Impact of Medical Care Coordination Intervention on Patient Activation

ABSTRACT

Objective: To evaluate the impact of an adult medical care coordination (AMCC) intervention on patient activation.

Study Design: This observational evaluation compared AMCC with usual care (UC). Eligible patients were adults discharged home who had 2 or more chronic conditions and a high risk of readmission. AMCC involved registered nurse care coordinators providing self-management support to patients via 1 home visit and regular phone calls. The outcome was the 10-item Patient Activation Measure (PAM), a validated patient-reported outcome tool with 4 levels ranging from 1 (lower activation) to 4 (higher activation). Measurement occurred at baseline and 30, 90, and 180 days.

Methods: We evaluated patient activation as an ordinal outcome using an ordered logistic regression model, a dichotomous outcome using a linear probability model, and a continuous outcome using ordinary least squares.

Results: We identified 915 (432 AMCC, 483 UC) patients who completed both the baseline and at least 1 follow-up PAM. For the ordinal analysis, AMCC was associated with a significant increase in the percentage with a PAM of level 3 at 30, 90, and 180 days and a decrease in the percentage with a PAM of level 1 or 2 at 180 days. For the dichotomous analysis, AMCC was associated with a significant increase in the percentage of patients with a PAM of level 3 or 4 at 180 days (15.2 percentage points; 95% CI, 5.6-24.7).

Conclusions: AMCC significantly increased patient activation, particularly at the final measurement. These findings highlight the potential value of AMCC as a self-management intervention, enhancing patients’ confidence to manage their health.

Am J Manag Care. 2025;31(6):In Press

Takeaway Points

We analyzed the impact of an adult medical care coordination model implemented at Mayo Clinic on patient activation as measured by the Patient Activation Measure and found that:

• The intervention was associated with significant improvement in patient activation compared with usual care.
• The benefit was concentrated at the final measurement point of 180 days, which suggests that some patients may need more time to benefit from the intervention.
• The results suggest that adult medical care coordination has value as a self-management support intervention to increase patients’ confidence to self-manage their health.

Patient activation refers to the degree to which individuals have the knowledge, confidence, and skills necessary to manage their health effectively. This construct can be measured using the Patient Activation Measure (PAM), which is a validated patient-reported outcome.1,2 This measure has been validated specifically among older adults with multiple chronic conditions.3 At lower levels of the PAM, patients perceive themselves as having less knowledge, confidence, and skills needed to actively manage their health.1,2

Higher levels of activation as measured by the PAM have been associated with better patient outcomes. Specifically, higher levels of PAM are linked to lower rates of hospitalization,4-6 lower rates of emergency department (ED) visits,6,7 higher rates of clinical indicators such as hemoglobin A1c in the normal range,5 higher quality of life,8 and higher rates of healthy behaviors such as refraining from smoking.5,7 Further, increases in the PAM measured over time have been associated with improved outcomes and lower costs,5,9,10 which suggests that interventions to improve patient activation may translate to better patient outcomes. However, interventions to increase patient activation have had mixed success. Many have increased PAM levels,11-15 and several systematic reviews have found that such interventions are generally successful.16-18 In contrast, some interventions were unsuccessful or had mixed results.19-24

Given the success of many of the prior interventions11-18 and the link between patient activation and improved outcomes,4-10 Mayo Clinic implemented an adult medical care coordination (AMCC) intervention. The goal of the AMCC intervention was to reduce 30-day hospital readmissions by increasing patient self-management. The AMCC intervention is based on the Coleman Care Transitions Intervention (CTI) model.25-27 CTI was motivated by the finding that individuals with medical complexity often need care in multiple settings and that the transitions between those settings present challenges for patient care.28 The program focuses on 4 pillars: medication self-management, a patient-centered record (ie, a document containing essential medical information for use across different settings), primary care and specialist follow-up, and knowledge of red flags (ie, warning symptoms or signs indicative of a worsening condition).27 CTI addresses the pillars by providing patients with hospital admissions for a chronic condition with a personal health record and care transition coaches who work to increase activation and coordinate care.25-27

AMCC is a self-management support intervention that is an adaptation of the CTI model with a focus on the care transition coach. Registered nurse (RN) care coordinators provided self-management support to patients via 1 home visit within 7 days after hospital discharge and regular phone calls up to 6 months post discharge. There were notable modifications between CTI and AMCC, including extending the intervention from 30 days to up to 180 days to support patients who may need a longer duration.

Our objective was to evaluate whether AMCC was associated with changes in the PAM at 30, 90, and 180 days relative to baseline. We compared AMCC patients with usual care (UC) patients who did not receive care coordination but otherwise met the criteria for AMCC. Given the mixed results from prior evaluations of interventions to increase patient activation, it is important to understand whether and to what extent AMCC can increase patient activation relative to UC. The impact of AMCC on patient activation was a prespecified secondary outcome thought to be critical to reducing 30-day readmissions. The results for utilization outcomes including 30-day readmissions will be presented in a separate manuscript.

METHODS

Design

The analysis included patients receiving care at 1 of 27 primary care clinics at Mayo Clinic in Rochester, Minnesota, or at Mayo Clinic Health System in southeast Minnesota and northwest Wisconsin. The study was a pragmatic29 stepped-wedge cluster randomized30 clinical trial (NCT04224220). However, we conducted the analysis using an observational design due to large differences in baseline characteristics between the AMCC and UC arms (Table).31 The study began in January 2020, and the last new participant was enrolled in June 2022, with outcomes assessed through 2022. Because of the stepped-wedge design, there were 4 phases for the rollout of AMCC: phase 0, no clinics offering AMCC; phase 1, first set of clinics began offering AMCC, with this phase being split between before and after a pause due to the COVID-19 pandemic; phase 2, second set of clinics began offering AMCC; and phase 3, all clinics offering AMCC. The PAM was collected for this study from both AMCC and UC sites starting in phase 1. We employed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist to inform the reporting of this study.32 This analysis was approved by the Mayo Clinic Institutional Review Board. The data sets generated and/or analyzed during the current study are not publicly available due to the sensitive nature of the medical records data we used but are available on reasonable request.

Participants

Patients in the AMCC group were required to be discharged home or to assisted living following hospitalization, be a Mayo Clinic primary care patient aged 18 years and older, and have a LACE+ score of at least 59. The LACE+ index predicts readmission risk using key components including length of stay, acuity of admission, comorbidity, and ED utilization.33 Additionally, patients had to have 2 or more chronic conditions such as coronary artery disease, congestive heart failure, atrial fibrillation, diabetes, chronic kidney disease, chronic obstructive pulmonary disease, and morbid obesity. Additional criteria are listed in eAppendix Table 1 (eAppendix available at ajmc.com). Similar criteria were used for UC patients using information from the electronic health record. AMCC patients did not have to provide consent because AMCC and the data collection were considered part of patients’ clinical care. However, UC patients needed to provide consent for the data collection, as it was considered research related. The study initially included 923 AMCC and 1004 UC participants. After excluding those without a complete baseline PAM and at least 1 follow-up PAM, the analysis included 432 AMCC and 483 UC participants (Figure 1).

Main Measures

The primary outcome was the change in PAM scores from baseline to follow-up at 30, 90, and 180 days. The PAM ranges from 0 to 100, with higher scores indicating greater self-activation. The raw score is used to categorize patients into 1 of 4 stages of activation, with level 1 being the lowest level and level 4 being the highest level of activation.1,2 We used the 10-item version of the PAM34 that is available from Insignia Health. We operationalized analysis of the PAM scores 3 ways. First, we assessed PAM as an ordinal outcome with 4 levels. Second, we assessed PAM as a dichotomous outcome with one group including levels 1 and 2 and the other group including levels 3 and 4. Third, we assessed PAM as a continuous outcome from 0 to 100. CMS uses the PAM as a continuous measure for the Merit-Based Incentive Payment System (MIPS) and sets thresholds for improvement at 3 and 6 points.35 We excluded observations with PAM scores of 51 (“agree” for all responses) or 100 (“agree strongly” for all responses) because these are considered outliers.34

Intervention and Other Variables

The key exposure was participation in AMCC or UC. We also included other variables that may act as confounders. The first set was demographics including age, gender, and whether the patient lived in a rural area as defined by rural-urban commuting area codes.36 The second set included measures of health status including the LACE+ index and the Charlson Comorbidity Index.37 The third set was indicators for the Mayo Clinic region, as care delivery patterns may vary across sites. The last set incorporated measures that may affect patient engagement, including whether the patient had an active patient portal account and health literacy.38,39

Statistical Approach

For the ordinal outcome, we used ordered logistic regression with random effects. For this model, we included the control variables noted earlier. In addition, we also included interactions for time of measurement (baseline, 30 days, 90 days, or 180 days) with age, the LACE+ index, the Charlson Comorbidity Index, whether the patient had an active portal, and health literacy. The reason for doing so was to control for how these variables may affect how the PAM changes over time. For the dichotomous outcome, we used a linear probability model with fixed effects. A linear probability model is an ordinary least squares regression model that uses a dichotomous outcome. In these models, we did not directly control for the variables mentioned earlier because they were all time invariant and therefore accounted for as part of the fixed effects. We were able to include the same interaction terms because the interaction between time and other variables changes with respect to time. For the continuous outcome, we used an ordinary least squares model with fixed effects.

Sensitivity Analysis

We conducted a sensitivity analysis with the outliers included in the models to assess whether their inclusion impacted results.

RESULTS

Sample Characteristics

The descriptive statistics show meaningful differences between the AMCC and UC arms at baseline (Table). AMCC patients were more likely to have inadequate health literacy (20.1% vs 8.5%), were older (mean age, 75.9 vs 66.5 years), had higher LACE+ scores (mean, 73.6 vs 71.0), and had higher Charlson Comorbidity Index scores (mean, 1.39 vs 1.11).

The patients who had missing data were similar to those who did not have missing data (eAppendix Table 2). Many of the patients without complete PAM data (418 of 1012) were in phase 0 before the PAM was collected. There were differences by age, with patients without a baseline PAM being older (mean age, 74.4 years) than patients with a baseline but no follow-up (68.9 years), patients with at least 1 follow-up (69.6 years), and patients with all follow-up PAMs completed (71.9 years).

PAM Descriptive Distribution

The distribution of PAM at baseline also differed considerably between the AMCC and UC arms (Figure 2 and eAppendix Table 3). At baseline, AMCC patients were more likely than UC patients to be at PAM level 1 (11% vs 4%) or level 2 (32% vs 15%). At 180 days, the percentage of patients at levels 1 (5% vs 1%) and 2 (19% vs 12%) decreased. There were also significant improvements in both arms for the continuous PAM (eAppendix Table 4). The mean (SD) length of stay in AMCC was 120.4 (59.8) days, with phone contact between RN care coordinators and patients on 10.4% of days (eAppendix Table 5). Patients with an initial PAM level 1 who had greater improvement at follow-up tended to have a higher percentage of days with phone contacts, but this pattern was not consistent across different baseline PAM levels.

Regression Results

The analysis of PAM as an ordinal outcome showed significant differences by arm (Figure 3 and eAppendix Table 6). The increase in the percentage of patients at level 3 was greater for AMCC than UC at 30 days (3.6 percentage points; 95% CI, 1.6-5.7), 90 days (3.0 percentage points; 95% CI, 0.9-5.0), and 180 days (3.5 percentage points; 95% CI, 1.0-6.1). The decrease in the percentage of patients who were at level 1 (–3.4 percentage points; 95% CI, –5.4 to –1.4) and level 2 (–5.7 percentage points; 95% CI, –10.3 to –1.2) was also greater for those in the AMCC arm at 180 days. The results from the sensitivity analysis were similar in direction and significance.

The analysis of PAM as a dichotomous outcome also showed a statistically significant difference (Figure 4 and eAppendix Table 7). However, unlike with the ordinal analysis, the difference was only significant at 180 days. The increase in the percentage of patients at level 3 or 4 was greater for AMCC than UC at 180 days (15.2 percentage points; 95% CI, 5.6-24.7). In contrast, the results from the sensitivity analysis showed no significant difference at any point.

Similar to the other analyses, the analysis of PAM as a continuous outcome found a significant difference at 180 days (eAppendix Table 8). The increase in the PAM was 2.3 points (95% CI, 0.1-4.4) higher for AMCC than UC. The sensitivity analysis also showed that the difference was not statistically significant when including the outliers.

DISCUSSION

We found that a self-management support intervention for recently discharged patients increased patient activation compared with UC. Specifically, AMCC was associated with decreases in the percentage of patients at PAM levels 1 and 2 at 180 days and increases in the percentage of patients at PAM level 3 at 30, 90, and 180 days. Using a dichotomous specification, AMCC was associated with an increase in the percentage of patients with a PAM of level 3 or 4 at 180 days only. Finally, AMCC was associated with an increase in the PAM on the continuous scale at 180 days only. These findings were sensitive to the inclusion of outliers, with the results for the dichotomous and continuous specifications no longer statistically significant when outliers were included. Taken together, these findings suggest with moderate confidence that AMCC improved patient activation at 180 days relative to UC.

Although the findings suggest improvements in patient activation at 180 days, the mixed findings from the different specifications and the sensitivity analyses suggest there may be nuances to the findings. First, the differences in the results between the ordinal and dichotomous specifications may be related to the different modeling approaches. The ordinal analysis found statistically significant increases for level 3 at 30, 90, and 180 days, whereas for the dichotomous and continuous analyses, there was only a significant improvement at 180 days. Additionally, the magnitude of the impact on the percentage of patients with a PAM of level 3 or 4 was actually larger for the dichotomous analysis at 180 days (15.2 percentage points) than for the combined impact on the percentage of PAM level 3 or level 4 in the ordinal analysis. One possible explanation is that the model for the ordinal specification was a random effects model whereas the model for the dichotomous outcome was a fixed effects model. Fixed effects models are considered stronger study designs because they account for all time-invariant factors including unobserved factors. As such, if the findings diverge, then generally the fixed effects models are considered more likely to be accurate. However, fixed effects models produce point estimates with larger CIs, which makes the results less likely to be statistically significant. This may explain why the point estimate for the sensitivity analysis for the dichotomous specification including outliers was not statistically significant even though the magnitude of the point estimate was moderate in size (7.9 percentage points).

Altogether, the findings provide moderate support for the success of the intervention in improving patient activation. The results suggest that there is not strong evidence for an impact at 30 and 90 days because the models diverge and the fixed-effects models did not find a significant difference. In contrast, there is moderate evidence for an effect at 180 days because all 3 models found a significant difference in the main specifications. The fact that the improvement seemed concentrated at the final measurement and that the mean length of stay was approximately 120 days may suggest that some patients in the AMCC arm benefited more from longer exposure to the program. However, the lack of statistical significance in 2 of the sensitivity analyses suggests that caution is needed when interpreting this finding. Further, the magnitude of change on the continuous scale was small (2.3 points) relative to the lower MIPS threshold of 3 points.35 Overall, these mixed findings support the continued use of AMCC as a self-management support intervention for recently discharged patients.

The findings are largely consistent with what has been observed in previous evaluations of interventions to increase patient activation. A review of such interventions found that improvements ranged from 2.5 to 6.5 points on the PAM in the group receiving the intervention,18 which is consistent with the improvement we observed in the AMCC group over time. Additionally, the review found that many successful interventions included skill development and tailoring support to the individual’s activation level, both of which were components of AMCC.18 The results for individual studies are more mixed. These studies cover an array of interventions, including a care management intervention for medically complex patients,21 a self-management intervention,14,23,24 2 population-level health campaigns combined with individual coaching for high-risk individuals,19 tailored coaching,12,13,15,22 and the introduction of a patient portal.11,13 Although these interventions consistently found improvements in PAM scores within the intervention arm, only some of them reported significant improvement relative to the control arm,11-15 whereas others reported no significant differences or mixed results.19-24 Similar to the studies that found a significant improvement compared with UC, we observed significant improvement at 180 days. However, similar to the studies that did not observe significant improvements, we did not find differences when including outliers. Taken together, our mixed results fit between the findings observed in prior work.

Limitations

The strength of this analysis was the collection of a validated patient-reported outcome to measure patient activation for many patients. There are also important limitations. First, the PAM was missing at baseline or at follow-up for a large proportion of patients. However, we found that the characteristics for individuals with or without a baseline and at least 1 follow-up PAM were similar except for age and portal status. Second, patients in the AMCC group were not required to consent, whereas patients in the UC group were required to consent. This difference in consent requirements contributed to large differences in the populations at baseline in measurable characteristics including age, medical comorbidities, and baseline PAM scores. Because baseline PAM scores were lower among individuals in the AMCC group on average, there may have been room for more improvement. Additionally, it is possible some of the results were affected by these baseline differences. We controlled for the differences in age and comorbidity burden. Further, the fixed effects models controlled for other unobservable characteristics that remain constant over time and focused on the differences within individuals. Third, we were unable to assess within-group differences by level of interaction with AMCC because more interaction may be due to greater need rather than more engagement.

CONCLUSIONS

There was moderate evidence that AMCC was associated with a significant increase in patient activation at 180 days. The findings suggest that AMCC has value as a self-management support intervention to increase patients’ confidence to self-manage their health.

Acknowledgments

The authors would like to acknowledge the contributions of Ruchita Dholakia, Karen Schaepe, Yvonne Larson Smith, and Danielle Loudermilk to this manuscript.

Author Affiliations: Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery (STS, MAL, SAI, BJB), Division of Primary Care Internal Medicine (VLH, RJS), and Department of Nursing (ABM, PJM, SGW), Mayo Clinic, Rochester, MN.

Source of Funding: This work was supported by the Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery at Mayo Clinic, Rochester, MN.

Prior Presentation: Poster presentation at the 2024 Health Care Systems Research Network Conference (April 9-11, 2024) and at the 2024 AcademyHealth Annual Research Meeting (June 29-July 2, 2024).

Author Disclosures: The authors 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 (STS, MAL, VLH, ABM, RJS, PJM, SGW, BJB); acquisition of data (SAI, VLH); analysis and interpretation of data (STS, SAI, RJS, PJM, BJB); drafting of the manuscript (STS, ABM); critical revision of the manuscript for important intellectual content (STS, MAL, ABM, RJS, PJM, SGW, BJB); statistical analysis (STS, SAI); provision of patients or study materials (VLH); administrative, technical, or logistic support (MAL, SAI, RJS, SGW); and supervision (MAL, VLH, ABM, SGW, BJB).

Address Correspondence to: Samuel T. Savitz, PhD, Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery, Mayo Clinic, 200 1st St SW, Harwick 2nd Floor, Rochester, MN 55905. Email: savitz.samuel@mayo.edu.

REFERENCES

1. Hibbard JH, Stockard J, Mahoney ER, Tusler M. Development of the Patient Activation Measure (PAM): conceptualizing and measuring activation in patients and consumers. Health Serv Res. 2004;39(4, pt 1):1005-1026. doi:10.1111/j.1475-6773.2004.00269.x

2. Hibbard JH, Mahoney ER, Stockard J, Tusler M. Development and testing of a short form of the Patient Activation Measure. Health Serv Res. 2005;40(6, pt 1):1918-1930. doi:10.1111/j.1475-6773.2005.00438.x

3. Skolasky RL, Green AF, Scharfstein D, Boult C, Reider L, Wegener ST. Psychometric properties of the patient activation measure among multimorbid older adults. Health Serv Res. 2011;46(2):457-478. doi:10.1111/j.1475-6773.2010.01210.x

4. Mitchell SE, Gardiner PM, Sadikova E, et al. Patient activation and 30-day post-discharge hospital utilization. J Gen Intern Med. 2014;29(2):349-355. doi:10.1007/s11606-013-2647-2

5. Greene J, Hibbard JH, Sacks R, Overton V, Parrotta CD. When patient activation levels change, health outcomes and costs change, too. Health Aff (Millwood). 2015;34(3):431-437. doi:10.1377/hlthaff.2014.0452

6. Kinney RL, Lemon SC, Person SD, Pagoto SL, Saczynski JS. The association between patient activation and medication adherence, hospitalization, and emergency room utilization in patients with chronic illnesses: a systematic review. Patient Educ Couns. 2015;98(5):545-552. doi:10.1016/j.pec.2015.02.005

7. Greene J, Hibbard JH. Why does patient activation matter? an examination of the relationships between patient activation and health-related outcomes. J Gen Intern Med. 2012;27(5):520-526. doi:10.1007/s11606-011-1931-2

8. Mosen DM, Schmittdiel J, Hibbard J, Sobel D, Remmers C, Bellows J. Is patient activation associated with outcomes of care for adults with chronic conditions? J Ambul Care Manage. 2007;30(1):21-29. doi:10.1097/00004479-200701000-00005

9. Harvey L, Fowles JB, Xi M, Terry P. When activation changes, what else changes? the relationship between change in patient activation measure (PAM) and employees’ health status and health behaviors. Patient Educ Couns. 2012;88(2):338-343. doi:10.1016/j.pec.2012.02.005

10. Hibbard JH, Greene J, Shi Y, Mittler J, Scanlon D. Taking the long view: how well do patient activation scores predict outcomes four years later? Med Care Res Rev. 2015;72(3):324-337. doi:10.1177/1077558715573871

11. Solomon M, Wagner SL, Goes J. Effects of a web-based intervention for adults with chronic conditions on patient activation: online randomized controlled trial. J Med Internet Res. 2012;14(1):e32. doi:10.2196/jmir.1924

12. Hibbard JH, Greene J, Tusler M. Improving the outcomes of disease management by tailoring care to the patient’s level of activation. Am J Manag Care. 2009;15(6):353-360.

13. Shively MJ, Gardetto NJ, Kodiath MF, et al. Effect of patient activation on self-management in patients with heart failure. J Cardiovasc Nurs. 2013;28(1):20-34. doi:10.1097/JCN.0b013e318239f9f9

14. Ehde DM, Elzea JL, Verrall AM, Gibbons LE, Smith AE, Amtmann D. Efficacy of a telephone-delivered self-management intervention for persons with multiple sclerosis: a randomized controlled trial with a one-year follow-up. Arch Phys Med Rehabil. 2015;96(11):1945-1958.e2. doi:10.1016/j.apmr.2015.07.015

15. Dwinger S, Rezvani F, Kriston L, Herbarth L, Härter M, Dirmaier J. Effects of telephone-based health coaching on patient-reported outcomes and health behavior change: a randomized controlled trial. PLoS One. 2020;15(9):e0236861. doi:10.1371/journal.pone.0236861

16. Downie S, Shnaigat M, Hosseinzadeh H. Effectiveness of health literacy- and patient activation-targeted interventions on chronic disease self-management outcomes in outpatient settings: a systematic review. Aust J Prim Health. 2022;28(2):83-96. doi:10.1071/PY21176

17. Shnaigat M, Downie S, Hosseinzadeh H. Effectiveness of patient activation interventions on chronic obstructive pulmonary disease self-management outcomes: a systematic review. Aust J Rural Health. 2022;30(1):8-21. doi:10.1111/ajr.12828

18. Hibbard JH, Greene J. What the evidence shows about patient activation: better health outcomes and care experiences; fewer data on costs. Health Aff (Millwood). 2013;32(2):207-214. doi:10.1377/hlthaff.2012.1061

19. Terry PE, Fowles JB, Xi M, Harvey L. The ACTIVATE study: results from a group-randomized controlled trial comparing a traditional worksite health promotion program with an activated consumer program. Am J Health Promot. 2011;26(2):e64-e73. doi:10.4278/ajhp.091029-QUAN-348

20. Schnock KO, Snyder JE, Fuller TE, et al. Acute care patient portal intervention: portal use and patient activation. J Med Internet Res. 2019;21(7):e13336. doi:10.2196/13336

21. Corbett CF, Daratha KB, McPherson S, et al. Patient activation, depressive symptoms, and self-rated health: care management intervention effects among high-need, medically complex adults. Int J Environ Res Public Health. 2021;18(11):5690. doi:10.3390/ijerph18115690

22. Moreno-Chico C, Roy C, Monforte-Royo C, González-De Paz L, Navarro-Rubio MD, Gallart Fernández-Puebla A. Effectiveness of a nurse-led, face-to-face health coaching intervention in enhancing activation and secondary outcomes of primary care users with chronic conditions. Res Nurs Health. 2021;44(3):458-472. doi:10.1002/nur.22132

23. Young L, Hertzog M, Barnason S. Effects of a home-based activation intervention on self-management adherence and readmission in rural heart failure patients: the PATCH randomized controlled trial. BMC Cardiovasc Disord. 2016;16(1):176. doi:10.1186/s12872-016-0339-7

24. Eikelenboom N, van Lieshout J, Jacobs A, et al. Effectiveness of personalised support for self-management in primary care: a cluster randomised controlled trial. Br J Gen Pract. 2016;66(646):e354-e361. doi:10.3399/bjgp16X684985

25. Coleman EA, Parry C, Chalmers S, Min SJ. The care transitions intervention: results of a randomized controlled trial. Arch Intern Med. 2006;166(17):1822-1828. doi:10.1001/archinte.166.17.1822

26. Parry C, Coleman EA, Smith JD, Frank J, Kramer AM. The care transitions intervention: a patient-centered approach to ensuring effective transfers between sites of geriatric care. Home Health Care Serv Q. 2003;22(3):1-17. doi:10.1300/J027v22n03_01

27. Coleman EA, Smith JD, Frank JC, Min SJ, Parry C, Kramer AM. Preparing patients and caregivers to participate in care delivered across settings: the Care Transitions Intervention. J Am Geriatr Soc. 2004;52(11):1817-1825. doi:10.1111/j.1532-5415.2004.52504.x

28. Coleman EA. Falling through the cracks: challenges and opportunities for improving transitional care for persons with continuous complex care needs. J Am Geriatr Soc. 2003;51(4):549-555. doi:10.1046/j.1532-5415.2003.51185.x

29. Thorpe KE, Zwarenstein M, Oxman AD, et al. A pragmatic-explanatory continuum indicator summary (PRECIS): a tool to help trial designers. J Clin Epidemiol. 2009;62(5):464-475. doi:10.1016/j.jclinepi.2008.12.011

30. Hemming K, Haines TP, Chilton PJ, Girling AJ, Lilford RJ. The stepped wedge cluster randomised trial: rationale, design, analysis, and reporting. BMJ. 2015;350:h391. doi:10.1136/bmj.h391

31. Savitz ST, Lampman MA, Inselman SA, et al. Overcoming challenges in real-world evidence generation: an example from an adult medical care coordination program. Learn Health Syst. 2024;8(suppl 1):e10430. doi:10.1002/lrh2.10430

32. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573-577. doi:10.7326/0003-4819-147-8-200710160-00010

33. van Walraven C, Wong J, Forster AJ. LACE+ index: extension of a validated index to predict early death or urgent readmission after hospital discharge using administrative data. Open Med. 2012;6(3):e80-e90.

34. Patient Activation Measure (PAM). Insignia Health. Accessed December 8, 2023. https://www.insigniahealth.com/pam/

35. Explore measures & activities. CMS. Accessed October 31, 2024. https://qpp.cms.gov/mips/explore-measures

36. Rural-urban commuting area codes. US Department of Agriculture Economic Research Service. Updated August 17, 2020. Accessed July 13, 2022. https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes.aspx

37. Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol. 1994;47(11):1245-1251. doi:10.1016/0895-4356(94)90129-5

38. Chew LD, Bradley KA, Boyko EJ. Brief questions to identify patients with inadequate health literacy. Fam Med. 2004;36(8):588-594.

39. Chew LD, Griffin JM, Partin MR, et al. Validation of screening questions for limited health literacy in a large VA outpatient population. J Gen Intern Med. 2008;23(5):561-566. doi:10.1007/s11606-008-0520-5