Cardiometabolic comorbidities, readmission, and costs in schizophrenia and bipolar disorder: a real-world analysis
In this large, nationally representative administrative database study of hospitalized patients with schizophrenia and bipolar disorder, cardiometabolic comorbidities were common. Over 60% of patients had ?1 cardiometabolic comorbidity and over 30% had ?2 cardiometabolic comorbidities. Increasing cardiometabolic comorbidity burden was associated with a significantly higher mortality rate (for bipolar disorder), and longer hospital stays (for bipolar disorder). Patients with schizophrenia appeared to have almost double the rates of mortality in comparison to patients with bipolar disorder. The average total all-cause hospitalization cost was $12,781 and $9725 per patient for schizophrenia and bipolar disorder, respectively. Each incremental cardiometabolic comorbidity was associated with an 8.3 and 13.4% increase in total hospital cost for patients with schizophrenia and bipolar disorder, respectively. While 1 in 10 schizophrenia or bipolar disorder patients had an all-cause readmission within 30-days after index hospitalization, the odds of 30-day readmission increased with each incremental cardiometabolic comorbidity.
The reported frequencies of cardiometabolic comorbidities in this study are generally consistent with those previously reported in the literature [1, 2, 15, 29, 30]. However, this study identified a higher prevalence of diabetes in patients with schizophrenia and bipolar disorder (28 and 22%, respectively) than previously reported (7–15%) [4, 6, 30–33]. This discrepancy in the reported prevalence of diabetes may be due to the study population; previous studies have typically been drawn from broader health system populations [30], clinical trial participants [4], or primary care settings [6]. Among the general population, nearly 20% of hospital stays in the US are associated with diabetes [34]. The prevalence of diabetes found among inpatients with schizophrenia (28%) or bipolar disorder (22%) in this study is therefore plausible, given that they are already considered a high-risk population for diabetes [33].
To the best of our knowledge, this is the first study to evaluate the risk of 30-day readmission among hospitalized patients with schizophrenia or bipolar disorder and its association with incremental cardiometabolic comorbidity burden. Our study showed that incremental cardiometabolic comorbidity burden was associated with a 3.1 and 6.4% increased risk of early readmission in patients with schizophrenia and bipolar disorder, respectively. A recent chart review study of 945 patients hospitalized in a psychiatric care facility found that psychiatric readmission in the following year was independently predicted by higher body mass index (BMI) [35]. The authors hypothesized that inflammation, which has been associated with both higher BMI and obesity as well as psychiatric disorders [36] may represent the link between the greater BMI and need for readmission, but research examining the mechanisms of early readmissions and cardiometabolic comorbidities is needed.
For patients with bipolar disorder, additional cardiometabolic comorbidity burden was associated with an increase in the length of stay (4.6 days for no comorbidities to 5.9 days for 3+ comorbidities). Surprisingly, increasing cardiometabolic comorbidity burden was associated with a small decrease in the length of stay among patients with schizophrenia (8.6 days for no comorbidities to 7.9 days for 3+ comorbidities). While the reasons for these differences in length of stay are not known, it is possible that bipolar disorder patients may have been more likely to receive medical assessment and/or intervention for comorbid conditions than were patients with schizophrenia in this study; alternatively it is possible that patients with schizophrenia were more likely to have medical comorbidities that were well established and known to treatment staff compared to patients with bipolar disorder.
Prior research has clearly established a link between cardiometabolic conditions and mortality in the general population [37]. The lack of statistical significance between the odds of mortality and cardiometabolic comorbidity burden in schizophrenia after correcting for demographic and hospital characteristics was unexpected; however, a univariate analysis showed a significant association with comorbidity burden and mortality. Given the small sample sizes and the rarity of mortality incidence in this dataset, it is also plausible that the potential association between mortality and cardiovascular comorbidity is underestimated. Cardiovascular disease, along with cancer and suicide, has also been established as one of the leading causes of death for patients with schizophrenia [38]. Although information about the cause of death was unavailable, confounding of the results by suicide is likely small, as all the patients were hospitalized at community hospitals and not psychiatric hospitals, indicating severity of a medical, rather than psychiatric condition at the time of admission.
Previous studies have reported increased outpatient or total costs for psychiatric patients with cardiometabolic comorbidities [39–41]. This study is unique in that it demonstrated the possible relationship between each additional cardiometabolic comorbidity and incremental costs per admission for patients with schizophrenia or bipolar disorder.
The results of this study highlight the importance of identifying optimal treatment regimens for patients with serious mental illness. Efforts should be taken to adequately monitor for and reduce the rates of cardiometabolic comorbidities in this vulnerable patient population, and perhaps consider antipsychotic therapeutic options with a limited liability for such comorbidities [13]. From a holistic approach of treating patients with serious mental illness, clinicians should coordinate care and consider a patients’ medical profile when prescribing medications. Coordinated care can also improve quality of care and patient satisfaction [42], which may have a positive effect on reducing healthcare costs through shorter hospital stays and/or reduced early readmissions. Furthermore, improved physical health may also positively impact psychiatric health outcomes [35]. In 2013, four new measures of the Healthcare Effectiveness Data and Information Set were added to assess quality of care for patients with serious mental illness; two of these four measures focus on diabetes monitoring and cardiovascular monitoring, which highlights the importance of monitoring cardiometabolic risks among this susceptible patient population [43]. Monitoring patients’ cardiometabolic profiles, consideration of these risk factors when selecting antipsychotic drug therapies, and striving to coordinate care delivery for both mental and physical symptoms may maximize patients’ outcomes.
Limitations
The data used in this study were collected for administrative reasons rather than for research purposes. The study design precludes any determination of causal relationship between cardiometabolic comorbidities and the outcomes. The analysis was restricted to variables present in this particular database, and other factors that were not available in the database may have confounded the observed relationships. Information about the prior treatments for the mental and physical morbidities, prior hospitalizations, and reasons for death was unavailable. In particular, data on atypical antipsychotic utilization prior to hospitalization were not available, therefore it was not possible to assess the relationship between specific antipsychotic medications and cardiometabolic comorbidities, which have been previously described to vary substantially [2, 13, 20, 33].
The burden of cardiometabolic comorbidities in this study was measured using diagnostic ICD-9-CM codes of 6 disease entities. Studies that have used the more robust National Cholesterol Education Program’s Adult Treatment Panel III report (ATP III) or International Diabetes Federation (IDF) definitions [15] have reported metabolic syndrome prevalence of approximately 33% for schizophrenia [15] and 37% for bipolar disorder [2]. The ICD-9-CM coding did not allow for the clear identification of bipolar II disorder (falls under “bipolar other”), precluding the assessment of differences in the number of cardiovascular comorbidities and outcomes between bipolar I and bipolar II subtypes. Moreover, there was no control group of inpatients without serious mental illness. The analysis of each patient was limited to a single index hospitalization and 30 days post-discharge, rather than attempting to determine hospital costs or outcomes over a longer duration of follow-up. Readmission rates in this study may be an underestimate, as these data only include admissions to the hospitals in the Premier network, and the likelihood of readmission to hospitals has been reported to be as high as 20% elsewhere [22].