Article Title

Linking Medication Orders to Prescription Fills: Lessons From HMO Enrollees Applicable to PPO Patients

Publication Date



prescription fill, medication orders


Background/Aims: This study develops an approach to estimate the prescription fill rate when detailed medication orders from electronic health records are available but pharmacy dispensing data are limited.

Methods: Data included medication orders from the electronic health records and pharmacy claims from a large ambulatory group practice with mixed payment sources. Pharmacy claims were available only for HMO patients, but no direct link was available to track whether a pharmacy claim was tied to a particular medication order. We analyzed 608,871 medication orders from 2007 to 2013 for 26,875 HMO patients who were enrolled in an HMO plan in the calendar year preceding, during and after the medication ordering date. We developed an algorithm matching the pharmacy claims to medication orders by the patient identifier, the generic product identifier and the prescription fill date within the eligible timeframe (i.e. within 365 days of the ordering date or between the ordering date and a specified termination date). Fill rates were computed separately for the index (i.e. newly initiated therapy) and renewal orders for each drug. Sensitivity analyses were conducted using different matching algorithms, e.g. using simple generic name and medication strength.

Results: Overall, 75% of the 608,871 medication orders had matched pharmacy claims using the proposed algorithm. Fill rates were higher among renewal orders than index orders (76% vs. 74%, P<0.01). Preliminary findings suggested a higher fill rate for chronic conditions compared to acute conditions (78% vs. 74%, P<0.01). In addition, there were variations across therapeutic classes, ranging from 83% fill rate of anti-infective agents to 40% of nutritional products. Sensitivity analyses using different matching criteria yielded a similar overall fill rate (73.6%).

Discussion: We developed a generic approach to estimate the prescription fill rates based on medication orders and pharmacy claims from HMO patients. This is an important first step, as the estimated fill rates can serve as a benchmark and be applied to patients who did not have pharmacy dispensing data in the same delivery system. Applications of estimated fill rates can be further extended to estimate pharmacy costs associated with medication orders or examine patterns of medication adherence over a range of conditions.




April 6th, 2015


April 28th, 2015