Study population

All participants with schizophrenia (n = 510) were identified from an epidemiological investigation of 123 572 people aged 15 years and older in six townships of Xinjin County in March 1994. Participants were identified through screening procedures for psychosis and general psychiatric interview. The details of this investigation have been described in previous papers. ^{Reference Ran, Xiang, Huang and Shan10,Reference Ran, Xiang, Li, Shan, Huang and Li20} All participants lived in rural communities and met ICD–10 criteria ²¹ for a diagnosis of schizophrenia based on standardised administration of the Present State Examination (PSE–9) ^{Reference Wing, Cooper and Sartorius22} by trained research interviewers. Using the baseline data in 1994 we followed up and interviewed all the participants with schizophrenia and their informants in May 2004. All respondents gave informed consent.

Measurements

The principal assessment tools included the PSE–9 and Social Disability Screening Schedule (SDSS) ^{Reference Liu, Wang, Shen and Wang23} in the baseline investigation in 1994. ^{Reference Ran, Xiang, Huang and Shan10,Reference Ran, Xiang, Li, Shan, Huang and Li20} For individuals still alive at the visit in 2004, at least one person familiar with each participant's life and circumstances and/or the participant themselves were interviewed. ^{Reference Ran, Chen, Conwell, Chan, Yip and Xiang7} For participants now deceased, the next-of-kin or at least one person familiar with the person was interviewed. All interviews were conducted by trained psychiatrists using the Patients Follow-up Schedule (PFS) ^{Reference Ran, Chen, Conwell, Chan, Yip and Xiang7} for about 30 min. For all participants, medical and psychiatric treatment records were obtained from hospitals, village doctors' clinics and traditional healers. For participants now deceased, information from the death certification and suicide note, where applicable, was also obtained.

Participants were defined as ‘never-treated’ if the individual and/or informants reported that they had not received any antipsychotic medication since their first episode of schizophrenia. The never-treated participants were classified according to the treatment information collected in 1994 and 2004, which included participant and/or informant reports, treatment records in hospitals, village doctor's clinics and traditional healers.

The classification of each death as a result of suicide or other causes represented the consensus opinion of interviewers and independent researchers after reviewing all information obtained during the interviews. Information from the death certification and suicide note (where applicable) was also obtained. Participants were defined as homeless and lost to follow-up if informants reported that they had wandered and slept in public places and that their whereabouts was unknown. ^{Reference Ran, Chen, Conwell, Chan, Yip and Xiang7} Participants' physical illnesses (e.g. diseases of heart and lung, hypertension and cancer) at some time during the follow-up period were defined according to the informants' report and doctors' diagnoses. Marked symptoms (significant positive and/or negative symptoms, mood symptoms or resulting behavioural disturbance) were defined according to the PSE–9. ^{Reference Wing, Cooper and Sartorius22}

Statistical analysis

The follow-up period for every participant started at recruitment and ended either at interview, death or the point at which they were lost to follow-up. Mortality rates were calculated overall and by subgroups defined according to various characteristics. Mortality rates were estimated using the person–time method (number of deaths divided by person-years of follow-up). The effects of antipsychotic treatment on mortality and suicide rates were tested using univariate Cox hazard regression analyses. Survival analyses were also used to explore treatment differences in survival rates.

Standardised mortality ratios (SMRs) were calculated by dividing observed deaths by expected deaths, with the general population in Xinjin County used as a standard population. Death registration data of Xinjin County were used for the general population. All the variables analysed were based on the measures at baseline or follow-up.

Characteristics of the cohort participants

Of 510 individuals identified as having schizophrenia in 1994, 10 people did not complete the evaluation. The remaining 500 participants (98.0%, 500/510) were available for follow-up (1994–2004), of which 46.6% were male, 64.2% were married, and in 55.8% the family's economic status was less than the mean.

All 500 participants included in 1994 were followed up from 1994 to 2004. Informants were available for all these participants (100%). Information on 305 participants was provided by both the individual and their informants, and information on 195 participants was provided by proxy informants alone.

Differences between never-treated and treated participants

At the end of the follow-up period, 132 participants (or their proxy informants) (26.4%) reported never having received antipsychotic treatment and 368 (73.6%) reported having received antipsychotic treatment for their schizophrenia (Table 1). Among the treated individuals, there were 117 people (31.8%) who had been admitted to a mental hospital, and 133 individuals (36.1%) who had accepted antipsychotic treatment for less than 2 months. Compared with treated participants (traditional Chinese medicine 45.7%, treatment by traditional healers 61.7%), never-treated people had received significantly less traditional Chinese medicine (13.6%) and treatment by traditional healers (34.1%).

The characteristics of never-treated and treated participants are described in Table 2. Compared with treated participants, never-treated participants were significantly older, less likely to be married, had a lower education level and fewer family members. Caregivers of never-treated people were less likely to be a parent or spouse. The never-treated individuals were significantly older at age at onset, more likely to live alone, had longer duration of illness, more marked symptoms and fewer previous suicide attempts.

Current status and mortality

In 2004, as indicated in Table 3, there were no significant differences between never-treated (70.5%) and treated participants (75.8%) in the percentage that had survived. There were also no significant differences between never-treated (5.3%) and treated participants (6.3%) in the proportion that were homeless. The percentage of never-treated individuals who died by suicide (3.0%) was similar to that in those who had been treated (4.6%). Deaths from other causes were significantly more frequent in never-treated (21.2%) than treated (13.3%) people during the follow-up period.

There were no significant differences between male never-treated and treated participants in the percentage surviving at follow-up, who had died by suicide or other causes, or who were homeless. However, deaths from other causes were significantly more frequent in female never-treated (20.0%) than treated (10.4%) participants during the follow-up period. Among never-treated individuals, there were no significant differences between men (44.8%) and women (44.6%) in the percentage who were unable to work. Among treated participants, there were significantly more men (41.6%) with an inability to work than women (32.2%) (χ² = 10.1, d.f. = 2, P<0.01).

The mortality rates and SMR of never-treated and treated participants are shown in Table 4. There was no significant difference between the mortality rate in never-treated and treated individuals using Cox hazard regression analyses.

There were no significant differences in suicide rates between never-treated (345.1 per 100 000 person–years) and treated participants (520.4 per 100 000 person–years) using Cox hazard regression analyses. The SMR for never-treated individuals who died by suicide was 32.5, and for treated individuals 46.7.

There were no significant differences in the mortality rate from other causes (accidental and natural) between never-treated (2415.9 per 100 000 person–years) and treated participants (1199.9 per 100 000 person–years) using Cox hazard regression analyses. The SMR for never-treated individuals who died from other causes was 9.5, and for treated individuals 5.9.

The survival probability for never-treated people in 2004 was 0.71 (95% CI 0.61–0.80). There was no significant difference in survival rate between never-treated and treated participants (survival probability in 2004: 0.76, 95% CI 0.71–0.81) during the 10 years of follow-up (Log-rank test: χ² = 2.13, P>0.05).

Mortality and suicide

The results of Cox hazard regression analyses did not support differences in mortality between never-treated and treated participants with schizophrenia. The results of survival analyses also indicated that there was no significant difference in 10-year survival rates between never-treated and treated individuals. The overall mortality rates in those never-treated and those treated are extremely high, exceeding by 6.5 times the rate observed among people over 15 years old in the general population.

The results indicate that suicidal behaviour is common in never-treated and treated people with schizophrenia. The results of Cox hazard regression analyses also did not support differences in rates of suicide between never-treated and treated individuals. The suicide rate that we observed in never-treated participants is similar to the rate in those who have received treatment. Standardised mortality ratios for suicide were 32.5 in never-treated participants and 46.7 for treated participants.

Why are there no significant differences in total mortality rates between never-treated and treated participants? Given the higher proportion of marked symptoms and longer duration of illness, it could not be because never-treated individuals had less severe illness. Thus, despite the lack of treatment, the mortality in never-treated participants still remained the same as those who received treatment. The reasons may be as follows. First, the mortality of those with schizophrenia may be influenced by multiple factors including antipsychotic medication, family care and physical status. Second, the results indicate that antipsychotic treatment may not reduce the long-term mortality rates and increase survival rates in people with schizophrenia. One study in The Netherlands indicated that there was no significant difference in the suicide rate between placebo and active treatment groups. ^{Reference Storosum, van Zwieten, Wohlfarth, de Haan, Khan and van den Brink24} Third, the effectiveness of antipsychotic treatment may be underestimated because a lot of people in the treated group had not received regular antipsychotic treatment. For example, only 31.8% of participants had been admitted to a mental hospital and 36.1% had accepted antipsychotic treatment for less than 2 months. Fourth, many people in the never-treated group received traditional Chinese medicine (13.6%) and treatment by traditional healers (34.1%), interventions that could potentially influence the outcome of schizophrenia. ^{Reference Rathbone, Zhang, Zhang, Xia, Liu and Yang25} Further studies of never-treated individuals may be helpful to explore the differences.

Although evidence indicates that a significant proportion of treated incident cases of schizophrenia achieve favourable long-term outcomes, ^{Reference Harrison, Hopper, Craig, Laska, Siegel and Wanderling5} certain classes of antipsychotics have been associated with death. ^{Reference Montout, Casadebaig, Lagnaoui, Verdoux, Philippe and Begaud26} Suicide risk among people with schizophrenia-spectrum disorders declines quickly after treatment and recovery. ^{Reference Qin and Nordentoft27} However, the results of this study indicate that there are no significant differences in mortality and rates of suicide between people who had received antipsychotic treatment and those who did not. The results did not support the expectation that antipsychotic drugs could reduce the long-term mortality rates in these individuals. The long-term mortality of never-treated participants is similar to, if not higher than, the mortality of treated participants with schizophrenia.

In general, older patients are much more likely to die. ^{Reference Palmer, Pankratz and Bostwick28} Evidence indicated that the mortality rate was significantly higher among individuals with later onset of schizophrenia (>45 years) than those with age at onset before 45 years of age. ^{Reference Ran, Chen, Conwell, Chan, Yip and Xiang7} However, although people with a later onset may have a more benign course of illness, symptom severity and cognitive deficits may be similar in participants with early-onset and late-onset schizophrenia. ^{Reference Jeste, Harris, Krull, Kuck, McAdams and Heaton29} The results of this study indicated that there were no significant differences in mortality rates between never-treated and treated participants even though never-treated people were older and had a later onset of illness than treated individuals.

The results of our study indicate that never-treated people are more likely to be older, unmarried, be of an older age at onset, longer duration of illness, fewer family members, more marked symptoms and accept less support than treated individuals in rural China. All these possible risk factors of never-treated participants identified in this study reflect the influence of both socioeconomic characteristics of rural China and the clinical characteristics of these participants. It is striking that there are no differences in mortality rates even though the untreated group is associated with a range of significant risk factors: more likely to be older, less likely to be married, have less social support and more marked symptoms. The influence of socioeconomic characteristics of participants with schizophrenia on mortality and suicide risk needs further study.

Given the similar mortality rates between never-treated and treated participants, we suggest that antipsychotic treatment may actually be ineffective in reducing mortality. However, the results of this study indicate that antipsychotic treatment might reduce long-term symptom severity. The results signified the importance of medication on reducing the psychotic symptoms. Differences in symptom severity might have an impact on other dimensions of outcome such as social function. The impact of medication on social function needs further study.

Our results indicate that there were no significant differences between male never-treated and treated participants in the percentage of survivals, suicide, deaths from other causes and homelessness. However, female never-treated participants had a higher percentage of deaths from other causes than treated participants. Male treated participants had a worse ability to work than female treated participants. Differences between genders regarding medication needs further study.

Other characteristics

Evidence indicates that the longer the psychotic symptoms continue unchecked by medication, the greater the likelihood of profound clinical deterioration. ^{Reference Black, Peters, Rui, Milliken, Whitehorn and Kopala30} The results of the present study indicate that never-treated participants have significantly more marked symptoms, consistent with a previous study in Bali in which never-treated participants showed significantly higher total Positive and Negative Syndrome Scale (PANSS) scores than did those in the treated group. ^{Reference Kurihara, Kato, Reverger, Tirta and Kashima11} Our results may support the possibility that the severity of symptoms remains the same in untreated individuals irrespective of the duration of illness. ^{Reference Tirupati, Padmavati, Thara and McCreadie19}

The results of this study indicate that never-treated participants may be associated with lower family economic level and fewer caregivers in rural China. ^{Reference Ran, Chen, Conwell, Chan, Yip and Xiang7,Reference Ran, Xiang, Conwell, Lamberti, Huang and Shan31} The small number of family members, reflecting the nuclear family, may be a risk factor related to non-treatment of individuals with schizophrenia that is different from a study in India in which the larger extended/joint family seemed to be a crucial factor associated with non-treatment. ^{Reference Padmavathi, Rajkumar and Srinivasan9} Caregivers were less likely to be a parent or spouse in never-treated participants which might also be a risk factor related to non-treatment of these individuals. Never-treated people were more likely to be older in age and ill for a longer duration than those who had been treated, which is consistent with the Indian study. ^{Reference Padmavathi, Rajkumar and Srinivasan9} We suggest that the traditionally supportive family network may be broken down by prolonged illness and poor clinical status. ^{Reference Ran, Xiang, Li, Shan, Huang and Li20,Reference Gureje and Bamidele32} The role of families needs further study.

Evidence indicates that people with schizophrenia have high rates of potentially reversible medical morbidity that increase mortality as well. ^{Reference Green, Canuso, Brenner and Wojcik33,Reference Goff, Cather, Evins, Henderson, Freudenreich and Copeland34} The results of this study indicate that there were no significant differences in physical illness between never-treated and treated individuals.

Implications for services

Our results have implications for reducing mortality and suicide among never-treated and treated people with schizophrenia in China and elsewhere. The characteristics of these individuals should be taken into account when developing interventions to prevent mortality. Resources and services for mental disorders are insufficient considering the burden caused by these disorders around the world. ³⁵ Long-term outcomes of schizophrenia may be worsened as the absence of mental health services delays treatment. ^{Reference Oosthuizen, Emsley, Keyter, Niehaus and Koen36} We suggest that treatment including antipsychotic medication and other interventions (such as traditional Chinese medicine) may improve outcomes for untreated individuals even though they have been ill for many years. ^{Reference Tirupati, Rangaswamy and Raman37} Given the limited resources in contemporary China, prevention programmes should emphasise community-based mental healthcare to provide earlier diagnosis, antipsychotic treatment, treatment of comorbid medical conditions, function rehabilitation and family support. Given severe stigma associated with psychiatric illness, ^{Reference Xiang, Ran and Li18} efforts to reduce stigma in the community will be necessary for individuals with schizophrenia to be accepted by the community again and interventions made to decrease their mortality rate.

The results of our long-term studies among people with schizophrenia challenge the axiom in international psychiatry that schizophrenia has a better course and outcome in low- and middle-income countries. ^{Reference Mueser and McGurk1,Reference Isaac, Chand and Murthy2} Given the high rates of mortality, including suicide, homelessness and never-treated people with schizophrenia in low- and middle-income countries, it is premature to come to this conclusion if withdrawals or attrition due to death and homelessness and the outcome of many never-treated participants are not included in follow-up analyses. ^{Reference Ran, Chen, Conwell, Chan, Yip and Xiang7,Reference Cohen, Patel, Thara and Gureje8,Reference Ran, Chan, Chen, Xiang, Caine and Conwell38} It is time to re-examine presumed wisdom about schizophrenia prognosis in low- and middle-income countries. ^{Reference Ran, Chen, Conwell, Chan, Yip and Xiang7,Reference Cohen, Patel, Thara and Gureje8}

Given the representative sample used in our study, we are confident that our findings are generalisable to the population of people with schizophrenia in rural areas, and even other low- and middle-income countries that have a similar social environment. Overall, mortality, suicide and homelessness are serious concerns in never-treated and treated individuals with schizophrenia in rural China. It is crucial to supply the necessary community mental health services and medication for these people in rural China.

Limitations

Possible misclassification of never-treated and treated participants, and of suicide, may exist due to recall bias. Discrimination concerning suicidal behaviour and lack of coroners' reports may have also had an impact on the study findings. The mortality and suicide rates may be underestimated because most homeless individuals were lost to follow-up. Given the diversity of sociocultural, economic and care provision characteristics, the results of this rural China study may not generalise to high-income countries.