NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Williams JW, Jackson GL, Powers BJ, et al. Closing the Quality Gap: Revisiting the State of the Science (Vol. 2: The Patient-Centered Medical Home). Rockville (MD): Agency for Healthcare Research and Quality (US); 2012 Jul. (Evidence Reports/Technology Assessments, No. 208.2.)

Cover of Closing the Quality Gap: Revisiting the State of the Science (Vol. 2: The Patient-Centered Medical Home)

Closing the Quality Gap: Revisiting the State of the Science (Vol. 2: The Patient-Centered Medical Home).

Show details

Results

Results of Literature Searches

Figure 2 depicts the flow of articles through the literature search and screening process. Searches of electronic databases for Key Questions (KQs) 1–3 yielded 5,052 citations. Manual searching identified an additional 3 citations, and searches of all sources relevant to KQ 4 yielded 31 relevant citations, for a total of 5,086 citations. After applying inclusion/exclusion criteria at the title-and-abstract level, 695 full-text articles were retrieved and screened. Of these, 610 were excluded at the full-text screening stage, leaving 85 articles (representing 58 unique studies) for data abstraction.

Describes the flow of articles through the literature search and screening process. Searches of electronic databases for published studies potentially relevant to KQs 1-3 yielded 5052 citations. An additional 3 citations were identified by manual searching for articles relevant to KQs 1-3, and searches of a variety of sources identified 31 citations relevant to KQ4, for a total of 5086 citations. After applying inclusion/exclusion criteria at the title-and-abstract level, 695 full-text articles were retrieved and screened. Of these, 610 were excluded at the full-text screening stage, with 85 articles (representing 58 unique studies) remaining for data abstraction.

Figure 2

Literature flow diagram. aAll studies/articles included for KQ 1 were also included for KQs 2 & 3 Notes: KQ = Key Question; PCMH = patient-centered medical home

As indicated here, many studies included for KQs 1–3 were described in more than one publication. Appendix E provides a detailed listing of the included primary and secondary publications for these questions. Appendix F provides a complete list of published articles excluded at the full-text screening stage, with reasons for exclusion.

Description of Included Studies

For KQs 1–3, we identified 27 peer-reviewed studies; 17 were comparative and 10 descriptive. Studies were conducted in the United States (n = 23), Canada (n = 2), Israel (n = 1), and France (n = 1). Studies most commonly recruited older adults (n = 13) or children with special health care needs (n = 8). Among the comparative studies, there were 8 trials (3 good- and 5 fair-quality) involving 10,084 subjects and 9 observational studies (2 good-, 5 fair-, and 1 poor-quality).

For the KQ 4 horizon scan, we identified 31 ongoing studies, of which 2 were RCTs. These studies are described in detail under KQ 4.

Further details are provided in the relevant KQ sections, below. The following Appendixes provide details of the characteristics of included studies:

  • Appendix G. Characteristics of Included Studies (KQs 1–3, RCTs)
  • Appendix H. Characteristics of Included Studies (KQs 1–3, Observational Studies)
  • Appendix I. Characteristics of Included Studies (KQs 2–3 Only)
  • Appendix J. Characteristics of Included Studies (KQ 4)

Key Question 1. Effects of PCMH Interventions

KQ 1: In published, primary care–based evaluations of comprehensive PCMH interventions, what are the effects of the PCMH on patient and staff experiences, process of care, clinical outcomes, and economic outcomes?

  1. Are specific PCMH components associated with greater effects on patient and staff experiences, process of care, clinical outcomes, and economic outcomes?
  2. Is implementation of comprehensive PCMH associated with unintended consequences (e.g., decrease in levels of indicated care for nonpriority conditions) or other harms?

Key Points

  • Studies varied widely in the range of outcomes reported and the specific measures used.
  • The medical home in primary care settings has been evaluated in observational studies (n = 9) and RCTs (n = 8), and older adults in the United States with multiple chronic conditions were the most commonly studied population (8 of 17 studies [1 additional Canadian study among older adults]). Fewer studies evaluated the effects in general adult populations or among children with special health care needs.
  • With the exception of one study that examined facilitated versus nonfacilitated PCMH implementation, all studies compared interventions meeting the definition of PCMH to usual care.
  • Based on a combination of good- and fair-quality studies, there is evidence of moderate strength indicating that interventions meeting PCMH criteria are generally associated with small improvements in patient experiences, both on overall and care coordination measures.
  • Based on a combination of good- and fair-quality studies, there is evidence of low strength that PCMH implementation is associated with improved clinical staff experiences.
  • Based on a combination of good- and fair-quality studies, there is evidence of overall low strength that PCMH may improve care processes. This is based on a combination of moderate evidence of an effect for preventive services and insufficient evidence to evaluate impacts on care for patients with chronic illness.
  • Based on a combination of predominantly good- and fair-quality studies, there is insufficient evidence to determine the impact of PCMH implementation on clinical outcomes.
  • Based on a combination of good- and fair-quality studies, there is a low strength of evidence that PCMH implementation may lead to lower utilization (inpatient and emergency department) for some subgroups of patients, but this effect was not uniform. Moreover, total costs were not lowered in the reviewed studies.

Detailed Analysis

As a reminder, we categorized included studies into those that explicitly tested the PCMH model (“PCMH” studies) and those that met our functional definition for PCMH but did not use the terms “PCMH” or “medical home” (“functional PCMH” studies). Further, studies were excluded if the intervention was designed to address the needs only of patients with a single chronic condition (e.g., a study of disease management for patients with diabetes or asthma). However, studies were included if a broad-based intervention reported outcomes for a specific tracer condition.

In addition to examining interventions that met our definition of a PCMH or functional PCMH, studies included in the analysis for KQ 1 had to include a control group. Of 27 otherwise eligible studies, 17 comparative studies described in 42 publications reported outcomes relevant to this question. These studies include 6 with PCMH interventions and 11 with functional PCMH interventions. Sixteen studies were conducted in the United States and one in Canada. There were 8 clinical trials (all RCTs) and 9 observational studies. The majority of studies had a followup period for abstracted outcomes of approximately 2 years, with no meaningful difference between RCTs and observational studies. Most studies (9 of 17) enrolled older adults with multiple chronic health conditions; fewer studies were conducted in general adult or general pediatric populations. While a large number of patients are represented by the 4 studies with children, 98 percent of these are from one secondary data analysis study.59 For most outcomes, the small number of studies conducted among children (4 of 17 studies [2 of 8 RCTs]) precluded formal comparison with studies conducted in adults. However, results in these two populations were generally congruent. Additional characteristics are described in Table 3.

Table 3. Characteristics of studies.

Table 3

Characteristics of studies.

Abstracted Outcomes

Over the past 5 years, multiple research agendas and recommendations for evaluation measurement have been proposed for PCMH evaluations.11,26,27,36,61 Because of the variability in recommended measures for evaluating PCMH, it was necessary to restrict the abstraction of outcomes to those that had face validity to the investigators and were reported across studies, and/or were collected using validated instruments or methods. With the exception of selected economic outcomes (namely, inpatient and emergency department utilization), studies were too heterogeneous in design and in outcomes reporting for quantitative syntheses. Therefore, with the exception of the economic outcomes noted immediately above, results are described qualitatively.

Results are described below for five major domains: (1) patient experiences (including reports from caregivers); (2) staff experiences; (3) care processes; (4) clinical outcomes; and (5) economic outcomes. Within each outcome domain, we focus first on PCMH studies (n = 6) and then on functional PCMH studies (n = 11). The qualitative description of results is further stratified by presenting information from clinical trials followed by observational studies.

No studies reported all five types of outcomes. Seven studies reported one type of outcome, three studies reported two types of outcomes, three studies reported three types of outcomes, and two studies reported four types of outcomes. Table 4 describes the number of studies and number of abstracted outcomes by specified study type.

Table 4. Number of studies with specific types of outcomes.

Table 4

Number of studies with specific types of outcomes.

Comparators

With one significant exception, all comparisons presented in this report are between an intervention specifically labeled as PCMH or meeting the functional definition of PCMH and usual care. However, we also included the American Academy of Family Physicians (AAFP) National Demonstration Project (NDP), a fair-quality multicenter RCT that compared facilitated verses nonfacilitated implementation of the PCMH.62 When reading the NDP report it should be noted that while facilitated practices adopted more PCMH components than nonfacilitated practices (10.7 components vs. 7.7 components, p = 0.005), there was still substantial adoption of PCMH by nonfacilitated control clinics.62 As a result, the NDP does not represent a comparison between having PCMH and not having PCMH. However, we believed that including this large trial of PCMH implementation provides a fuller picture of the state of evidence regarding PCMH.

Patient Experiences

One or more patient experience outcomes were reported by seven studies (Table 5).60,63-68 Our summary of patient experience focuses on overall patient experience and coordination of care. If a study reported overall measures of patient experience, those measures were abstracted as opposed to individual component scales. However, care coordination was also abstracted because of the overall goal, highlighted in all major definitions of PCMH, of improving the coordination of health care services.22 For some studies, especially those involving children, experience measured may have been provided by caregivers.

Table 5. Results—patient experiences.

Table 5

Results—patient experiences.

Overall Patient Experience

Overall patient experience was reported in four studies (all RCTs) at followup periods ranging from 6 to 26 months.60,64,65,68 Two of these studies evaluated PCMH interventions and two tested functional PCMH interventions.

The AAFP NDP, the fair-quality multicenter RCT that tested the impact of facilitated PCMH versus nonfacilitated PCMH, evaluated effects on overall practice experience on a 0–1 scale after 26 months.62 There was essentially no longitudinal change within arms over 26 months based on a 0–1 overall patient experience scale (intervention -0.02; control +0.01; within-group p-value 0.92). At 26 months, there was no difference between the facilitated and nonfacilitated arms (0.26 vs. 0.33, p = 0.31).

A fair-quality trial of a PCMH intervention among children with special health care needs in a state Medicaid program compared parent-reported satisfaction with various types of care after 6 months. While satisfaction with primary care was evaluated, results were not presented. Results indicating greater satisfaction with mental health services on a three-point scale (1 = excellent, 3 = fair/poor) were presented indicating greater satisfaction among intervention patients (1.3 [SD 0.5] vs. 1.5 [SD 0.7], p = 0.004).64

A good-quality trial of guided care, meeting the definition of functional PCMH (designed using the Wagner Chronic Care Model),74 reported the overall score from the Patient Assessment of Chronic Illness Care (1–5 scale)75 at 18 months. The mean scores were higher for the guided care than usual care patients (adjusted mean difference 0.20; 95% CI, 0.07 to 0.33).69 A separate good-quality trial of geriatric management found that after 8 months of the program intervention patients were significantly more satisfied with care than with control on a 1–4 satisfaction scale developed for the study (3.28 [SD 0.68] vs. 3.13 [SD 0.77], p < 0.05).70

Coordination of Care

Aspects of patient-perceived coordination of care were reported in seven studies (five RCTs, two observational studies) for followup periods ranging from 6 months to 5½ years.60,63-68 Three of these studies were PCMH studies, and four were functional PCMH studies. This review does not address the provision of services or processes that are designed to improve care coordination. Rather, the goal is to assess the degree to which patients perceive an improved experience as a result of improved care coordination.

The AAFP NDP, the fair-quality multicenter RCT that tested the impact of facilitated PCMH versus nonfacilitated PCMH, evaluated effects on patient-reported coordination of care on a 0–1 scale (Components of Primary Care Index) after 26 months. There was no difference between the facilitated and nonfacilitated arms (0.75 vs. 0.73, p = 0.46). There was also essentially no longitudinal change in the arms over 26 months (-0.01 vs. -0.02, within-group p-value 0.11).62

A fair-quality trial of a medical home intervention among children with special health care needs in a state Medicaid program compared parent-reported satisfaction with care coordination after 6 months. Reflecting results for satisfaction with mental health services described above, the trend toward greater satisfaction with care coordination on a five-point scale (1 = excellent, 5 = poor) approached statistical significance (2.2 [SD 0.95] vs. 2.7 [SD 1.4], p = 0.058).64

The fair-quality Reid et al. evaluation of implementation of PCMH in one practice of an integrated delivery system compares results on the care coordination scale of the Ambulatory Care Experiences Survey-Short Form76 from the intervention and two control clinics (100-point scale, higher is better). Patients in the intervention clinic reported more care coordination after both 1 year (83.1 vs. 77.9, adjusted difference 3.32, p < 0.001) and 2 years (83.9 vs. 78.9, adjusted difference 3.06, p < 0.01).71

Three trials of functional PCMH interventions evaluated care coordination. A good-quality trial of guided care meeting with components meeting the functional definition of the medical home (designed using the Wagner Chronic Care Model)74 reports the coordination of care score from the Patient Assessment of Chronic Illness Care (1-5 scale)75 at 18 months. The mean scores were higher for the guided care than usual care patients (2.96 vs. 2.57, adjusted treatment effect 0.34 [95% CI, 0.18 to 0.50]).69 A separate good-quality trial of geriatric management found that after 8 months of the program intervention patients reported significantly more help obtaining services than did control patients on a 1–4 satisfaction scale developed for the study (3.11 [SD 3.41] vs. 1.57 [SD 2.48], p < 0.05).70 The third trial, a fair-quality study of enhanced developmental services for young children, examined whether parents indicated that they received needed support from their pediatrician/nurse practitioner (including with accessing needed services). The reported between-group adjusted odds ratios (95% CIs) at both 30-33 months (2.70 [2.17 to 3.45]) and 5–5½ years (1.25 [1.02 to 1.53]) indicate better care coordination in the intervention group.73

The fair-quality nonrandomized pilot study of the same guided care intervention examined integration of services after 6 months using the Primary Care Assessment Survey.77 There were no differences in changes in integration scale values between the study arms (0.10 [95% CI, -5.72 to 5.92]).72

Summary

Based on a combination of good- and fair-quality studies, there is evidence of moderate strength indicating that interventions meeting PCMH criteria are generally associated with small improvements in patient experiences, both on overall and care coordination measures based on patient or family reports. These studies included a variety of patient populations. With one exception, followup time periods were still approximately 2 years and less.

Staff Experiences

Our summary of staff experience focuses on overall staff experience. If a study reported overall measures of staff experience, those measures were abstracted as opposed to individual component scales.

Overall Staff Experience

Measures that we classify as representing overall staff experience were reported for followup periods ranging from 1 year to 26 months in three studies (all RCTs; see Table 6).60,63,65 Two of these were PCMH studies and one evaluated a functional PCMH intervention.

Table 6. Results—staff experiences (overall experience).

Table 6

Results—staff experiences (overall experience).

The AAFP NDP, the fair-quality multicenter RCT that tested the impact of facilitated PCMH versus nonfacilitated PCMH, evaluated effects on practice-level adaptive reserve.78 Practice-level adaptive reserve was based on aggregation of individual staff surveys using a 23-item scale developed for the study that included components of relationship infrastructure, facilitated leadership, sensemaking, teamwork, work environment, and culture of learning (summary scale of 0-1; higher score equates to more adaptive reserve). Intervention and control practices had the same mean level of adaptive reserve at baseline (0.69). At 26 months, intervention practices (n = 16) had greater adaptive reserve (mean 0.74, SD 0.38) than control practices (n = 15, mean 0.68, SD 0.46, p = 0.02).

In an observational study, Reid et al. examined the impact of PCMH implementation among clinicians at one intervention clinic compared to clinicians at two control clinics.63,71 Using the 22-item Masiach Burnout Inventory,79 investigators reported three components (emotional exhaustion, depersonalization, and lack of personal accomplishment) representing staff experience. At baseline, 104 clinicians responded, declining to 82 at 12 months and 48 at 24 months. At 12 months followup, scores for emotional exhaustion (value/effect size) and lack of personal accomplishment (p = 0.06) improved more for PCMH than control clinicians.63 Patterns were similar for the 48 clinicians responding to the survey after 24 months, with statistically significant lower levels of emotional exhaustion and depersonalization. However, the difference for personal accomplishment was not statistically significant (effect size not reported).71 These results are limited by the relatively low response rate at 24 months followup, which could bias the estimate of effect.

A good-quality clinical trial led by Boult et al. compared comprehensive guided care for older adults to usual care and examined physicians' satisfaction with care at 1 year (18 intervention and 20 usual care physicians). There was no statically significant difference in satisfaction with chronic illness care between intervention and control physicians. However, intervention physicians were more likely to report satisfaction with patient/family communication (mean 4.40 [95% CI, 3.99 to 4.81] vs. 3.94 [3.58 to 4.30], p = 0.014) and knowledge of patients' clinical characteristics (scale mean 3.17 [95% CI, 2.88 to 3.46] intervention vs. 2.77 [2.50 to 3.03] control, p = 0.042). The small number of providers may have limited the lack of power to detect differences. However, it should be noted that results of all nine measures of chronic illness care processes assessed had point estimates in the direction of being favorable to the intervention.80

Summary

Based on a combination of good- and fair-quality studies, there is evidence of low strength that PCMH implementation is associated with improved clinical staff experiences. However, none of the studies reporting information on staff experiences were conducted in pediatric practices. Two of the three were conducted in an older adult population. None of the studies reported outcomes more than approximately 2 years following the implementation of the intervention under study. Relatively few practices and few clinicians have been involved in these studies, and these practices may not be representative of the wider primary care practices in the United States.

Process of Care

One or more process of care outcomes were reported by seven studies.59,60,63,67,81-83 We categorized process of care outcomes into preventive services and chronic illness care services. Prioritization was given to generally accepted, guideline-recommended processes as opposed to processes that would have been implemented or enhanced specifically because of the PCMH implementation.

Preventive Services

Information on preventive services was reported in five studies (three RCTs, two observational studies) for followup periods ranging from 12 to 26 months (Table 7).60,63,67,82,83 Two of these studies were explicit evaluations of PCMH, and three tested functional PCMH interventions.

Table 7. Results—care processes, preventive services.

Table 7

Results—care processes, preventive services.

The AAFP NDP, the fair-quality multicenter RCT that tested the impact of facilitated PCMH versus nonfacilitated PCMH, evaluated effects on preventive services recommended by the U.S. Preventive Services Task Force (USPSTF).60 The facilitated PCMH practices did not significantly improve the rate of preventive services compared to the nonfacilitated PCMH practices (41.1 percent vs. 39.8 percent, p = 0.09).62

In a fair-quality evaluation of PCMH at one Group Health Cooperative of Puget Sound clinic, Reid and colleagues reported on Healthcare Effectiveness Data and Information Set (HEDIS) results compared to the rest of the Group Health system. HEDIS includes both measures of preventive and chronic illness services. Prior to PCMH implementation, the intervention clinic had better overall quality, as measured by the average percentage of 22 quality indicators achieved for each patient (68.7 vs. 64.5, statistical significance not provided). Over the 1-year intervention period, the PCMH practice showed greater improvements than the rest of the Group Health clinics (p < 0.05). However, an analysis that adjusted for differences in baseline quality did not show a statically significant improvement compared to control practices (mean difference = 1.3 percentage points, p < 0.05).71

In addition, three functional PCMH studies examined the percentage of patients receiving specified preventive services: (1) a fair-quality trial conducted as part of a care coordination Medicare demonstration project;82 (2) a fair-quality trial of enhanced developmental services for young children;67 and (3) a fair-quality observational study of team care implemented among adult patients of an integrated delivery system.83 For the Medicare demonstration trial, there were no statistically significant differences in guideline-concordant preventive services reported. For example, comparing intervention to control, results were virtually identical for receipt of adult vaccines (influenza 87.3 percent vs. 87.7 percent, p ≥ 0.10; pneumococcal 88.9 percent vs. 88.4 percent, p ≥ 0.10). There was also no difference in cancer screening based on claims data (colon 23.7 percent vs. 23.5 percent, p ≥ 0.10; mammography 74.8 percent vs.71.2 percent, p ≥ 0.10).84 In the trial of adding developmental services for very young children (0-2 years for abstracted outcomes), intervention patients were more likely to have appropriate well-child care at 12 months (90 percent vs. 81.4 percent; OR 2.06 [95% CI, 1.65 to 2.56]) and 24 months (85.2 percent vs. 78.7 percent; OR 1.68 [95% CI, 1.35 to 2.09]). At 24 months, rates of age-appropriate vaccinations were higher in the intervention practices (83.0 percent vs. 75.3 percent; OR 1.68 [95% CI, 1.59 to 1.95]).85 Finally, a team-based intervention in an integrated delivery system found higher rates of breast cancer and colorectal cancer screening rates in intervention patients after 2 years (breast 90.0 percent vs. 69.4 percent, p < 0.05; colorectal 38.1 percent vs. 23.9 percent, p < 0.05).83

Chronic Illness Care Services

Information on chronic illness care services was reported in five studies (three RCTs, two observational studies) for followup periods ranging from 1 year to 26 months (Table 8).60,63,81-83 Two of these studies were explicit evaluations of PCMH and three tested functional PCMH interventions.

Table 8. Results—care processes, chronic illness care services.

Table 8

Results—care processes, chronic illness care services.

The AAFP NDP, the fair-quality multicenter RCT that tested the impact of facilitated PCMH versus nonfacilitated PCMH, evaluated effects on health status.60 Among patients enrolled in facilitated PCMH practices, the percentage of eligible patients who received 17 recommended services for chronic conditions was not significantly improved (58.7 percent vs. 47.3 percent p = 0.92).62 Further, as noted above, in the Reid et al. evaluation of PCMH implementation, while the PCMH clinic had greater improvement in the patient average HEDIS measure that included preventive and chronic care quality measures, the difference was between the clinics was modest.71

A good-quality evaluation of a PCMH program in North Carolina that used pediatric asthma as a tracer condition found that patients in the PCMH program used 325 percent more maintenance medication than patients in the traditional fee-for-service program (5.6 percent vs. 1.6 percent, p < 0.01).59

In addition, three functional PCMH studies examined the percentage of patients receiving specified services for chronic conditions: (1) a fair-quality trial conducted as part of a care coordination Medicare demonstration project;82 (2) a fair-quality observational study of team care implemented among adult patients of an integrated delivery system;83 and (3) a fair-quality observational study of comprehensive disease management for high utilizers of different ages in a commercial health plan.81 For the Medicare demonstration trial, results for reported chronic illness care services were mixed. Among patients with diabetes, intervention patients had higher levels of lipid testing (93.1 percent vs. 86.9 percent, p < 0.01) and urine microalbuminuria testing (81.0 percent vs. 60.2 percent, p < 0.01). However, there was not a statistically significant difference for receipt of diabetes education (25.0 percent vs. 22.0 percent), eye exams (86.5 percent vs. 83.3 percent), or glycated hemoglobin (HbA1c) testing (94.9 percent vs. 94.7 percent). However all point estimates are in the direction of the intervention arm. In addition, patients with coronary artery disease had higher levels of lipid testing in the intervention compared to the control arm (89.4 percent vs. 82.5 percent, p < 0.01).84

Although a team-based intervention significantly improved preventive services in an integrated delivery system, analogous results were not seen for the two indicators of chronic illness care, warfarin monitoring (no change from baseline among intervention patients or health system as a whole) and diabetic eye exams (no statistically significant improvement among intervention patients, but improvement for health system as a whole [p < 0.0001]). However, the number of eligible patients in the intervention panel was small, and the authors contend that improvements in the delivery system as a whole for eye exams among patients with diabetes were potentially the result of low baseline rates.83

Finally, while the evaluation of enhanced disease management for high utilizers in an insurance plan provided percentages of patients meeting specific HEDIS measures for patients with diabetes, they did not provide p-values for these results. While the estimates were generally in favor of the intervention, the point estimate for the percentage of patients with eye exams was lower in the intervention than control group (57.9 vs. 65.0, p-value not reported).81

Summary

Based on a combination of good- and fair-quality studies, there is evidence of overall low strength that PCMH may improve care processes. This is based on a combination of moderate evidence of an effect for prevention services and insufficient evidence to evaluate impacts on care for patients with chronic illness. Evidence points to a potential for PCMH to positively impact care processes, especially for preventive services. While results are mixed in terms of whether differences are statistically significant, the point estimates for all but two of the comparisons are in the direction of the intervention. As noted, a lack of power may account for at least some of the differences not being statistically significant. For the two studies claiming to examine PCMH, the AAFP NDP indicated that there was not a statistically significant difference between groups for preventive (although p = 0.09) or chronic illness services. However, among all practices in the study, there was an average of 46 percent of PCMH elements in place at baseline. Further, it should be noted that organizations that did not have facilitated implementation were given credit for having a significant number of PCMH components in place at the end of the study.62 The Reid et al. evaluation of PCMH implementation at one clinic in the Group Health Cooperative of Puget Sound found that the PCMH clinic had better HEDIS performance than the rest of the organization.71 Studies of functional PCMH interventions had mixed results for individual care processes; this often included mixed results within the same study. As a result, we conclude that evidence points to a hypothesis that PCMH may improve care processes. However, more research is needed to examine this possibility.

Clinical Outcomes

One or more clinical outcomes were reported by six studies (Table 9).60,68,81,86-88 Our summary of clinical outcomes is divided into biophysical markers, patient reported health status, and mortality.

Table 9. Results—clinical outcomes.

Table 9

Results—clinical outcomes.

Biophysical Markers

One fair-quality observational study focusing on differences in costs among managed patients with high health care costs reported that patients receiving enhanced care coordination meeting the PCMH definition were more likely to have HbA1c ≤ 9.5 percent after 1 year (87.9 percent vs. 76.4 percent) and have low-density lipoprotein (LDL) cholesterol ≤ 130 mg/dL (94.2 percent vs. 67.5 percent) after 1 year of the intervention. However, no information on the size of the group or p-values was provided.81 As a result, we conclude that there is no evidence base to assess the impact of comprehensive PCMH programs on biophysical markers.

Health Status

Overall health status was reported for followup periods ranging from 1-2 years in four studies (three RCTs, one observational study).60,68,87,88 One of these studies was an explicit evaluation of the medical home and three tested functional PCMH interventions.

The AAFP NDP, the fair-quality multicenter RCT that tested the impact of facilitated PCMH versus nonfacilitated PCMH, evaluated effects on health status.60 Based on a single item measure (1-5 Likert scale), self-reported health status did not improve significantly (0.2 point improvement in each group; p = 0.80). The study authors concluded that the adoption of NDP-suggested components was not associated with change in health status.60

Two RCTs comparing functional PCMH interventions to usual care among older adults assessed differences in health status using a validated health-related quality-of-life measure (versions of the Medical Outcomes Study [MOS] Short Form questionnaire89,90). Neither study had a significant intervention effect.68,88 One of these studies88 also found no difference when examining physical functioning using the Health Activities Questionnaire.91

One observational study of a Canadian program designed to improve care coordination for frail elderly patients found that of 272 patients with moderate to severe disability at baseline, 31 percent had a functional decline (combination of mortality, institutionalization, or increase in disabilities) at 12 months compared to 49 percent of control patients (p = 0.002). While this difference was also seen at 24 months, it was not statistically significant (26 percent vs. 36 percent; p = 0.06). Also with a p-value of 0.06, the risk ratio (RR) of being institutionalized among control patients was 1.44 when compared to intervention patients.87

In summary, PCMH interventions were not associated with improved self-reported health status. Three clinical trials, two of good and one of fair quality, found no difference in self-reported health status.60,70,88 One poor-quality study found that a program designed to improve care coordination and patient autonomy decreased the proportion experiencing functional decline at 12 months (31 percent vs. 49 percent, p = 0.002) but not 24 months (26 percent vs. 36 percent, p = 0.07).87

Mortality

Two functional PCMH studies reported data on mortality among older adults receiving enhanced older adult services meeting the PCMH definition.68,86 One good-quality clinical trial with 160 total older patients (mean age 72.2) who frequently used medical services (≥ 10 outpatient visits in the last 12 months), which also found no difference in health status as measured by the MOS SF-20, found no statistically significant impact of the intervention on 24-month mortality. However, fewer patients in the intervention arm died (15 percent vs. 22.5 percent, p = 0.24).70 By contrast, a large, good-quality observational study of 1144 intervention and 2288 usual care control older patients (mean age 76.2) who were often quite sick (1.8 percent received hospice services within 90 days of the study start date) found that after 1 year 6.5 percent of intervention patients died compared to 9.2 percent of control patients (OR 0.68, p = 0.01). At 2 years, fewer patients in the intervention arm had died, but the difference was not statistically significant (OR0.77, p = 0.07). A similar pattern was seen when mortality was compared for the subset of patients with diabetes.86

Summary

Based on a combination of predominantly good- and fair-quality studies, there is insufficient evidence to determine the impact of PCMH implementation on clinical outcomes. Only one of the studies had a stated goal of testing PCMH. That study did not compare PCMH against true usual care. Further, none of the studies reporting information on clinical outcomes were conducted among children. Most were conducted in an older adult population. Among the older adult population, there is some limited indication that PCMH may have a positive impact on mortality. However, the difference was only statistically significant in one good-quality observational study after 1 year of the intervention and no longer statistically significant in that study after 2 years.86 This finding, along with nonsignificant findings of a good-quality clinical trial68 and a poor-quality observational study that reports functional decline via a measure that includes mortality,87 points to potential benefit of continuing to examine the possible link with mortality among seniors, particularly those with frailty.

Economic Outcomes

One or more abstracted economic outcomes were reported by 13 studies.59,63,65-68,81,82,86,88,92-94 Our summary of economic outcomes is divided into differences in inpatient utilization, emergency department utilization, and total costs. Inpatient and emergency department utilization may be expected to be reduced if exacerbations of disease, complications, or long-term consequences are avoided. Previous reviews of the impact of disease management programs have primarily found evidence of cost savings in situations where a primary clinical goal is prevention of disease exacerbation.95 Differences in total cost reflect the overall impact of the program on per-patient economic impact.

Utilization Meta-Analysis

Utilization of services as reported by clinical trials represents one way of examining the economic impact of interventions meeting the functional definition of PCMH. Data on inpatient utilization were available from five trials. Data on emergency department utilization were available from three trials. None of these trials were specifically designed to test PCMH; rather, all evaluated functional PCMH interventions.

Meta-analyses were used to calculate summary risk ratios, initially for studies overall, and then for the subgroup of studies that enrolled adults. The results for the effect of PCMH interventions on hospital inpatient admissions are shown in Table 10. There was no evidence of an effect of treatment when including both adult and pediatric populations (RR 0.98; 95% CI, 0.86 to 1.12). Results were similar (RR 0.96; 95% CI, 0.84 to 1.10) when analyses were limited to older adults. There was some evidence of heterogeneity, but it was not statistically significant.

Table 10. Results—trials reporting inpatient admissions.

Table 10

Results—trials reporting inpatient admissions.

The results for the effect of PCMH interventions on emergency department utilization are shown in Table 11. When both adult and pediatric populations were included, there was no evidence of an effect for PCMH (RR 0.93; 95% CI, 0.72 to 1.20). There was evidence of heterogeneity (p = 0.022). In a subgroup analysis of studies examining older adults, the intervention significantly decreased emergency department visits (RR 0.81; 95% CI, 0.67 to 0.98).

Table 11. Results—trials reporting emergency department visits.

Table 11

Results—trials reporting emergency department visits.

Utilization Analysis of Observational Studies

Because of differences in study design and populations, we thought that it was not appropriate to include observational studies in the meta-analysis with trial results. Results for the observational studies are summarized in Table 12.

Table 12. Results—observational studies reporting inpatient or ED utilization.

Table 12

Results—observational studies reporting inpatient or ED utilization.

Two fair-quality studies of limited PCMH implementation in two large integrated delivery systems reported information on inpatient and emergency department utilization.63,92 The evaluation of PCMH implementation in one Group Health Cooperative of Puget Sound evaluated adult utilization against the rest of the system. Overall inpatient admissions for all causes were essentially the same over the first 12 months (relative percent difference 99; 95% CI, 94 to 104) and first 18 months (relative percent difference 96; 95% CI, 91 to 101) of the intervention. However, when examined for the first 21 months of the intervention, there were fewer admissions in the PCMH clinic (relative percent difference 94; 95% CI, 89 to 98). Based on the literature about disease management,95 reduced use of resources may result from prevention of disease exacerbations. This possibility is reflected by the result that inpatient admissions for ambulatory care sensitive conditions were significantly lower (p < 0.001) for all followup time periods (21-month relative percent difference 87; 95% CI, 81 to 93). Likewise, there were approximately 30 percent fewer emergency department and urgent care visits for each followup period (21-month relative percent difference 71; 95% CI, 68 to 74).71

An evaluation of PCMH in the Geisinger Health Plan system utilized data from practice patients and a matched cohort to model the expected difference in hospital admissions per 1000 patients per year. Investigators estimated that there would be a difference of 56 fewer admissions among older adults (257 vs. 313, 18 percent [95% CI, -30 percent to -5 percent] difference) with PCMH as opposed to what would be expected without it.97 A separate analysis comparing patients in the health plan that had access to PCMH at non-Geisinger providers and those that did not in 2009 noted 28.0 percent fewer inpatient admissions per 1000 Medicare beneficiaries (227.5 vs. 316.7, p-value not reported) and 37.9 percent fewer inpatient admissions for commercial beneficiaries (40.5 vs. 65.2, p-value not reported). There were also 8.1 percent fewer emergency department visits among Medicare beneficiaries (282.2 vs. 307.0, p-value not reported) and 34.4 percent fewer among commercial beneficiaries (157.5 vs. 240.0, p-value not reported).92

Using childhood asthma as a tracer condition, Domino et al.59 conducted a good-quality evaluation of the impact of the often cited PCMH program Community Care of North Carolina99,100 on utilization and costs. Based on results of a multivariable regression model, investigators found that children in the medical home program had 8 percent fewer total monthly emergency department visits, 6 percent fewer monthly emergency department visits related to asthma, and 18 percent fewer monthly inpatient admissions than children with asthma in the Medicaid fee-for-service program. The p-value for all three comparisons was < 0.01.59

The final observational study with the specified goal of evaluating PCMH was a small, fair-quality study (49 PCMH patients and 146 control patients for utilization analysis) among children with special health care needs in family practice. Although point estimates were in the direction of the PCMH intervention, there was not a statistically significant difference in emergency department visit rates in the 2 years after implementation (year 1, 15.5 percent vs. 17.8 percent [adjusted rate ratio 0.795]; year 2, 12.3 percent vs. 16.6 percent [adjusted rate ratio 0.651]), although the p-value was 0.086 in year 2. The authors did not provide significance tests for inpatient admissions. However, point estimates for hospitalization rates were higher for PCMH patients than for control patients in both years 1 and 2 following implementation (year 1, 7.7 percent vs. 3.4 percent; year 2, 4.0 percent vs. 2.6 percent).93

Reflecting the meta-analysis of utilization reported in trials, the two fair-quality studies of interventions that met the functional definition of PCMH had no utilization results that favored the intervention.86,98 The one statistically significant result in fact indicated that over the 2 years following implementation of comprehensive care management at Intermountain Health Care, intervention patients had more emergency department visits (OR 1.28, p = 0.02).86

Total Costs

The impact of PCMH on total costs was addressed for followup periods ranging from 6 months to 2 years in nine studies (four RCTs, five observational studies; see Table 13).59,63,65,66,68,81,82,92,94 Three observational studies were explicit evaluations of PCMH, and six studies evaluated functional PCMH interventions.

Table 13. Results—economic outcomes: total costs.

Table 13

Results—economic outcomes: total costs.

There was no indication of a positive impact of PCMH on total costs. Despite showing a positive impact of PCMH interventions on inpatient and emergency department utilization at the Group Health Cooperative of Puget Sound and Geisinger Health Care, neither intervention was associated with reduced total cost.71,97 However, differences in costs reported comparing the one PCMH clinic to the rest of the health system (~10 percent) approached statistical significance (p = 0.114 over 12 months, p = 0.059 over 18 months, p = 0.076 over 21 months), indicating a potential trend toward lower costs.

The good-quality evaluation of Community Care of North Carolina (CCNC) using children with asthma as tracers found that while the mean costs for patients that had any services in a month were $43 (9 percent) lower for patients in the PCMH program compared to fee-for-service program, per-member per-month Medicaid costs were actually higher by $145 (95% CI, $139 to $153) than for patients in the fee-for-service system. However, as the authors point out, this may reflect greater access to service as well as billing for PCMH program components. Children in the medical home program were 58 percent more likely to have a Medicaid claim in any given month (p < 0.01). Further, this was an evaluation relatively early in the development of the CCNC program (data from 1998-2001).59

Reflecting results of the utilization meta-analyses, results from the five clinical trials of interventions that meeting the functional definition of PCMH also generally do not point to PCMH related cost savings.65,68,82,94

One fair-quality trial of enhanced care coordination found that intervention patients had higher overall annual costs when taking into account the $148 mean program fee ($209; 90 percent CI, $153 to $265; p < 0.001). Even when the fee is not taken into account, greater costs among the intervention group approached statistical significance ($61; 90 percent CI, $4 to $117; p = 0.08).84

One of the other two observational studies reporting total costs81,98 did report cost savings from an intervention that met the functional definition of PCMH. While a fair-quality evaluation of differences in costs of high utilizing patients receiving enhanced case management compared to a control commercial insurance population reports relative saving of $63 per member per month. However, statistical significance was not reported.81

Summary

Based on a combination of good- and fair-quality studies, there is a low strength of evidence that PCMH implementation may lead to lower utilization (inpatient and emergency department) for some subgroups of patients, but this effect was not uniform. Moreover, total costs were not lowered in the reviewed studies. Moreover, total costs are not consistently lowered in the reviewed studies. However, three observational studies specifically designed to test PCMH do report lower inpatient and emergency department utilization among patients in the PCMH program.59,71,92,97 However, total costs were not statistically different for PCMH and non-PCMH patients in the three studies. None of the clinical trials of functional PCMH interventions had statistically significant differences between intervention and control arms for inpatient or emergency department utilization.

No studies reported statistically significant cost savings among PCMH patients. In fact, when taking into account program costs, two studies, one good-quality trial and one fair-quality observational study, reported greater total costs among intervention patients.59,84

Effects of Specific PCMH Components (KQ 1 a)

We intended to examine the relationship between inclusion of specific elements as part of the PCMH framework and effectiveness in the five domains reviewed above. In preparation for this analysis, we generated a priori hypotheses about which specific elements would have an impact. However, there were not enough studies for each outcome domain that also had appropriate variation in PCMH elements to conduct such an evaluation. As a result, we conclude that there is not currently sufficient evidence to evaluate whether specific PCMH components are associated with greater effects on patient and staff experiences, process of care, clinical outcomes, and economic outcomes.

For more information on the specific PCMH components implemented in the included studies, please see the results section for KQ 2, below.

Unintended Consequences (KQ 1b)

The issue of unintended consequences was not specifically addressed in any of these controlled studies. However, two studies, one a good-quality observational evaluation of a Medicaid medical home program59 and another a fair-quality clinical trial of a Medicare disease management demonstration program meeting the functional definition of PCMH,84 report that when costs of the program are taken into effect, overall costs are greater for the PCMH intervention. Questions concerning the potential of the costs of PCMH programs themselves leading to increased costs are an important potential area of future study.

Key Question 2. PCMH Components Implemented

KQ 2. In published, primary care–based evaluations of comprehensive PCMH interventions, what individual PCMH components have been implemented?

Key Points

  • Eight of 27 studies addressed children and adolescents only, one study addressed all ages, and the remaining 18 studies addressed adult-only patient populations (9 of these 18 were specific to older adults).
  • Twenty-one of 27 studies reported approaches that addressed all 7 major PCMH components. These included team based-care, sustained partnership, reorganized or structural changes to care, enhanced access, coordinated care, comprehensive care, and a systems-based approach to quality. We abstracted 51 different strategies or approaches across these seven major PCMH components and found considerable variability across studies based on what was reported.
  • PCMH interventions used a greater number of approaches than functional PCMH interventions to address the seven major medical home components.
  • Team-based care: 93 percent of the studies reported multiple disciplines as part of the team in addition to a physician and nurse.
  • Comprehensive care: 93 percent of studies addressed chronic illness care, and only 26 percent included specialty care.
  • Coordinated care: 63 percent of studies coordinated care transitions across settings. Only 11 percent reported integration of mental health.
  • Quality: 41 percent of studies reported the use of electronic health records and 15 percent were reportedly new.

Detailed Analysis

This section of the report presents a synthesis of the individual PCMH components reported in the 27 included studies. Of the 27 studies, 8 included only children and adolescents, 1 included all ages, and the remaining 18 included adult-only patient populations, with 9 of the 18 specific to older adults.

PCMH is defined as a comprehensive intervention that includes items 1, 3, and 4, below, along with at least two components of item 2. The number of strategies or approaches (areas) examined for each component is noted:

  1. Team–based care (six areas examined)
  2. At least two of the following:
    1. Enhanced access to care (nine areas examined)
    2. Coordinated care (eight areas examined)
    3. Comprehensiveness (four areas examined)
    4. A systems-based approach to improving quality and safety (10 areas examined)
  3. A sustained partnership oriented toward the whole person (six areas examined)
  4. Reorganized care delivery (through structural changes to the traditional practice; eight areas examined)

For each component a comparison is made between PCMH studies (n = 10) and studies of functional PCMH (n = 17), and between studies with pediatric-only patient populations (n = 8),59,64,67,93,102-105 adult-only patient populations (n = 18),60,63,65,66,68,82,83,86-88,92,94,106-111 and the study with patients of all ages (n = 1).81 Seven of the eight pediatric-only studies were studies of PCMH.59,64,93,102-105 The oldest study, by Rubin and colleagues (1992),94 was the only study to report implementation of just two of the four elements listed under item 2. Four additional studies implemented three of these elements, and the remainder (81 percent) included all four. With the exception of the enhanced access to care component, there was little to no difference between PCMH and functional PCMH studies in reporting details for each component. It is important to note that while some studies reported multiple approaches or strategies for implementing a particular component, evidence of only one approach was required. Each component is analyzed independent of the next for this KQ and is described in more detail below.

Team-Based Care

The composition of teams varied widely across studies; within comparisons by physician, nurse, and mid-level provider groupings; and within analytic groups (PCMH vs. functional PCMH and pediatric vs. adult vs. both) (Table 14). It was most common to report having a physician and a nurse (56 percent). All but two studies reported other disciplines as part of the team. Four studies, two PCMH (one pediatric, one adult) and two functional PCMH (adult only), did not explicitly report having a designated physician for the patients. Nurses and case managers were more frequently reported as the primary contact, but no single discipline was reported in this role for ≥ 15 percent of the studies. Five of the nine studies with pediatric patients did not report a primary contact for the patients and/or their families. The majority of studies (67 percent) reported team members to have defined roles. A different set of 16 studies (67 percent) reported that team members had dedicated time for PCMH activities, and 63 percent had dedicated team meetings. Not all teams were co-located.

Table 14. Team-based care.

Table 14

Team-based care.

Enhanced Access to Care

Several strategies were described that may enhance patient and family access to services and providers; these are presented by those reported most to least frequently in Table 15. A higher proportion of PCMH studies compared with functional PCMH reported advanced clinic access (40 percent vs. 12 percent), group visits (20 percent vs. 6 percent), telephone visits (40 percent vs. 29 percent), disease management (30 percent vs. 18 percent), and enhanced telephone or electronic communication options (50 percent vs. 29 percent). Access to a provider at all times (24/7 coverage) was rare and was only reported in two studies; both included only adults. Only one pediatric study offered advanced clinic access, and none offered group visits.

Table 15. Strategies reported that may enhance access to services and providers.

Table 15

Strategies reported that may enhance access to services and providers.

Coordinated Care

Care coordination was not a required component for inclusion in this review but was addressed by all 27 studies. Examples are presented by those reported most to least frequently in Table 16. Coordination with community resources either with a community liaison or referral was addressed by 67 percent of the studies, more common among functional PCMH than PCMH (71 percent vs. 60 percent), and in 6 of the 9 studies that included pediatric patients. Also common, but not equally distributed between groups, was the focus on coordinated care transitions—only 3 of 9 studies that included pediatric patients and 76 percent of functional PCMH vs. 40 percent of PCMH studies. Previsit planning, tracking the results of tests, and tracking referrals were reported in six or fewer studies. None of the studies of pediatric patient populations coordinated home health, included pharmacist activities, tracked tests, or integrated mental health.

Table 16. Coordination of care strategies.

Table 16

Coordination of care strategies.

Comprehensiveness

Four service areas were examined to describe the comprehensiveness of the intervention (Table 17). All but two studies (one pediatric PCMH, one adult functional PCMH) addressed chronic illness care. In studies that addressed only one service area (n = 6), the focus was on chronic illness care rather than preventive care (five vs. one studies, respectively). Preventive wellness care was addressed by 18 studies, a higher proportion of PCMH than functional PCMH (80 percent vs. 59 percent). Also more frequently addressed by PCMH than functional PCMH was acute care (90 percent vs. 65 percent). Specialty care was only included in studies that addressed all other service areas (n = 6), and only one of these six studies was PCMH. PCMH studies more commonly addressed three of the service areas but not specialty care and this was true for all three of the PCMH studies of adult populations.

Table 17. Comprehensiveness—addressing patients' needs measured across four service areas.

Table 17

Comprehensiveness—addressing patients' needs measured across four service areas.

Systems-Based Approaches to Improving Quality and Safety

Several systems-based approaches to improving quality were reported but only two of these by more than 50 percent of the studies: 59 percent identified high-risk patients, and 52 percent reported to use evidence-based practice guidelines (Table 18). Performance monitoring and the use of electronic health records were each reported in 11 studies. Reid and colleagues reported several approaches, including an orientation to the practice for new patients, a reduced panel size, longer appointment times, and electronic prescribing.63 Electronic prescribing was also reported by Steele and Jaen.60,92 Like Reid, Zuckerman reported longer appointment times and providing an orientation to the practice for new pediatric patients.67 Such an orientation was also addressed by Sommers 2000.88

Table 18. Systems-based approaches to improving quality and safety.

Table 18

Systems-based approaches to improving quality and safety.

Sustained Partnership

Approaches to supporting a sustained partnership with patients were examined and are presented in order of how they are likely to present in working with a new patient (Table 19). Although all studies were required to address this component with indication of treating the “whole” patient, only three studies, each for adult populations, reported specific strategies to include patients in the decisionmaking for their care. Reported most frequently were care plans and comprehensive assessments of patients (67 percent and 63 percent respectively). The latter was more common among functional PCMH studies (71 percent) than PCMH studies (50 percent). Self-management support was more common among PCMH studies (50 percent vs. 35 percent of functional PCMH studies). The provision of family caregiver support was reported in 10 studies, 5 pediatric and 5 adult, and similar proportionally among PCMH and functional PCMH studies, 40 percent and 35 percent, respectively.

Table 19. Strategies reported to facilitate a sustained partnership.

Table 19

Strategies reported to facilitate a sustained partnership.

Reorganized Care Delivery

Examples of reorganized care and structural changes were not reported in isolation of other PCMH components. Table 14 addresses team-based care and important elements of staff, roles, and the location of the team. In describing the design of the intervention, 78 percent of studies reported that new staff were added, 12 studies indicated the roles that were defined were new roles, and two studies reported a new physical location for providing patient services (Table 14). New organizational affiliations were reported in four studies,87,92,108,109 and Domino and colleagues in their study addressing chronic illness care among pediatric patients reported to have established a “new entity.”59 The creation of new services was reported in 63 percent of studies,59,60,64,65,81,82,86,87,92,102,105-111 similar among PCMH and functional PCMH studies (60 percent vs. 65 percent, respectively).

Key Question 3. Financial Models and Implementation Strategies

KQ 3: In published, primary care–based evaluations of comprehensive PCMH interventions, what financial models and implementation strategies have been used to support uptake?

Key Points

  • Of the 27 studies included in our literature review, 22 studies (45 articles) reported information about the financial models and/or implementation processes (either organizational learning strategies or actual implementation strategies) used to support uptake of PCMH interventions. Nine of the 22 interventions studied were explicitly described as PCMH; the remaining 13 were not so described, but met our functional definition of PCMH.
  • Seven of the 22 studies involved pediatric populations (6 PCMH and 1 functional PCMH). The financial models and implementations strategies were similar between the pediatric and adult studies; we therefore report the results for the full set of studies.
  • Relatively few studies (11 of 22) described any aspect of change in financial models. The financial models described varied greatly in the scope of the financial changes implemented and in the level of detail reported.
  • In both PCMH and functional PCMH studies, the most commonly used organizational learning strategies, implemented in 19 of 22 studies, were formal learning collaboratives and/or collaborative program planning for practice team members to learn about the new intervention and the processes of change being implemented.
  • In both PCMH and functional PCMH studies, the most commonly employed implementation strategies, used in 13 of 22 studies, involved some form of audit and feedback, often in the form of quality improvement methodology.

Detailed Analysis

The shift of focus for primary care clinics away from a fee-for-service driven practice directed at acute medical care toward the medical home model, which is focused more holistically on prevention and the management of both acute and chronic medical conditions, requires many changes at the levels of the provider, practice, and health system. In our review of the literature, we were interested in processes of care that studies implemented to help practices become medical homes. We are not aware of studies that have rigorously tested these processes of care for their efficacy, so we will qualitatively describe what has been done to date in this area.

We abstracted data related to financial models and implementation strategies used to change primary care clinics into medical homes or into clinics with functions similar to medical homes. In what follows, we begin by describing the financial models used for PCMH changes, that is, any changes made to the financing of providers, the practice, or health system as part of PCMH implementation. Next, we focus on two areas related to processes of care in the area of implementation: (1) organizational learning strategies, and (2) implementation strategies. Organizational learning strategies are mechanisms through which providers and staff gain knowledge about, or provide feedback about, how to make their practice more consistent with PCMH. Implementation strategies are strategies that are used, generally at the level of the practice, to implement the changes needed to be more consistent with PCMH, as well as the methods used to measure the impact of the PCMH transformation on clinical care processes or outcomes. In abstracting this information from the studies, we found that there was often overlap in the processes of change that could be considered both organizational learning strategies and implementation strategies, as described below.

Our literature review identified 22 studies (45 articles) that described strategies used for organizational learning or implementation of PCMH interventions; 11 of these also described some component of a financial model for these PCMH interventions. Nine of the 22 interventions studied were explicitly described as PCMH;59,60,63,92,93,102-105 six of these involved pediatric populations.59,93,102-105 The remaining 13 were not described to be a PCMH intervention, but met our functional definition of PCMH.65-67,82,83,86,88,94,107-111 Of these interventions, only one67 involved a pediatric population. Table 20 summarizes the number of studies included in this section and the strategies employed. Below we describe in more detail the financial, organizational learning, and implementation strategies employed in these interventions. While we did not find any clear pattern of strategies that distinguished these interventions, we describe the interventions qualitatively according to whether the intervention was explicitly PCMH or functionally PCMH. We also did not find any clear pattern of strategies which distinguished interventions employed for pediatric versus adult populations, and so we have combined studies for all patient populations in our descriptions.

Table 20. Numbers of studies describing financial, organizational learning, and implementation strategies.

Table 20

Numbers of studies describing financial, organizational learning, and implementation strategies.

Financial Models

Under the heading of “financial models,” we considered any change to the financial structure of clinics required for the financing of the PCMH or functional PCMH interventions. The types of financial restructuring we anticipated being reported included bundled payments for most health services; PCMH per member, per month payment for PCMH activities; pay for performance; enhanced fee-for-service compensation; accountable care organization; and revised pharmacy benefits. On reviewing the included studies, however, we found that the amount of detail provided about the short-term financing and the envisioned long-term financing of these interventions varied greatly and often did not correspond to these categories. In what follows, we describe the information actually provided as clearly as possible.

PCMH Studies

Five PCMH studies59,63,92,103,104 reported some aspect of the financing of the PCMH intervention. One study was small-scale and funded by an external grant.104 Two studies received financial stipends for certain aspects of their interventions—one to fund a local parent consultant for each clinic,103 and another to offer additional services such as enhanced phone access;92 only the latter study detailed the source of the stipend.

Some studies described more significant changes to the overall financial model of the clinic practices. One study59 introduced reimbursement on a per-member, per-month basis and used the fees generated to cover the cost of case management. Two studies59,92 describe the use of an enhanced fee-for-service program as part of their financial model. The Group Health PCMH pilot study63 reduced providers' panel size and increased appointment time length to accommodate the different design component of the intervention; this study also changed provider compensation from a fee-for-service model to fixed-salary compensation without relative value unit (RVU)-based adjustments. In the Geisinger's ProvenHealth Navigator study,92 there were several changes to the reimbursement model. They created a hybrid program with fee-for-service payments, payments for achieving certain quality and efficiency targets determined jointly by the providers and health plan teams, and stipends to support the PCMH implementation changes within the practices.

Functional PCMH Studies

Six functional PCMH studies66,67,82,94,107,111 described some aspect of their financial model. Four studies received funding to support components of their interventions.66,67,94,111 One study was funded by a grant to support its intervention with a Geriatrics Assessment Team,94 and another received separate funds from their health care system without significant changes to the care reimbursement of the clinic practices for funding of its Guided Care Nurse and for administrative support.66 One large national intervention, called the Healthy Steps pediatric program,67 was funded by The Commonwealth Foundation and by local organizations, which developed and supported certain aspects of the intervention. The Colorado Regional Integrated Care Collaborative (CRICC) pilot program111 received some of its program funding from the Colorado Health Foundation.

One functional PCMH intervention implemented a reimbursement program on a per member, per month basis, and used these fees to cover the cost of the services provided as part of the intervention.82 Two studies82,107 offered extra compensation for providers' time spent on aspects of the intervention that detracted from their clinical time and productivity, such as collaborating with other providers who were often located in different clinics,107 or developing and implementing guidelines for the intervention.82 The CRICC pilot program,111 which provided care to certain Medicaid recipients, received much of its funds through a capitated risk contract with the state.

Organizational Learning Strategies

Organizational learning strategies were defined as the mechanisms through which providers and staff gained knowledge about, or provided feedback about, how to make their practice more consistent with PCMH. Categories of organizational learning strategies abstracted for this review included:

  1. Formal learning collaboratives, such as lectures and training sessions
  2. Collaborative program planning, such as team meetings to educate and to get feedback regarding ongoing processes for the purpose of improvement
  3. Community of practice, in which groups of professionals from different practices could consult each other and work together to improve care with a common goal
  4. Designated research/project team assistance for PCMH development and implementation, usually from the study team
  5. Use of implementation toolkits, often designed by the study team, to help practices develop PCMH functions, conduct audit and feedback, and learn other techniques to help with implementation of PCMH
  6. Other

When we abstracted data, we found that the first two categories were often combined, so we have grouped them together below and in Table 20.

PCMH Studies

Among the nine PCMH interventions, eight60,63,92,93,102-105 described the use of formal learning collaboratives and/or collaborative program planning, which were often combined. A majority of these strategies took the form of regularly scheduled team meetings to discuss issues such as clinic work-flow,92 to provide feedback regarding program design and interventions,60,105 and to provide a forum to discuss experiences.60 Formal didactic sessions (with continuing medical education) were often offered on topics about PCMH,104 community-based services and clinic policies,93 or health literacy.105 For example, the National Demonstration Project (NDP)60 held four 2-day learning sessions over a 2-year period with two representatives from each intervention clinic. In the didactic sessions, presenters discussed PCMH programs and demonstrated technologies that enabled the implementation. Some sessions were interactive and allowed members of different teams to network and share ideas.

Three studies59,60,103 describe a community of practice in which intervention practices had regular contact for sharing their experiences. Two studies59,60 had monthly conference calls among practice providers to discuss their progress and barriers toward achieving PCMH intervention goals, while the third103 had face-to-face meetings among physicians of six practices to discuss issues around practice management.

Two studies60,102 had designated research/project team assistance from study team members (external to the clinic staff) who provided training in PCMH process implementation and were available to help or advise clinic staff either on- or off-site, via email or phone. For example, the NDP60 had a total of 3 facilitators for the 36 intervention clinic sites who assisted with clinic implementation of the PCMH components. These facilitators made initial site visits of 2-3 days' duration in order to get to know the practice via in-depth interviews and observations. They also made subsequent on-site visits during the intervention period. However, the majority of their facilitation was provided during monthly conference calls, when multiple intervention practices shared their ideas and experiences, or through email, where facilitators could provide more clinic-specific recommendations.

Two PCMH studies60,102 described their use of implementation tool-kits. These studies provided online resources and manuals to help clinic staff with implementation changes.

Functional PCMH Studies

Eleven of the 13 functional PCMH studies65-67,82,83,86,88,107,108,110,111 describe employing interventions that involved formal learning collaboratives and/or collaborative program planning sessions, which often overlapped in their function. For example, the Guided Care intervention65,66 contained an intensive 9-week program for nurses who were the designated Guided Care Nurses for a group of intervention clinics. The planning sessions consisted of didactic lectures, assigned readings, and learner participation in motivational interviewing, along with skill development through interactive role-playing. In addition, this intervention included meetings of the clinic managers, their assigned Guided Care Nurses, and study team members to discuss current implementation problems and plan future implementation steps. The CRICC pilot program111 utilized an established training program, Care Management Plus, to train care managers. This involved using many learning modules which covered aspects of care such as patient coaching, motivational interviewing, and chronic disease management issues.

Two studies67,108 described a mechanism for community of practice. For example, the Healthy Steps pediatric intervention67 facilitated monthly telephone calls during which the practices received technical assistance from the study team and discussed issues surrounding implementation strategies and best practices.

Three of the larger, multi-site studies67,82,111 provided designated research/project team assistance. The Medicare Coordinated Care Demonstration (MCCD)82 designated a study team member (an advanced practice nurse [APN] consultant) to work closely on-site with multiple practices to guide program improvement, guideline development, and implementation. The Healthy Steps program67 created a National Advisory Committee, which conducted an initial evaluation of the 15 implementation sites and provided resources, oversight, and leadership, but which did not provide on-site direct assistance. The CRICC pilot program assigned “highly experienced registered nurses” to supervise all care managers.

Two of these large studies67,82 created implementation tool-kits to help intervention practices with programmatic changes. Examples of tool-kits include pocket cards, Web resources,82 and a training videotape with manual.67

Implementation Strategies

Implementation strategies are methods employed by the practices to implement the changes needed to be more consistent with PCMH, as well as the methods used to measure the impact of the PCMH transformation on clinical care processes or outcomes. The categories of implementation strategies initially used for data abstraction for this review include:

  1. Audit and feedback to providers, teams, and/or clinics
  2. Quality improvement measures
  3. Academic detailing
  4. Lectures/classes for staff (i.e., didactic education)
  5. Designated clinical champion (facility/practice level)
  6. Designated project manager (facility/practice level)
  7. Plan-Do-Study-Act cycles/rapid cycle improvement mechanisms
  8. Flow mapping of care system
  9. Total quality improvement/continuous quality improvement
  10. Strengths-weakness-opportunities-threats analysis
  11. External benchmarking at the organizational level
  12. Other

Through the data abstraction process, we found that we often had to draw some inferences regarding the implementation strategy from the description of the process of change in order to categorize them. We also combined some of these categories when clear distinctions could not be made, as described below, and as indicated in Table 20.

PCMH Studies

The most commonly described implementation strategy among the nine PCMH interventions was some form of audit and feedback or more formal measures of quality improvement either at the provider level or the practice level. Six interventions59,60,63,92,104,105 involved some form of practice performance review and feedback to the practice team, with the overall goal of improving implementation of PCMH changes. Examples of the audit and feedback mechanism included a visual reporting system to track changes63 and a compilation of outcomes and quality metrics, with performance reports and recommendations regarding modification of methods provided back to the practices.92 One study104 conducted monthly meetings led by practice quality improvement (QI) teams, while most studies did not describe such formal meetings.

Some interventions employed an implementation strategy very similar to the previously described organizational learning forums. Four PCMH interventions60,102-104 employed academic detailing or lectures and classes for clinic staff, sometimes within the informal setting of team meetings, as forums to discuss changes in implementation strategies. For example, the Illinois Medical Home Project104 held three learning sessions over an 18-month period for implementation training and practice quality improvement.

Four interventions59,60,93,105 had designated clinical champions or project managers to assist with implementation of PCMH changes. These individuals, primarily from the study team and not a part of the clinical practice, provided guidance on PCMH implementation and improvement strategies. For example, for the Medical Home project of the Texas Children's Health Plan (TCHP),105 an individual from the TCHP Health Promotion Program was responsible for implementing PCMH changes within their assigned practices, taking into account each practice's unique environment.

Three interventions63,92,104 implemented rapid cycle improvement mechanisms for evaluating changes that occurred. The Group Health PCMH initiative63 used “team-based rapid process improvements” to incorporate changes into their clinic practice. Geisinger's ProvenHealth Navigator program92 also used the process of rapid cycle innovation to make short-cycle changes to care coordination processes for patients with chronic medical conditions. Similarly, the Illinois Medical Home Project utilized the Plan-Do-Study-Act cycle of practice improvement for their PCMH implementation.104

Functional PCMH Studies

Seven of the 13 functional PCMH studies65,67,82,83,86,108,109 employed techniques of audit and feedback or QI initiatives to enhance implementation of PCMH changes in their practices. One study86 tracked tasks that were due but not yet completed from individual patient care plans and kept a “tickler list” for the practice care manager. The other six studies65,67,82,83,108,109 generated performance reports with process of care, clinical outcomes, and financial information for practice team members to review and improve performance.

Six interventions67,86,88,107,108,111 used academic detailing or lectures/classes for staff to implement the care coordination changes. As previously noted, this strategy was similar to collaborative program planning forums and could not necessarily be distinguished from them. Within these academic detailing sessions, the study team provided updated care guidelines or made recommendations of changes to their care processes for further implementation. For example, one study108 conducted quarterly meetings to present data on quality indices and resource utilization in order to help optimize these measures in future performance audits. The CRICC pilot program111 held weekly multidisciplinary consultations with a medical director and also held regular treatment team meetings at the larger clinic sites.

Only 1 of the 13 functional PCMH studies clearly described having a designated clinical champion or project manager. The Group Health Cooperative of Puget Sound intervention83 designated a member of the practice team as the leader of the new intervention who would “…assume responsibility for organizing meetings, setting long-term strategy, and maintaining a vision.” While in the four PCMH studies the clinical champion was a member of the study team and external to the practice, in the Group Health Cooperative study the champion was a member of the practice.

One of the functional PCMH studies111 described a type of rapid cycle improvement mechanisms for evaluating changes that occurred during the implementation phase of the program. This internal evaluation process was said to be modeled on the multimethod assessment process/reflective adaptive process.112 This study also collected both quantitative and qualitative data through meeting minutes, key informant interviews, and surveys as part of its internal evaluation process. However, this study did not describe exactly how these data were used to inform changes.

Key Question 4. Horizon Scan of Ongoing PCMH Studies

KQ 4: What primary care–based studies evaluating the effects of comprehensive PCMH interventions on patient and staff experiences, process of care, clinical outcomes, or economic outcomes are currently under way? In these ongoing studies, what are the study designs, PCMH components, comparators, settings, financial models, and outcomes to be evaluated?

Key Points

  • We identified 31 ongoing studies of comprehensive PCMH interventions that specified a comparison group and met our other inclusion criteria.
  • Studies included a broad representation of geographic areas, with individual studies mostly conducted within a single state.
  • Only 2 of the 31 studies were RCTs; the remainder were quasi-experimental or observational studies.
  • Seventy-one percent (71%) of studies are scheduled for completion in 2012.
  • The studies differed in the specific PCMH components they specified. The median number of components specified across all studies was 3.5 (of a possible 7). The most infrequently reported PCMH components were comprehensiveness and a sustained partnership (27% each).
  • Several different financial models for PCMH implementation were reported. Enhanced fee-for-service was reported in 19 percent of studies. Bundled payment per member and pay for performance were each reported in 23 percent of the ongoing studies.
  • Most studies intend to collect outcomes pertaining to patient or staff experiences, processes of care, and economic outcomes. Only one-third of studies reported an intention to collect and report on clinical outcomes.
  • Limited information reported on ongoing studies restricted our ability to ascertain study design, components of the PCMH included, comparison interventions, and planned outcomes with certainty. Many ongoing demonstration projects were excluded because they lacked sufficient detail to meet our inclusion criteria.

Detailed Analysis

The sources searched for KQ 4 are detailed in the Methods chapter. Searches of all sources identified 900 citations, of which 204 were selected for further independent review by two investigators. After this review, we included 31 records that described ongoing or planned evaluations of PCMH interventions that were conducted in the United States and included a comparison group for the evaluation. Among the reviewed PCMH demonstration projects, the most common reason for exclusion was the lack of a comparison group specified in the evaluation plan. Most of the included records came from online databases that catalogued ongoing projects affiliated with the sponsoring organization. This included: 10 citations/studies from the Patient Centered Primary Care Collaborative (PCPCC);113-122 10 citations/studies from enGrant scientific (a database of federally sponsored research);123-132 4 from The Commonwealth Fund;133-136 2 each from Robert Wood Johnson Foundation137,138 and Clinicaltrials.gov;139,140 and one from the CMS Web site.141 Direct email contact to representatives of CMS and the Department of Veterans Affairs yielded one additional study.142 In addition to this primary search, we used a published horizon scan on PCMH based on semi-structured interview of lead personnel as an additional resource.46 This review identified one additional study for inclusion.143 These sources varied significantly in the level of detail provided, with most providing one to two paragraphs of description, while others provided reports exceeding 100 pages. Nearly three-quarters of these studies are targeted for completion in 2012.

The number of participating patients, providers, and clinics was reported for 56 percent of the included studies. Twelve studies were conducted exclusively in adults, 1 study in children, 5 studies in both adults and children, and 13 studies did not specify the population. Among studies for which data were available, the median number of patients was 27,000 (range 300–2,000,000); the median number of participating providers was 66 (range 8–7618); and the median number of participating clinics was 14 (range 1–1200). The number of patients was often based on the number of covered lives under a particular insurance program and may not reflect the number of patients receiving care within a PCMH.

Table 21 summarizes the most important characteristics of the 31 ongoing studies. The majority of these are being conducted in a single state, in cooperation with a single insurance payer. While several payers, such as Humana and Blue Cross/Blue Shield, supported projects in multiple states, the extent of collaboration across states was not clear. Overall, the included studies broadly represented different geographic areas of the United States. Two studies were RCTs with randomization at the patient level. There were no cluster randomized controlled trials, and the remainder of studies were quasi-experimental or observational evaluations of PCMH interventions. For many of the studies, it was difficult to ascertain clearly the level of care received by the comparator groups. The term “usual care” can vary substantially across different settings, yet this was the most common comparator reported. This was followed by studies comparing differing levels of PCMH implementation, in which practices were considered to be more or less of a comprehensive medical home.

Table 21. Characteristics of ongoing studies (n = 31).

Table 21

Characteristics of ongoing studies (n = 31).

The studies differed in the PCMH components specifically included in the ongoing study. The median number of components reported across all studies was 3.5 (of a possible 7). The most infrequently reported PCMH components were comprehensive care and a sustained partnership, each of which was reported in only 29 percent of the included studies. Nearly half of the ongoing studies did not specify any financial support for PCMH implementation. Among studies that did report details of their financial models, the most common approaches were enhanced fee-for-service, bundled payment (usually per member/per month), and pay for performance based on prespecified targets. Most studies intend to collect outcomes on patient or staff experiences, process of care measures, and economic outcomes; only one-third specified clinical outcomes as part of their planned analysis.

Further details of these studies are provided in Appendix J.

PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.4M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...