Length of stay of COPD hospital admissions between 2006 and 2010: a retrospective longitudinal study

BACKGROUND Hospitalizations for COPD are associated with poor patient prognosis. Length of stay (LOS) of COPD admissions in a large urban area and patient and hospital factors associated with it are described. METHODS Retrospective longitudinal study. All COPD patients registered with London general practitioners and admitted as an emergency with COPD (2006-2010), not having been admitted with COPD in the preceding 12 months were included. Association of patient and hospital characteristics with mean LOS of COPD admissions was assessed. Association between hospital and LOS was determined by negative binomial regression. RESULTS The total number of admissions was 38,504, from 22,462 patients. The mean LOS for first admissions fell by 0.8 days (95% confidence interval [CI]: 0.7-1.5) from 8.2 to 7.0 days between 2006 and 2010. Seventy-nine percent of first admissions were ≤10 days, with a mean LOS of 3.7 days (2009-2010). The mean LOS of successive COPD admissions of the same patients was the same or less throughout the study period. The interval between successive admissions fell from a mean of 357 days between the first and second admission to a mean of 19 days after eight admissions. Age accounted for 2.3% of the variance in LOS. Socioeconomic deprivation did not predict LOS. Fewer discharges happened at the weekend (1,893/day) than on weekdays (5,218/day). The mean LOS varied between hospitals, from 4.9 days (95% CI: 3.8-5.9) to 9.5 days (95% CI: 8.6-10.3) when adjusting for clustering, age, sex, and socioeconomic deprivation. CONCLUSION The fall in LOS of the first COPD admission between 2006 and 2010 reflects international trends. The stability of LOS in successive admissions suggests that increasing severity of disease does not affect recovery time from an exacerbation. Variations between hospitals of nearly 5 days in LOS for COPD admissions suggests that significant improvements in patient outcomes and in savings in health care utilization could be made in hospitals with longer LOS.