Data Source
Medicaid claims from Iowa, Kansas, Missouri, and Mississippi (09/01/2013–03/31/2016) were used. The data includes medical and prescription drug claims, enrollee’s plan eligibility, and demographics. All data is de-identified and in compliance with the Health Insurance Portability and Accountability Act. No institutional review board approval was required for this study.
Study Design
A retrospective cohort study design was used to evaluate baseline characteristics, treatment patterns, HRU, and costs in pediatric patients diagnosed with CIU/CSU (the CIU/CSU cohort). A retrospective matched-cohort design was used to compare HRU and costs between the CIU/CSU cohort and CIU/CSU-free patients (the control cohort).
For both cohorts, the baseline period was defined as the 6-month period prior to the index date, and the observation period spanned from the index date until the earliest of health plan disenrollment or end of data availability.
The index date for CIU/CSU cohort was defined as the date of the first diagnosis of idiopathic, other, or unspecified urticaria or angioedema. The index date for the control cohort was randomly selected on the basis of the distribution of time between the start of the continuous eligibility period and the CIU/CSU diagnosis date in the CIU/CSU cohort.
Patient Selection
Patients in the CIU/CSU cohort were identified on the basis of the validated algorithm and had to have either at least two independent claims at least 6 weeks apart with a diagnosis of idiopathic (ICD-9-CM: 708.1x or ICD-10-CM: L50.1xx), other (ICD-9-CM: 708.8x or ICD-10-CM: L50.8xx), or unspecified (ICD-9-CM: 708.9x or ICD–10-CM: L50.9xx) urticaria, or at least one claim with a diagnosis of idiopathic, other, or unspecified urticaria and at least one claim with a diagnosis of angioedema (ICD-9-CM: 995.1x or ICD-10-CM: T78.3) at least 6 weeks apart [20]. In the control cohort, patients had no claims with a diagnosis of idiopathic, other, or unspecified urticaria or angioedema. Patients in both cohorts were required to be 11 years old or younger and have at least 6 months of continuous Medicaid eligibility before and after the index date.
Study Measures
Study measures included demographics and clinical characteristics, treatment patterns, as well as all-cause and CIU/CSU-related HRU and costs (i.e., pharmacy and medical costs) per patient per year (PPPY).
Treatment patterns were described using the following components: (1) proportions of patients using CIU/CSU-related prescription treatments, (2) proportions of patients switching CIU/CSU-related treatments and using CIU/CSU-related treatments in combination; (3) pill burden (the ratio of the total number of pills prescribed to the number of days during a given period); and (4) types of medical specialists visited. These outcomes were described during the first 6, 12, and 18 months of follow-up among CIU/CSU patients with at least 6, 12, and 18 months of follow-up, respectively.
HRU and costs were reported by type of service: inpatient (IP), emergency department (ED), and outpatient (OP; broken down into urgent care facility, home care, office, ambulatory surgery center, and other OP). All-cause HRU and costs were assessed during the observation period for both cohorts. CIU/CSU-related HRU and costs were defined using diagnosis codes for idiopathic, other, or unspecified urticaria, and were described for the CIU/CSU cohort only. All costs were inflation-adjusted to 2016 dollars based on the US Consumer Price Index.
Statistical Analysis
Mean, standard deviation (SD), first and third quartile, interquartile range (IQR), and median were used to describe continuous variables, and frequencies and percentages were used to describe categorical variables.
Propensity score (PS) matching was used to adjust for the differences in baseline characteristics between the CIU/CSU and control cohorts. PS, defined as the probability of being in the CIU/CSU cohort, was estimated using a multivariate logistic regression model, including age, sex, race, state, presence of Home Maintenance Organization (HMO) insurance plan, year of the index date, and the Quan-Charlson comorbidity index (Quan-CCI). Patients in the CIU/CSU cohort were matched in a 1:1 ratio with control patients on the basis of PS. A standardized difference calculated by dividing the absolute difference in means (for continuous variables) or proportions (for categorical variables) of two cohorts by the pooled SD of both cohorts was used to assess the balance of baseline characteristics between cohorts before and after matching (less than 10% indicated sufficient balance) [21].
All-cause HRU and costs were compared between the PS-matched CIU/CSU and control cohorts using univariate Poisson and linear regression models. For HRU, the 95% confidence intervals (CIs) and p values for incidence rate ratios (IRRs) were calculated using the robust variance estimator to account for the correlation in PS-matched data. For costs, the 95% CIs and p values for mean cost differences (MCDs) were based on a bootstrap approach (500 resamples) to account for non-normal distribution of cost data, and used robust variance estimator to account for the correlation in PS-matched data. All analyses were performed using SAS version 9.4 (SAS Institute, NC, US).