Aims: The aim of this study was to examine the correlations of birth seasonality in schizophrenia, considering influences of gender and income status. Methods: The sample consisted of 1 000 000 people in the general population randomly selected from the Taiwan National Health Insurance Research Database. Data for the birth‐year period 1950–1989 were extracted for analysis (n = 631 911; 306 194 male, 325 717 female). Subjects with schizophrenia (2796 male, 2251 female) were compared with the general population. Subgroups divided by birth‐year periods (10‐year interval), gender, and income status (low, medium, high) were analyzed using both the Walter and Elwood seasonality and chi‐squared tests. Results: The winter/spring birth excess in schizophrenia was 5.3% when compared with the general population. There was a statistically significant excess in winter/spring births than summer/autumn births inschizophrenia patients (relative risk [RR], 1.12; 95% confidence interval [CI]: 1.06–1.18). This winter/spring birth excess in schizophrenia was observed only in female subjects (RR, 1.20; 95%CI: 1.10–1.30), not in male subjects (RR, 1.03; 95%CI: 0.98–1.14), in all subgroups of income status, but was most pronounced in the low income subgroup (RR, 1.20, 1.09, 1.13; 95% CI: 1.05‐1.37, 1.01–1.17, 1.02–1.25 for low, medium, and high income status, respectively). Conclusion: A gender difference with female predominance of the effect of birth seasonality in schizophrenia, and a more pronounced effect in low income status were noted.
epidemiology; schizophrenia; season of birth; seasonality
BIRTH SEASONALITY IN schizophrenia generally refers to the higher number of winter and spring births among individuals who later develop schizophrenia. Many studies have reported this effect among schizophrenia patients. Review articles have concluded that among schizophrenia patients worldwide there are 5–8% more births during the winter and spring months compared with the general population.[
The latitude effect for birth seasonality in schizophrenia has also been described. This refers to birth seasonality in schizophrenia varying by latitude bands, with a small but significant positive correlation between effects of birth seasonality and latitude in the northern hemisphere. There is a more significant seasonality effect in high‐latitude regions, but no such effect has been observed in equatorial regions, where seasons are absent.[
There have been attempts, however, to correlate effects of birth seasonality in schizophrenia with gender, social class, urbanicity, race, family history, marital status, chronicity, severity, subtypes, as well as neurological, neuropsychological, and neurophysiological features.[
All citizens and foreign residents who have lived in Taiwan for at least 4 months are required to be insured by the National Health Insurance (NHI) program. The NHI program is a single‐payer compulsory social insurance plan that centralizes the disbursement of health‐care funds, and the population coverage had reached 99% by the end of 2003. The National Health Research Institute (NHRI) routinely transfers health insurance claims data from the NHI bureau and makes the National Health Insurance Research Database (NHIRD) available for research purposes.
The study was based on the Longitudinal Health Insurance Database 2005 (LHID2005). The LHID2005 contains all the original claim data from 1 000 000 beneficiaries, randomly sampled from the year 2005 Registry for Beneficiaries (ID) of the NHIRD. There is no significant difference in the gender or age distribution or average insured payroll‐related amount between the patients in the LHID2005 and the original NHIRD.[
The definition of the four seasons in the northern hemisphere used in the literature was adopted, as follows: spring, March–May; summer, June–August; autumn, September–November; and winter, December–February.
To examine correlations of effects of birth seasonality in schizophrenia with age, gender, and income status, both the schizophrenia patients and the general population were divided into subgroups by age, gender, and income status. Age subgroups were divided into 10‐year birth‐year periods as follows: 1950–59, 1960–69, 1970–79, and 1980–89. Each of these subgroups was further divided by gender. Income status subgroups included low, medium, and high yearly income (USD 0–1007, USD 1007–20 000, and USD >20 000).
We used SAS (SAS System for Windows, version 9.2; SAS Institute, Cary, NC, USA) to perform all analyses in this study. To assess differences between schizophrenia patients and the general population in terms of effects of birth seasonality and birth month variation in gender and income status, we used both the Walter and Elwood seasonality[
Figure [NaN] plots the monthly birth distributions of births for the schizophrenia patients and the general population. There was an obvious peak in February and a trough from June to July, but in the general population, no such patterns can be observed. The distributions across the 12 months were significantly different between the schizophrenia patients and the general population (Walter and Elwood test, P = 0.0001; Fig. [NaN] ).
Table [NaN] lists the monthly distributions of births in schizophrenia patients and in the general population for the whole sample and for each of the four birth‐year subgroups (1950–59, 1960–69, 1970–79, 1980–89). The differences between the observed and expected numbers of patients are expressed as a percentage of excess or deficit for each month. The expected number of schizophrenia births was calculated based on the expected proportion observed in the general population. An excess of schizophrenia births in winter/spring of 5.3% and a deficit in summer/autumn of 5.6% in comparison with the general population was observed. A comparison of the winter/spring schizophrenia birth excess with the summer/autumn deficit showed that the magnitude of the deficit births was greater than that of the excess births (T test, P = 0.007, SE = 2.66).
Distribution of births (1950–1989)
Period/age Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total Schizophrenia Birth‐year period/age 1950–59/50–59 113 134 100 90 94 91 84 85 87 113 102 129 1 222 1960–69/40–49 161 157 140 118 126 106 110 125 136 155 155 142 1 632 1970–79/30–39 148 144 131 114 104 93 104 133 130 112 122 129 1 464 1980–89/20–29 64 78 54 62 48 65 57 63 61 64 59 55 729 Total (observed number) 486 513 425 383 372 355 355 406 414 444 438 455 5 047 Expected number 445 442 411 384 374 382 385 428 439 461 458 438 Percentage excess or deficit (+/−) (%) +9.31 +16.01 +3.39 −0.14 −0.64 −6.99 −7.80 −5.09 −5.60 −3.72 −4.44 +3.98 All births Birth‐year period/age 1950–59/50–59 13 533 13 584 12 238 11 087 10 370 10 254 10 041 11 833 12 525 12 977 13 484 12 859 144 784 1960–69/40–49 15 141 15 555 14 335 13 111 12 608 12 673 12 260 14 192 14 061 15 337 15 353 14 991 169 617 1970–79/30–39 14 790 14 304 13 632 13 053 13 152 13 574 14 000 15 100 15 411 16 196 15 828 14 965 174 005 1980–89/20–29 12 203 11 879 11 261 10 818 10 748 11 351 11 907 12 437 12 959 13 232 12 740 11 971 143 506 Total 55 667 55 322 51 466 48 069 46 878 47 852 48 208 53 562 54 956 57 742 57 405 54 786 631 911
To confirm the effects of birth seasonality, we performed chi‐squared analysis to examine the risks of each month relative to the month with fewest schizophrenia births. We used July as the reference month and calculated the RR (95%CI) for each month. The risks were significantly greater in January and February, with RR of 1.19 (95%CI: 1.03–1.36) and 1.26 (95%CI: 1.10–1.44) separately. When the RR was compared by seasons, there was a significant excess in winter/spring schizophrenia births compared to summer/autumn (RR, 1.12; 95%CI: 1.06–1.18; P < 0.0001).
Effects of birth seasonality in schizophrenia births were observed only in female subjects but not in male subjects (P < 0.00001 in women but non‐significant in men; Walter and Elwood test). Excess winter/spring schizophrenia births were observed only in female subjects (RR, 1.20; 95%CI: 1.10–1.30) but not in male subjects (RR, 1.03; 95%CI: 0.98–1.14; P = 0.15). Also, the monthly distributions of schizophrenia births were significantly different between male and female subjects. May, August, and November showed a marked excess of male schizophrenia births, while December had a marked deficit. The birth‐month‐specific birth rate of schizophrenia in both genders, and the male to female ratio of monthly births in schizophrenia and the general population are plotted in Figure [NaN] . The pattern of male to female birth ratio in schizophrenia patients fluctuated greatly compared to that in the general population, which was almost identical throughout the year (Fig. [NaN] ).
Effects of birth seasonality of schizophrenia were noted in all income status (data not shown). Low income status was associated with the most significant effects of birth seasonality, followed by medium and high income status (P = 0.002, 0.006, and 0.047 for low, medium, and high income status, respectively; Walter and Elwood test). The winter/spring to summer/autumn RR in the three groups were 1.20 (95%CI: 1.05–1.37), 1.09 (95%CI: 1.01–1.17), and 1.13 (95%CI: 1.02–1.25), respectively.
This study identified a statistically significant winter/spring schizophrenia birth excess of 5.3% in a population‐based sample in a low‐latitude region (Taiwan). This result was consistent with findings in previous large studies that showed that schizophrenia patients worldwide have a 5–8% excess of winter/spring births compared with that of the general population.[
To better understand the effect of latitude on schizophrenia birth excess/deficit, we compared the present findings to the results of a systematic review and meta‐analysis of schizophrenia birth studies conducted in the northern hemisphere.[
Few studies have examined gender differences in effects of birth seasonality in schizophrenia, and the findings are inconsistent.[
Birth rate of schizophrenia in T aiwan (2005)
Birth year/age(years) Birth rate (%) Total Male Female M/F ratio 1950–59/50–59 0.84 0.83 0.86 0.96 1960–69/40–49 0.96 1.09 0.83 1.32 1970–79/30–39 0.84 1.00 0.69 1.46 1980–89/20–29 0.51 0.65 0.39 1.65 Total 0.80 0.91 0.69 1.32
In terms of gender differences in each birth‐year period subgroup, the male : female ratios were 0.96, 1.32, 1.46, and 1.65 in the 1950–59, 1960–69, 1970–79 and 1980–89 subgroups, respectively (Table [NaN] ). Male birth rate was predominant in all subgroups except for the 1950–59 subgroup (50–59 years old). The results suggest that male schizophrenia patients have an earlier onset. As for the exception of the 1950–59 subgroup, possible explanations include (i) a second peak of the disease onset in middle age for women; (ii) a longer lifespan in women; and (iii) an underreported rate for male subjects in the past years because of the traditionally conservative society.
There were only few studies examining the correlation between social class and effects of birth seasonality in schizophrenia, with inconclusive results;[
The primary advantages of the present database include a purer ethnic composition with few immigrants and emigrants; accuracy of birth data and subject count; a compatible control group from the general population; large sample size; and geographic centralization. This study made up a deficiency in previous seasonality studies at lower latitude. It also provides further evidence in issues of gender and socioeconomic effects, which have not been extensively studied in the seasonality literature. Some limitations should also be considered. Because Taiwan's NHI program was launched in 1995 and more comprehensive data were collected after 2000, the available data did not allow for the applications for analysis of period effect, cohort effect, and possible shifts of birth seasonality over time.
The present study has confirmed the well‐replicated finding of a winter–spring excess that varies with latitude, and provides comprehensive subgroup analyses with gender and socioeconomic status. A significant winter/spring birth excess in schizophrenia at low latitude, a gender difference with regard to female predominance, and a more pronounced effect in low income status were found in this study. Further follow‐up studies are required to explore the underlying mechanisms, combining analyses of the potential risk variables in birth seasonality and development of schizophrenia, such as vitamin D deficiency, prenatal and perinatal insults, viral exposure, meteorological influences, social factors, and so on. These analyses can be done using data for temperature, rainfall, sunlight, ultraviolet, ozone, pregnancy and birth complications, and vaccine records. If we identify these underlying risks, clinical implications of primary prevention in schizophrenia may be practicable in the future.
This study was based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health and managed by the National Health Research Institutes, Taiwan. The study was funded by Taichung Veteran General Hospital, Taichung Taiwan (Grant number TCVGH‐100‐3105). We acknowledge the help of the Biostatistics Task Force of Taichung Veterans General Hospital, Taiwan. All authors declare no conflicts of interest.
Graph: Proportion of births vs month (1950–89) for () the general population and () schizophrenia patients.
Graph: image%5ft/pcn12076-fig-0001-t.gif
Graph: Gender‐specific prevalence and male : female ratio of prevalence vs birth month for () schizophrenia patients and () the general population.
Graph: image%5ft/pcn12076-fig-0002-t.gif
Graph: Latitude vs ratio of observed/expected number of schizophrenia births per month. Reproduced with permission from McGrath J and Oxford University Press. Davies G, Welham J, Chant D, Torrey EF, McGrath J. A systematic review and meta‐analysis of Northern Hemisphere season of birth studies in schizophrenia. Schizophr Bull. 2003; 29: 587–93.
Graph: image%5ft/pcn12076-fig-0003-t.gif
By Chin Cheng; El‐Wui Loh; Ching‐Heng Lin; Chin‐Hong Chan and Tsuo‐Hung Lan