OBJECTIVES: Remission, the spontaneous resolution of symptoms among patients, is one of the characteristics of chronic spontaneous (or idiopathic) urticaria (CSU/CIU). Limited literature currently exists regarding remission rates among chronic urticaria (CU) or CSU/CIU patients and available estimates vary. Due to the requirements of economic modelling (e.g. cycle length), these estimates cannot be incorporated directly. The objective was to adapt the remission rates to model cycle length, using various statistical methods.
METHODS: From a systematic review, 4 studies reporting proportions of patients undergoing remission at different time points were identified. One of the studies reported data for 2 populations (CSU/CIU and all chronic urticaria patients), therefore, 5 populations were considered in total. A four-step approach was undertaken: (1) converting reported data to standard time units; (2) using the extracted data to run the Kaplan-Meier (K-M) analysis; (3) applying four statistical distributions (exponential, log-normal, weibull and log-logistic) to identify the distribution best fitting the literature estimates. Lowest Kolmogorov-Smirnov (KS) distance was chosen as the criterion for the best fit distribution; (4) values obtained from the best fit distribution were further converted into rates for each 4-week cycle length. The analysis was carried out for 78 years to correspond to the lifetime horizon of the cost-effectiveness model.
RESULTS: Based on the KS distance, log-normal distribution was the best fit for 2 populations and log-logistic for 3 populations. Remission rates were generated for these 5 populations which ranged from 9.5% to 37.7% for year 1, 29.5% to 70.8% for year 5 and 49.6% to 91.5% for year 20.
CONCLUSIONS: This approach provides a robust statistical method for adapting the literature estimates as per the requirements of an economic model. Due to the wide range of remission estimates in the literature, face validation via expert clinical opinion is recommended to determine appropriate model inputs.
