50-Year Trends in Rochester,
in the Incidence of Anorexia Minn.: A Population-Based
Alexander R. Lucas, Michael O’Fallon, Ph.D.,
w.
Objective: long-term
The aim of the trends in incidence
community
of Rochester,
M.D., C. Mary Beard, M.P.H., and Leonard T. Kurland, M.D.,
study was of anorexia
Minn.,
to determine nervosa by
during
the 50-year
disorder. Method: From a community-based with diagnoses ofamenorrhea, starvation, were
screened
diagnostic f ulfilled
to
identify
true
cases
of
anorexia
nervosa
eighty-one nervosa;
954
increased
and
to 26.3
in 1 980-1
and for males remained The overall age-adjusted
determined
residents these were
had
the
by
using
(1 66 female the incidence
standard
and 1 5 male) cases. Due to
among
J Psychiatry
females 1991;
Foundation,
I 5-24
Rochester,
print requests to Dr. Lucas, Mayo Clinic, ester, MN 55905. Supported in part by grant NS-17750
of Neurological The authors
and thank
in screening
Hultz for checking merable diagnostic analysis. © 1991
Psychiatry
Communicative Judy K. Stand medical
records,
200
years
from
the
Disorders and Joanne Laurie
Psychiatric
1 48: 7, July
1991
Minn. First
Address
St.,
old
5.W.,
National
reRoch-
Institute
and Stroke. C. Ward for assis-
E. Jones
residency, James R. Wentz lists, and Shauna L. Stensrud American
incidence 15-24 100,000
rates
for
women
years old, person-years
20 years there was a was 14.6
but
not
among
older
women
or among
males.
148:917-922)
Received July 3, 1990; revision received Dec. 3 1, 1990; accepted Feb. 4, 1991. From the Section of Child and Adolescent Psychiatry, the Section of Clinical Epidemiology, and the Section of Biostatistics,
Mayo
The
For females rate per
and I .8 for males. The prevalence rate per 1 00,000 population was 269.9 for 22.5 for males. Conclusions: Anorexia nervosa is more common than previously Among girls I 5-1 9 years old it is a very common chronic illness. Its incidence
norexia nervosa has become an important health problem in Western countries (1). It is associated with severe long-term morbidity and substantial mortality (2-5). Debate has focused on whether the mcidence of anorexia nervosa is increasing (6-9). Incidence represents the number of new cases appearing during a time interval, usually expressed as the number of cases pen 100,000 person-years (same as cases pen year pen 100,000 population). Prevalence represents the total number of cases of the disease in a group at a certain time expressed as the rate per 100,000 population on as a percentage. Most estimates have sug-
and
984.
constant. incidence
A
J
who
increased
(Am
Am
1 984
in 1 950-1
recognized.
statistical Copyright
1 935 through
in the rates for girls 1 0-i 9 years old, the incidence rate among female 1 6.6 per 1 00,000 person-years in the 1 935-1 939 period to a low of 7.0
has
tance
period
and prevalence rates and all persons residing in the
a quadratic trend residents fell from
f or females f emales and
Clinic
incidence identifying
Dr.P.H.
epidemiologic resource, 13,559 medical records weight loss, anorexia nervosa, or other conditions
criteria. Results: One hundred the diagnostic criteria for anorexia
old and older linear increase.
Mayo
Nervosa Study
and
for searching for assistance
Association.
Karen
S.
innuin the
gested
an increase in the incidence of anorexia nenvosa 1950 (10-17) and perhaps since 1930 (18). Although most clinicians interpreted greater numbers of patients with anorexia nervosa seen in practice as signifying a remarkable increase in the occurrence of the disorder, Williams and King (7) in England and Nielsen (19) in Denmark concluded that the apparent rise in anorexia nervosa was not real but was related to changes in demographics and in medical practice. Our population-based study of the incidence of anorexia nervosa in Rochester, Minn., during the 45-year penod 1935-1979 (8) failed to reveal a significant trend and identified surprisingly high rates during the 15year interval between 1935 and 1949. A trend toward an increase in the rates for girls 10-19 years old, albeit not statistically significant, was noted for the period 1950-1979. A considerable increase in incidence in females with the illness during the final 2 years of the study was a compelling reason to extend the study for another S years to determine whether that trend would continue. Rochester is a Midwestern city located about 75 miles southeast of Minneapolis. It is the county seat of Olmsted County, a grain and dairy farming region. its population was approximately 60,000 in 1985 and since
917
SO-YEAR
TRENDS
IN
ANOREXIA
TABLE 1. Diagnostic Categories Reviewed to Identify Anorexia Nervosa in Rochester, Minn., 19351984a
Diagnostic
Number Recoras Screened
Category
Anorexia
nervosa
Amenorrhea Oligomenorrhea Anovulation, irregular menstruation, menstrual dysfunction
dysfunction,
anism Delayed
puberty,
delayed
malnutrition diet disturbance,
Anasarca,
edema,
Addison’s
disease
Simmonds’ Malaise
b213
incidence
peared
in more
97
42.0
15
1.8
70 14
1.8 1.6
2,120
and than
Nervosa
70 60 Cl)
0 Females
5O
OMales
ci
(N
166)
=
(N=15)
C) .340
.30 2O
S
0.2
3
0.5
L
10 0
menar-
1 2
428 181
0-
5-
10-
4
9
14
:::
1519
2024
:
25- 30- 35- 4029
34
Age
0.2 1.1
39
44
4549
5054
5559
60
(years)
under1,483 1,775 marasmus
psychogenic disorder
Mayo
FIGURE 1. Distribution of Age at Diagnosis of Anorexia Among Female and Male Residents of Rochester, Minn.
z
51 4
3.4 0.2
377 138
0 0
0.0 0.0
54
0
0.0
52 61
1 0
1.9 0.0
344
1
0.3
12 203
0 0
0.0 0.0
disease
from
%
826
hydrops
Nervous exhaustion Nervous stomach, gastrointestinal Vomiting are
N
3,789 878
607
weight, Inadequate Nutritional
“Data
or
hypo-ovari-
che Delayed menstruation Starvation, weight loss,
Total
Cases of Anorexia Nervosa Identified
231
Anorexia
Ovarian
All Cases of
Clinic prevalence one
records
and
cases
were
diagnostic
13,559 other
sources.
identified;
1.9
264b
some
ap-
category.
23,000 in 1935. Although the local economy once was based chiefly on farming and light industry, the Mayo Medical Center and IBM are now the largest employens in the city. The homogeneous population is pnimarily of northern European origin. In 1980, the racial composition was 98% white. Compared with the U.S. population, there was a relatively greater proportion of female residents between the ages of 15 and 25 years.
tients had never been hospitalized for their illness, and many had not seen a psychiatrist. Records with diagnostic coding for depressive illnesses were not screened separately, although many of the anorexia nervosa patients had depressive symptoms and some developed major depression after the onset of anorexia nervosa. The depressed patients who had notable weight loss or menstrual dysfunction appeared on those diagnostic lists and thus were identifled as potential anorexia nervosa cases. Diagnostic certainty was rated as definite, probable, and possible on the basis of the completeness of the information available in the records. When incidence data were analyzed separately for these groups, for the first 45 years of the study the trends did not differ appreciably from the overall trend for all subjects cornbined (8). Therefore, we combined the groups to analyze data for the entire 50-year span of the study. Criteria were applied consistently throughout the duration of the study, maximizing the likelihood that a comparable group of subjects was identified over time. The background of the study and the Mayo Clinic epidemiologic archives have been described previously
(8, 21, 22). METHOD The medical records in the Mayo Clinic epidemiologic archives and in the files of other health-cane providens in the surrounding geographic area were exammed for residents of Rochester, Minn. We screened for approximately 30 diagnostic terms and particularly for amenorrhea, oligomenorrhea, starvation and weight loss from any cause, and anorexia nervosa (table 1). For patients with these diagnoses the original medical records were reviewed and the diagnosis of anorexia nervosa determined or confirmed by using DSM-III-R and the criteria of the Pathology of Eating Group (20). Thus, we were able to identify a large number of patients not previously diagnosed as having anorexia nenvosa. Those who were so diagnosed previously but did not meet the criteria were excluded. Most of these pa-
918
Tests of trends using generalized structure. Tests of freedom, and freedom.
in incidence rates were performed by linear models (23) with Poisson error for linearity of trend have one degree quadratic models have two degrees of
RESULTS During the 50-year period 1935-1984, 166 females and 15 males who had been residing in Rochester for at least 1 year developed anorexia nervosa. The mean± SD age at diagnosis of the females was 21.7±8.3 years (median=19 years, nange=10-57 years); for the males it was 20.9±9.0 years (median=18 years, range=1249 years). The age distribution is shown in figure 1. Age-adjusted (to the 1970 U.S. white population) incidence rates by S-year periods are shown in figure 2.
AmJPsychiatry
148:7,July
1991
LUCAS,
FIGURE 2. Age.Adjusteda Incidence Rates for Anorexia Nervosa in Female and Male Residents of Rochester, Minn., by 5-Year Periods
11 35
-
30
-
BEARD,
O’FALLON,
ET AL.
FIGURE 3. Age-Specific Incidence Ratesa for Anorexia Nervosa, by 5-Year Periods, in Two Age Groups of Female Residents of Rochester, Minn.b
U)
Females
S-S
(N
=
166)
j
le,15,//
200-
-.-Age
10-19
Age 20-59
-0-
.
S S S S
19351939
19401944
19451949
1950-
1955-
1954
1959
5-year
19601964
19651969
. period
1970-
1975-
1980-
1935-
1974
1979
1984
1984
to
1970
U.S.
white
I
I
1935-
19401944
‘ -
Nervosa
Incidence per 100,000 Person-Years Age Group (years) Females Males ________________________________________________________
Total
0-9 10-14
0.0 25.7
0.0 3.7
0.0 14.6
15-19
69.4
7.3
43.5
20-24
27.6
2S-29
50-59 60 and older Total aAge.adjusted
(to
1970
20.7 10.9 5.2 2.0 0.0 15.1 U.S. white
1.8.
Age-
and
_
14.6;
males,
population)
sex-adjusted
rate:
4.6
19.9
3.6 0.0 1.0 0.0 0.0 1.7 rates:
13.0 5.6 3.2 1.1 0.0 9.0
on
a log
bObseedcircles,
Psychiatry
148:7,
1954
July
I
I
I
1955-
19601964
19651969
1959
200
-
100
-
1991
I 19701974
I 19751979
I 19801984
period
scale.
predicted=lines.
FIGURE 4. Age-Specific Incidence 5-Year Periods, in 15-24-Year-Old Minn.”
Ratesa for Anorexia Nervosa, by Female Residents of Rochester,
Cl) C
.5
10
8.2.
5
J
0
females,
The overall age-adjusted incidence rate for females was 14.6 pen 100,000 person-years; for males it was 1.8 per 100,000 person-years. Thus, anorexia nervosa was found 8.1 times more frequently in females than in males. The overall age- and sex-adjusted rate was 8.2 per 100,000 person-years. Age- and sex-specific mcidence rates are shown in table 2. The incidence rates for females 10-19 years old and 20-59 years old are shown in figure 3. The age-adjusted incidence rates of anorexia nenvosa in females residing in Rochester have increased from a low of 7.0 pen 100,000 person-years in 19501954 to a high of 26.3 in 1980-1984. This increase is due primarily to the highly significant linear trend in the incidence rates for females 10-19 years old (p< 0.001). The estimated coefficient indicates that, on the average, the incidence rates increased by 36% every S years over the 35 years from 1950 to 1984. When the entire scope of the study is considered (1935-1984), the incidence rates for females 10-19 years old show a highly significant quadratic trend (p