American Journal of Gastroenterology
C ?2005 by Am. Coll. of Gastroenterology
Published by Blackwell Publishing
Lack of Seasonal Variation in the Endoscopic Diagnoses
of Crohn’s Disease and Ulcerative Colitis
Joel N. Auslander, M.D., David A. Lieberman, M.D., and Amnon Sonnenberg, M.D., M.Sc.
Portland VA Medical Center and Oregon Health & Science University, Oregon
BACKGROUND: Conflicting data have been reported about the seasonal variation of inflammatory bowel diseases
(IBD). The purpose of the present analysis was to assess the occurrence of seasonal variations in the
endoscopic diagnosis of Crohn’s disease (CD) and ulcerative colitis (UC).
METHODS:The Clinical Outcomes Research Initiative (CORI) uses a computerized endoscopic report generator
to collect endoscopic data from 73 diverse practice sites throughout the United States. We utilized
the CORI database to analyze the date-specific occurrence of colonoscopy, as well as the
colonoscopic diagnoses of CD and UC. Time trends were analyzed by autocorrelation, linear, and
RESULTS:Between January 2000 and December 2003, the number of colonoscopies increased 4.1-fold. The
proportion of colonoscopies with a CD diagnosis fell by 28%, and the proportion of colonoscopies
with a UC diagnosis fell by 50%. The occurrence of neither CD nor UC was shaped by any clear-cut
seasonal periodicity. However, the trends of the two diseases revealed strikingly similar patterns
with four resembling peaks superimposed on their monthly fluctuations.
CONCLUSIONS: Endoscopic diagnosis of IBD is unaffected by any seasonal variation. The decline in the diagnostic
rate of colonic IBD may reflect a relative increase in the utilization of colonoscopy for colon cancer
screening. The similarity in the monthly fluctuations of both IBD suggests that their incidence or
flare-ups may be influenced by identical exogenous risk factors.
(Am J Gastroenterol 2005;100:2233–2238)
Conflicting data have been reported about the seasonal vari-
ation of inflammatory bowel diseases (IBD). While some
authors have reported an increased occurrence of flare-ups
during winter or spring and a decreased occurrence during
summer, others have failed to observe such seasonal varia-
tions in their patient populations (1–11). The American So-
ciety for Gastrointestinal Endoscopy (ASGE) initiated the
Clinical Outcomes Research Initiative (CORI) to develop
a database of endoscopic procedures (12). The database
has been designed to store records from gastrointestinal en-
doscopy procedures that reflect current endoscopic practice
among a diverse sample of gastroenterologists distributed
opportunity to revisit the issue of seasonality in the temporal
trends of IBD and analyze the possibility of its seasonal vari-
ation in yet another type of dataset. The aim of the present
tions in the diagnosis of Crohn’s disease (CD) and ulcerative
endoscopic procedures and diagnoses, also offered the pos-
sibility to separatelyfollow the procedural versus diagnostic
IBD stemmed from an underlying pattern in the performance
database formed in 1995 in collaboration with the ASGE.
The CORI database collects endoscopy reports containing
the standard reporting elements recommended by the ASGE.
Reports are entered by participating gastroenterologists and
then submitted electronically on a weekly basis to the CORI
headquarters in Portland, Oregon. Prior to electronic submis-
sion, reports are stripped of information that could compro-
mise the confidentiality of patient, physician, or participat-
ing institution. Data objects are then pooled for analysis. The
the population in the United States. The database has grown
into one of the largest of its kind, recording endoscopic pro-
cedures from 73 sites in 25 states. Fifty-three percent of re-
ports are derived from community-based practice, 29% from
academic medical centers, and 18% from Veterans Affairs
database to be considered in the present analysis. The first
2234 Auslander et al.
group comprised the entirety of all colonoscopies performed
between January 1, 2001 and December 31, 2003 and asso-
ciated with any type of indication or diagnosis. The second
and third group comprised all colonoscopies associated with
a diagnosis of either CD or UC. The last two groups were
further broken down by the indication for endoscopy, that is,
surveillance for cancer in known IBD versus other reasons
for endoscopy. In a subsequent sensitivity analysis to test the
robustness of the trends, data extraction was limited to those
practice sites, which had continuously contributed data to
the CORI repository throughout the entire 4-yr study period.
Lastly, we accumulated separately the entirety of all flexible
sigmoidoscopies, as well as those flexible sigmoidoscopies
associated with a diagnosis of CD or UC.
Diagnostic rates were calculated as monthly proportional
fractions of colonoscopies with a diagnosis of CD or UC
per 100 colonoscopies with any diagnosis. The rates of CD
or UC diagnosed by flexible sigmoidoscopy were similarly
doscopies. Standard techniques of time series analysis were
used to assess the trends for any underlying cyclic variations
(13). The time trends of colonoscopy and diagnostic rates of
CD or UC were first approximated by linear and polynomial
or fall in the proportional rates, the time-dependent linear or
polynomial trend was subtracted from each actual data point.
The trend-adjusted rates were subsequently smoothened by
a 3-month moving average. In a final step, the unadjusted,
as well as the adjusted trends were subjected to autocorrela-
tion with a step-wise increase of the time lag by 1 month. A
first-order autocorrelation corresponds to an ordinary Pear-
son correlation of the series with itself, shifted by 1 month.
Similarly, correlating the series with itself shifted by 2, 3,
4 months generates 2nd, 3rd, 4th-order autocorrelations, re-
spectively. As the monthly shift progresses, the correlation
coefficients of a series with seasonal periodicity change in a
used similarly to compare the two different time series of CD
of the SAS Institute in Cary, North Carolina, was used to test
various seasonal ARIMA models of the time trends of UC
cluded because of incomplete computer records or because
patients were younger than 20 yr. A total of 373,000 colono-
scopies performed between January 1, 2000 and December
Table 1 lists the number of colonoscopies for each individual
year, as well as the number of cases with a diagnosis of CD
or UC. Figure 1 shows the time trends of monthly colono-
scopies and cases with IBD. The rise in the monthly number
of colonoscopies was best approximated by a second-degree
Disease (CD) or Ulcerative Colitis (UC)
Year COL CDUC
adjusting the colonoscopy trends for their underlying poly-
nomial rise, no seasonal or any other cyclical pattern could
be discerned. The lack of seasonality applied similarly to
subgroups of colonoscopies performed in patients with sus-
pected or established disease. The rise in the monthly num-
0.820 for CD (p < 0.001) and R = 0.698 for UC (p < 0.001).
Although the absolute numbers of endoscopic diagnoses
of both types of IBD increased during the observation pe-
riod (Table 1 and Fig. 1), their proportional rates per 100
colonoscopies and month decreased. Comparing the first
and last value, the CD rate fell by 28% (R = 0.551, p <
0.001), and the UC rate fell by 50% (R = 0.647, p < 0.001).
Figure 2 displays the autocorrelation function of the unad-
justed rates of UC and CD. No distinctive correlation at 12
or 24 months could be discerned. The seasonal ARIMA of
both diseases failed to revel any significant seasonal model
parameters. After adjustment for their underlying time trend
cyclical pattern during the 4-yr time period. However, four
irregular peaks that occurred in September 2000, May 2001,
February 2002, and November 2002 were super-imposed on
the frequency curve of Crohn’s disease. Four peaks of simi-
lar shape, but shifted by 1–2 months toward the future, were
also noted in the time trend of UC (Fig. 3). The peaks rose
5–34% above the slope-corrected, baseline levels of the two
IBD. The overall correlation between the temporal patterns
of CD and the lag-corrected pattern of UC was R = 0.346,
p = 0.016.
To test the robustness of our results, we subjected them
to three additional types of confirmatory analyses. In a first
analysis, the data were limited to only those practice sites,
which contributed data throughout the entire 4-yr study pe-
riod. Of the original 73 practice sites, 36 sites contributed
data continuously and regularly to the CORI repository. The
data from these 36 “stable-practice” sites amounted to 75%
of all colonoscopies and to 79% of both IBD diagnoses. The
“stable-practice” time trends were almost identical to those
of the overall data set. They showed a significant increase in
the number of colonoscopies, as well as a significant decline
in the proportional rates of colonoscopies associated with
the two types of IBD. The analysis limited to stable practice
sites also revealed the same four peaks as shown in Figure 3
without any underlying seasonal or other cyclical pattern.
Seasonal Variations of IBD2235
Crohn's disease (CD) & ulcerative colitis (UC)
Figure 1. Time trends of colonoscopy (COL) and endoscopic diagnoses of Crohn’s disease (CD) and ulcerative colitis.
In a second confirmatory study, colonoscopies associated
with inflammatory bowel were analyzed by the indication for
endoscopy. Of all colonoscopies associated with CD, 47%
were done for surveillance in known disease, whereas 53%
were done for other reasons. In cases with UC, the corre-
eral, surveillance and nonsurveillance colonoscopies were
characterized by similar declining time trends and the ab-
sence of any seasonal variations. The trend-adjusted colono-
Figure 2. Autocorrelation function of ulcerative colitis (left) and
Crohn’s disease. The bars, and lines represent the serial correlation
coefficients and their standard errors, respectively.
scopies for surveillance and nonsurveillance in UC revealed
a similar set of four peaks as the entirety of all colonoscopies
associated with UC (Fig. 4). The correlation between the
two indications for colonoscopy in UC was R = 0.485, p <
0.001. With a lower overall number of colonoscopies associ-
ated with CD, a similar analysis for CD as for UC failed to
reach statistical power.
In a third confirmatory study, we analyzed the time trends
of CD and UC diagnosed through flexible sigmoidoscopy.
Between January 2000 and December 2003 the total number
of flexible sigmoidoscopies per month declined slightly with
doscopies associated with a diagnosis of CD and UC remain
largely unchanged (Fig. 5). No seasonal variation could be
were too small for a more refined analysis to reliably assess
other temporal trends.
The present analysis utilized the CORI database of endo-
scopic procedures to investigate the temporal trends and sea-
The time trends of both diseases in the CORI database were
characterized by a steady decline during the 4-yr observa-
tion period. Although the occurrence of neither CD nor ul-
cerative was shaped by any clear-cut seasonal changes, both
diagnoses revealed a strikingly similar fluctuations in their
The CORI database offers several unique features that
render it particularly suitable to study the seasonal patterns
of gastrointestinal disease. Foremost, the database has been
a variety of gastroenterology centers distributed throughout
the United States. The data are, therefore, less likely to be
2236Auslander et al.
Rate per 100 colonoscopies
Figure 3. Time trends of the proportional rate of Crohn’s disease and ulcerative colitis per month and 100 colonoscopies. The crude rates
were adjusted for an underlying linear trend and smoothened by 3-month moving average.
previous analyses, the database contains a much larger num-
ber of patients with IBD. Because of the large number of
cases in the database, the time trends are less likely to be in-
a monthly basis with sufficient statistical power. Lastly, be-
cause the diagnoses are established through colonoscopy, the
infectious or functional diarrhea, irritable bowel syndrome,
and many other gastrointestinal diagnoses that can mimic
Our analysis shows an overall decline in the diagnostic
rate, that is, the proportion of colonoscopies associated with
Ulcerative colitis per 100 colonoscopies
Figure 4. Proportional rates of ulcerative colitis per month and 100 colonoscopies, broken down by endoscopy indication. The crude rates
were adjusted for an underlying linear trend and smoothened by 3-month moving average.
a diagnosis of IBD. The decline may reflect a decreasing
incidence of IBD or a relative increase in the utilization of
cal databases in Europe and the United States reveal a stable
time trend or moderate decline during the past two decades.
The occurrence of CD from many Western countries is char-
(14). The trends of both types of IBD are relatively smooth
and appear drawn out over many years. In contradistinction,
the downward trend of IBD within the CORI database oc-
Seasonal Variations of IBD2237
Flexible sigmoidoscopies (FS)
Crohn's disease (CD) & ulcerative colitis (UC)
Figure 5. Time trends of all flexible sigmoidoscopies (FS) and diagnoses of Crohn’s disease (CD) and ulcerative colitis (UC) using flexible
leads one to speculate that it reflects primarily a dispropor-
fall in the diagnosis of IBD. In recent years, there have been
several campaigns to promote screening for colorectal can-
cer and several studies have provided convincing evidence
for the superior benefit of colonoscopy as compared to other
to have increased the number of screening colonoscopies.
No characteristic seasonal variations were detected in the
time trends of endoscopic diagnosis of UC or CD. Similarly
to the present analysis, three previous analyses of hospital
admissions and physician visits failed to reveal any striking
seasonal variation in the occurrence of either CD or UC (2,
6, 11). In the past, seasonal variations were found mostly in
hospitals or endoscopy centers, whereas authors using larger
datasets have been unable to confirm any clear-cut periodic-
ity in the occurrence of the two diseases. In contradistinction
and nongastrointestinal diseases exhibit a seasonal variation,
characterized by a summer low and a winter high that affect
almost all morbidity parameters (19–23). As a matter of fact,
seasonal variation is so common among all other diagnoses
that in statistics where IBD is listed as secondary diagno-
sis, the seasonal variations of the primary diagnoses tend to
mask the lack of seasonality in the occurrence of IBD (24).
Until now the reasons for the appearance of seasonal vari-
ations in the time trends of other diagnoses have remained
unknown. The lack of seasonality in IBD is equally mysteri-
ous and noteworthy for its absence in comparison with most
data from several consecutive years and lumped together
monthly data from different years to increase the statistical
power of their study. This type of trend analysis obscures the
monthly variations of IBD. Although in the CORI database
the occurrence of neither CD nor UC was influenced by any
did reveal some striking similarity of their monthly fluctua-
tions. These patterns suggest that incidence or flare-ups of
IBD are influenced by short-term variations of one or mul-
tiple exogenous risk factors. Similar patterns were seen in
our previous study of hospital admissions among Medicare
patients (6). In the previous as well as in the present analysis,
the trend in CD appeared to precede the trend in UC by 1–2
months. At the present time, the nature of the mechanisms,
which underlie these fluctuations, remains speculative. Vari-
29). In future studies it would be worthwhile to correlate the
temporal changes of IBD with those of gastrointestinal or
In conclusion, the present analysis reveals that the endo-
scopic diagnoses of neither CD nor UC are affected by any
underlying seasonal variation. The decline in the diagnostic
rate of both diagnoses from 2000 to 2003 most likely reflect
a relative increase in the number of screening colonoscopies
as compared to diagnostic colonoscopies in symptomatic pa-
tients. The time trends of the two diagnoses show a striking
parallelism in their short-term variability that suggests that
flare-ups of both diseases may be modulated by similar envi-
ronmental risk factors.
The authors are indebted to Nora Mattek and Jennifer Holub
for their help in retrieving the CORI data.
The study was supported by NIDDK grants 2-U01-
DK057132-06A1 and 5-R33-DK061778-03. In addition, the
2238Auslander et al.
Clinical Outcomes Research Initiative (CORI) has received
support from the following entities to support the infras-
tructure of its practice-based network: AstraZeneca, Bard
International, Pentax USA, ProVation, Endosoft, GIVEN
Imaging, and Ethicon. The commercial entities have no in-
volvement in this research.
M.Sc., Portland VA Medical Center-P3-GI, 3710 SW US Veterans
Hospital Road, Portland, OR 97239.
Received January 25, 2005; accepted April 28, 2005.
2. Don BAC, Goldacre MJ. Absence of seasonality in emer-
gency hospital admissions for inflammatory bowel disease.
3. Riley SA, Mani V, Goodman MJ, et al. Why do patients
with ulcerative colitis relapse? Gut 1990;31(2):179–83.
cerative proctocolitis. J R Coll Surg Edinb 1986;31(3):158–
5. Tysk C, Jarnerot G. Seasonal variation in exacerbations of
ulcerative colitis. Scand J Gastroenterol 1993;28(1):95–6.
6. Sonnenberg A, Jacobsen SJ, Wasserman IH. Periodicity of
hospital admissions for inflammatory bowel disease. Am J
of Crohn’s disease. Scand J Gastroenterol 1996;31(1):79–
sonal variation in exacerbations of ulcerative colitis. Hep-
9. Moum B, Aadland E, Ekbom A, et al. Seasonal variations
in the onset of ulcerative colitis. Gut 1996;38(3):376–8.
10. Vergara M, Fraga X, Casellas F, et al. Seasonal influence
in exacerbations of inflammatory bowel disease. Rev Esp
Enferm Dig 1997;89(5):357–66.
12. Lieberman D, Fennerty MB, Morris CD, et al. Endo-
scopic evaluation of patients with dyspepsia: Results from
the national endoscopic data repository. Gastroenterology
13. Box GEP, Jenkins GM, Reinsel GC. Time series analysis.
Forecasting and control. 3rd Ed. New Jersey: Prentice Hall,
14. Delco F, Sonnenberg A. Commonalities in the time trends
15. Cram P, Fendrick AM, Inadomi J, et al. The impact of a
celebrity promotional campaign on the use of colon can-
cer screening: The Katie Couric effect. Arch Intern Med
16. Harewood GC, Lieberman DA. Colonoscopy practice pat-
terns since introduction of medicare coverage for average-
risk screening. Clin Gastroenterol Hepatol 2004;2(1):72–
17. Lieberman DA, Weiss DG, Bond JH, et al. Use of
colonoscopy to screen asymptomatic adults for colorectal
cancer. Veterans Affairs Cooperative Study Group 380. N
Engl J Med 2000;343(3):162–8.
18. Sonnenberg A, Delco F, Inadomi JM. Cost-effectiveness of
colonoscopy in screening for colorectal cancer. Ann Intern
19. Sonnenberg A, Wasserman IH, Jacobsen SJ. Monthly vari-
ation of hospital admission and mortality of peptic ulcer
disease: A reappraisal of ulcer periodicity. Gastroenterol-
20. Jacobsen SJ, Goldberg J, Miles TP, et al. Seasonal variation
in the incidence of hip fracture among white persons aged
65 years and older in the United States, 1984–1987. Am J
21. Douglas AS, Dunnigan MG, Allan TM, et al. Seasonal vari-
ation in coronary heart disease in Scotland. J Epidemiol
Community Health 1995;49(6):575–82.
22. Vilkman S, Keistinen T, Tuuponen T, et al. Seasonal vari-
ation in hospital admissions for chronic obstructive pul-
in mortality in Scotland. Int J Epidemiol 2000;29(2):274–9.
24. Sonnenberg A, Wasserman IH. Epidemiology of inflamma-
tory bowel disease among U.S. military veterans. Gastroen-
25. Banks BM, Bargen JA. Relapses in chronic ulcera-
tive colitis—causes and prevention. Arch Intern Med
26. Mee AS, Jewell DP. Factors inducing relapse in inflamma-
tory bowel disease. Br Med J 1978;2(6140):801–2.
27. Gebhard RL, Greenberg HB, Singh N, et al. Acute viral
enteritis and exacerbations of inflammatory bowel disease.
28. Kangro HO, Chong SK, Hardiman A, et al. A prospective
study of viral and mycoplasma infections in chronic inflam-
matory bowel disease. Gastroenterology 1990;98(3):549–
inflammatory bowel disease. Experience from the Milwau-
kee outbreak. Dig Dis Sci 1997;42(8):1580–6.