Sincere Integration of Patients’ Perspectives into Kidney Care: Affirming and Adopting Patient Activation

In health care, patients must possess knowledge to understand their treatment options, confidence to communicate effectively with providers, and skills to apply this information to experience the best health outcomes. This combination of knowledge, confidence, and skills, collectively referred to as “patient activation,” is fundamental to the Chronic Illness Care Model (1). Activated patients access online health information more frequently, experience fewer emergency department visits, and engage more effectively in disease self-management behaviors. Measuring activation in nephrology can identify patients who receive the greatest benefit from a skill-building intervention, monitoring activation can pinpoint those at greatest risk for decreased behavioral engagement, and targeting activation can improve the efficacy of behavioral intervention trials (2).

Quantifying complex psychological constructs to be actionable, including patient activation, requires patient-reported outcome measures. Ensuring that patient-reported outcome measures accurately and parsimoniously capture the underlying construct they intend to measure requires tests of validity, reliability, and responsiveness (3). The Patient Activation Measure-13 (PAM-13), a 13-item scale developed by Hibbard and colleagues (1), is a CKD quality metric in the United States and the United Kingdom. We recently exposed a dearth of studies applying PAM-13 in kidney disease; emphasized the need to identify outcomes affected most by patient activation in this population; and demonstrated that the validity, reliability, and responsiveness of PAM-13 in CKD had not yet been established (4).

In this issue of CJASN, Lightfoot et al. (5) report on PAM-13’s validity and reliability in patients with nondialysis CKD or a kidney transplant. In a multicenter study between 2018 and 2020, 942 adults (762 with CKD and 163 with a transplant) self-administered PAM-13 (5). Participants were 66 years old on average, and many had multiple comorbid conditions. Participants’ mean PAM-13 scores were high at 55.1 (level 3; “Ready for action”), although only 14% reported the highest level of activation (level 4; “Maintaining behaviors”). Activation levels did not differ across CKD stage or transplant status.

Reliability of scale items is measured using (1) Cronbach α values to judge consistency and ensure items capture the same latent (or unobserved) construct and (2) interitem correlations to assess the extent to which items relate to each other (3). A Cronbach α of ≥0.8 and interitem correlations of 0.15–0.50 indicate acceptable internal consistency (3). In the analysis of Lightfoot et al. (5), PAM-13 demonstrated excellent internal consistency (Cronbach α =0.925; average interitem correlations =0.502). Items 5, 6, 8, and 9 (related to knowledge of medications and treatments as well as confidence in information seeking and communication ability) had high interitem correlations of ≥0.5, suggesting potential overlap and less new information contributed by each item.

Construct validity, or the degree to which a scale measures what it claims to measure, is assessed using many methods, including factor analyses, structural equation modeling, item-response theory, and Rasch models. Rasch models analyze survey responses as a function of a respondent’s abilities (e.g. overall activation) and the difficulty of each question item (6). The analysis of Lightfoot et al. (5) supports that activation was valid as a singular construct, although PAM-13 items 7 and 13 (related to confidence in self-managing at home and maintaining healthy lifestyle behaviors) had high “outfit statistics” with values of ~1.3. These results suggest that patients with “low” activation actually endorse items that are expected to require “higher” levels of activation to endorse (and vice versa). Investigators hypothesized that participants may have had trouble understanding how these questions applied to living with kidney disease. These participants who were derived from an observational study examining lifestyle behaviors (at home) may also have been primed differently than a general population. The investigators note that although thresholds for “fit statistics” are debated, all PAM-13 items were within recommended limits.

Floor and ceiling effects describe how study participants score at or near the lower and upper limits on a scale (3). Although PAM-13 items demonstrated no significant floor effects, over 15% of participants chose “Agree Strongly” in response to items 1–8. This threatens statistical comparisons due to a potentially skewed distribution of the data and to evaluating change in activation scores in response to an intervention if a participant already reports high activation. Yet, if this score is the actual representation of a patient’s self-assessed activation and this would not change with more information, practice of skills, or retest, then a performance asymptote is reached and is valid. Targeting this group for behavioral intervention support may not be necessary. Current evidence supports that the poorest health outcomes occur in those who report PAM-13 levels 1–3 (2).

The analysis of Lightfoot et al. (5) contains several strengths. The investigators provide new evidence of the reliability and validity of PAM-13 in kidney disease. In addition to providing information related to the measure’s unidimensionality, potential inclusion of scale items that measure similar constructs, and existence of ceiling effects, the investigators also used differential item functioning to test if the PAM-13 may be measuring different abilities among members of certain subgroups (3). The investigators did not observe differential item functioning across CKD stage or sex. Item 3 (confidence in ability to minimize symptoms or problems) and item 12 (confidence in finding solutions to new health problems) showed differential item functioning in patients with a kidney transplant and patients who were older than 69, respectively. As cognitive or qualitative interviews were not included as part of this analysis, robust conclusions related to differences in these subgroups’ interpretation of or response to these items remain unknown.

This study involved mostly non-Hispanic White participants (96%), with 40% completing “tertiary education” or education involving colleges, universities, or trade schools. Patient activation is differentially affected by self-identified race, educational attainment, and health literacy (7). Determining whether PAM-13 measures activation across populations disproportionately burdened by kidney disease will require extending analyses to participants belonging to diverse racial and ethnic backgrounds and with limited social and economic resources. Importantly, patients receiving dialysis for ESKD were not included, leaving a significant unaddressed gap in PAM-13’s psychometric evidence.

There is an ongoing debate of whether disease-specific measures of psychologic constructs have greater validity than those that are generic and nondisease specific. As an example, evidence supports that the frequency of “diabetes distress” is distinct from and more significantly associated with adverse health outcomes compared with measures of general depressive symptom frequency (8). It remains unknown whether the knowledge, confidence, and skills needed to self-manage CKD or a kidney transplant are fundamentally different from what is needed to successfully live with another chronic condition. As kidney disease most frequently occurs along with comorbid diseases, it may not be necessary to isolate participants’ activation levels to the experience of kidney disease. Further, many generic measures included in the patient-reported outcomes measurement information system have shown adequate validity across multiple chronic conditions (9).

As a unidimensional measure of three key drivers of behavior change (knowledge, confidence, and skill), patient activation is an emerging influential construct. Future investigations building upon the foundational work by Lightfoot et al. (5) must determine whether patient activation as measured by PAM-13 is a clinically meaningful predictor of outcomes in CKD and thus, a primary trial end point. Just as nephrologists were called to lead the implementation of patient experience measures into effective care, so must we integrate patient activation into learning health systems for individual and population health (Figure 1) (10). Incorporating PAM-13 in kidney disease research through learning health systems has high potential to advance the precision and reproducibility of behavioral intervention trials to actualize patient-centered CKD care delivery.

Figure 1.

Applications of patient activation measurement in kidney disease. CDSS, clinical decision support system; CMS, Centers for Medicare & Medicaid Services; EMR, electronic medical record; PAM-13, Patient Activation Measure-13; PROM, patient-reported outcome measure.

Disclosures

K.L. Cavanaugh reports consultancy agreements with Kidney Health Initiative, REATA Pharmaceuticals, and Responsum Health; ownership interest in HCA Healthcare; and serving as a scientific advisor or member of American Journal of Kidney Diseases editorial board, CJASN editorial board, Medical Decision Making editorial board, and the National Kidney Foundation–Kidney Disease Outcomes Quality Initiative Education Committee. The remaining author has nothing to disclose.

Funding

The work was supported by Agency for Healthcare Research and Quality and Patient Centered Outcome Institute grant K12HS026395 (to K.L. Cavanaugh and D. Nair).