One of the biggest challenges to pharmacies has always been limited access to patient information such as diagnoses, laboratory values, and physical exam records. While pharmacists may not always have access to patient charts, we do have access to one very valuable piece of information not held by our medical colleagues: prescription claim records. There is much that pharmacists can do with this rich data set.
Medication adherence, as measured by medication possession ratio (MPR), can easily be calculated from the data in the dispensing system report. The data points needed are the unique patient identifier, drug name, days supplied, and date of fill. Using these data, you can calculate MPR using the formula in Figure 1. The resulting MPR usually ranges from 0 to 1, with 1 corresponding to 100% adherence. It is possible for MPR to exceed 1; this case is discussed below.
Figure 1. Calculating the medication possession ratio
For a single patient, MPR can be manually calculated quickly and easily. But for a large patient population, manual calculations are not practical. Instead, you can use Microsoft Excel functions such as sort, remove duplicates, VLOOKUP, and pivot tables. You can learn more about these functions on the Microsoft Office online help page (http://office.microsoft.com/en-us/support/). Figure 2 explains the steps involved in calculating MPR within a spreadsheet, which can be applied to a large patient population.
Figure 2. Calculating medication possession ratios in Excel
While prescription claim records are objective measures of adherence that are relatively inexpensive to obtain, they do have some shortcomings. First and foremost, dispensing a medication does not necessarily entail its use. Also, MPR may be overstated or understated early in therapy due to dose adjustments (e.g., titrations). Patients filling their prescriptions at more than one pharmacy will have low MPR values due to incomplete claim records. Last, because days’ supply is manually entered by pharmacy staff and can require adjustment due to third party payer regulations, it may not reflect the true number of days’ supply of the prescription.
Given these shortcomings, there are a variety of ways to improve the accuracy of your results. First, if your dispensing system can report a pick-up or delivery date instead of fill date, the MPR calculation will better reflect the patient’s actual possession of a medication. Another adjustment is to include a washout period. To do this, run your report beginning with fill dates at least 3 months prior to the timeframe of interest. Then, after completing steps one through six in Figure 2, eliminate patients with a first fill occurring in that lead-in period. This will help ensure that your patient population includes only patients who started therapy in the timeframe of interest. Similarly, if you are evaluating adherence for patients on a medication with a predefined duration of therapy (e.g., teriparatide [Forteo—Eli Lilly], hepatitis C regimens), you can use an extended window of time and eliminate patients with a last fill date occurring in the last 3 months. This will generally limit your population to patients who have completed or discontinued their regimen.
It is usually recommended to remove patients with only one fill. By definition, the MPR for these patients is always 1, and including these data will artificially inflate the overall MPR for your patient population. MPR for the patient population can also be exaggerated by patients with repetitive early refills. A solution for this is to cap MPRs at 1. To do so, create a column within the original worksheet (column K in our Figure 2) and enter the formula “=IF(J2>1, 1, J2)” without the quotation marks. This will reduce all values greater than 1 to 1, thus eliminating overstated MPRs.
Once MPR has been calculated, the duration of therapy is easy—it is simply equal to the denominator of the MPR equation: last fill date minus first fill date plus days’ supply at last fill. To ensure that this calculation reflects the true duration of therapy, utilize a washout period as described above.
With little additional information, you can use similar techniques to determine other patient and population characteristics, such as proportion of patients with a dose change or number of patients receiving a specific adjunct therapy. This could be useful with, for example, patients with hepatitis C virus (HCV) who have a multidrug regimen that requires high levels of adherence for a specific period of time dictated by genotype, response, and therapy tolerance. Ribavirin, one of the agents in this regimen, is known to cause hematologic adverse effects that often lead to dose reductions or the initiation of erythropoietic agents.
A pharmacy may want to provide an adverse effect management program to patients with HCV to help them remain adherent to their therapy. Using the steps outlined above, this pharmacy can evaluate patient adherence and duration of therapy, two important measures of this program’s effectiveness. Another valuable metric is the proportion of patients who have had a dose reduction or are also prescribed epoetin alfa.
To determine dose reductions, generate a report for patients using ribavirin similar to the report described above for MPR calculations, but include the ribavirin dose as a separate column. Note that in this example, the total daily dose must be used; this may need to be calculated in a new column using tablet strength and quantity dispensed.
Next, create a pivot table including all of the columns, using patient ID for row label and dose twice for value—one with the value field settings set to maximum and one to minimum. The difference between these columns is the dose reduction. A number of additional metrics can be drawn from this calculation, including proportion of patients receiving a dose reduction and average dose reduction.
To evaluate the use of epoetin alfa in conjunction with dose reductions, generate an epoetin alfa dispense report and then use Excel’s VLOOKUP function to combine epoetin alfa data with the ribavirin report. This time, create the pivot table using the added adjunct therapy column as a filter to compare dose reductions with the use of epoetin alfa.
This is just the beginning of what a pharmacy can do with its prescription claim records. With a little time and motivation, pharmacists can dive much deeper into these data to calculate metrics such as median gap and correlation between select patient characteristics and refill habits. Many dispensing systems even allow users to assign patients to certain groups, such as disease states or programs, which could also help stratify the results of data analyses.