Occupation, Employment Status and Chronic

Vol. 25, No. 3
Printed In Great Britain
International Journal of Epidemiology
O International Epidemiotoglcal Association 1996
Occupation, Employment Status and
Chronic Inflammatory Bowel Disease
in Denmark
HENRIK B0GGILD,* FINN TUCHSEN" f AND ELSA 0RHEDE*
Beggild H (Department of Occupational Medicine, Aalborg Regional Hospital, Denmark), TQchsen F and Orhede E.
Occupation, employment status and chronic inflammatory bowel disease in Denmark. International Journal of Epidemiology 1996; 25: 630-637.
Background. Certain occupational groups have formerly been Identified as having higher risks of suffering from chronic
inflammatory bowel diseases. These were evaluated in an independent data set.
Methods. A cohort, comprising all 2 273 872 male and female Danes aged 20-59 years on 1 January 1981, and a cohort
of 2 387 620 men and women 1 January 1986 were followed up for hospitallzations due to chronic Inflammatory bowel
disease until 31 December 1990.
Results. From 1981 to 1990 6296 first time admissions occurred. The incidence Increased from 1981-1985 to 19861990. Of 363 male and 213 female occupational groups eight and five groups respectively had statistically significant
raised standardized hospitalizatlon ratios. Among 15 groups previously found to have significant odds ratios only female
office staff and health occupations were found to have statistically significant raised standardized hospitalization ratios.
Ratios for occupational groups with non-daytime work were not statistically significant. Compared to occupations without
sedentary work occupations with predominantly sedentary work had a standardized hospitalization ratio of 125
(95% confidence interval [95% Cl]: 116.9-133.1). Self-employed had low hospitalization rates, while'other salaried staff
and 'not economically active' had high rates.
Conclusions. We found no consistent pattern of occupations at increased risk except that sedentary work may Increase
the risk of attracting chronic inflammatory bowel disease.
Keywords: Crohn's disease, ulcerat'rve colitis, epidemiology, hospitalization
shiftwork7 may confer a risk of contracting inflammatory bowel disease. In the present study we took
advantage of the unique National Inpatient Register and
the record linkage facilities in Denmark to set up a cohort
study in order to test some of these hypotheses in an independent data set. In addition we have identified other
occupational groups which may have increased risks.
The incidence of Crohn's disease (CD) appears to rise
in most industrial communities,1 while ulcerative colitis
(UC) seems to have a smaller rise or an unaltered incidence. 2 The rise may not be due to change in diagnostic
procedures or changes in referral to hospitals.1 Although
both CD and UC show familial aggregation3 the sharp
rise suggests that an environmental aetiology is also
present and among others, several infectious agents
have been proposed.4 The chronic inflammatory bowel
diseases (CEBD) have been reported to be more numerous
among white collar workers.5 Recently Sonnenberg
identified 7 occupational groups at high risk and 8 at
low risk,6-7 and suggested that physically less demanding work, sedentary occupations,6 working in an airconditioned environment or with extended or irregular
MATERIALS AND METHODS
A cohort of all 2 273 872 men and women aged 20-59
on 1 January 1981 identified in the Central Population
Register (CPR) of Denmark was followed during the
10 years to 31 December 1990. All first time hospitalizations with CIBD as the principal diagnosis were
recorded. Another cohort, aged 20-59 on 1 January
1986 consisting of 2 387 620 men and women was
followed up for hospitalizations during 5 years. This
was achieved by the construction of the Occupational
Hospitalization Register (OHR) which links data from
the register-based population census of Denmark to the
* Department of Occupational Medicine, Aalborg Regional Hospital,
P O Box 561, DK-9100 Aalborg, Denmark.
• • Centre of Biochemical and Occupational Epidemiology, Denmark.
' National Inititute of Occupational Health, Department of
Occupational Medicine, Denmark.
630
OCCUPATION AND INFLAMMATORY BOWEL DISEASE
National Inpatient Register. Coding of occupations was
by means of the Employment Classification Module
(ECM) at Denmark's Central Bureau of Statistics.
The Central Population Register (CPR)
From the CPR, we included information on the personal
identification number (PIN) from which we extracted
information about gender and date and year of birth.
The CPR was used to identify men and women actually
living in Denmark on 1 January 1981. We restricted the
analyses to age groups with a high percentage of
economically active people (20-59 years old).
The Employment Classification Module (ECM)
Using various administrative registers, all men and
women in Denmark aged 16+ years are classified annually according to occupation, in the ECM. The ECM
holds information on economic activity, occupational
classification and industry. The occupational code is an
extended classification of the International Standard
Classification of Occupations (ISCO) 1958 version, and
includes the self-employed. The industrial code is a
national extension of the International Standard Classification of all Economic Activity (ISIC) 1968 version.
The National Inpatient Register (NIP)
The NIP holds data on all admissions to Danish hospitals and is updated each year. First time hospitalizations with CIBD (ICD-8: 563) as the main cause for
all men and women except foreigners treated in Danish
hospitals between 1981 and 1990 were included in this
study. Differentiation between CD and UC is not
possible because five digit codes for CIBD were not
included in the OHR before 1991.
The PIN-number from the NIP was used in the crosslinking procedure but excluded in the final file. The use
of the PIN enabled exclusion of readmissions in the
follow-up period.
Standardized hospitalization ratios (SHR) were
calculated for each 5 or 10 year period by dividing the
observed number of hospitalizations 1981-1990 in a
given sub-cohort defined by occupation and/or industry
code by an expected number. The expected number was
based on age-specific incidence rates for men or women
economically active on 1 January 1981 and on the age
distribution in the cohort. Individuals were no longer at
risk of being admitted to a hospital in Denmark from
the date of the first emigration or from the date of death.
From the date of such an event, these men and women
were excluded from the denominator in the period of
risk calculations. This information was collected for
10 years from the Migrations files and the Death
Certificate files in the CPR.
631
In order to analyse time trends and exclude possible
'by chance' findings, two cohorts based on occupation
in 1980 and 1985 were followed-up in two periods,
1981-1985 and 1986-1990 respectively.
Analysis
Assuming the findings of Sonnenberg7 were genuine
the same occupations should also have higher risks of
hospitalization in our data. We constructed groups of
occupational codes, comparable to the occupations for
which Sonnenberg7 found increased or decreased risks
of attracting CIBD. For 15 of the 17 statistically significant groups in Sonnenberg's data we constructed similar groups. The remaining two, sales representatives
and unemployed, are not coded separately in the ECM.
Sonnenberg7 suggested that people working in airconditioned rooms or on extended and irregular shiftwork had a higher risk of contracting CIBD. A Survey
of Living Conditions in Denmark in 1976 among 5166
respondents8 included information about non-daytime
work, and occupational groups with at least 20% giving
positive answers were included in the present study if
they could also be identified in the OHR. We were
unable to find information, which in a similar objective
manner, identified occupational groups working in an
air-conditioned environment.
Sedentary work may be a more likely common
explanation of high risk in many of the occupational
groups in Sonnenberg's studies.6'7 Data from a Danish
national survey on work environment and health in
19909 on 6000 Danish employees were used to identify
occupational groups in which more than 50% of the
respondents stated that they performed sedentary work
all day.
As the ECM held no information on social class,
employment status was used as a proxy measure. Due to
small numbers 'assisting spouses' were aggregated with
the group of self-employed.
Ninety-five per cent confidence limits (two-tailed)
were calculated assuming a binomial distribution
estimated by a Poisson distribution.
RESULTS
The 10-year cohort had a total of 6296 first time admissions to hospital with CIBD in 1981-1990. The total
number of cases was 2806 in 1981-1985 and 3490 in
1986-1990, an increase of 24% although the number of
people at risk had only increased by 5%. The age and
gender distributions in the two 5-year cohorts 19811985 and 1986-1990 are shown in Figure 1. For all age
groups and both genders we found a higher incidence in
the latter period. Hospitalization due to CIBD was more
632
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
10001D
•
900 -
-D Male 1981-1985
•
O---O
A
800 -
V
£—o
*
Male 1986-1990
Female 1981-1985
*
Female 1988-1990
\
—q
g
S.
f''
400 r
300 -
200
1
i
i
i
i
i
to 100 and they were not consistently in the same
direction as the OR.
In Table 3 groups with non-daytime jobs are listed.
Neither individual occupational groups nor aggregations among groups with predominantly night, afternoon, 24-hour or other forms of irregular working hours
had increased risks of hospitalizations for CIBD.
Occupations with a high proportion of sedentary
work are listed in Table 4. Of the five identified groups
in which more than 50% worked in a sedentary position
for the whole day two groups of lawyers and office staff
had statistically significant raised SHR. The aggregation of all five groups yielded an SHR of 111 which was
statistically significant. If compared with occupational
groups in which more than 75% never worked in
sedentary positions the SHR increased to 125 (95%
C I : 116.9-133.1) (groups not shown).
Figure 2 shows that employment status is related to
risk of CIBD so that statistically significant high risks
of first time admission with CIBD among both males
and females were found in 'not economically active'
and 'other salaried staff. A statistically significant low
risk was found in 'self-employed men' and in 'selfemployed women and assisting wives'.
20-24 25-29 30-34 35-39 40-*4 45-^9 50-54 55-59
Age group
FIGURE 1 Incidence of chronic inflammatory bowel disease per
100 000 person-years by 5-year age groups in 1981-1985 and
1986-1990
frequent among females than males. The incidence was
highest in the younger age groups. Among men an
additional but smaller peak in incidence was found in
the age groups of 45-54 years.
'Exposure' information included 363 occupational
codes for men and 213 for women all consisting of
more than 100 people. Of these, eight male and five
female occupational groups had statistically significant
elevated SHR, and two male and three female groups
had statistically significant reduced SHR (Table 1). The
groups that were statistically significant in the 10-year
cohort were identified in the 5-year cohorts (Table 1).
The risk estimates were inconsistent in the male group
of 'other unskilled workers' and in the female group
'other office staff which indicate that the finding is
either by chance or subject to bias.
Sonnenberg 7 found eight groups with an elevated risk
of attracting CIBD. In only one of these: 'female office
workers', did our result show a statistically significant
elevated SHR (Table 2). Among the nine groups with a
low risk our calculations did not reach significance for
any of the groups. The point estimates were all close
DISCUSSION
German social security data7 and mortality data for
England and Wales6 were formerly used to estimate a
possible association between occupation and CIBD.
The Danish Occupational Hospitalization data set may
be a good supplement to these sources. It has the
advantage that it is a follow-up study. This makes it
possible to eliminate or at least reduce secondary
healthy worker effect and readmissions in the follow-up
period.
Comparisons of up to 363 figures are likely to
produce statistically significant results and the 18
occupational groups with statistically significant high
or low SHR might be due to chance. By comparing the
two 5-year cohorts we got some indication of the
consistency in the results. Although the results were
not statistically significant they pointed in the same
direction suggesting that the results are not by chance.
We tested our material assuming we would find high
and low SHR in the same occupational groups as in
Sonnenberg's study on morbidity,7 as we were unable
to compare our data with the occupational coding in the
mortality study6 from England and Wales. In both the
morbidity7 and the mortality data6 bakers had increased
risk of suffering CD, while security personnel and
workers in construction both had a low risk. We found
an increased risk in self-employed bakers, but not in
633
OCCUPATION AND INFLAMMATORY BOWEL DISEASE
TABLE 1 Standardized hospitalization ratios (SHR) with 95% confidence intervals (Cl) for occupational groups. Denmark 1981-1990,
1981-1985 and 1986-1990. Men and women. Statistically significant results
Code
Occupation/industry
No.
Obs.
Exp.
Statistically significant high SHR, male
Lecturer at institute of higher education
42130
Book-keeper, cashier
43320
Office staff, others
44319
Wooden shipbuilding
45813
45852
Electrician, power station
Unskilled workers, others
46999
51000
Students
Pensioner
54O00
15 590
8700
22 432
492
1268
17464
20 760
41 286
42
28
61
4
7
48
64
143
29.3
17.3
46.1
1.0
2.5
35.0
45.1
69.3
143
162
132
415
276
137
142
206
103-194
108-235
101-170
113-1063
111-570
101-182
109-181
175-243
162*
179*
149*
572*
288
108
141
193*
121
160
115
396
395*
81
112
206*
Statijucally significant high SHR, female
44065
Assistant with general practitioner, etc.
44314
Office staff, trade, hotels and catering
44319
Office staff, others
44850
Telephone staff
54000
Pensioner
3636
29 509
4655
399
61 889
17
94
19
5
357
9.3
72.2
11.2
0.9
119.5
183
130
170
557
299
107-293
105-159
102-265
181-1301
269-332
126
121
195*
387
375*
182
107
78
342
356*
6465
8321
5
7
12.5
16.0
40
44
13-93
18-90
44
34
78
72
32 575
4162
23 063
46
3
42
70.6
10.5
57.8
65
29
73
48-87
6-84
52-98
84
47
84
53*
28*
67
Statistically significant low SHR, male
12920
Carpenter, joiner
Haulier
13300
Statistically significant low SHR, female
Assisting spouse, agriculture, horticulture, forestry
21000
44012
Medical laboratory staff
46073
Assistant in other day- or full-time care centres
SHR
95% Cl
1981-90
SHR in cohort SHR in cohort
1981-85
1986-90
' P < 0.05.
TABLE 2 Standardized hospitalization ratios (SHR) and 95% confidence intervals (Cl) for occupational groups formerly found to have a
statistically significant odds ratio.7Denmark 1981-1990. Men and women
Occupation/industry
Code
No.
Obs.
Exp.
SHR
95% Cl
189
35 932
2679
8091
89
5
18
0.4
74.4
5.4
16.5
0
120
93
109
97-147
30-216
65-172
Male occupations with high OR
Instrument makers
Electricians
Baken
Technical assistants
45940
43961,46850,45851-9
45774, 46774
44030
Female occupations with high OR
Office workers
Health occupations
Hairdressers
44311-9
43061-9, 44062-9, 46060
45590, 46590
190 553
113 282
7564
512
309
16
462.5
276.4
18.8
111
112
85
102-121
100-125
49-138
Male occupations with low OR
Bricklayers
Road construction
Unskilled workers in rock and stone
Unskilled workers
Transportation workers
Bricklayers' assistant
Security
45943
46700
46951-9
46xxx
46970-9
46953
46530
9587
951
25 449
284 283
71 145
6340
1493
21
2
42
545
134
14
2
19.7
1.8
50.5
571.2
141J
12.5
2.8
107
111
83
95
95
112
71
66-163
13-401
62-113
88-104
80-112
66-189
9-258
Female occupations with low OR
Cleaning
46991,46521-9
47 815
115
110
105
87-126
0
0-925
634
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
TABLE 3 Standardized hospitalization ratios (SHR) and 95% confidence intervals (CI) for chronic inflammatory bowel disease in male
occupational groups with different work schedules. Denmark 1981-1990
Code
Occupation/industry
No.
Predominantly night and morning work
12102
Self-employed baker
45774
Skilled baker
Total
Obs.
Exp.
SHR
95% CI
1927
2055
3982
6
3.6
4
10
4.1
7.7
167
98
130
61-363
26-250
62-239
Late evening work
13310
Taxi operators and transport ictivities n.e.c.
13200
Self-employed in hotels, restaurants and cafes
45510
Cooks and waiters
Total
2829
4650
6658
14 137
6
8
12
26
5.2
8.6
13.2
27.0
115
93
91
96
42-252
40-183
47-159
63-142
24-hour services
Fishermen
11100
42340
Senior traffic staff
43341
Railway staff
43342
Bus, coach and road transport staff
44530
Rescue services, customs and excise, police
Total
4361
1396
1385
313
15 275
22 730
10
2
1
1
32
46
8.3
2.6
2.6
0.6
31.1
45.2
120
77
38
167
103
102
58-222
9-278
1-214
4-928
70-146
76-136
224
1493
71 145
9434
37 031
119 327
0
2
134
14
65
215
0.5
28
141.5
18.4
76.9
240.1
0
71
95
76
85
90
0-740
9-258
80-112
42-128
66-108
78-102
Other forms of irregular working hours
43330
Salaried data processing staff
46530
Doormen, ticket-checkers, security staff
46970
Drivers
46940
Production workers
47990
Wage-earners n.e.c
Total
TABLE 4 Standardized hospitalization ratios (SHR) and 95% confidence intervals (CI) for occupational groups with predominantly
sedentary work (male and female). Denmark 1981-1990
Occupation/industry
Statistician, datanomist, programmer etc.
Lawyer
Office manager, office staff etc.
Accountant, book-keeper, cashier etc.
Tailor, furrier, cutter etc.
Total
Code
No.
Obs.
Exp.
SHR
95% CI
42080, 43080, 44080
42120
42310,43310,44310
42320, 43320, 44320
43790, 45790, 46790
38 966
2434
250 903
45 198
24 075
361 576
83
11
669
110
48
921
78.5
5.2
584.9
104.7
56.2
829.5
106
212
114
105
85
111
85-131
106-379
106-124
87-127
64-113
104-118
skilled bakers. None of these results were however
statistically significant. In accordance with the
morbidity study7 we found that 'female office workers'
and 'female health occupations' had a statistically
significant increased risk.
We also tried to test the hypothesis that Sonnenberg7
proposed. We did not find any risk connected to shift
work. We were not able to test whether working in an
air-conditioned environment would confer a larger risk.
The statistically significant groups in our study did not
tend to suggest this. Both Sonnenberg's results6'7 and
our own statistically significant results suggested that
sedentary work could be a risk factor, and based on
interviews, we identified occupational groups with a
high proportion of people performing sedentary work
all day. We found that 2 out of 5 groups as well as the
aggregated group had statistically significant high risks.
This might suggest a connection that could be biologically plausible as the sedentary position is suspected to
increase the abdominal transit time, 10 giving potential
immunological substances more time to react in the
intestines.
635
OCCUPATION AND INFLAMMATORY BOWEL DISEASE
Men
Setf-employed and assisting spouse
Director (>20 employed)
Senior salaried staff
Leading salaried staff
Other salaried staff
Skilled workers
Unskilled workers
Other employed groups
Not economically active
100
SHR and 95% Cl
200
Women
Self-employed and assisting spouse
Director (>20 employed) *
Senior salaried staff
Leading salaried staff
Other salaried staff
Skilled workers
Unskilled workers
Other employed groups
Not economically active
100
SHR and 95% Cl
200
* Not shown (95 Cl: 0-925).
FIGURE 2 Standard hospilalization ratios (SHR) by employment status and 95%
confidence intervals (95% Cl) for men and women
We found an association between employment status
and the risk of first admission to hospital with CIBD.
This might suggest that environmental factors which
are not tied to specific occupations but rather to lifestyle (as diet) or general occupational conditions
like manual/non-manual work or income could be
aetiologically associated with CIBD.
As it is possible that sedentary work could be confounded with employment status, we compared 'senior
salaried staff occupations with sedentary work with the
whole group of 'senior salaried staff. This yielded an
SMR of 128 (95% Cl : 101.9-160.8), suggesting that
confounding is not responsible for the result.
The design of the cohort did not exclude people who
had already contracted the disease and people who
might have been hospitalized before the follow-up
period. This means that people who were already ill
might have changed to occupations with easier access
to lavatories. This would bias the results towards an
effect of indoor occupations. Pre-employment selection
could not be controlled but the selection in the followup period may be well controlled by the long follow-up
period.11
The main disadvantage of our data is that CD and UC
cannot be separated. Although the diseases have many
epidemiological and pathological similarities they seem
to have different trends in incidence progression and
thus they might have different aetiology. This disadvantage has, however, only a limited influence on the
risk estimate for occupational groups. Sonnenberg7 found
636
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
that the correlation of odds ratios for occupational
groups over 10 000 was 0.73 for men (/> < 0.0001) and
0.8 for women (P < 0.0001).
The incidence seems to rise (Figure 1) but as the
diseases cannot be differentiated it is not possible to
compare the figures with previously published data
suggesting different rates.l>2>12
The data are based on first-time admissions to hospitals in the follow-up period. Although diagnosis of
UC will not always be preceded by hospitalization it is
unlikely that diagnosis without hospitalization would
be biased towards certain occupations, even though this
might be the case for different social strata. As admission to hospital is free of charge in Denmark the bias is
probably not very large and since use of hospitals
is slightly more common in upper social classes a
bias would result in a larger difference between the
employment status groups than shown in Figure 2.
An explanation of the difference between our results
and Sonnenberg's can be differences in the occupational groupings. Our data were made comparable to
those of Sonnenberg by collapsing occupations into
employment status groups (excluding self-employed,
as they were not part of Sonnenberg's data). This
was done in a way which included all employment
status groups if not explicitly excluded by Sonnenberg.
In this way our groups might be broader defined
which in turn would tend to dilute an effect related to
a specific occupation. The possible exposures probably
do not differ between Danish and German work
environments.
The validity of the basic information is high. Age and
gender are part of the PIN and are carefully checked by
the authorities. The completeness and accuracy of these
data are shown by the fact that the matching of different
registers on PIN was 100% complete. Data on occupation are not routinely checked by the primary data
producers, but information from various sources is
compared and pooled in the ECM.
The validity of diagnosis was not examined, but
other diagnoses in the OHR have previously been
checked and validity has been found to be high.13 The
increased risk found in women occupied in the health
sector may be due to a referral bias earlier found in
nursing assistants. 14 For other occupational groups the
misclassification is likely to be non-differential. This
may bias the results towards unity.
Missing confirmation of findings from the other data
sets could either be due to low sensitivity in the OHR or
because the original findings were due to selection bias
or chance. The sensitivity of the present record linkage system based on individual records was found to
be high in other studies of various diagnosis. 15 ' 16 Low
sensitivity is thus an unlikely explanation of the missing confirmation of the hypothesis.
In conclusion we found no significant accumulation
of CIBD in any particular occupational groups except
female office workers, neither did we find any association between CIBD and groups working late evenings,
night work, early morning work or in industries providing 24-hour services. It seems, however, that sedentary
work can confer a risk, albeit small and not uniform in
our data. Since this hypothesis may be biologically
plausible and has been proposed in other, independent
datasets,6 further investigation is suggested.
ACKNOWLEDGEMENTS
This study was granted by the Danish Strategic
Environmental Research Programme and the set up of
the basic, linked register was supported by the Danish
Work Environment Fund, grant 1991-35. Thanks to Mr.
Otto Andersen, Danish Bureau of Statistics, who did the
basic linkage of the Occupational Hospitalization File.
REFERENCES
1
Munkholm P, Langholz E, Nielsen O H, Kreiner S, Binder V.
Incidence and prevalence of Crohn's disease in the county
of Copenhagen, 1962-87: a sixfold increase in incidence.
Scand J Gastrocntcrol 1992; 27: 609-14.
2
Langholz E, Munkholm P, Nielsen O H, Kreiner S, Binder V.
Incidence and prevalence of ulcerative colitis in Copenhagen county from 1962 to 1987. Scand J Gastroenterol
1991; 26: 1247-56.
3
Orholm M, Munkholm P, Langholz E, Nielsen O H, Sorensen
I A, Binder V. Familial occurrence of inflammatory bowel
disease. N Engl J Med 1991; 324: 84-88.
* Sandier R S, Golden A L. Epidemiology of Crohn's Disease.
J Clin Gastroenterol 1986; 8: 160-65
3
Keigley A, Millere D S, Hughes A O, Langman M J S. The
demographic and social characteristics of patients with
Crohn's disease in the Nottingham area. Scand J
Gastroenterol 1976; 11: 293-96.
'Sonnenberg A. Occupational mortality of inflammatory bowel
disease. Digestion 1990; 46: 10-18.
7
Sonnenberg A. Occupational distribution of inflammatory bowel
disease among German employees. Gut 1990; 31: 1037-1040.
1
Hansen E J. The distribution of living conditions. Main results
from the welfare survey. Pan I. Theory, method and
summary. Copenhagen: Teknisk foiiag, 1978 (English
summary).
9
Nord-Larsen M, 0rhede E, Nielsen J, Burr H. National survey on
work environment and health 1990 - part I. Copenhagen:
Arbejdsmiljafondet, 1992 (In Danish).
l0
Garabrandt D H, Peters J M, Mack T M, Bernstein L. Job activity
and colon cancer risk. Am J Epidemiol 1984; 119:
1005-14.
"Fox J, Goldblatt P O, Jones DR. Social class mortality
differentials: artefact, selection or life circumstances?
J Epidemiol Community Health 1985; 39: 1-8.
OCCUPATION AND INFLAMMATORY BOWEL DISEASE
l2
Primatesta P, Goldacre M J. Crohn's disease and ulcerative
colitis in England and the Oxford record linkage study
area: a profile of hospitalized morbidity. Int J Epidemiol
1995; 24: 922-28.
13
Jensen M V, Tfichsen F. Occupation and lumbar disc prolapse.
Ugeikr Laeger 1995; 157: 1519-23 (English summary).
M
TQchsen F, Andersen O, Olsen J. Referral bias among health
workers in studies using hospitalization as a proxy measure
of the underlying incidence rate. J Clin Epidemiol 1996; 49.
637
"Tdchsen F, Jeppesen H J, Bach E. Employment status, nondaytime work and gastric ulcer in men. Int J Epidemiol
1994; 23: 365-70.
"TUchsen F, Bach E, Marmot M. Occupation and hospitalization
with ischaemic heart diseases: A new nationwide surveillance system based on hospital admissions. Int J Epidemiol
1992; 21: 450-59.
(Revised version received November 1995)