Burden of schizophrenia among Japanese patients: a cross-sectional National Health and Wellness Survey

The study is a secondary analysis of existing data collected in 2019 from the cross-sectional Japan National Health and Wellness Survey (NHWS). The NHWS is a self-administered online survey covering self-reported information on respondent characteristics, disease status, and health outcomes, administered to adult respondents (aged 18 and above) in Japan and 11 other countries.

In Japan, a stratified random sampling framework with quotas approximating the age and gender distribution according to the Japanese census data was implemented to ensure that the NHWS respondents were representative of the general adult population in Japan. The NHWS was granted exemption status upon review by Pearl International Review Board (Indianapolis, IN). All respondents provided informed consent prior to participation. Anonymized data were utilized in this study, and the Juntendo University Ethics Committee assessed and determined that institutional ethics approval could be waived, according to the Ministry of Education, Culture, Sports, Science, and Technology and the Ministry of Health, Labour, and Welfare’s ethical guidelines for medical and health research involving human subjects [19].

Respondents who self-reported a physician diagnosis of schizophrenia were included in this study. Among diagnosed schizophrenia patients, they were divided into subgroups based on their Patient Health Questionnaire (PHQ)-9 score [20], the self-reported experience of sleep disturbances, as well as self-reported experience of anxiety problems.

Demographic characteristics analyzed in this study included age, gender, marital status, level of education, employment status, household income, geographic location, and health insurance status. The general health characteristics included body mass index (BMI), Charlson Comorbidity Index (CCI) [21, 22], smoking status, alcohol use, and exercise behavior. The healthcare resources utilization (HCRU) was assessed as the number of healthcare provider (HCP) visits, the number of emergency room (ER) visits, and the number of hospitalizations in the past 6 months for own medical conditions, as well as the out-of-pocket payment for own prescription medications in an average month.

The lifetime prevalence of schizophrenia in Japan was estimated using the total number of diagnosed schizophrenia patients as the numerator and the total number of NHWS respondents in Japan as the denominator. Demographic and general health characteristics, as well as HCRU, were reported using counts and percentages for categorical variables and means and standard deviations (SDs) for continuous variables, among patients with schizophrenia. Bivariate comparisons between the subgroups concerning these characteristics and health outcomes were conducted using independent t-tests for continuous variables and chi-square tests for categorical variables.

Multivariate analyses included the comparison of outcomes (HRQoL, WPAI, indirect costs) using generalized linear models (GLMs) to control for covariates between schizophrenia patients with different severity of depressive symptoms. GLMs with identity link functions were used for approximately normally distributed outcomes (including HRQoL summary scores, EQ-5D index, and VAS); while GLMs with log link functions were used for outcome variables with skewed distribution (including WPAI measures and indirect costs). Age, gender, region, marital status, level of education, employment status, household income, insurance type, BMI, CCI, smoking status, alcohol use, exercise behavior, and out-of-pocket payment were included in the GLMs as covariates. Estimated adjusted means with 95% confidence intervals (CIs) and p-values for all outcomes were reported. Association between the PHQ-9 categories (< 5, 5–9, 10–14, 15–19, and ≥ 20) and patient health outcomes among schizophrenia patients were further investigated using Kruskal-Wallis H Test. This approach for analysing NHWS data has previously been described in other studies [32].

All statistical analyses were performed using IBM SPSS Statistics Version 25 [33]. P-values of less than 0.05 were considered statistically significant.

The average age of schizophrenia patients was 42.70 and almost 90% of patients were below 65 years old. There were similar proportions of males and females among patients with schizophrenia. A small proportion of patients (29.2%) were married or living with a partner, and 32.6% of schizophrenia patients had a university degree. More than half of the patients were employed and about one-third of the patients were overweight (Table 1).

Demographic and general health characteristics were compared between Japanese schizophrenia patients with milder (PHQ-9 < 10) or more severe depressive symptoms (PHQ-9 ≥ 10). Among the patient characteristics examined, significant differences were only observed in age and marital status – patients with more severe depressive symptoms were younger (38.80 vs. 46.26) and fewer of them were married or living with a partner (20.0% vs. 37.6%) (Table 2).

The bivariate analysis demonstrated that Japanese norm-based MCS and RCS scores of the SF-12v2 were lower in patients with schizophrenia and with PHQ-9 ≥ 10, compared to schizophrenia patients with PHQ-9 < 10 (p <  0.001). No significant differences were observed in PCS score (Table 3). In addition, lower EQ-5D utility scores and VAS scores were reported in schizophrenia patients with PHQ-9 ≥ 10 compared to those with PHQ-9 < 10 (both p <  0.001). Comparing to patients with PHQ-9 < 10, schizophrenia patients with PHQ-9 ≥ 10 had higher levels of impairment in terms of absenteeism, presenteeism, total work productivity impairment, activity impairment and higher indirect costs (all p <  0.05).

After adjusting for age, gender, region, marital status, level of education, employment status, household income, insurance type, BMI, CCI, smoking status, alcohol use, exercise behavior and out of pocket payment, schizophrenia patients with PHQ-9 ≥ 10 had significantly lower MCS (41.75 vs. 50.45, p <  0.001), PCS (44.15 vs. 49.50, p = 0.006) and RCS (26.28 vs. 40.09, p <  0.001) scores, lower EQ-5D utility score (0.56 vs. 0.78, p <  0.001) and VAS score (44.29 vs. 67.77, p <  0.001), compared to schizophrenia patients with PHQ-9 < 10 (Table 3). In terms of work productivity, after adjustment, schizophrenia patients with PHQ-9 ≥ 10 had significantly higher absenteeism (14.22% vs. 1.08%, p = 0.001), presenteeism (38.26% vs. 14.53%, p <  0.001), total work productivity impairment (58.38% vs. 25.75%, p = 0.004) and activity impairment (53.55% vs. 22.82%, p <  0.001), compared to schizophrenia patients with PHQ-9 < 10. Indirect costs due to work productivity impairment were significantly higher for patients with PHQ-9 ≥ 10 than those with PHQ-9 < 10 (1965.1 vs. 892.8 thousand-yen, p = 0.010).

Comparing Japanese schizophrenia patients with PHQ-9 < 14 and PHQ-9 ≥ 14, similar findings were observed (Supplementary Tables 1 and 2).

Further investigation on the association between PHQ-9 and the health outcomes also confirmed that the PHQ-9 score was significantly associated with health outcomes among patients with schizophrenia (Table 4). Specifically, MCS and EQ-5D index scores decreased while presenteeism and total work productivity impairment increased across the PHQ-9 score categories (Supplementary Fig. 1).

No significant differences in terms of baseline patient characteristics were observed between schizophrenia patients with and without experience of sleep disturbances (Supplementary Table 3). Patients with experience of sleep disturbances had significantly more physician visits and hospitalizations in the past 6 months than patients without experience of sleep disturbances.

Bivariate comparisons between schizophrenia patients with and without experience of sleep disturbances demonstrated that those who experienced sleep conditions had lower MCS (43.02 vs. 47.91, p = 0.004) and RCS (29.14 vs. 38.56, p <  0.001) scores, EQ-5D index (0.65 vs. 0.75, p <  0.001) and VAS (46.64 vs. 65.79, p <  0.001) scores compared to patients without experience of sleep disturbances (Table 5). Schizophrenia patients with experience of sleep disturbances also had significantly higher absenteeism (26.11% vs. 7.20%, p = 0.001), presenteeism (53.11% vs. 29.77%, p = 0.001), total work productivity impairment (63.16% vs. 32.00%, p <  0.001) and total activity impairment (53.30% vs. 34.67%, p <  0.001), as well as significantly higher indirect cost (2370.0 vs. 1144.1 thousand yen, p <  0.001).

GLMs were not used for adjustment as the baseline characteristics were similar between the two groups.

No significant differences in terms of baseline patient characteristics were observed between schizophrenia patients with and without experience of anxiety problems, except for CCI (Supplementary Table 3). Patients with experience of anxiety problems had significantly more physician visits and hospitalizations in the past 6 months compared to patients without experience of anxiety problems.

In the bivariate comparisons, schizophrenia patients with experience of anxiety problems experienced significantly lower RCS score (26.90 vs. 38.03, p <  0.001), as well as EQ-5D index (0.62 vs. 0.75, p <  0.001) and VAS (44.12 vs. 63.51, p <  0.001) scores. In addition, presenteeism (53.61% vs. 33.46%, p = 0.004), total work productivity impairment (63.66% vs. 37.26%, p <  0.001), and total activity impairment (55.91% vs. 36.79%, p <  0.001), as well as indirect cost (2321.0 vs. 1399.0 thousand yen, p = 0.001), were significantly higher among schizophrenia patients with experience of anxiety problems compared to those without experience of anxiety problems.

GLMs were not used for adjustment as the baseline characteristics were similar between the two groups.

It is also worth mentioning that despite the average age (43 years old) of the patients with schizophrenia in Japan, about half were not currently employed. This contributes to the total morbidity cost of schizophrenia in Japan, in addition to the indirect cost incurred by impairment of work productivity for those who are employed. As post-hoc exploratory analysis, the national morbidity cost of schizophrenia was estimated to be 1074 billion Japanese yen, using available data from this study and government-published employment and income data (details of the methodology are described in Supplementary material 2). Cost due to unemployment contributed to more than half of total morbidity cost (552 billion yen; 51.4%).

Our findings demonstrate an estimated schizophrenia lifetime prevalence of 0.59% (95% CI: 0.51%, 0.68%). Local data are aligned to ours as they detected a lifetime prevalence in Japan ranging from 0.19 to 1.79% [3]. Also, Goldner et al., through a global systematic literature review, found a similar lifetime prevalence of 0.55% [34]. It is important to notice that the data have a 6- to 14-fold variation across geographical regions that may be attributed to the different populations’ characteristics (i.e., age, urbanicity, different risks factors for schizophrenia) and/or study design [34, 35]. Therefore, our findings are extremely important as health planners need local data on schizophrenia rates to assist in planning and implementing effective interventions [34]. According to the Ministry of Health, Labour and Welfare patient survey [36], the estimated number of patients with schizophrenia in Japan in 2017 was 793,000, with 639,000 outpatients and 154,000 inpatients. Our estimation of 0.59% was close to the prevalence rate of outpatient adult schizophrenia patients in Japan. This is understandable as institutionalized patients may not have access to online surveys. In addition, most hospitalized patients were older, which corroborates with the relatively younger average age (42.7 years old) of patients with schizophrenia in this study.

On the other hand, a recently published systematic review and meta-analysis reported a pooled estimate of the prevalence of comorbid major depressive disorder in schizophrenia to be 32.6% [15]. Notably, in our study, 47.8% (85 of 178) and 34.3% (61 of 178) of schizophrenia patients had PHQ-9 score at least 10 and 14, respectively.

The indirect cost of schizophrenia estimated in this study is 1.07 trillion Japanese yen. This cost is lower than the estimated value in a previous study, which revealed an estimation of 1.85 trillion Japanese yen morbidity and mortality cost in Japan in 2008 [10]. One possible explanation for the discrepancy is differences in employment rate and wage calculations between the two studies. In our study, the overall employment rate was 51.1% (Full-time: 20.8%, Part-time: 20.2%, Self-employed: 10.1%) (Supplementary Table 4); in contrast, the employment rate reported in Sado et al. was much lower, ranging from 3.8 to 30.6% across gender and age groups [10]. However, it is worth noting that owning to the self-reported nature of the NHWS, it is possible that some respondents might have counted their rehabilitation/other general hardship training as a part-time employment. In addition, although both studies calculated average wage for full-time and part-time employees separately, Sado et al. did not separate wage calculation for regular and non-regular (e.g., self-employment) employees.

The inclusion of inpatients could be an alternative explanation for the discrepancy in indirect costs estimated in Sado et al. and this study. Inpatients are less likely to be included in our study given that they may not have access to online surveys. According to the Labour and Welfare patient survey conducted by the Ministry of Health in 2017, there were about 154,000 inpatients with schizophrenia in Japan [36]. Using the average annual income (around 5 million yen) reported by the Ministry of Health Labor and Welfare [28] as a proxy, the estimated indirect costs for 154,000 inpatients is about 770 billion yen. This estimated value closely resembles the discrepancy of 780 billion yen (i.e., 1.85 vs. 1.07 trillion) between Sado et al and this study. Taken together, this difference in employment rate, the inclusion of inpatients, and wage calculation could potentially lead to the discrepancy in the indirect costs related to productivity loss due to unemployment.

Schizophrenia patients were younger compared to the general population in Japan where around 28% were above 65 years old [37]. This is consistent with global findings where approximately 70.8% of schizophrenia cases occurred in the 25–54 years age group [2]. It is worth noting that, however, the average age of schizophrenia patients in these studies are lower owing to the fact that inpatients, many of whom are elderly over 75 years old [4], may not be included in the study population. Notably, one-third of the patients were overweight. It is important to notice that metabolic syndrome is more likely to occur in schizophrenia patients and psychiatrists should take note of the potential antipsychotic-induced metabolic adverse effects [38, 39]. Additionally, less than half presented at least moderate depressive symptoms (47.8%) and experienced sleep disturbances (49.4%) or anxiety problems (37.1%). The prevalence of comorbid depressive symptoms in our study is aligned to the prevalence reported in the systematic literature review of Etchecopar-Etchart et al. (2021) which ranged from 16.3 to 69% [15]. When compared to the data demonstrated by the meta-analysis of McEnery et al., our rates on anxiety symptoms are greater, possibly due to demographic characteristics, socioeconomic status, cultural factors, differences in assessment and diagnostic measures, duration of illness and mixed population [17]. Also, it is important to state the limitation for the diagnosis of anxiety once the disturbance is not better explained by another mental disorder, in this case schizophrenia patients [40]. However, our study reported almost identical rates (38.3%) to the data evidenced by the meta-analysis of Achim et al. [16]. Finally, the rates of sleep disturbances in our study are similar to those found by Wang et al. (58%), which is much higher than the general population [41, 42].

Our study demonstrated that schizophrenia patients with more severe depressive symptoms were younger and were more often not married in Japan. The unadjusted and adjusted analysis showed that schizophrenia patients with more severe depressive symptoms, compared to those with milder depressive symptoms, had lower HRQoL, higher levels of absenteeism, presenteeism, total work productivity and activity impairment, and twice more indirect costs. We also found an association between severity of depressive symptoms and health outcomes. Depressive symptoms, such as social withdrawal and reduction in work and activities [9], can have severely negative impact on patients with schizophrenia and thus poses a challenge for their social reintegration. Moreover, the psychopathological, social, and health impairment posed by schizophrenia can be worsened by depressive symptoms, which can lead to degradation in patient wellbeing [9, 43]. Last but not least, it has been reported that usage of relapse-related mental health services is higher in schizophrenia patients with severe depressive symptoms and they also had a significantly higher rate of ‘suicidality’ than those without depressive symptoms [6, 9, 44].

Our study found that schizophrenia patients who experienced sleep disturbances had significantly severe depressive symptoms and consequently, more physician visits and hospitalizations in the past 6 months, compared to patients who did not experience sleep disturbances. Also, schizophrenia patients who experienced sleep disturbances had lower HRQoL, higher levels of absenteeism, presenteeism, total work productivity, and activity impairment, and twice more indirect costs, compared to schizophrenia patients who did not experience sleep disturbances.

Fang et al. similarly demonstrated that schizophrenia patients had sleep characteristics significantly different from healthy controls, such as longer sleep onset latency and an average length of awakening, and higher number of awakening and total activity counts [45]. A possible explanation is that these patients have a delta wave deficit reflecting a possible thalamocortical dysfunction [46] or the antipsychotic used [44].

The evidence of our study of lower HRQoL in schizophrenia patients with sleep disturbances is similar to previous studies [47, 48]. Sleep disturbances in schizophrenia patients were reportedly associated with lower HRQoL, depressive symptoms, overall distress, medications’ adverse events and mental and somatic discomfort than those who had no sleep complaints [47], and reduced coping strategies to manage a stressful event [48]. Besides, more than half of schizophrenia patients had sleep disturbances and it was significantly associated with depression and coping mechanisms, and these factors were related to impaired quality of life [41]. Those impaired mechanisms may be explained by the cognitive and memory deficit faced by these patients, which lead to high rigidity, consequently harmful attitudes and impulsive solutions, such as self-injury, influencing sleep quality [49].

It is well known that sleep quality influences work productivity and daily activities [50, 51] and in schizophrenia patients, this had a profound impact on daytime dysfunctions and disability [52]. A study demonstrating the influence of sleep in business found that impaired employee sleep resulted in increased disease burden, health care utilization, absenteeism, and presenteeism, reducing workplace efficiency [53]. Schizophrenia patients have poor sleep quality, and this could impact work productivity levels. Consequently, the indirect costs expressed by morbidity (productivity loss, absenteeism, and presenteeism) increases due to the disease burden, which could also influence the total costs of the disease.

In Australia, the total annual financial cost of inadequate sleep was estimated at US$17.9 billion, of which 68% was comprised of productivity losses largely due to absenteeism and presenteeism [54]. A study in Japan found that subjective sleep quality, daytime dysfunction, use of sleep medicine and sleep disturbances were important variables affecting productivity loss among office employees [51].

Our study evidenced that schizophrenia patients who experienced anxiety problems had more physician visits and hospitalizations in the past 6 months, compared to patients who did not experience anxiety problems. In addition, these patients who experienced anxiety problems had lower HRQoL, higher levels of absenteeism, presenteeism, total work productivity, and activity impairment, and 1.7-fold more indirect costs, compared to schizophrenia patients who did not experience anxiety conditions.

The major clinical problem with schizophrenia and comorbid social anxiety is marked by the fear of acting in a way that may be humiliating or embarrassing [55]. A study with Veterans Health Administration patients with schizophrenia demonstrated that comorbid anxiety disorders were associated with higher rates of psychiatric and medical hospitalization, and increased utilization of outpatient mental health services [56].

Role functioning and quality of life can be seriously affected by social anxiety [57]. Anxiety disorders are associated with increase in level of depression and suicidality, utilization of medical service, and impairment to cognitive and quality of life [14]. Previous studies have reported critical impacts of comorbid social anxiety on social functioning and HRQoL in schizophrenia patients [17]. A worsening in social anxiety and general psychiatric symptoms were associated with a decrease in patient functioning, likely owing to exacerbation of general severity and negative symptoms [14]. The results from the Aikawa et al. study revealed that social anxiety in patients with schizophrenia was significantly correlated with the age of onset, duration of untreated psychosis (DUP), social functioning, psychiatric symptoms, and HRQoL. Furthermore, based on a systematic review, social functioning, gender, age of onset, and DUP were identified as important variables associated with the severity of social anxiety symptoms [8].