|
Demographics
Table
1 displays the total numbers of subjects recruited in each strata,
the proportion that were women, and similar data from the Ontario
census conducted in 1996. Despite the use of random sampling of
the resident with the next birthday, the sample generally overrepresents
women (59.3%). In addition, in 2 strata (stratum 2, with 17% rural,
and stratum 8, with 16.9% urban), we were not able to recruit the
planned minimum of 18%. However, the sample was not significantly
different from the Ontario population with regard to education level,
marital status, and employment status (data not shown).
Prevalence of SAD
Overall the crude prevalence of current SAD was 1.9% (95%CI, 1.3
to 2.7), and the prevalence of lifetime SAD was 2.6% (95% CI, 1.9
to 3.5). Since lifetime SAD reflects a more inclusive diagnostic
entity, we used this term for further analyses. The OR for the presence
of lifetime SAD among women was 1.67 (95% CI, .85 to 3.3; x2
= 2.2, df 1, P = 0.13). Regarding the relation between lifetime
prevalence and age, with the highest prevalence was in the group
aged 40 to 49 years (3.6%) and the lowest prevalence in those older
than 70 years (0%; see Figure
1). Using the technique of Levy and Lemeshow (10), adjusting
for age and sex, the adjusted prevalence of lifetime SAD was the
same as the crude prevalence. The prevalence estimate of lifetime
SAD was not affected by the season of interview (winter = 2.5%,
spring = 2.5%, summer = 2.9%, fall = 2.3%; x2
[for linear-by-linear association] = 0.00, df 1, P = 0.9).
The prevalence of SPAQ-SAD was 7.4% (95%CI, 6.2 to 8.8). There
was also a significant decline in prevalence of SPAQ-SAD with age
(x2 [for
linear trends] = 19.7, df 1, P < 0.0001, data not shown),
with the highest prevalence in the group aged 20 to 29 years (10.3%)
and the lowest prevalence in those over 70 years (0%). The prevalence
of SPAQ-SAD was significantly different, depending on the season
of interview: for winter, the rate was 5.4%; for spring, 6.4%; for
summer, 8.2%; and for fall, 8.7% (c2 [for linear-by-linear association]
= 4.0, df 1, P < 0.05).
Impact of Latitude
Table
2 presents the prevalence estimates for lifetime SAD, SPAQ-SAD,
and major depression for each stratum. There were no significant
trends for the group as a whole in any diagnostic group. For women,
the lowest prevalence occurred in the strata with the highest latitude
(1.7% in strata 8), and the highest prevalence occurred in the strata
with the lowest latitude (5.4% in strata 1). Using t-tests,
the mean latitude for women with lifetime SAD (45.0 oN, SD 2.1)
was minimally but statistically significantly lower, compared with
women who did not have lifetime SAD (45.8 oN, SD 2.1; t =
2.0, df 952, P < 0.05). There was no such effect for men.
As noted above, the SPAQ generates a measure of overall seasonality—the
GSS—based on seasonal changes in 6 symptoms. There was a practically
insignificant, although statistically significant, negative correlation
between latitude and the GSS score (r = –0.05, P
< 0.05). Controlling for age and sex and longitude did not change
the r value or significance.
Rural vs Urban Differences
In rural areas, the OR for the presence of lifetime SAD was 0.85
(95% CI, 0.45 to 1.60; x2
= 0.24; P = ns), for SPAQ-SAD it was 0.86 (95% CI, 0.58 to
1.26; x2
= 0.57; p = ns), and for lifetime major depression it was
0.99 (95% CI, 0.80 to 1.24; x2
= 0.003; P = ns).
Reliability
Twenty-nine subjects were contacted on 2 occasions within 1 month.
Two subjects did not have an exact match on demographic variables
and were therefore excluded from the reliability analysis. Of the
27 remaining subjects, 11 had major depression at the first interview,
and 12 at the second. The kappa value was 0.92, with the positive
predictive value of 100% and a negative predictive value of 94%.
No subjects had SAD according to DSM-IV criteria at either interview.
One subject met criteria for SPAQ-SAD at the first interview, and
2 met criteria at the econd interview, with a kappa value of 0.65.
The correlation between the GSS on first and second interview was
highly significant (r = 0.68, P < 0.0001).
Discussion
Prevalence
This study finds a 2.6% prevalence of lifetime SAD in the province
of Ontario between the latitude of 41.5 °N and 49.5 °N.
This is in keeping with a recent survey using a similar methodology
in the city of Toronto, Ontario (1). In that study, our group found
a 2.9% lifetime prevalence of SAD. In fact, 151 respondents in the
current survey were from the Toronto area, and 4 (2.6%) had lifetime
SAD. The prevalence of lifetime major depression was 26.2%. This
is similar to the 26.4% lifetime prevalence of major depression
found in the Toronto survey and the 29.6% found by Fournier and
colleagues (11) using a similar telephone survey in Montreal, Quebec.
These findings suggest that major depression is common in Ontario
and that the seasonal subtype of major depression represents approximately
10% of all cases. It should be noted that the prevalence estimate
for SAD differs, based on whether the focus is on current symptoms
or lifetime symptoms. As noted above, the DSM-IV requires that there
be 2 episodes in the last 2 years. However, it is possible that
a person has had a lifetime pattern of seasonal episodes but does
not have an episode in the current year. If the strict DSM-IV criteria
are applied in the current study, only 1.9% of the sample meet criteria
for SAD. However, a further 0.7% of the population clearly had recurrent
annual depressions and had had a period of time wherein they had
2 consecutive years with seasonal episodes. Although the distinction
between current and lifetime SAD has not previously been examined
in detail, we believe that it is meaningful to include all patients
with a clearly distinct seasonal pattern of major depression in
SAD prevalence estimates.
In this study, the SPAQ tended to overestimate the prevalence of
SAD. Of the 119 subjects identified as having SAD according to the
SPAQ, 32 did not have a single lifetime episode of major depression
by diagnostic criteria. Of the 87 with an episode of major depression,
27 did not have recurrent depression. Therefore, approximately one-half
of the SPAQ-identified subjects could not fulfill DSM-IV diagnostic
criteria for SAD. In fact, if the DSM-IV SAD as determined by the
DSI is taken as the gold standard, the SPAQ has a specificity of
0.98, but a sensitivity of 0.13. The SPAQ therefore appears to be
a useful screening instrument that is excellent at ruling out SAD,
if it is negative; the instrument is less useful when it is positive.
|
|
This has serious implications for its use as a diagnostic instrument
in epidemiologic surveys. Nonetheless, it provides interesting and
important information regarding seasonality of mood symptoms.
Latitude
To our knowledge, this is the first study of the epidemiology of
SAD to structure the sample to survey prospectively according to
latitude, rather than sampling a specific city or region and assigning
latitude retrospectively. We did not find an increased prevalence
of SAD with increasing latitude, as has been suggested in previous
studies (2). The findings of this study suggest that latitude has
no meaningful impact on the pattern of SAD’s occurrence. The
statistically significant findings were of minimal effect size and
pointed to an opposite conclusion. First, the prevalence of lifetime
SAD in women tended to decline with increasing latitude. Further,
there was a significant trend for decreasing seasonal difficulties
(as measured by the GSS) with increasing latitude. In other words,
the population living at higher latitudes tended to report less
difficulty with the seasons. Taken together these findings do not
support the assumption that seasonal depression is more common at
higher latitudes. Similarly, Blazer and others (7) did not find
a significantly different OR for the prevalence of seasonal depression
according to geographic region in the US. However, the overall prevalence
of seasonal depression in their study was very low, specifically,
0.3%. The methodology employed in their study also differed substantially
from ours. Nonetheless, they did use a similar instrument to diagnose
depression and determine the seasonal recurrence of major depression.
As well, both studies employed the DSM-IV criteria for the diagnosis
of SAD. It is possible that the US, in general, has a lower prevalence
of seasonal depression, compared with Canada, because of the differences
in the latitude between the 2 countries. However, there are many
other differences between the 2 samples, and no firm conclusions
can be drawn as yet.
It is possible that the rates of seasonal depression decline with
increasing latitude as a result of a general migration south of
people with seasonal difficulties. In other words, people who have
recurrent difficulties in the winter may move south to milder conditions
in an attempt to ameliorate their seasonal difficulties. Alternatively,
some individuals who are more tolerant of the northern climatic
condition may move north. We did exclude subjects who had not lived
in their geographical area for more than 3 years, but it is possible
that people moved more than 3 years ago to a more acceptable climate.
Another explanation may be that prolonged exposure to northern climates
leads to adaptation and the ability to tolerate the more dramatic
seasonal changes. Finally, it is possible that this finding was
spurious and results from sampling across only a limited range of
latitude. It is possible that had we been able to sample across
a broader range of latitudes we might have been able to detect somewhat
different patterns.
It is possible that women have an accentuated response to latitude
and therefore may be more likely to demonstrate a relation between
latitude and the full syndrome of SAD. It is important to note that
latitude is only a crude measure of some of the climatic risk factors—such
as exposure to sunshine or daily temperatures—thought to play
a role in the development of SAD. These variables may not change
in a linear fashion segregated by horizontal bands. Therefore, the
current focus on latitude may obscure the true relation between
climatic variables and the occurrence or expression of SAD.
Limitations
There are several possible limitations to this study. First, over
one-half of the telephone numbers initially contacted refused to
participate and did not hear details of the study. This is typical
for telephone surveys in this province, but it means that the current
sample may not represent Ontario as a whole. However, over 70% of
people who did hear details of the nature of the study completed
the interview. Further, when the demographic profile of the subjects
in the current sample is compared with the 1996 Ontario census,
there are no significant differences with regard to housing, schooling,
household composition, and employment status (data not shown). The
only substantial difference between the current sample and the Ontario
census population is a slight but significant preponderance of female
respondents. Nonetheless, when the prevalence estimates are adjusted
for age and sex, there is no difference between the crude rate and
the adjusted rate. Another limitation is some oversampling in 2
strata and some undersampling of rural or urban regions in 2 strata.
This was an unfortunate complication, but the prevalence of SAD
in the over- or undersampled strata did not differ from the overall
rates, or from the rates in any other stratum. A further limitation
may be that approximately 4% of Ontario households do not have a
telephone and could not be sampled. It is possible that these households
comprise individuals in lower-income groups who might also have
high rates of mental illness. However, the number of households
that would have been excluded is small, and there could at best
be only a marginal impact on prevalence estimates. This study sampled
1605 respondents—not a large number, given the prevalence rate
of 2.6% for this condition. Further, the number of subjects sampled
per strata was as low as 157 (strata 3), which meant a 95%CI range
from 0.6% to 5.2% for that stratum. Nonetheless, this study is the
largest single survey to specifically examine the pattern of occurrence
of SAD across a range of latitudes, and the findings are in keeping
with a previous report from our group. Finally, although several
findings were significant, they were not of a substantial magnitude.
It is therefore possible that there is no appreciable or important
influence of latitude on the expression or occurrence of SAD or
seasonal difficulties.
SAD is a relatively common subtype of major depression that does
not demonstrate the expected increasing prevalence with increasing
latitude—at least in the current sample spanning 8 degrees
of latitude in the province of Ontario. Future studies should attempt
to sample across a wider range of latitudes, potentially spanning
the broadest range of landmass possible. The DSI may be an appropriate
tool for such an investigation: it appears to have good reliability,
when compared with face-to-face diagnostic interview (1) and when
compared with repeat telephone interview within 1 month.
Acknowledgements
The authors acknowledge the assistance of Zillah Baumil, Irja Helin,
and Randy Leiter at Goldfarb Consultants; John Smart at Sampling
Modelling Research Technologies; and Cathy MacDonald for assistance
in sampling and interviewing. Data from this study were presented
at the American Psychiatric Association Annual conference, Toronto,
1998. This study was funded by the Ontario Mental Health Foundation,
Ontario, Canada.
|
