OBJECTIVES

A multimodal prescriber-focused intervention was implemented in Parma, Italy during 2008-2009. The intervention substantially reduced the prevalence of potentially inappropriate medication (PIM) use. In light of the relationship between PIM exposure and iatrogenic events resulting in hospitalization, our objective was to evaluate changes in hospitalization rates associated with the intervention’s reduction in PIM use.

METHODS

We conducted a population-based retrospective cohort study using the administrative healthcare database from the Local Health Authority of Parma, Italy. Residents of Parma aged >65 years were followed until death, out-migration, or long-term institutionalization. Crude hospitalization rates and changes in quarterly hospitalization rates were evaluated in 3 periods (pre-intervention: 2005 Q1 – 2007 Q4, intervention: 2008 Q1 – 2009 Q4, post-intervention: 2010 Q1 – 2017 Q4). Multivariable Poisson regression was used to estimate a piecewise linear spline of quarterly hospitalization rates during the intervention and post-intervention periods adjusted for changes in population characteristics, repeated measures within individuals, and seasonality.

RESULTS

In the pre-intervention, intervention, and post-intervention periods, 111,755 patients contributed 285,203 person-years (PYs), 107,390 patients contributed 193,072 PYs, and 142,898 patients contributed 795,034 PYs, respectively. Crude hospitalization rates for the pre-intervention, intervention, and post-intervention periods were 261.7, 255.9, and 248.6 per 1000 PYs, respectively. The adjusted pre-intervention hospitalization rate was estimated to be declining by 0.8% (IRR=0.992, p<0.01, approximately 2.1 hospitalizations per 1000 PYs per quarter). The intervention hospitalization rate declined twice as fast (1.6%, IRR=0.984, p<0.01, approximately 4.0 hospitalizations per 1000 PYs per quarter). The post-intervention hospitalization rate continued to decline, however, the rate of decline (approximately 0.7 hospitalizations per 1000 PYs per quarter) was slower than the intervention period (p<0.01).

CONCLUSIONS

In a large population-based cohort of older adults, a multimodal intervention to decrease PIM use was associated with an accelerated decline in the rate of hospitalization.