Association of COPD Maintenance Medication Adherence With Resource Use and Cost Among COPD Patients (MARU)

Objectives: The primary objective of this study is to examine the association of COPD maintenance inhalation medication (Inhaled Corticosteroid or ICS) adherence with COPD exacerbation healthcare resource utilization among COPD patients with exacerbation history.

Study design: This study is a retrospective database analysis using 2014-2016 Guangzhou City health insurance database.

Data Source(s): Guangzhou city health insurance database will be used for this study. All outpatient and inpatient visits for one patient during 2014-2016 were available.

Study Population: This study will include patients with a physician COPD diagnosis in Guangzhou city health insurance database during 2015 and with complete treatment data during 1 year before and 1 year after the index date. The rough sample size of the study population was 49,000.

Outcome(s):

1. Indicators of COPD control: exacerbations;
2. Health resource utilization: annual number of inpatient days, inpatient visit number, emergency department (ED) visit number;
3. Costs: COPD exacerbation hospitalization expenditures, all-cause hospitalization expenditure.

Statistical Analysis:

Descriptive statistics will be used for all variables, as appropriate. Continuous variables will be summarized by the number of observations, mean, standard deviation, median, minimum, maximum, first quartile and third quartile. Categorical variables will be summarized by frequency counts and percentages at each category. Number of patients and number of missing for each variable will be summarized.

Logistic regression will be used to estimate the risk of any hospitalization. Generalized Linear Models (GLMs) with a gamma distribution and log link will be used to approximate the highly right-skewed distribution of medical expenditure.

For exploratory objectives, Propensity Score Matching (PSM) will be used to explore the association of medication class and health resource utilization if applicable. The interaction of medication class * adherence or subgroup analysis (by medication class) will be considered when conducting the regression model.