A Comparison of Symptom Scores
and Radiographic Staging Systems
in Chronic Rhinosinusitis
Neil Bhattacharyya, M.D.
ABSTRACT
Background: The aim of this study was to determine if one of
the popular computed tomography (CT) scan staging systems
exhibits better correlation with sinonasal symptom severity in
chronic rhinosinusitis (CRS).
Methods: Adult patients meeting diagnostic criteria for CRS
were prospectively studied with the Rhinosinusitis Symptom
Inventory and their paranasal sinus CT scan was staged ac-
cording to the Kennedy, Harvard, and Lund staging systems.
From the Rhinosinusitis Symptom Inventory, Likert symptom
severity scores were obtained and the nasal, facial, oropha-
ryngeal, systemic, and total symptom scores were computed.
Correlation analysis was conducted between symptom scores
and CT scan stage for each of the three staging systems.
Results: Two hundred patients met inclusion criteria (mean
age, 41.1 years; 42% men). The mean Lund score for the entire
cohort was 10.0 (SD, 5.3). The Kennedy stage distributions
were 40, 42, 84, and 34 patients for stages 1–4, respectively.
The Harvard stage distributions were 17, 53, 85, and 45
patients for stages 1–4, respectively. All three staging systems
exhibited statistically significant Pearson correlation (all p ⬍
0.001) for the nasal symptom domain (nasal obstruction, rhi-
norrhea, and hyposmia), although the magnitude of the corre-
lation coefficient generally was small (range of Pearson’s r,
0.242–0.345). The Kennedy and Harvard systems exhibited
From Division of Otolaryngology, Brigham and Women’s Hos-
pital, Boston, Massachusetts, and Department of Otology and
Laryngology, Harvard Medical School, Boston, Massachusetts
Address correspondence and reprint requests to Neil Bhatta-
charyya, M.D., Division of Otolaryngology, Brigham and
Women’s Hospital, 333 Longwood Avenue, Boston, MA 02115
Copyright © 2005, OceanSide Publications, Inc., U.S.A.
American Journal of Rhinology
significant but negative correlation with the oropharyngeal
symptom domain. All three staging systems failed to signifi-
cantly correlate CT stage with facial symptoms, systemic symp-
toms, and total symptom scores.
Conclusion: The Lund staging system exhibited the best cor-
relation between nasal symptom scores and CT stage in CRS,
but the degree of correlation remained small. All three systems
were lacking in staging value for nonnasal sinonasal symp-
toms. (American Journal of Rhinology 19, 175–179, 2005)
T he paranasal sinus computed tomography (CT) scan
has come to be relied on as an important diagnostic
and staging tool in the evaluation of chronic rhinosinusitis
(CRS).1 Because of its objective nature, substantial interest
has been focused on the usefulness of the CT scan in staging
patients with CRS. However, the value of and the method-
ology for preoperative CT staging of patients who plan to
undergo endoscopic sinus surgery (ESS) still has to find
widespread consensus. Much effort has been focused on the
development of different CT staging systems and interrater
and intrarater reliability measures for the systems.2–4 Cur-
rently, three paranasal sinus CT scan staging systems are in
common use: the Kennedy staging system, the Harvard
staging system, and the Lund staging system.5–7 Each of the
systems was developed with the intention of staging or
stratifying patients according to volumetric radiographic
levels of disease. With such stratification schemes, it was
hoped that the CT scan could predict several important
clinical factors such as symptom severity with CRS, sub-
jective response to ESS, and recurrence rates after ESS.
However, several studies using the staging systems have
failed to show solid correlations between patients’ reported
symptoms and CT scan stage.8,9 Others have found that
certain sinonasal symptoms do correlate with CT scan stage
in select populations.10 It is conceivable that some of this
variability in degree of correlation could be caused by
175
 Figure 1. Distribution of CT scan stage for the Lund system.
intrinsic differences in the usefulness of each of the staging
systems. If one staging system exhibited superior correla-
tion with patients’ reported subjective symptoms or disease
severity parameters, such a staging system might be more
useful for disease stratification and outcome studies for the
treatment of CRS. Therefore, this study was conceived to
examine specifically which of these three major radio-
graphic staging systems for CRS correlates best with indi-
vidual symptoms attributed to CRS.
METHODS
A convenience sample of a consecutive series of adult
patients undergoing evaluation and treatment for
symptoms compatible with CRS formed the population for
this study. This study was approved by our hospital’s com-
mittee on clinical investigations. Inclusion criteria consisted
of (1) symptoms meeting the American Academy of Oto-
laryngology criteria for the symptoms-based diagnosis of
CRS lasting ⬎12 weeks and refractory to standard medical
management and (2) a paranasal sinus CT scan conducted
during the routine course of evaluation.11 During the course
of evaluation, each patient was asked to complete the Rhi-
nosinusitis Symptom Inventory (RSI), which tabulates the
major and minor symptoms of CRS on a five-point Likert
scale.12–14 The RSI has previously established good internal
reliability with a value for Cronbach’s ␣ of 0.85. Test-retest
reliability was determined by administering the RSI to 14
patients ⬎2 weeks apart. No significant differences were
noted in nasal, facial, oropharyngeal, systemic, or total
symptom domains between the two tests (all p ⬎ 0.5; paired
Student’s t-test). For patients meeting inclusion criteria, the
CT scans were retrieved and staged in random order and
blinded to the RSI data according to each of the following
established CT scan scoring systems: the Kennedy system,
the Harvard system, and the Lund system.
Cases were excluded from the analysis if the complete
176
Figure 2. Distribution of CT scan stage for the Kennedy system.
CT scan could not be staged according to any one of the
individual staging systems. In addition, according to pub-
lished criteria, cases with Lund scores of 0 or 1 on review
of the CT scan were excluded from subsequent analysis as
not having sufficient radiographic evidence of CRS.1 From
the prospectively obtained RSI information, Likert-scale
data for the six major and six minor symptoms of CRS were
tabulated. Nasal (nasal obstruction, rhinorrhea, and dysos-
mia), facial (facial pressure, congestion, and headache),
oropharyngeal (dental pain, ear pain, cough, and halitosis),
systemic (fever and fatigue), and total symptom domain
scores were computed also as previously described with a
scale score of 0 representing absence of symptomatology
and 100 representing maximum symptomatology.13
Data were tabulated in a Microsoft Excel spreadsheet and
imported to SPSS version 10.0 (SPSS, Chicago, IL). Stan-
dard descriptive data for this cohort were computed. For
each of the three staging systems, correlation analysis was
conducted for the five symptom domains and the corre-
sponding numerical CT scan stage with the Pearson’s prod-
uct moment correlation coefficient. Statistical significance
for correlation was set at p ⬍ 0.05. Correlation coefficients
for each of the staging systems were compared to determine
if one staging system exhibited superior correlation with
symptom scores.
RESULTS
A total of 211 adult patients initially met inclusion
criteria. Of these, 11 patients were excluded based on
failure to meet radiographic criteria for disease. All of the
remaining 200 patients were staged successfully according
to each of the three staging systems under study. Mean
patient age was 41.1 years, consisting of 116 women and 84
men. Figures 1–3 depict the CT staging distribution histo-
grams for each of the systems.
To facilitate data presentation for graphical and tabular
March-April 2005, Vol. 19, No. 2
 TABLE I

Mean Scores for Rhinosinusitis Symptom Inventory

Symptom Domains According to CT Scan Stage*
Kennedy Harvard Lund*
Nasal Symptom Domain
Stage I 53.9 53.9 51.4
Stage II 51.1 50.8 55.4
Stage III 58.9 57.4 64.1
Stage IV 72.2 71.4 79.1
Facial Symptom Domain
Stage I 58.1 54.5 52.9
Stage II 51.3 55.8 56.1
Stage III 52.5 53.4 48.1
Stage IV 48.6 47.1 47.2
Oropharyngeal Symptom
Figure 3. Distribution of CT scan stage for the Harvard system. Domain
Stage I 37.6 33.9 31.8
Stage II 30.6 34.2 34.6
comparison with the other staging systems, the Lund stag- Stage III 32.0 32.5 28.8
ing system was subdivided into four “stage” categories Stage IV 22.6 24.3 21.6
evenly across the range of scores from 0 to 24 (0–6, 7–12, Systemic Symptom Domain
13–18, and 19–24). Table I lists the RSI symptom domain Stage I 37.6 32.4 35.9
mean scores for each staging system according to stage. Stage II 38.0 40.4 37.4
Statistically significant correlations between CT scan stage Stage III 33.6 33.6 31.7
and nasal symptom domain were identified for each of the Stage IV 29.1 30.4 25.9
staging systems. Despite statistical significance, the values Total Symptoms
for the correlation coefficients generally were small, rang- Stage I 46.0 43.1 41.8
ing from 0.242 to 0.345. The Lund staging system exhibited Stage II 40.7 43.3 44.1
the truest linear increase in nasal symptoms according to CT Stage III 43.0 43.1 42.5
scan stage. These findings for the nasal symptom domain Stage IV 42.0 42.1 42.1
are graphically depicted in Fig. 4. Both the Kennedy and the
*Domain scores range: 0 (no symptoms) to 100 (maximal
Harvard systems showed little distinction in nasal symptom
symptoms).
scores between stage 1 and stage 2 disease. Mean facial
symptom domain scores failed to correlate well with CT
scan stage in any of the staging systems. Interestingly,
oropharyngeal and systemic symptoms tended to decline
(negative correlation) with increasing CT scan staging each
of the individual staging systems, especially as CT scan
disease progressed toward the higher end of the spectrum.
This negative correlation with oropharyngeal symptoms ac-
tually was statistically significant for both the Kennedy and
the Harvard systems (p ⬍ 0.05). Finally, no significant
correlation was identified between the systemic symptom
domain or the total symptom score with any of the three
staging systems. The results of the statistical correlation
analysis for each of the RSI symptom domains and total RSI
score are depicted in Table II.

DISCUSSION Figure 4. Comparison of nasal symptom scores by staging system.

M any current controversies exist in the diagnosis, man-

agement, pathophysiology, and outcomes for
CRS.15 Several investigators have shown that the clinical quality of life as measured by several general and disease-
diagnosis of CRS carries with it a health burden that is not specific quality-of-life instruments.16,17 CRS also carries
solely limited to sinonasal symptoms. CRS affects global with it a tremendous economic burden for the patient and

American Journal of Rhinology 177

 TABLE II
Results of Correlation Analysis for Symptoms of CRS and CT Stage
Kennedy
r* p**
Nasal Domain 0.242 0.001
Facial Domain ⫺0.104 0.143
Oropharyngeal Domain ⫺0.202 0.004
Systemic Domain ⫺0.128 0.074
Total Symptom Score ⫺0.055 0.435
*r, Pearson product moment correlation coefficient; **p, underlined values are significant at the 0.05 level.
the health care system in general.13 As such, significant
research effort has been devoted to classification schemes
for outcomes research in CRS. For diseases with wide
spectra of symptom manifestations, staging is important for
severity stratification and prognostication. Accurate staging
allows for selection of therapy and prognostic information
useful for counseling patients as to the likelihood of suc-
cessful treatment outcomes. It also allows for comparison of
treatment outcomes between different treatment modalities
and clinical trials. In CRS, the CT scan traditionally has
been the most commonly used staging tool.
As such, several previous investigators have elaborated
various staging systems for the paranasal sinus CT scan to
stratify patients according to disease severity. Other staging
systems have been proponed even beyond those studied
here, e.g., the May or Friedman staging systems.2 Although
many of these staging systems have been used for clinical
and research purposes, a specific comparison of staging
systems with respect to their individual correlations with
sinonasal symptoms has been lacking. Therefore, this study
was conducted to determine if one of these commonly used
staging systems exhibited superior correlations with symp-
toms of CRS and thus might be more valuable as a tool for
clinical prognostication and disease stratification for re-
search.
The CT scan is considered the gold standard radiographic
test in the diagnosis of CRS. Previously, we have quantified
accuracy parameters for the paranasal sinus CT scan in the
adult patient population for the diagnosis of CRS and the
test-retest reliability of the CT scan in CRS.1,18 Other au-
thors previously have studied the correlation between CT
scan findings and CRS symptomatology with mixed re-
sults.8,9 Although some authors have found no correlation
between CT and patient reported symptoms, other authors
have found some correlation.10 Although the CT scan may
not correlate well with initial disease symptoms severity
parameters, other investigators have found that the degree of
disease evident on CT scan may well correlate with the
degree of response to ESS and the level of postoperative
symptoms exhibited after ESS.19 Additional study will need
to be conducted to determine if the CT scan stage also will
178
Harvard Lund
r p r p
0.272 ⬍0.001 0.345 ⬍0.001
⫺0.109 0.124 ⫺0.078 0.274
⫺0.163 0.021 ⫺0.138 0.051
⫺0.103 0.15 ⫺0.132 0.064
⫺0.022 0.757 0.02 0.774
predict response to standard medical therapy or alternative
forms of treatment, because this was not the focus of the this
study.
The current data indicate that the paranasal sinus CT scan
correlates to some degree with the severity of nasal-specific
symptomatology (the nasal symptom domain) exhibited in
CRS. However, the magnitude of the Pearson correlation
coefficient is relatively small, indicating that the variability
in nasal symptom domain scores is only partially explained
by the level of disease on the CT scan. Thus, CT scan alone
may not provide adequate prognostic or disease severity
stratification for clinical or research uses. Of the three
staging systems under study, the Lund system exhibited the
most direct linear correlation between nasal symptom do-
main severity and radiographic disease and it also exhibited
the highest Pearson correlation coefficient (Table I). The
Kennedy and Harvard systems showed little distinction in
nasal symptoms severity scores between stage 1 and stage 2
disease. Therefore, the Lund system seems the most apt of
the three for staging nasal symptom scores. However, it
should be pointed out that the degree of correlation between
sinonasal symptom scores and radiographic staging remains
modest at best.
We also found that none of the specific staging systems
exhibited high degrees of (positive) symptom correlation in
the facial, oropharyngeal, systemic, or total symptom do-
mains. Two potential explanations exist for this lack of
correlation. First, it may be possible that nonnasal symp-
toms such as facial pressure, headache, fatigue, etc. are
symptoms that simply are not specific to CRS, and therefore
the symptoms do not correlate with the degree of disease on
the CT scan. Alternatively, it is possible that the CT scan
and therefore the individual staging systems fail to identify
the nonradiographic physiological factors that contribute to
these disease symptoms. Because many of the symptoms
traditionally have been attributed to CRS, this lack of cor-
relation with the CT scan is important in the stratification of
disease for clinical and research purposes. Again, the CT
scan alone may not effectively stratify patients according to
overall disease severity regardless of the staging system
used. It is rather peculiar that both the Harvard and the
March-April 2005, Vol. 19, No. 2
Kennedy staging systems exhibited a negative but statisti-
cally significant correlation between CT scan severity and
oropharyngeal or systemic symptom scores. Although the
specific explanation for this negative correlation is not
immediately evident, these data suggest that oropharyngeal
and systemic symptoms may not be reliable specific indi-
cators of the diagnosis of CRS.
The lack of correlation between CRS symptoms and
radiographic findings on CT scan is frustrating for most
investigators. Some of this variability in correlation may
result from the inclusion of patients that have symptoms
compatible with CRS but who do not truly have CRS based
on endoscopic, radiographic, or even histopathological cri-
teria. In an effort to minimize this source of potential
variability, this study only included patients who met pre-
viously published radiographic criteria for the diagnosis of
CRS.1 This methodology merges a symptoms-based diag-
nosis of CRS with some corroborating radiographic evi-
dence in keeping with currently endorsed recommenda-
tions.15 Despite this precaution, Stewart and associates also
found that elimination of normal scans and other exclusion-
ary adjustments failed to enhance symptom correlation with
CT stage.19 CT scan traditionally has been used as an
objective measure of disease burden for patients with CRS,
mainly because it is one of the few purely objective preop-
erative measures that exist in this disease process. However,
it is possible that the CT scan is not a good measure of
disease burden for these patients, and this may, in fact, be
the true reason for lack of solid correlation between symp-
toms and CT scan stage. Therefore, with the currently
available data, reliance on the preoperative CT scan as the
sole measure of disease burden is likely to be inadequate.
In summary, we failed to find a specific staging system
that offered dramatically better correlation between the CT
scan stage and level of symptoms in patients with CRS.
Among the three staging systems, the Lund system showed
the best linear relationship between CT stage and nasal
symptom scores. Other symptoms including global health
symptoms such as fatigue are poorly correlated with CT
scan findings, and other methods of disease stratification
should be used to study these CRS symptoms that are more
difficult to correlate and quantify objectively.
REFERENCES
1. Bhattacharyya N, and Fried MP. The accuracy of computer
tomography in the diagnosis of chronic sinusitis. Laryngoscope
113:125–129, 2003.
American Journal of Rhinology
2. Metson R, Gliklich RE, Stankiewicz JA, et al. Comparison of
sinus computed tomography staging systems. Otolaryngol Head
Neck Surg 117:372–379, 1997.
3. Oluwole M, Russell N, Tan L, et al. A comparison of comput-
erized tomographic staging systems in chronic sinusitis. Clin
Otolaryngol Allied Sci 21:91–95, 1996.
4. Lund VJ, and Kennedy DW. Quantification for staging sinusitis.
The Staging and Therapy Group. Ann Otol Rhinol Laryngol
Suppl 167:17–21, 1995.
5. Kennedy DW. Prognostic factors, outcomes and staging in eth-
moid sinus surgery. Laryngoscope 102(suppl 57):1–18, 1992.
6. Gliklich RE, and Metson R. A comparison of sinus computed
tomography (CT) staging systems for outcomes research. Am J
Rhinol 33:291–297, 1994.
7. Lund VJ, and Kennedy DW. Staging for rhinosinusitis. Otolar-
yngol Head Neck Surg 117:S35–S40, 1997.
8. Bhattacharyya T, Piccirillo J, and Wippold FJ II. Relationship
between patient-based descriptions of sinusitis and paranasal
sinus computed tomographic findings. Arch Otolaryngol Head
Neck Surg 123:1189–1192, 1997.
9. Stewart M, Sicard MW, Piccirillo JF, et al. Severity staging in
chronic rhinosinusitis: Are CT findings related to patient symp-
toms? Am J Rhinol 13:161–167, 1999.
10. Kenny TJ, Duncavage J, Bracikowski J, et al. Prospective anal-
ysis of sinus symptoms and correlation with paranasal computed
tomography scan. Otolaryngol Head Neck Surg 125:40–43,
2001.
11. Lanza DC, and Kennedy DW. Adult rhinosinusitis defined. Oto-
laryngol Head Neck Surg 117:S1–S7, 1997.
12. Bhattacharyya N, and Kepnes LJ. Bipolar radiofrequency cold
coblation turbinate reduction for obstructive inferior turbinate
hypertrophy. Oper Tech Otolaryngol Head Neck Surg 13:170–
174, 2002.
13. Bhattacharyya N. The economic burden and symptom manifes-
tations of chronic rhinosinusitis. Am J Rhinol 17:27–32, 2003.
14. Bhattacharyya N. Symptom outcomes after endoscopic sinus
surgery for chronic rhinosinusitis. Arch Otolaryngol Head Neck
Surg 130:329–333, 2004.
15. Benninger MS, Ferguson BJ, Hadley JA, et al. Adult chronic
rhinosinusitis: Definitions, diagnosis, epidemiology, and patho-
physiology. Otolaryngol Head Neck Surg 129S:S1–S32, 2003.
16. Gliklich RE, and Metson R. The health impact of chronic sinus-
itis in patients seeking otolaryngologic care. Otolaryngol Head
Neck Surg 113:104–109, 1995.
17. Senior B, Glaze C, and Benninger M. Use of the rhinosinusitis
disability index in rhinologic disease. Am J Rhinol 15:15–20,
2001.
18. Bhattacharyya N. The test-retest reliability of computed tomog-
raphy in the assessment of chronic sinusitis. Laryngoscope 109:
1055–1058, 1999.
19. Stewart MG, Donovan DT, Parke RB Jr, and Bautista MH. Does
the severity of sinus computed tomography findings predict
outcome in chronic sinusitis? Otolaryngol Head Neck Surg 123:
81–84, 2000. e
179
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.