The effect of insurance-driven medication changes on patient care

We based our decision to mail 1200 surveys on a power calculation assuming a 40% response rate and approximately 25% of patients reporting a problem. A sample size of 480 or more provides 80% power to detect moderate differences in characteristics between those reporting a problem and those not reporting a problem.

Results

Four-hundred thirty-four patients returned the survey (36% response rate). We excluded 6 participants from analysis due to incomplete data for the primary outcome variable (problem with a prescription). Respondents and nonrespondents were similar in sex ratio, but respondents on average were 3 years older (P<.001). The average number of prescriptions taken was 3.4, and most patients (85%) had some form of private insurance ( TABLE 1 ). Most patients were female, in good health, and well educated.

Of the 428 study participants, 100 (23%) reported at least one problem obtaining a prescribed medication due to insurance. Generally, those who experienced a problem were younger, more likely to be female, and reported poorer health status than those reporting no problem ( TABLE 1 ). Additionally, those who encountered a problem were more than twice as likely to rely solely on Medicaid or Medicare, and were also taking more prescription medications. Problems filling a prescription were also reported more often in an urban setting than in suburban or semirural areas.

TABLE 1
Demographic variables for patients who did and didn’t report problems filling prescriptions

Using logistic regression, we analyzed a model that included all significant variables (age, total number of prescription drugs taken, sex, health status, insurance type, and practice location). The final logistic regression model showed statistical significance for only 3 variables: type of insurance, total number of prescription drugs taken, and age. (When we included type of insurance in the analysis, practice location was not associated with a problem filling a prescription.)

Specifically, the independent predictors of an insurance-related problem in filling a prescription were reliance solely on government-provided insurance, as opposed to private insurance or government insurance supplemented with private insurance (odds ratio [OR]=1.90; 95% confidence interval [CI], 1.02-3.61); taking more prescription medications (OR=1.19; 95% CI, 1.10-1.29); and being younger (OR=0.96; 95% CI, 0.94-0.99).

Respondents reporting at least one insurance-driven impediment to filling a prescription encountered an average of 2.9 different types of resultant problems ( TABLE 2 ). Insurance-related problems with medications were not limited to new prescriptions. Of the 100 patients reporting a problem with a medication, 21% had a problem with a new prescription, 42% with a medication they were already taking, and 37% with both a new and a previously prescribed medication.

TABLE 2
Resultant problems when patients had at least one insurance-related issue filling a prescription in the previous year

Forty-one percent of patients reporting a problem experienced adverse medical outcomes. The most serious adverse medical outcomes were reported least often, but occurred nonetheless: worsening of medical condition (8%), new medication adverse effects (6%), and requiring a visit to the ED (5%). More commonly reported was decreased satisfaction with the health care system (68%). Patients were less likely to report being upset with their physician than their insurance company or pharmacist. Problems that burdened the physician practice were reported most frequently (83%).

TABLE 3 shows the medication categories that were affected when respondents reported at least one problem. Formulary changes or restrictions involving cardiac/hypertension/lipid and neurologic/psychiatric medications were linked to the most problems.