Secondary Logo

Journal Logo

Epidemiology and Social

A taxonomy of pragmatic measures of HIV preexposure prophylaxis use

Pyra, Mariaa,b; Rusie, Lauraa; Castro, Michaela; Keglovitz Baker, Kristina; McNulty, Moirab; Bohm, Nicka; Prokurat, Andreia; Schneider, Johna,b

Author Information
doi: 10.1097/QAD.0000000000002618



Preexposure prophylaxis (PrEP) implementation is undergoing intensive scale-up as part of efforts to end the HIV epidemic. However, measures of PrEP use have largely been developed in clinical trial settings and are not accessible or comparable for real-world healthcare settings. Indeed, definitions of persistence and retention have varied widely thus far in the literature, making comparisons difficult. For instance, persistence has been assessed by medication possession ratio, proportion of days covered (PDC), and time to discontinuation [1–5]; retention has been described as quarterly visits, semi-annual visits, and quarterly HIV tests [6–10]. Adherence to long-term medications has been studied for other medical conditions and may serve as a useful guide [11–13], with some special considerations.

For the purposes of this analysis, we consider adherence broadly to be a process with specific, measurable elements [14]. First is PrEP uptake or initiation, which is not a focus of this article. While understanding the denominator, or number of individuals who would benefit from PrEP, remains a challenge, the PrEP-to-Need ratio is a useful measure of uptake [15]. Next, adherence (sometimes referred to as execution or compliance) is how well patients actually take PrEP. There are a range of options to assess adherence, from self-report as the most subjective to biomarkers as the most objective. There are key considerations in determining how to measure PrEP adherence, including how many doses per week are needed to be effective and how to account for 2-1-1 (or event-driven) dosing strategies. Another aspect of adherence, persistence, is the length of time on PrEP [16]; discontinuation is a related concept. Questions regarding seasonal or cyclic use of PrEP can make assessing persistence and discontinuation difficult; the concept of prevention-effective adherence, or using PrEP during periods of risk, is foundational in PrEP delivery [17]. Finally, a related concept is retention, which we define as maintaining HIV prevention care, with or without PrEP use.

Our goal was to determine population-level measures of persistence and retention to evaluate and compare PrEP implementation programs. These measures are not intended to evaluate how an individual patient is using PrEP, which depends on their specific context and choices. An ideal measure will be accessible to various healthcare settings and can be extended to future modalities of PrEP.


Electronic health record (EHR) data was collected for all HIV-uninfected patients who received their first PrEP prescription between January 2015 and August 2018. PrEP prescriptions were defined as tenofovir and emtricitabine with no third antiretroviral (thereby excluding PrEP prescriptions); for patients with a concurrent hepatitis B diagnosis, charts were reviewed to clarify the indication. This algorithm has previously been validated [9]. All subsequent PrEP prescriptions through September 2019 were included; we also collected data on all HIV tests starting from PrEP initiation. Any patient who was ever prescribed more than 180 pills in one prescription (including refills) was excluded as an outlier.

A summary of PrEP use measures can be found in Table 1. For all the measures, we relied on prescription date, number of pills prescribed, and the number of refills; PrEP supply was the total number of pills available from each prescription (the number of pills times the number of fills). All measures were censored at HIV acquisition. Total PrEP time was calculated as the number of months from the date of the first PrEP prescription until the end of the last supply; on PrEP at 6 months (or 12 months) was determined as whether total PrEP time was greater or equal to the respective number of months. The medication prescription ratio (MRxR) was the total number of pills prescribed divided by total PrEP time on days; this value can exceed 100% and was therefore capped at 100% [12]; a binary measure was created to determine if MRxR was at least 85% (equivalent to six doses per week) or at least 57% (equivalent to four doses per week).

Table 1
Table 1:
A taxonomy of pragmatic preexposure prophylaxis measures based on electronic health record data.

Next we calculated the PDC for each month (see Supplemental Fig. 1, for examples); this method involves moving an overlapping prescription to the end of the previous prescription, thereby extending the total length of time covered [13,18]. PDC traditionally uses more than 80% as a threshold [18], but given the data on effective PrEP use, we chose a cutoff of PDC for each month at least 85% (six doses per week) or alternatively at least 57% (four doses per week); we defined these as early (first 6 months) or late (first 12 months). These values can also be summed over the first 6 (or 12) months to provide a continuous value.

We also created two measures specifically of retention; for all measures, there was one-week window before or after for HIV test dates. Using total PrEP time (in quarters) as the denominator, retention over total PrEP time is the proportion of quarters with an HIV test. Consistent with PDC, we measured whether each quarter has an HIV test over the first 6 months (early retention) or 12 months (late retention). To note, retention over total PrEP time only includes visits during the period when a patient is on PrEP (though it includes gaps in PrEP use), while early and late retention include patients who may have completely stopped using PrEP but are still engaging in HIV prevention services. All of these measures were then compared by race/ethnic group among MSM, using chi-square tests for binary measures and Wilcoxon rank sum tests for continuous measures.

In addition, claims data from approximately 100 Walgreens pharmacies were matched to the EHR prescriptions by the date the prescription was written, with a 2-week window before or after. From this, we determined how many PrEP prescriptions were actually filled, with fill rates capped at 100%. Finally, as a measure of validation, we used dried blood spots (DBS) a subset of 169 patients who had been enrolled in the first Centers for Disease Control & Prevention (CDC) PrEP implementation study [Sustainable Health Center Implementation PrEP Pilot Study (SHIPP)] to compare the objective biomarker tenofovir-diphosphate (TFV-DP) to PDC over the prior month. In addition, we report PrEP measures comparing patients who seroconverted while on PrEP to those who remained HIV negative, as well as the area under the curve (AUC) for each measure; this analysis was limited to those who seroconverted after 6 months to avoid immortality bias. This study was approved by the Howard Brown Health Institutional Review Board.

Role of the funding source

Study sponsors were not involved in the design, analysis, interpretation of results, or writing of this report.


Data were collected from 6068 patients who ever had a PrEP prescription (Table 2). ‘While the majority of patients were white, the proportion identifying as Latinx (21.7%) and Black/African-American (18.2%) were similar’. Most patients were privately insured (including private Medicaid expansion insurance) and 26.9% were uninsured or self-pay. The majority, 89.1%, were cismen and 75.7% identified as gay. At PrEP initiation, 8% were diagnosed with chlamydia and 6% with gonorrhea. The subset of patients with DBS data were similar overall, except that they had been on PrEP longer on average. We present the examples of each PrEP measure, as well as by race/ethnicity among MSM solely as an example and not to specifically compare across race/ethnicity.

Table 2
Table 2:
Preexposure prophylaxis patient characteristics, Chicago 2011–2018.

Preexposure prophylaxis measures over all time

Total PrEP time and the related measures in Table 3 describe PrEP use over the entire duration. Overall persistence, measured by average Total PrEP time was 19.8 months, with 79% of patients on PrEP at 6 months and 64% on PrEP at 12 months. During the full period of PrEP use, adherence as measured by MRxR was 89% on average; 77% of patients had an MRxR at least 85% and 90% have an MRxR at least 57%, as measures of effective adherence. Finally for retention over all PrEP use, patients had quarterly HIV tests 71% of follow-up time, on average, with little variation across groups. All measures differed significantly by race/ethnicity.

Table 3
Table 3:
Examples of preexposure prophylaxis use measures, by race/ethnicity among MSM, Chicago 2011–2019.

Preexposure prophylaxis measures at specific time points

Also presented in Table 3 are PDC and related measures. Early PDC at least 85% was 53% on average, slightly less than early PDC at least 57% at 57% on average. Early PDC at least 85% was higher than late PDC at least 85%, at 32% on average. To compare with MRxR, we also reported average PDC, which as 81% over the first 6 months and 70% over the first 12 months. For all PrEP use measures, there were large racial/ethnic differences, significant at P less than 0.05. Retention was also higher over the first 6 vs. the first 12 months, at 59 vs. 30%; there were racial/ethnic differences in retention at both time points, also at P less than 0.05.

Comparisons of measures

Direct comparisons for some measures (for instance total PrEP time and PDC) would not be appropriate as they capture different constructs; however, Table 4 presents meaningful contrasts, although timeframes may differ across measures’. The on PrEP measure overestimates persistence relative to the PDC measures. For instance, 79% of patients are on PrEP at 6 months, while 57% have PDC at least 57% and 53% have PDC at least 85%. Similarly, the average MRxR (over all PrEP use) was 89%, compared with an average PDC of 81% over the first 6 months. The relative differences between White and Black MSM were greater for PDC measures compared with measures over all time. For instance, the odds ratio (OR) between Black and White MSM would be 0.50 [95% confidence intervals (CI) 0.42, 0.58] for PDC at least 85% and 0.68 (95% CI 0.57, 0.82) for on PrEP at 6 months. In comparing the continuous measures, the risk difference (RD) between Black and White MSM would be 9.3% (95% CI 7.3%, 11.3%) for PDC and 4.9% (95% CI 3.3%, 6.4%) for MRxR.

Table 4
Table 4:
Comparison of preexposure prophylaxis measures, by race/ethnicity among MSM in the cohort, Chicago 2011–2019.

Claims and tenofovir-diphosphate comparisons

On average, patients filled 45% of all PrEP prescriptions, compared with Walgreens claims data. Among patients (n = 3901) who filled their first PrEP prescription at a Walgreens, the average fill rate was 60% and among those whose prescription was electronically sent to a Walgreens (n = 5639), the average fill rate was 51%.

There were 224 DBS samples from the first 6 months of PrEP use (Supplemental Table 1, from which TFV-DP was quantified. Using a cutoff of 1050 fmol/punch [19,20], the sensitivity of PDC at least 85% in the prior month was 97% and the specificity was 7%. Similarly, using a cut-off of 700 fmol/punch, the sensitivity of PDC at least 57% in the prior month was 97% and the specificity was 13%.

Finally, we looked at HIV acquisition to understand the discriminatory values of these measures, though importantly, these are not meant to be predictive. We limited these measures to values that were calculated over the first 6 months only, and excluded cases that occurred during the first 6 months, to avoid immortality bias (i.e. HIV-uninfected patients would potentially have more time on PrEP than patients censored at HIV infection) (Table 5). The proportion of patients meeting each measure was higher among those who did not acquire HIV compared with those who did seroconvert. All the metrics we compared had similar, poor AUCs, ranging from 0.62 to 0.68 and overlapping 95% CIs.

Table 5
Table 5:
Binary preexposure prophylaxis measures and HIV acquisition, among all patients.


In the analysis, we present a taxonomy of measures that can make reporting and comparing PrEP use adherence across programs more meaningful. For implementation researchers, community stakeholders, and public health departments interested in the full duration of PrEP use, total PrEP time is our recommended measure of persistence, with the understanding that it includes gaps in PrEP use. Total PrEP time is also the simplest calculation, while MRxR can easily measure adherence over the same time period. This may be appropriate when considering a prevention-effective adherence framework, in which patients are expected to start and stop PrEP in conjunction with their HIV risk [17]. However, MRxR overestimates PrEP use compared with PDC. In addition, using Total PrEP Time to calculate on PrEP at 6 months can results in different values depending on when the calculation is made (see Supplemental Fig. 1,

We chose to report retention over the full duration of PrEP as a proportion of quarters with an HIV test rather than a binary measure of successfully having a test every quarter, given that the length of time on PrEP can vary widely and might confound retention. Furthermore, we defined retention as an HIV test rather than a follow-up visit, as HIV testing is part of the CDC recommendations [21] and follow-up visits may be defined differently at different clinics.

For programs that are further along with a focus on specific PrEP use time points, we favor PDC. Inherent in PDC, these measures include specific unit of time, with early measures defined as the first 6 months and late measures as the first 12 months, and include specific level of adherence as well. While we think 6 months in particular is an important value to assess early drop off, we recognize that other timepoints are also valid choices and should be clearly stated in reporting results. Likewise, we set two thresholds of effective adherence – at least 85% and at least 57% – corresponding to six and four weekly doses, respectively. The former is an appropriate choice for broad engagement with most populations; if a program is more specialized to focus on predominantly MSM the latter might be more appropriate. Furthermore, 2-1-1 dosing with at least one sexual encounter per week would also be captured by the at least 57% threshold; only if a large portion of the patient population was using 2-1-1 for rare sexual encounters would a lower adherence threshold be needed [22]. As nondaily regimens become more popular, specifying and reporting the adherence threshold used as well as the proportion of patients using nondaily PrEP will improve the comparability of adherence literature. More work to validate these measures and establish ideal thresholds of ‘success’ would be valuable, especially as new formulations of PrEP become available.

It is important to consider the future of PrEP and the ability of metrics to incorporate other PrEP modalities, such as injections or implants. For instance, total PrEP time could be from first prescription of any PrEP modality to end of any PrEP supply. For PDC-based measures, we would recommend that new PrEP modalities truncate rather than extend any overlapping PrEP prescriptions, assuming patients are more likely to switch to the new prescription immediately. Finally, the smaller the unit used to assess adherence the less likely that it will hide gaps in use; for instance, there are situations where a patient would have PDC at least 85% if measured over 6 months but have PDC less than 85% for each individual month (Supplemental Fig. 1,

We do not expect that these measures to be predictive of HIV acquisition at an individual level, which would require a much higher AUC; however, we did expect that a good measure of PrEP use should relate to the primary outcome, HIV acquisition. All of these measures demonstrated evidence of this relationship with AUCs at least 0.62 and confidence intervals excluding 0.50 (with 0.50 being nondiscriminatory). Likewise, we do not expect any population to reach 100% on any of these measures; they are meant mostly for comparison.

While self-report, claims data, and biomarker data would all provide useful adherence data, we used EHR prescriptions as a common resource that many clinical settings will have access to, though nonelectronic prescription data would also be sufficient for these metrics. Using one of the largest real-world PrEP cohorts, we were able to evaluate EHR data using claims and biomarker data. As expected, EHR data overestimate actual adherence; not all prescriptions are filled and not all filled prescriptions are taken. We found that at least 60% of prescriptions were picked up, though importantly prescriptions may have been filled at outside pharmacies where claims data was not available for the analyses described. When possible, reporting fill rates may help contextualize EHR-based adherence measures. Certainly, use of pharmacy claims data will require organizations without prescribers to work closely with clinics and pharmacies if these EHR-based measures are to be used. However, TFV-DP results showed that PDC over the prior month had a high sensitivity but low specificity; individuals who do not meet the PDC threshold are likely not taking the drug, while most but not all individuals who meet the PDC threshold are taking the drug.

In summary, total PrEP time most closely captures a strict definition of persistence, while PDC can assess persistence and adherence at a specific time point. PDC is a more conservative measure of adherence compared with MRxR, though it will likely still overestimate true adherence. While we did find larger differences by OR or RD across race/ethnic groups using PDC-based measures relative to Total Time/MRxR measures, these may be a function of our data rather than inherent properties of the measures and should be examined further. Retention in care can be measured by quarterly HIV tests, either over all time or at specific time points, and may be relevant for agencies that partner with clinics and have visit history data. If other choices in time point or adherence level are appropriate for a specific study, these should be clearly stated so that comparisons can still be made. Consistent PrEP use terminology and definitions will help compare PrEP outcomes as this important HIV prevention strategy is increasingly implemented within ending the epidemic contexts.


M.P. designed the research question and conducted the analysis. L.R., N.B., and A.P. contributed to data collection. All authors contributed to writing and editing.

J.S. was supported in part by R01 AI120700. This work was supported in part by the Third Coast Center for AIDS Research, an NIH funded center (P30 AI117943). M.M. was supported by K23 MH118969. The CDC foundation funded the DBS measurement of tenofovir-diphosphate, which was performed by the University of Colorado. The conclusions, findings, and opinions expressed by authors contributing to this article do not necessarily reflect the official position of the Centers for Disease Control and Prevention, Gilead Sciences, the University of Colorado or the authors’ affiliated institutions.

Conflicts of interest

There are no conflicts of interest.


1. Marcus JL, Hurley LB, Hare CB, Nguyen DP, Phengrasamy T, Silverberg MJ, et al. Preexposure prophylaxis for HIV prevention in a large integrated healthcare system: adherence, renal safety, and discontinuation. J Acquir Immune Defic Syndr 2016; 73:540–546.
2. van Epps P, Maier M, Lund B, Howren MB, Beck B, Beste L, et al. Medication adherence in a nationwide cohort of veterans initiating preexposure prophylaxis (PrEP) to prevent HIV infection. J Acquir Immune Defic Syndr 2018; 77:272–278.
3. Krakower D, Maloney KM, Powell VE, Levine K, Grasso C, Melbourne K, et al. Patterns and clinical consequences of discontinuing HIV preexposure prophylaxis during primary care. J Int AIDS Soc 2019; 22:e25250.
4. Dombrowski JC, Golden MR, Barbee LA, Khosropour CM. Patient disengagement from an HIV pre-exposure prophylaxis program in a sexually transmitted disease clinic. Sex Transm Dis 2018; 45:e62–e64.
5. Spinelli MA, Scott HM, Vittinghoff E, Liu AY, Morehead-Gee A, Gonzalez R, et al. Provider adherence to preexposure prophylaxis monitoring guidelines in a large primary care network. Open Forum Infect Dis 2018; 5:ofy099.
6. Chan PA, Mena L, Patel R, Oldenburg CE, Beauchamps L, Perez-Brumer AG, et al. Retention in care outcomes for HIV pre-exposure prophylaxis implementation programmes among men who have sex with men in three US cities. J Int AIDS Soc 2016; 19:20903.
7. Montgomery MC, Oldenburg CE, Nunn AS, Mena L, Anderson P, Liegler T, et al. Adherence to pre-exposure prophylaxis for HIV prevention in a clinical setting. PLoS One 2016; 11:e0157742.
8. Hevey MA, Walsh JL, Petroll AE. PrEP continuation, HIV and STI testing rates, and delivery of preventive care in a clinic-based cohort. AIDS Educ Prev 2018; 30:393–405.
9. Rusie LK, Orengo C, Burrell D, Ramachandran A, Houlberg M, Keglovitz K, et al. Preexposure prophylaxis initiation and retention in care over 5 years, 2012–2017: are quarterly visits too much?. Clin Infect Dis 2018; 67:283–287.
10. Zucker J, Carnevale C, Richards P, Slowikowski J, Borsa A, Gottlieb F, et al. Predictors of disengagement in care for individuals receiving preexposure prophylaxis (PrEP). J Acquir Immune Defic Syndr 2019; 81:e104–e108.
11. Stolpe S, Kroes MA, Webb N, Wisniewski T. A systematic review of insulin adherence measures in patients with diabetes. J Manag Care Spec Pharm 2016; 22:1224–1246.
12. Hofmeyer BA, Look KA, Hager DR. Refill-based medication use quality measures in kidney transplant recipients: examination of proportion of days covered and medication possession ratio. J Manag Care Spec Pharm 2018; 24:367–372.
13. CDC. Calculating proportion of days covered (PDC) for antihypertensice and antidiabetic medications: an evaluation guide for grantees. Atlanta, GA: CDC; 2015. [Accessed 19 November 2019].
14. Raebel MA, Schmittdiel J, Karter AJ, Konieczny JL, Steiner JF. Standardizing terminology and definitions of medication adherence and persistence in research employing electronic databases. Med Care 2013; 51:S11–S21.
15. Siegler AJ, Mouhanna F, Giler RM, Weiss K, Pembleton E, Guest J, et al. The prevalence of PrEP use and the PrEP-to-need ratio in the fourth quarter of 2017, United States. Ann Epidemiol 2018; 28:841–849.
16. Cramer JA, Roy A, Burrell A, Fairchild CJ, Fuldeore MJ, Ollendorf DA, Wong PK. Medication compliance and persistence: terminology and definitions. Value Health 2008; 11:44–47.
17. Haberer JE, Bangsberg DR, Baeten JM, Curran K, Koechlin F, Amico KR, et al. Defining success with HIV preexposure prophylaxis: a prevention-effective adherence paradigm. AIDS 2015; 29:1277–1285.
18. Nau DP. Proportion of days covered (PDC) as a preferred method of measuring medication adherence. Pharmacy Quality Alliance; 2012.
19. Castillo-Mancilla JR, Zheng JH, Rower JE, Meditz A, Gardner EM, Predhomme J, et al. Tenofovir, emtricitabine, and tenofovir diphosphate in dried blood spots for determining recent and cumulative drug exposure. AIDS Res Hum Retroviruses 2012; 29:384–390.
20. Anderson PL, Liu AY, Castillo-Mancilla JR, Gardner EM, Seifert SM, McHugh C, et al. Intracellular tenofovir-diphosphate and emtricitabine-triphosphate in dried blood spots following directly observed therapy: the DOT-DBS study. Antimicrob Agents Chemother 2017; 21:e01710–e01717.
21. Centers for Disease Control and Prevention; US Public Health Service. Preexposure prophylaxis for the prevention of HIV infection in the United State-2017 Update: a clinical practice guideline. Atlanta, GA: CDC; 2017. [Accessed 13 June 2018]
22. Antoni G, Tremblay C, Delaugerre C, Charreau I, Cua E, Castro DR, et al. On-demand preexposure prophylaxis with tenofovir disoproxil fumarate plus emtricitabine among men who have sex with men with less frequent sexual intercourse: a posthoc analysis of the ANRS IPERGAY trial. Lancet HIV 2019; 7:e113–e120.

adherence; HIV preexposure prophylaxis; persistence; retention

Supplemental Digital Content

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.