Journal of NeuroEngineering and Rehabilitation

also contained a period of follow-up care, used to track trends in overall health care utilization, expendi- tures, and outcomes. The episodes were structured similarly for the lower extremity and spinal orthotic analyses. For study group beneficiaries in these two service types, the post-service episode started upon receipt of the orth- otic service, and the pre-service window comprised the 12 months prior to this date. The post - service period captured up to 18 months of Medicare claims after receiving the orthotic service. Because compari- son group beneficiaries did not receive orthotic ser- vices, a proxy episode start date was established. To ensure the same post-service window for which health care utilization and expenditures were tracked and com- pared across cohorts, the length of time between etio- logical diagnosis and episode start, or “ lag time, ” for the comparison group was set to the average of the length of time for study group participants of similar age and gen- der. This lag time was added to the date of etiological diagnosis to create an episode start date for each compari- son group beneficiary. Similar to the study group, the pre-service window comprised the 12 months prior to the episode start date, and the post-service window comprised the 18 months following the start date. This episode structure was modified for the pros- thetic analysis. In the 2007 – 2010 study, analysis using a temporal autocorrelation function indicated that the optimal length of the post-period for the prosthetic analysis was 12 months following the episode start, which was approximately three months after amputa- tion. However, the Affordable Care Act (ACA) was implemented since our prior analysis, requiring modi- fications to this 2011 – 2014 study. The ACA had a considerable impact on hospital inpatient and out- patient mix, stay duration, and re-admission policies, among other factors. To address this, we used a 15-month episode period starting with the date of hospital discharge associated with amputation for the 2011 – 2014 lower extremity prosthetic population, as contrasted to the 3-month waiting period post-amputation and an im- mediately subsequent 12-month episode period we had used for the 2007 – 2010 study. Thus, both study and com- parison groups had a pre-service window comprising the 12 months prior to this hospital discharge and a 15-month post-service window immediately following it. Calculating descriptive statistics and analyzing impact of orthotic/ prosthetic devices on overall patient Medicare expenditures For each of the three analyses (lower extremity orth- oses, spinal orthoses, and lower extremity pros- theses), descriptive statistics were calculated for the study and comparison groups after the propensity score matching. The two groups were compared to each other based on the distribution of patient char- acteristics including but not limited to age, gender, race, and comorbidities. We then compared the total average episode Medicare payments of the study and comparison groups over the post-service period, as well as the distribution of payments by care settings, and a range of outcome measures, such as falls, hospitalizations, and days of rehabilitative/physical therapy. Results Demographic analysis Table 2 presents the descriptive statistics of matched patients for each O&P service. Since the propensity score matching criteria included patient demographic characteristics and controlled for observable selection bias, the study and comparison group patients were highly similar within each O&P service type. No sig- nificant differences were found between the matched study and comparison groups for any variables used in the propensity score matching process, including age, gender, dual eligibility, and race, for any O&P service ( p < 0.05). Table 2 Descriptive Statistics across Matched Pairs (2011 – 2014) Lower extremity orthotic model Spinal orthotic model Lower extremity prosthetic model Demographic characteristic Study group Comparison group Study group Comparison group Study group Comparison group Number of beneficiaries 43,487 43,487 34,575 34,575 545 545 Average age 68.6 68.7 67.2 67.2 65.9 65.9 Dual eligibility status 29.7% 29.7% 34.9% 34.9% 39.2% 39.2% Gender: female 43.1% 43.1% 37.6% 37.6% 17.4% 17.4% Race/Ethnicity: white 84.7% 84.7% 81.2% 81.2% 68.8% 68.8% Race/Ethnicity: black or african american 8.3% 8.3% 11.8% 11.8% 24.8% 24.8% Race/Ethnicity: hispanic 4.4% 4.4% 5.0% 4.4% 6.4% 6.4% Differences were not significant at α = 0.05 Source: Dobson | DaVanzo analysis of custom cohort Standard Analytic Files (2011 – 2014) for Medicare beneficiaries who received O&P services from January 1, 2012 through June 30, 2013 (and matched comparisons), according to custom cohort database definition Dobson et al. Journal of NeuroEngineering and Rehabilitation 2018, 15 (Suppl 1):55 Page 64 of 72