydén M, Palmér M, Tovi J.
https://www.viss.nu/kunskapsstod/vardprogram/hypertyreos. 2019.
75. Burch HB. Drug Effects on the Thyroid. The New England journal of medicine.
2019;381(8):749-61.

65
76. Wilson SA, Stem LA, Bruehlman RD. Hypothyroidism: Diagnosis and
Treatment. Am Fam Physician. 2021;103(10):605-13.
77. Calissendorff J, Ljung R. Thyroxin substitution and open-angle glaucoma: a
nationwide register-based study on filled prescriptions. The British journal of ophthalmology.
2011;95(4):592-4.
78. Bengtsson C, Padyukov L, Kallberg H, Saevarsdottir S. Thyroxin substitution
and the risk of developing rheumatoid arthritis; results from the Swedish population-based
EIRA study. Annals of the rheumatic diseases. 2014;73(6):1096-100.
79. Dayan CM, Daniels GH. Chronic autoimmune thyroiditis. The New England
journal of medicine. 1996;335(2):99-107.
80. Bourji K, Gatto M, Cozzi F, Doria A, Punzi L. Rheumatic and autoimmune
thyroid disorders: a causal or casual relationship? Autoimmunity reviews. 2015;14(1):57-63.
81. Alfaris N, Curiel R, Tabbara S, Irwig MS. Autoimmune thyroid disease and
Sjögren syndrome. Journal of clinical rheumatology : practical reports on rheumatic &
musculoskeletal diseases. 2010;16(3):146-7.
82. Watad A, Cohen AD, Comaneshter D, Tekes-Manova D, Amital H.
Hyperthyroidism association with SLE, lessons from real-life data--A case-control study.
Autoimmunity. 2016;49(1):17-20.
83. Watad A, Mahroum N, Whitby A, Gertel S, Comaneshter D, Cohen AD, et al.
Hypothyroidism among SLE patients: Case-control study. Autoimmunity reviews.
2016;15(5):484-6.
84. Fallahi P, Ferrari SM, Ruffilli I, Elia G, Biricotti M, Vita R, et al. The
association of other autoimmune diseases in patients with autoimmune thyroiditis: Review of
the literature and report of a large series of patients. Autoimmunity reviews.
2016;15(12):1125-8.
85. Fallahi P, Ferrari SM, Ruffilli I, Elia G, Miccoli M, Sedie AD, et al. Increased
incidence of autoimmune thyroid disorders in patients with psoriatic arthritis: a longitudinal
follow-up study. Immunologic research. 2017;65(3):681-6.
86. Bowness P, Shotliff K, Middlemiss A, Myles AB. Prevalence of
hypothyroidism in patients with polymyalgia rheumatica and giant cell arteritis. British
journal of rheumatology. 1991;30(5):349-51.
87. Yavne Y, Tiosano S, Watad A, Comaneshter D, Shoenfeld Y, Cohen AD, et al.
Association between giant cell arteritis and thyroid dysfunction in a "real life" population.
Endocrine. 2017;57(2):241-6.
88. Raterman HG, van Halm VP, Voskuyl AE, Simsek S, Dijkmans BA,
Nurmohamed MT. Rheumatoid arthritis is associated with a high prevalence of
hypothyroidism that amplifies its cardiovascular risk. Annals of the rheumatic diseases.
2008;67(2):229-32.
89. Chan AT, Al-Saffar Z, Bucknall RC. Thyroid disease in systemic lupus
erythematosus and rheumatoid arthritis. Rheumatology (Oxford, England). 2001;40(3):353-4.
90. Shiroky JB, Cohen M, Ballachey ML, Neville C. Thyroid dysfunction in
rheumatoid arthritis: a controlled prospective survey. Annals of the rheumatic diseases.
1993;52(6):454-6.

66
91. Bahemuka M, Hodkinson HM. Screening for hypothyroidism in elderly
inpatients. British medical journal. 1975;2(5971):601-3.
92. Przygodzka M, Filipowicz-Sosnowska A. Prevalence of thyroid diseases and
antithyroid antibodies in women with rheumatoid arthritis. Polskie Archiwum Medycyny
Wewnetrznej. 2009;119(1-2):39-43.
93. Somers EC, Thomas SL, Smeeth L, Hall AJ. Are individuals with an
autoimmune disease at higher risk of a second autoimmune disorder? Am J Epidemiol.
2009;169(6):749-55.
94. Mahagna H, Caplan A, Watad A, Bragazzi NL, Sharif K, Tiosano S, et al.
Rheumatoid arthritis and thyroid dysfunction: A cross-sectional study and a review of the
literature. Best practice & research Clinical rheumatology. 2018;32(5):683-91.
95. Masi AT, Hartmann WH, Hahn BH, Abbey H, Shulman LE. Hashimoto's
disease. A clinicopathological study with matched controls. Lack of significant associations
with other "autoimmune" disorders. Lancet (London, England). 1965;1(7377):123-6.
96. Mulhern LM, Masi AT, Shulman LE. Hashimoto's disease. A search for
associated disorders in 170 clinically detected cases. Lancet (London, England).
1966;2(7462):508-11.
97. McCoy SS, Crowson CS, Gabriel SE, Matteson EL. Hypothyroidism as a risk
factor for development of cardiovascular disease in patients with rheumatoid arthritis. The
Journal of rheumatology. 2012;39(5):954-8.
98. Becker KL, Ferguson RH, Mc CW. The connective-tissue diseases and
symptoms associated with Hashimoto's thyroiditis. The New England journal of medicine.
1963;268:277-80.
99. Fallahi P, Elia G, Ragusa F, Ruffilli I, Camastra S, Giusti C, et al. The
aggregation between AITD with rheumatologic, or dermatologic, autoimmune diseases. Best
practice & research Clinical endocrinology & metabolism. 2019:101372.
100. Boelaert K, Newby PR, Simmonds MJ, Holder RL, Carr-Smith JD, Heward
JM, et al. Prevalence and relative risk of other autoimmune diseases in subjects with
autoimmune thyroid disease. The American journal of medicine. 2010;123(2):183.e1-9.
101. Pongratz R, Buchinger W, Semlitsch G, Meister E, Nadler K, Rainer F.
[Increased occurrence of autoimmune thyroiditis in patients with chronic rheumatoid
arthritis]. Acta medica Austriaca. 2000;27(2):58-60.
102. Lazurova I, Benhatchi K, Rovensky J, Kozakova D, Wagnerova H, Tajtakova
M, et al. Autoimmune thyroid disease and autoimmune rheumatic disorders: a two-sided
analysis. Annals of the New York Academy of Sciences. 2009;1173:211-6.
103. Innocencio RM, Romaldini JH, Ward LS. Thyroid autoantibodies in
autoimmune diseases. Medicina. 2004;64(3):227-30.
104. Magnus JH, Birketvedt T, Haga HJ. A prospective evaluation of antithyroid
antibody prevalence in 100 patients with rheumatoid arthritis. Scandinavian journal of
rheumatology. 1995;24(3):180-2.
105. Cardenas Roldan J, Amaya-Amaya J, Castellanos-de la Hoz J, Giraldo-Villamil
J, Montoya-Ortiz G, Cruz-Tapias P, et al. Autoimmune thyroid disease in rheumatoid
arthritis: a global perspective. Arthritis. 2012;2012:864907.

67
106. Atzeni F, Doria A, Ghirardello A, Turiel M, Batticciotto A, Carrabba M, et al.
Anti-thyroid antibodies and thyroid dysfunction in rheumatoid arthritis: prevalence and
clinical value. Autoimmunity. 2008;41(1):111-5.
107. Pan XF, Gu JQ, Shan ZY. Increased risk of thyroid autoimmunity in
rheumatoid arthritis: a systematic review and meta-analysis. Endocrine. 2015;50(1):79-86.
108. Liao KP, Kurreeman F, Li G, Duclos G, Murphy S, Guzman R, et al.
Associations of autoantibodies, autoimmune risk alleles, and clinical diagnoses from the
electronic medical records in rheumatoid arthritis cases and non-rheumatoid arthritis controls.
Arthritis and rheumatism. 2013;65(3):571-81.
109. Vanderpump MP, Tunbridge WM, French JM, Appleton D, Bates D, Clark F,
et al. The incidence of thyroid disorders in the community: a twenty-year follow-up of the
Whickham Survey. Clinical endocrinology. 1995;43(1):55-68.
110. Saevarsdottir S, Olafsdottir TA, Ivarsdottir EV, Halldorsson GH, Gunnarsdottir
K, Sigurdsson A, et al. FLT3 stop mutation increases FLT3 ligand level and risk of
autoimmune thyroid disease. Nature. 2020;584(7822):619-23.
111. Kerola AM, Nieminen TV, Kauppi MJ, Kautiainen H, Puolakka K, Virta LJ, et
al. Increased risk of levothyroxine-treated hypothyroidism preceding the diagnosis of
rheumatoid arthritis: a nationwide registry study. Clinical and experimental rheumatology.
2014;32(4):455-9.
112. Crepaldi G, Scirè CA, Carrara G, Sakellariou G, Caporali R, Hmamouchi I, et
al. Cardiovascular Comorbidities Relate More than Others with Disease Activity in
Rheumatoid Arthritis. PloS one. 2016;11(1):e0146991.
113. Radner H, Smolen JS, Aletaha D. Impact of comorbidity on physical function
in patients with rheumatoid arthritis. Annals of the rheumatic diseases. 2010;69(3):536-41.
114. Ranganath VK, Maranian P, Elashoff DA, Woodworth T, Khanna D, Hahn T,
et al. Comorbidities are associated with poorer outcomes in community patients with
rheumatoid arthritis. Rheumatology (Oxford, England). 2013;52(10):1809-17.
115. van den Hoek J, Roorda LD, Boshuizen HC, van Hees J, Rupp I, Tijhuis GJ, et
al. Long-term physical functioning and its association with somatic comorbidity and
comorbid depression in patients with established rheumatoid arthritis: a longitudinal study.
Arthritis care & research. 2013;65(7):1157-65.
116. De Vito P, Incerpi S, Pedersen JZ, Luly P, Davis FB, Davis PJ. Thyroid
hormones as modulators of immune activities at the cellular level. Thyroid : official journal
of the American Thyroid Association. 2011;21(8):879-90.
117. Jara EL, Munoz-Durango N, Llanos C, Fardella C, Gonzalez PA, Bueno SM, et
al. Modulating the function of the immune system by thyroid hormones and thyrotropin.
Immunology letters. 2017;184:76-83.
118. Joshi P, Agarwal A, Vyas S, Kumar R. Prevalence of hypothyroidism in
rheumatoid arthritis and its correlation with disease activity. Tropical doctor. 2017;47(1):6-
10.
119. Emamifar A, Hangaard J, Jensen Hansen IM. Thyroid disorders in patients with
newly diagnosed rheumatoid arthritis is associated with poor initial treatment response
evaluated by disease activity score in 28 joints-C-reactive protein (DAS28-CRP): An
observational cohort study. Medicine. 2017;96(43):e8357.

68
120. Díez JJ, Hernanz A, Medina S, Bayón C, Iglesias P. Serum concentrations of
tumour necrosis factor-alpha (TNF-alpha) and soluble TNF-alpha receptor p55 in patients
with hypothyroidism and hyperthyroidism before and after normalization of thyroid function.
Clinical endocrinology. 2002;57(4):515-21.
121. Chen K, Wei Y, Sharp GC, Braley-Mullen H. Decreasing TNF-alpha results in
less fibrosis and earlier resolution of granulomatous experimental autoimmune thyroiditis.
Journal of leukocyte biology. 2007;81(1):306-14.
122. Caturegli P, Kimura H. A nonclassical model of autoimmune hypothyroidism.
Thyroid : official journal of the American Thyroid Association. 2010;20(1):3-5.
123. Raterman HG, Jamnitski A, Lems WF, Voskuyl AE, Dijkmans BA, Bos WH,
et al. Improvement of thyroid function in hypothyroid patients with rheumatoid arthritis after
6 months of adalimumab treatment: a pilot study. The Journal of rheumatology.
2011;38(2):247-51.
124. Bliddal S, Borresen SW, Feldt-Rasmussen U. Thyroid Autoimmunity and
Function after Treatment with Biological Antirheumatic Agents in Rheumatoid Arthritis.
Frontiers in endocrinology. 2017;8:179.
125. Kaklamanos M, Thomas D, Pikazis D, Kaltsas G. Thyroid-specific changes
following treatment with biological therapies in patients with rheumatic diseases. Endocrine.
2015;50(1):146-53.
126. Caramaschi P, Biasi D, Colombatti M, Pieropan S, Martinelli N, Carletto A, et
al. Anti-TNFalpha therapy in rheumatoid arthritis and autoimmunity. Rheumatol Int.
2006;26(3):209-14.
127. Salvi M, Vannucchi G, Beck-Peccoz P. Potential utility of rituximab for
Graves' orbitopathy. The Journal of clinical endocrinology and metabolism.
2013;98(11):4291-9.
128. Salvi M, Vannucchi G, Curro N, Campi I, Covelli D, Dazzi D, et al. Efficacy of
B-cell targeted therapy with rituximab in patients with active moderate to severe Graves'
orbitopathy: a randomized controlled study. The Journal of clinical endocrinology and
metabolism. 2015;100(2):422-31.
129. Rubinov A, Zommer H, Aghazadeh H, Weis E. Role of methotrexate in
thyroid-related orbitopathy. Can J Ophthalmol. 2018;53(1):34-8.
130. Hopkins UJ.
https://pathology.jhu.edu/autoimmune/classification/#:~:text=Autoimmune%20diseases%20c
an%20be%20classified%20according%20to%20several,autoimmune%20diseases%20are%2
0distinguished%20into%20systemic%20or%20organ-specific. 2021 [cited 2021 28 th of
December].
131. Kahaly GJ. Management of Graves Thyroidal and Extrathyroidal Disease: An
Update. The Journal of clinical endocrinology and metabolism. 2020;105(12).
132. Askling J, Fored CM, Geborek P, Jacobsson LT, van Vollenhoven R, Feltelius
N, et al. Swedish registers to examine drug safety and clinical issues in RA. Annals of the
rheumatic diseases. 2006;65(6):707-12.
133. Neovius M, Simard J, Sundstrom A, Jacobsson L, Geborek P, Saxne T, et al.
Generalisability of clinical registers used for drug safety and comparative effectiveness
research: coverage of the Swedish Biologics Register. Annals of the rheumatic diseases.
2011;70(3):516-9.

69
134. Neovius M, Sundstrom A, Simard J, Wettermark B, Cars T, Feltelius N, et al.
Small-area variations in sales of TNF inhibitors in Sweden between 2000 and 2009.
Scandinavian journal of rheumatology. 2011;40(1):8-15.
135. Askling J, Baecklund E, Granath F, Geborek P, Fored M, Backlin C, et al. Anti-
tumour necrosis factor therapy in rheumatoid arthritis and risk of malignant lymphomas:
relative risks and time trends in the Swedish Biologics Register. Annals of the rheumatic
diseases. 2009;68(5):648-53.
136. Askling J, Dixon W. The safety of anti-tumour necrosis factor therapy in
rheumatoid arthritis. Current opinion in rheumatology. 2008;20(2):138-44.
137. Askling J, Fored CM, Brandt L, Baecklund E, Bertilsson L, Feltelius N, et al.
Time-dependent increase in risk of hospitalisation with infection among Swedish RA patients
treated with TNF antagonists. Annals of the rheumatic diseases. 2007;66(10):1339-44.
138. Askling J, van Vollenhoven RF, Granath F, Raaschou P, Fored CM, Baecklund
E, et al. Cancer risk in patients with rheumatoid arthritis treated with anti-tumor necrosis
factor alpha therapies: does the risk change with the time since start of treatment? Arthritis
and rheumatism. 2009;60(11):3180-9.
139. Raaschou P, Simard JF, Holmqvist M, Askling J. Rheumatoid arthritis, anti-
tumour necrosis factor therapy, and risk of malignant melanoma: nationwide population
based prospective cohort study from Sweden. BMJ (Clinical research ed). 2013;346:f1939.
140. Neovius M, Simard JF, Klareskog L, Askling J. Sick leave and disability
pension before and after initiation of antirheumatic therapies in clinical practice. Annals of
the rheumatic diseases. 2011;70(8):1407-14.
141. Eriksson JK, Askling J, Arkema EV. The Swedish Rheumatology Quality
Register: optimisation of rheumatic disease assessments using register-enriched data. Clinical
and experimental rheumatology. 2014;32(5 Suppl 85):S-147-9.
142. Frisell T, Holmqvist M, Kallberg H, Klareskog L, Alfredsson L, Askling J.
Familial risks and heritability of rheumatoid arthritis: role of rheumatoid factor/anti-
citrullinated protein antibody status, number and type of affected relatives, sex, and age.
Arthritis and rheumatism. 2013;65(11):2773-82.
143. Baecklund E, Iliadou A, Askling J, Ekbom A, Backlin C, Granath F, et al.
Association of chronic inflammation, not its treatment, with increased lymphoma risk in
rheumatoid arthritis. Arthritis and rheumatism. 2006;54(3):692-701.
144. Knight A, Sandin S, Askling J. Increased risk of autoimmune disease in
families with Wegener's granulomatosis. The Journal of rheumatology. 2010;37(12):2553-8.
145. Thomas SL, Edwards CJ, Smeeth L, Cooper C, Hall AJ. How accurate are
diagnoses for rheumatoid arthritis and juvenile idiopathic arthritis in the general practice
research database? Arthritis and rheumatism. 2008;59(9):1314-21.
146. Widdifield J, Bernatsky S, Paterson JM, Tu K, Ng R, Thorne JC, et al.
Accuracy of Canadian health administrative databases in identifying patients with rheumatoid
arthritis: a validation study using the medical records of rheumatologists. Arthritis care &
research. 2013;65(10):1582-91.
147. Widdifield J, Bombardier C, Bernatsky S, Paterson JM, Green D, Young J, et
al. An administrative data validation study of the accuracy of algorithms for identifying
rheumatoid arthritis: the influence of the reference standard on algorithm performance. BMC
Musculoskelet Disord. 2014;15:216.

70
148. Ibfelt EH, Sørensen J, Jensen DV, Dreyer L, Schiøttz-Christensen B, Thygesen
PH, et al. Validity and completeness of rheumatoid arthritis diagnoses in the nationwide
DANBIO clinical register and the Danish National Patient Registry. Clinical epidemiology.
2017;9:627-32.
149. Holmqvist ME, Neovius M, Eriksson J, Mantel A, Wallberg-Jonsson S,
Jacobsson LT, et al. Risk of venous thromboembolism in patients with rheumatoid arthritis
and association with disease duration and hospitalization. Jama. 2012;308(13):1350-6.
150. Hellgren K, Smedby KE, Feltelius N, Baecklund E, Askling J. Do rheumatoid
arthritis and lymphoma share risk factors?: a comparison of lymphoma and cancer risks
before and after diagnosis of rheumatoid arthritis. Arthritis and rheumatism.
2010;62(5):1252-8.
151. Ludvigsson JF, Otterblad-Olausson P, Pettersson BU, Ekbom A. The Swedish
personal identity number: possibilities and pitfalls in healthcare and medical research.
European journal of epidemiology. 2009;24(11):659-67.
152. Svensk Reumatologis Kvalitetsregister (SRQ) årsrapport 2019. Retrieved 2022
January from: https://srq.nu/wp-content/uploads/2020/06/SRQ-arsrapport-2019.pdf.
153. Socialstyrelsen. Täckningsgrad för nationella kvalitetsregister 2020. Retrieved
2022 January from: https://www.socialstyrelsen.se/globalassets/sharepoint-
dokument/artikelkatalog/statistik/2020-12-7049.pdf [
154. Ludvigsson JF, Almqvist C, Bonamy AK, Ljung R, Michaëlsson K, Neovius
M, et al. Registers of the Swedish total population and their use in medical research.
European journal of epidemiology. 2016;31(2):125-36.
155. Socialstyrelsen; Patientregistret. Extracted 2022 January from:
https://www.socialstyrelsen.se/statistik-och-data/register/alla-register/patientregistret/ [
156. Wallerstedt SM, Wettermark B, Hoffmann M. The First Decade with the
Swedish Prescribed Drug Register - A Systematic Review of the Output in the Scientific
Literature. Basic & clinical pharmacology & toxicology. 2016;119(5):464-9.
157. Barlow L, Westergren K, Holmberg L, Talbäck M. The completeness of the
Swedish Cancer Register: a sample survey for year 1998. Acta oncologica (Stockholm,
Sweden). 2009;48(1):27-33.
158. Rothman K. Types of Epidemiologic Studies: Oxford University Press; 2012.
93 p.
159. World Medical Association Declaration of Helsinki: ethical principles for
medical research involving human subjects. Jama. 2013;310(20):2191-4.
160. Lag (2003:460) om etikprövning av forskning som avser människor. Retrieved
January 2022 from: https://www.lagboken.se/Lagboken/start/skoljuridik/lag-2003460-om-
etikprovning-av-forskning-som-avser-manniskor/d_4060-lag-2003_460-om-etikprovning-av-
forskning-som-avser-manniskor [
161. Rothman K. Epidemiology: An Introduction. : Oxford University Press; 2012.
87 p.
162. Rothman K. Epidemiology: An Introduction: Oxford University Press; 2012.
69-70 p.
163. Wolfe F, Michaud K, Pincus T, Furst D, Keystone E. The disease activity score
is not suitable as the sole criterion for initiation and evaluation of anti-tumor necrosis factor

71
therapy in the clinic: discordance between assessment measures and limitations in
questionnaire use for regulatory purposes. Arthritis and rheumatism. 2005;52(12):3873-9.
164. Rothman K. Epidemiology: An Introduction: Oxford University Press; 2012.
126 p.
165. Singh JA, Furst DE, Bharat A, Curtis JR, Kavanaugh AF, Kremer JM, et al.
2012 update of the 2008 American College of Rheumatology recommendations for the use of
disease-modifying antirheumatic drugs and biologic agents in the treatment of rheumatoid
arthritis. Arthritis care & research. 2012;64(5):625-39.
166. Wadström H, Frisell T, Askling J. Malignant Neoplasms in Patients With
Rheumatoid Arthritis Treated With Tumor Necrosis Factor Inhibitors, Tocilizumab,
Abatacept, or Rituximab in Clinical Practice: A Nationwide Cohort Study From Sweden.
JAMA internal medicine. 2017;177(11):1605-12.
167. Wadström H, Frisell T, Sparén P, Askling J. Do RA or TNF inhibitors increase
the risk of cervical neoplasia or of recurrence of previous neoplasia? A nationwide study
from Sweden. Annals of the rheumatic diseases. 2016;75(7):1272-8.
168. Pedersen M, Klarlund M, Jacobsen S, Svendsen AJ, Frisch M. Validity of
rheumatoid arthritis diagnoses in the Danish National Patient Registry. European journal of
epidemiology. 2004;19(12):1097-103.
169. Kochi Y, Suzuki A, Yamada R, Yamamoto K. Ethnogenetic heterogeneity of
rheumatoid arthritis-implications for pathogenesis. Nature reviews Rheumatology.
2010;6(5):290-5.
170. Conigliaro P, D'Antonio A, Pinto S, Chimenti MS, Triggianese P, Rotondi M,
et al. Autoimmune thyroid disorders and rheumatoid arthritis: A bidirectional interplay.
Autoimmunity reviews. 2020;19(6):102529.
171. Rezaei H, Saevarsdottir S, Forslind K, Albertsson K, Wallin H, Bratt J, et al. In
early rheumatoid arthritis, patients with a good initial response to methotrexate have excellent
2-year clinical outcomes, but radiological progression is not fully prevented: data from the
methotrexate responders population in the SWEFOT trial. Annals of the rheumatic diseases.
2012;71(2):186-91.
172. Studenic P, Radner H, Smolen JS, Aletaha D. Discrepancies between patients
and physicians in their perceptions of rheumatoid arthritis disease activity. Arthritis and
rheumatism. 2012;64(9):2814-23.
173. Parodis I, Studenic P. Patient-Reported Outcomes in Systemic Lupus
Erythematosus. Can Lupus Patients Take the Driver's Seat in Their Disease Monitoring?
Journal of clinical medicine. 2022;11(2).
174. Weldring T, Smith SM. Patient-Reported Outcomes (PROs) and Patient-
Reported Outcome Measures (PROMs). Health services insights. 2013;6:61-8.
175. Wessels JA, Huizinga TW, Guchelaar HJ. Recent insights in the
pharmacological actions of methotrexate in the treatment of rheumatoid arthritis.
Rheumatology (Oxford, England). 2008;47(3):249-55.

72