Measuring Patient Activation as Part of Kidney Disease Policy: Are We There Yet?

Abstract

Optimal care occurs when patients possess the skills, knowledge, and confidence needed to effectively manage their health. Promoting such patient activation in kidney disease care is increasingly being prioritized, and patient activation has recently emerged as central to kidney disease legislative policy in the United States. Two options of the Centers for Medicare and Medicaid Services Kidney Care Choices model—the Kidney Care First option and the Comprehensive Kidney Care Contracting option—now include patient activation as a quality metric; both models specifically name the patient activation measure (PAM) as the patient-reported outcome to use when assessing activation in kidney disease. Because nephrology practices participating in these models will receive capitated payments according to changes in patients’ PAM scores, it is time to more critically evaluate this measure as it applies to patients with kidney disease. In this review, we raise important issues related to the PAM’s applicability to kidney health, review and summarize existing literature that applies this measure to patients with kidney disease, and outline key elements to consider when implementing the PAM into practice and policy. Our aim is to spur further dialogue regarding how to assess and address patient activation in kidney disease to facilitate best practices for supporting patients in the successful management of their kidney health.

Patient activation, which refers to having the knowledge, skills, and confidence needed to effectively manage one’s own health, is considered an important patient-reported outcome to measure as part of quality-of-care metrics in the United States.¹ The Chronic Care Model has highlighted the importance of clinicians interacting with informed and activated patients to optimize care delivery and facilitate the best possible health outcomes.² On April 7, 2016, the National Quality Forum (NQF) Quality Positioning System endorsed the patient activation measure (PAM) to assess activation in patients across a wide spectrum of care settings and chronic illnesses, including ESKD.^3,4 The NQF supports using the PAM to predict and track improvements in a broad range of patient-centered health outcomes, including access to care, care coordination, hospital readmissions, functional status, and medication overuse.

Interest in the PAM has been growing in nephrology. As part of an initiative to develop a framework related to patient-reported outcomes, patient-reported outcome measures, and patient-reported outcome performance measures for patients with ESKD, the Kidney Care Quality Alliance also highlighted the PAM.⁵ Recently, patient activation was included in two options of the Kidney Care Choices payment model of the Centers for Medicare and Medicaid Services (CMS): the Kidney Care First and the Comprehensive Kidney Care Contracting models.⁶ These models aim to determine whether certain incentives will improve both the cost and quality of care along the entirety of a patient’s kidney disease care continuum from nondialysis through dialysis, transplantation, and the end of life. Participating nephrologists and nephrology practices will receive capitated payments on the basis of key quality measures, one of which is the PAM.

The inclusion of the PAM as a quality metric in kidney disease care raises important unanswered considerations. A recent assessment of 60 kidney disease quality metrics conducted by the American Society of Nephrology Quality Committee did not comment on the PAM.⁷ Thus, it remains to be determined whether the PAM accurately captures the construct of patient activation in a kidney disease population, whether the measure predicts outcomes important to patients and other stakeholders, and if use of the PAM can lead to interventions that result in meaningful improvement in health outcomes in this patient group. To encourage dialogue in this timely area, we provide a brief introduction to and description of the PAM, a review of existing studies applying the PAM in kidney disease, and a summary of remaining knowledge gaps and challenges.

Measuring and Interpreting Patient Activation

The PAM, currently the most widely used instrument for measuring patient activation, is a commercial scale distributed by Insignia Health that aims to assess an individual’s knowledge, ability, skills, and confidence in self-managing chronic medical conditions.³ The PAM was developed using rigorous methods, which included defining a conceptual framework, generating and testing instrument items, validating the measure in populations with and without chronic disease, and assessing performance of the measure in a national probability sample.⁸ Scores on the original 22-item scale (22-item patient activation measure [PAM-22]) range from 0 to 100 and are divided into four levels of activation (with lower levels indicating low activation and higher levels indicating high activation): disengagement and disbelief about one’s own role in self-management (level 1, score≤47); increasing awareness, confidence, and knowledge in self-management tasks (level 2, score=47.1–55.1); readiness and taking action (level 3, score=55.2–67); and sustainment (level 4, score>67.1) (Figure 1). The instrument is scored along a Guttman scale, meaning that higher scores along a unidimensional continuum signify a more activated individual.

Figure 1.

Description of four levels of activation on the Patient Activation Measure (PAM).³

The PAM-22 has since been reduced to a 13-item version (13-item patient activation measure [PAM-13]) by its original developers (Supplemental File 1). The shortened scale has also demonstrated internal consistency (Cronbach α-coefficient, 0.87) as well as adequate reliability and validity.⁹ In the PAM-13, raw scores range from 38.6 to 53.0 and are standardized to a 0–100 scale. Per the original developers, the range of values that correspond to each activation level is similar to those of the PAM-22. The PAM-13 has been translated to multiple languages and adapted to crosscultural differences, and it has demonstrated adequate validity in such chronic conditions as diabetes, HIV disease, and chronic lumbar disease.^10–18 The original developers no longer license the 22-item version of the instrument for research or use in health care settings.

According to the NQF, achieving adequate patient activation should be on the basis of a change score calculated from a baseline score and followed by a subsequent measurement taken within 12 months (but not <6 months). Individuals who score at a level of four at baseline would be excluded from the calculation of the change score from a population metric perspective.⁴ An accountable nephrology unit would be one that reports two PAM scores per patient for at least 50% of their eligible patient population.

Evidence to Support Measuring Patient Activation in Kidney Disease

Patient activation as measured by the PAM-13 is linked to improved health outcomes in chronic conditions other than CKD. Lower PAM-13 scores are associated with an increase in hospitalizations in diabetes (odds ratio [OR], 1.7; 95% confidence interval [95% CI], 1.3 to 2.2). In a study of individuals living with HIV, every five-point increase in the PAM-13 was associated with a greater odds of having a CD4 count >200 cells per milliliter (OR, 1.10; 95% CI, 1.01 to 1.21) and improved medication adherence (OR, 1.18; 95% CI, 1.09 to 1.29).^19,20 Because few studies have administered the PAM-13 to patients with kidney disease, information regarding the determinants and outcomes associated with patient activation in this population is sparse.

We need additional evidence that higher levels of patient activation are associated with clinically meaningful outcomes in kidney disease. Table 1 describes the few published studies that have applied the PAM-13 to patients in this group, with information on study design, the kidney disease subpopulation tested, the prevalence of high and low activation levels in each study, and outcomes associated with these activation levels.^21–31 Most studies report cross-sectional associations between patient activation and clinical characteristics. Nearly all studies were completed outside the United States and focus on patients with nondialysis-dependent CKD. In studies that included this information, 20%–30% of individuals reported high activation levels (level 4) at baseline. In general, lower activation levels associated with older age, receiving in-center hemodialysis versus peritoneal dialysis or transplantation, poorer perceptions of health-related quality of life, higher decisional conflict (uncertainty about which course of action to take) with respect to dialysis modality choice, and lower medication adherence. No studies found associations between patient activation and treatment satisfaction, frequency of hospitalizations, or clinical biomarkers specific to kidney health. Only one study, a randomized, controlled trial of home-based primary care among Zuni Indians with CKD, targeted patient activation as a primary outcome.²⁷

Table 1.

Studies using the PAM-13 in kidney disease

Investigators, Country of Origin, and Study Design	Population and Sample Size (N)	PAM-13 Activation Level		Characteristics and Outcomes Associated with Patient Activation OR/β (95% CI)a or r (P Value)
Bos-Touwen et al.21 (Netherlands), cross-sectional	CKD (eGFR<60 ml/min per 1.73 m2; n=219)	Level 1 versus Levels 2–4		Characteristics: patients with CKD had the lowest activation levels compared with those with DM2, COPD, CHF
				Outcomes: multivariable linear regression (R2=0.2 if NS variables of social support and comorbidity score were included in model): BMI, 1.05 (1.01 to 1.08); living alone, 1.50 (1.10 to 2.06); some financial distress, 1.60 (1.17 to 2.18); some education, 1.41 (1.03 to 1.92); disease vintage >5 yr, 0.66 (0.45 to 0.96); depression, 1.05 (1.01 to 1.10); illness perception/understanding, 1.03 (1.01 to 1.04)
Hamilton et al.22 (United Kingdom), cross-sectional	HD/PD (n=173); KT (n=417)	Levels 1–4; median: level 3		Outcomes: multivariable linear regression (R2 not reported): dialysis use, −4.52 (−6.94 to −2.10); younger age of RRT, −3.09 (−5.89 to −3.00); medication adherence (P value for trend <0.01)
		Level 1: 26%
		Level 2: 18%		Level 2: 0.5 (−0.0 to 0.9)
		Level 3: 36%		Level 3: 0.7 (0.3 to 1.1)
		Level 4: 20%		Level 4: 0.6 (0.1 to 1.1)
Johnson et al.23 (United States), cross-sectional	Comorbid HTN, DM2, CKD (eGFR<60 ml/min per 1.73 m2; n=62); HD (n=19)	Levels 1–4; Level 1: 10%		Characteristics: patients with stage 5 CKD had lower activation levels compared with those at earlier CKD stages
		Level 2: 28%
		Level 3: 28%
		Level 4: 34%		Outcomes: no significant associations between patient activation and GFR decline
Lo et al.24 (Australia), cross-sectional	Comorbid DM2 and CKD (eGFR<60 ml/min per 1.73 m2; n=199)	Levels 1–4; Level 1: 20%		Characteristics: no significant associations between activation scores and glycemic control or BP control
		Level 2: 23%
		Level 3: 29%
		Level 3: 29%
Magnezi et al.25 (Israel), cross-sectional	Kidney disease (unknown type; n=25)			Characteristics: individuals aged 20–29 had lower activation levels compared with older adults
The Renal Association,26 National Health Service (England), longitudinal cohort	CKD (unknown eGFR; n=320); HD (n=921); PD (n=51); KT (n=617)	Levels 1–4; Level 1: 25%		Characteristics: those aged 25–44 and those who received KT had the highest activation levels. Patients on HD had lowest activation levels. Those who reported better quality of life had higher activation levels. Inverse relationship between neighborhood deprivation and patient activation level. Inverse relationship between symptom burden and patient activation level
		Level 2: 18%
		Level 3: 33%
		Level 4: 17%
				Outcomes: no association between patient activation level and calcium, phosphorus, or hemoglobin overall or by treatment modality. No association between clinician support and patient activation. Resurveying (without an intervention) resulted in improvements in patient activation among those who were previously at levels 1 and 2
Nelson et al.27 (United States), randomized controlled trial	CKD (mean eGFR =101–105 ml/min per 1.73 m2; n=125)	Levels 1–4; Level ≥3: 84% (usual care) versus 68% (home care); mean in both groups: level 3		PAM-13 as primary outcome: 8.7 points higher on activation score at 12 mo with receipt of home-based care (1.90 to 15.5)
Rivera et al.28 (United States), cross-sectional	CKD (unknown eGFR; n=67)	Levels 1 and 2 versus levels 3 and 4		Outcomes: no significant associations between activation scores and hospitalizations or emergency department visits
Van Bulck et al.29 (Belgium), cross-sectional	HD (n=192)	Levels 1–4; mean: level 2		Characteristics: univariable linear regression: age, −0.33; self-reported health, 0.33; nonuniversity higher education, 0.22; university education, 0.21; part-time work, 0.19; full-time work, 0.15; leisure activities, 0.33; having children, −0.22; living alone, 0.33; living with someone, 0.49; >1 care service at home, −0.29; receipt of KT, 0.16; treatment in hospital #2, −0.17. Multivariable linear regression (R2=0.31): age, −0.28; self-reported health, 0.28; leisure activities, 0.21; treatment in hospital #3, −0.02; living with someone, 0.14
Vélez-Bermúdez et al.30 (United States), cross-sectional	CKD (eGFR=7–25 ml/min per 1.73 m2; n=64)	Levels 1–4; Mean: level 3		Characteristics: patients who stated they would choose PD had highest activation scores. Pearson correlations: presence of heart disease, −0.28 (<0.05); decisional conflict, −0.47 (<0.01); CKD-related treatment satisfaction, −0.36 (<0.01)
				Outcomes: patient activation found to mediate relationship between treatment satisfaction and decisional conflict
Zimbudzi et al.31 (Australia), cross-sectional	Comorbid DM2 and CKD (eGFR<60 ml/min per 1.73 m2; n=305)	Levels 1 and 2 versus levels 3 and 4; levels 1-4 for multivariable linear regression; Level 1: 22%		Characteristics: univariable linear regression: self-care score, 0.21 (0.06 to 0.37); symptoms of kidney disease of KDQOL-36, 0.15 (0.05 to 0.25); burden of kidney disease of KDQOL-36, 0.11 (0.05 to 0.16); effects of kidney disease of KDQOL-36, 0.09 (0.02 to 0.17); PCS, 0.17 (0.01 to 0.33); MCS, 0.26 (0.09 to 0.42). Multivariable linear regression (R2 not reported): self-care score, 0.18 (0.02 to 0.35); burden of kidney disease of KDQOL-36, 0.11 (0.05 to 0.17)
		Level 2: 23.6%
		Level 3: 36.4%
		Level 4: 18%

DM2, type 2 diabetes; COPD, chronic obstructive pulmonary disease; CHF, chronic heart failure; NS, nonsignificant; BMI, body mass index; HD, in-center hemodialysis; PD, peritoneal dialysis; KT, kidney transplantation; HTN, hypertension; KDQOL-36, Kidney Disease–Related Quality of Life-36; PCS, physical composite summary; MCS, mental composite summary.

^aIf available. All numerical results are statistically significant.

We lack additional evidence that patient activation can be meaningfully improved in kidney disease, but ongoing studies may help answer this question. One upcoming randomized, controlled trial that includes patients with CKD will assess the effect of peer coaching to improve activation as a primary outcome.³² Two other studies, one using a clinical decision support tool to improve the quality of primary care for patients with CKD and another evaluating the effects of a home-based exercise program on cardiac biomarkers in kidney transplant, measure patient activation as a secondary outcome.^33,34 The results of these studies will expand our knowledge of the types of interventions likely to be most effective in improving patient activation in the kidney disease population.

Remaining Knowledge Gaps and Challenges in Measuring Patient Activation

Studies have found kidney disease–specific measures of constructs such as knowledge and self-efficacy that contribute to patient activation to be valid and reliable, but these measures have not been used with or compared directly with the PAM-13 in kidney disease populations.^35–37 Compared with measuring patient activation for other chronic illnesses, patient activation in the setting of kidney disease may require knowledge and self-management skills that are disease specific. Whether the more general PAM-13 will relate to important clinical outcomes in kidney disease with the same strength as it does in other conditions is unknown. Additionally, the original developers of the PAM use commercial license fees that are on the basis of the size of the target population of interest and operational requirements of the health care organization. This may pose a barrier to widespread implementation and sustained uptake of the PAM.³

Longitudinal studies of behavioral interventions aimed to improve the PAM-13 at a follow-up period of 6 months to 4 years.^38–41 This is critical because the length of follow-up selected for a study may have implications for how the success of interventions targeted to improve patient activation is interpreted. For example, a single-center, randomized, controlled trial assessing a home-based integrated disease management intervention’s effect on health-related quality of life, anxiety, depression, and patient activation among patients with chronic obstructive pulmonary disease detected greater improvements in activation at a 24-month follow-up compared with follow-up at 6 or 12 months.⁴⁰

Studies of behavioral interventions targeting patient activation also do not specify when patient activation must be measured after an intervention, but the Kidney Care First model requires assessing the PAM-13 at 12 months. Neither the length of time required to detect minimally significant or clinically important improvements in patient activation nor the time point of maximal effect have been established in CKD. Furthermore, the threshold of activation associated with the greatest improvement in health outcomes for a patient with kidney disease is also unknown. The NQF has suggested targeting an average increase in six points on the PAM-13 over a period of 6–12 months to achieve an “excellent” improvement in activation.⁴ In the only published randomized, controlled trial using the PAM-13 as a primary outcome measure in kidney disease, patients in the intervention arm exceeded this metric at 12 months. Mean (SD) activation scores in individuals who received home-based primary care increased from 61.1 (21.2) to 70.3 (21.6), with an average increase in activation score of 8.7 points (95% CI, 1.90 to 15.5).²⁷ Whether patients would have reported higher activation levels at an even longer follow-up duration is unknown.

In its recommendations regarding which patients should subsequently be administered the PAM-13,⁴ the NQF excludes those with high activation levels at baseline (level 4). Existing studies demonstrate a moderate prevalence of individuals with CKD who score at level 4 on the PAM-13 (17%–34%) (Table 1). In addition, patients with high baseline activation levels may still experience a decline in activation at a later date. One study from the Netherlands of >600 patients with various chronic conditions reported that 31% of patients had an activation level of four at baseline, which declined to 20% at follow-up.⁴² In another study, individuals who maintained the highest levels of activation over a 2-year follow-up period reported improved health behaviors and better mental health quality of life.⁴³ How to maintain or remediate decline in high patient activation among patients living with kidney disease is unknown. Furthermore, exclusion of patients with high activation scores at baseline from CKD payment models may introduce selection bias. Because systematic differences in sampling methods across facilities could affect payments to facilities, any analysis should consider this issue and include adjustment strategies. Inclusion of all eligible patients in descriptions of patient activation levels over time may be warranted. Rather than excluding individuals with high baseline activation levels, the CMS may consider stratifying individuals by change in score from baseline activation level.

Finally, it is also important to note that the NQF excludes administering the PAM-13 to individuals with mild cognitive impairment. Prevalence estimates for mild cognitive impairment in CKD range from 25% to 62%.^44,45 Neither the NQF nor the Kidney Care First model provide details on how cognitive impairment may impair survey response, and individuals with cognitive impairment may truly experience poor confidence and skills in disease-related self-management. Given the increasing age, comorbid disease burden, and accelerated cognitive decline faced by many patients with kidney disease, the rule against administering the PAM-13 to those with mild cognitive impairment may exclude those patients who may most benefit from interventions targeted to improve activation. Experts have outlined strategies to improve the cognitive accessibility of patient-reported outcomes measures that may be applied to PAM-13 administration: leveraging allied health professionals to assist with survey completion, administering the measure during earlier hours of the day, avoiding jargon or complex language prompts, and removing any requirements for extensive written responses.⁴⁶ Given that the PAM-13 already incorporates some of these suggestions, it may be suitable for patients with mild cognitive impairment.

Lessons from the United Kingdom’s Efforts to Measure Patient Activation in Kidney Disease

We may be able to learn from the United Kingdom’s National Health Service efforts to pilot test implementation of the PAM-13. In partnership with health care organizations and nephrology practices, the National Health Service conducted qualitative and quantitative studies involving interviews with 112 patients and providers (nephrologists and allied health professionals) to assess barriers and facilitators of PAM-13 use as well as test outcomes most strongly associated with the PAM-13.^26,47 According to providers, the PAM-13 aligned well with the United Kingdom’s Five Year Forward View, a care model aimed to improve population health, detection of chronic disease, and the overall cost effectiveness of health care.⁴⁸ Providers also felt that incorporating the PAM-13 into quality-of-care metrics would spur behavioral interventions such as health coaching and catalyze deeper conversations about self-efficacy and motivation. The study participants described difficulties with data management and overall flexibility of collecting PAM-13 data and concerns regarding data privacy and storage. Some providers said they felt the measure was overly complex, expressed concerns that patients who scored poorly would feel stigmatized, and indicated a preference for unstructured conversations with patients as opposed to using a specific measure. Patient views toward the PAM-13 varied, ranging from strong advocacy for the measure’s routine use to confusion regarding the meaning of the concept of patient activation. Patients with multiple chronic diseases were unsure which condition the measure was referring to when they considered their responses.

These qualitative results informed a subsequent quantitative study on the basis of the National Health Service Change Model, an implementation framework that defines shared goals, sets incentives for systems-wide culture change, and incorporates performance measures to guide the sustainment of new processes of care within health systems.⁴⁹ Between 2015 and 2017, the National Health Service enrolled 14 nephrology practices across England to pilot test the PAM-13.²⁶ According to the results, younger individuals and those who had received a kidney transplant reported higher patient activation levels. Although the study found no demonstrated associations between patient activation and clinical biomarkers linked to kidney health (Table 1), it revealed several findings related to the processes of care needed for the sustained uptake of the PAM-13. First, the United Kingdom Renal Registry infrastructure allowed for the routine collection and return of paper and online survey data to both nephrologists and patients. Second, nurses and allied health care professionals rather than physicians collected the data on patient-reported measures. Challenges included irregular CKD clinic attendance by patients and having a depleted workforce to administer the PAM-13.

On the basis of study results, investigators recommended incorporating patient preferences into the delivery structure of the PAM-13, training allied health professionals in measure collection, training physicians in core behavior change models, and embedding measures into clinical information technology systems (electronic patient-reported outcome measures) to encourage regular use and to allow for patients accessing results. The recommendation about embedding measures into clinical information technology systems is a key point given that the NQF requirement to assess baseline activation levels and measure a change score may require long-term data storage in the electronic medical record.

In addition to incorporating lessons learned from the United Kingdom to leverage allied health professionals, using electronic patient-reported outcome measures to store long-term information and minimize missing data, administering the measure to individuals with mild cognitive impairment, and allowing for a follow-up period of longer than 12 months for scoring, the CMS should consider calculating payments on the basis of models that include patients who report very high activation levels but also, consider adjusting for baseline activation levels and such demographics as socioeconomic status across facilities. Standardized methods for imputing missing values similar to those of the ESRD Quality Incentive Program should be considered.⁵⁰ Drawing on the lessons learned from the United Kingdom’s implementation efforts and implementing our additional recommendations will maximize the likelihood of successfully incorporating the PAM-13 in line with legislative policy in the United States. All of these strategies would align with kidney community’s mission to be a learning health system.

The PAM-13 has been used as a variable in predictive modeling to both characterize individuals who would most benefit from tailored interventions to improve activation and identify those at greatest risk for increased health care utilization.^39,51 Additionally, studies have demonstrated improvements in the PAM-13 in chronic illnesses, including for kidney disease in the setting of home-based primary care, for cancer survivorship in programs that incorporate psychosocial support, and for rheumatoid arthritis self-management with the use of mobile technology.^27,52,53 Including patient activation as a quality metric has the exciting potential to pinpoint individuals at greatest risk for poor health outcomes and to catalyze the development of novel patient-centered interventions to improve activation and disease self-management. However, to achieve this, additional rigorous analyses on a larger scale are needed to better understand the specific role and effect of patient activation on patients living with kidney disease.

Disclosures

All authors have nothing to disclose.

Funding

Dr. Cavanaugh is supported by National Institute of Diabetes and Digestive and Kidney Diseases grants P30DK114809 and R01DK103935 and National Institute on Aging grant R56AG061522-01A1.