- Top of page
Abstract As many as 96% of all residents of Taiwan have been enrolled in the National Health Insurance (NHI) program since 1996. The NHI database was used to examine the prevalence and incidence of schizoprenia. The National Health Research Institute provided a database of 200 432 random subjects, about 1% of the population, for study. By means of exclusion criteria, a random sample of 136 045 subjects as a fixed cohort dated from 1996–2001 was obtained. Those study subjects who had at least one service claim during these years for either ambulatory or inpatient care, with a principal diagnosis of schizophrenia, were identified. The cumulative prevalence increased from 3.34 per 1000 to 6.42 per 1000 from 1996 to 2001. The annual incidence density decreased from 0.95 per 1000/year to 0.45 per 1000/year from 1997 to 2001. Male subjects had higher treated prevalence in younger age groups than did female subjects. Higher prevalence was associated with the 25–44 and 45–64 age groups, insurance amount less than US$640, the eastern region, and suburban areas. Lower incidence was associated with the 45–64 age group. Higher incidence was associated with insurance amount less than US$640, and the eastern region. According to the trends of cumulative prevalence and incidence density, the treated prevalence and incidence rate will be approximate to community rates gradually. Most persons with schizophrenia had received treatment in Taiwan after the NHI program was implemented. Future studies should focus on outcome and cost evaluation.
- Top of page
Schizophrenia is the most severe mental disease, and it results in a great economic burden all over the world. In the US, schizophrenia accounts for 2.5% of all health care expenditures, and total costs amount to US$50 billion annually.1 Moreover, about 75% of persons with schizophrenia cannot work and are unemployed.1 Many studies have focused on the prevalence and incidence of schizophrenia.2,3 However, there are great variations in prevalence and incidence rates, depending on different diagnostic standardization, sampling methods, and demographic factors. Generally, point prevalence rates of schizophrenia ranging 1–17 per 1000 population was found.4 In the US, the lifetime prevalence of schizophrenia is about 1%, and only half of all patients with schizophrenia obtain treatment.1 The annual incidence rate reviewed was 10–58 new cases per 100 000 population.4
According to the results of previous studies, the rates of treated prevalence peak in the 45–64 age group, and the age group of highest incidence rate for men was younger than that of women.5 Immigrant groups were found to have higher prevalence and incidence of schizophrenia, possibly because of greater stress and high vulnerability.6 The incidence of schizophrenia has been associated with race and ethnicity.7 The incidence of schizophrenia in urban areas has been found to be higher than that in rural areas.7,8 Persons with schizophrenia are concentrated in urban areas of the poorest living conditions, which possibly indicates that industrialization has some effect on the onset and chronicity of schizophrenia. Schizophrenia is more prevalent among persons of lower socioeconomic status (SES). The ‘social drift hypothesis’ proposes that persons with schizophrenia are unable to compete for resources, resulting in downward social mobility.
In a previous census survey in Taiwan, which covered the period from 1946 to 1948, the lifetime prevalence rate for schizophrenia was 0.22%,9 and the rate had decreased to 0.14% when a survey was performed in the same three areas 15 years later; from 1961 to 1963.10 Persons with schizophrenia may have a higher mortality rate than the general population. Another community survey, the Taiwan Psychiatric Epidemiological Project conducted from 1981 to 1986, used the Chinese modified Diagnostic Interview Schedule as a standardized tool. The results revealed that the lifetime prevalence rate of schizophrenia was 0.27%.11 There has been no updated study of the prevalence and incidence of schizophrenia in Taiwan during the past 15 years.
Several epidemiological methods for evaluating schizophrenia are available, including case register studies to evaluate the treated prevalence and incidence, the key informant method to search for community case rates, and field surveys to identify all cases.5 Although the field survey is the most comprehensive method, schizophrenia is a disease with a low prevalence rate. Because of practical and economic factors, it would be difficult to perform such a large community survey. Thus, several studies have used health care registration data to analyze the epidemiology of schizphrenia.5,6,12–14
Taiwan implemented a National Health Insurance (NHI) program in March 1995, offering a comprehensive, unified, and universal health insurance program to all citizens. Therefore, a fair share of risk-pooling for NHI should be expected. All citizens who have established a registered domicile for at least 4 months in the Taiwan area should be enrolled in NHI. The Bureau of NHI (BNHI) has contracted with most medical institutions in Taiwan. As many as 96% of the people in Taiwan have joined the NHI program since 1996.
In this study, we first present the prevalence and incidence of schizophrenia in Taiwan, based on NHI data from 1996 to 2001. Second, we discuss the associated factors of prevalence and incidence in schizophrenia. Because schizophrenia is the most chronic and disabling mental illness, we can offer evidenced-based data for future policy making and resource allocation.
- Top of page
Table 1 shows demographic distribution of the study sample, including age, sex, race, insurance amount, region, and urbanicity.
Table 1. . Demographic distribution of study sample
| 15–24|| 31 879|| 23.4|
| 25–34|| 32 178|| 23.7|
| 35–44|| 29 246|| 21.5|
| 45–54|| 16 339|| 12.0|
| 55–64|| 12 944|| 9.50|
| ≥65|| 13 459|| 9.90|
| Female|| 66 921|| 49.2|
| Male|| 69 124|| 50.8|
| Aborigine|| 1 942|| 1.40|
| Non-aborigine†||132 274|| 98.6|
|Insurance amount (US$)|
| ≥1281 (≥40 000 NTD)|| 20 462|| 15.3|
| 640–1280 (20,000–39 999 NTD)|| 33 128|| 24.7|
| <640 (<20 000 NTD)|| 59 801|| 44.7|
| Fixed amount‡|| 20 514|| 15.3|
| North|| 60 782|| 45.6|
| Center|| 25 284|| 19.0|
| South|| 43 470|| 32.6|
| East§|| 3 692|| 2.80|
| Urban|| 67 663|| 50.8|
| Suburban|| 22 076|| 16.6|
| Rural¶|| 43 489|| 32.6|
Table 2 shows the cumulative prevalence from 1996 to 2001 and annual incidence density of schizophrenia from 1997 to 2001 in NHI enrollees. The 1-year prevalence was 3.34 per 1000 in 1996, and the cumulative prevalence increased to 6.42 per 1000 in 2001. The annual incidence density decreased from 0.95 per 1000 per year in 1997 to 0.45 per 1000 per year in 2001. Figure 1 illustrates the increasing trend in cumulative prevalence and the decreasing trend in annual incidence density.
Table 2. . Cumulative prevalence and incidence density of schizophrenia, 1996–2001
|Year||Prevalent cases||Cumulative prevalence (per 1000)||Incident cases||Person-year||Annual incidence density (per 1000)|
|1998||666||4.90|| 84||133 907.5||0.63|
|1999||754||5.54|| 88||132 864.0||0.66|
|2000||814||5.98|| 60||131 822.5||0.46|
|2001||873||6.42|| 59||130 791.0||0.45|
Figure 1. Cumulative prevalence and annual incidence density of schizophrenia, 1996–2001. The cumulative prevalence increased from 3.34 per 1000 to 6.42 per1000 from 1996 to 2001. The annual incidence density decreased from 0.95 per 1000/year to 0.45 per 1000/year from 1997 to 2001.
Download figure to PowerPoint
Table 3 shows age and sex-specific prevalence during 1996, the 6-year prevalence, and incidence density of schizophrenia in NHI enrollees from 1997 to 2001. In the 1-year prevalence, the higher rates for males were in the 25–34 (5.35 per 1000) and 35–44 (4.65 per 1000) age groups, whereas, the higher rates for females were in the 35–44 (4.17 per 1000) and 45–54 (4.90 per 1000) age groups. In the 6-year prevalence analysis, the higher rates for males were in the 25–34 (9.11 per 1000) and 35–44 (7.87 per 1000) age groups, whereas, the higher rates for females were in the 35–44 (7.31 per 1000) and 45–54 (7.10 per 1000) age groups. In the incidence density analysis, the higher rates for males were in the 25–34 (0.76 per 1000 per year) and 65 and over (0.84 per 1000 per year) age groups, whereas, the higher rate for females was in the 65 and over (0.81 per 1000 per year) age group.
Table 3. . Age-specific and sex-specific prevalence in 1996, 6-year cumulative prevalence, and incidence density (1997–2001)
|Age (years)||Prevalence in 1996 (per 1000)||6-year cumulative prevalence (1996–2001) (per 1000)||Incidence density (1997–2001) (per 1000 per year)|
Table 4 shows the logistic regression of factors associated with prevalence of schizophrenia in 1996. Higher prevalence was associated with the 25–44 (odds ratio [OR], 2.21; 95% confidence interval [CI], 1.70–2.86) and 45–64 age groups (OR, 1.54; 95% CI, 1.14–2.09), insurance amount less than US$640 (OR, 2.36; 95% CI, 1.50–3.71), and a fixed amount (OR, 13.74; 95% CI, 8.85–21.33), the eastern region (OR, 1.74; 95% CI, 1.09–2.78), and suburban areas (OR, 1.33; 95% CI, 1.03–1.71). There was no significant difference associated with sex or race in the 1-year prevalence of schizophrenia.
Table 4. . Logistic regression model of factors associated with prevalence of schizophrenia in 1996 and Cox regression model of factors associated with incidence of schizophrenia from 1997 to 2001
|Variable||Logistic regression model||Cox regression model|
|Odds ratio||95% CI||Hazard ratio||95% CI|
| 15–24|| 1.00||–||1.00||–|
| 25–44|| 2.21§||1.70–2.86||1.11||0.88–1.42|
| 45–64|| 1.54‡||1.14–2.09||0.72†||0.53–0.99|
| ≥ 65|| 0.44‡||0.27–0.72||0.89||0.63–1.25|
| Female|| 1.00||–||1.00||–|
| Male|| 1.18||0.98–1.42||1.06||0.88–1.29|
| Aborigine|| 1.00||–||1.00||–|
| Non-aborigine|| 2.25||0.96–5.25||1.04||0.54–1.99|
|Insurance amount (US$)|| |
| ≥1281|| 1.00||–||1.00||–|
| 640–1280|| 1.51||0.91–2.49||1.16||0.70–1.92|
| <640|| 2.36§||1.50–3.71||2.56§||1.65–3.98|
| Fixed amount||13.74§||8.85–21.33||7.83§||5.05–12.15|
| North|| 1.00||–||1.00||–|
| Center|| 1.02||0.80–1.32||1.00||0.76–1.31|
| South|| 0.92||0.73–1.14||1.07||0.86–1.35|
| East|| 1.74†||1.09–2.78||1.99‡||1.26–3.13|
| Urban|| 1.00||–||1.00||–|
| Suburban|| 1.33†||1.03–1.71||1.20||0.91–1.58|
| Rural|| 1.12||0.90–1.40||1.12||0.89–1.40|
Table 4 also shows the Cox regression analysis of factors associated with the incidence of schizophrenia from 1997 to 2001. A lower incidence was associated with the age 45–64 group (hazard ratio [HR], 0.72; 95% CI, 0.53–0.99). Higher incidence was associated with an insurance amount less than US$640 (HR, 2.56; 95% CI, 1.65–3.98) and a fixed amount (HR, 7.83; 95% CI, 5.05–12.15), and the eastern region (HR, 1.99; 95% CI, 1.26–3.13). There was no significant difference associated with sex, race, or urbanicity in the incidence of schizophrenia during 1997–2001.
- Top of page
Our study is the first to have used NHI data to detect the prevalence and incidence of schizophrenia in Taiwan, nor have there been previous studies that have discussed the incidence of schizophrenia in Taiwan. Because 96% of the Taiwanese population joined the NHI program, we have reason to believe that the treated prevalence and incidence data presented here are approximate to the true distribution of schizophrenia in Taiwan. Moreover, we analyzed a fixed cohort to observe the study sample for 6 years from 1996 to 2001.
Our results revealed a 1-year prevalence rate of 3.34 per 1000 and a 6-year (cumulative) prevalence rate of 6.42 per 1000. These data are higher than those found in a community study during the 1980s in Taiwan (lifetime prevalence of 2.7 per 1000), but they are closer to those of a study performed in the 1980s in the US (lifetime prevalence of 9.8 per 1000).17 Why was the treated prevalence in the NHI study higher than that found in the community study? There are several factors that should be considered. First, the community survey may have omitted those persons with schizophrenia who had been admitted to hospitals. Second, social stigma can lead to a denial of symptoms or underreporting of previous episodes, which results in underestimation.5 Third, the 15-year gap, different sampling methods, and different diagnostic criteria probably contributed to the differences. Fourth, the possibility of an overdiagnosis of schizophrenia in NHI data for insurance application should be taken into consideration. Reliability varies by diagnosis of psychiatric disorders and is highest for diagnosis of psychosis.18 One study has revealed a high level of accuracy, about 86.8%, of the diagnosis of schizophrenia in Medicaid claims.19 To examine accuracy of the diagnosis of schizophrenia in NHI, we used a diagnostic algorithm for subjects identified as schizophrenia who also had other major psychiatric diagnoses. Priority of judgment of the diagnosis was done according to diagnosis of catastrophic illness registration, inpatient diagnosis (by frequency), ambulatory diagnosis (by frequency), and whether they had been diagnosed by a psychiatrist. Overall, 89% of prevalent cases of schizophrenia during 1996–2001 were diagnosed by psychiatrists, and 60% of these cases had a catastrophic illness registration card. So, it reveals that we have higher accuracy of diagnosis of schizophrenia in this study.
The annual incidence density was 0.95 per 1000 per year in 1997 and decreased to 0.45 per 1000 per year in 2001. In comparison to studies published by the World Health Organization, which reported a range of incidence between 0.1 per 1000 population and 0.6 per 1000 population in other countries,4 we found the incidence of schizophrenia to be higher in NHI study in Taiwan. We propose that, because Taiwan implemented the NHI program in 1995, more persons with schizophrenia first sought treatment during the initial years of our data. Given the decreasing trend during these years, we believe that the treated incidence will gradually decrease to a rate approximate to the general incidence found in community surveys in other countries.
With respect to age- and sex-specific rates of prevalence, we found that males aged 25–44 years and females aged 35–54 years had higher prevalence rates. This finding was in line with those of other studies, that males with schizophrenia have earlier onset and come into contact with services much earlier than do females.5,20 Actually, no data are available for the distribution of the age of onset in schizophrenia among NHI enrollees in Taiwan. Also, based on community studies, no definite data on the distribution of the age of onset in schizophrenia has been reported for Taiwan. According to other studies, the average age onset of schizophrenia is 15–25 years for males and 25–35 years for females.1,4 Females have later age of onset than males. The peak age of initial treatment was older than that of onset age in both sexes, which reveals that most patients sought service several years after onset. A 2-year study in Germany reported that the earliest signs of mental disturbance occurred 4.5 years prior to first admission, on avearage.21 Thus, we must emphasize the need for early detection and early treatment for patients with schizophrenia.
According to our regression results, prevalence was highest in the age group 25–44 years and lowest in the age group 65 and over, and the incidence was lowest in the age group 45–64 years. This indicates that most patients with schizophrenia have onset and seek treatment during early adulthood, which is consistent with other studies. That the lowest prevalence was in older people may mean that persons with schizophrenia have shorter life spans than the general population or that older people with schizophrenia received inadequate treatment in 1996. Most studies have shown that there is no significant difference between males and females in the prevalence and incidence of schizophrenia.1 Our study revealed a similar result overall, except in age- and sex-specific distribution.
Aborigines constitute 2% of the Taiwanese population, however, they represented only 1.4% of the insured people in the NHI program. Thus, a large proportion of aborigines did not join the NHI program. Some studies have found a greater prevalence of schizophrenia in immigrant minorities,6,7 whereas non-immigrant minority groups have no greater risk for schizophrenia than the general population. Our study revealed no significant difference between aborigines and nonaborigines in prevalence and incidence in Taiwan, which agrees with the above conclusion, in that aborigines in Taiwan are non-immigrant minority groups. However, this question should be further investigated when more aborigines have joined the NHI program.
In considering SES, our findings show that schizophrenia is more prevalent in persons with lower SES. Simultaneously, lower SES is a risk factor for schizophrenia, because it has a higher incidence. Therefore, we support both the ‘social drift hypothesis’ and the ‘poverty causation theory.’ Persons with schizophrenia are forced into downward social mobility and more new cases are caused by lower SES with its attendant psychosocial stress. The causes and result interact each other, which makes conditions more complicated.
With respect to regional differences, this may represent resource allocation and equity of psychiatric services. There was no significant difference among regions in the prevalence and incidence of schizophrenia, except that people living in the eastern region had a higher prevalence and incidence of schizophrenia. However, the two largest mental hospitals in Taiwan (several thousand beds) are in the eastern region, which may explain this phenomenon. The eastern region is the least industrialized area, so there is no problem of resource allocation for the treatment of patients with schizophrenia in different regions in Taiwan.
Some studies have reported that an increased risk of schizophrenia is associated with living in urban areas.7,22,23 Patients with schizophrenia are concentrated in areas with the poorest living conditions and are at high risk of being homeless. Our study showed higher prevalence in suburban areas. This may indicate that Taiwan has high-density populations even in suburban areas. People who live in suburban areas suffer from greater change and stress.
Because the prevalence and incidence of schizophrenia are lower than those of minor psychiatric disorders and general medical diseases, there are many advantages in using insurance data for the study of schizophrenia, including a large available number for the sample, the saving of time and money needed to perform psychiatric assessments, and longitudinal data to detect incidence and its risk factors. We used random sampling methods to prevent selection bias. However, we still faced some limitations, such as dual diagnoses, the existence of over- and underdiagnoses, the reliability and validity of the secondary data, and the presence of primary or secondary diagnoses.24 We must also consider the implications of comparing rates with other studies, given the different study designs and instruments.