The relationships between access to and use of a patient-centered medical home and healthcare utilization and costs: A cohort study using Medical Expenditure Panel Survey data from 2007 to 2010

Abstract

Objective

The patient-centered medical home has emerged as an innovative healthcare delivery model that holds the conceptual promise to improve healthcare quality while reducing costs. This study is among the first to examine how patient perceived patient-centered medical home is related to utilization and costs for emergency department and inpatient care nationwide in the U.S.

Methods

This retrospective cohort study used data from the 2007–2010 Medical Expenditure Panel Survey. This study focused upon insured individuals aged 18 and older. In each two-year cohort, we measured and identified a full patient-centered medical home group, a partial patient-centered medical home group (with a usual source of care but not a patient-centered medical home), and an unknown patient-centered medical home group. Using negative binomial regression and generalized linear models, we conducted the analysis while controlling for covariates at baseline. Given the nature of the complex survey design, we adjusted weights and variance.

Results

The study sample consisted of 15,595 individuals, representing a total of 368 million people in the U.S. In the trend of outcome changes from the baseline to the follow-up year, the full and partial patient-centered medical home groups demonstrated reduced efficiency for the 2007–2008 cohort, but increased efficiency for the 2009–2010 cohort, as compared to the no regular provider group.

Conclusions

Overall, the empirical evidence does not indicate whether patient-centered medical home models reduce healthcare utilization and costs, but it does suggest their potential as mechanisms for achieving healthcare system efficiency, when primary care practices have grown from early to middle stage of patient-centered medical home transformation. A longer observation window and holistic view on all stages of patient-centered medical home growth may be more informative on patient-centered medical home’s efficiency.

Keywords

Emergency department and inpatient care health system efficiency nationwide patient-centered medical home patients’ experience utilization and costs

Introduction

Rapidly rising healthcare costs continue to be a significant concern in the U.S. where spending on healthcare is a greater proportion of gross domestic product compared to other countries in the world (17.6% in 2009) and yet high quality and efficient healthcare is not consistently provided.^1–6 The patient-centered medical home (PCMH) is an innovative healthcare delivery model that holds the conceptual promise of improving healthcare quality while reducing costs.

While there are variations in the definition and measurements of this model, the PCMH basically consists of the following key principles: a personal physician, a physician-directed medical practice, a whole person orientation, coordination and/or integration of care, quality and safety as hallmarks, enhanced access to care, appropriate payment and incentives, and wide-ranging team-based care.^7,8 PCMH models integrate the attributes of high-quality primary care and the chronic care model. Described as a “lifeline for primary care,” the PCMH has the potential to transform and increase the appeal and viability of primary care practice.⁹ Further, early demonstration programs have indicated that the PCMH may introduce healthcare system efficiencies to contain rising healthcare costs and bring about substantial savings in total healthcare costs.^10,11 System efficiency is achieved by reconfiguring the primary care system, which emphasizes robust primary care that is organized and paid for in a new way and that enhances patients’ access to efficient and coordinated care.¹² High-quality primary care may effectively prevent patients using unnecessary emergency department (ED) visits or hospital inpatient care.

Although promising conceptually, the economic evidence for the benefits of comprehensive PCMH interventions is limited and uncertain. A systematic review conducted by the U.S. Agency for Healthcare Research and Quality in 2012 provided an overall low strength of evidence for an association between PCMH and lower healthcare use among 11 identified studies.¹³ Specifically, it found lower ED visits in older adults, but no reduction in inpatient admissions, with imprecise estimated effects; and it also reported insufficient evidence regarding reductions in total costs in adults and for all economic outcomes including ED visits, inpatient admissions, and total costs in children. Nevertheless, other recent studies reported reductions in ED visits, hospitalization, and/or total cost savings.^14–24

The current study uses nationally representative data to empirically analyze whether the PCMH is associated with lower utilization and costs for ED and inpatient care. Besides mixed findings, prior studies suffered from gaps that limit both internal and external validity of their findings. Almost all of them examined the research question within regional health systems, in a pilot demonstration program, or among certain subpopulations, such as chronically ill patients or private enrollees, and thereby lack broad generalizability.^14–24 Moreover, most PCMH intervention groups did not consist of an adequate number of practices or have adequate sample size, for example, the study using HealthCore data contained only 10 practices in the PCMH category.¹⁴ The current study advances the policy debate by filling these gaps and by being among the first to examine empirical evidence that the PCMH is potentially related to lower utilization and costs for ED and inpatient care using a large nationally representative sample.

Methods

Data

We used Medical Expenditure Panel Survey-household component (MEPS-HC) data for this study.²⁵ The MEPS-HC is a nationally representative survey of the noninstitutionalized civilian population of the U.S. and is designed to produce national estimates of the healthcare use, costs, sources of payment, and insurance coverage of the noninstitutionalized U.S. population.

Each new panel covers a series of five rounds of in-person interviews.²⁶ This design, which covers two full calendar years, allows for the tracking of individuals’ utilization and costs over time. Like many other national survey designs, MEPS adopts a complex multistage, unequal probability, and cluster sampling study design.²⁷ Since Hispanics, African Americans, Asians, and indigent populations have been oversampled to increase statistical power and improve the precision of estimates for specific subgroups, sample weights have been provided to allow for calculation of population estimates.

Study population

This study used a retrospective cohort design. Individuals were included in this analysis if they were 18 years and older, were covered by some type of public or private insurance, and had data from all five survey rounds. We selected the 2007–2008 and 2009–2010 cohorts, two nonoverlapping MEPS-HC sequential samples, to compare the PCMH’s effects on outcome changes over time between these two cohorts.

Outcome variables

Four types of outcome measures were examined: utilization of and costs for ED and inpatient care, in both the baseline and the follow-up year for each cohort. Utilization is measured by the number of ED visits and hospital admissions. Costs are defined as payments received by the provider. In order to ensure data quality on care use and costs, the MEPS Medical Provider Component collected cost information from medical providers to supplement or replace MEPS-HC data. Given costs measured at different times covering a four-year period of time, we converted costs to the value in a single year 2009 using consumer price index to account for the price inflation.

Primary exposure: PCMH

Consistent with the key features of PCMH models, patient reports of practice characteristics and their experiences with care were used to develop robust indicators of medical home practices.²⁸ One study by Beal et al.²⁹ used the MEPS data to measure the PCMH as follows: (1) having a regular provider; (2) provider’s role in total care for the patient (i.e. new health problems, preventive healthcare, ongoing health problems, and referrals to other health professionals); (3) patient engagement in care (provider asks about medications and treatments prescribed by other doctors or asks respondent to help decide treatment); and (4) care accessibility (able to contact providers during regular business hours, at night or on weekends). We adopt the same approach to capture the main features of the PCMH, but make a minor revision by deleting the criterion “Has no difficulty contacting regular source of care over the telephone during regular business hours” to improve the face validity of PCMH models.

Consistent with the Beal et al. study, we further categorized respondents into three groups: (1) having a PCMH (those who said yes to all of these four components); (2) having a regular source of care that is not a full PCMH (heretofore, “partial medical home;” e.g. those with a regular provider, but who said “no” to any of the other three components); and (3) having no regular source of care (those who reported no regular provider), which was the reference group. In addition, we also added an unknown PCMH category that included values of “not ascertained,” “don’t know,” and “inapplicable” in any of these four components, since these unknown values cannot be used to clearly classify an individual into any other three PCMH groups, but observations within this category had valid values for other variables in the study sample. The rationale of adding this unknown PCMH group was discussed in “Statistical analysis” section. The PCMH status was measured in the baseline year of each cohort. We have provided Figure 1 to illustrate the PCMH group definition and classification process.

Figure 1.

PCMH groups definition and classification.

Control variables

Our analysis controlled for a range of factors in the baseline year known or likely to affect changes in healthcare utilization and costs, including a person’s age (18–34 years, 35–64 years, 65 years and older), gender (male and female), race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and other race), geographic area (Northeast, Midwest, South, and West), rural/urban location, marital status (married or not, including widowed, separated, never married, and divorced), education level (0–8 years, 9–12 years, 13 years, and more), income level (poor, near poor, low income, middle income, and high income), employment status (employed or not), and insurance type (public and private).

Statistical analysis

There is a potential selection issue between healthcare utilization or costs and the PCMH enrollment because sicker patients with higher demand for care utilization may be more likely to opt for the PCMH group. Differences in individuals’ utilization could reflect the combination of a causal effect of the PCMH enrollment and/or the effect of unmeasured characteristics that are correlated with the PCMH status and use of care, thus biasing the estimate of the impact of the PCMH on utilization. If the selection issue is not accounted for, it is likely to bias the estimated relationship between PCMH and our outcomes of interest. We attempted to attenuate this problem by including controls for the unknown PCMH category, health status, as well as health service costs in the baseline year in our regression model. Specifically, healthy individuals with unknown PCMH information may have a low level of ED and inpatient care use, and may not need a medical home. Inclusion of the unknown PCMH category will help account for the potential selection issue. Health status was included based on the same rationale that healthy individuals with a low level of ED and inpatient care use are likely to enroll into the PCMH group. We measured health status using the physical component score (PCS) derived from the SF-12 in the Medical Outcomes Study to reflect health-related quality of life.³⁰ In addition to health status, we also controlled two health service cost variables from the baseline year in the model to account for the selection issue. They are the costs of physician office visits and outpatient visits.

The characteristics of the count data and expenditure data were accounted for in the analysis. Measured by number of ED visits and inpatient stays, utilization data contain nonnegative count values, with a substantial number of zeros. As a random variable, smaller count values may have higher probability than larger ones. In addition, the data variance is not necessarily equal to its mean. Therefore, a negative binomial regression model, a more general form than a Poisson model, was used to account for these count data characteristics. Differences in costs were analyzed with generalized linear modeling (GLM), using gamma distribution and log link function because of its applicability to continuous variables with highly skewed distributions. The GLM is analyzed using maximum likelihood estimation.

Typically, individuals within the survey’s sampling cluster (i.e. a county or a family) are more similar to one another than those in other clusters; as a result, the error term in healthcare utilization or cost for individuals within a cluster is correlated. Failure to consider this correlation at a cluster level will result in underestimation of variance, which will overestimate the significance of the estimates. Thus, the survey procedures in STATA version 13 were used to conduct the analysis, which accounts for the complex weight and variance in the sampling design and yields nationally representative results.

Results

Study sample description

The demographic and socioeconomic characteristics of the sample used in the analysis are presented in Table 1. The study sample consisted of 15,595 individuals. Among them, there were 6659 individuals in the full PCMH group, 2786 individuals in the partial PCMH group, 3191 individuals in the unknown PCMH group, and 2959 individuals in the no regular provider group. They represented the weighted population size of 160 million, 62 million, 79 million, and 67 million, respectively, in the U.S.

Table 1.

Study sample characteristics (N = 15,595).

	Full medical home n = 6659		Partial medical home n = 2786		Unknown medical home n = 3191		No regular source of care n = 2959
Variable	Number	%	Number	%	Number	%	Number	%
Age (years)
18–34	1528	22.9	665	23.9	613	19.2	1340	45.3
35–64	3782	56.8	1502	53.9	1768	55.4	1403	47.4
≥65	1349	20.3	619	22.2	810	25.4	216	7.3
Gender
Male	2795	42.0	1140	40.9	1408	44.1	1597	54.0
Female	3864	58.0	1646	59.1	1783	55.9	1362	46.0
Race
Non-Hispanic white	3905	58.6	1486	53.3	2063	64.7	1415	47.8
Non-Hispanic African-American	1082	16.2	506	18.2	489	15.3	595	20.1
Hispanic	1053	15.8	547	19.6	374	11.7	640	21.6
Other	619	9.3	247	8.9	265	8.3	309	10.4
Geographic area
Northeast	1365	20.5	355	12.7	534	16.7	312	10.5
Midwest	1592	23.9	541	19.4	786	24.6	527	17.8
South	2053	30.8	1088	39.1	1113	34.9	1280	43.3
West	1649	24.8	802	28.8	758	23.8	840	28.4
Rural/urban location
Urban	5734	86.1	2236	80.3	2755	86.3	2547	86.1
Rural	925	13.9	550	19.7	436	13.7	412	13.9
Marital status
Married	4016	60.3	1506	54.1	1815	56.9	1468	49.6
Not married	2643	39.7	1280	45.9	1376	43.1	1491	50.4
Education level (years)
0–8	422	6.3	263	9.4	173	5.4	195	6.6
9–12	2736	41.1	1208	43.4	1314	41.2	1330	44.9
13–17	3465	52.0	1293	46.4	1673	52.4	1409	47.6
Income
Poor	777	11.7	448	16.1	322	10.1	491	16.6
Near poor	272	4.1	169	6.1	142	4.5	140	4.7
Low income	795	11.9	451	16.2	386	12.1	526	17.8
Middle income	2235	33.6	829	29.8	1063	33.3	931	31.5
High income	2580	38.7	889	31.9	1278	40.1	871	29.4
Employment status
Employed	4317	64.8	1628	58.4	1973	61.8	2264	76.5
Not employed	2334	35.1	1158	41.6	1200	37.6	691	23.4
Insurance type
Private	5160	77.5	1904	68.3	2469	77.4	2289	77.4
Public	1499	22.5	882	31.7	722	22.6	670	22.6
Health status
Mean PCS (SD)	48.2	11.2	46.5	12.1	48.4	11.2	52.3	8.3

Note:

1. The percentage was calculated using each PCMH group total as the denominator (column percent).

2. Some variables have a sum percentage slightly lower than 100% since there are a few cases whose values are “don’t know, refuse to answer, not ascertained, or inapplicable.”

The study sample had the highest proportion in the following groups across all PCMH categories: middle-aged individuals (47.4–56.8%), female (55.9–59.1%, except for the group without a regular source of care), non-Hispanic White (47.8–64.7%), living in the South (30.8–43.3%), and urban area (80.3–86.3%), employed (58.4–76.5%), married (54.1–60.3%, except for the group without a regular source of care), with college education (46.4–52.4%), middle or high income (29.4–40.1%), and private insurance (68.3–77.5%). The mean PCS, the measure for health status, was similar across all PCMH groups (46.5–52.3), with the highest score in the group without a regular source of care.

The multivariate analysis results

The 2007–2008 cohort results

The results for the 2007–2008 cohort are reported in Tables 2 and 3. In the baseline year 2007, compared to the group without a regular source of care, there was no significant difference in the visits and costs among ED and inpatient care, for any of the full, partial, and unknown PCMH group. However, in the follow-up year 2008, significantly more ED visits and higher ED costs were observed for PCMH groups: mean cost ratio = 1.59 (p = 0.021) in the ED costs for the full PCMH group, IRR = 1.47 (p = 0.003) in the ED visits, and mean cost ratio = 1.83 (p = 0.023) in the ED costs for the partial PCMH group.

Table 2.

Healthcare use and costs result summary in 2007 among the 2007–2008 cohort.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	0.91	0.92	1.30	1.10
Partial medical home	0.87	0.83	1.38	1.31
Unknown medical home	0.80	0.88	1.05	1.27

Table 3.

Healthcare use and costs result summary in 2008 among the 2007–2008 cohort.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	1.21	1.59*	1.23	1.01
Partial medical home	1.47**	1.83*	1.16	1.19
Unknown medical home	0.98	1.34	1.00	0.95

: p ≤ 0.05, **: p ≤ 0.01.

The 2009–2010 cohort results

The results for the 2009–2010 cohort are reported in Tables 4 and 5. In the baseline year 2009, compared to the group without a regular source of care, significantly more inpatient stays and ED visits and higher inpatient costs were observed for PCMH groups: IRR = 1.54 (p = 0.004) in the inpatient stays and mean cost ratio = 2.08 (p = 0.001) in the inpatient costs for the full PCMH group, IRR = 1.32 (p = 0.038) in the ED visits, IRR = 1.63 (p = 0.006) in the inpatient stays, and mean cost ratio = 2.65 (p < 0.001) in the inpatient costs and for the partial PCMH group. However, in the follow-up year 2010, there was no significant difference in the visits and costs among ED and inpatient care, for any of the full, partial, and unknown PCMH group.

Table 4.

Healthcare use and costs result summary in 2009 among the 2009–2010 cohort.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	1.08	0.76	1.54**	2.08***
Partial medical home	1.32*	1.03	1.63**	2.65***
Unknown medical home	0.90	0.78	1.07	1.32

: p ≤ 0.05, **: p ≤ 0.01, ***: p ≤ 0.001.

Table 5.

Healthcare use and costs result summary in 2010 among the 2009–2010 cohort.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	1.11	1.32	1.11	1.02
Partial medical home	1.05	1.21	1.02	1.07
Unknown medical home	1.14	1.48	1.21	1.01

We also conducted sensitivity analysis using the initial PCMH definition, with its results presented in Tables 6 to 9. The findings were essentially consistent between the initial and refined PCMH definitions: the 2007–2008 cohort had the decreasing efficiency in the trend of outcome changes for PCMH groups in some outcomes, while the 2009–2010 cohort had the increasing efficiency in the trend of outcome changes for PCMH groups in some outcomes. All these results suggested that our refined PCMH definition has generated consistent results with the initial PCMH definition, while improving PCMH face validity by deleting the criterion “Has no difficulty contacting regular source of care over the telephone during regular business hours.”

Table 6.

Healthcare use and costs result summary in 2007 among the 2007–2008 cohort using initial PCMH definition.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	0.91	0.94	1.29	1.24
Partial medical home	0.80	0.67	1.34	1.12
Unknown medical home	0.72*	0.84	0.88	0.61

: p ≤ 0.05.

Table 7.

Healthcare use and costs result summary in 2008 among the 2007–2008 cohort using initial PCMH definition.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	1.14	1.43	1.16	1.03
Partial medical home	1.68***	2.23*	1.35	1.42
Unknown medical home	1.14	1.91*	0.85	0.64

: p ≤ 0.05, ***: p ≤ 0.001.

Table 8.

Healthcare use and costs result summary in 2009 among the 2009–2010 cohort using initial PCMH definition.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	1.06	0.80	1.41*	1.77**
Partial medical home	1.41*	1.08	1.97***	3.93***
Unknown medical home	0.99	0.69	1.02	1.28

: p ≤ 0.05, **: p ≤ 0.01, ***: p ≤ 0.001.

Table 9.

Healthcare use and costs result summary in 2010 among the 2009–2010 cohort using initial PCMH definition.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	1.09	1.39	1.14	1.03
Partial medical home	1.16	1.11	1.09	1.29
Unknown medical home	1.10	1.23	0.95	0.71

Discussion

Overall, the change in utilization and costs in ED and inpatient care from the baseline year to the follow-up year has formed a contrast between the two nationally representative cohorts. In the 2007–2008 cohort, the healthcare use and costs have not been significantly different between PCMH groups and the no regular provider group at the baseline, but have become significantly higher in ED use and costs for PCMH groups during the follow-up period, suggesting the decreasing efficiency in the trend of outcome changes for PCMH groups. On the contrary, in the 2009–2010 cohort, the healthcare use and costs in PCMH groups have started at significantly higher levels than the no regular provider group, but have finished at the similar level with insignificant difference in healthcare use and costs during the follow-up period, suggesting the increasing efficiency in the trend of outcome changes for PCMH groups. The outcome trend contrast between these two sequential cohorts indicates divergent findings and may not point in any direction conclusively.

In fact, these results may be interpreted within a broad picture of PCMH transformation timelines. As Alexander and Bae point out, a central problem in PCMH-effectiveness studies is determining when, or to what degree, PCMH has been implemented.³¹ The PCMH model is a new concept with a history of just a few years since the PCMH joint principles were released in 2007. In fact, most of the large demonstrations to transform practices did not begin until 2009 and later, plus it is a lengthy and difficult transitioning process before a primary care practice can be fully functioning as a PCMH. To put things into perspective, it is expected that there are not many practices with a long history of PCMH implementation or with a high degree of transformation in the overall study sample that covers a time period from 2007 to 2010. Specifically, patients in the 2007–2008 cohort would encounter a majority of PCMH practices at the early stage of transformation, even if patients perceive them as having essential qualities of a PCMH. While patients in the 2009–2010 cohort would be expected to experience quite a number of practices that have transitioned from early stage to middle stage of transformation. Thus, varying functioning degrees of PCMH practices encountered by each cohort may explain to some extent different trends of outcome changes over time.

In the case of the 2007–2008 cohort, patients may encounter a substantial amount of early stage PCMH practices that have undergone difficulties and temporary setbacks in healthcare use and cost performance. Thus, it would not be unexpected to find that ED and inpatient care use and costs may increase to some extent or in some outcomes during the follow-up period than the baseline period. The redesign of primary care organization and delivery may slow down care efficiency or lower care quality. For example, incorporating an electronic medical record system in a primary care practice may decrease rather than improve care coordination and work flow temporarily at the transitional period. This may delay effective treatment in the primary care clinics, and thus, may increase patient downstream use of ED care or inpatient care.

It is likely that primary care practice transformation costs, such as investment on administration and coordination in practice organization and delivery, may be absorbed into the regular operational costs in ED or inpatient care services during the early transitioning period, thus increasing rather than decreasing their use and costs. After the initial trial-and-error process during this transition, the 2009–2010 cohort could encounter many more PCMH practices that have finished early stage setbacks and fluctuations and have moved to a more stable functioning stage in healthcare use and cost performance. Thus, one can find that although ED and inpatient care use and costs for the PCMH groups started at higher levels during the baseline period, they have decreased to the similar level with the comparison group during the follow-up period, suggesting that PCMH practices may have caught up and made some progress in healthcare use and cost efficiency at the middle stage of their growth.

Thus, our results may be interpreted as the effects on ED and inpatient care use and costs of the PCMH model at its different transformation stages. That is, patients visiting early stage of PCMH practices may not actually receive PCMH envisioned care since the PCMH was not being implemented as designed as highly functioning practices, but later stage of PCMH practices may have potential improvements in healthcare use and costs to some extent. Although currently it is still early for the PCMH model to reap the benefits of reduced ED and inpatient care use and costs, having more practices with longer history of PCMH implementation in the coming years will allow more opportunities for the benefits of healthcare efficiency to be fully demonstrated by the PCMH model.

A sensitivity analysis using the 2010–2011 cohort has generated the consistent results as those using the 2009–2010 cohort (Tables 10 and 11), with PCMH groups starting at higher levels in some outcomes during the baseline period, and ending with the similar level with the comparison group during the follow-up period in all outcomes. These findings suggest that although practices at the middle stage of PCMH transformation have not yet generated pronounced effects by themselves, such as lower levels of healthcare use and costs, they tend to yield steady performance in improving healthcare efficiency when comparing the trend of outcome changes with practices at their early stage of transformation.

Table 10.

Healthcare use and costs result summary in 2010 among the 2010–2011 cohort.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	1.13	1.08	1.31	1.85*
Partial medical home	1.41*	1.15	1.41	2.02*
Unknown medical home	1.18	1.45	1.07	1.10

: p ≤ 0.05.

Table 11.

Healthcare use and costs result summary in 2011 among the 2010–2011 cohort.

Variable	ED visits Incidence rate ratio	ED costs Mean costs ratio	IP stays Incidence rate ratio	IP costs Mean costs ratio
Medical home status
No regular source of care	Reference	Reference	Reference	Reference
Full medical home	0.96	1.18	1.23	1.30
Partial medical home	0.99	1.31	1.13	1.52
Unknown medical home	0.99	1.19	1.26	1.32

Current evidence suggests that exposure to a PCMH does not “flip a switch” with immediate changes in utilization or costs and healthcare system efficiency based on early evidence from our national sample, but that these effects may play out over time, in large part likely because of the lengthy and difficult transformation process. Specifically, this is a systematic reengineering process that calls upon the redesign of primary care organization and delivery and system infrastructure. The primary care practice infrastructure includes human resource assets, medical equipment, protocols, supporting systems such as a health information system, and so on that all support care organization and delivery practices such as physician–nurse coordination and collaboration, and management mechanism. The comprehensive structure and process building blocks are systematic and inherent and cannot be improved over a short period of time. Thus, as the output of all the comprehensive structure and process building blocks, healthcare system efficiency is expected to be achieved potentially after this lengthy PCMH transformation process. Although this study does not indicate whether PCMH models reduce healthcare utilization and costs, it does suggest their potential mechanisms of achieving healthcare system efficiency.

Although PCMH models have not indicated any clear effect currently, their potential mechanisms still have policy implications. Continuously rising healthcare costs have imposed great challenges to policy makers, health plans, providers, and communities. In 2010, health expenditures in the United States neared $2.6 trillion.³² Identifying strategies on costs containment have thus become more imperative. While the PCMH concept is believed to hold the potential to control costs by reducing the need for expensive healthcare services, this is one of the first studies that examines the association between PCMH and costs and utilization using national level data. The PCMH transformation is a reengineering process within primary care practices that may take time before healthcare efficiency can be improved. Patience may be needed to allow more opportunities for the potential benefits of healthcare efficiency to be realized. It is cursory to make a hasty conclusion currently as whether the PCMH model can bend the cost curve, especially given that the PCMH transformation process may be resource intensive and time consuming, and there are mixed findings in the literature. A longer observation window and holistic view on inconsistent findings at different stages of PCMH growth may be more informative to clarify this policy debate.

One particular strength of this paper is that it uses patient-reported experience to measure PCMHs. Currently, patients’ experience is widely used as a measure of healthcare quality.³³ The PCMH model, therefore, has gained its ascendency by highlighting its core features from patients’ perspective: a whole person orientation by meeting all patient needs, encouraging patients’ participation, and enhancing patients’ access to care. The PCMH model seeks to improve the continuum of personalized care from primary care or preventive care to treatment of chronic and acute illness. While healthcare organization and delivery relate to care effectiveness, patients’ values, beliefs, and circumstances also influence their expectations of, their needs for, and their use of services.³⁴ If the patient does not perceive that he is receiving the tenets underlying PCMH, then it really does not matter what the PCMH practice is doing, it is unlikely to be truly effective. If the patient does not value the PCMH, then the efficiency and performance of a PCMH model can be discounted. Thus, patients’ buy-in on PCMH models would be likely to affect their satisfaction and the extent that they interact and cooperate with providers to achieve the healthcare system efficiency, regardless of improvements in the organization and delivery of primary care practices.

This study has some limitations. First, the measurement of PCMH is constrained by data availability. The PCMH is a new concept and consists of a number of principles mentioned earlier in the paper, such as the definition developed by the Agency for Healthcare Research and Quality in 2011. Some criteria require measures from the provider side and at the system level that is beyond the scope and capacity of survey information from individual respondents, for example, coordination and/or integration of care, quality and safety as hallmarks, appropriate payment and incentives, and wide-ranging team-based care. Thus, the PCMH measure is not completely captured by the data. However, like all other studies, very few health systems or demonstration programs where the PCMH measure drew from met all the PCMH criteria due to various reality constraints. We acknowledge that the national dialog is turning to one where the PCMH is not considered a “one size fits all.” Just because each PCMH model has unique features, it does not necessarily mean that common components within unique PCMH models that vary widely across health systems and locations cannot be captured or summarized. The MEPS data is still a good source to reflect main and common features of the PCMH model, especially patients’ perspectives. PCMH measures based on these common features can still contribute to estimating this model’s effect on ED and inpatient care use and costs nationwide. Second, this study has a short observation window of only one year follow-up period, which is constrained by the two-year panel design in the MEPS data. Longer follow-up period may yield more pronounced effects. Third, the selection issue may not be thoroughly accounted for although we have controlled a set of covariates to attenuate it. Fourth, the MEPS data does not contain the information on the length of time of PCMH implementation, which cannot be reflected by the PCMH measure in the one-year baseline period. However, the PCMH transformation timeline nationwide has provided a broad picture and thus, aided our result interpretation as the effect on ED and inpatient care use and costs of the PCMH model at its different transformation stages. Future research should have better measurement of the PCMH concept, such as features reflecting integrated care and healthcare quality and safety, have a longer follow-up period and PCMH implementation duration, and thoroughly account for the selection issue to explore the effects of the PCMH model. In addition, future research can also examine uninsured populations that may benefit from the PCMH model.

In conclusion, the empirical evidence does not indicate whether PCMH models reduce healthcare utilization and costs, but it does suggest their potential as mechanisms to achieve healthcare system efficiency, when primary care practices have grown from early to middle stage of PCMH transformation. Given the lengthy and difficult PCMH transformation process, a long observation period is needed to allow more opportunities for the PCMH model before its potential benefits in healthcare efficiency can be achieved.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical approval

Not required.

References

Martin

Lassman

Whittle

. Recession contributes to slowest annual rate of increase in health spending in five decades. Health Aff (Millwood) 2011; 30: 11–22.

McGlynn

Asch

Adams

. The quality of health care delivered to adults in the United States. N Engl J Med 2003; 348: 2635–2645.

Jencks

Huff

Cuerdon

. Change in the quality of care delivered to Medicare beneficiaries, 1998–1999 to 2000–2001. JAMA 2003; 289: 305–312.

Saaddine

Cadwell

Gregg

. Improvements in diabetes processes of care and intermediate outcomes: United States, 1988–2002. Ann Intern Med 2006; 144: 465–474.

Grant

Buse

Meigs

. Quality of diabetes care in U.S. academic medical centers: low rates of medical regimen change. Diabetes Care 2005; 28: 337–442.

Nolte

McKee

. Measuring the health of nations: updating an earlier analysis. Health Aff (Millwood) 2008; 27: 58–71. Erratum appears in Health Aff (Millwood) 2008; 27: 593.

Stange

Nutting

Miller

. Defining and measuring the patient-centered medical home. J Gen Intern Med 2010; 25: 601–612.

Moreno

Peikes

Krilla

. Necessary but not sufficient: The HITECH Act and Health Information Technology’s Potential to Build Medical Homes (Prepared by Mathematica Policy Research under Contract No. HHSA290200900019I TO2.) AHRQ Publication No. 10-0080-EF, Rockville, MD: Agency for Healthcare Research and Quality, 2010.

Bodenheimer

Grumbach

Berenson

. A lifeline for primary care. N Engl J Med 2009; 360: 2693–2696.

10.

Reid

Fishman

. Patient centered medical home demonstration: a prospective, quasiexperimental, before and after evaluation. Am J Manag Care 2009; 15: e71–87.

11.

Reid

Coleman

Johnson

. The Group Health medical home at year two: cost savings, higher patient satisfaction, and less burnout for providers. Health Aff (Millwood) 2010; 29: 835–843.

12.

Rittenhouse

Shortell

. The patient-centered medical home: will it stand the test of health reform? JAMA 2009; 301: 2038–2040.

13.

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. (Evidence Reports/Technology Assessments, No. 208.2.) Available at: https://www.ncbi.nlm.nih.gov/books/NBK99094/ (accessed 23 August 2015).

14.

DeVries

Sridhar

. Impact of medical homes on quality, healthcare utilization, and costs. Am J Manag Care 2012; 18: 534–544.

15.

Jaudes

Champagne

Harden

. Expanded medical home model works for children in foster care. Child Welfare 2012; 91: 9–33.

16.

Roby

Pourat

Pirritano

. Impact of patient-centered medical home assignment on emergency room visits among uninsured patients in a county health system. Med Care Res Rev 2010; 67: 412–430.

17.

Fields

Leshen

Patel

. Analysis and commentary. Driving quality gains and cost savings through adoption of medical homes. Health Aff (Millwood) 2010; 29: 819–826.

18.

Cooley

McAllister

Sherrieb

. Improved outcomes associated with medical home implementation in pediatric primary care. Pediatrics 2009; 124: 358–364.

19.

Roby

Pourat

Pirritano

. Impact of patient-centered medical home assignment on emergency room visits among uninsured patients in a county health system. Med Care Res Rev 2010; 67: 412–430.

20.

Maeng

Davis

Tomcavage

. Improving patient experience by transforming primary care: Evidence from Geisinger’s patient-centered medical homes. Popul Health Manag 2013; 16: 157–163.

21.

Epperly

. The patient-centred medical home in the USA. J Eval Clin Pract 2011; 17: 373–375.

22.

Harbrecht

Latts

. Colorado’s patient-centered medical home pilot met numerous obstacles, yet saw results such as reduced hospital admissions. Health Aff (Millwood) 2012; 31: 2010–2017.

23.

Gilfillan

Tomcavage

Rosenthal

. Value and the medical home: effects of transformed primary care. Am J Manag Care 2010; 16: 607–614.

24.

Maeng

Graham

Graf

. Reducing long-term cost by transforming primary care: evidence from Geisinger’s medical home model. Am J Manag Care 2012; 18: 149–155.

25.

Medical Expenditure Panel Survey household component. Available at: http://www.meps.ahrq.gov/mepsweb/survey_comp/household.jsp. Rockville, MD: Agency for Healthcare Research and Quality, 2010.

26.

Agency for Healthcare Research and Quality. MEPS HC-139. Rockville, MD: Agency for Healthcare Research and Quality.

27.

Machlin

Zodet

. Computing standard errors for MEPS estimates, Rockville, MD: Agency for Healthcare Research and Quality, 2005. Available at: http://www.meps.ahrq.gov/survey_comp/standard_errors.jsp (accessed 23 August 2015).

28.

Schoen

Osborn

Doty

. Toward higher-performance health systems: adults’ health care experiences in seven countries. Health Affairs 2007; 6: w717–734.

29.

Beal

Hernandez

Doty

. Latino access to the patient-centered medical home. J Gen Intern Med 2009; 24: 514–520.

30.

Ware

Jr Sherbourne

. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care 1992; 30: 473–483.

31.

Alexander

Bae

. Does the patient-centred medical home work? A critical synthesis of research on patient-centred medical homes and patient-related outcomes. Health Serv Manage Res 2012; 25: 51–59.

32.

Martin

Lassman

Washington

. Growth in US health spending remained slow in 2010; health share of gross domestic product was unchanged from 2009. Health Aff (Millwood) 2012; 31: 208–219.

33.

Principles for the National Quality Strategy (NQS). Available at: http://www.ahrq.gov/workingforquality/nqs/principles.htm (accessed 8 August 2014).

34.

National Clinical Guideline Centre (UK). Patient experience in adult NHS Services: Improving the experience of care for people using adult NHS services: patient experience in generic terms, London: Royal College of Physicians (UK), 2012.