A province-wide study of the association between hospital resource allocation and length of stay

The relationship between hospital resource allocation and clinical efficiency is poorly understood. Within the single-payer healthcare system in Ontario, Canada, the association between hospital spending patterns and length of stay was studied using data from 1 117 090 patient discharges in 1997/8 at 162 of 171 acute care hospitals. A weighted regression model was created using an overall hospital length of stay index (actual length of stay divided by predicted length of stay) as the dependent variable. Control variables included: hospital size, teaching activity, occupancy rate, rural location and geographic region. Four independent spending variables were defined as a percentage of total hospital spending: nursing, ambulatory care, administration and support, and diagnostics and therapeutics. The reduced regression model had an r-squared of 0.45. Across all spending variables, hospitals spending relatively too little or too much had significantly longer length of stay. Hospitals' overall pattern of resource allocation was also significantly associated with length of stay. Thus, measurable clinical effects can be seen with resource allocation decisions made by hospital management, supporting the need for rigorous decision-making processess. Future research should focus on exploring the nature of this relationship and the potential interdependencies among hospital services that cause this effect.