This was a retrospective claims database analysis conducted in the Truven Health Analytics MarketScan® Medicare Supplemental and Coordination of Benefits (COB) Database using data from July 1, 2011 through June 30, 2017. The Medicare Supplemental database includes data capturing the healthcare experience of individuals with Medicare supplemental insurance paid for by their former or current employers. Both the Medicare-covered portion of payment and the employer-paid portion are included in this database. In compliance with the Health Insurance Portability and Accountability Act of 1996 (HIPAA), all patient data used in this study were de-identified and thus Institutional Review Board approval was not required.
Study Design and Patient Selection Criteria
Figure 1 shows the selection criteria of the eligible study population. Eligible patients had at least 1 outpatient claim for DC diagnosis (i.e International Classification of Diseases Ninth or Tenth Revision (ICD-9/10) codes 728.6 or M72.0) between July 2011 and June 2017. Patients were assigned in either a Xiaflex cohort, fasciectomy cohort or a no treatment cohort.
Patients were included in the Xiaflex cohort if they had any of the following points of evidence— a claim of Xiaflex medication, identified through a Healthcare Common Procedure Coding System (HCPCS) code J0775 or a National Drug Code (NDC) 66887-0003-01; injection of the medication in the palm tissue within a 90-day period after the medication claim, identified through the Current Procedural Terminology (CPT) code 20527; DC diagnosis obtained on the same date as procedure. Fasciectomy-treated patients had at least one claim with CPT code for fasciectomy (Appendix A) and DC diagnosis on same date. The remaining study population had no any DC-associated treatments (Xiaflex, fasciectomy, fasciotomy, or needle aponeurotomy) and was therefore defined as „no treatment” cohort.
For DC patients treated with Xiaflex or fasciectomy the index date was the date of the first procedure. The index date for patients who were not treated was set to be a date of the initial DC diagnosis. All procedures performed after initial procedure were required to have DC diagnosis obtained on the same date.
Patients had to have continuous enrollment in the database with medical and prescription benefits for 24 months before the index date (pre-index baseline period) and 12 months after the index date (follow-up period). Patients were excluded if had any claims for Peyronie disease (i.e., ICD-9/10 607.85 or N48.6) during the baseline or follow-up period. To ensure that patients were treatment naïve, patients were not allowed to have any treatment for DC (Xiaflex, fasciectomy, fasciotomy, or needle aponeurotomy) during the pre-index period.
Patients were excluded if had any claims for Peyronie disease (i.e., ICD-9/10 607.85 or N48.6) during the baseline or follow-up period, or had any treatment for DC (Xiaflex, fasciectomy, fasciotomy, or needle aponeurotomy) during the pre-index period.
Demographic variables such as age, gender, geographic region, and insurance plan type were recorded on the index date. Clinical characteristics, including the Charlson Comorbidity Index and other comorbid conditions (based on the presence of ICD-9/ ICD-10 diagnosis codes), were measured during the pre-index period.
The primary outcomes of this analysis were all-cause and DC-related healthcare utilization and costs in the 12 months following the index date. Total costs included medical (inpatient, emergency room, and outpatient office visit) and pharmacy costs. ‘Dupuytren-related’ costs were defined as a medical claim with a diagnosis of Dupuytren contracture in any position. In addition, healthcare utilization and cost were categorized by place of service (pharmacy, laboratory, physician office, patient home, outpatient, emergency department, inpatient, etc.) and by service provider (facility inpatient, facility outpatient, laboratory, radiology, physician outpatient, physician inpatient, etc.). Healthcare resource use was reported as the number of patients with outpatient visits, inpatient hospital admissions and ER visits, length of hospitalizations, average number of outpatient prescription, average number of total outpatient services, average number of inpatient services and ER visits, and average length of hospitalization in days.
Propensity score matching was used to adjust for baseline differences between Xiaflex and fasciectomy cohort, and Xiaflex and no treatment cohort separately. The variables included in a logistic regression model as predictors were CCI, geographical region of the patient’s residence, insurer’s plan type, gender, hypertension, hyperlipidemia, carpal tunnel syndrome, trigger finger syndrome, depression and sleep disorder. Using propensity scores generated from this logistic regression model, Xiaflex patients were matched 1:1 to fasciectomy patients based on the nearest neighbor approach, without replacement metric. Xiaflex and no treatment cohort were matched in ratio 1:3 ratio based on the following variables: CCI, gender, age of patients, geographical region of the patient’s residence, insurer’s plan type, hypertension, carpal tunnel syndrome, trigger finger syndrome and mental disorders.
Demographic and clinical characteristics and healthcare utilization and costs were reported for unmatched and propensity score–matched samples. Categorical variables were reported as counts and percentages. Continuous variables were summarized as means and standard deviations. Chi-square tests and ANOVA or t-tests were used to calculate potential differences in these distributions between cohorts. Generalized linear models (GLMs) with log link and gamma error distribution to handle non-normal cost distributions were used to assess the relationship between treatment cohorts and healthcare costs.