On 2015 Mar 18, Adrian Barnett commented:

Suffering a hospital-acquired complication is a time-dependent variable as some patients will experience it some days after their admission. These patients cannot be discharged between admission and the complication. Unfortunately the statistical analysis used ignores this fact and hence the estimates are subject to the length bias and are likely to be over-estimates of the extra length of stay. A solution is to use multi-state models combined with Cox survival models with day of admission as the time variable. A hospital-acquired complication is a transient state between admission and discharge. This accounts for the timing of complications and correctly estimates the extra length of stay due to a complication.

On 2015 Mar 18, Adrian Barnett commented:

Suffering a hospital-acquired complication is a time-dependent variable as some patients will experience it some days after their admission. These patients cannot be discharged between admission and the complication. Unfortunately the statistical analysis used ignores this fact and hence the estimates are subject to the length bias and are likely to be over-estimates of the extra length of stay. A solution is to use multi-state models combined with Cox survival models with day of admission as the time variable. A hospital-acquired complication is a transient state between admission and discharge. This accounts for the timing of complications and correctly estimates the extra length of stay due to a complication.