What is already known on this topic

Individuals with schizophrenia die ∼15–20 years prematurely compared to the general population, mostly due to cardiovascular disease, obesity-related cancer, diabetes and chronic obstructive pulmonary disease.1 Owing to the known metabolic effects of antipsychotics,2 clarifying their role in increased mortality is important. Paradoxically, prior studies in schizophrenia have suggested a relative decrement in mortality with moderate antipsychotic doses compared to no antipsychotic exposure. Cohort studies have indicated an elevated mortality risk with benzodiazepines in several populations, but adjustment for other treatment variables has been limited.

Methods of the study

A cohort of 21 492 individuals aged 17–65 years old who lived in Sweden in 2005 and who received treatment for schizophrenia prior to the beginning of 2006 was followed for 5 years. This cohort was derived from nationwide registers representing all contacts with healthcare in Sweden during the study period. The cumulative prescription of antipsychotics, benzodiazepines and antidepressants was the exposure variable, measured by national prescription records, comparing no use during 5 years, to low, moderate and high cumulative use. The outcome was mortality (all cause, cardiovascular and suicide-related) measured by a national registry. The authors controlled for several relevant variables, including age, disease severity (measured by inpatient stays), treatment compliance (measured by outpatient contacts) and use of concomitant medications. The magnitude of the mortality differences in patients exposed or not exposed to antipsychotics, antidepressants and benzodiazepines was expressed as HRs, representing the per cent of increased or decreased risk versus an equal risk (HR=1.0) between groups for the time until the onset of a given outcome.

What this paper adds

• Finding an overall, categorically defined, 4.8-fold higher mortality risk in patients with schizophrenia than in the general Swedish population adds to concerns about an ‘avoidable mortality gap’ in patients with schizophrenia.

• A statistically significant protective effect of antipsychotics for time until all-cause mortality was confirmed, even at high doses (HR=0.75, ie, a 25% reduction of risk). At low and moderate doses, antipsychotics were protective against cardiovascular mortality (HR=0.61 and HR=0.73, respectively), despite their known cardiometabolic side effects.2 This statistically significant protective cardiovascular effect, which may be due to symptomatic improvement-related healthier lifestyle and participation in medical care, was only lost at high doses, where instead a significant suicide preventive effect of antipsychotics was observed (HR=0.43).

• Antidepressants were protective at all doses against all-cause mortality (low doses: HR=0.60, moderate doses: HR=0.85, high doses: HR=0.71), but not for suicide-related mortality. Rather, at low and high doses, antidepressants reduced cardiovascular mortality (HR=0.49 and HR=0.71, respectively). This finding indicates that psychosis, but not so much depression, is a relevant driver of suicide in schizophrenia and/or that antidepressants are not particularly effective for depression in patients with schizophrenia.3 Further, antidepressants may exert their reduction of mortality through reducing inflammatory processes associated with cardiovascular mortality rather than by reducing suicide.4

• Benzodiazepine prescription was associated with a dose-dependent, significantly increased risk of all-cause mortality. In particular, moderate and high doses significantly increased all-cause mortality (HR=1.23 and HR=1.74, respectively) and high doses increased cardiovascular and suicide-related mortality (HR=1.98 and HR=2.16, respectively). This consistent association could be explained by greater illness severity or comorbid anxiety, accidents due to sedation or cognitive impairment and, especially, the association with comorbid substance use, which is related to greater illness severity, antipsychotic non-adherence and barriers to access care.5

Limitations

• Inferences are limited by lack of randomisation and the related risk of ‘confounding by indication’ (ie, the treatment of sicker patients with benzodiazepines who have a medication-independent mortality risk), which prevents differentiation of causal from non-causal associations.

• Despite statistical adjustment for more baseline variables than in previous cohort studies, highly relevant variables were unaccounted for, including direct measures of schizophrenia and depressive illness severity, unhealthy lifestyle behaviours, degree of substance use, medical comorbidities and treatments, health services use patterns and treatment adherence.

• Differential effects of specific antipsychotics, which vary in their cardiometabolic risk profile, on all-cause and specific-cause mortality were not examined.

What next in research

To untangle the complicated relationship between premature mortality and schizophrenia, database studies are needed with more detailed data resolution. Further, comparative effectiveness research is called for that integrates randomisation, serum drug levels and externally valid healthy lifestyle and suicide prevention interventions. Particularly, given the potentially avoidable deaths due to high-dose benzodiazepine prescribing, a large and long-term prospective, pragmatic, randomised study is needed to clarify the presence/absence of a causal relationship between benzodiazepines and increased mortality in people with schizophrenia and other psychiatric disorders.

Do these results change your practices and why?

Yes. Despite study limitations, these results are consistent enough to merit attention, especially considering that the outcome is mortality. Findings indicate the need for risk contingency measures for patients with schizophrenia who also require high doses of benzodiazepines (attention to medical needs, management of comorbid substance use, limitation of benzodiazepine treatment duration, etc). Results also emphasise the importance of long-term antipsychotic treatment that reduces premature mortality risk in schizophrenia through different mechanisms.