Currently Viewing:
The American Journal of Managed Care March 2018
False-Positive Mammography and Its Association With Health Service Use
Christine M. Gunn, PhD; Barbara Bokhour, PhD; Tracy A. Battaglia, MD, MPH; Rebecca A. Silliman, MD, PhD; and Amresh Hanchate, PhD
Incorporating Value Into Physician Payment and Patient Cost Sharing
Zirui Song, MD, PhD; Amol S. Navathe, MD, PhD; Ezekiel J. Emanuel, MD, PhD; and Kevin G. Volpp, MD, PhD
Development and Implementation of an Academic Cancer Therapy Stewardship Program
Amir S. Steinberg, MD; Anish B. Parikh, MD; Sara Kim, PharmD; Damaris Peralta-Hernandez, RPh; Talaat Aggour, BPharm; and Luis Isola, MD
Overuse and Insurance Plan Type in a Privately Insured Population
Meredith B. Rosenthal, PhD; Carrie H. Colla, PhD; Nancy E. Morden, MD; Thomas D. Sequist, MD; Alexander J. Mainor, JD; Zhonghe Li, MS; and Kevin H. Nguyen, MS
Patients Discharged From the Emergency Department After Referral for Hospitalist Admission
Christopher A. Caulfield, MD; John Stephens, MD; Zarina Sharalaya, MD; Jeffrey P. Laux, PhD; Carlton Moore, MD, MS; Daniel E. Jonas, MD, MPH; and Edmund A. Liles Jr, MD
Trends in Opioid and Nonsteroidal Anti-Inflammatory Use and Adverse Events
Veronica Fassio, PharmD; Sherrie L. Aspinall, PharmD, MSc; Xinhua Zhao, PhD; Donald R. Miller, ScD; Jasvinder A. Singh, MD, MPH; Chester B. Good, MD, MPH; and Francesca E. Cunningham, PharmD
Ambulatory Care–Sensitive Emergency Visits Among Patients With Medical Home Access
Dina Hafez, MD; Laurence F. McMahon Jr, MD, MPH; Linda Balogh, MD; Floyd John Brinley III, MD; John Crump, MD; Mark Ealovega, MD; Audrey Fan, MD; Yeong Kwok, MD; Kristen Krieger, MD; Thomas O'Connor, MD; Elisa Ostafin, MD; Heidi Reichert, MA; and Jennifer Meddings, MD, MSc
Improving Quality of Care in Oncology Through Healthcare Payment Reform
Lonnie Wen, RPh, PhD; Christine Divers, PhD; Melissa Lingohr-Smith, PhD; Jay Lin, PhD, MBA; and Scott Ramsey, MD, PhD
Currently Reading
Assessing Medical Home Mechanisms: Certification, Asthma Education, and Outcomes
Nathan D. Shippee, PhD; Michael Finch, PhD; and Douglas R. Wholey, PhD

Assessing Medical Home Mechanisms: Certification, Asthma Education, and Outcomes

Nathan D. Shippee, PhD; Michael Finch, PhD; and Douglas R. Wholey, PhD
Using statewide quality data for medical home–eligible clinics, we tested asthma education as a clinical mechanism whereby medical homes achieve better asthma outcomes.

Objectives: Patient-centered medical homes (PCMHs) represent a widespread model of healthcare transformation. Despite evidence that PCMHs can improve care quality, the mechanisms by which they improve outcomes are relatively unexamined. We aimed to assess the mechanisms linking certification as a Health Care Home (HCH), a statewide PCMH initiative, with asthma care quality and outcomes. We compared direct certification effects versus indirect clinical effects (via improved care process).

Study Design: This was an observational study using statewide patient-level data on asthma care quality and asthma outcomes.

Methods: This study examined care quality for 296,662 adults and children with asthma in 501 HCH-certified and non-HCH clinics in Minnesota from 2010 to 2013. Using endogenous treatment effects models, we assessed the effects of HCH certification on care process (patient education using asthma action plans [AAPs]) and outcomes (asthma controlled; having no exacerbations) and asthma education’s effect on outcomes. We used logistic regression to formally decompose direct (certification) versus indirect (via education/AAPs) effects.

Results: Adults’ adjusted rates of process and outcomes targets were double for HCH versus non-HCH clinics; children’s rates were also significantly higher for HCHs. Tests of the indirect/care process effect showed that rates of meeting outcomes targets were 7 to 9 times higher with education using an AAP. Decomposition indicated that the indirect effect (via education/AAPs) constituted 16% to 35% of the total HCH effect on outcomes.

Conclusions: HCHs were associated with better asthma care and outcomes. Asthma education with AAPs also was associated with better outcomes despite being a minority of HCHs’ total effect. These findings suggest that HCHs improve outcomes partially via increased care management activity, but also via other mechanisms (eg, electronic health records, registries).

Am J Manag Care. 2018;24(3):e79-e85
Takeaway Points
  • We used statewide quality reporting data in medical home and non–medical home clinics to assess whether education using asthma action plans (AAPs) is a mechanism linking medical home certification to better asthma control and lower risk of exacerbation.
  • Using endogenous treatment effects models, we confirmed that medical homes and education using AAPs are associated with better outcomes. We then decomposed medical home effects via clinical (education/AAP) versus other mechanisms, and we found that patient education using AAPs accounted for 16% to 35% of medical homes’ association with outcomes.
  • Healthcare redesign requires both clinical and nonclinical transformation to improve outcomes. These findings also support patient education using AAPs.
Patient-centered medical homes (PCMHs) represent a type of payment and delivery system reform that emphasizes improved access, primary care–based teams, and care continuity and coordination.1,2 Evidence suggests that medical homes may be associated with higher quality of care and lower resource use and costs.3-11 However, although certain medical home components—including electronic health record (EHR) adoption, team-based care, financial incentives, state support, and others—may improve care,12-14 the mechanisms by which medical homes may result in better outcomes remain untested.

Two mechanisms that might link medical homes to better asthma outcomes are: 1) organizational transformation, including workforce management, team structure, and use of registries or EHRs to track patient populations,12,14,15 and 2) patient-centered care management.16 Although the former might require qualitative methods or provider surveys to assess, care management is observable via certain process measures in secondary data, making it amenable to research in understanding medical homes’ mechanisms.

One condition with evidence of benefit in medical homes is asthma,9,17 which afflicts more than 25 million people in the United States.18 Beyond its symptom burden, the condition is associated with 1.75 million visits to the emergency department (ED) and more than 400,000 hospitalizations annually.19,20 Proper asthma care includes routine follow-up visits and specialty referrals as needed, assessment and self-monitoring, addressing environmental irritants and triggers, pharmacotherapy, and patient–provider partnerships in striving for treatment goals and asthma control.21

Within asthma care, a notable care management tool is asthma education by a clinician, often including an asthma action plan (AAP), which is a written or electronic plan focusing on daily treatment, early detection, crisis management, and treatment of asthma exacerbations. AAPs are a standard of care recommended by several scientific and professional organizations.21 Although highly variable,22,23 AAPs have evidence of benefit in improving provider counseling and patient outcomes, particularly as part of comprehensive asthma education.23-30 As such, education using an AAP serves as a process measure for asthma quality, although AAP use remains low.31,32 Education with an AAP may represent a mechanism of coordinated care and self-management linking medical homes with better outcomes. Yet, medical home–based improvements in asthma outcomes, if present, are likely not solely due to care process via education and AAPs; they also likely reflect organizational and structural changes, including primary care physician–led teams, improved access, and population assessment and patient tracking.12-14

In this study, we used statewide clinic-reported patient-level data on adult and pediatric patients with asthma at primary care clinics certified as Health Care Homes (HCHs), versus those not HCH-certified, under a statewide initiative. Our research questions were: 1) what are the effects of being in HCH versus non-HCH clinics on asthma care quality process (education/AAP) and outcomes measures (having asthma controlled; being free from exacerbations)?; 2) is education with an AAP associated with better asthma outcomes?; and 3) what are the direct (via HCH certification) and indirect (via education/AAP) effects on asthma outcomes, and how do they compare?


This was an observational study; clinics self-selected into HCH certification. The study was approved by the University of Minnesota Institutional Review Board.

Health Care Home Initiative

The Minnesota HCH initiative was part of a larger 2008 health reform law.9 Minnesota developed its own HCH certification process for primary care clinics. Certification, conducted by the Minnesota Department of Health, is based on 5 standards: access/communication, participant registry/tracking participant care activity, care coordination, care plan, and performance reporting/quality improvement.
The HCH initiative included payment reform via additional reimbursement for care coordination, based on tiers of patient complexity. However, financial incentives do not appear to be a primary reason for certification: In a 2012 survey, most HCH clinics had not conducted a financial analysis before seeking certification.9,33


Data came from the HCH certification database and MN Community Measurement (MNCM). Under the Statewide Quality Reporting and Measurement System (SQRMS), data are collected by clinics and submitted directly to MNCM, the measures developer and data steward. These data include records about patients from all payers and insurance types, including commercial insurance, Minnesota Health Care Plans (MHCP; Minnesota’s Medicaid/SCHIP and low-income insurance programs), Medicare, self-pay, and uninsured.

SQRMS is a separate initiative from HCH; participation in SQRMS does not mean a clinic will seek HCH certification, and certification does not require high quality scores. Incentives for SQRMS participation are separate from those for HCH (eg, Minnesota’s Quality Incentive Payment System includes payments for meeting quality targets or improvement over time).


Only primary care clinics that were located in Minnesota and eligible for HCH certification were included. Based on available years and data for quality measures, our sample (depending on the year and measure) consisted of between 64,000 and 80,000 patients in more than 500 HCH and non-HCH clinics reporting data on asthma quality from 2010 to 2013. Data from 2009 were not used for quality measurement (no clinics were HCH-certified in 2009), but they were used to estimate self-selection into HCH certification.


“HCH certification” is a yearly indicator of whether a patient’s clinic was HCH-certified for at least part of that year (“HCH”) or was an HCH-eligible primary care clinic that either chose not to become certified or had not yet completed certification during the year (“non-HCH”). A clinic could be non-HCH one year and certified the next. Under SQRMS, clinics themselves reported enrollee-level data for their own patients, meaning that patients were attributed directly by the clinics who cared for them, and submitted quality measures based on their EHRs.

Asthma quality data were available from 2010 to 2013. They consisted of a process measure and 2 outcomes measures, based on the observation period of July 1 of the previous year to June 30 of the current year. The process measure pertained to having asthma education and AAP in the patient’s chart at any time during the observation period. The AAP contained information on medication doses and purposes of medications, information on recognizing exacerbations and what to do when they occurred, and information on patient triggers. For outcomes measures, having “asthma well controlled” came from the most recent asthma control tool during the observation period: for those 17 years and older, the latest Asthma Control Test or Childhood Asthma Control test score of 20 or greater or the latest Asthma Control Questionnaire score of 0.75 or lower; for those younger than 17 years, the latest Asthma Therapy Assessment Questionnaire score of 0. “Not having elevated risk of exacerbation” indicated having less than 2 inpatient hospitalizations or ED visits summed across the observation period.

Patient covariates included sex, insurance (commercial, Medicare, MHCP, self-pay/uninsured), distance to clinic, and child/adult. Clinic characteristics, including rurality, came from the HCH certification database. For rurality, we classified zip codes based on rural–urban commuting areas34: urban/metropolitan, micropolitan, small town, and rural.


We conducted analyses of the total, direct, and indirect effects of HCHs and AAPs using the endogenous treatment effects (eteffects) procedure in Stata 1435 (StataCorp LP; College Station, Texas) to map direct and indirect effects and the ldecomp procedure in Stata to formally decompose total, direct, and indirect effects of HCHs and AAPs, respectively.36

Eteffects models use a probit model for treatment (treatment model), whose residuals are then used in models predicting outcomes (outcomes models) for treated and nontreated groups. This was done to help address endogeneity and ensure that there were comparable observations between “treated” and “nontreated” groups (the overlap condition). We employed 2 sets of eteffects models. The probit treatment models, where treatments were 1) the direct effect (HCH as treatment) and 2) the indirect effect (education/AAP as treatment), included clinic rurality, square root of patient volume, clinic membership in a medical group, and a selection bias correction term (see below). Outcomes models for both were probit models predicting asthma outcomes (not being at risk of exacerbation and having asthma well controlled), controlling for patient sex, insurance, and distance to clinic (same zip code, <10 miles, 10-20 miles, >20 miles).

Treatment models for the eteffects procedure employed an inverse Mills ratio to adjust for the fact that clinics self-selected into applying for HCH certification.37 This selection term was calculated yearly, because certification, variables associated with certification, and their effects on certification could vary across years. The selection term was calculated from a model estimating the probability, each year, that a clinic would be HCH-certified during the year, as regressed on clinic-level measures from the prior year, including average rates of meeting diabetes care quality and vascular care quality targets for each clinic, number of providers, number of patients, membership in a medical group with 10 or more clinics, and proportion of patients insured by MHCP, Medicare, or uninsured/self-pay. The selection term was imputed for clinics that had not participated in quality reporting in the prior year via regression-adjusted imputation, with an imputed-lambda indicator included in the models.

Based on the eteffects models, we present adjusted potential outcomes means for probabilities of, separately, HCH (direct) and AAP (indirect) effects on asthma outcomes. To formally decompose the direct and indirect (via AAPs) effects of HCH on asthma outcomes, we applied the ldecomp procedure in Stata,36 used for decomposing such effects of categorical variables in logistic regression models, with standard errors obtained via the bootstrap method. Based on this analysis, we present computed odds ratios (ORs) with 95% CIs, as well as log ORs (where direct and indirect effects sum to total effect), allowing us to present direct/indirect effects as percentages of the total effect.

Copyright AJMC 2006-2020 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up