Independent validation of the prolonged length of stay score.
Koton S., Luengo-Fernandez R., Mehta Z., Rothwell PM.
BACKGROUND: The prolonged length of stay (PLOS) score has recently been derived and validated in 2 independent national cohorts of acute stroke patients in Israel. The present study aimed to determine the performance of the PLOS score in an independent population-based cohort of stroke patients in a health care system considerably different from that in which the score was derived. METHODS: The study was performed on all 434 patients with ischemic stroke or intracerebral hemorrhage hospitalized during the first 5 years of the population-based Oxford Vascular Study (OXVASC) in Oxfordshire, UK. Median (interquartile range) length of stay (LOS) and rates of LOS ≥7 days and LOS ≥30 days by the PLOS score were calculated. Model discrimination was assessed by the c-statistic and goodness of fit was evaluated with the Hosmer-Lemeshow test. RESULTS: Median LOS and rates of LOS ≥7 days increased with the increase in PLOS score among all inpatients and hospital survivors. In the analysis of LOS ≥7 days, the PLOS c-statistic (95% CI) was 0.676 (0.618-0.734) for all inpatients and 0.722 (0.664-0.779) for hospital survivors. Findings were similar for LOS ≥30 days. The Hosmer-Lemeshow showed good calibration. CONCLUSIONS: The PLOS score successfully predicted PLOS in the OXVASC population of acute stroke patients. Although the score was originally derived for the prediction of prolonged acute hospitalization, it successfully predicted prolonged total LOS.