Adapting the Proportion of Days Covered (PDC) to account for multiple pharmacy providers
Dr Vanessa Cooper
The proportion of days covered (PDC) for estimating adherence to medicines
There is no gold standard method for assessing medication adherence. Data from pharmacy databases are commonly used to estimate patient adherence. Although this information cannot tell us whether the patient took their medication as prescribed, it can tell us whether the medicine was available for the person to take. A widely used method for estimating adherence using pharmacy data is the proportion of days covered (PDC). The PDC has some advantages over other methods, such as its ability to account for discontinuation of medicines, but the extent to which it stands up as an effective measure depends on the availability of dispensing information.
The PDC looks at the proportion of days a person has access to their medication over a given period of interest. It is generally calculated as:
Sum of days covered by medication during the period of interest | x 100 |
| Number of days medicine should be taken during the period of interest |
The rise of online pharmacies
The use of online pharmacies has rapidly increased in recent years1. This increase was accelerated during the coronavirus pandemic where online pharmacies provided patients with a method for obtaining medicines while reducing contact with others.2 Online pharmacies have potential benefits for patients including the convenience of being able to order medications online or over the phone at any time of day, delivery of medicines to the door and round the clock customer care.
The need to identify and evaluate solutions for estimating adherence
We now have greater choice of how to fill our prescriptions. We may choose to fill our prescriptions from more than one pharmacy. Unless these pharmacies share records, when a patient switches from one pharmacy to another (Pharmacy 1 to Pharmacy 2), the records at Pharmacy 1 will show a large gap between the end of a supply of medicine and the next dispensation. When applying the traditional PDC this large gap will be assumed to a result of nonadherence (a drug holiday) when in reality the patient obtained their medicine from Pharmacy 2 and had a continuous supply. Since more patients are obtaining medicines from more than one pharmacy, it is now important to identify and evaluate solutions for estimating adherence that can account for large gaps in the supply of medicines from a single pharmacy.
“Previous papers have explained the limitations and problems of PDC calculations but do not propose any solutions. Papers highlighting the limitations of methods are useful, but we also need papers proposing solutions. Our paper addresses this gap by both proposing and testing solutions for addressing limitations of the PDC”
David Prieto-Merino, Associate Professor at the Faculty of Epidemiology & Population Health, London School of Hygiene & Tropical Medicine
Adapting the PDC for use in today’s pharmacy landscape
In partnership with Pharmacy 2U, the UK’s largest online pharmacy, Sprout researchers conducted a study to identify and evaluate solutions to address limitations of the PDC3. The team identified three variations of the algorithm which differ in terms of the assumptions they make about the denominator (the number of days the medication should be taken during the period of interest).
The conventional approach (PDC1) used the total number of days between first dispensation and a defined end date as the denominator. A second algorithm (PDC2) counted only the days between first dispensation and the end of supply date. The third algorithm (PDC3) removed specific large gaps between fills (where patients may have purchased their medication from another pharmacy) from the denominator.
The team applied each algorithm to an anonymised, real-world dataset from Pharmacy 2U. The dataset included patients taking once daily medicines (ACE inhibitors (n= 65,905), statins (n= 100,362), and/or thyroid hormones (n= 30,637)) during a two-year period. The proportion of people estimated to have a high level of adherence (defined as a PDC of ≥ 0.8) was compared when applying PDC1, PDC 2 and PDC 3.
Estimates of high adherence according to PDC 1, 2 and 3 in people taking ACE inhibitors
Version of the algorithm applied | High adherence (PDC ≥ 0.8) |
PDC1 | 50-74% |
PDC2 | 81-91% |
PDC3 | 86-100% |
*Similar ranges were identified in people taking statins and thyroid hormones.
What do the findings mean?
The PDC3 algorithm was able to pick up large gaps between medication refills. The results were in the expected direction, showing the lowest levels of adherence when applying the traditional PDC and the highest level of adherence when large gaps in supply were assumed to be due to the patient obtaining their refill from a different pharmacy.
These findings have practical value. One example of how PDC2 and PDC3 could be applied is in NHS England, where patients can nominate a pharmacy where their prescriptions are sent and have the right to switch to another pharmacy or to a paper-based prescription without informing the originally nominated pharmacy. The choices patients make may create gaps in pharmacy data that are not due to nonadherence.
Providing a choice of algorithms for calculation of the PDC enables researchers and healthcare providers to evaluate pharmacy services and estimate levels of adherence in real world settings where there may be multiple suppliers of medicines to patients. This is important when the researcher does not have information from all the suppliers, leading to temporary gaps in the data available to estimate treatment coverage. If no flexibility was allowed in the PDC calculation, these gaps in available data would lead to underestimation of adherence. This could be completely unrealistic for critical treatments. The definition of a ‘large gap’ when using PDC3 can be adapted to the specific disease and treatment of interest, providing even more flexibility and personalisation of the algorithm.
Further research with patients and prescribers is now required to validate the algorithms’ assumptions, for example, the assumption that large gaps are due to patients obtaining their medication from an alternative supplier. However, we believe that this paper is an important first step in adapting the PDC to address changes in the way that people use pharmacies.
Read the paper here:
https://joppp.biomedcentral.com/track/pdf/10.1186/s40545-021-00385-w.pdf
References
Wickware, C. Online pharmacy dispensing volume grows by 45% in 2020, fuelled by COVID19 pandemic. The Pharmaceutical Journal PJ, April 2021, Vol 306, No 7948;306(7948)::DOI:10.1211/PJ.2021.1.78119
Bhatt, H. From shopping-in-store to online shopping: change in consumer behaviour during the pandemic. Jan 11, 2021. https://socialnomics.net/2021/01/11/from-shopping-in-store-to-online-shopping-change-in-consumer-behavior-during-the-pandemic/
Prieto-Merino, D., Mulick, A., Armstrong, C., Hoult, H., Fawcett, S., Eliasson, L., Clifford, S. Estimating proportion of days covered (PDC) using real-world online medicine suppliers’ datasets. J Pharm Policy Pract.2021; 14: 113.
