Fadia T. Shaya, PhD, MPH

 Patients are vulnerable to fragmentation of care when transitioning from in-patient settings to ambulatory care offices. Primary care teams can support care transitions by enhancing self-management capability, explaining instructions provided by the hospital staff, addressing lack of patient ability to self-manage chronic disease, and providing resources to contact their primary care physicians. This study explores the link between patient-centered medical home (PCMH) team functioning and the impact of physician leadership on care transitions.

The Maryland Multi-Payor Program includes 52 community-based PCMHs. The Team Perceptions Questionnaire (TPQ) and Care Transitions Survey (CTS) were distributed to 36 parent practices in the program. Of these, 26 practices provided complete data on the surveys.

 We observed that positive responses on the TPQ were associated with positive responses to the questions on the treatment and management domain of the CTS (average association effect ranging from 0.24 to 0.35) and in the patient-centered communications and education domain, with average association effects of 0.52 and 0.57, respectively. Physician leadership had a significant impact on team functioning and on care transitions.

 PCMHs with high scores on the TPQ have improved care transitions functioning, specifically in the treatment and management of patients, and have a greater likelihood of impacting the overall costs of care. Healthcare reform efforts to develop integrated care transitions teams along with PCMHs and hospitals/long-term care facilities are likely to lead to enhanced teamwork and more seamless transitions for patients that have the potential for cost savings, higher quality of care, and greater satisfaction for both patients and providersBecause outcomes have not necessarily improved as costs of care have risen in the United States, we must revisit primary care delivery. Patient-centered medical homes (PCMHs) are potentially the foundational building blocks for integrated systems of care and the anchors for population health,

and the adoption of PCMH principles, including team functioning and care management, can lead to enhanced access to care, comprehensive chronic disease management, and patient tracking.

Patient-centered care delivery has demonstrated early positive impact on the quality of care delivery, modest reduction in costs, and enhanced patient and provider satisfaction.

However, in the current healthcare system, patients transitioning from in-patient to ambulatory facilities and vice versa are at high risk for fragmentation of care.

PCMH teamwork is facilitated by leadership, role definition and training of all team members, shared goals, good communication, and measurable outcomes.

 PCMHs in the Maryland Multi-Payor Program (MMPP) have primary care teams with goals designed to prevent fragmentation of care for patients who are transitioning from in-patient settings to ambulatory care offices. These goals include increasing self-management capability, fostering comprehension of instructions provided by hospital staff, addressing sparse resources, and ensuring ways to contact their primary care physician.

 Patient-centered primary care teams with physician leadership may be well positioned to provide an efficient method to transition patients to and from the ambulatory arena.

This study explores the link between PCMH team functioning and the impact of physician leadership on care transitions.

The MMPP for PCMHs was established by the Maryland Health Care Commission, pursuant to Maryland Legislative Resolution HB929/SB855 enacted in April 2010.

 The MMPP established the Maryland Learning Collaborative (MLC; sometimes known as the Maryland Health Care Innovations Collaborative), housed in the Department of Family and Community Medicine of the University of Maryland School of Medicine, to provide educational and logistic support to transform primary care practices to PCMHs and for implementation of the advanced primary care model.

 The 5 commercial insurance carriers participating in the program are Aetna, CareFirst, Cigna, Coventry, and UnitedHealthcare; in addition, public insurers Medicaid and Tricare provide fixed transformation payments toward the advanced primary care model. One-third of fixed transformation payments are dedicated to the development of embedded care management teams that can provide comprehensive, coordinated primary care.

The MMPP includes 52 practices, representing a mix of rural, semi-rural, urban, and suburban practices. The 52 practices have 36 parent organizations, each with 1 to 4 practice sites. Practices are statewide, are diverse in populations served, and range in size; they include hospital-owned, practitioner-owned, and academic practices, as well as federally qualified health centers, and each practice has transformed into a PCMH recognized by the National Committee for Quality Assurance (NCQA). In total, the practices include 339 practitioners: 277 physicians, 40 nurse practitioners, and 22 physician assistants. The MLC provides PCMH teams with education and teamwork training designed to develop a team process that includes an embedded care manager and is based on the acquisition of practice-specific data to stratify patients by disease using registries or data from the state-designated Health Information Exchange (SDHIE) about utilization of hospitals and emergency departments (EDs).

There are 52 primary care practice sites with 36 parent practices, caring for 250,000 attributed patients. Payment structures include fixed transformation payments and quality-linked shared savings. All 52 practices are NCQA-recognized at Level 2/3, use electronic health records (EHRs), and utilize their EHR registry function; additionally, 77% are linked to the SDHIE. Most practices are linked to their local hospital discharge teams; they receive daily data from the hospitals and from the SDHIE regarding admissions/transfers/discharges for their patients.

The MLC provides technical assistance to the PCMHs regarding teamwork, evidence-based medicine adoption, dissemination of patient-centered outcomes research, and quality improvement. It also coordinates support for health information technology through the regional extension center.

Each of the participating practices in the MMPP had received coaching and learning collaborative participation in teamwork; each was also working with an embedded care manager. In all practices that responded, primary care teams were led by physicians. This study’s data sources were the Teamwork Perception Questionnaire (TPQ) and Care Transitions Survey (CTS) (see 

 The SurveyMonkey tool, which included the Care Transitions Survey and Team Perceptions Questionnaire, were e-mailed to the 36 parent practices and followed by reminder e-mails, phone calls, and mailed notes asking all participants to complete the surveys. In total, 27 practices responded. One person from each practice completed both the TPQ and CTS, and that individual was required to consult with other members of the team. Data was gathered anonymously using SurveyMonkey.

The TPQ consists of 35 Likert scale questions, with 7 in each of these domains: team structure, leadership, situation monitoring, mutual support, and communication. Response choices ranged from (1) 

. The CTS includes 17 questions—each with 3 response choices: works well, works somewhat, works poorly&mdash;in 3 domains: care coordination; treatment and management; and patient-centered communication and education.

In preliminary analysis, we identified the proportions of agree/strongly agree responses to the TPQ and works well responses to the CTS. Using univariate ordinary least squares regression with robust standard errors, we modeled the association between the scores on the CTS and the average score for the TPQ domains. We performed the modeling individually for each of the 17 questions from the CTS as well as for when its questions were grouped in the 3 domains. The average scores for the domains of the TPQ were calculated by assigning a score for the responses to each of the domain component questions (

 = 5) and taking the average. Similarly, the responses to the CTS were given a numeric score: 

 = 3. Multivariate analyses were not undertaken due to the high degree of multicollinearity between the domains of the TPQ and CTS tools.

The MMPP had 52 practice locations and included 36 parent practices. A survey was sent to each practice and to each parent company. We received responses from 27 of 36 parent practices (75%); 26 surveys were complete and used in analysis. Each survey returned by a parent practice was completed by a primary care team member—usually a care manager in consultation with other team members. The majority of the questions on the TPQ had a ≥72% response rate of either 

The distribution of the responses to the CTS was much more varied (

). Questions in the treatment and management domain were marked as 

 about 70% of the time. Practices reported high performance on CTS domain areas under medication reconciliation. Responses to questions in the patient-centered communication and education domain received a 

 response less than 50% of the time; for the questions in the care coordination domain, this response rate was about 60%.

In correlating the survey results from the practices in the TPQ and CTS domain areas, we observed that on average, positive responses on the TPQ were associated with positive responses on the questions from the treatment and management domain of the CTS (average association effect ranging from 0.24 to 0.35; 

). Teamwork in a PCMH appeared to have the highest impact on the management of patients in a patient-centered manner. The highest association effect with teamwork was 0.61 in medication reconciliation and communication to the patient/family.

As reported by each responding PCMH parent practices, high scores on the TPQ in teamwork and physician leadership were significantly associated with the CTS domains of care coordination, patient-centered communications, and patient education (Table 3). Specifically, the scores for TPQ had a high association effect (0.46 and 0.45, respectively) with these questions in the care coordination domain: “Knowing when patients have visited the ED, been admitted to the hospital, are going to be discharged and have been discharged” and “Reviewing discharge summaries prior to patient visit.” Similarly, positive responses on the TPQ were associated with positive scores on the CTS questions in the patient-centered communication and education domain: “Eliciting patient goals for post discharge visit” and “Educating patients to recognize and respond appropriately to warning signs/red flags using teach back,” with average association effects of 0.52 and 0.57, respectively.

Advanced models of care require redesign of primary care work flows, reorganization of work responsibilities, and empowering the patient to self-manage health.

We found that stronger physician leadership responses to the TPQ correlated specifically with the following responses as given in the CTS: stronger team knowledge of patient care transitions; effective inclusion of the patient as a member of the team by eliciting patient goals and providing education on self-management; the efficient provision of a reconciled medication list for the patient to family members, caregivers, and home health nurses; and more direct physician role in ordering tests and assessing and adjusting medications.

 Medication management in particular is a mainstay of reorganized patient care transitions, and it is an area in which patient-centered care with physician-led teams is demonstrating an impact.

All the PCMH parent practices responding to this survey received real-time data from the SDHIE regarding admissions/discharges and transfers of their patients from EDs and hospitals.

 In addition, primary care teams managed patients on the Encounter Notification Systems list from the SDHIE. The TPQ and CTS responses indicated that there was greatest team awareness on practice knowledge of patient discharge from EDs and hospitals, on eliciting patient goals at discharge, on patient self-management, and on medication reconciliation.

The data from the 26 PCMHs present an interesting dichotomy of the impact of teamwork in a PCMH on care coordination and on patient-centered communication. Although the reorganization of a primary care practice using practice transformation to a PCMH is designed to improve care coordination and patient communication within the PCMH, the impact of such reorganization is not immediately apparent in care transitions. Care transitions focus on patients who are utilizing hospitals and EDs, who are de facto high utilizers of healthcare dollars. Included among these high utilizers are patients with end-stage disease and chronic disease and those with sub-optimal social supports, for whom a PCMH may not be effective as the sole care-transition strategy.

 Patients transitioning to and from hospitals and EDs may benefit from wrap-around services supported by the hospitals, the public health infrastructure, and community resource linkages.

Thus, a PCMH with enhanced communication with hospital discharge teams, and with care coordination supports from local hospitals, has a greater likelihood of success in care transitions. In addition, patients who are repeat ED users demonstrate a greater correlation with social determinants of health, and these patients are most likely to change their behaviors with a concerted and coordinated effort by both the PCMH and the ED.

While this study is relevant to the field of care transformation and new models of care delivery, it has some limitations. The sample size is small and a single member of the team may have responded on behalf of all team members. As with all self-reported data, there may have been some test-retest bias, and the data were not audited. Further, multivariate analyses were not undertaken due to the high degree of multicollinearity between the domains of the TPQ and CTS tools.

Primary care team structure was demonstrated to have the highest impact of all domains on enhanced care transitions. PCMHs with high scores on the TPQ have improved care transitions functioning and greater likelihood of impacting the overall cost of care. Increased TPQ scores and higher care transitions capability are linked to shared savings.

The PCMH has the potential to become the ambulatory care coordination center for the healthcare continuum with support from hospitals, long-term care facilities, and community-based resources.

PCMHs are also positioned to enhance patient self-management, and have the capability to provide ambulatory supports for optimal utilization of the healthcare system.

Healthcare reform efforts, particularly the development of integrated care transitions teams with hospitals, long-term care facilities, and PCMHs, can lead to optimal teamwork and seamless transitions. This scenario would offer the potential for cost savings, higher quality of care, and greater satisfaction for both patients and providers.

Family and Community Medicine, University of Maryland School of Medicine (NK), Baltimore, MD; Pharmaceutical Health Services Research, University of Maryland School of Pharmacy (FTS, VVC), Baltimore, MD; Department of Health and Mental Hygiene, Center for Health Information Technology and Innovative Care Delivery (DS), Baltimore, MD; Department of Health and Mental Hygiene, Maryland Health Care Commission (BS), Baltimore, MD.

The Maryland Health Care Commission (MHCC) established the Maryland Multi-Payor Patient Centered Medical Home Program in 2011 as required by Health-General Article §19-1A-01, et sequentes. The law required MHCC to develop a 3-year pilot, which aims to improve the health and satisfaction of patients and slow the growth of healthcare costs in Maryland, while supporting the satisfaction and financial viability of primary care providers in the State. The 5 largest commercial insurance carriers—CareFirst BlueCross BlueShield, United HealthCare, Aetna, Coventry, and CIGNA&mdash;are required under the enabling legislation to participate. TRICARE, a healthcare program for uniformed services members, also participates in the program. Maryland‘s Medicaid agency, the Health Care Financing Administration, supports reimbursement to Medicaid Managed Care Organizations (MCOs) subject to existing budget constraints.

 The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. Dr Khanna takes full responsibility for the work as a whole, including study design, access to data, and the decision to submit and publish the manuscript. All authors have made a substantial contribution to the research design, the acquisition, analysis or interpretation of data; have drafted the paper and revised it critically; and have approved the final version of this article.

Concept and design (NK, BS, DS); acquisition of data (NK, FTS, DS); analysis and interpretation of data (NK, FTS, VVC); drafting of the manuscript (NK, VVC); critical revision of the manuscript for important intellectual content (NK, FTS, DS). Address correspondence to: Fadia T. Shaya, PhD, MPH, University of Maryland School of Pharmacy, 220 Arch St, 12th Fl, Rm 01-204, Baltimore, MD 21201. E-mail: 

1. Developing and running a primary care practice facilitation program. Agency for Healthcare Research and Quality website. 

http://pcmh.ahrq.gov/sites/default/files/attachments/ContextualFactors.pdf

. Published December 2011. Accessed September 12, 2014.

2. Grumbach K, Bodenheimer T. Can health care teams improve primary care practice? 

3. Bodenheimer T. Lessons from the trenches--a high-functioning primary care clinic. 

4. Naylor MD. Transitional care for older adults: a cost-effective model. 

5. Mitchell P, Wynia M, Golden R, et al. Core principles & values of effective team-based health care. National Area Health Education Center Organization website. https://www.nationalahec.org/pdfs/VSRT-Team-Based-Care-Principles-values.pdf. Published October 2012. Accessed December 23, 2014.

6. Bodenheimer T, Laing BY. The teamlet model of primary care. 

7. Rojas Smith L, Ashok M, Morss Dy S, Wines C, Teixeira-Poit S. Contextual frameworks for research on the implementation of complex system interventions. AHRQ publication No. 14-EHC014-EF. Rockville, MD: Agency for Healthcare Research and Quality. 

http://effectivehealthcare.ahrq.gov/ehc/products/490/1882/contextual-frameworks-complex-interventions-report-140318.pdf

. Published March 2014. Accessed May 2015.

8. Stange KC, Glasgow RE. Contextual factors: the importance of considering and reporting on context in research on the patient-centered medical home. AHRQ publication No. 13-0045-EF. Rockville, MD: Agency for Healthcare Research and Quality. 

http://www.pcmh.ahrq.gov/sites/default/files/attachments/ContextualFactors.pdf

. Published June 2013. Accessed May 2015.

9. Maryland multi-payor patient centered medical home program. Maryland state government website. 

10. House bill 929. General Assembly of Maryland website. 

http://mlis.state.md.us/2010rs/bills/hb/hb0929t.pdf

. Published 2010. Accessed December 23, 2013.

11. Chapter 5 (Senate Bill 855) Patient Centered Medical Home Program. General Assembly of Maryland website. http://mlis.state.md.us/2010rs/chapters_noln/Ch_5_sb0855T.pdf. Published 2010. Accessed December 23, 2013.

12. Maryland Learning Collaborative: Multi-Payer Program for Patient Centered Medical Home. University of Maryland School of Medicine website. 

http://medschool.umaryland.edu/familymedicine/mdlearning/

13. Khanna N. Community tailored partnerships that work: implementing new models of primary care in the state of Maryland. 

14. Bailit M, Hughes C, Burns M, Freedman DH; Bailit Health Purchasing, LLC. Shared-Savings Payment Arrangements In Health Care: Six Case Studies. New York, NY: The Commonwealth Fund; 2012.

15. Khanna N, Shaya F, Chirikov V, Steffen B, Sharp D. Dissemination and adoption of the advanced primary care model in the Maryland multi-payer patient centered medical home program. 

16. Furukawa MF, King J, Patel V, Hsaio CJ, Adler-Milstein J, Jha AK. Despite substantial progress in EHR adoption, health information exchange and patient engagement remain low in office settings. 

17. TeamSTEPPS Teamwork Perceptions Questionnaire (T-TPQ) Manual. Agency for Healthcare Research and Quality website. 

http://teamstepps.ahrq.gov/Teamwork_Perception_Questionnaire.pdf

. Published June 28, 2010. Accessed September 12, 2014.

Articles

Physician-led patient care teams have the potential to impact care transitions to prevent fragmentation of care, and ensure seamless care delivery.

Patient-Centered Teamwork in Care Transitions

To examine comorbidity and therapy use rates in a Medicaid population with heart failure (HF), evaluate hospitalization risk as a function of comorbidity and therapy use, and assess the impact of modification on costs to the Medicaid program.

Historical prospective cohort study. Claims were from adult enrollees in Maryland Medicaid (managed care organization or fee-for-service) diagnosed with HF between 2005 and 2009.

The end point was first hospitalization after index HF. Average hazard ratios (HRs) were estimated by multivariate weighted Cox regression. Budget impact of modifications was assessed using annual number-needed-to-treat calculations and external estimate of average cost of HF hospitalization.

Most patients were >45 years (71%), women (56%), and black (60%). Medication use: beta-blockers (26%), angiotensin-converting enzyme inhibitors and/or angiotensin II receptor antagonists (ACEi/ARBs) (29%), aldosterone antagonists (5%), and others including nitrates-hydralazine (37%). Nearly all (98%) were diagnosed with 1 or more comorbidities. Relative risk of hospitalization was higher with most, but not all, comorbidities investigated. ACEi/ARBs (HR, 0.77; CI, 0.73-0.81), beta-blockers (HR, 0.83; CI, 0.79-0.87), and other cardiovascular drugs (HR, 0.76; CI, 0.72-0.80) had beneficial effects. A 20% increase in the use prevalence of ACEi/ARBs and beta-blockers translated to annual Medicaid savings of at least $85 and $57 per HF patient, respectively.

Findings call attention to comorbidities and optimization of disease-modifying therapy in Medicaid patients with HF. Certain disease-modifying medications mitigated risk, but were used infrequently. Substantive outcome improvement and savings to Medicaid may be achieved with small changes in prescribing rates or comorbidity prevalence.

The projected increase in heart failure prevalence coupled with greater numbers of patients presenting with comorbidities are likely to place greater strain on state Medicaid budgets and disease management programs.

		This study elucidates the impact of comorbidities on the risk of hospitalization among patients with heart failure.

		Certain disease-modifying medications mitigated the risk of hospitalization, but were used infrequently.

		Substantive outcome improvement and savings to Medicaid may be achieved with small changes in prescribing rates or comorbidity prevalence.

Heart failure (HF) was estimated to affect nearly 5.7 million Americans in 2008,

 and its prevalence has been projected to rise by 25% through 2030.

 This rise in prevalence is accompanied by a projected 215% increase in direct medical costs over the next 20 years in the United States, from approximately $24.7 to $95.6 billion (2008 US dollars).

It was estimated that most (77%) medical costs following diagnosis of HF accrue during hospitalization.

While hospitalizations are frequent among HF patients, occurring about once per year, only one-third had HF as the primary admitting diagnosis.

The clinical and corresponding economic burden of HF is often compounded by concurrent morbidities (eg, chronic obstructive pulmonary disease [COPD], renal dysfunction, psychological disorders, and stroke, among others).

 Little is known on how comorbidities may affect the risk of hospitalization among patients with HF—specifically patients in expanding Medicaid plans, which have typically shown high prevalence of HF.

 The literature has not cited many population-based studies with real-world data, which may stand to inform disease management. Much of the evidence based on clinical trial populations is invaluable in clinical practice; however, it is often difficult to interpret and apply in populations whose demographics and risk-factor profiles vary from those cited.

A case in point is the Medicaid population.

Recent healthcare policy changes in the American Recovery and Reinvestment Act and the Affordable Care Act have relegated more responsibilities and expanded funding to states themselves, resulting in increased enrollment in Medicaid plans.

 For instance, Medicaid enrollment grew by 9 million over the past year, increasing total enrollment to 66.5 million Americans as of October 2014, or 21.2% of the total population.

The projected increase in HF prevalence coupled with greater numbers of patients presenting with comorbidities are likely to place greater strain on state budgets and disease management programs.

 Given the high risk profile among Medicaid patients,

increasing enrollment in state Medicaid programs, sparse literature on population-based HF studies, and the burden of hospitalization among HF patients, we investigated risk factors for HF in a contemporary Medicaid population and developed a risk score for hospitalization associated with specific medication use and comorbidities. Finally, we assessed the impact of a potential modification of risk factors on hospitalization and cost.

This study was a historical-prospective cohort study sampling from the population of Maryland Medicaid recipients enrolled in 1 of 7 prepaid state-contracted managed care organizations (MCOs) or fee-for-service (FFS) programs. Patients between the ages of 18 and 64 years diagnosed with HF between July 1, 2005, and December 31, 2009, were enrolled in the study. We obtained all of their encounter and prescription data, and followed them from their date of enrollment through first hospitalization to end of study (June 20, 2010), or the date of disenrollment as recorded in their last Medicaid MCO or FFS encounter claim. The study design allowed for at least 6 months and for up to 5 years of follow-up post diagnosis.

We recorded demographics (age, sex, race), dates of service (physician visits, hospitalizations, medication dispensing), and primary through tertiary 

International Classification of Diseases, Ninth Revision, Clinical Modification

) codes. Pharmacy claims also included American Society of Health-System Pharmacists Pharmacologic-Therapeutic Classification (AHFS) numbers.

The date of diagnosis for HF was defined as the earliest physician visit associated with 

 code 428.xx. In addition to this prevalence cohort, we built an incidence cohort to include all patients whose first HF claim was at least 6 months after their first Medicaid claim. Dates of hospitalization were derived from the first date of service on records of inpatient hospital encounters. Less than half (47%) of HF diagnoses coincided with patients' first hospitalization, in which case a second hospitalization was used as the end of follow-up. Of 24,635 HF patients identified, the final prevalence cohort included 14,149 observations and the incidence cohort included 7470 patients.

Systolic HF versus HF with preserved ejection fraction (HFpEF) was denoted if specified by secondary coding (428.2x vs 428.3x, or 428.40-428.43 combined). Although systolic HF and HFpEF share the same clinical phenotype, they are thought to differ in pathophysiology and thus tend to respond differently to treatment.

 Medications that have been shown in clinical trials to produce unequivocal improvements in systolic HF have not produced similar effects in HFpEF, and the optimal treatment strategy for HFpEF has yet to be defined.

 The type of HF was specified in only 5% of our sample; consequently, we could not distinguish which patients were considered most susceptible to disease-modifying therapy. Since it is known that about half of all patients with HF display HFpEF and the remainder systolic HF,

there is little reason to suspect that HF was unspecified on claims for 95% of our sample in any systematic fashion. All patients were retained, regardless of specified or unspecified HF.

Covariates included first-line disease-modifying therapies (guideline-recommended), comorbidities, age, race/ethnicity, gender, and an indicator for HF diagnosis during hospitalization. First-line therapies were ascertained using pharmacy claims with drugs dispensed between HF diagnosis and the end of follow-up. Therapies examined include ACEi/ARBs (AHFS 24:32.04 or 24:32.08); betablockers (AHFS 24:24); aldosterone receptor antagonists (AAs) (AHFS 24:32.20); nitrates + hydralazine combination (AHFS 24:08.20 or 24:08.20 with 24:12.08); “other” cardiovascular therapies, including nitrates or hydralazine alone; and all medications with AHFS 24:xx but not described above. Drugs do not include medications that are not reimbursed by Medicaid.

Diagnoses of comorbidities were determined from 3-digit 

 codes recorded in the primary through tertiary fields of claims for medical encounters between the earliest claim and within 3 months after HF diagnosis. Comorbidities include COPD, stroke, renal dysfunction, diabetes, psychological disorder(s), hyperlipidemia, hypertension, and chronic ischemic heart disease (

: 410.xx-414.xx, which includes past myocardial infarction, angina pectoris, and other forms of ischemic heart disease). An indicator for “other” cardiovascular disease captures all ICD-9-CM diagnosis codes between 390.xx and 459.xx, but not hypertension, ischemic heart disease, HF, or stroke. Classification of therapies and comorbidities is described in 

). Age (18-44 years, 45-54 years, and 55-64 years), gender, and race/ethnicity (white, black, Hispanic, and other) were identified at HF diagnosis.

We report the prevalence of comorbidities in the prevalence cohort as well as cross-tabulations of these comorbidities (

). Demographic distributions are examined in the prevalence cohort and across comorbidities. Kaplan-Meier estimates were used to estimate median days until hospitalization after HF diagnosis.

We built survival analysis models to assess the impact of first-line use of disease-modifying therapies with HF patients on their risk of hospitalization. We estimated the risk of hospitalization, adjusting for demographic risk factors and the presence of comorbidities. We also adjusted for the diagnosis of 1 or more comorbidities after 3 months of HF diagnosis, since new or undiagnosed morbidity might have confounded estimates for the risk of hospitalization over follow-up.

Weighted Cox regression (WCR) was used to model the average hazard ratios (HRs) of factors for hospitalization after HF diagnosis. WCR provided a simplified method for estimating HRs adjusted for covariates and averaged over the follow-up period, regardless of whether risk varies over time (ie, nonproportional hazards).

WCR is described in the eAppendix and was implemented using SAS version 9.2 (SAS Institute, Cary, North Carolina) macro program, WCM, provided on the Web by Heinze.

Using the WCR estimates, we computed the budget impact to the Medicaid plan, given a 20% increase in the prevalence of comorbidity or therapy use in the HF cohort. The mean cost of a primary HF hospitalization among nondually enrolled Maryland Medicaid patients in 2011 was $16,341.

Of note, average hospitalization costs have been estimated to be up to 43% higher when HF was a secondary, rather than the primary, diagnosis for hospitalization.

The numbers-needed-to-treat approach was adopted to determine the number of patients with comorbidity or using a first-line therapy associated with at least 1 hospitalization per year. We calculated the savings per patient, as attributed to the expected impact on hospitalization rates.

Patient characteristics and hospitalization rates for the HF sample and comorbidity cohorts are reported in 

. Most patients were 45 years or older (71%), women (56%), and black (60%); 33% were white; and Hispanics (2%) and other races (6%) represented a small portion. Nearly all HF patients (98%) were diagnosed with 1 or more other conditions. The prevalence of cardiovascular disease (78%) and hypertension (73%) was highest, followed by psychological disorder(s) (55%), ischemic heart disease (43%), diabetes (41%), hyperlipidemia (37%), COPD (27%), renal dysfunction (27%), and stroke (21%).

About two-thirds (63%) of all patients received any form of cardiovascular therapy in follow-up—mostly ACEi/ARBs (29%) or beta-blockers (26%). There were fewer AAs (5%) and others, including nitrates + hydralazine (37%). Of the only 5% of patients with specified HF, about two-thirds were diagnosed with systolic (or combined) HF. Both systolic and HFpEF patients were likely to be prescribed some form of disease-modifying therapy (70% vs 63%). The use rates of ACEi/ARBs (36% vs 32%), beta-blockers (35% vs 27%), AAs (6% vs 3%), nitrates + hydralazine (1% vs 1%), and other cardiovascular therapies (39% vs 38%) were somewhat higher in the systolic HF (vs HFpEF) patients (

Nearly 70% of all patients initially diagnosed with HF were hospitalized during follow-up, and 31% of patients were rehospitalized (since diagnosis coincided with their first hospitalization). Time to hospitalization (in median days) was shortest for those with comorbid renal failure (73 days) and longest for those with comorbid hyperlipidemia (238 days). Median time to hospitalization for the 290 patients with no comorbidity was 791 days.

report the adjusted risk of hospitalization over the follow-up period. The following results were estimated using the prevalence cohort of HF patients—note that estimates from the incidence cohort were similar. The presence of renal failure was the strongest predictor of hospitalization (average HR, 1.43; 95% CI, 1.36-1.51), followed by other cardiovascular disease (HR, 1.40; CI, 1.31-1.50), COPD (HR, 1.33; CI, 1.26-1.40), ischemic heart disease (HR, 1.28; CI, 1.22-1.35), stroke (HR, 1.27; CI, 1.20-1.34), diabetes (HR, 1.26; CI, 1.20-1.32), and hypertension (HR, 1.11; CI, 1.05-1.17). Patients with a psychological disorder (HR, 0.77; CI, 0.73-0.81) and/or hyperlipidemia (HR, 0.81; CI, 0.77-0.85) experienced significantly lower risk of hospitalization.

Hospitalization risk was significantly reduced through the use of ACEi/ARBs (HR, 0.77; CI, 0.73-0.81), betablockers (HR, 0.83; CI, 0.79-0.87), and/or other cardiovascular drugs (HR, 0.76; CI, 0.72-0.80) as first-line therapies for HF. However, initiation on AA and/or nitrates + hydralazine had no significant effect on hospitalization risk.

In the adjusted models, a diagnosis of HF coinciding with hospitalization significantly increased the risk for subsequent hospitalization (HR, 1.27; CI, 1.35-1.47), as did middle age (ages 45-54 years; HR, 1.13; CI, 1.06-1.19). When adjusted for other risk factors, Hispanics (HR, 0.81; CI, 0.68-0.98) or other race groups (HR, 0.83; CI, 0.75-0.93) were at lower risk for hospitalization than white or black patients.

The predicted HRs for significant risk factors were used to estimate the 6-month, 1-year, and 2-year risks of hospitalization among Maryland Medicaid patients diagnosed with HF (eAppendix). The model’s ability to correctly discriminate between individuals in the sample with high and low risk for hospitalization was good with a C statistic of 0.80 for any time horizon using both the prevalent and incident cohorts.

 presents the yielded annual numbers-needed-to-treat to prevent at least 1 hospitalization in the Maryland Medicaid prevalence-based HF cohort. As HF patients may experience multiple hospitalizations per year, our relative cost/savings estimates are conservative since they are based on the first hospitalization after HF diagnosis. While not all patients are candidates for certain cardiovascular therapies,

 estimates show that for every 12 patients started on ACEi/ARBs, at least 1 hospitalization can be prevented annually. A 20% increase in prescription rates in the sample resulted in savings of at least $85 (CI, $70-$101) per HF patient. Using our database of Medicaid claims to ascertain the number of non-dual-enrolled HF patients in Maryland Medicaid at any time in 2007 (12,054 patients), this estimate translated to a total annual savings of $1,024,590 (CI, $843,780-$1,217,454) for Medicaid.

For perspective, total medical expenditures by Maryland Medicaid on HF in 2007 was approximately $23.6 million.

Thus, Medicaid spent approximately $2200 (in 2011 dollars) per HF patient. This may be a slightly high comparator since dual-enrolled individuals were not included in our data, but comprised roughly 11% of all Maryland Medicaid enrollees in 2007.

For further comparison, Table 2 also presents the reduction in the number of cases of each comorbidity that was estimated to prevent 1 hospitalization per year. For instance, for every 8 HF patients presented with concomitant renal dysfunction, at least 1 hospitalization can be expected annually. Thus, a 20% decrease in the prevalence of renal dysfunction in the HF cohort resulted in an estimated saving of at least $111 (95% CI, $96-$127) per HF patient.

Given the high cost burden of hospitalizations and the potential ensuing complications and course of HF, our findings, in a unique Medicaid population, call for attention to strategies to attend to comorbidities and optimize use of disease-modifying therapy in such patients with HF.

 We found that a diagnosis of COPD, stroke, renal dysfunction, diabetes, ischemic heart disease, hypertension, or other cardiovascular diseases in HF significantly increased the risk of hospitalization: by about 30% for COPD or stroke, 25% for diabetes and ischemic heart disease, 10% for hypertension, and 40% for renal dysfunction and other cardiovascular diseases. Conversely, disease-modifying therapies reduced that risk by about 20%. Medicaid patients between 45 and 54 years of age were at the highest risk of hospitalization. Interestingly, the risk was 30% greater for individuals who were diagnosed with HF upon their first recorded hospitalization, perhaps suggesting a need for more proactive differential diagnosis approaches or faster referral patterns in the first point-of-care setting.

Medications that have been shown in clinical trials to improve systolic HF have not been shown to improve HFpEF, and prior data suggest that patients with HFpEF tend to have a history of hypertension and be older and female.

 Thus, our results likely overestimate the impact of disease-modifying therapies on hospitalization risk in all HF patients given that the broader population of HF patients might be more heavily represented by an older population with a higher prevalence of HFpEF. Moreover, our database consisted of only non-dual-enrolled Medicaid beneficiaries; therefore, caution should be used when generalizing our findings on relative risk to a dual-enrolled population. The objective of this study was to quantify the hospitalization risk in a high-risk Medicaid population with HF and should not be generalized to the Medicare population since it is likely represented by distinctly different risk groups.

We used the WCR method to estimate the average HRs on the comorbidities and other variables to reduce the possibility of biased estimates from a proportional hazards model. Given the nature of the claims-based data, we did not have access to lab values or medication logs per se, and thus could not assess disease severity nor ascertain that prescriptions filled were indeed taken by the patients. However, assuming this lack of information is systemic in the data and thus affects all comparison groups, it should not affect the risk comparisons.

To the extent that the study population was selected from among Medicaid enrollees, all study patients were, by definition, from a relatively homogeneous socioeconomic group and had the same insurance coverage, with presumably equal chance at access to medical care, medication therapy, or hospital services. Some cultural norms, beliefs and values, factors of cultural competency, literacy, self-efficacy, and other related measures may directly or indirectly affect the risk of hospitalization. We tried to adjust for these confounders indirectly, and at least partly, by adjusting for race/ethnicity. Further, we relied on the literature on the association between socioeconomic status, race/ethnicity, and socio-cultural-behavioral determinants of access to care to confirm that although worth pursuing in further research, the residual confounding due to unobservable variables may not present a notable bias in this study. We further matched samples by comorbidity in order to account for the possible imbalance of covariates across those comorbidities. Theoretically, this approach would have eliminated confounding of our estimates, to the extent possible, due to lifestyle factors or other unobservable variables that may contribute to the risk of hospitalization. The results were similar to those reported.

Perhaps physician practice styles, type of training, time in practice, or cultural competency may be concerns for bias.

 However, by design, in the general Medicaid setting, particularly that of managed care and related practice, reimbursement and incentive structures somewhat limit variations in practice style. Thus, to the extent that patients are likely to see various healthcare professionals during their course of care, we do not think that the lack of information on individual physicians in our study population poses a serious validity threat.

This study is the first to address the epidemiology of comorbidities in a high-risk Medicaid population, reflecting a demographic largely under-represented in large-scale studies or clinical trials.

 We show the unmet needs of this population and the clinical and hospitalization issues associated with prevalent disease and therapies. The burden of comorbidity was much higher than that observed in national statistics on HF patients, and the prescribing prevalence was lower than expected given the high-risk profile of the population. These findings may point to a high-priority area for Medicaid plans. Most notably, in the context of the study’s population demographic and clinical profile, we found that even small increments in disease-modifying therapies would result in significant reduction in costs to state plans. For instance, in the Maryland Medicaid program, a 20% increase in prescribing rates of ACEi/ARBs or beta-blockers would have led to an approximate annual savings of at least $85 or $57 per HF patient, respectively, or over $1 million total.

Substantive outcome improvement and savings to Medicaid may be achieved with small changes in prescribing rates or comorbidity prevalence among patients with heart failure.

Disease-Modifying Therapy and Hospitalization Risk in Heart Failure Patients

	As the costs of type 2 diabetes mellitus (T2DM) care and related clinical trials continue to rise, economically viable methods are being sought to effectively predict the relative utility of various treatment options. The high price of clinical trials has led to the development of alternative methods to collect and consolidate data. Comparative effectiveness research (CER) synthesizes existing evidence to address knowledge gaps and drive patient-focused clinical decisions and outcomes. CER methods compare the health outcomes and costs associated with interventions to determine the option with the maximum patient benefit at optimal cost. In addition to traditional CER approaches such as systematic reviews, meta-analyses, and retrospective claims analyses, Markov modeling and Bayesian analysis can be applied to predict patient outcomes in scenarios where clinical trials are not feasible. Additionally, cost-benefit, cost-effectiveness, and cost-utility analyses comprise “cost-effectiveness analyses.” Cost-benefit analysis looks solely at monetary value, while cost-effectiveness and cost-utility analyses include gains in health and quality of life, providing a ratio of cost to benefit. This paper will discuss a range of approaches to CER including Markov modeling, mixed treatment comparisons, the Archimedes model, and Bayesian statistics, and provide guidance in interpreting data from these studies in a managed care context, with a particular focus on evaluating treatments for T2DM. It will also provide guidance on common indices of comorbidity used in health economics research. Data from these models can be used to reduce treatment costs and improve the overall quality of population-level health.

Healthcare delivery is extremely expensive, and costs continue to rise as demand for care increases and newer treatments are developed.

 By 2020, US healthcare expenses are expected to increase to nearly 20% of the total gross domestic product.

 Healthcare delivery for type 2 diabetes mellitus (T2DM) is particularly costly; currently, 1 out of 5 US healthcare dollars is used to treat patients with diabetes.

 Likewise, the costs of clinical trial programs required for drug development have continued to rise sharply.

 This has led to an increased focus on alternate approaches to effectively predict the relative utility of various treatment options.

Considered broadly, these new approaches can be labeled as comparative effectiveness research (CER). The US Federal Coordinating Council describes CER as “The conduct and synthesis of systematic research comparing different interventions and strategies to prevent, diagnose, treat and monitor health conditions.”

 CER enables stakeholders to work toward a common goal of improving access to effective treatments, while keeping costs in mind. CER can identify ways to lower costs through improved outcomes, such as fewer emergency department visits and hospitalizations, and can be used to stratify treatment options based on efficacy, cost, and other factors.

CER can also provide data related to drug effectiveness, making it highly relevant to managed care decision making. Until recently, clinical trials applying active comparators were considered to be the most reliable type of CER. Such trials, however, are not only expensive to conduct, but also only predict outcomes over a short period of time.

Emerging CER approaches have started to address these issues by consolidating data from sources such as existing placebo-controlled clinical trials or patient data sets using computer-assisted analytic techniques.

 Some of these alternate approaches to evidence acquisition have achieved a high level of evidentiary acceptance, and many are conducted using fairly straightforward research methodologies; for example, meta-analyses and retrospective claims analyses.

 Other, more sophisticated approaches, such as Markov modeling or Bayesian analysis of mixed treatment comparisons (MTCs), are not as well understood by healthcare professionals.

This paper will summarize our current understanding of T2DM health economics research, discuss the role and benefits of CER in managed care decision making for T2DM, and provide an overview of Markov and Bayesian modeling techniques, as well as guidance on how to interpret data obtained from such studies in a managed care context.

Health Economic Analyses of Antidiabetic Medications

Economic evaluations of T2DM treatment and care compare the health outcomes and costs of different interventions to determine which option, when utilized, achieves maximum patient benefit at optimal cost.

There are 3 primary analytic methods for evaluating the economic value of an intervention that fall under the umbrella term “cost-effectiveness analysis.” These are cost-benefit, cost-effectiveness, and cost-utility analyses. Cost-benefit analysis typically measures and reports results in the form of the overall monetary value, whether costs or savings, associated with use of the treatment. Cost-effectiveness and cost-utility analyses also project monetary outcomes, but go further to measure gains in health and quality of life using a ratio of overall cost (in dollars) divided by health effect (eg, number of heart attacks prevented), or assessed as quality-adjusted life-years (QALYs).

In a systematic review, Klonoff et al evaluated 17 cost-benefit analyses performed to evaluate common diabetes interventions; the authors used this prior research to develop a scale estimating the economic impact of various diabetes interventions.

 summarizes the key findings of this systematic review, including the ratings for each type of intervention. This provides a stark demonstration of the lack of cost-effectiveness of many interventions, as well as the high utility of some underappreciated concerns in diabetes, such as preconception care. However, it is important to keep in mind that a lack of identified cost-effectiveness may simply indicate a need to modify an intervention, rather than discontinue its use.

Using Modeling to Estimate the Value of T2DM Screening Strategies

				The Archimedes model is a tool that assesses the potential long-term health and economic impact of new treatments on the healthcare system from the perspective of changes in medical guidelines, processes, and practice patterns. The model has been validated by more than 50 clinical trials.

				A recent analysis by Kahn et al used the Archimedes model to simulate and compare 8 T2DM screening strategies, compared with a control group receiving no screening. Screening strategies included routine screening once patients reached a certain age, or at hypertension diagnosis, as well as repeated screening at various time intervals. All simulated patients were followed for 50 years, or until death, with outcomes recorded annually.

				The model found that the T2DM screening strategy associated with the highest benefit, in terms of QALYs gained, was to initiate T2DM screening for all patients at 30 years of age, with repeat screening every 6 months. Investigators found that the use of this strategy resulted in earlier diagnosis (mean, 7.8 years) compared with the control group. Additionally, the simulation showed the effects of each screening strategy on myocardial infarction, stroke, microvascular outcomes, and death over a 50-year period.

				The ability to compare the outcomes of various screening strategies, plus the benefits of early T2DM diagnosis, against the costs of frequent screening provides useful information for informed managed care decision making.

Comparative Effectiveness Research on Treatments for T2DM

One of the main purposes of CER is to synthesize existing evidence in order to address knowledge gaps and drive patient-focused clinical decisions and outcomes.

 Thus, CER can identify the most beneficial treatment based on patient characteristics and inform real-world practice.

Two recent large, systematic T2DM reviews conducted for the Agency for Healthcare Research and Quality (AHRQ) summarize existing evidence for the use of oral agents and insulin.

 The AHRQ review of oral medications included outcomes from 140 randomized controlled trials (RCTs) and 26 observational trials; the results support the use of metformin as first-line monotherapy.

 In contrast, comparisons of dual drug combinations (metformin thiazolidinedione, sulfonylurea, meglitinide, dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 agonists, or basal or premixed insulin) identified no greater benefit with any specific treatment combination. Similarly, the insulin CER review included results from 50 studies and found, for example, that premixed insulin analogs were more effective in lowering fasting glucose compared with long-acting insulin analogs and noninsulin antidiabetic agents.

 As with the oral agents review, however, this paper identified research gaps. Specifically, certain comparator designs were not available (eg, basal-bolus vs premixed insulin), and a need was observed for more real-world effectiveness data on insulin use. In some cases, these limitations precluded the ability to draw definitive conclusions.

Observational studies have also been instrumental in CER when evidence gaps from RCT findings exist, despite their methodological ability to establish a reliable relationship between cause and effect.

 Observational studies use real-world data to assist in decision making. Sources of “realworld data” include administrative claims, patient registries, large simple trials, resource use information collected as part of clinical trials, supplements to clinical registration studies, health surveys, and electronic medical records.

 However, due to the lack of standardized instruments available to appraise quality of evidence, real-world data are not yet widely used for decision-making purposes.

 This highlights the need for innovative, sophisticated techniques such as Markov modeling and Bayesian MTCs to overcome these barriers. A study by Zhou et al illustrates how, in the absence of high-quality real-world data, an analytical computer model can have significant predictive power. Zhou and colleagues built a Markov-like model of diabetes progression, based on the baseline characteristics of patients from a populationbased study in southern Wisconsin. When the model was followed over time, predicted values were consistent with actual observed outcomes such as mortality, quality of life, and cost.

Common Indices of Comorbidity Used in Health Economics Research

				A number of comorbidity indices have been developed and validated for use in health economics research. These assessments evaluate patient risk and enhance study designs by accounting for patient comorbidity assessment, mortality risk, and comedication indicators. Some key indices used in CER evaluations include:

				The CCI predicts the 10-year mortality risk of patients with certain illnesses, based on medical record review. Patients are given scores of 0 to 5, with 0 representing the absence of comorbidities and 5 indicating a moribund state. Conditions that contribute to the CCI score are weighted by illness severity. This measure can be used to predict mortality and healthcare utilization costs.

				The ICED assesses the burden of coexistent diseases by combining 2 scales that reflect: (1) burden of coexistent disease, based on disease severity; and (2) overall physical impairment, based on the level of disability caused by illness. The burden of coexistent diseases can affect multiple aspects of care, such as recovery from surgery, and affect patient outcomes and resource use.

				The CDI uses data from a medication database to estimate a patient’s chronic disease burden. Although this model can miss chronic diseases that are not treated with medication, it can be used to risk-adjust a large patient population by predicting outcomes such as morbidity and mortality, costs of medical care, and healthcare utilization.

				The RxRisk model is an algorithm that uses prescription medication data to predict the cost of chronic diseases, making it useful for medical risk analyses. The use of specific prescription medications classifies a patient into a specific risk category, which, combined with the patient’s age and sex, provides an estimate of future healthcare costs.

Interpreting Complex Analytical Approaches

As statistical approaches to data analysis become more sophisticated, they also become more complex and, for those unfamiliar with these techniques, more difficult to interpret. This prompts questions regarding how to use the data generated by this type of research, so that evidence obtained can be translated into coverage decisions in a managed care setting.

 To facilitate this process, discussions of Markov models and MTCs are presented below.

Markov modeling is used to characterize a random process that evolves over time; it is therefore especially useful for chronic disease state modeling. In a Markov model, the disease is broken into separate “states;” each state is mutually exclusive (ie, a patient cannot be in more than 1 state at a time) and represents an important clinical and/or economic event in the process of the disease being modeled. For example, in a diabetes model, states might include “on treatment—healthy,” “on treatment with neuropathy,” or “on treatment with hypoglycemia.” Transition probabilities are assigned to the movement between these states over a set period of time (known as a Markov cycle). Estimates of health outcomes and/or resource use are assigned to the states and transitions. When the model is run for a large number of cycles, it becomes possible to estimate the long-term health outcomes or costs that may be associated with a disease or a particular intervention.

shows a simple example of a Markov model, with only 3 states. In this figure, there is asymptomatic disease, in which the patient has the disease, but is not experiencing any effects and has a mortality risk similar to an otherwise healthy individual. Next is progressive disease, in which the patient is experiencing symptoms and has an increased risk of mortality due to the disease. The final state is death. Death, in a Markov model, is an example of an “absorbing state,” one that is impossible to leave. In this sample model, patients can progress from asymptomatic to progressive disease. It may also be possible for a patient to move backward (ie, from progressive disease to asymptomatic disease) if their other condition improves. Also note the circular arrows in Figure 1, which indicate the possibility for a patient to remain in a particular state over more than 1 cycle.

An important limitation of Markov modeling is that it is “memoryless,” meaning that the probability of moving from one state to another is not dependent on the history of states previously experienced. A technique used to overcome the memoryless properties of Markov models is the use of time-dependent transition probabilities. In a Markov chain, transition probabilities are assumed to be constant without regard to time.

For example, a study used data from the National Health and Nutrition Examination Survey to create 2 cohorts (diabetes-prevalent and diabetes-incident) to put through a Markov model. This model further drew data from the UK Prospective Diabetes Study to model diabetes progression. Analysts were then able to estimate population changes in body mass index, diabetes incidence, disease progression and complications, and related costs over a 25-year period.

Another study used data from large-scale clinical trials, epidemiologic studies, and national surveillance studies to create 2 comparator cohorts: 1 group with diabetes receiving injected insulin and another on oral antidiabetic medication. In this study, the model was run twice, once as a baseline and once with medication adherence parameters included. Following this, the outputs of life expectancy, quality-adjusted life expectancy, complications, costs, and incremental cost-effectiveness ratios were compared between interventions, both with and without adherence issues taken into account.

In contrast to systematic reviews, wherein 1 pharmacological intervention is consistently compared against the same competitor using traditional meta-analysis techniques, MTCs, also known as network meta-analysis, compare the safety, efficacy, and costs of 2 or more interventions by linking them in a network. This method obviates the need to conduct a direct comparison study, and instead borrows evidence from other trials in which each respective intervention was evaluated against a common comparator (eg, placebo). The information obtained from these analyses can prove useful for medical decision making when direct comparison data are not available.

Fadia T. Shaya, PhD, MPH

Articles

Patient-Centered Teamwork in Care Transitions

Disease-Modifying Therapy and Hospitalization Risk in Heart Failure Patients

Decision Support Tools to Optimize Economic Outcomes for Type 2 Diabetes

Quantifying the Role of Natalizumab in Health and Economic Outcomes in Multiple Sclerosis