Mortality in schizophrenia: A study of verbal autopsy from cohorts of two rural communities of South India.

BACKGROUND: Schizophrenia is a life-shortening disease. Although the rate of mortality of persons with schizophrenia in India is established to be more compared to that in the general population, there is a little exploration of the causes for the same. AIM: The aim of this study is to explore the causes of death in two rural cohorts of schizophrenia.
MATERIALS AND METHODS: In-person interviews of primary caregivers of 53 deceased persons with schizophrenia were conducted using the World Health Organization's verbal autopsy 2014 instrument. Physician-based method was used to determine the causes of death.
RESULTS: Average age of 53 deceased schizophrenia patients was 50.45 ± 13.65 years with almost equal gender ratio. Just more than two-thirds of patients were married, just more than one-third are educated up to primary school and also had no formal education each. Noncommunicable diseases (NCDs) were the most common causes of death (30, 56.6%) in this sample, followed by communicable disease (7, 13.2%), and then unnatural deaths (suicide - 8, 15%, and road traffic accidents - 3, 5.6%).
CONCLUSIONS: It is the first study in India to explore the causes of death in schizophrenia. NCDs being the most common cause of death in schizophrenia suggests to the need of integration of schizophrenia care into general health care.

INTRODUCTION

In general medicine from the mid-1980s, mortality is considered as an endpoint for treatments of cardiovascular diseases.[1] In general, reliable data on causes of death are an important step toward health policy, planning, monitoring, and evaluation. The vital registration systems about causes of death in low- and middle-income countries (LAMIC) which represent for more than two-thirds of global death are not well documented and are almost nonexistent or at rudimentary stages.[23] More than 75% of deaths in LAMIC including India occur at home without medical attention, in contrary, in hospital with medical attention in developed countries.[4] Hence, at-home deaths in India are undocumented and do not have a medically certified cause of death.[4] In this scenario, verbal autopsy (VA) method has been used extensively to assign the causes of death, especially from developing countries like India at the community or population level.[5] VA method used in Million Death Study (MDS) from India was conducted biannual in-person surveys of more than 1 million households across India and determined four most top causes of at-home death for Indians in the age of 30–69 years which are vascular diseases, chronic respiratory diseases, tuberculosis (TB), and cancer.[5]

Mortality is the gold standard measure of clinical outcome in schizophrenia which represents an overall quality of treatment received.[6789] Schizophrenia is often called as “life-shortening disease.”[1011] Most of the mortality reports come from inpatient registers, death registers, or cohort of patients. For example, a systematic review of 22 studies from developed countries, 3 studies from emerging economies, and 1 from least developed countries reported that the median all-cause standardized mortality ratio (SMR) of schizophrenia was 2.58.[12] This demonstrates a more than 2½-fold excess mortality among schizophrenia patients across the globe. This excess premature mortality is represented by a higher number of natural causes of death (such as medical diseases, unhealthy lifestyle, substance misuse, poor compliance or treatment refusal, untreated somatic conditions, and adverse effects of psychotropic medications) and a significant number of unnatural causes such as suicide, especially in young males.[111213] In contrast, in India, this issue has not been well studied. Only 3 reports have provided systematically collected data on mortality in cohorts of schizophrenia patients in India. Dube et al.[14] reported SMR of 2.3 and 4.5 in males and females, respectively, in their 13–14 years’ follow up of a cohort of 140, of which complete information was available only in 62 patients. Harrison et al.[15] report SMRs of 1.9, 1.86, 1.88, and 3.02 in Madras, Agra, Chandigarh urban, and Chandigarh rural centers, respectively. Most of the patients in these studies were treatment-seeking individuals from urban centers (except Chandigarh rural, where treatment-seeking rural individuals were included). A substantial number of patients with schizophrenia in India do not seek treatment, particularly from rural places, which constitute more than 70% of India's population. Studies conducted in urban centers including treatment-seeking patients cannot be generalized to them. A recent mortality study from Thirthahalli taluk, Shivamogga district, Karnataka, of a schizophrenia cohort over 3 years suggests that an average SMR for all 3 years taken together was 1.79, indicating about two-fold excess mortality among schizophrenia patients.[16] Although this research studied the causes of mortality in this cohort, the inquiries regarding the cause of death were preliminary. Detailed qualitative assessment of the circumstances of death would provide valuable clues regarding preventive strategies. In this study, we performed VA of all deaths of patients with schizophrenia in two rural cohorts of schizophrenia patients.

Aim and objectives

The aim of this study was to understand the probable cause of death in patients suffering from schizophrenia in rural areas.

MATERIALS AND METHODS

Data gathered for this study came from two cohorts from ongoing community intervention programs of the Department of Psychiatry, National Institute of Mental Health and Neurosciences (NIMHANS), Bengaluru, in the southern state of Karnataka, India. A community intervention program called as Community Intervention Psychotic Disorders at Thirthahalli taluk, Shivamogga district, was started in June 2005. Another community intervention program called as treating the Untreated Psychosis in Rural Community: Variation in the Experience of Care at Turuvekere taluk, Tumakuru district, was started in December 2009. Till date, 584 patients with schizophrenia are identified, being treated and followed-up in these two taluks.

Details of these cohorts have been published elsewhere.[171819] Doctors and staff of taluk health administration of these two taluks were trained in identifying persons with schizophrenia. Patients thus identified were diagnosed clinically by psychiatrists; the diagnosis was confirmed at first visit using the Mini-International Neuropsychiatric Interview[20] and Indian Disability Evaluation and Assessment Scale[21] by a qualified psychiatrist and provided treatment free of cost. Follow-up of these cohorts is done periodically by psychiatrists coordinated by a social worker at each taluk. These social workers are familiar with local culture and have a rapport with family members from the beginning of their consultations.

One of the authors (NM) conducted a training session to these two social workers about the World Health Organization (WHO) VA 2014 instrument and its data collection/entering procedures. This training session covered individual symptoms and description in local spoken language for easy recognition by the respondents. These field interviewers contacted NM, CNK or JT if they had doubts while interviewing primary caregivers using the VA 2014.

Primary caregivers (usually a family member or relatives or friends) of these patients were contacted. In-person interviews of these primary caregivers were conducted for about 1 h at the venue of their choice (commonly, their homes) with sample VA questionnaire 3 (death of a person aged 12 years and above) of the WHO VA instrument 2014 (http://www.who.int/healthinfo/statistics/verbalautopsystandards/en/index1.html). VA 2014 a fully structured questionnaire administered through a combination of guided self-report and interview methods depending on the literacy and preferences of the respondents.

Baseline sociodemographic and relevant clinical data of deceased patients were collected from their case records. Updated information such as clinical details and medication compliance from last follow-up to patient death was collected from family members in a specially designed pro forma prepared for this mortality study.

VA is a public health tool for obtaining a reasonable direct estimation of the cause of mortality at the community or population level, although it may not be an accurate method for attributing causes of death at the individual level. Hence, VA semi-structured instrument to conduct the systematic retrospective inquiry of family members about the symptoms and signs of illness prior to death has been used to help determine the underlying cause of death.[2223]

The 2014 WHO VA instrument (death of a person aged 12 years and above) is intended to allow for simple and inexpensive identification of causes of death in places where civil registration systems are dysfunctional, especially in underdeveloped countries. The 2014 VA instrument consists of personal information, information on the deceased, vital registration and certification, and information on the respondent and background about interview. Cause of death is determined by subsections of medical history associated with final illness, general signs and symptoms associated with final illness, history of injuries and accidents, risk factors, and health service utilization as well as background and context with recommended optional open narrative text field.

Physician-based method is used to determine the cause of death. Two physicians (NM and CNK) reviewed the outcome of VA questionnaire and formulated the causes of death independently; they were blind to each other's reviews. In cases of disagreement between them, a third physician (JT) arbitrated the results. The final cause of death was assigned based on agreement between the third physician and any of the other two physicians. The cause was considered “undetermined” if all three physicians assigned a different cause or if all of them could not identify a cause.

The quality of information was assessed by noting the relationship of informants with patients (19, 14, and 1 from first-, second-, and third-degree relatives, respectively), whether informants lived with deceased at the time of death and recall period (period from death to interview).

The confidence level about the cause of death was determined as “confident” or “probable” depending on the subjective level of confidence of the physicians about the cause of death from the available information about each patient. The causes of death were classified as immediate cause and underlying cause of death as per the WHO's international death certificate.[24]

The Ethical committee of NIMHANS approved this study protocol. Written consent from the primary caregiver was obtained.

RESULTS

Fifty-three out of 55 patients died were interviewed (family of two were unable to contacted). The causes of death were analyzed from 53 deceased schizophrenia patients from these two cohorts. Mean ± standard deviation age of deceased patients was 50.45 ± 13.65 years (25 patients were highest in the age range of 41–60 years) with almost equal gender distribution (male:female = 28:25). Thirty-eight (71.7%) of them were married, 9 (17%) were never married, 3 (5.7%) were windows, 2 (3.8%) were separated from their spouses, and 1 (1.9%) was divorced. Twenty-one (39.6%) of them were educated up to primary school, 19 (35.8%) had no formal education, 11 (20.8%) went up to secondary school, and 1 had completed high school education. Overall, none of them studied beyond 12th standard. Twenty-five (47.2%) of them were unemployed (assessed as predominant work status during 1 year prior to death), 11 (20.8%) were employed mainly, 10 (18.9%) were homemakers, 6 (11.3%) were semi-skilled worker, and 1 (1.9%) was a pensioner. Twenty-six (49%) of them were from lower-income status; 20 and 7 were from lower- and upper-middle socioeconomic status, respectively.

Nineteen and 14 were first- and second-degree relatives, respectively, who provided the information, 45 informants out of 53 lived with deceased at the time of death, and recall period was 54.76 ± 31.94 months (range: 4–132 months).

Age at onset of illness was 35.43 ± 15.58 years. The duration of schizophrenia till death was 13.59 ± 9.89 years. Forty-two (79%) and 7 patients died at home and hospitals, respectively. Thirty-five (66%) of them died suddenly.

Thirty (57%) of them were not on medication 1 month before death. Forty-seven (89%) patients were not receiving any social benefits. Thirty-five (66%) of the deaths were registered at taluk/gram panchayat.

Determining the cause of death

The authors determined that 22 (41.5%) and 31 (58.5%) causes as confident and probable diagnoses were made for both immediate and underlying causes of death, respectively. The causes of death determined from VA as immediate and underlying are as below. In this way, some patients had more than one causes of death.

Immediate causes of death

The authors determined an immediate cause in 39 out of 53 patients; among which, the most common cause was suicide (n = 6); six patients died due to respiratory, cardiovascular, and multiorgan failure each [Table 1].

Table 1

Immediate causes of death of schizophrenia patients

	n (%)
Nil immediate cause	14 (25.2)
Suicide	8 (14.4)
Undetermined	6 (10.8)
Multiorgan failure	6 (10.8)
Respiratory illness	6 (10.8)
CVS (cardiac attack and arrest)	6 (10.8)
Seizure	3 (5.4)
Road traffic accident	2 (3.76)
Meningitis/encephalitis	1 (1.88)
Dengue fever	1 (1.88)
Total	39+14 (100)

Nil immediate cause means there may be having underlying cause. CVS – Cardio Vascular System

Underlying cause of death

The authors did not find any underlying cause of death in 28 patients. Among the remaining 25, 8 patients died due to chronic kidney disease (CKD) (2 of which also had another cause, chronic obstructive pulmonary disease [COPD], and TB one each) and 5 patients died due to stroke [Table 2].

Table 2

Underlying causes of death of schizophrenia patients

	n (%)
Nil underlying cause	28 (52.8)
Stroke	5 (9.4)
CKD	5 (9.4)
CKD + COPD	1 (1.9)
CKD + TB	1 (1.9)
TB	1 (1.9)
TB+HIV	1 (1.9)
AIDS	1 (1.9)
Gastrointestinal cause	3 (5.7)
Cancer	2 (3.8)
Alcohol liver disease	2 (3.8)
Dementia	1 (1.9)
Nutritional	1 (1.9)
Road traffic accident	1 (1.9)
Total	53 (100)

COPD – Chronic obstructive pulmonary disease; HIV – Human immunodeficiency disease; AIDS – Acquired immunodeficiency syndrome; CKD – Chronic kidney disease; TB – Tuberculosis

Overall, death from noncommunicable diseases (NCDs) (n = 30: CKD – 7, cardiovascular disease – 6, respiratory illness – 6, COPD – 1, stroke – 5, cancer – 2, dementia – 1, epilepsy – 1, and nutritional – 1) dominated the cause of death, followed by communicable diseases (n = 7: TB – 3, human immunodeficiency virus disease/acquired immunodeficiency syndrome – 2, dengue fever – 1, and meningitis – 1) and then unnatural deaths (n = 11: suicide – 8 and road traffic accidents – 3). The authors determined undetermined (unknown) cause and multiorgan failure in 6 patients each. Thirty-five (66%) of the deaths had death certificate from taluk/gram panchayat; none of them had postmortem report.

DISCUSSION

The above results suggest that NCDs are more common causes of death among schizophrenia patients than communicable diseases. Similar to the MDS results,[5] this study in schizophrenia also found all four top causes of death of MDS in this sample. However, two surprising findings were the occurrence of CKD and suicides.

In a population-based nested case–control study at Taiwan, risk of CKD in schizophrenia was found to be higher for those who used second-generation antipsychotics (SGAs) longer cumulatively for more than 1000 days than those who did not.[25] Even in a population-based cohort study from Canada, it was found that SGA use within 90 days of prescription in older adults is associated with a higher risk of acute kidney injury. However, there is a need of such study from the Indian population. Considering that majority of our patient population receive risperidone (SGA), this is a new finding in the Indian population and should be looked further. Considering that diabetes mellitus is one of the risks factors to develop CKD,[26] there is a need of a continuous monitoring of metabolic syndrome in cohorts.

It is well documented that persons with schizophrenia receive less care for NCDs than those without. It may be true of our population too, though we do not have details of their consultations; we do not have comparative data from the general population either.

Only consistent protective factor in studies of suicide in schizophrenia is the adherence to treatment.[27] In this study, more than half of the patients were not on medications 1 month before death. Consistent with earlier findings that suicide is associated with poor adherence, high proportion of patients who committed suicide in our sample had stopped medications. This association between treatment adherence and suicide needs further exploration, as it has obvious clinical and public health implications.

Despite sample dominated by lower socioeconomic status, half of these patients were not receiving any kinds of government-based social disability benefit. This also needs further exploration whether receiving social benefit leads longer longevity among schizophrenia patients?

Half of the patients were unemployed in this study. This concurs with functional impairment in people with serious mental illness is a marker of increased mortality risk.[28]

Majority of deaths in this study population occurred in homes, similar to the MDS results.[4] In an African study from a rural community-based schizophrenia cohort, Teferra et al.[29] reported much younger age of death for both sexes (35 ± 7.35 years). Most common causes were infection (47.4%), followed by severe malnutrition (13.2%) and suicide (10.5%). But in this study NCDs were commoner than communicable diseases.

While it was well known that persons with schizophrenia die earlier, there was no study that had explored the causes of death. This is the first study in India to explore the cause of death among persons with schizophrenia. Community-living (vis-à-vis hospital-based) rural sample of this study makes the findings more generalizable to other parts of the country. The social workers, who collected these data from close family members, hailed from the same regions; they were aware of the sociocultural sensitivities of conducting interviews about death of close family members. The cause of death was determined by two independent reviewers and by arbitration through a third physician in case of disagreements. There was a high degree of agreement between the two physicians (giving some data in result section); the WHO VA is a standardized tool which has been used in earlier studies in the Indian population.[30]

Limitations

The authors acknowledge that there is a possibility of underreporting of suicide rate in this study considering that true disclosure exposes informants to stigma and of anticipated legal risk. The authors also consider the possibility of recall bias considering the longer period between death and VA interview. Possibility of manual errors among physician agreement about the cause of death of patients from VA reports[31] and some studies reports that VA may be ineffective in identifying causes of death in persons over 70 years of age.[32]

Future directions

A prospective study which shall collect mortality data soon after bereavement could be planned in the future.

CONCLUSIONS

NCDs are the most common causes of death among schizophrenia in two rural community cohorts. The higher number of CKDs in this study is worrisome finding which needs further exploration. This study also suggests the need of prevention approaches such as to ensure the adherence to medication using an innovative approach at community level and need of integrating schizophrenia care in general care to reduce the preventable cause of schizophrenia such as NCDs.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.