When Does Nonadherence Indicate a Deviation From Patient-Centered Care?

Aisha T. Langford, PhD, MPH

Stella K. Kang, MD, MSc

R. Scott Braithwaite, MD, MS

The American Journal of Managed Care, May 2021, Volume 27, Issue 5

The authors suggest that assessment of patient-centered care may be improved by flagging probable discordance between a patient’s preferences and their treatment care plan.


Patient-centered care, defined as “providing care that is respectful of, and responsive to, individual patient preferences, needs and values, and ensuring that patient values guide all clinical decisions,” is advocated by clinicians and professional organizations and is part of a composite criterion for augmented reimbursement for various health care settings, including patient-centered medical homes. Despite general agreement that patient-centered care is a good idea and worthy of incentivization, patient-centered care is difficult to assess accurately, scalably, and feasibly. In this commentary, we suggest that assessment of patient-centered care may be improved by identifying circumstances that indicate its probable absence—in particular, by flagging probable discordance between a patient’s preferences and their treatment care plan. One potential marker of this discordance is persistent lack of control of a comorbid condition that is easily controllable by existing therapies and where existing therapies are sufficiently diverse to be compatible with a wide range of patient preferences (eg, stage 1 hypertension, type 2 diabetes with glycated hemoglobin < 8.5%). We outline how this approach may be tested, validated, and harmonized with existing quality improvement activities.

Am J Manag Care. 2021;27(5):e141-e144. https://doi.org/10.37765/ajmc.2021.88635


Takeaway Points

Preference-concordant care is a key component of patient-centered care. Conversely, if preference-concordant care is not occurring, that means that patient-centered care is not occurring.

  • Patient-centered care is advocated by clinicians and professional organizations.
  • Patient-centered care is difficult to assess accurately, scalably, and feasibly.
  • Preference concordance is relevant when screening or treatment decisions are “preference sensitive,” meaning more than 1 reasonable screening or treatment option exists.
  • It is possible that screening for deviations from patient-centered care may be realized through manual or automated monitoring of the electronic health record.


For decades, clinicians and professional organizations have endorsed the concept of patient-centered care.1,2 In 2001, the Institute of Medicine defined patient-centered care as “providing care that is respectful of, and responsive to, individual patient preferences, needs and values, and ensuring that patient values guide all clinical decisions.”3 In 2017, the American Institutes for Research issued a report titled “Principles for Making Health Care Measurement Patient-Centered.”4 In that report, 5 principles for patient-centered measurement were put forth, including the concept that measurement should be patient driven, which was defined as “patient goals, preferences, and priorities drive what is measured and how performance is assessed.”

Despite the benefits of patient-centered care, it is widely acknowledged that patient-centered care often does not happen in practice.5,6 A potential reason for its persistent absence is that patient-centered care can be difficult to measure.7,8 For example, quality-of-care measures that include patient preferences or valuations, such as the Patient-Reported Outcomes Measurement Information System (PROMIS) and the Consumer Assessment of Healthcare Providers and Systems Survey, fall short because the concept of patient-centered care requires that medical decisions be preference concordant.9,10 That is, final screening or treatment decisions should be consistent with patients’ values and preferences.8 PROMIS, Healthcare Effectiveness Data and Information Set (HEDIS), and other commonly employed quality measures do not measure preference concordance.11,12

Preference Concordance, Preference Sensitivity, and Patient-Centered Care

Preference concordance is relevant when screening or treatment decisions are “preference sensitive,” meaning that there is more than 1 reasonable screening or treatment option and each option has varying trade-offs and consequences that would be weighed differently by individual patients.13,14 For example, patients with stage 1 hypertension have many nonpharmacological (eg, losing weight, heart-healthy diet) and pharmacological (eg, diuretics, angiotensin-converting enzyme inhibitors) options to lower their blood pressure to reduce risk of early death.15 Given the multiple appropriate treatment options, the care plan could readily rest upon the patient’s preferences and goals for treatment. If the final treatment received matches the patient’s pretreatment preference, preference-concordant care has been realized. For example, if a patient with hypertension reports that they have no time or desire to make lifestyle changes and want “the easiest option” to lower blood pressure, a first-line antihypertensive medication would be preference concordant. Consequently, patient-centered care may not be occurring if a patient is persistently nonadherent with a recurring decision that is preference sensitive (Figure), in the setting in which there are untried alternatives with plausibly greater preference concordance. Continuing with the example, if that patient received lifestyle recommendations without the option of a medication, care would not be preference concordant and patient-centered care would not be occurring. Other examples of preference-sensitive decision scenarios include type 2 diabetes management, weight management, anticoagulant selection for atrial fibrillation, hyperlipidemia management, enlarged prostate management, and joint replacement for osteoarthritis.

Measuring Preference Concordance: Implications for Patient-Centered Care

Although measuring the continuum of preference concordance is difficult, its absence is not always difficult to detect. In particular, we argue that preference-concordant care may not be occurring, and therefore patient-centered care is not occurring, (1) if a patient is persistently nonadherent; (2) if a decision is recurrent, either because it involves a chronic condition or is routinely discussed in medical encounters for other reasons; and (3) where there are untried alternatives that are potentially more preference concordant, meaning that their consequences may be valued more favorably if preferences were plausibly but alternatively specified. Because alternatives with greater preference concordance occur only when decisions are preference sensitive, these criteria apply only to the subset of decisions that are preference sensitive.

Signs That Preference-Concordant Care Is Not Happening

Deviations from patient preferences could feasibly be surveilled because of existing quality measures that are routinely reported for HEDIS or PROMIS. For example, a practice may already seek to detect individuals with a stage 1 hypertension diagnosis who are not meeting particular goals for blood pressure control. Once a roster of recurrent and preference-sensitive decisions that are already important to a practice is established, surveillance for nonadherence to these decisions is possible. For example, an electronic health record (EHR) can be programmed to identify low medication possession ratios for prescribed blood pressure medications or persistently high blood pressure readings, then assess whether nonadherence is accompanied by untried alternatives that are potentially more preference concordant. Here, we would expect to see a series of revisions to the patient assessment and care plan. This process may shed light on the subset of patients who may benefit from more focused decision-making supports or other interventions such as question prompt lists and health coaches.16,17

Possible Future Directions

New opportunities for surveilling patient preferences have emerged.18 For example, it is possible that screening for deviations from patient-centered care may be realized through manual or automated monitoring of the EHR, such as natural language processing. Briefly, we describe what such a process could look like using a hypertension-specific protocol, which is a recurrent and preference-sensitive decision (Table). First, select a sample of patients with the International Classification of Diseases, Tenth Revision code for essential (primary) hypertension. Then, assess whether nonadherence is occurring, and if so, determine whether it is persistent. In addition to objective metrics such as medication possession ratios and blood pressure measurements, this may include surveillance for patient and clinician language that may serve as proxies for patient preferences. Examples of clinician language may include terms like noncompliant, nonadherent, and failed to reach goal. Examples of patient language may include terms like patient wants to try, patient refuses, and patient is concerned about, particularly if this language occurs in the context of a visit where the goals for managing a chronic condition are not being met.

Current Quality Improvement Processes Can Be Adapted to Include Surveillance for Preference-Discordant Care

Widely utilized quality-of-care measures such as accountable care organization (ACO) criteria can potentially be used to complement screening for preference-discordant care.19 For example, screening for and managing high blood pressure, blood glucose levels, and major depression are ACO criteria, as is colorectal cancer screening. As more health care organizations become ACOs and tracking of quality care measures becomes more standardized, more opportunities will arise to develop important quality measures that incorporate patient preferences into performance metrics. Moreover, HEDIS measurement infrastructure could be leveraged to screen for preference-discordant care. Several HEDIS measures have relevance for health issues that affect large segments of the US population, including heart disease and diabetes. Controlling high blood pressure is an example of a key HEDIS measure that is routinely reported.20,21

Potential Limitations

Admittedly, measuring preference-concordant care can be challenging; likely, this is why quality metrics emphasize other, more easily measurable aspects of care. However, choosing a convenience-based sampling frame of decisions that are common and recurrent in primary care offers an opportunity to mitigate many of the challenges. One challenge with implementing preference concordance measures into routine clinical care is that new tracking systems within the EHR may be needed. Historically, EHRs were designed around maximizing billing revenue rather than facilitating preference-sensitive care or measuring preference concordance. Despite these limitations, routine surveillance of preference concordance in the EHR may have implications beyond routine clinical care. For example, there may be implications for clinical trial participation because outreach can be done based on patients’ preferences for behavioral or medical interventions, which in turn may increase the likelihood that patients will respond to clinical trial opportunities and remain adherent throughout the trial.22 Further, although assessing preference concordance of care may be difficult, it is not impossible, as evidenced by the research on breast cancer and other treatment decisions,8,23,24 and is arguably as important as other quality improvement measures such as Plan-Do-Study-Act cycles, Six Sigma, and fishbone diagrams.


Preference-concordant care is a key component of patient-centered care. Conversely, if preference-concordant care is not occurring, that means that patient-centered care is not occurring. Although surveilling for preference-concordant care may be difficult, it is important, and using a sampling frame of recurrent common decisions could mitigate its complexity.

Author Affiliations: Department of Population Health (ATL, SKK, RSB) and Department of Radiology (SKK), NYU Grossman School of Medicine, New York, NY.

Source of Funding: None.

Author Disclosures: Dr Kang receives royalties from Wolters Kluwer for unrelated work. Drs Langford and Braithwaite report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (ATL, SKK, RSB); analysis and interpretation of data (ATL); drafting of the manuscript (ATL, SKK); critical revision of the manuscript for important intellectual content (ATL, SKK, RSB); administrative, technical, or logistic support (ATL); and supervision (RSB).

Address Correspondence to: Aisha T. Langford, PhD, MPH, Department of Population Health, NYU Grossman School of Medicine, 227 E 30th St, New York, NY 10016. Email: Aisha.Langford@nyulangone.org.


1. Elder CR, Debar LL, Ritenbaugh C, et al. Health care systems support to enhance patient-centered care: lessons from a primary care-based chronic pain management initiative. Perm J. 2017;21:16-101. doi:10.7812/TPP/16-101

2. What is patient-centered care? NEJM Catalyst. January 1, 2017. Accessed October 26, 2020. https://catalyst.nejm.org/doi/full/10.1056/CAT.17.0559

3. Institute of Medicine Committee on Quality of Health Care in America. Crossing the Quality Chasm: A New Health System for the 21st Century. National Academies Press (US); 2001.

4. Principles for making health care measurement patient-centered. American Institutes for Research. 2017. Accessed October 26, 2020. https://www.air.org/sites/default/files/Patient-Centered-Measurement-Principles-April-2017.pdf

5. Elwyn G, Frosch DL, Kobrin S. Implementing shared decision-making: consider all the consequences. Implement Sci. 2016;11:114. doi:10.1186/s13012-016-0480-9

6. Jaensch D, Baker N, Gordon S. Contemporaneous patient and health professional views of patient-centred care: a systematic review. Int J Qual Health Care. 2019;31(10):G165-G173. doi:10.1093/intqhc/mzz118

7. Robinson JH, Callister LC, Berry JA, Dearing KA. Patient-centered care and adherence: definitions and applications to improve outcomes. J Am Acad Nurse Pract. 2008;20(12):600-607. doi:10.1111/j.1745-7599.2008.00360.x

8. Winn K, Ozanne E, Sepucha K. Measuring patient-centered care: an updated systematic review of how studies define and report concordance between patients’ preferences and medical treatments. Patient Educ Couns. 2015;98(7):811-821. doi:10.1016/j.pec.2015.03.012

9. Patient satisfaction surveys. NEJM Catalyst. January 1, 2018. Accessed May 17, 2018. https://catalyst.nejm.org/doi/full/10.1056/CAT.18.0288

10. Mehta SJ. Patient satisfaction reporting and its implications for patient care. AMA J Ethics. 2015;17(7):616-621. doi:10.1001/journalofethics.2015.17.7.ecas3-1507

11. HEDIS measures and technical resources. National Committee for Quality Assurance. Accessed October 26, 2020. https://www.ncqa.org/hedis/measures/

12. Cella D, Choi SW, Condon DM, et al. PROMIS adult health profiles: efficient short-form measures of seven health domains. Value Health. 2019;22(5):537-544. doi:10.1016/j.jval.2019.02.004

13. Ostermann J, Brown DS, van Til JA, et al. Support tools for preference-sensitive decisions in healthcare: where are we? where do we go? how do we get there? Patient. 2019;12(5):439-443. doi:10.1007/s40271-019-00372-z

14. Stiggelbout AM, Pieterse AH, De Haes JCJM. Shared decision making: concepts, evidence, and practice. Patient Educ Couns. 2015;98(10):1172-1179. doi:10.1016/j.pec.2015.06.022

15. Sundström J, Arima H, Jackson R, et al; Blood Pressure Lowering Treatment Trialists’ Collaboration. Effects of blood pressure reduction in mild hypertension: a systematic review and meta-analysis. Ann Intern Med. 2015;162(3):184-191. doi:10.7326/M14-0773

16. Kambhampati S, Ashvetiya T, Stone NJ, Blumenthal RS, Martin SS. Shared decision-making and patient empowerment in preventive cardiology. Curr Cardiol Rep. 2016;18(5):49. doi:10.1007/s11886-016-0729-6

17. Stacey D, Hill S, McCaffery K, Boland L, Lewis KB, Horvat L. Shared decision making interventions: theoretical and empirical evidence with implications for health literacy. Stud Health Technol Inform. 2017;240:263-283.

18. Scott IA. Machine learning and evidence-based medicine. Ann Intern Med. 2018;169(1):44-46. doi:10.7326/M18-0115

19. Medicare Shared Savings Program quality measure benchmarks for the 2019 performance year. CMS. December 2018. Accessed October 7, 2019. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/Downloads/2019-quality-benchmarks-guidance.pdf

20. Patel MM, Datu B, Roman D, et al; CDC. Progress of health plans toward meeting the Million Hearts clinical target for high blood pressure control—United States, 2010-2012. MMWR Morb Mortal Wkly Rep. 2014;63(6):127-130.

21. Jaffe MG, Lee GA, Young JD, Sidney S, Go AS. Improved blood pressure control associated with a large-scale hypertension program. JAMA. 2013;310(7):699-705. doi:10.1001/jama.2013.108769

22. Pentz RD, Hendershot KA, Wall L, et al. Development and testing of a tool to assess patient preferences for phase I clinical trial participation. Psychooncology. 2015;24(7):835-838. doi:10.1002/pon.3731

23. Sepucha K, Ozanne EM. How to define and measure concordance between patients’ preferences and medical treatments: a systematic review of approaches and recommendations for standardization. Patient Educ Couns. 2010;78(1):12-23. doi:10.1016/j.pec.2009.05.011

24. Sepucha KR, Feibelmann S, Abdu WA, et al. Psychometric evaluation of a decision quality instrument for treatment of lumbar herniated disc. Spine. 2012;37(18):1609-1616. doi:10.1097/BRS.0b013e3182532924