227
Estimating the Effects of Anti-Cholesterol Drugs on Hospitalization Risk
Emilia Simeonova[1], Niels Skipper[2] and Peter Rønø Thingholm[3]
In most developed countries, Cardio Vascular Disease (CVD) related hospitalizations account for a large share of the inpatient care and amounts to billions of dollars yearly in health care expenditures. This makes reducing the incidence of CVD hospitalizations a primary concern for insurance providers, patients and policy makers. This study provides evidence that adherence to cholesterol lowering drugs is an efficient and cost-effective path to reduce the individual risk of CVD hospitalization.
It is widely accepted that high levels of cholesterol is a primary cause of CVDs and efforts to lower cholesterol levels, reduce lipid accumulation and prevent CVDs are made by policy makers and physicians alike. Anti-cholesterol drugs such as statins provide an effective and fast approach to obtain this, but as continuous consumption of statins is required, adherence to the prescribed treatment regimen becomes a central determination of clinical outcome. However, previous work investigating the relation between adherence and hospitalizations suffer from endogeneity in the adherence decision. This endogeneity works through channels such as the healthy adherer effect, which is widely documented to cause a negative relationship between adherence and hospitalization risk as highly adhering individuals are more likely to be healthy in other aspects as well.
Using all pharmacy records in Denmark from 2004 to 2010, we identify all users of anti-cholesterol drugs, and through the unique social security number, we link each individual to the prescriber, hospitalization records and demographics. Using this high quality administrative data, we propose an instrumental variables approach to circumvent the endogeneity issues apparent in the adherence decision. It is generally accepted that the general practitioner is a significant contributor to the adherence decision by individuals, and we propose to instrument individual adherence by the variation in the physician specific ability to facilitate and promote adherence to prescription drugs.
Using the instrument we are capable of estimating robust significant negative effects of adherence on CVD hospitalization risk among the group of patients previously hospitalized with a cardiovascular disease. Furthermore, we provide evidence indicating that improving adherence to anti-cholesterol drugs is highly cost-effective and potentially results in large health care expenditure savings.