Impact of additive alcohol and substance use disorders on the mortality of people with schizophrenia and mood disorders

What is already known on this topic

Excess mortality has been well described among people with severe mental illness (SMI) or substance use disorders (SUDs) in various settings.1 However, research about how SUDs impact mortality for people with SMI is scarce.2 To further address the issues on the potential contribution of SUDs on elevated mortality in people with SMI (defined as schizophrenia, bipolar disorder and unipolar depression), studies evaluating mortality in SMI cohorts in relation to SUDs are warranted.

Methods of the study

This cohort study was based on the nationwide Danish register systems with identified schizophrenia, bipolar disorder or unipolar depression patients and their demographics, diagnoses of SUDs, date and primary cause of death. A total of 41 470 patients with schizophrenia, 11 739 with bipolar disorder and 88 270 with depression born on or after 1 January 1955 were detected and traced till 17 July 2013 for date of death and 31 December 2011 for cause-specific mortality. Calculations of standardised mortality ratios (SMRs) were performed to compare general mortality in the SMI cohort comorbid with a diagnosis of SUDs and others with SMI only, adjusting for age and sex with Danish population during 1990–2012. HRs for all-cause mortality and subHRs (SHRs) for cause-specific mortality were then estimated on a lifetime diagnosis of SUDs of alcohol, cannabis or hard drugs since 1996 by Cox regressions.

What this paper add

• The elevated age-and-sex-SMRs for people with SMI (schizophrenia 3.63, 95% CI 3.42 to 3.83; bipolar disorder 2.93, 2.56 to 3.29; unipolar depression 1.93, 1.82 to 2.05) were further boosted by an additive lifetime substance use disorder (8.46, 8.14 to 8.79; 6.47, 5.87 to 7.06; and 6.08, 5.82 to 6.34, respectively).

• In Cox modelling analyses, all types of SUDs were found to be significantly associated to excess all-cause mortality in people with schizophrenia (with HRs ranging from 1.24 to 2.65), especially for alcohol and hard drugs (HR 2.65, 95% CI 2.43 to 2.90). Similar patterns were found for people with bipolar disorder (HRs from 1.52 to 3.03) and depression (HRs from 2.01 to 3.44), except cannabis use disorder (HRs 1.09 for bipolar disorder and 0.86 for depression, respectively). All kinds of SUDs were also significantly related to deaths from accidents among people with schizophrenia.

• Prevention strategies of premature mortality among people with SMI should focus on patients with a dual diagnosis of SUDs, particularly for people with schizophrenia.

Limitations

• Population-based registry systems were established for the purpose of administration, not for research, which might limit the depth of data.3 Besides demographics and psychiatric comorbidities, other potential confounders were not available. Substance use disorders could be underdiagnosed, even in people with SMI, making the estimation of relative risk shift towards the null.

• Cohort members in the exposed categories of defined research interest (ie, those with dual diagnoses of SMI and SUDs) would have survived till the time point of diagnosis, since the others without a diagnosis of SUDs (the comparison group) might have died at any time during the follow-up period. This so-called ‘immortal time bias’, referring to the situation that the study subjects must live long enough to qualify by their specific characteristics at baseline in prospective studies,4 might yield an underestimation of the impact of SUDs to mortality.

• The issue of multiple comparisons could be another concern of this study by containing many parallel statistical tests in analysis. These specific results could be random findings under the significance level of 0.05 and thus should be interpreted cautiously.

What next in research

• Further analyses by a more sophisticated factor analysis method, like structural equation modelling, deliberating the role of SUDs in the pathways of SMI leading to premature mortality are warranted in future research.

• More detailed information about physical comorbidities and lifestyle factors retrieved from other data sources through linkages might be helpful to address the issue of residual confounders.

• Underlying biomedical mechanism for the interaction between antipsychotics or antidepressants and substance abuse might be of research interest in the future.

Do these results change your practices and why?

Yes. Advanced clinical efforts to reduce excess mortality should be made, targeting this vulnerable group of SMI patients for treatment of their substance abuse behaviours and physical health problems. Specific clinical guidelines should be generated for people with dual diagnosis of SMI and SUDs for the prevention of premature deaths. The essential components of successful treatments of SMI comorbid with SUDs consist of primary and secondary care plus physical and mental healthcare to inform an integrated health service system.