OBJECTIVES: Proportion of days covered (PDC) is a common method to measure medication acquisition behavior. Hospitalizations during an observation period may impact PDC calculation, because independent medication acquisition behavior during hospital stays is involuntary and cannot be accurately measured. The purpose of this study was to examine three methods of adjusting for hospital days when calculating PDC, and assess their impact on PDC. METHODS: Medical and pharmacy claims from a commercial Midwestern insurer were used. Population identification criteria included continuously enrolled subjects from January 2007 to December 2009, with diabetes (medical claim with primary or secondary ICD-9 code 250.xx during 2008 also used to establish index date), no long-term care or pregnancy, and age greater than 18 years. Observation period was defined as 365 days from index date. PDC (number of days supplied with diabetes drug / number of days from first diabetes drug acquisition to observation end) was calculated in three ways: 1) without hospital days adjustment; 2) adjusting the denominator for hospital days; 3) adjusting the numerator and denominator for hospital days. One way ANOVA and sensitivity analysis were used to examine the difference in PDC estimates. RESULTS: A total of 24,072 subjects met study criteria and had mean hospital days of 0.86 (SD=7.09). PDC results were as follows: No hospital days adjustment 66.16% (SD=0.39), adjusting denominator 66.24% (SD=0.39), and adjusting both numerator and denominator 66.33% (SD=0.39). No significant difference among the three calculation methods was found (p=0.889). Sensitivity analysis indicated that when average total hospital days reached approximately two days, estimated non-adjusted PDC was significantly lower (p<0.05). CONCLUSIONS: Among subjects with diabetes followed for 365 days, mean total hospital days was small, and impact on PDC calculations was non-significant. As total hospital days increase, PDC estimates can be affected and adjusting for hospital days should be considered.