The overall intent of developing a conceptual model was to set out on a series of general definitions that could be broadly applied across the fields of electronic data methods and medical informatics and that were applicable in estimating adherence to oral medications for chronic diseases. These definitions were intended to encompass the many opportunities for patients to accept or decline medication within the prescribing and dispensing components of the adherence, persistence, and discontinuation continuum of behaviors. In developing the conceptual model, we applied the following principles: (1) develop a clear, concise, and broadly applicable model that described a sequence of discrete behaviors overtime; (2) articulate key constructs and subconstructs of the continuum; (3) express and organize categories in a logical and systematic manner; (4) provide sufficient detail to clarify the approach and enable its use; and (5) avoid restricting appropriate study-specific decisions (eg, observation windows).
Adherence and persistence definitions require study-specific decisions to yield operational definitions of, for example, the period of observation (observation window), the patient sample under study, and the period over which the prescription must be filled after the initial order is written. The duration of the observation window is often conditional on the context, type of medication or disease state, unique details (eg, usual days’ supply of drug dispensed), and selection criteria (eg, patients with at least 2 dispensings). Other study-specific decisions often include specifying an observation timeframe after the medication was dispensed as well as measurement metrics and tools such as formulas to calculate medication possession or gaps in medication availability. The definitions outlined here are focused on the structure of the definitions rather than on the specifics of the operational definitions. However, some operational definitions are provided to remind researchers that these decisions must also be clearly documented. Although the primary focus of developing this conceptual model was to provide broadly applicable definitions, we also identify and classify existing metrics that enable use of these definitions. The Kaiser Permanente Colorado Institutional Review Board (IRB) determined that the activities involved in writing this paper met the federal and institutional criteria for exemption from IRB review.
Our proposed conceptual model, terminology, and definitions of medication adherence for electronic data–based methods (Fig. 1 and Table 1) is predicated on 2 key constructs: medication adherence and medication persistence. Adherence connotes the degree or extent to which the patient conforms to the medication use recommendations specified by the prescriber (eg, frequency/interval of administration, time of day ingested, strength of dosage).20 In contrast, persistence encompasses the time over which a patient continues treatment or continues to refill the prescription, from starting to stopping therapy.20,47
Medication persistence implies that the patient must have exhibited at least primary adherence because persistence over time cannot be measured unless the patient has received at least the first dispensing (Fig. 2). We propose that early-stage persistence be defined to include individuals with at least 2 dispensings and later-stage persistence as including individuals with 3 or more dispensings of the medication and with evidence of medication availability. For consistent terminology, we propose the converse of each persistence category to also be “non” as in “early-stage nonpersistence” and “later-stage nonpersistence.”
Medication discontinuation implies that a patient has terminated therapy as evidenced by not refilling a prescription, but no subjective inference regarding appropriateness is made, as discontinuation may be initiated either by the clinician or the patient. Further, in claims database studies, it is usually not possible to determine whether discontinuation was prescriber-initiated or patient-initiated. Medication discontinuation in electronic database studies can only be assessed within the context of a prespecified operational definition for the required number of days without medication available. Thus, very low measured levels of adherence (MPR or PDC<40%; CMG or NPMG>60%) can in some circumstances represent, or be confused with, discontinuation.
Metrics that have been used to calculate medication adherence and/or persistence using electronic databases are summarized in Table 2. In general, these metrics enable calculation of either medication possession (ie, possession measures) or gaps in medications availability (ie, gap measures)18,40,48,49 and most estimate adherence only among individuals with secondary adherence. Most metrics are continuous measures, but they are often categorized (eg, low or inadequate vs. moderate vs. high or adequate adherence). These measures require data including the date of medication dispensing, days’ supply dispensed with each dispensing, and previous (stockpiled) medications (or an indication that it will be set to 0) to estimate medication availability and consumption, usually estimated between the first and terminal dispensings within an observation window. A minority of metrics estimate availability within a single dispensing interval (eg, the continuous, single-interval measure of medication acquisition). The metrics also vary in whether or not the days’ supply dispensed with the terminal dispensing is included in the calculation. The time between any 1 dispensing and the subsequent dispensing is known as the refill interval. Person-time is censored at the last dispensing date, at the time of exhaustion of the last days’ supply, or at a fixed number of days after exhaustion of the last days’ supply. Most gap measures of secondary adherence censor after the last dispensing once stockpiled medications have been exhausted.
The 2 most commonly used secondary adherence medication possession measures are the medication possession ratio (MPR) and the proportion of days covered (PDC).18,40–42,44,47 Both report medication availability by estimating the proportion of prescribed days’ supply obtained during a specified observation period over refill intervals and both are becoming widely applied in health care settings50 in large part because they are easily calculated (a SAS macro has been written for MPR and PDC). For example, as operationalized by the Pharmacy Quality Alliance,50 the PDC has been endorsed by the National Quality Forum as a tool to measure health care quality.51 The main difference between the PDC and the MPR is that with the PDC any oversupply is truncated, whereas adherence values of >100% are allowed with the MPR. There is controversy about whether “over adherence,” often considered as MPR between 100% and 120%, has clinical meaning.41 A shortcoming of these (and other) secondary adherence measures is that, when integrating across several observation periods of multiple refills each, delayed dispensing(s) in 1 observation period can be numerically counterbalanced by early dispensing in a later observation period, thus potentially underascertaining adherence in 1 observation period and overestimating it in another. The converse can also occur. This drawback is of particular importance in longitudinal assessments where changes in adherence behavior are assessed across multiple observation periods by calculating the adherence metric separately within each period. Other strengths and weaknesses of the medication possession and gap measures are summarized in Table 2.
Because most measures of adherence require at least 2 dispensings, the least adherent patients (primary nonadherent) are excluded. Within the last few years, measures have been developed that include patients with either primary or secondary (non)adherence.6 One such metric, the New Prescription Medication Gap (NPMG) measure, is defined as the proportion of days within an interval bounded by the prescriber’s initial EHR prescription medication order date and the end of the observation period (or end of follow-up if censored or the therapy is switched or discontinued).6 As with older gap measures, NPMG is a continuous measure, ranging from 100% for patients who obtain no medication to 0% for those who consistently refill their medication in a timely manner. Unlike secondary adherence measures, NPMG was designed to evaluate medication supply starting at prescribing and ending at a fixed censoring point, thus comprehensively capturing (non) adherence for those who never start the prescribed medication or who discontinue it early as well as for those who have at least 2 dispensings. An additional strength of NPMG is that because it enables evaluation from the point of prescribing in the EHR, person-time can be censored if the prescriber switches or discontinues therapy and documents those orders in the EHR.
In this paper, we offer a standardized set of definitions and terminology for assessing medication adherence and persistence in electronic database studies, whether the databases employed are medications ordered within an EHR, medications dispensed and documented in a pharmacy database, or pharmacy claims processed through an insurance database. These conceptual models and terminology are more comprehensive than current, commonly used definitions. The models we propose include clear and systematic definitions of adherence, and persistence developed to facilitate EHR-based research are specific to medication adherence, and extend to adherence and persistence subcategories and medication discontinuation. We also point out the importance and utility of developing precise operational definitions for adherence research as these definitions enhance the precision of ascertaining whether a patient was likely exposed to a specific medication on a particular date for a specific research purpose (eg, on the date of some clinical measure, event, or outcome).
Research focusing on adherence is voluminous. As obtaining the initial prescription medication and taking the medication are prerequisite health behaviors for medication effectiveness, these are key explanatory variables when observed effectiveness is lower than the efficacy demonstrated in controlled trials. As a consequence, comparative effectiveness studies may be designed to evaluate the effectiveness of various interventions to improve these adherence behaviors in their own right. Meta-analytic studies are also useful in assessing adherence as an outcome, and the level of adherence necessary to achieve treatment goals. To facilitate these types of studies, it is critical that the adherence measure be accurately estimated and consistent across studies.
There are limitations to this work. Our conceptual model is specific to prescription-filling and has not been compared with frameworks for other prescription behaviors such as medication-taking.20,25 Definitions are only part of the decision-making process in adherence research. Many important methodological considerations that should be addressed were beyond the scope of this paper such as identifying clinically meaningful categorizations for adherence (eg,<20% CMG; ≥80% MPR) based on observed relationships between adherence and clinical outcomes for specific disease states, the role of informative censoring (eg, medication stop orders and switches), comprehensive assessment of discrete events (eg, appropriate timeframes to consider for medication discontinuation), and bias associated with assuming dispensing data are complete (eg, prescriptions transferred outside an integrated health care system, paying cash for prescriptions resulting in no prescription insurance claims being filed).
We offer a set of standardized medication adherence terminology and definitions for use with electronic database research. The medication adherence and persistence conceptual models and definitions we present will enable future meta-analytic and comparative effectiveness research, as standardized terminology facilitates rigorous comparisons. As such, this paper is foundational for adherence methods.
1. Steiner JF.Rethinking adherence.Ann Intern Med.2012;157:580–585.
2. McGinnis B, Olson KL, Magid D, et al..Factors related to adherence to statin therapy.Ann Pharmacother.2007;41:1805–1811.
3. Batal HA, Krantz MJ, Dale RA, et al..Impact of prescription size on statin adherence and cholesterol levels.BMG Health Serv Res.2007;7:175.
4. Briesacher BA, Andrade SE, Fouayzi H, et al..Comparison of drug adherence rates among patients with seven different medical conditions.Pharmacotherapy.2008;28:437–443.
5. Chan DC, Shrank WH, Cutler D, et al..Patient, physician, and payment predictors of statin adherence.Med Care.2010;48:196–202.
6. Karter AJ, Parker MM, Moffet HH, et al..New prescription medication gaps: a comprehensive measure of adherence to new prescriptions.Health Serv Res.2009;44:1640–1661.
7. Kane S, Shaya F.Medication non-adherence is associated with increased medical health care costs.Dig Dis Sci.2008;53:1020–1024.
8. Ho PM, Magid DJ, Masoudi FA, et al..Adherence to cardioprotective medications and mortality among patients with diabetes and ischemic heart disease.BMC Cardiovasc Disord.2006;6:48.
9. Kramer JM, Hammill B, Anstrom KJ, et al..National evaluation of adherence to beta-blocker therapy for 1 year after acute myocardial infarction in patients with commercial health insurance.Am Heart J.2006;152:454e1–454e8.
10. Raebel MA, Carroll NM, Ellis JL, et al..Importance of early non-adherence in estimations of medication adherence.Ann Pharmacother.2011;45:1053–1060.
11. Siegel D, Lopez J, Meier J.Antihypertensive medication adherence in the Department of Veterans Affairs.Am J Med.2007;120:26–32.
12. Shah NR, Hirsch AG, Zacker C, et al..Predictors of first-fill adherence for patients with hypertension.Am J Hypertens.2009;22:392–396.
13. Shah NR, Hirsch AG, Zacker C, et al..Factors associated with first-fill adherence rates for diabetic medications: a cohort study.J Gen Intern Med.2009;24:233–237.
14. Carroll NM, Ellis JL, Luckett C, et al..Feasibility of determining medication adherence from electronic health record medications orders.J Am Med Inform Assoc.2011;18:717–720.
15. Fischer MA, Stedman MR, Lii J, et al..Primary medication non-adherence: analysis of 195,930 electronic prescriptions.J Gen Intern Med.2010;25:284–290.
16. Karter AJ, Parker MM, Adams AS, et al..Primary non-adherence to prescribed medications.J Gen Intern Med.2010;25:763.
17. Raebel MA, Ellis JL, Carroll NM, et al..Characteristics of patients with primary nonadherence to medications for hypertension, diabetes, and lipid disorders.J Gen Intern Med.2012;27:57–64.
18. Andrade SE, Kahler KH, Frech F, et al..Methods for evaluation of medication adherence and persistence using automated databases.Pharmacoepidemiol Drug Saf.2006;15:565–574.
19. Caetano PA, Lain JMC, Morgan SG.Toward a standard definition and measurement of persistence with drug therapy: examples from research on statin and antihypertensive utilization.Clin Ther.2006;28:1411–1424.
20. Cramer JA, Roy A, Burrell A, et al..Medication compliance and persistence: terminology and definitions.Value Health.2008;11:44–47.
22. Phillipson TJ, Mozaffari E, Maclean JR.Pharmacy cost sharing, antiplatelet therapy utilization, and health outcomes for patients with acute coronary syndrome.Am J Manag Care.2010;16:290–297.
23. Schneeweiss S, Patrick AR, Maclure M, et al..Adherence to beta-blocker therapy under drug cost-sharing in patients with and without acute myocardial infarction.Am J Manag Care.2007;13:445–452.
24. Yu AP, Yu YF, Nichol MB, et al..Delay in filling the initial prescription for a statin: a potential early indicator of medication nonpersistence.Clin Ther.2008;30:761–774.
25. Vrijens B, De Geest S, Hughes DA, et al..A new taxonomy for describing and defining adherence to medications.Br J Clin Pharmacol.2012;73:691–705.
26. Briesacher BA, Andrade SE, Harrold LR, et al..Adherence and occurrence of fractures after switching to once-monthly oral bisphophonates.Pharmacoepidemiol Drug Saf.2010;19:1233–1240.
27. Huybrechts KF, Ishak KJ, Caro JJ.Assessment of compliance with osteoporosis treatment and its consequences in a managed care population.Bone.2006;38:922–928.
28. Deambrosis P, Saramin C, Terrazzani G, et al..Evaluation of the prescription and utilization patterns of statins in an Italian local health unit during the period 1994-2003.Eur J Clin Pharmacol.2007;63:197–203.
29. Hertz RP, Unger AN, Lustik MB.Adherence with pharmacotherapy for type 2 diabetes: a retrospective cohort study of adults with employer-sponsored health insurance.Clin Ther.2005;27:1064–1073.
30. Carnahan RM, Lund BC, Chrischilles EA, et al..Consistency of antidepressant and chronic nonpsychiatric medication use in a high-risk clinical population.Res Soc Admin Pharm.2008;4:367–374.
31. Christensen L, Sasane R, Hodgkins P, et al..Pharmacological treatment patterns among patients with attention-deficit/hyperactivity disorder: retrospective claims-based analysis of a managed care population.Curr Med Res Opin.2010;26:977–989.
32. Erkens JA, Panneman MMJ, Klungel OH, et al..Differences in antihypertensive drug persistence associated with drug class and gender: a PHARMO study.Pharmacoepidemiol Drug Saf.2005;14:795–803.
33. Gold DT, Martin BC, Frytak JR, et al..A claims database analysis of persistence with alendronate therapy and fracture risk in post-menopausal women with osteoporosis.CMRO.2007;23:585–594.
34. Yeaw J, Benner JS, Walt JG, et al..Comparing adherence and persistence across 6 chronic medication classes.J Manag Care Pharm.2009;15:728–740.
35. van Wijka BLG, Shrank WH, Klungela OH, et al..A cross-national study of the persistence of antihypertensive medication use in the elderly.J Hypertens.2008;26:145–153.
36. Barron TI, Connolly RM, Bennett K, et al..Early discontinuation of tamoxifen: a lesson for oncologists.Cancer.2007;109:832–839.
37. Boccuzzi SJ, Wogen J, Fox J, et al..Utilization of oral hypoglycemic agents in a drug-insured US population.Diabetes Care.2001;24:1411–1415.
38. El-Serag H, Fitzgerald S, Richardson P.The extent and determinants of prescribing and adherence with acid-reducing medications: a national claims database study.Am J Gastroenterol.2009;104:2167.
39. Kalsekar ID, Madhavan SS, Amonkar MM, et al..Impact of depression on utilization patterns of oral hypoglycemic agents in patients newly diagnosed with type 2 diabetes mellitus: a retrospective cohort analysis.Clin Ther.2006;28:306–318.
40. Hess LM, Raebel MA, Conner DA, et al..Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures.Ann Pharmacother.2006;40:1280–1288.
42. Karve S, Cleves MA, Helm M, et al..An empirical basis for standardizing adherence measures derived from administrative claims data among diabetic patients.Med Care.2008;46:1125–1133.
43. Bryson CL, Au DH, Young B, et al..A refill adherence algorithm for multiple short intervals to estimate refill compliance (ReComp).Med Care.2008;45:497–503.
44. Benner JS, Glynn RJ, Mogun H, et al..Long-term persistence in use of statin therapy in elderly patients.JAMA.2002;288:455–461.
46. Steiner JF, Prochazka AV.The assessment of refill compliance using pharmacy records: methods, validity, and applications.J Clin Epidemiol.1997;50:105–116.
47. Peterson AM, Nau DP, Cramer JA, et al..A checklist for medication compliance and persistence studies using retrospective databases.Value Health.2007;10:3–12.
48. Karve S, Cleves MA, Helm M, et al..Prospective validation of eight different adherence measures for use with administrative claims data among patients with schizophrenia.Value Health.2009;12:989–995.
49. Hansen RA, Kim MM, Song L, et al..Comparison of methods to assess medication adherence and classify nonadherence.Ann Pharmacother.2009;43:413–422.