Sociodemographic and clinical factors predict psychiatric ED utilization within six months of inpatient discharge.

This study represents a secondary analysis of a pragmatic stepped-wedge cluster-randomized trial conducted across ten health care sites in Alberta, Canada. The research population consisted of 1,098 psychiatric inpatients. The primary objective was to identify predictors of psychiatric emergency department (ED) utilization within six months of discharge. While the broader trial evaluated interventions such as text messaging (SMS) and peer support, this specific analysis focused on sociodemographic and clinical factors, including age, ethnicity, relationship status, employment, housing status, and prior ED use, rather than comparing intervention arms directly. The follow-up period for the analysis was six months.

The analysis utilized logistic regression to determine associations between patient characteristics and ED utilization. Results indicated that age was a significant predictor. Patients aged 26 to 40 years had lower odds of revisiting the ED compared to those under 25 years (OR 0.66; 95% CI 0.46-0.95). Similarly, patients over 40 years had reduced odds of ED use compared to those under 25 years (OR 0.58; 95% CI 0.37-0.92). Regarding ethnicity, individuals identifying as mixed or other ethnicity had lower odds of returning to the ED compared to White patients (OR 0.52; 95% CI 0.28-0.96).

Employment status emerged as a critical factor. Unemployed individuals demonstrated higher odds of ED use compared to employed individuals (OR 1.66; 95% CI 1.18-2.34). The strongest predictor identified was prior ED attendance. Patients with a history of ED use prior to the current admission had significantly higher odds of revisiting the ED within six months (OR 2.45; 95% CI 1.03-5.80). No secondary outcomes were reported in the provided data.

Safety and tolerability data were not reported for this specific analysis, as the focus was on observational predictors rather than intervention safety profiles. Consequently, adverse event rates, serious adverse events, discontinuations, and general tolerability could not be assessed. The study was funded by sources not reported in the available data, and potential conflicts of interest were not disclosed.

These results align with broader literature suggesting that social determinants of health and historical utilization patterns are robust predictors of post-discharge outcomes. However, this analysis does not compare the efficacy of specific interventions like SMS or peer support against treatment as usual for these specific subgroups, as the primary focus was on risk stratification. The study design, being a secondary analysis of a cluster-randomized trial, relies on the data collected during the primary intervention evaluation.

Key methodological limitations include the observational nature of the predictor analysis, which precludes causal inferences regarding the factors themselves. The absence of reported data on housing status and relationship status in the results limits the ability to fully assess the impact of social support networks. Additionally, the lack of reported funding or conflict of interest information prevents a complete assessment of potential biases. The certainty of the findings is not explicitly reported.

Clinically, these findings emphasize the importance of targeted transitional care interventions. Resources should be prioritized for high-risk groups, specifically younger patients (under 25), unemployed individuals, and those with a history of recent ED use. Scalable approaches, such as text messaging and peer support, may be particularly beneficial when integrated into discharge planning for these specific populations. Further research is needed to determine if these risk factors interact with specific intervention types to improve outcomes.

Several questions remain unanswered. The study did not report on housing status or relationship status outcomes, leaving gaps in understanding how specific social supports influence readmission. The long-term impact of these predictors beyond six months is unknown. Additionally, the interaction between identified risk factors and the specific interventions tested in the primary trial (SMS, peer support) was not the focus of this analysis, leaving the optimal combination of risk stratification and intervention unclear.