Effectiveness of Integrative Therapy for Parkinson’s Disease Management

Woo, Yeonju; Hyun, Min Kyung

To adjust for biomedicine treatment in both treatment groups, the treatment period in neurology, hospitalization period in neurology and anti-parkinsonism medications (levodopa, ropinirole, pramipexole, selegiline) were also considered as confounders (Zhuo et al., 2017) . The propensity score (PS) was used to increase the level of causality revealed in the observational study (D'Agostino, 1998) . Specifically, the PS was used to match the subjects with similar characteristics in the two groups. Several variables were used to create the PS index: age, sex, index year, CCI, treatment period in neurology, hospitalization period in neurology, severity of disability and anti-parkinsonism medication. Comparison of the PS between groups was repeated using the greedy algorithm and 1: 1 matching continued until there were no more matching pairs (Parsons, 2001) . The PS fit was confirmed by c-statistics and the Hosmer and Lemeshow Goodness-of-Fit test. In the case of a c-statistic value over 0.5, the Hosmer and Lemeshow Goodness-of-Fit test was considered to be good when the p-value was 0.05 or higher (D'Agostino, 1998; Maldonado and Greenland, 2002) . The prevalence and annual length of hospitalization were analyzed using logistic regression analysis and t-tests. A pvalue < 0.05 was considered significant. Additionally, the integrative therapy group was divided into two groups by treatment period, less than 30 days and over 30 days. The prevalence and annual length of hospitalization were analyzed using logistic regression analysis and ANOVA. After ANOVA, Duncan's post hoc test was applied. A p-value < 0.05 was considered significant. Finally, survival analysis of the integrative therapy group and monotherapy group was performed using the Cox proportional hazard analysis.