To better understand the prognosis of strokes with delirium, we performed a large systematic review and meta-analysis of 13 observational studies. Three outcome parameters related to delirium in stroke patients were estimated: inpatient mortality and length of hospital stay.
As for inpatient mortality, the results of published article were controversial. The pooled estimates of 10 studies demonstrated that post-stroke delirium had higher inpatient mortality compared to patients without delirium (OR = 6.35, 95% CI: 4.35–9.25, p<0.0001). With regard to length of hospital stay, the pooled results of 10 studies suggested that stroke with delirium had a 5.93 days longer than stroke without delirium (MD = 5.93, 95% CI: 2.79–9.07; p<0.0001). This is similar to previous studies on delirium in postoperative patients and intensive care unit [30–32]. In 2012, Shi et al [11] showed that post-stroke delirium had higher hospital mortality (OR = 4.71), and longer hospital stays (MD = 9.39 days). However, only a small number of eligible studies were included. Subgroup analysis suggested that there was no significant difference in inpatient mortality between Caucasian and Others (OR: 6.63 vs 5.59, respectively). In addition, significant difference was found in length of hospital stay between Caucasian and Others (MD: 3.57 days vs 10.51 days, respectively).
We found significant heterogeneity among studies exists in length of hospital stay. However, meta-regression did not identify any source of heterogeneity. We supposed that the reasons of significant heterogeneity may be explained by followings: diagnostic criteria for delirium, stroke severity, location of stroke lesions, sample size and delirium onset, development level of country. NOS showed that all selected studies were considered to be moderate or high. Potential publications were not detected in length of hospital stay and inpatient mortality. According to above, we have reasons to think, the pooled results were robust.
The following limitations of our study should be emphasized. First, the included studies showed that delirium was screened by different tools, thus, the results should be interpreted with caution. Second, the severity of delirium was not assessed. In fact, no available tools was used for the severity of delirium. Different degrees of delirium may have different outcomes. Thirdly, the language was restricted to English. Thus, the other available studies published in non-English may be ignored.
Our comprehensive systematic review and meta-analysis suggested that delirium was associated with poor prognosis including inpatient mortality and length of hospital stay. However, our results should be interpreted with caution because of the above limitations. In consideration of these poor outcomes, early screening and effective intervention for delirium may protect against delirium and improve the prognosis. Future larger, well-designed multi-centred cohort studies is needed.