Introduction and objectives The major cost to health services of COPD care is hospital admission for exacerbation. Reducing length of stay (LOS) will reduce cost, yet there is wide variability across patients and hospitals. We test the hypothesis that these variations may be attributed to either patient characteristics, hospital characteristics and/or the so-called hospital-clustering effect, which indicates that patients with similar characteristics may experience different processes of care and outcomes depending on the hospital to which they are admitted.

Methods The European COPD Audit which was carried out in 432 hospitals from 13 countries, included data from 16,018 patients admitted over an 8 week period. The recorded variables included information on the patient and disease characteristics, resources available and clinical practice. Variables in each category associated with LOS were evaluated by a multivariate multilevel analysis and expressed as odds ratios (OR).

Results Mean LOS was 8.7 days (median: 7, standard deviation: 8.3, interquartile range: 4–11). Factors associated with an LOS higher than the median (see figure) were clinical with the highest impact in patients with use of mechanical ventilation (OR: 4.74) and higher oxygen flows (OR 2.63). In-hospital treatments, comorbidities and patient-related variables including GOLD class IV (OR 1.77) were also significant. These relationships were maintained with respect to longer LOS (> 21 days). Neither the day of admission, nor any of the resource variables were associated with significant differences in LOS. The crude variability of LOS between the different countries was reduced after accounting for these clinical factors and the clustering effect.

Conclusions This study demonstrates a noteworthy reduction in the observed crude inter-hospital variation in LOS after accounting for the hospital-cluster effect and patient related variables. This emphasises the predictor importance of the patients’ clinical conditions and interventions, and understates the impacts of hospital resources and organisational factors. This “real-life” reflection may highlight some valid learning points that may help us to determine which achievable strategies are most relevant to improve outcomes.