Professional Documents
Culture Documents
Prevalence and Determinants of Exocrine Pancreatic Insufficiency Among Older Adults Results of A Population Based Study
Prevalence and Determinants of Exocrine Pancreatic Insufficiency Among Older Adults Results of A Population Based Study
To cite this article: Dietrich Rothenbacher, Michael Löw, Philip D. Hardt, Hans-Ulrich Klör,
Hartwig Ziegler & Hermann Brenner (2005) Prevalence and determinants of exocrine pancreatic
insufficiency among older adults: Results of a population-based study, Scandinavian Journal of
Gastroenterology, 40:6, 697-704, DOI: 10.1080/00365520510023116
ORIGINAL ARTICLE
Medical Department, Justus-Liebig University Giessen, Giessen, Germany, and 3Gesundheitsberichterstattung Saarland /
Krebsregister, Saarbrücken, Germany
Abstract
Objective. The prevalence and main determinants of exocrine pancreatic insufficiency were investigated in a large
population-based sample of older adults by measuring pancreatic elastase-1 in stool. Material and methods. The study
comprised 914 participants aged 50 to 75 years recruited by their general practitioner during a general health examination.
All participants and their physicians were asked to fill out a standardized questionnaire which contained information on
socio-demographic and lifestyle factors as well as medical history. Native stool was examined for pancreatic elastase-1 with a
commercially available ELISA (ScheBo† Tech, Giessen, Germany). Results. Overall, 524 women and 390 men aged 50 to
75 years (mean age 61.9 years) were included in the analysis. In total, 105 (11.5%) of the 914 subjects showed signs of
exocrine pancreatic insufficiency (EPI) with 5/200 m g elastase-1/g stool, and 47 (5.1%) subjects showed signs of a severe
exocrine pancreatic insufficiency (SEPI, B/100 m g elastase-1/g stool). There was a clear increase in EPI with age. Patients
taking angiotensin-converting enzyme (ACE) inhibitors had a lower prevalence than subjects without this medication; these
associations persisted after adjustment for covariates. Conclusions. Prevalence of EPI increases with age and seems to be
tentatively higher in men than in women. However, smoking seems to be an independent risk factor for EPI and SEPI
whereas ACE-inhibitor intake might be a protective factor. The latter finding may even point to new options in the
treatment of chronic pancreatitis.
Key Words: Elastase-1, exocrine pancreatic insufficiency, observational study, prevalence, risk factors
Correspondence: Dietrich Rothenbacher, MD, MPH, Department of Epidemiology, German Centre for Research on Ageing, University of Heidelberg,
Bergheimer Str. 20, DE-69115 Heidelberg, Germany. Tel: /49 62 2154 8146. Fax: /49 62 2154 8142. E-mail: rothenbacher@dzfa.uni-heidelberg.de
examined the main determinants using a multi- antibodies binding to two distinct epitopes of this
variate analysis strategy. enzyme. The following cut-off points have been
established in numerous clinical trials (for an over-
view see Table VI): normal levels: /200 m g
Material and methods elastase /1/g stool, moderate EPI: 100 /200 mg
Study design and study population elastase /1/g stool, severe exocrine pancreatic insuf-
ficiency (SEPI): B/100 mg elastase-1/g stool.
This study was carried out in the context of the
baseline examination of a large-scale, population-
based, cohort study (ESTHER /‘‘Epidemiologische Statistical methods
Studie zu Chancen der Verhütung, Früherkennung
The subsample with faecal pancreatic elastase-1
und optimierten Therapie chronischer Erkrankun-
measurements and the whole cohort of the
gen in der älteren Bevölkerung’’) with the aim of
ESTHER study were described with respect to the
evaluating new approaches to the prevention and
early detection of chronic diseases among older main socio-demographic characteristics. Prevalence
adults. of EPI and SEPI was determined according to age
Downloaded by [Ruhr-Universitaet Bochum] at 09:27 29 November 2017
The ESTHER study, which has been approved by and gender. In addition, the association of EPI
local and regional ethics committees, is conducted in (EPI and SEPI) with age and gender and the
Saarland, a state located in South-West Germany. following potential determinants was assessed calcu-
Between July 2000 and December 2002, 9958 lating a x2 statistic: duration of school education
inhabitants of Saarland, aged 50 to 75 years, were ( 5/9 years, /9 years), self-reported alcohol con-
recruited by their general practitioner during a sumption (none, B/60 g/week, 60 /140 g/week,
general health examination offered biennially to /140 g/week), smoking status (never, former,
older adults in Germany. current), body mass index (B/25 kg/m2, 25 /27.5
kg/m2, /27.5 /30 kg/m2, /30 kg/m2), history of
diabetes mellitus (no, yes (no-insulin treatment), yes
Data collection (insulin treatment)), history of hyperlipidaemia (no,
All participants and their physicians were asked to yes), history of gallstones (no, yes), history of
complete a standardized questionnaire containing cholecystectomy (no, yes). In addition, the regular
information on socio-demographic and lifestyle fac- intake of the following medications for which poten-
tors as well as medical history. The physicians tial side effects on exocrine pancreatic function were
provided information on medication of each patient considered was evaluated: antidiuretics (no, yes),
and filled in brand names, doses and dose instruc- angiotensin-converting enzyme (ACE) inhibitors
tions of currently prescribed drugs in a structured (no, yes), non-steroidal anti-inflammatory drugs
questionnaire. Drugs were attributed to preparations (NSAIDs) (no, yes), statins (no, yes). If the expected
available on the German market and linked to the cell number was less than 5, then Fisher’s exact test
‘‘Gelbe Liste-Pharmindex’’ (MediMedia, Neu-Isen- was used.
burg, Germany) providing the active compounds We then used multivariate logistic regression with
and anatomic therapeutic chemical (ATC) codes [7]. a backward variable selection procedure to identify
In addition, patients were asked to mail a stool the independent association of various factors with
sample to the study centre for laboratory analyses EPI and SEPI. All the variables listed above were
which was mainly used to extract DNA from stool. considered in the initial model. We then eliminated
The current study focuses on a subsample of 914 step by step all variables that did not contribute to
participants for whom a sample of native, solid stool the model in a statistically significant way (p /0.1).
from the original probe was available after DNA Odds ratios (OR) and 95% confidence intervals (CI)
extraction; the sample was stored at /708C and were calculated for all variables included in the final
later examined for pancreatic elastase-1. This sub- model. The analyses were carried out with the SAS
sample did not differ from the whole study popula- statistical software package (version 8.02).
tion with respect to the distribution of gender and
age and can be regarded as a (quasi-) random sample Results
as it was only determined by the amount of stool in
the container that was sent to the laboratory. Overall, 914 participants aged 50 to 75 years were
included in this analysis (Table I) and the subsample
with faecal pancreatic elastase-1 measurements was
Laboratory analysis
similar to the overall ESTHER study population.
We used a commercially available ELISA (ScheBo† The mean age of the study sample was 61.9 (SD
Tech, Giessen, Germany) using two monoclonal 6.67) years. 57.3% were female, 71.1% had a school
Exocrine pancreatic insufficiency among older adults 699
Table I. Main socio-demographic characteristics of the study
A history of diabetes was reported by 11.4% and a
population with faecal elastase-1 test result and of the whole
ESTHER cohort. history of hyperlipidaemia by 45.8%.
Table II shows the prevalence of EPI (defined as
N (%) Study N (%) ESTHER 5/200 mg elastase-1/g stool) and the prevalence of
cohort cohort SEPI ( B/100 m g elastase-1/g stool) according to age
Age (years) (Ntotal /914) (Noverall /9958) and gender. Overall, 11.5% of the 914 subjects
Mean (SD) 61.9 (6.67) 62.1 (6.63)
showed signs of EPI and 5.1% of SEPI indicated
by faecal elastase-1 measurements. The prevalence
Age
50 /54 years 167 (18.3%) 1712 (17.2%) was tentatively higher in men than in women (EPI
55 /59 years 150 (16.4%) 1689 (17.0%) 13.6% versus 9.9%, p /0.08, SEPI 5.9% versus
60 /64 years 261 (28.6%) 2704 (27.2%) 4.6%, p/0.36).
65 /69 years 194 (21.2%) 2278 (22.9%) There was a clear increase in EPI with age, from
70 /75 years 142 (15.5%) 1564 (15.7%)
6.0% in the 50 /54 years age group to 15.5% in the
Gender
65 /69 years age group, and 13.4% in the 70 /75
Female 524 (57.3%) 5471 (54.9%)
Male 390 (42.7%) 4487 (45.1%) years age group (p/0.005 after adjustment for
Downloaded by [Ruhr-Universitaet Bochum] at 09:27 29 November 2017
School education
5/9 years 78/666 (11.8%) 32/666 (4.8%)
/9 years 21/220 (9.6%) 0.4 12/220 (5.5%) 0.7
Alcohol consumption
None 34/309 (11.0%) 19/309 (6.2%)
B/60 g/week 17/229 (7.4%) 7/229 (3.1%)
60 /140 g/week 22/189 (11.6%) 9/189 (4.8%)
/140 g/week 15/112 (13.4%) 0.3 6/112 (5.4%) 0.4
Smoking status (cigarettes)
Never 45/483 (9.3%) 20/483 (4.1%)
Former 0/257 (11.7%) 12/257 (4.7%)
Current 24/153 (15.7%) 0.09 13/153 (8.5%) 0.09
Body mass index
Downloaded by [Ruhr-Universitaet Bochum] at 09:27 29 November 2017
prevalence of EPI and SEPI in the bivariate analysis. The bivariate associations between various medi-
A history of hyperlipidaemia was associated with cations that were suspected of being potentially
both a lower prevalence of EPI and of SEPI (the associated with pancreatitis and EPI and SEPI are
latter was statistically significant). listed in Table IV. Notably, treatment with ACE
Table IV. Association of intake of various medications with exocrine pancreatic insufficiency (EPI, B/200 m g elastase /1/g stool) and severe
exocrine pancreatic insufficiency (SEPI, B/100 m g elastase /1/g stool).
Antidiuretics
No 93/827 (11.3%) 43/827 (5.2%)
Yes 12/87 (13.8%) 0.5 4/87 (4.6%) 0.8
ACE inhibitors
No 90/724 (12.4%) 44/724 (6.1%)
Yes 15/190 (7.9%) 0.08 3/190 (1.6%) 0.01
NSAIDs
No 101/893 (11.3%) 47/893 (5.3%)
Yes 4/21 (19.1%) 0.1* 0/21 (0%) 0.6*
Statins
No 98/843 (11.6%) 45/843 (5.3%)
Yes 7/71 (9.9%) 0.7 2/71 (2.8%) 0.5*
inhibitors showed a tentative inverse association with smoking. In addition, ACE-inhibitor intake ap-
prevalence of EPI (7.9% versus 12.4%, p /0.08) peared to be a preventive factor and may point to
and a significant inverse association with SEPI an interesting new treatment option for chronic
(1.6% versus 6.1%, p /0.01). pancreatitis.
Table V shows the results of the multivariate So far, there are few data on the prevalence of EPI
logistic regression after taking all the aforementioned at the population-based level. Estimated incidence
variables into the initial model and applying a rates of chronic pancreatitis in adulthood ranged
backwards selection strategy. As the final model from 1.6 new cases per year per 100,000 subjects in
shows, increasing age was a clear determinant for Switzerland to 23 per year per 100,000 in Finland
EPI and SEPI as already seen in the bivariate [8], and a hospital-based study from Germany
analysis. Current smokers also showed a statistically estimated a prevalence of 6.4 cases per 100,000
significant increased OR for EPI (OR 2.06 (95% CI inhabitants [9]. A nationwide clinical survey from
1.19 /3.57)) and SEPI (OR 2.39 (95% CI 1.14 / Japan estimated a prevalence of 29 cases per 100,000
5.02)). Subjects who took ACE inhibitors (according [10] with a higher rate in men compared to women.
to the physicians’ information) showed a consider- The estimates of the annual incidence would result
ably lower odds for EPI compared to other subjects in an estimated cumulative prevalence between 1
Downloaded by [Ruhr-Universitaet Bochum] at 09:27 29 November 2017
(OR 0.55 (95% CI 0.31 /0.99)). The inverse asso- and 20 cases per thousand subjects (1 to 2%) in a
ciation with ACE-inhibitor intake was even more group of subjects aged 70 years.
pronounced for SEPI (OR 0.23 (95% CI 0.07 / However, all previous estimates had been based on
0.77)). No other variables met the criteria for clinical studies (using invasive methods that can
inclusion in both models. hardly be applied in a population-based sample) or
autopsy studies [6,8 /12]. Both may be difficult to
generalize as they are based on clinical subjects with
Discussion
diagnosed disease or fatal diseases, and the relation-
This population-based study of 914 asymptomatic ship to the source population is usually difficult to
subjects aged 50 to 75 years who had undergone a assess and essentially unknown. In contrast, our
routine health screening examination suggests that study is the first one conducted on a population-
the prevalence of pancreatic insufficiency as deter- based sample, and we have no indication that it is
mined by means of faecal elastase-1 measurements not representative of the general population in this
might be higher in the general population than age group.
previously estimated. The prevalence increased The faecal elastase-1 test seems to be a useful and
with age, was tentatively higher in men than in reliable non-invasive test to describe exocrine pan-
women and was positively associated with cigarette creatic function with a sufficient test performance in
the context of an epidemiological study (for a
Table V. Main determinants of pancreatic insufficiency as deter-
summary of test performance, see Table VI [3/5,
mined by means of unconditional logistic regression. 13 /19]) and a good sensitivity for the detection of
moderate to severe chronic pancreatitis has been
Odds ratio (95% CI) reported. According to some studies, faecal elastase-
1, however, showed less sensitivity for mild to
Factors* For EPI$ For SEPI% moderate chronic pancreatitis [4,13,14,16,17]. As
Age a note of caution, most of the available studies have
50 /54 years 1Reference 1Reference been small and estimates prone to random variation.
55 /59 years 1.66 (0.70 /3.94) 1.84 (0.57 /6.00) In addition, heterogeneous patient groups had been
60 /64 years 2.79 (1.31 /5.92) 2.27 (0.78 /6.63) included in some of the studies.
65 /69 years 3.67 (1.70 /7.93) 3.59 (1.23 /10.53)
Based on the test characteristics as described in
70 */75 years 3.26 (1.42 /7.48) 3.01 (0.92 /9.85)
the work of Gullo et al. [15] or Stein et al. [18] (the
Smoking status (cigarettes)
two studies that included the highest numbers of
Never 1Reference 1Reference
Former 1.26 (0.78 /2.04) 1.18 (0.57 /2.44) both cases and control subjects), our estimated
Current 2.06 (1.19 /3.57) 2.39 (1.14 /5.02) prevalence of about 13.4% for EPI in the 70 /75
ACE-inhibitor intake years age group would correspond to a true pre-
No 1Reference 1Reference valence of 12.9% or 8.2%, respectively after correc-
Yes 0.55 (0.31 /0.99) 0.23 (0.07 /0.77) tion for misclassification. Furthermore, a positive
Abbreviations: EPI/exocrine pancreatic insufficiency; SEPI/
and negative predictive value of 74.0% and 96.6% or
severe exocrine pancreatic insufficiency. 58.9% and 99.6%, respectively could be derived.
*All variables in the model controlled for each other; $EPI (B/200 The corrected prevalences are still much higher than
m g elastase /1/g stool); %SEPI (B/100 m g elastase /1/g stool). the aforementioned estimates of 1 to 2% in a
Table VI. Test performance of faecal elastase-1 to determine exocrine pancreatic insufficiency (EPI) as described in various studies*.
702
Study population Comparison
D. Rothenbacher et al.
Authors [Ref.], country N (Gold standard) Results Comments
Amann et al. [13] 14 patients with CP (mild or moderate 7, Secretin-pancreozymin SE (mild or moderate) 43%, Very few subjects. Some
USA severe 7) test or ERCP or SE (severe) 100% controls suffer other
7 patients with non-pancreatic disorders leading pancreatic surgery SP 29% gastrointestinal disorders
to malabsorption
15 healthy controls
Downloaded by [Ruhr-Universitaet Bochum] at 09:27 29 November 2017
Dominguez-Munoz et al. [14] 20 patients with CP (mild 6, moderate or ERCP, ultrasonography, SE (mild) 0% Very few subjects. SE mild CP
Germany severe 14) CT SE (moderate/severe) 100% based on n/6
36 patients with non-pancreatic disease SP 83%
Gullo et al. [15] 44 patients with CP (mild 9, moderate 13, ERCP, ultrasonography, SE 77.3% Very good SP
Italy severe 22) medical history SE (severe/moderate) 91.4%
43 patients with non-pancreatic digestive disease SP 95.8%
53 healthy subjects
Hardt et al. [4] 210 patients with CP (mild 97, moderate 53, ERCP SE 45.3% Low SE for mild EPI Good SP
Germany severe 51) SP 75% for moderate and severe CP
12 patients without CP SE (moderate/severe) 57.4%
SP (moderate/severe) 86.4%
Lankisch et al. [16] 30 patients with EPI (mild 10, moderate or Secretin-pancreozymin SE 53% SE in mild CP low (based on
Germany severe 19) 30 patients without EPI test SE (severe) 82.7% n /10)
SP 94% Very good SP
Leodolter et al. [17] 33 patients with CP (mild or moderate 20, ERCP, ultrasonography, SE 50% Few subjects. SE in mild CP
Germany severe 13) CT SE (mild) 35%, lower
7 patients without CP SE (severe) 85%
SP 100%
Löser et al. [5] 26 patients with EPI (mild 13, severe 13) Secretin-caerulein test, SE 92.% Few subjects
Germany 25 patients with gastrointestinal disease ultrasonography, ERP SE (severe) 100% Good SP
SE (mild) 85%
SP 90%
Lüth et al. [3] 62 patients with EPI (mild 23, moderate 14, Secretin-caerulein test SE 84% SP in severe EPI better
Germany severe 25) SP 57%
65 patients without EPI SE (severe) 60%
SP (severe) 78%
Stein et al. [18] 63 patients with EPI Secretin-pancreozymin SE 96% SP 94% Very good SP
Germany 86 patients with cystic fibrosis test
Walkowiak et al. [19] 28 patients with cystic fibrosis Secretin-cholecystokinin SE 89.3% Very few subjects
Poland 55 children without gastrointestinal disease test SE (mild) 25%
SE (moderate/severe) 100%
SP 96.4%
Abbreviations: CP/chronic pancreatitis; ERCP/endoscopic retrograde cholangiopancreaticography, CT/computed tomography; SE/sensitivity; SP/specificity.
*Only studies with at least 30 subjects are considered.
Exocrine pancreatic insufficiency among older adults 703
group of subjects aged 70 years. If the accuracy of and a recent retrospective cohort study in a large
the test (as described in [15] and [18]) could be group of elderly patients did not show an increased
corroborated in even larger validation studies, then risk for pancreatitis associated with ACE-inhibitor
EPI may have been underestimated to some degree intake [30]. Meanwhile, animal data showed that
by many other studies. Our results are supported by ACE inhibitors alleviated chronic pancreatitis and
the findings of two unselected autopsy studies which fibrosis and ACE inhibitors were suggested as
suggested a prevalence of chronic pancreatitis of 6/ potential treatment for chronic pancreatitis [31].
12% [11,12]. Our data have delivered further evidence of the
Furthermore, we found a clear association of EPI potential beneficial effects of ACE inhibitors for
with age and a tentative higher prevalence of EPI in chronic pancreatitis in humans and suggest that the
men compared with women. This is in concordance anti-inflammatory effects of ACE inhibitors should
with previous reports [20]. In other, mainly clinical be investigated further as a potential component in
reports, excessive alcohol consumption has been the treatment of chronic pancreatitis.
reported as the main cause of chronic pancreatitis When looking on the results of the study the
[21,22]. In the current study, alcohol consumption following limitations have to be considered: as this
showed a U-shaped, non-monotonic relationship
Downloaded by [Ruhr-Universitaet Bochum] at 09:27 29 November 2017
1303 /11.
Gastroenterol 1998;33:327 /34.
[22] Dufour MC, Adamson MD. The epidemiology of alcohol-
/ /
40:325 /32.
creatic atrophy in relation to duration of chronic type 1
/
Epidemiology of pancreatic diseases in Luneburg County. A [25] Hardt PD, Hauenschild A, Nalop J, Marzeion AM, Jaeger C,
study in a defined German population. Pancreatology 2002; /
Teichmann J, et al. S2453112/S2453113 Study Group. High
2:469 /77.
/
prevalence of exocrine pancreatic insufficiency in diabetes
[10] Lin Y, Tamakoshi A, Matsuno S, Takeda K, Hayakawa T, mellitus. A multicenter study screening fecal elastase 1
Kitagawa M, et al. Nationwide epidemiological survey of concentrations in 1.021 diabetic patients. Pancreatology
chronic pancreatitis in Japan. J Gastroenterol 2000;35:136 / / /
2003;3:395 /402.
/ /
1995;90:1834 /7.
/ /
[29] Tilkemeier P, Thompson PD. Acute pancreatitis possibly
[15] Gullo L, Ventrucci M, Tomassetti P, Migliori M, Pezzilli R. related to enalapril. N Engl J Med 1988;318:1275 /6. / /
Fecal elastase-1 determination in chronic pancreatitis. Dig [30] Cheng RM, Mamdani M, Jackevicius CA, Tu K. Association
Dis Sci 1999;44:210 /3. / /
between ACE inhibitors and acute pancreatitis in the elderly.
[16] Lankisch PG, Schmid I, König H, Lehnick D, Knollmann R, Ann Pharmacother 2003;37:994 /8. / /
Lohr M, et al. Faecal elastase 1: not helpful in diagnosing [31] Kuno A, Yamada T, Masuda K, Ogawa K, Sogawa M,
chronic pancreatitis associated with mild to moderate exo- Nakamura S, et al. Angiotensin-converting enzyme inhibitor
crine pancreatic insufficiency. Gut 1998;42:551 /4. / /
attenuates pancreatic inflammation and fibrosis in male
[17] Leodolter A, Kahl S, Dominguez-Munoz JE, Gerards C, Wistar Bonn/Kobori rats. Gastroenterology 2003;124: / /