|
Abstract
|
iolence perpetrated against others by people with psychotic disorders such as schizophrenia often attract a great deal of public attention and also contribute to the enduring stigma associated with the diagnosis. Given the increased risk of violent acts perpetrated against others by patients with schizophrenia spectrum disorders compared with community control individuals,1 there is a particular need to identify factors that may modulate this risk. In studies examining risk factors for violence perpetrated by patients with schizophrenia, poor medication adherence has been reported as one of the disease-related factors associated with greater risk of violence,2,3 consistent with its association with a number of other indices of poor outcomes in people with schizophrenia.4 Poor medication adherence is common among people with schizophrenia4 and is among the factors associated with risk of violence and poor outcomes in general; thus, it may arguably be a factor that is more amenable to intervention than some others.
Although previous studies2,3 have reported the association between medication nonadherence and increased risk of violence or crimes, whether medication nonadherence is associated with increased risk of violence perpetrated by people with schizophrenia has not been systematically examined before. Li and colleagues5 address this gap in understanding by investigating the nature of the association between medication nonadherence and violence in a large cohort of patients with schizophrenia from China. As per previous literature, they used a proportion of regular medication intake (according to follow-up records) of 0.80 or more over the follow-up period as the threshold to determine adherence, whereas intake below that threshold was designated as nonadherence. Using propensity score matching and generalized linear mixed-effects models, they examined the association between medication adherence and risk of violence in people with schizophrenia. Following an established protocol of evidence synthesis and data analysis, they created a directed acyclic graph showing an association between poor medication adherence and increased risk of violence among people with schizophrenia and also identified a number of confounders. One of the key methodological issues that confounds the interpretation of associations identified in observational studies such as this5 relates to the nonrandom allocation of study patients to the different exposure conditions--that is, medication adherence or nonadherence groups. As a result, it is possible that patients in the nonadherent group were more unwell to start with than those in the adherent group and, hence, more likely to have committed violent acts against others. To address this limitation, the authors5 used propensity score matching, a technique that allowed them to compare medication adherent and nonadherent subsets of patients (112 710 patients) who were more evenly matched in terms of measured confounders. The confounders considered for this matching included those identified in the final directed acyclic graph, as well as geographic region from where patients were drawn. In these balanced groups of patients, the authors examined whether violent acts were more likely among patients with medication adherence compared with those without adherence, accounting for the fact that individual patients may have had different propensity to commit violent acts and those from the same region in China may have been more similar to each other in this regard than those from other regions. Although it was unclear how the varying periods of follow-up for different patients were considered during analysis, Li et al5 found that medication nonadherence was significantly associated with different types violence, ranging from minor nuisances to those that violated the civil or the criminal law. The odds were 1.5-fold higher for the more severe criminal infractions among individuals without adherence compared with those with adherence at least 80% of the time, with even higher odds for less severe infractions.5 However, they did not find evidence of a dose-response relationship. Increasing nonadherence was not significantly associated with increasing risk of violence to others; instead, there was a comparable increase in risk across all 4 levels of nonadherence tested compared with patients with adherence. Furthermore, sensitivity analyses suggested that the reported results were generally robust to unmeasured confounding, although this was less so for violent criminal offenses. Subgroup analyses based on gender and urban vs rural settings suggested that the associations held true across both genders and settings with significant differences in the odds of association between urban and rural settings for civil offenses...
Language: en |