Introduction

Bipolar disorder (BD) is a serious mental illness causing a high degree of medical burden during the course of illness and with the aging process [1]. Notably, patients with BD are documented to have 2 to 4 times higher mortality and at least a 10-year reduction in life expectancy compared to the general population [1,2]. Furthermore, evidence suggests that the mortality gap has widened in recent decades [3]. Among the medical causes of excessive and premature mortality deaths among BD patients, cardiovascular diseases are recognized as the leading cause, reported in both Western and Eastern studies [4–7].

Stroke is a cardiovascular disease with a high risk of mortality and long-term disability [8,9]. General population studies have shown that obesity, diabetes, dyslipidemia, hypertension, cigarette smoking, and alcohol abuse are the major risk factors for stroke [10]. Studies also suggest that patients with BD are more susceptible to develop or have the stroke risk factors than the general population [11–15]. One proposed reason is suspected to be suboptimal medical care of metabolic and vascular diseases in patients with BD [16–20].

Studies, worldwide, have shown that patients with BD experience suboptimal treatment for stroke [7,15,21]. Stroke care guidelines recommend specific and detailed medical care plans to limit complications and improve stroke outcomes [22]. Although patients with BD are about twice as likely to develop stroke as the general population [1,5,7,15,23,24], in-patient admission rates for cardiovascular care are only slightly higher in patients with BD [7,15]. A few studies have reported that patients with BD may suffer from poor outcomes following a stroke [7,25]. Failure to deliver adequate medical care to these patients is postulated to contribute to their poorer outcomes following stroke and cardiovascular diseases. However, there are no studies examining concurrently, the medical care costs and in-hospital adverse outcomes of patients with BD admitted for a stroke. In this study, we used a nationwide, population-based database to investigate both adverse outcomes and acute hospitalization costs patients with stroke. We hypothesized that BD could significantly affect the in-hospital outcomes of stroke. We investigated total medical cost as a proxy for care intensity comparing patients with BD with the comparison group to evaluate the role of potentially differential care provision in the documented adverse outcomes of patients with BD with stroke. Such an analysis is essential to understand the patterns of care provided to patients with BD when they experience non-psychiatric medical conditions.

Methods

Results

Table 1 shows the demographic characteristics and comorbidities of the study group and comparison group. Given the use of propensity-score matching, the two groups were similar on most characteristics: mean ages of the study group and comparison group were 62.3±14.8 and 62.0±14.7 years, respectively (p = 0.63), with no difference in sex, urbanization level, hypertension, hyperlipidemia, diabetes, and coronary heart disease distribution. However, the groups differed on geographic region (p = 0.007) and monthly income (p = 0.03). Furthermore, of the 580 patient with BD, 26.8%, 24.0%, 16.5%, 13.8% and 18.9% were diagnosed with bipolar affective disorder, depressed (ICD-9-CM code 296.5), bipolar affective disorder, manic (ICD-9-CM code 296.4), bipolar affective disorder, mixed (ICD-9-CM code 296.6) manic-depressive psychosis, unspecified (ICD-9-CM code 296.80) and others (ICD-9-CM codes 296.0, 296.7 or 296.89), respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic characteristics and comorbidities of stroke patients with and those without bipolar disorder (N = 2,320).

https://doi.org/10.1371/journal.pone.0213072.t001

Table 2 presents the occurrence rate of the adverse outcomes of interest and hospitalization variables among patients with and without BD. Stroke patients with BD had significantly lower in-hospital mortality (3.28% vs. 5.63%), and event rates for acute respiratory failure (2.59% vs. 5.57%), and use of mechanical ventilation (6.55% vs. 10.23%). There were no statistically significant differences in pneumonia, UTIs, length of stay, and medical costs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Occurrence of adverse events and profile of hospitalization variables among stroke patients, stratified by the presence of bipolar disorder.

https://doi.org/10.1371/journal.pone.0213072.t002

Adjusted associations of BD with the adverse outcomes, and hospitalization variables are presented in Table 3. Logistic regressions showed that after adjusting for the demographic and comorbidity covariates ORs for in-hospital mortality, acute respiratory failure, and use of mechanical ventilation for patients with BD were 0.56 (95% CI: 0.34–0.92), 0.46 (95% CI: 0.26–0.80), and 0.63 (95% CI: 0.44–0.91), respectively, relative to those without BD. BD was negatively associated with in-hospital mortality, acute respiratory failure, and use of mechanical ventilation among stroke patients.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Adjusted relationships between the occurrence of bipolar disorder, adverse outcomes, and hospitalization outcomes.

https://doi.org/10.1371/journal.pone.0213072.t003

Discussion

The present study may be the first to investigate in-hospital mortality and adverse event rates in stroke patients with BD. With comparable total hospitalization costs and length of stay, we found that stroke patients with BD actually had lower rates of in-hospital mortality, acute respiratory failure, and mechanical ventilation use than comparison patients without BD. On previous study had shown that the annual hospitalization cost of patients with BD for non-psychiatric causes was higher than patients without BD in Taiwan [15]. However, prior studies also revealed that admission rates for stroke among patients with BD were similar to those of the general population despite their higher risk of suffering a stroke [15, 21]. Taken together, the previous research findings suggest that the poor outcomes of stroke in patients with BD may result from the inadequate stroke care. In the present study, we have observed that, with the comparable hospitalization costs and length of hospital stay, stroke patients with BD had lower in-hospital mortality and serious adverse events compared to stroke patients without BD. The results suggest that the high mortality rates of stroke among patients with BD may possibly be reduced by optimal stroke care. Therefore, further studies utilizing the database from other health care systems are warranted to validate our present observations.

Contrary to the literature [7,21,25], we found that stroke patients with BD had lower in-hospital mortality and adverse event rates compared to the comparison group. Schoepf et al. found patients with BD were at an increased risk of in-hospital mortality following a stroke compared to the control group among their patient sample drawn from general hospitals [25]. In particular, the mean age of patients with BD at baseline in their study were 47.3±0.2 years and the duration of follow-up was 1814 days, much younger than our sample (62.3±14.8 years). The age discrepancy between the present study and previous studies remains inexplicable and may reflect differences in sampling methodology. Age is a critical factor in the management of patients with BD. Recently, the International Society for Bipolar Disorders Task Force proposed the age of 50 years as the demarcation to define older-age BD because of the higher medical burden and shorter lifespan in patients with BD [26]. Evidence suggests that patients with BD have an increased risk of cardiovascular disease from the fourth decade [7].

Prior studies suggested that patients with BD usually received suboptimal medical care for non-psychiatric diseases compared to the general population [16,18,19,20]. This suggests another explanation for the fewer adverse outcomes in our stroke patients with BD: under-diagnosis of adverse events and inadequate treatment of those conditions. However, the adverse events studied (i.e. in-hospital mortality, acute respiratory failure, and use of mechanical ventilation) are high severity conditions and not readily obscured. Furthermore, the hospitalization variables, length of stay and costs did not differ between stroke patients with BD and those without BD. Taken together, it is less likely that inequities in standard stroke care occurred among our study patients with BD. Recent systemic reviews and large-scale meta-analysis have suggested that antipsychotics and antidepressants generally increase the risk of stroke [27,28]. Further researches would be of interest to examine the effect of medications on in-hospital mortality and adverse event rates in stroke patients with BD.

The study provided a sufficient sample to investigate the outcomes of interest. Use of propensity score adjustment to improve the statistical estimation of risks mitigates selection bias. Nevertheless, several methodological limitations should be acknowledged in interpreting the results. First, diagnostic validity in claims data is often questioned. In the present study, the diagnosis of BD was based on the ICD code. No information from psychiatric evaluations was available. Although the National Health Insurance Bureau randomly reviews medical records to verify the diagnoses and accuracy of coding, our epidemiological research may still exist limitation of under-diagnosis as mentioned previously in the literature. The prevalence of BD was only about 0.2% among the adult patients with stroke in this study. Second, our data lack information on the severity of BD. One potential bias is mortality censoring of patients with BD. More severe patients with BD may be censored due to premature death from other causes such as suicide or other medical conditions. Third, direct measurements of stroke severity are not available in NHIRD database. We used proxy variables (e.g. length of hospital stay and hospitalization costs) which are documented to be correlated with stroke severity measures, such as National Institute of Health Stroke Scale [29–31]. Fourth, lifestyle variables such as cigarette smoking, alcohol consumption, dietary habit, and physical activity are not available in this database. Unhealthy lifestyle behaviors usually have adverse implications for stroke outcomes [10]. Given the direction of findings, life-style factors are less likely to be the confounders. Fifth, our findings may not generalized to other ethnic populations given that differences in stroke characteristics by ethnicity have been reported in the literature [8,32]. We only included hospitalized patients with BD in the study. As previous studies shown, some patients with BD with stroke may be neglected or not brought to medical establishments, who would be excluded from study. Their outcomes may be worse than those of our study sample. Sixth, we selected adverse outcomes associated with significant morbidity and mortality. The functional status following the stroke remains unknown in our sample. Patients with BD with stroke may suffer higher level of disability due to the double disability and worth further investigation.