This prospective research was designed to provide longitudinal data about potential sex differences in the clinical course and outcome of psychiatric disorders. It also explored whether potential sex differences are unique to schizophrenia or whether they also characterize other psychotic disorders and nonpsychotic disorders. The research was based on large samples of patients who were assessed during index hospitalization and then followed up prospectively five times over the next 15 years: at a mean of two, 4.5, 7.5, ten, and 15 years after the index hospitalization (
1,
2 ). Data on sex differences in global outcome, recovery, and long-term course of illness involving numerous assessments have not previously been available to the field of psychiatry.
Numerous studies suggest that among persons with schizophrenia, women show less severe courses of illness than men and receive initial inpatient and outpatient treatment at an older age (
3,
4,
5,
6,
7,
8 ). Women with schizophrenia have been reported to show fewer negative symptoms (
9,
10,
11,
12 ) and better responses to antipsychotics (
13 ). Longitudinal research also suggests that women with schizophrenia show better social functioning (
14,
15,
16,
17 ). Other studies suggest that women with schizophrenia have shorter and fewer hospitalizations and more consistent family involvement (
18,
19,
20 ). The data suggesting that women with schizophrenia have better outcomes than their male counterparts have been generally accepted by the field. However, some studies have found no sex differences in, for instance, negative, affective, and psychotic symptoms (
21,
22 ), neurocognitive functioning (
23 ), magnetic resonance imaging findings (
23 ), number of hospitalizations (
24 ), and lengths of stay (
24 ). Other studies have suggested that the outcome of women with schizophrenia declines throughout the course of the disease and eventually approximates that of men with schizophrenia (
25,
26,
27 ). Very few studies have evaluated sex differences among individuals with other psychotic disorders.
Many studies (
3,
4,
7 ), although not all (
14,
24 ), suggest that women with schizophrenia have a later onset of illness than men with schizophrenia. The research presented here is one of the few naturalistic follow-up studies to control for age of onset by focusing on a sample of patients who were young when they entered the study (mean age of 23 years for both sexes) (
28,
29 ). Thus, in this study, women whose schizophrenia had relatively early onset for women were compared with men whose schizophrenia emerged at a more typical age for men.
This research investigated sex differences by using a prospective longitudinal design to address the following questions: Do psychiatric patients show sex differences in long-term global outcome, rate of recovery, course of illness, and rehospitalization rates? If so, are these sex differences limited to schizophrenia or are they also characteristic of other psychotic disorders? Do these potential differences also extend to patients with nonpsychotic disorders?
Methods
Patient samples
All 239 patients in this study (77 percent of the original sample) had data available at the 15-year follow-up. Of these 239 patients, 190 (79 percent) had data available at all five follow-ups. Forty (17 percent) had data available from four of the five follow-ups, and nine (4 percent) had data available from three of the five follow-ups. Each analysis reflects the data that were available at the relevant follow-up points. The patients were participants in the Chicago Follow-Up Study, a prospective longitudinal research program studying major dimensions of psychopathology, including symptoms, longitudinal course of illness, and global outcome (
2,
28,
29,
30,
31,
32,
33 ). Patients were diagnosed at index hospitalization according to the Research Diagnostic Criteria (
34 ). Research diagnoses were made for all patients prospectively at the index hospitalization by using structured diagnostic interviews, including the Schedule for Affective Disorders and Schizophrenia (SADS), the Schizophrenia State Inventory (
35 ), or both interviews, as well as information from admission clinical diagnostic interviews and detailed inpatient observations. Interrater reliability for the diagnosis of schizophrenia was
κ =.88.
A total of 120 men and 119 women were in the sample. Groups included 69 individuals with schizophrenia (29 percent; 24 women and 45 men), 56 individuals with other psychotic disorders (23 percent; 29 women and 27 men), and 114 individuals with nonpsychotic disorders (48 percent; 66 women and 48 men). The individuals with other psychotic disorders included 29 with bipolar disorder with manic episodes at index hospitalization (52 percent), 14 with psychotic depression (25 percent), and 13 with other psychotic disorders (23 percent). Patients without psychosis included 63 with unipolar depression (55 percent); six with nonpsychotic bipolar disorder (5 percent); 11 with minor depression or other depression-related disorders (10 percent); seven with personality disorders (7 percent), including three with borderline personality disorder (3 percent); seven with substance use disorders (6 percent); five with eating disorders (4 percent); and 15 with other nonpsychotic disorders (13 percent).
All patients were assessed at the index hospitalization by using a standardized battery of semistructured interviews, questionnaires, and psychological tests. Patients were reassessed at five follow-ups over a 15-year period.
Of the 239 patients in our sample, 190 (80 percent; 95 women and 95 men) were assessed at all five of the follow-ups over 15 years. Another 40 patients (17 percent; 18 women and 22 men) were assessed at four of the five follow-ups. Overall, 113 women (95 percent) and 117 men (98 percent) were assessed at four or five of the five follow-ups.
To control for age of onset and to reduce the potential effects of long-term treatment, this sample was first assessed at a relatively young age (index hospitalization): the mean±SD age of the women was 23.4±4.4 years, and the mean age of the men was 22.6±3.6 years. Thus they entered the study as a young group without any chronic psychiatric conditions: 65 women (55 percent) and 64 men (53 percent) were experiencing their first hospitalization at the index hospitalization. A total of 182 (76 percent) had either one or no previous hospitalizations.
The sample included 177 white patients (74 percent), 59 African American patients (25 percent), and three patients of other races (1 percent). Parental socioeconomic status was measured with the Hollingshead-Redlich Scale (
36 ). Possible scores on the scale range from 1 to 5, with higher scores indicating lower socioeconomic status; participants had a mean score of 3.07±1.35. The mean education level was 13.2±2.3 years. There were no significant sex differences in the age, education, number of previous hospitalizations, or social class for any of the three diagnostic groups.
Follow-up instruments and assessment
Follow-up assessments involved a semistructured interview providing detailed information about patients' global functioning and adjustment, symptoms, work and social functioning, family adjustment, and rehospitalization. Patients were administered a modified version of the Schedule for Affective Disorders and Schizophrenia (SADS) to assess psychotic symptoms, affective symptoms, and other psychiatric symptoms.
We analyzed global posthospital functioning and adjustment at each of the five follow-ups with the Levenstein-Klein-Pollack scale (LKP) (
37 ). The LKP has been used successfully in previous research (
1,
2,
31,
32,
37,
38,
39 ). The 8-point LKP measures work and social functioning, life adjustment, self-support, major symptoms, relapses, and rehospitalizations within the year before each follow-up. In assessing interrater reliability for the LKP we obtained an intraclass correlation coefficient of r=.92 (p<.01).
The LKP allowed for separation of the sample into three groups. The first group had good global outcome, remission, or recovery within the year before each follow-up, which was demonstrated by a score of 1 or 2 on the LKP, indicating adequate or near-adequate functioning in all areas. Recovery, defined operationally, involves the absence of major symptoms and adequate psychosocial functioning, including instrumental work half-time or more within the year before each follow-up. The second group had moderate impairment, which was demonstrated by a score of 3 to 6 on the LKP, indicating difficulties in some but not all areas of adjustment within the year before each follow-up. The third group had uniformly poor outcome within the year before each follow-up, which was demonstrated by a score of 7 or 8 on the LKP, indicating uniformly poor functioning or poor functioning in almost all areas, including poor psychosocial functioning and severe symptoms. On the basis of a large sample of patients assessed at follow-up we have found a correlation of r=.85 (p<.001) between the LKP and the Global Assessment Scale (
40 ).
To measure rehospitalization, we used the Strauss-Carpenter rehospitalization scales (
41 ), which examined the percentage rehospitalized during the year before each follow-up. We also derived an index based on the sum of the Strauss-Carpenter rehospitalization scales for each patient at each follow-up. This index reflected lengths of hospitalizations for each patient across the 15-year follow-up period. The behavior rating scale of the Psychiatric Assessment Interview was completed at each follow-up (
42 ). This measure provided a negative symptoms scale that assesses flat affect, poverty of speech, and psychomotor retardation, with specific behavioral items quite similar to those used in studies by other research groups (
43 ).
Medications
At the 15-year follow-up, 40 individuals with schizophrenia were taking antipsychotic medications (59 percent), five were taking psychotropic medications that were not antipsychotics (7 percent), and 23 were not taking psychotropic medications (34 percent). There was no significant sex difference in the percentage of individuals with schizophrenia who were taking antipsychotic medications at any of the five follow-ups. At the 15-year follow-up, 13 individuals with other psychotic disorders were taking antipsychotic medications (24 percent); 13 were taking mood stabilizers, antidepressants, or both (24 percent); four were taking other psychotropic medications (7 percent); and 24 were not taking any psychotropic medications (44 percent). For individuals with other psychotic disorders, no significant sex difference occurred in the percentage of individuals who were taking antipsychotic medications at the 15-year follow-up. Similarly, no significant sex difference occurred for individuals with other psychotic disorders, according to whether or not they were taking any psychotropic medications at the 15-year follow-up.
Results
Sex differences in global outcome
To investigate potential sex differences, we analyzed the data on global outcome (LKP scores) for the two sexes and three diagnostic groups at the five successive follow-ups, with a 2 by 3 by 5 mixed-design repeated-measures analysis of variance (ANOVA). The three main factors were the two sexes, the three diagnostic groups, and the five follow-up periods (a within-subjects analysis).
The ANOVA indicated a significant main effect for sex differences on global outcome (F=7.41, df=1, 184, p=.007), a significant main effect for diagnostic differences in outcome (F=19.68, df=2, 184, p<.001), and a significant main effect for time course of global outcome (F=6.66, df=4, 736, p<.001). Mean scores for outcome by sex indicated that women with schizophrenia and women with other psychotic disorders showed better outcomes than parallel samples of men. However, there was no sex difference in outcome for the individuals with nonpsychotic disorders and no interaction between sex and diagnosis. The same results were obtained when we conducted an analysis of covariance, controlling for age at index hospitalization.
Because the overall mixed-design repeated-measures ANOVA reported above included 199 of the 239 participants (that is, those who were assessed at all five follow-ups), we made the analyses more comprehensive by conducting an ANOVA designed to include all participants, even those with missing data at one or more of the five time points. This mixed-effects regression model (
44 ) again showed significant differences in global outcome according to sex (p=.02) and diagnosis (p<.001).
To study sex differences in global outcome for only the combined group of patients with psychotic disorders (schizophrenia and other psychotic disorders), which was the focus of our initial hypotheses, we also conducted a 2 (schizophrenia versus other psychotic disorders) by 2 (sex) by 5 (time period) mixed-design repeated-measures ANOVA, similar to the preceding ANOVA. This analysis included only patients who showed initial vulnerability to psychosis, namely patients who were psychotic at the index hospitalization (98 patients, or 41 percent). The results indicated a significant effect for sex differences (F=5.62, df=1, 94, p=.02) and a significant difference (as expected) for diagnostic group (F=13.93, df=1, 94, p<.001). No significant sex-by-diagnosis interaction effect was found. Overall, there was a significant change in level of functioning over time (F=4.25, df=4, 376, p=.002), indicating general improvement over time for both women and men (
Table 1 ).
Women with schizophrenia had better global outcomes than men with schizophrenia at all five follow-ups over the 15 years. Similarly, women with other psychotic disorders had better outcomes than their male counterparts. On the basis of t tests, women with schizophrenia had significantly better outcome (p<.05) at two of the follow-ups (the two-year and the ten-year follow-ups), and a trend in the same direction (p<.10) at the 7.5-year follow-up. On the basis of t tests, women with other psychotic disorders showed significantly better outcomes at the ten-year follow-up (p<.05) and showed a trend toward having better outcomes (p<.15) at the 7.5- and 15-year follow-ups. For the individuals with nonpsychotic disorders, no significant sex difference was found in global outcome at any follow-up.
To provide more specific information about potential sex differences between individuals with other psychotic disorders and those with nonpsychotic disorders, we also studied sex differences among patients with affective disorders. In this analysis we compared women and men with psychotic affective disorders. We also compared women and men with nonpsychotic affective disorders. There were no significant sex differences in either of these analyses.
Percentage of patients with poor outcomes
Few studies have focused on sex differences among individuals with schizophrenia who show very poor outcome. We studied sex differences among individuals with schizophrenia with uniformly poor global functioning (LKP scores of 7 or 8) at each of the five follow-ups. We compared the percentage of women and men with schizophrenia with very poor outcome (that is, difficulties in all areas of functioning) with those with moderate impairment or who were in a period of recovery (that is, symptoms or problems in some but not all areas). For men, the percentage of individuals with schizophrenia showing very poor global functioning at any given follow-up ranged from 38 to 59 percent (median=50 percent). For women, the percentage ranged from 24 to 39 percent (median=32 percent). At all five follow-ups, the percentage of women with schizophrenia who showed very poor global functioning was lower than that of men. This difference was significant at the ten-year follow-up (χ 2 =4.70, df=1, p<.05).
Percentage of patients showing interims or periods of recovery
To provide information about the percentage of women and men who showed periods of recovery, we divided the sample according to which patients had shown very good global outcome, or intervals of recovery, throughout at least two different assessment years over the course of the 15 years. As shown in
Table 2, a larger percentage of the women with schizophrenia showed at least two periods of recovery, compared with a comparable group of men (eight women, or 40 percent, compared with ten men, or 23 percent). However, this comparison was not statistically significant (p<.15), although the percentages of women and men fell in the predicted directions. A similar pattern emerged for individuals with other psychotic disorders, in that a higher percentage of women in this group had periods of recovery on at least two follow-up assessments (20 women, or 69 percent, compared with 13 men, or 48 percent). Again, this difference was not statistically significant (p=.11). Generally, among individuals with other psychotic disorders, those with the highest recovery rates tended to be those with unipolar psychotic depression among both women and men.
Sex differences in rate of rehospitalization
To assess sex differences in duration of psychiatric hospitalization across the 15-year follow-up period, we compared the percentage of men and women who had been rehospitalized at any point during the year preceding each follow-up. For the entire sample, the percentage of women hospitalized during the year preceding each of the follow-ups ranged from 12 to 31 percent (median=15 percent) and the range for men was 21 to 40 percent (median=30 percent). The sex differences were significant at the 7.5-year follow-up (χ 2 =6.40, df=1, p<.01) and the ten-year follow-up (χ 2 =12.09, df=1, p<.001). There was also a trend toward a significant difference at the 4.5-year follow-up (p=.06). At each follow-up, the percentage of men rehospitalized was higher than that of women.
For individuals with schizophrenia, the percentage of women hospitalized during the year preceding each follow-up ranged from 14 to 35 percent (median=29 percent), and the percentage of men ranged from 32 to 59 percent (median=48 percent). No women with schizophrenia had been rehospitalized at all five follow-ups, whereas 13 percent of the men were rehospitalized at all five follow-ups. The sex differences were statistically significant at the 7.5-year follow-up (χ 2 =7.36, df=1, p=.01). There was also a nonsignificant trend at the two-year follow-up (p<.08) and the ten-year follow-up (p<.13).
For individuals with other psychotic disorders, the percentage of women hospitalized during the year preceding each follow-up ranged from 14 to 31 percent (median=19 percent), and the percentage of men ranged from 15 to 40 percent (median=27 percent). There was no significant sex difference at any of the five follow-ups.
For individuals with nonpsychotic disorders, the range was 6 to 31 percent for women (median=13) and 10 to 23 percent for men (median=18 percent). The sex difference was statistically significant only at the ten-year follow-up (χ 2 =3.7, df=1, p<.05).
We also used the Strauss-Carpenter rehospitalization scales throughout the 15 years to compare psychotic and nonpsychotic women and men. These scales were designed to combine information on frequency and length of rehospitalization. The results indicated significant sex differences (t=2.84, df=158, p<.01) among psychotic patients (schizophrenia and other psychotic disorders). Women in this group had significantly shorter hospitalizations. However, there was no sex difference among individuals with nonpsychotic disorders.
Sex differences in negative symptoms
We also analyzed potential sex differences in the course of negative symptoms over time. The data indicated that at the two-year follow-up, one woman with schizophrenia (8 percent) and four men with schizophrenia (14 percent) showed evidence of negative symptoms. By the time of the 15-year follow-up, six women (35 percent) and eight men (27 percent) showed evidence of negative symptoms. Thus both women and men displayed an increase in negative symptoms over time. However, this finding needs replication and should be interpreted with caution because the samples were small.
Discussion
This research examined whether sex differences in outcome and recovery occur among individuals with schizophrenia and other psychotic disorders. This research is important because previous research on sex differences has not produced uniform results. Previous studies have not usually focused on whether this pattern is unique to schizophrenia. The study reported here addressed whether sex differences in course, outcome, and recovery occur and whether they are specific to schizophrenia or whether they are also characteristic of individuals with other psychotic disorders and nonpsychotic disorders.
Many studies on sex differences do not control for age of onset, and there is evidence that early onset of schizophrenia is a negative prognostic indicator (
45,
46,
47,
48 ). Age of onset is important because older age before a first psychotic break allows time for development of greater knowledge, social skills, and experience. In addition, older age at first break may suggest greater internal resiliency. We controlled for age of onset by assessing a sample of patients who were young when first evaluated and who were most often (76 percent) undergoing their first or second hospitalization. Although the mean ages for the women and men with schizophrenia were almost identical, the women with schizophrenia in this sample included many whose first break occurred early for women, because many women with schizophrenia experience their first break after age 26. In our study, this sample of women was compared with men with schizophrenia whose first break occurred at an age more typical for men. Thus the current comparison involves more seriously ill women with schizophrenia than is usual in many studies, which makes the differences in outcome between the sexes even stronger.
Overall, the current data show a significant tendency for women with schizophrenia and other psychotic disorders to have better global outcomes than their male counterparts. Although the analyses did not produce uniformly significant results in each area at each follow-up, many sex differences were significant. All analyses showed that the men's outcome was poorer than the women's outcome. However, the differences between women and men with schizophrenia were only moderate, rather than very large. Thus the men with schizophrenia generally showed poorer outcomes than women, with a significant overall F test, but the differences were not statistically significant at every assessment.
Many men with schizophrenia showed persistent trends toward uniformly poor outcome, and very few men with schizophrenia showed complete recovery at all five follow-ups. Only a small number of women with schizophrenia showed complete recovery at all five follow-ups, although fewer of these women than their male counterparts showed a consistent trend toward uniformly poor outcome over time. The individuals with other psychotic disorders also showed moderate trends toward sex differences, with women with other psychotic disorders showing better global outcomes than men with other psychotic disorders. Compared with the men with other psychotic disorders, a larger percentage of their female counterparts showed periods of recovery or moderate impairment, and fewer of these women showed very poor global outcome.
In contrast to the outcome data for the two psychotic patient groups, (schizophrenia and other psychotic disorders), the data for individuals with nonpsychotic disorders showed no or few sex differences. The data suggest these differences primarily apply to psychotic patients, and not to those with nonpsychotic disorders.
Several factors contribute to the data showing poorer global outcome for men with schizophrenia than for women with schizophrenia. Environmental factors affect global outcome. These include familial and cultural characteristics and expressed emotion—that is, critical overinvolvement (
49 ). Additionally, men are subject to greater social and vocational expectations than women (and expectations for greater vocational success), possibly resulting in more external stressors. In contrast, women may be better trained in social skills (
50,
51 ). Severely ill patients who also have poor social skills may be at a greater disadvantage than severely ill patients who have at least partially adequate social skills. Women with schizophrenia more frequently marry than men with schizophrenia (
52 ). Data from the 15-year follow-up with our sample provide support for this finding (χ
2 =4.01, df=1, p<.05). Women also remain closer to their social support systems as they experience their illnesses.
Genetic research suggests that familial transmission of schizophrenia may be more frequent among women than among men (
53 ). In addition, sex differences may be affected by structural brain differences (
54,
55 ). Sex differences in premorbid functioning (
56 ) are also important. Substance abuse, more prevalent among men than among women, probably contributes to sex differences in outcome (
57,
58 ). Hormonal differences are also important. Seeman and colleagues (
57,
59 ) and Akhondzadeh and colleagues (
60 ) have proposed that estrogen inhibits postsynaptic dopamine transmission and thereby serves as a "natural" antipsychotic. Furthermore, psychopathology has been shown to improve when estradiol levels rise, and vice versa (
61,
62 ). Other hormones may contribute to psychopathology and outcome (
63 ).
Conclusions
This study indicates that female patients with psychotic disorders (schizophrenia and other psychotic disorders) show a significant tendency to have better outcomes than their male counterparts. Furthermore, individuals with other psychotic disorders show sex differences that are as equally robust as those with schizophrenia. These sex differences were not statistically significant in all analyses, but in all cases, the direction was such that women with schizophrenia and other psychotic disorders had better global outcomes and more periods of recovery than parallel samples of men. The current data add a new dimension in suggesting that sex differences are not specific to schizophrenia but rather occur among patients with psychotic disorders in general.
Acknowledgments
This research was supported, in part, by U.S. Public Health Service grants MH-26341 and MH-068688 from the National Institute of Mental Health. The authors thank Robert Gibbons, Ph.D., and Subhash Aryal, M.S., for statistical consultation.