Copayments are a commonly used patient level incentive for modulating the demand for services. Prior studies have demonstrated that copayments and other forms of cost sharing lead to decreased utilization, but none have had sufficient sample sizes to evaluate effects on patient outcomes. The dearth of information on outcomes is particularly concerning given data suggesting that the lack of insurance is associated with poor outcomes. Furthermore, questions remain about whether copayments result in higher downstream treatment costs. To address these issues, we propose to evaluate the effects of the size of a copayment for emergency department (ED) use on patient outcomes and treatment costs within the Kaiser Permanente- Northern California (KP) health system, using a quasi-experimental pre- post design, with concurrent controls. KP is in the midst of a natural experiment, with nearly half of the 2.7 million member population experiencing a sizable increase in ED copayment level on January 1, 2000; 47 percent of members now face a 25-35 dollars ED copayment. The size of the ED copayment will be the main predictor; and the main outcomes of interest are population rates of hospital admissions, ICU admissions, mortality, and the mean costs of hospital care (treatment costs). We will investigate these rates by the level of copayment within the Total Population (KP), as well as within select Vulnerable Populations. We also will investigate the adverse outcome rates and treatment costs for select Sentinel ED Diagnoses. We will use the Cox proportional hazard model to test the hypotheses that the copayment size is related to rates of adverse outcomes. To test our hypothesis relating copayment size to treatment costs, we will use a two-part model. We also will make adjustments in our analyses for relevant patient and organizational factors, such as SES (census block group characteristics, employer type), case-mix, medical center, and use of a primary care physician. These factors may influence ED use independently or modify the association of ED copayment and patients' decisions to access ED care. Given the sample size, our study will have the ability to detect small changes in relatively rare adverse outcomes, e.g. 95 percent power to detect a difference of 2 deaths/10,000 person- years (baseline rate = 56 deaths/ 10,000py); also, the assignment of copayments at the employer level (rather than the patient level) reduces concerns of confounding by self-selection in this natural experiment. In short, the issue of the safety of copayments is of tremendous and increasing importance as the use of financial incentives becomes more common, and concerns about patient safety and choice increase among patients, providers, and policy makers.
