Journal Pre-proof

Overall and Annual Post-Colonoscopy Colorectal Cancer Rates in a Large

Integrated Healthcare Setting: A Cross-sectional Study

Jeffrey K. Lee, MD, MPH, James H-E. Kang, MA(Oxon), BM BCh, Sophie A.
Merchant, MPH, Christopher D. Jensen, PhD, MPH, Nicholas E. Burr, MBBS, MD,
Douglas A Corley, MD, PhD.

PII: $1542-3565(23)00225-2
DOI: https://doi.org/10.1016/j.cgh.2023.03.017
Reference: YJCGH 58903

To appearin: Clinical Gastroenterology and Hepatology

Accepted Date: 16 March 2023

Please cite this article as: Lee JK, H-E. Kang J, Merchant SA, Jensen CD, Burr NE, Corley DA, Overall
and Annual Post-Colonoscopy Colorectal Cancer Rates in a Large Integrated Healthcare Setting: A
Cross-sectional Study, Clinical Gastroenterology and Hepatology (2023), doi: https://doi.org/10.1016/
j.gh.2023.03.017.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2023 by the AGA Institute

Overall and Annual Post-Colonoscopy Colorectal Cancer Rates in a Large Integrated
Healthcare Setting: A Cross-sectional Study

Authors:
) effrey K. Lee, MD, MPH;* ] ames H-E. Kang, MA(Oxon), BM BCh;*23 Sophie A. Merchant,
MPH;! Christopher D. J ensen, PhD, MPH;! Nicholas E. Burr, MBBS, MD,*5 Douglas A. Corley,
MD, PhD.t

Short title: Post-colonoscopy colorectal cancer rates

Author names in bold designate shared co-first authorship.

Author institutions:
1. Division of Research, Kaiser Permanente Northern California, Oakland, CA.
2. School of Public Health, University of California, Berkeley, Berkeley, CA
3. Eastof England Deanery, United Kingdom
4. Cancer Epidemiology Group, Institute of Cancer and Pathology and Institute of Data
Analytics, University of Leeds, Leeds
5. Mid Yorkshire Hospitals NHS Trust, Pinderfields General Hospital, Wakefield, UK.

Grant support: This study was funded by a Community Health grant from Kaiser Permanente
Northern California.

Conflict of interest statement: No conflicts of interest exist for any of the authors.

Corresponding author:
) effrey K. Lee, MD, MPH
Kaiser Permanente Northern California, Division of Research
2000 Broadway, Oakland, CA 94612
510-891-5918
) effrey.K.Lee@ kp.org

Author contributions:
Study concept and design: ) KL, | HEK, SAM, CD)
Acquisition of data: SAM
Analysis and interpretation of data: | KL, HEK, SAM, CDJ
Drafting of the manuscript: ) KL, J HEK, SAM, CD)
Critical revision of the manuscript for important intellectual content: J KL, JHEK, SAM, CDJ, NB,
DAC
Colonoscopy reduces colorectal cancer (CRC) incidence and mortality through removal of pre-
cancerous adenomatous lesions and earlier detection of cancer (1). However, some patients
develop CRC after a colonoscopy in which cancer was not diagnosed, termed post-colonoscopy
CRC (PCCRC). Reasons for PCCRC include missed or incompletely excised lesions and de-
novo lesions, and therefore, the PCCRC rate has been proposed as a quality measure for
colonoscopy services (2).

PCCRC rates and interpretation vary considerably depending on the calculation method used.
Using four different calculation methods applied to a single data set, Morris et al. obtained
PCCRC rate estimates ranging from 2.5 to 7.7% (3). To promote consistency in PCCRC rate
reporting and to permit benchmarking, the World Endoscopy Organization (WEOQ) recently
published a consensus statement for calculating PCCRC rates using an interval of 3 years
ollowing colonoscopy, referred to as the PCCRC-3y rate (3). To date few studies have reported
PCCRC-3y rates using WEO methodology and none are from the United States (US) (4). As
CRC screening programs and colonoscopy services increasingly emphasize quality, population-
based studies are needed to inform PCCRC-3y rate benchmarks and aspirational targets.

o0 address this knowledge gap, we conducted a cross-sectional study and calculated PCCRC-
3y rates among health plan members of Kaiser Permanente Northern California (KPNC), a
arge, demographically diverse, integrated healthcare system that includes patients with
commercial, Medicare, and Medicaid insurance. The study population included all patients 18
years and older who underwent a colonoscopy between J anuary 1, 2006, and December 31,
2017, and were diagnosed with CRC (defined as adenocarcinoma of the colon or rectum) at or
within 36 months after the procedure. CRC cases were obtained from the KPNC cancer registry,
which reports to the Surveillance Epidemiology and End Results (SEER) registry and maintains
a >97% population-based completeness standard as verified by random audits by the cancer
registry and SEER.

We used the WEO methodology to calculate PCCRC-3y rates both overall for the time interval
2006-2017 and annually. Colonoscopies were categorized as true positive (TP) if a CRC was
diagnosed at or within 6 months after the procedure and as false negative (FN) if a CRC was
diagnosed >6 months to 36 months after the procedure. Per WEO recommendations, for
patients who underwent multiple colonoscopies in the study period, only the TP colonoscopy
and/or FN colonoscopy closest to and on or before the CRC diagnosis date, were included in
the PCCRC-3y rate calculation. Additional details on rate calculations can be found in
Supplemental Methods.

We calculated PCCRC-3y rates using the following equation in accordance with the WEO
methodology:
FN colonoscopies / (TP colonoscopies +FN colonoscopies) x 100

Trends in annual PCCRC-3y rates over time were assessed using the Cochrane-Armitage test.

In total, 902,023 colonoscopies were performed in 2006-2017 among 666,088 KPNC members
aged 18 years and older; for 15,526 of these colonoscopies (14,074 patients), CRC was
diagnosed at or within 36 months after the procedure. Among the 14,074 patients, 47.9% were
female, the majority were non-Hispanic White, and the primary indications for colonoscopy were
diagnostic (63.4%) and follow-up to a positive fecal immunochemical test (24.5%)
(Supplementary Table 1).
We identified 10 unique clinical scenarios where one or more colonoscopies were performed
and a CRC was diagnosed at or within 36 months afterwards. Detailed clinical scenarios and
how each colonoscopy leading to the CRC diagnosis contributed to the PCCRC-3y rate
calculation are described in Supplemental Table 2. In total, 14,740 of the 15,526 (94.9%)
colonoscopies contributed 13,299 TP and 818 FN colonoscopies to the overall PCCRC-3y rate
calculation. The overall rate across the 2006-2017 study interval was calculated as follows:

Overall PCCRC-3y rate =[(FN colonoscopies / (TP- colonoscopies +FN colonoscopies)) x 100]
=(818/13,299 +818) x 100
=5.5%

Annual PCCRC-3y rates were stable over time, ranging from a high of 6.3% to a low of 4.7%
(test for trend P=0.261) (Figure 1). In a sensitivity analysis where patients who had a prior
history of CRC were notincluded in the analysis and only the first cancer was considered in
patients diagnosed with CRC at two different times during the study interval, the overall
PCCRC-3y rate and annual rate estimates did not change substantively (Supplemental Table
3).

Since the WEO consensus document was published in 2018 (2), a meta-analysis of three
European and one Hong Kong population-based cohort studies reported a pooled PCCRC-3y
rate estimate of 8.2% (6-9), with a range of 7.0% to 10.4% across studies (4). The lower overall
PCCRC-3y rate and stability of the annual rate estimates in the present study may stem from
multiple factors, including monitoring/reporting of physician adenoma detection rate as an
important colonoscopy quality indicator; adoption of higher-definition colonoscopes over time;
establishment of an advanced endoscopy referral center for managing complex polyps; and
increased population tracking and follow-up colonoscopies for those requiring high-risk polyp
surveillance.
Annual PCCRC-3y rate estimates were stable across the study period, ranging from 4.7% to
6.3%. Studies describing declining annual rates over time all reported higher initial rate
estimates (e.g., 9.0% to 22.5%) (7-9). Given a strong emphasis within KPNC on colonoscopy
quality and the stable annual rate estimates observed, a PCCRC-3y rate of 5.0% to 6.0% may
inform the lower end of the target range for this quality metric. Burr et al. argued thata 5.5%
PCCRC-3y should be a reasonable benchmark; this was based on the 25" percentile from the
range of PCCRC-3y rates in their findings (7). Our findings suggest that this minimum
benchmark is achievable in a large, community-based, integrated healthcare setting and given
estimates that nearly 70% of PCCRCs may be avoidable (10), aspirational rates lower than
5.5% may be achievable. Also, given PCCRCs are relatively rare events, calculating PCCRC-3y
rates annually may lead to variability in rate estimates that could be masking small changes in
PCCRC rates.
Study strengths include a diverse, community-based cohort, an integrated healthcare system
thatincluded all colonoscopies performed; a validated cancer registry with >97% capture of
cancers; and explicit elucidation of multiple clinical scenarios for PCCRC-3y rate calculation that
provide practical standardized methods to inform standardized calculations in other settings.
The fact that all patients were insured and from Northern California may limit generalizability to
other settings.
In summary, the overall PCCRC-3y rate was 5.5% over the 12-year study period and annual
rates were stable in a large, demographically diverse, US community-based healthcare setting;
these estimates provide a benchmark to use as an indicator of the quality of the colonoscopy
service and can inform discussion of PCCRC-3y rate standards, methods, and aspirational
targets.

REFERENCES (10 max)

1. Rex DK, Boland CR, Dominitz J A, Giardiello FM, J ohnson DA, Kaltenbach T, et al.
Colorectal Cancer Screening: Recommendations for Physicians and Patients From the
U.S. Multi-Society Task Force on Colorectal Cancer. Gastroenterology 2017;153(1):307-
23.
Rutter MD, Beintaris |, Valori R, Chiu HM, Corley DA, Cuatrecasas M, etal. World
Endoscopy Organization Consensus Statements on Post-Colonoscopy and Post-Imaging
Colorectal Cancer. Gastroenterology 2018;155(3):909-925.e3.
Morris EJ A, Rutter MD, Finan PJ, Thomas | D, Valori R. Post-colonoscopy colorectal
cancer (PCCRC) rates vary considerably depending on the method used to calculate
them: A retrospective observational population-based study of PCCRC in the English
National Health Service. Gut 2015;64(8):1248-56.
Kang J HE, Evans N, Singh S, Samadder NJ, Lee ] K. Systematic review with meta-
analysis: the prevalence of post-colonoscopy colorectal cancers using the World
Endoscopy Organization nomenclature. Aliment Pharmacol Ther. 2021;54(10):1232-42.
Krieger N. Overcoming the absence of socioeconomic data in medical records: Validation
and application of a census-based methodology. Am ] Public Health. 1992;82(5):703-10.
Lee CK, ChoiKS, Eun CS, Park D-I, Han DS, Yoon M, et al. Risk and Characteristics of
Postcolonoscopy Interval Colorectal Cancer after a Positive Fecal Test: A Nationwide
Population-Based Study in Korea. Cancer Res Treat 2018;50(1):50-9.
Burr NE, Derbyshire E, Taylor |, Whalley S, Subramanian V, Finan P}, et al. Variation in
post-colonoscopy colorectal cancer across colonoscopy providers in English National
Health Service: Population based cohort study. BMJ 2019;367.
Pedersen L, Valori R, Bernstein I, Lindorff-Larsen K, Green C, Torp-Pedersen C. Risk of
post-colonoscopy colorectal cancer in Denmark: time trends and comparison with
Sweden and the English National Health Service. Endoscopy 2019;51(8):733-41.
Forsberg A, Widman L, Bottai M, Ekbom A, Hultcrantz R. Postcolonoscopy Colorectal
Cancer in Sweden From 2003 to 2012: Survival, Tumor Characteristics, and Risk
Factors. Clin Gastroenterol Hepatol 2020;18(12):2724-2733.e3..
10. Anderson R, Burr NE, Valori R. Causes of Post-Colonoscopy Colorectal Cancers Based
on World Endoscopy Organization System of Analysis. Gastroenterology. 2020
Apr;158(5):1287-1299.e2. doi: 10.1053/j.gastr0.2019.12.031. Epub 2020 J an 8. PMID:
31926170.
Flgure 1. AnnuarrLek L-3y rdiles, ZUuvo-ZulL/

PCCRC, post-colonoscopy colorectal cancer; TP, true positive colonoscopies; FN, false negative colonoscopies
PCCRC-3y Rate, % P=261

34

24

14

04
T T T T T T T T T T T T
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year of Colonoscopy
TP 964 1,035 1309 1257 1,138 1180 1,077 109 1178 1177 1207 1,304
FN 65 60 74 73 75 7 71 59 53 65 il 75
Supplemental Table 1. Patient demographic and colonoscopy characteristics, 2006-2017
N (%)
Total 14,074
Age, years
First Cancer Diagnosis
<50 1,233 (8.8)
50-75 8,714 (61.9)
276 4,127 (29.3)
Mean (SD) 67.1 (13.0)
Median (IQR) 67.4 (57.6, 77.0)
Sex
Female 6,748 (47.9)
Male 7,325 (52.1)
Qther or Unknown 1(0.0)
Race and Ethnicity
Asian or Pacific Islander 2,032 (14.4)
Black 1,156 (8.2)
Hispanic 1,826 (13.0)
Non-Hispanic White 9,000 (64.0)
QOther or Unknown 60 (0.4)
Colonoscopies
Total 15,526 (100)
Screening 1,032 (6.6)
Surveillance 697 (4.5)
Follow-up to positive FIT 3,808- (24.5)
Diagnostic 9,837 (63.4)
Unknown indication 152 (1.0)
FIT, fecal immunochemical test; SD, standard deviation; IQR, interquartile range
Supplemental Table 2. Description of PCCRC cases by clinical scenario and how colonoscopies
preceding the diagnosis contributed to the overall PCCRC-3y rate, 2006-2017
Colonoscopies, N
Within the 2006-2017 date range for study eligibility: Pcc;;g:’ rot Contributingto
N(%) T°® _PpccRC-3yrate
Clinical scenarios: FN TP
A colonoscopy was performed and cancer was diagnosed at or 12877
1 within 6 months after the procedure. All colonoscopies performed (9’0 4) 12,877 0 12,877
were TP_and count toward the rate calculation. |
2 or more colonoscopies were performed within 6 months before 552
2 the cancer diagnosis. Only the closestTP colonoscopy to the (3.9) 1,139 0 552
diagnosis date counts toward the rate calculation.
A colonoscopy was performed that did not resultin a cancer
diagnosis at or within 6 months after the procedure, but a second 380
3 colonoscopy was performed within 36 months after the first and it 2N 760 380 380
did diagnose cancer. Both the FN and TP colonoscopy count .
toward the rate calculation.
2 or more colonoscopies were performed that did not diagnose
cancer at or within 6 months after the procedures, but a
subsequent colonoscopy was performed within 36 months after 23(0.1) 71 23 23
the first that did diagnose cancer. Only the closestFN and TP '
colonoscopy to the diagnosis date count toward the rate
calculation.
A colonoscopy was performed that did not diagnose cancer at or
within 6 months after the procedure, but 2 or more subsequent
colonoscopies were performed within 36 months after the first 81(0.6) 266 81 81
procedure and the last procedure did diagnose cancer. Only the :
closestFN and TP colonoscopy to the diagnosis date count
toward the rate calculation.
2 or more colonoscopies were performed that did not diagnose
cancer at or within 6 months after the procedure, but 2 or more
6 subsequent colonoscopies were performed within 36 months after 9(0.1) 39 9 9
the first colonoscopy and the last procedure did diagnose cancer. .
Only the closestFN and TP colonoscopy to the diagnosis date
count toward the rate calculation.
A colonoscopy was performed that did not diagnose cancer at or
within 6 months after the procedure, but within 36 months after 182
7 the procedure cancer was diagnosed by a non-colonoscopy 13) 182 182 0
method (e.g., surgery). All FN colonoscopies count toward the A
rate calculation and there were no TP procedures.
2 or more colonoscopies were performed that did not diagnose
cancer at or within 6 months after the procedure, but within 36
8 months after the first procedure cancer was diagnosed by a non- 18(0.1) 37 18 0
colonoscopy method (e.g., surgery). Only the closestFN '
colonoscopy to the diagnosis date counts toward the rate
calculation.
A colonoscopy was performed that did not diagnose cancer at or
within 6 months after the procedure, but within 36 months after 104
9 the procedure cancer was diagnosed; however, the cancer and ©0.7) 104 104 0
diagnosing procedure were outside the 2006-2017 date range. All
colonoscopies were FN and all count toward the rate calculation.
2 colonoscopies were performed that did not diagnose cancer at
or within 6 months after the procedure, but within 36 months after
the first procedure cancer was diagnoses; however, the cancer 21(0.1) 51 21 0
10
and diagnosing procedure were outside the 2006-2017 date '
range. Only the closestFN colonoscopy to the diagnosis date
counts toward the rate calculation.
Total 14,247 15,526 818 13,922
CRC, colorectal cancer; PCCRC, post-colonoscopy colorectal cancer; TP, true positive; FN, false negative
*PCCRC-3y calculations require a specific date range for cancer follow-up after colonoscopies. Cancers after this date range
should be included but not colonoscopies, to provide both three years of follow-up (for colonoscopies within the date range)
and fixed numerators and denominators. Thus, for #3 and #10, we include three years of cancer follow-up diagnoses for

2
calculation of TP and FN colonoscopies completed during the date range but do not include colonoscopies outside of that date
range. The date range was selected to provide three years of follow-up for 2017 colonoscopies (through 2020) followed by
cancer registry validation of 2020 cancer cases (which was completed in 2022).
Supplemental Table 3. Sensitivity analysis comparing post-colonoscopy colorectal cancer-3y
(PCCRC-3y) rates overall (2006-2017) and annually based on all cancers versus new cancers
diagnosed during the study interval.*
PCCRC-3y Rates PCCRC-3y Rates
All Cancers New Cancers
N=14,247 N=11,931
Total 5.5% 5.4%
Year of colonoscopy
2006 6.3% 6.0%
2007 5.5% 5.4%
2008 5.4% 5.1%
2009 5.5% 5.4%
2010 6.2% 6.3%
2011 6.1% 5.8%
2012 6.2% 6.4%
2013 5.1% 4.9%
2014 4.3% 3.9%
2015 5.2% 5.0%
2016 5.6% 5.0%
2017 5.4% 5.1%
P-value for trend 0.261 0.123
*Explanatory footnote: Among the 14,074 patients diagnosed with colorectal cancer (CRC) during the study interval there were
14,247 total CRCs diagnosed; 13,901 patients had one CRC diagnosis and 173 patients had 2 CRC diagnoses during the
study interval. Of the 13,901 patients with one CRC diagnosis, 2,116 had a history of CRC. Of the 173 who had 2 CRC
diagnoses, 27 had a history of CRC priorto their first CRC diagnosis in the study time interval. Because the second or third
cancer diagnosed in these patients could have been metastatic disease rather than a true second or third CRC, a sensitivity
analysis was performed where patients who had a prior history of CRC were dropped from the analysis and only the first
cancer was considered in patients diagnosed with CRC at two different times during the study interval. Thus, of the 14,247
CRC cases diagnosed during the study time interval, 11,931 new cancers (14,247-2116-173-27=11,931) were included in the
sensitivity analysis. As shown in the table, the overall PCCRC-3y rate estimate (5.5% vs 5.4%) and annual rate estimates did
not change substantively in this sensitivity analysis.

Supplemental Methods

Data Sources
Patient demographics were obtained from electronic health records. Colonoscopies were extracted
from the procedure database using Current Procedural Terminology codes, International Classification
of Disease 9™ and 10% Edition procedure codes, Healthcare Common Procedure Coding System
codes, and KPNC-specific local codes. Colorectal cancer diagnoses were extracted from the KPNC
Cancer Registry using International Classification of Diseases for Oncology, 3™ E dition (ICD-0-3)
codes and histological codes specific to adenocarcinoma.

List of Codes Used to Ascertain Colonoscopies and Colorectal Cancers

ICD-10: 0DBC8ZX, 0DBC8ZZ, 0DBF8ZX, 0DBF8ZZ, 0DBH8ZX, 0DBH8ZZ,
0DBK8ZX, 0DBK8ZZ, 0DBL8ZX, 0DBL8ZZ, 0DBM8ZX, 0DBM8ZZ
ICD-9: 45.23, 98.04
Colonoscopy f;g—;:;;gg 44394, 44397, 44401-44402, 44404-44405, 44408, 45355, 45378

HCPCS: G0105, G0121, G9659, G9660-G9661, G9936-G9937

Local Codes: 106504, 129133, 204456, 230847, 235525, 279802, 299458-
299461, 566735, 643763-643764, 696739
ICD-0-3: C18.0,C18.2,C18.3,C18.4,C18.5,C18.6,C18.7,C18.8, C18.9,
Colorectal €19.9, €20.9
Cancer Morphology: 8000, 8010, 8020-8021, 8140-8141, 8143-8144, 8200, 8210-8211,
8215, 8220-8221, 8230, 8255, 8260-8263, 8323, 8410, 8430, 8440, 8470, 8480-
8481, 8490, 8510, 8560, 8570-8574, 8576
PCCRC-3y Rate Calculations in Accordance with the WEQ Methodology
For each CRC diagnosis, colonoscopies in the 36-month interval leading up to the diagnosis date were
classified as true-positive (TP) if CRC was detected at or within 6 months after the procedure, or as
false-negative (FN) if CRC was diagnosed >6-36 months after the procedure.

For CRC diagnoses that had multiple preceding colonoscopies, only the TP colonoscopy and/or FN
colonoscopy closestto and at or before the diagnosis date were selected, and the remaining
colonoscopies performed were removed from the PCCRC-3y rate calculation in accordance with the
WEO recommendations. For example, if a CRC diagnosed in 1/2009 was preceded by two FN
colonoscopies performed in 12/2006 and 7/2007 and two TP colonoscopies performed in 11/2008 and
1/2009, the FN colonoscopy performed in 7/2007 and the TP colonoscopy performed in 1/2009
contributed to the PCCRC-3y rate calculation and those performed in 12/2006 and 11/2008 did not
contribute to the PCCRC-3y rate calculation.

Per WEO recommendations, a person may also have been diagnosed with more than one CRC. In this
situation, each colonoscopy was only included once and was assigned to the closest subsequent CRC.
Forexample, if a patient was diagnosed with CRC in 7/2009, which was preceded by colonoscopies
performed in 5/2008 (FN) and 7/2009 (TP), and a second CRC was diagnosed in 5/2011, which was
preceded by colonoscopies performed in 5/2008, 7/2009, 6/2010 and 5/2011, only the colonoscopies
performed in 2010 (FN) and 2011 (TP) would be associated with the second CRC diagnosis.

Overall PCCRC-3y rates were calculated as the number of selected FN colonoscopies divided by the
sum of the selected TP and FN colonoscopies performed within the 12-year study period, expressed as
a percentage.
Annual PCCRC-3y rates were calculated as the number of selected FN colonoscopies divided by the
total selected TP and FN colonoscopies performed within each year of the 12-year study period,
expressed as a percentage. Although PCCRC-3y rate are commonly referred to as “rates” they
represent a percentage.

Finally, the WEO-recommended method for calculating PCCRC rates does not account for individuals
who electto leave the catchment area after a colonoscopy, and so there could be additional PCCRCs
that are unaccounted for. This is an inherent limitation of the PCCRC rate methodology. For our study,
902,023 colonoscopies were performed during the study interval and 137,031 (15.2%) of those
examined terminated their membership within 3 years after their procedure for reasons unrelated to
death or diagnosis of CRC. These individuals, had they remained members of KPNC, could have
contributed to the PCCRC-3y rate calculations if they had been diagnosed with CRC in the 3-year time
interval.
STROBE Statement—checklist of items that should be included in reports of observational studies
Item
No Recommendation Page

Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the 1
z(ib) Provide in the abstract an informative and balanced summary of what was N/A
one
Introduction
Background/rationale 2 Explain the scientific background and rationale for the investigation being 2
Objectives 3 State specific objectives, including any prespecified hypotheses 2
Methods
Study design 4 Present key elements of study design early in the paper 2
Setting 5 Describe the setting, locations, and relevant dates, including periods of 2
recruitment,
Participants 6 (a) Cohort study—Give the eligibility criteria, and the sources and 2
methods of selection of participants. Describe methods of follow-up S2
Case-control study—Give the eligibility criteria, and the sources and methods
of case ascertainment and control selection. Give the rationale for the choice
of cases and controls
(,;mssrswfiuml study—Give the eligibility criteria, and the sources and methods
o
(b) Cohort study—For matched studies, give matching criteria and number of N/A
exposed and unexposed
C;ase\comml study—For matched studies, give matching criteria and the number
o
Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and 2
effect S2
Data sources/ 8% For each variable of interest, give sources of data and details of methods of S4
measurement assessment (measurement). Describe comparability of assessment methods if
there
Bias 9 Describe any efforts to address potential sources of bias 3
Study size 10 Explain how the study size was arrived at 2
Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, ~ 2
describe which groupings were chosen and why S5
Statistical methods 12 (a) Describe all statistical methods, including those used to control for 2,85
(b) Describe any methods used to examine subgroups and interactions N/A
(¢) Explain how missing data were addressed N/A
(d) Cohort study—1If applicable, explain how loss to follow-up was addressed 2
Case-control study—If applicable, explain how matching of cases and controls
was addressed
C;ros«sec‘fiaml study—If applicable, describe analytical methods taking account
o
() Describe any sensitivity analyses 3
Continued on next page
Results Page

Participants 13% (a) Report numbers of individuals at each stage of study— | 2

eg numbers potentially eligible, examined for eligibility,
confirmed eligible, included in the study, completing
follow-up, and analysed

(b) Give reasons for non-participation at each stage N/A

(c) Consider use of a flow diagram N/A

Descriptive data 14* (a) Give characteristics of study participants (eg 2

demographic, clinical, social) and information on
exposures and potential confounders

(b) Indicate number of participants with missing data for N/A

each variable of interest

(c) Cohort study—Summarise follow-up time (eg, average N/A

and total amount)

Outcome data 15* Cohort study—Report numbers of outcome events or N/A

summary measures over time

Case-control study—Report numbers in each exposure category, | N/A

or summary measures of exposure
Cross-sectional study—Report numbers of outcome events or 3,55
summary measures

Main results 16 (a) Give unadjusted estimates and, if applicable, 3,55

confounder-adjusted estimates and their precision (eg,
95% confidence interval). Make clear which confounders
were adjusted for and why they were included

(b) Report category boundaries when continuous variables | 3

were categorized

(c) If relevant, consider translating estimates of relative N/A

risk into absolute risk for a meaningful time period

Other analyses 17 Report other analyses done—eg analyses of subgroups 3

and interactions, and sensitivity analyses

Discussion

Key results 18 Summarise key results with reference to study objectives 3

Limitations 19 Discuss limitations of the study, taking into account 3

sources of potential bias or imprecision.

Discuss both direction and magnitude of any potential bias

Interpretatlon 20 Give a cautious overall interpretation of results 3

considering objectives, limitations, multiplicity of analyses,

results from similar studies, and other relevant evidence

Generalisability 21 Discuss the generalisability (external validity) of the study 3

results

Other information

Funding 22 Give the source of funding and the role of the funders for 1
the present study and, if applicable, for the original study
on which the present article is based

S, Supplemental; N/A, not applicable

*Give information separately for cases and controls in case-control studies and, if applicable, for exposed and unexposed
groups in cohort and cross-sectional studies.

Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published
examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely available on the Web
sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at http://www.annals.org/, and Epidemiology
at http://www.epidem.com/). Information on the STROBE Initiative is available at www.strobe-statement.org.