Sex difference in age of onset of schizophrenia: findings from a community-based study in India

Abstract

This study examined the sex difference in age of onset of schizophrenia in a community sample. Community-level health workers identified patients with symptoms of schizophrenia living in the community in a defined geographical area in South India. Two hundred and nine of them were diagnosed as hav-ing schizophrenia according to ICD-10 criteria by a team of psychiatrists. The age of onset of schizophrenia was assessed using the Interview for Retro-spective Assessment of Onset of Schizophrenia (IRAOS). The mean age of onset of schizophrenia did not significantly differ between males (29.2±8.8 years) and females (30.8±11.4 years) (t = 1.12; p = 0.27). Among those with an age of onset ≤33 years, females had a significantly earlier onset; among those with an age of onset >33 years, females had a significantly later onset. The results from this community-based study confirm the previous findings in hospital-based patients in Asia. There is a need to revise the description of schizophrenia in the classificatory systems, keeping in view the regional varia-tions in the age of onset of the disorder.

A sex difference in the age of onset of schizophrenia has been reported since the time dementia praecox was described by Kra-epelin 1. A later age of onset in females has been reported in several recent studies 2,3. Studies have also observed that females with schizophrenia have an older age at first admission 4,5. Overall, these studies suggested a difference of 3-5 years between the sexes for age of onset of the disorder. The ICD-10 6and the DSM-IV-TR 7also note that females have a later age of onset of schizophrenia. This difference is proposed to be due to both males having an earlier and pronounced peak incidence in their early 20s and females having a second, later peak incidence in their late 40s 2.

However, some studies from Asia and Africa do not seem to support this finding. The International Pilot Study of Schizo-phrenia (IPSS, 8), the Madras Longitudinal study 9, and three studies from the National Institute of Mental Health and Neuro Sci-ences (NIMHANS), Bangalore 10-12found no difference in age of onset of the disorder between the sexes. More recently, a study from Pakistan 13also did not find a sex difference in the age of onset of schizophrenia. Actually, in one of the above studies 11, there was a female preponderance among patients with the earliest onset. A relatively greater loss of male infants due to poor perina-tal care, which eliminated a proportion of earliest onset male schizophrenics, was hypothesized as a possible explanation. Indeed, a comparison of patients from regions with high and low infant mortality rate showed that there was a reversed gender effect on age of onset of schizophrenia in the former but not in the latter region 12.

One limitation common to the above reports was that all of them included patients who sought help. In a country like India, with a huge population/psychiatrist ratio 14, a large proportion of patients with schizophrenia live without treatment in the commu-nity. Treatment-seeking patients may not be representative of all schizophrenia patients. This study was conducted to explore whether the finding of a lack of sex difference in age of onset of the disorder could be replicated in a community sample consisting of both treated and untreated schizophrenia patients.

METHODS

Subjects

The sample for this study included patients with schizophrenia recruited for the Community Intervention in Psychotic Disor-ders (CoInPsyD) project in Thirthahalli (an administrative block in South India with a population of 143,000). The project aims to identify all patients with schizophrenia living in this rural community and treat them. Fifty-four rural health workers were trained in identifying patients with severe mental disorders in the community by a team of senior consultants from the NIMHANS. This in-cluded didactic lectures on symptoms and course of schizophrenia, video clippings of patients with schizophrenia, and question-answer sessions. These were done on three different occasions separated by about a month each. At the end of the training, the health workers were shown video interviews of different psychiatric patients and were asked to identify those with schizophrenia: they were able to identify them accurately.

Two trained social workers interviewed the Thirthahalli health workers about the presence of persons with symptoms suggest-ing psychosis in each family (a total of 29,432 families for the entire community). All patients thus identified were clinically inter-viewed by a research psychiatrist, and diagnoses were assigned using the ICD-10-Diagnostic Criteria for Research (ICD-10-DCR, 6). The diagnosis of schizophrenia was confirmed by another psychiatrist after an independent clinical interview.

A total of 209 persons were diagnosed as having schizophrenia. Of these, five could not give reliable information about age of onset of the disorder. The diagnosis of two patients changed (one to bipolar disorder and the other to organic psychosis) during fol-low-up. The final sample thus consisted of 202 subjects. Of these, 103 were males and 99 were females. One hundred and fourteen (56.4%) were receiving treatment at the time of evaluation; the rest were living without any treatment.

The health workers reported about the presence, in the community, of 20 other persons with features suggesting schizophre-nia. These could not be interviewed because of several reasons, including refusal to give consent or being severely ill with no care-takers to give any information.

Assessments

Information concerning the age of onset of the disorder was collected by the Interview for Retrospective Assessment of Onset of Schizophrenia (IRAOS, 15), used by a research psychiatrist. Subjects, family members who were in continuous contact with them and the health workers were interviewed and the age of onset of the first psychotic episode was determined. Two social workers collected the sociodemographic details of the subjects, including lifetime use of alcohol and illicit substances. This included a question on the age of onset of schizophrenia. The age of onset assessed using the IRAOS by the psychiatrist had a high degree of interrater reliability with the age of onset as recorded by the social worker (intraclass correlation coefficient: 0.86).

The study obtained ethical clearance from the Institute's Ethics Committee and all subjects were recruited after obtaining writ-ten informant consent.

Statistical analysis

Independent-sample t-test and Kaplan-Meier survival analysis were used to analyze the difference between males and females in age of onset of schizophrenia. The Statistical Package for Social Sciences version 10.0.1 was used for the analysis.

RESULTS

Table 1 shows the sociodemographic and clinical features of males and females. The mean age of onset of schizophrenia was 29.2±8.8 years for males and 30.8±11.4 years for females (t=1.12; p=0.27). A cut-off age of onset at 33 years was taken to classify patients as having earlier or later age of onset of the disorder. Figure 1 shows the survival analysis using Kaplan-Meier survival curve for both groups: in the earlier age-of-onset group, females had a significantly lower age of onset of the disorder; in the later age-of-onset group, they had a significantly higher age of onset.

Table 1.

Table 1 Sociodemographic and clinical features of males and females

Variables	Males (n=105)	Females (n=99)	t/chi-square	p
Current age (years, mean±SD)	41.4±9.7	41.6±11.9	0.1	0.92
Age of onset of schizophrenia (years, mean±SD)
Total sample	29.2±8.8	30.8±11.4	1.12	0.27
Age of onset < 33 years	24.8±5.8	22.9±4.9	2.01	0.046
Age of onset >33 years	39.5±5.2	43.0±6.9	2.35	0.022
Duration of illness (years, mean ±SD)	12.7±6.9	10.8±9.3	1.07	0.29
Socio-economic status (%)
Lower	42.4	50.6
Middle	33.3	34.1	1.42	0.496
Upper	24.2	15.3
Education (years, mean±SD)	6.8±4.7	6.4±4.8	0.43	0.664
Alcohol abuse/dependence (%)	31.2	2.1	30–154	<0.001

Figure 1.

Kaplan-Meier survival curves for age of onset of schizophrenia in males and females

Figure 2 shows the number of males and females who had their onset at different ages. Females had two peaks: the first, higher peak in the 20-25 years range and another in the 35-40 years range. Males had a steady rise through the early ages to a peak at 30-35 years; this was followed by a steep decline through the older age-range.

Figure 2.

Distribution of age of onset of schizophrenia in males and females

The results were not different among those with illness duration less than 10 years: 32.1±8.8 years for males (n=41) and 33.1±12.5 years for females (n=51) (t=0.45; p=0.66).

DISCUSSION

This study shows that in India there is no significant difference between the sexes in the age of onset of schizophrenia. Among early-onset patients, females have a significantly earlier age of onset than males, and among late-onset patients, they have a later age of onset.

The important merit of this study is that it included all patients from a defined geographical area in a rural setting. The sample included both treated and untreated patients with schizophrenia living in the community. We found that about 39% of the patients were living untreated. Hospital-based studies would have missed these patients. Though this study was not aimed to assess the prevalence of schizophrenia in the community, the point prevalence, as can be made out, is 1.6 per thousand (95% CI: 1.3-1.8 per thousand). This is comparable to the prevalence reported from other areas in India (16) and other South Asian countries like Sri Lanka 17.

We identified patients who were either currently symptomatic (on or off treatment) or in remission while being on antipsy-chotic medications; we might have missed patients who had sustained remission of their schizophrenic episode despite being off treatment currently. We may have also missed a few patients as we conducted a “key informant” rather than a door-to-door survey. However, the number of such missing cases is likely to be low. The health workers visit the families once every month; they are thus quite knowledgeable about the families under their care and they would not have missed patients with symptoms of schizophrenia or receiving treatment for the same.

The diagnosis of schizophrenia was made by two psychiatrists after independent interviews and remained stable at six-month follow-up in the 202 subjects included in the analysis. There were only two subjects who had a duration of psychosis of less than 6 months – the sample was not “contaminated” by inclusion of acute psychosis patients.

A psychiatrist trained in administering the IRAOS assessed the age of onset of the disorder. There was a high degree of inter-rater reliability between his findings and the assessment by an independent interviewer. The age of onset recorded in this community sample is comparable with other hospital studies from India 10-12.

The duration of psychosis in this sample ranged from 4 months to 46 years. Eighty-one (50.3%) of the subjects had a duration of illness of ten years or more. To rule out a possible difficulty in recall in those with a long duration of psychosis, we compared the age of onset of males and females where the duration of psychosis was less than 10 years: this did not alter the findings.

Our results show that, similar to the literature from Western countries 2, women have an earlier, higher peak in age of onset in the early twenties and a later, lower peak in their late thirties. However, unlike Western countries, men had a fairly later peak in age of onset in the early thirties, followed by a steep decline through the older age ranges. It may be reasoned that this is because of lesser number of men with very early age of onset in our sample.

Our sample had 80% power to detect a mean difference of 3.9 years, which is the reported figure in the similar studies from the West 2. The difference in mean age of onset for the whole sample was 1.6 years, and this difference was not statistically signifi-cant. One might argue that our study did not have adequate power to detect this difference. Though larger samples could detect this difference, the magnitude of the difference is likely to be substantially lower than what literature suggests.

In our previous studies on this issue, we have argued that poor perinatal care in India might result in preferential attrition of birth-injured male children, who, in the Western samples, would have contributed to age of onset being lower among males 12. This explanation may be true of the current sample too. An alternative explanation can be sought in the present study. None of the patients had used illegal substances anytime in their lives. It is known that abuse of illicit substances is associated with earlier age at onset 18-20. Greater proportion of male patients abuse illicit substances than female patients 21,22. This might contribute to earlier age at onset of schizophrenia in males in Western countries. Absence of such abuse also may contribute to the lack of sex differences in our sample. A lower rate of substance abuse has been suggested to explain the near-equal sex ratio in the incidence of schizophrenia in the developing countries 23. A sex ratio of 1:1 in our sample perhaps reflects a similar trend.

This work was done in a rural South Indian setting and the results may thus be generalized to similar populations. However, our earlier reports, which showed similar results, were from a mixed rural-urban population from a tertiary center, which draws pa-tients from all over India. Taken together, these consistent findings suggest that the epidemiology of schizophrenia is different in In-dia from Western countries, at least with respect to age of onset of the disorder: there seems to be a relative lack of earliest onset male schizophrenia patients in our population. The note in the diagnostic systems about sex differences in age of onset of schizophrenia cannot thus be generalized.