To assess the impact of the MDVIP model of personalized preventive care on hospital utilization rates over a 5-year period.

This study was a comparative hospital utilization analysis between MDVIP members and nonmembers using the Intellimed database from 5 mandatory reporting states (New York, Florida, Virginia, Arizona, and Nevada) from 2006 to 2010.

Hospital discharge rates per 1000 persons were calculated and comparisons were made between members and nonmembers by age (Medicare [>65 years] vs non-Medicare [35-64 years]) and year.

Overall, MDVIP members were approximately 42%, 47%, 54%, 58%, and 62% less likely to be hospitalized relative to nonmembers for the years 2006, 2007, 2008, 2009, and 2010, respectively. By 2010, MDVIP hospital discharges for the Medicare population were 79% lower than the nonmember Medicare population, and this difference was shown to be trending up since 2006 (70% to 79%). A similar trend was seen in the non- Medicare population (49% to 72%). In addition, elective, non-elective, emergent, urgent, avoidable, and unavoidable admissions were all lower in the MDVIP members compared with nonmembers for each year.

The MDVIP model of personalized preventive care allows the physician to take a more proactive, rather than reactive, approach; we believe this increased physician interaction is the reason for the lower hospital utilization and ultimately lower healthcare costs seen here.

The MDVIP personalized preventive care model which allows for increased physician interaction results in better health outcomes and ultimately lower healthcare costs.

		 MDVIP hospital discharges for both the Medicare and Commercial populations were lower than those for the nonmembers over 5 years, and this difference was shown to be trending up from 2006 to 2010.

		 MDVIP members were readmitted 97%, 95%, and 91% less frequently for acute myocardial infarction, congestive heart failure, and pneumonia, respectively.

		The estimated cost savings to the system from the 5 states was $119.4 million for 2010, representing a $2551 savings per patient.

The high and growing cost of healthcare is a significant issue in the United States, with hospital care representing the largest component (31.1%) of national healthcare expenditures.

 In 2008, there were an estimated 39.9 million hospital discharges (131 per 1000 persons) in the United States, with an estimated total aggregate cost of $364.7 billion.2 Medicare was the expected primary payer for the largest number of discharges (14.9 million, 46.2%), followed by private insurance (14.1 million, 32.2%).

Some hospitalizations for ambulatory care—sensitive conditions, such as diabetes and chronic cardiac and respiratory diseases, can potentially be prevented with timely and effective ambulatory care. Diabetes with complications was in the top 20 most expensive conditions requiring hospitalization for Medicare in 2008.

 In addition, coronary artery disease and acute cerebrovascular disease (stroke) ranked among the top 10, while congestive heart failure, heart attack, pneumonia, and respiratory failure ranked among the top 20 most expensive conditions for both Medicare and private insurance in 2008.3 In 2006, hospital costs for potentially preventable (ie, ambulatory care—sensitive) conditions were approximately $30.8 billion (1 of every 10 dollars of total hospital expenditures), and as many as 4.4 million hospital stays could possibly have been prevented with better ambulatory care, improved access to effective treatment, and/ or implementation of healthy behaviors.

 Therefore, reducing the frequency of potentially preventable hospitalizations would be an effective strategy for lowering costs while improving quality of care and outcomes.

In 2001, the Institute of Medicine issued its vision for reinventing healthcare in the 21

century to foster innovation and improve the delivery of care, which included patient-centered, responsive, and timely approaches to care.

 One new model that has emerged in the United States that seeks to address this issue is personalized preventive care, which focuses on prevention and wellness while delivering a high quality of care. This model is based upon consumer/patient empowerment, greater focus on preventive services, increased physician availability, and access to other healthcare resources. MDVIP (MD — Value in Prevention) is the first company to establish a network of primary care practices that delivers personalized healthcare in this new way. The MDVIP model of comprehensive management focuses on personalized preventive healthcare by delivering a host of screenings (eg, depression, anxiety, sleep, nutrition, sexual function, vision, and hearing) and diagnostics (eg, expanded testing for diabetes, bone disease, and cardiovascular disease) for a membership fee of $125 to $150 per month. MDVIP practices limit the number of members in each practice (

600 patients per physician) to be able to logistically deliver the service, thus allowing physicians to provide more personalized attention to manage all relevant health issues including acute concerns, chronic disease, and preventive screening, not just the diagnosis and treatment of illness.

 Traditional practices have over 2000 patients

 and do not have the time to provide many preventive services.

 MDVIP founders calculated the amount of time it would take to deliver a comprehensive preventive wellness program yearly and be able to follow up with chronic issues, thus capping the number of patients at 600 for each physician. As a result of the smaller practice sizes required to be able to provide the services, MDVIP members also receive same-day or next-day appointments for urgent and nonurgent care and the ability to reach their primary care physician (PCP) 24 hours a day. This model is not a third-party payer, and the fee covers only the extended prevention and wellness services provided by the PCP; therefore, members still need traditional health insurance to cover the costs of inpatient and outpatient visits, services provided by specialists, and other medical services (eg, laboratories, Xrays).6 In this model, MDVIP believes that the focus on prevention and wellness and the additional attention and time (ie, high quality of care) will lead to lower hospital utilization and ultimately lower healthcare costs. In an early effort to assess the impact of the MDVIP model on hospital utilization, we compared the hospital inpatient discharge rates from MDVIP members from 5 mandatory reporting states to nonmembers within the same states over a 5-year period.

Comparative Utilization Methodology and Definitions

Hospital inpatient utilization data were purchased from Intellimed, an industry provider of hospital data. Intellimed reports utilization by state and payer (ie, for Medicare and non-Medicare). They also provided population data from the US Census Bureau and Claritas system, which were used to construct the comparator (nonmember) population against which the hospital utilization of MDVIP members was compared. In addition, members were identified by physician provider information since individual patient identifiers are not provided in the database.

Hospital utilization data were classified within the Intellimed database as being elective/ non-elective, emergent/urgent, and avoidable/unavoidable admissions. An elective admission was predefined in the IntelliMed database as the condition permitting adequate time to schedule the availability of a suitable accommodation. A non-elective admission was predefined in the IntelliMed database as the condition not permitting adequate time to schedule the availability of a suitable accommodation. Furthermore, non-elective admissions were classified as emergent or urgent. Emergent and urgent admissions were also predefined in the IntelliMed database. Emergent admissions were those that required immediate medical intervention as a result of a severe, life-threatening, or potentially disabling condition, and urgent admissions were those that required attention for the care and treatment of a physical or mental disorder. An avoidable admission was defined and validated by MDVIP physicians as an admission for inpatient care of chronic illnesses that could be averted if the patient had good quality outpatient care (eg, ruptured appendix, asthma, cellulitis, congestive heart failure [CHF], diabetes, gangrene, hypokalemia, immunizable conditions, malignant hypertension, pneumonia, pyelonephritis, perforated or bleeding ulcer) and included the following

 International Classification of Diseases, Ninth Revision, Clinical Modification 

) codes: 032, 033, 037, 045, 055, 072, 250.1, 251.0, 276.8, 401.0, 402.0, 403.0, 404.4, 428, 437.2, 481, 482, 483, 485, 486, 493, 531.0, 532.0, 533.0, 540.0, 590.0, 681, 682, 785.4, inclusive of all sub-codes; whereas, an unavoidable admission could not be averted with good-quality outpatient care.

Of the 40 states and the District of Columbia in which MDVIP does business, only 5 states provide physician-specific data that could be used for this analysis. Therefore, MDVIP members were identified from MDVIP-affiliated physician provider information from those 5 mandatory reporting states: New York, Florida, Virginia, Arizona, and Nevada.

Since there is a lack of patient-level data within the Intellimed data set utilized for this analysis, MDVIP members were identified as any admission where an MDVIP physician was the attending physician. We established the comparator population using both inclusion and exclusionary measures. Using Claritas population data and collateral information available from the US Census Bureau, we established a comparison population matched by state of residence, age, and third-party payer coverage. We excluded from the comparator population individuals with attributes not found within the MDVIP cohort (ie, individuals under the age of 35 years were excluded from the commercial payer group, and individuals under the age of 65 years from the Medicare comparison group). We also excluded from the comparison group all payer groups not represented within the MDVIP membership (Medicaid, Medicaid HMO, CHAMPUS, Charity, Kid Care, state/ local government, self pay, underinsured, unknown, Veteran’s Administration, worker’s compensation, and other). In order to further ensure valid comparison, we excluded from the comparison group admission counts any admission types not represented in the population of MDVIP members (ie, newborn/ newborn premature, observation, unknown [including “info” and “NA”], semi-urgent, and trauma). MDVIP members were also excluded if their physician had a start date with the MDVIP network within the reporting quarter.

Hospital inpatient discharge rates per 1000 persons were used as the measure of comparison, calculated as: (discharges/ [days in reporting period/365])/(population/1000). Sensitivity analyses were also performed to assess the robustness of the data and to gain better understanding of systematic influences on the utilization differences between MDVIP and nonmember populations.

Additionally, the rate of readmission for MDVIP Medicare members between January 1, 2008, and December 31, 2010, with a primary diagnosis (

code) of acute myocardial infarction (MI), CHF, or pneumonia was calculated using IntelliMed as the data source. Even though individual patient identifiers are not available in the IntelliMed database, readmissions were defined as another hospital discharge within 30 days of a preceding discharge for the same

 code and the same physician provider for a patient with the same age and sex. Thus, if any patient’s date of birth fell in between the first admission and the readmission, then the readmission was not captured, since only age is provided in the IntelliMed database. Also, if a patient was readmitted by another physician provider, then the readmission was not captured. MDVIP Medicare member average readmission rates from 2008 to 2010 for the 3 medical conditions were then compared with readmission rates for non-MDVIP Medicare members within the same 5 states for the same 3 conditions based on the 2009 Medicare readmission rates available from the Dartmouth Atlas Project.

 shows the number of hospital inpatient discharge cases and the base population used in the analysis by health plan for each of the 5 years. From 2006 to 2010, the base population of MDVIP members more than doubled due to the increase in enrollment in the network over time. Approximately 45% of the MDVIP member base was commercial, and approximately 55% was Medicare per year.

Overall (combining commercial and Medicare), MDVIP members were approximately 42%, 47%, 54%, 58%, and 62% less likely to be hospitalized relative to the nonmember population for the years 2006, 2007, 2008, 2009, and 2010, respectively. 

 show the discharges/ 1000 by member population and year for the commercial and Medicare health plan populations, respectively. By 2010, hospital discharges for the MDVIP Medicare population were 79% lower than the nonmember population in this study, and this difference was shown to be trending up since 2006 (Figure 2). A similar trend was seen in the commercial population (49% in 2006 to 72% in 2010) (

In addition, relative hospital utilization tended to decrease as MDVIP membership increased over time. The estimated utilization savings for just 2010 in these 5 states was calculated to be $109.2 million for Medicare and approximately $10.2 million for commercial members, totaling $119.4 million, which was derived from the 2009 inpatient cost per admission value from the US Department of Health and Human Services (since the 2010 value was not available).

For elective admissions, the overall MDVIP member discharges/ 1000 was approximately 59% lower than nonmembers for both 2006 and 2007. For 2008, 2009, and 2010, the relative reductions in elective admissions were 71%, 77%, and 83%, respectively.  For non-elective admissions, the overall MDVIP member discharges/1000 were approximately 37%, 43%, 49%, 52%, and 56% lower than nonmembers for the years 2006, 2007, 2008, 2009, and 2010, respectively.

Of the non-elective admissions, the overall MDVIP member discharges/1000 for urgent admissions were approximately 17%, 35%, 42%, 40%, and 42% lower than nonmembers for the years 2006, 2007, 2008, 2009, and 2010, respectively. For emergent admissions, the overall MDVIP member discharges/ 1000 for urgent admissions were approximately 41%, 45%, 51%, 54%, and 58% lower than nonmembers for the years 2006, 2007, 2008, 2009, and 2010, respectively.

MDVIP members experienced approximately 23%, 31%, 38%, 47%, and 49% fewer avoidable admissions than nonmembers for the years 2006, 2007, 2008, 2009, and 2010, respectively. Furthermore, MDVIP members experienced approximately 45%, 49%, 56%, 59%, and 63% fewer unavoidable admissions than nonmembers for the years 2006, 2007, 2008, 2009, and 2010, respectively.

While the comparative utilization analysis matched MDVIP and nonmember populations on all available measures, we recognize the possible effects of selection bias, ie, MDVIP members may differ from nonmembers on factors such as attitudes toward health, personal health behaviors, and baseline health status, none of which are discernible from hospital inpatient utilization data. To better understand systematic influences on the utilization differences between MDVIP and non-MDVIP populations, we performed sensitivity analyses. First, we tested for bias introduced from scaling. The baseline comparative utilization analysis used the comparator group population of the MDVIP physician’s state of practice (matched as described previously). This established a comparison population ratio of roughly 425:1 (nonmembers:MDVIP member). We would expect true utilization differences to be robust to a reduction in scaling differential. To test this, we limited the comparator group to the population of matched nonmembers residing in the same zip code as the physician practice, reducing the comparison population ratio to 15:1. Using this scale, across all payers for the 5 reporting states, inpatient utilization was 57% lower for MDVIP members versus nonmembers. From this we can conclude that the utilization difference for MDVIP versus non-MDVIP is robust to population scaling. Second, since year-over-year differences between MDVIP and nonmember populations increased in magnitude for the past several years, we examined the relationship between mean caseload maturity (ie, how long the average person has been an MDVIP member) and the difference in utilization. We found a strong relationship; roughly 86% of the variance in utilization is accounted for by increased caseload maturity. However, due to a relatively small sample size (4 years of data), analysis of variance did not find significance at the 0.05 level (there is roughly a 7% chance that the relationship would be found by chance).

To further understand the dramatic differences between the hospital utilization of MDVIP Medicare members relative to non-MDVIP Medicare members, readmission rates were evaluated. When compared with the 2009 readmission rates for non-MDVIP Medicare members within the 5 same states,9 MDVIP members were readmitted 97%, 95%, and 91% less frequently for acute MI, CHF, and pneumonia, respectively (

Articles

WEB EXCLUSIVE

We assessed the impact of the MDVIP personalized preventive care model on hospital utilization and found the MDVIP members' rates were substantially lower than nonmembers'.



Personalized Preventive Care Leads to Significant Reductions in Hospital Utilization

 To evaluate the association between social support and medication adherence.

 A search of articles published before November 2010 in peer-reviewed, healthcarerelated journals was conducted using PubMed, EMBASE, and Web of Science, and search terms related to social support (social support OR friend OR family OR agency) and adherence (patient compliance OR medication adherence), yielding 5331 articles.

Articles were included if they directly measured the relationship between medication adherence and some form of social support. Excluded were case studies, studies with participants <18 years of age, and non—English language studies. Four social support categories were reported: structural, practical, emotional, and combination. Medication adherence was reported in the manner in which it was described in each study.

 Fifty studies were included in the final analysis. A greater degree of practical support was most consistently associated with greater adherence to medication; evidence for structural or emotional support was less compelling. However, most studies were limited in size and design, and substantial variability in designs and outcome measurement prohibited pooling of results, necessitating qualitative evaluation of the studies.

 This qualitative analysis found that practical social support was most consistently associated with greater medication adherence. Interventions that use existing contacts (friends or family) to engage patients in the mundane and practical aspects of medication purchasing and administration may be an effective approach to promoting better medication adherence.

This qualitative analysis found that practical social support is consistently associated with higher medication adherence.

		 Medication adherence is most closely associated with having a repertoire of close friends or providers who can consistently provide emotional and practical support that does not leave the patient with the perception of unmet needs.

		 Providing people in the patient’s social network with emotional and practical skills (eg, focused on medication adherence, barriers, support, and confidence building) may have a significant impact on medication adherence.

Nonadherence to chronic medications imposes a substantial clinical and financial burden on the US healthcare system. Studies have repeatedly demonstrated that medication nonadherence is a common source of hospitalizations, morbidity, and mortality in a variety of populations and disease states.

 The financial consequences of medication nonadherence in the United States are tremendous; the cost of care for patients with cardiovascular disease and diabetes are substantially greater in those who do not adhere to therapy than in those who do adhere, and total annual costs of medication nonadherence are estimated at almost $300 billion in the United States.

Despite the importance of medication nonadherence, no simple solutions are available to fix the problem.

 Numerous studies have aimed to improve medication adherence, yet best practices to enhance better medication taking continue to evolve.

 Moreover, little is known about the most cost-effective interventions to encourage better adherence.

 Therefore, new, low-cost approaches to improve medication adherence are needed to promote improved health outcomes and reduce healthcare costs.

The emergence of popular online social networking websites has stimulated interest in the role of social capital or connectedness in promoting health. Greater social support has been shown to be associated with improved health outcomes and healthier behavior.

 Yet the role of one’s social connectedness in medication adherence has not been well defined. Considering that social support connections are wide ranging, encompassing many aspects of the relationship between the patient and the person providing the support, we know even less about how different features of one’s social support system can influence medication use.

Accordingly, we conducted a systematic review of the published literature to evaluate what is known about the association between social support and medication adherence in a variety of disease states, and to explore features of one’s social support that might encourage better behavior. To do so, we categorized social support structures into those that are more functional and those that are predominantly emotional in nature.

 We aimed to evaluate the relative influence of the various types of support on medication adherence, so that appropriate and effective interventions to improve medication adherence can be designed.

With the help of a professional librarian, we performed a search of articles published before November 2010 in peerreviewed, healthcare-related journals using PubMed, EMBASE, and Web of Science. We used search terms related to social support (

). Articles with at least 1 search term from both categories met the criteria for the initial title/abstract review. After screening for duplicate entries, the results from the 3 databases were combined and totaled 5331 articles. Of these abstracts, 169 were included for full review of inclusion and exclusion criteria (

Articles were included if they directly measured the relationship between medication adherence and some form of social connectedness and/or support. There were 131 excluded articles, including case studies, studies with patients less than 18 years of age, non—English language publications, and studies where no medication-specific adherence measure was reported as an outcome (leaving 38 articles). Pediatric studies were excluded because the different mechanism and influence of social support in pediatric versus adult medication adherence would make conclusions difficult. The inclusion or exclusion of each article was determined by 2 reviewers independently, and disagreements were resolved by a third reviewer. All reference lists from the included articles were evaluated, and 12 additional studies (not found in our original search) were included, as determined by consensus. Data on study populations and characteristics, results, and study quality were extracted from each article using a standardized protocol and reporting form. Specific information collected included study population, social support measure, adherence measure, and adherence outcome.

For our key explanatory variable, social support, we defined 4 categories: structural, practical, emotional, and combination, which are further defined with examples in the 

. Each social support item abstracted from an included study was placed into 1 of these 4 social support categories by 2 reviewers independently, and disagreements were resolved by a third reviewer. The combination category reflects those social support items or instruments that did not distinguish the precise type of support measured, or that only reported a summary measure that included multiple support types.

Our outcome variable, medication adherence, is reported in the manner it was described in each study (Appendix). Studies that reported more than 1 social support category are represented in more than 1 table in the Appendix.

We included a total of 50 studies; 14 evaluated structural support, 12 practical support, 14 emotional support, and 26 combination support (Figure). Most were conducted in disease-specific populations; the most common were human immunodeficiency virus infection/acquired immunodeficiency syndrome (HIV/AIDS), diabetes, dialysis, tuberculosis, cardiovascular (hypertension and congestive heart failure), asthma, and transplant. Most were cross-sectional observational studies. Sample sizes ranged from 26 to 1198 patients.

Almost all included studies required participants to evaluate their perception of support; 2 interviewed caregivers.

 Approximately half used some variation of a previously validated instrument, and half used study-specific questionnaires. A majority utilized a self-completed survey instrument (n = 28), while fewer relied on interviews (n =16). The remainder utilized a combination of surveys and interviews, or did not describe the data collection. Most included several different types of support, while 17 measured only 1 type. Adherence was largely measured using subjective self-report, while 13 corroborated adherence objectively with pill count, refill information, or urine/serum markers of adherence.

 One study used a nurse assessment of adherence.

We identified 14 studies that assessed the relationship between structural social support and medication adherence; this structural support was cohabitation in 4 studies,

 presence of a spouse or supportive relationship in 7 studies,

In 1 study, structural social support was defined by the size of patients’ social network (Table).

Of all these studies, 6 (43%) identified an association between structural support and medication adherence, but some identified a positive association and others found a negative association. Two identified relationship status as significantly associated with being more adherent, with logistic regression coefficients of 1.2 (

; however, a third study identified having a partner as being associated with lower odds of adherence compared with not having a partner (odds ratio [OR] of adherence 0.97, 

and a fourth study found that adherence was highest in divorced persons.

Living situation (eg, living with someone) had a negative association with adherence in 2 studies (both in an HIV population).

 Only 1 study found that living with a spouse or relative had a significant, but modest, association with adherence.

Of the 8 (57%) studies that did not identify an association between structural support and medication adherence, 

 most evaluated marital status or living status. Two of these studies involved large cohorts of patients with objectively measured medication adherence.

 The 1 study that evaluated social network size did not identify a significant association with adherence, although it evaluated only frequency of contact, but not quality or type of the interaction.

We identified 12 studies that assessed the relationship between practical social support and medication adherence. Practical support was assessed by subjective participant ratings about how much help they received or how satisfied they were with the help in 3 main areas: medication help (reminders to take medication, directive guidance on medications, picking up prescriptions), help with household functions (cooking, cleaning, paying bills, running errands), and help with transportation.

Of these 12 studies, 8 (67%) identified a significant association of help with adherence. Of those, all but 1 identified a significant positive association (either amount of support or satisfaction with support) with medication adherence.

The 1 study that identified a negative association was a small study with an HIV cohort, where the adherence measure was nurse-perceived adherence, a poorly validated measure of actual medication use.

Two studies analyzed the association using a multivariate technique. The first study found the number of sources of practical support was associated with adherence; having 1 source of support was not associated with higher adherence, but 2 or more sources were associated with twice the odds ofadherence compared with the adherence of patients with no sources of support (OR of adherence 2.12, 95% confidence interval [CI] 1.06-4.26; 

In the second study, participants who had an “unmet need” for benefits (eg, Medicaid, Social Security, or prescription help) had an OR of nonadherence of 2.8 (

<.01) compared with patients who did not have this unmet need; 51% of the cohort reported needing help in this area. Other unmet practical needs were not associated with adherence (housing, emergency provisions, legal assistance, or supplemental nutrition), although about two-thirds of the cohort did not report a need in those areas.

Of the 4 (33%) studies that did not identify an association, all were limited by very small sample sizes (58-112 patients), and subjective self-reports of both practical support and medication adherence.

We identified 14 observational studies that assessed the relationship between emotional support and medication adherence. Emotional support was assessed by self-report in a variety of ways, including perceived amount of or satisfaction with social support, number of social support unmet needs, or the number of friends and social support sources.

Of these studies, 6 (42%) identified a significant association between any emotional support measure and adherence. Five found a positive association between emotional support and adherence, and 1 found a mixed association between support and adherence, with a negative association between social attachment and short-term (3-day) medication adherence (OR 0.49, 95% CI 0.34-0.71; 

 <.01), and a positive association between higher reassurance of worth and longer term (30-day) medication adherence (OR 1.3, 95% CI 1.03-1.6; 

 Of the 5 that found a positive association, 2 found a univariate association between medication adherence and emotional support in small samples of HIV patients.

Another small study of renal transplant patients found affectionate support explained 12% of the variance in adherence in multivariable regression.

A large sample of HIV patients identified as having the unmet need of counseling had significantly reduced odds of being adherent (OR 0.32; 

 <.01), and having the unmet need of a support group significantly increased the odds of not taking medications (OR 3; 

 In 1 large cohort of patients on lipid-lowering medications, the larger the number of close friends, the lower the chance of being nonadherent, with 16% nonadherence in those with no close friends and 7% nonadherence in those with 3 or more close friends (

= .04 for trend). The frequency of contact also correlated with adherence, although it did not reach statistical significance (16% nonadherence in those with less than weekly contact vs 8% nonadherence in those with at least weekly contact; 

Medication adherence is most closely associated with emotional and practical support.

Association Between Different Types of Social Support and Medication Adherence

To demonstrate a threat to validity in using claims-based risk tools with chronically ill, underinsured populations.

We tracked disease burden of highrisk pool beneficiaries with potentially disabling health conditions receiving enhanced health insurance benefits through a federally funded research demonstration. At baseline, beneficiaries paid high premiums and cost sharing for risk pool coverage, and most met common criteria for underinsurance. Study benefits provided intervention group members premium and cost-sharing subsidies and additional coverage; control group members paid usual premiums and coinsurance and received usual benefits. We hypothesized that enhanced benefits for the intervention group would increase or stabilize health status measures and decrease case-mix weights, reflecting stabilized or reduced disease burden.

The SF-12v2 health survey was used to measure health status and the Johns Hopkins Adjusted Clinical Groups (ACGs), Version 8.2 with DX-PM model and prior cost for a non-elderly population, was used to measure disease burden.

 Over a 3-year period, SF-12v2 scores showed stable health status for the intervention group and significant decline for the control group, while ACG case-mix weights, major illnesses, and chronic condition counts rose significantly for the intervention group but remained stable for the control group. Increased resource utilization for the intervention group appears to have driven increases in ACG measures.

 When high cost-sharing constrains access to care, risk tools that rely on medical claims may not provide an accurate measure of disease burden.

		 Risk tools based on International Classification of Diseases, Ninth Revision, Clinical Modification codes assume that access to care is sufficient to adequately represent health status.

		 Access barriers may mean that many conditions, even major comorbidities, are not coded.

		While the Affordable Care Act (ACA) may bring coverage to millions of previously uninsured individuals, many will still face high cost sharing. In addition, they may have higher than average comorbidity because of lower socioeconomic status.

		 Since the ACA proposes to use individual risk scores in adjusting risk across exchanges, and because claims-based risk tools are widely used for other purposes, the potential bias associated with underutilization should be recognized.

A number of administrative tools have been developed to measure comorbidity—or burden of illness&mdash;through claims data. Instruments such as the Diagnostic Cost Group/Hierarchical Condition Category (DCG/HCC),

 the Johns Hopkins Adjusted Clinical Groups (ACGs),

 and the Chronic Illness and Disability Payment System (CDPS),

 were created to measure actuarial risk but have been extended to numerous other purposes, such as targeting high-risk patients for case management, profiling physician performance, and conducting outcomes research.

 Most recently, the Affordable Care Act (ACA) proposes to use individual risk scores to determine or validate average plan risk within health insurance exchanges.

 Plan average risk scores, in turn, will be used to reallocate premiums within exchanges so that plans with a disproportionate share of sicker individuals will be compensated for losses. Thus, the issue of the validity of commonly used risk tools is assuming increased importance and examining potential vulnerabilities is an urgent issue.

In the absence of readily available standardized clinical information, administrative data are often the only practical source of information for estimating illness burden. Typically tools such as the ACG and CDPS calculate a case-mix score by combining diagnostic and/or prescription drug codes with demographic variables associated with variation in utilization, such as gender and age. Some tools use prescription drug codes instead of or in addition to diagnostic codes.

Various authors have noted the limitations of these tools.

 This article makes the case for another common, unrecognized vulnerability. Specifically, the validity of administrative data as a proxy for comorbidity requires that beneficiaries have adequate access to healthcare services in order to generate a reasonably complete diagnostic profile. As this paper will demonstrate, scant or missing diagnosis codes and cost data can severely bias case-mix scores, even for very sick populations. Arguably, tools that use pharmacy data may compensate somewhat, because many chronically ill individuals use prescriptions even when they do not regularly see a provider. However, these tools are limited by the lack of specificity in the uses of many drugs. In addition, because many individuals now obtain prescriptions from $4 generic drug retail programs, insurer prescription records can no longer be assumed to represent reliable records of utilization.

We illustrate these vulnerabilities with the ACG by showing how scores were influenced by a dramatic increase in utilization among a group of chronically ill adults participating in a research demonstration.

The Demonstration to Maintain Independence and Employment (DMIE) was a study of disability outcomes sponsored by the Centers for Medicare & Medicaid Services (CMS) under the Ticket to Work and Work Incentives Improvement Act (TW-WIIA) of 1999 and independently conducted by 4 states between 2006 and 2009.

 The demonstrations tested the hypothesis that improved access to healthcare and personal supports would prevent or delay transition to reliance on Social Security disability benefits among a population with potentially disabling conditions. The intervention was targeted to uninsured or underinsured populations whose high healthcare costs may have resulted in their delaying or forgoing needed health services. We hypothesized that the intervention would work over time by reducing barriers to care seeking, thus improving health status and reducing the burden of illness. This paper describes the experience of the Kansas DMIE, which recruited participants from the state high-risk insurance pool. The case-mix findings reported below pertain to the subset of the study population that was continuously enrolled for the first 2 years of the study. A more complete description of the full study population and outcomes is reported elsewhere.

As with most other state high-risk pools, the Kansas pool offers a benefit package modeled on nongroup coverage, which typically has less generous benefits and is more expensive than group coverage. At baseline, premiums were 125% of standard market rates. Deductibles ranged from $500 to $7500, with no benefits available prior to meeting the deductible, other than a $250 preventive benefit in some plans. Prescription coinsurance was 50%, and medical coinsurance was 30% in-network and 50% out-of-network, with out-ofnetwork charges not accruing toward the annual coinsurance maximum. Some plans had a $100,000 annual maximum, and all had a lifetime limit of $1 million. The coinsurance cap for in-network services was $5000 for single policies and $14,000 for family plans, with some plans having unlimited 10% coinsurance thereafter.

The study design was a clinical trial with applicants randomly assigned to equal-sized intervention and control groups. Control group members received high-risk pool benefits as usual, plus cash stipends for participation in surveys ($1200 paid in progressively increasing installments) and focus groups ($25 per session).

The intervention consisted of premiums subsidized to a flat $152 per month; elimination of all deductibles and coinsurance; cost-sharing limited to a $3 copay per service and charges in excess of reasonable and customary for out-of-network services; and a $1 million lifetime limit for DMIE-paid services. The intervention also provided nursing case management to coordinate care and pre-authorize benefits.

To be eligible, participants had to be enrolled in the state high-risk pool at least 6 months prior to recruitment and be aged 18 to 60 years, so that no one would turn 65 years during the study and attain Medicare coverage. They had to experience 1 or more health conditions that represented a potential for disability, as designated by CMS and the state program administrators. Data collected included claims files, surveys including standardized health status assessments and other questions, and focus group discussions. This research design and evaluation plan was approved by the institutional review board affiliated with the authors.

The program began recruiting participants in the fall of 2005 and offered intervention services from April 2006 through September 2009. Over this time a total of n = 508 subjects participated. The present analysis is limited to the subset of subjects who were continuously enrolled during the first 2 years of the study (n = 317, cohorts 1 and 2). The recruitment process is described in more detail elsewhere.

Although intervention and control groups were equal sized at baseline, large control group attrition resulting from dropping high-risk pool coverage created an imbalance in group size over time (n = 184 intervention, n = 133 control by the end of the second year). Comparison of demographic and health status data, including interviews that tracked reasons for disenrolling and types of replacement coverage, showed that those who disenrolled did not differ significantly from those who remained. Most disenrollees obtained group coverage through their own or a spouse’s employment, where there was no underwriting for pre-existing conditions. The reason cited for leaving high-risk pool coverage was almost exclusively the unaffordability of coverage.

Study participants were demographically similar to the overall high-risk pool population, except that they were on average slightly younger, reflecting the study’s eligibility cap. We found no statistically significant between-group differences in demographic variables or distribution of comorbidities (

). Most participants were aged 50 to 60 years, married, well-educated, and owned homes. Mean incomes were approximately $50,000 for individuals and $71,000 for families, with median individual incomes of $30,000 and family incomes of $50,000; 57% were below 400% of poverty when family size was included and 21% had incomes below 200% of federal poverty level. Occupations included homebased service or manufacturing enterprises, farming, professional services, and small business ownership. About 70% of participants earned at least some income from self-employment, explaining this population’s lack of access to group health insurance.

Baseline claims and self-reported diagnoses showed that participants experienced numerous comorbid conditions (Table 2). Musculoskeletal pain, cardiovascular conditions, and diabetes were among the most common; 19% had cancer, either active or by history. Forty-five percent reported at least 1 activity of daily living (ADL) limitation and 46% at least 1 instrumental activity of daily living (IADL) limitation, most commonly those requiring physical strength and mobility. Based on self-reported height and weight, 75% were overweight to obese (body mass index [BMI] >25), including 32% obese (BMI >30-40) and 11% morbidly obese (BMI >40).

Baseline claims showed high cost burden. Premiums averaged $460 per month, but ranged from $189 to $1011. Premiums for those aged 50 to 59 years (53% of the sample) ranged from $264 per month for a $7500 deductible plan to $878 per month for a $500 deductible plan. Only 3% of the sample enrolled in the $500 deductible plan, while almost one-half (46%) enrolled in $5000 or $7500 deductible plans. Median and mean out-of-pocket expenses were $1639 and $2472, respectively, compared with $960 and $575 nationally for nonelderly individuals covered by individual and group plans.14 Because of high deductibles and cost-sharing, participants, on average, paid 61% of allowed costs. Only 58% received reimbursement for more than one-half of allowed costs, and one-third received no reimbursement because they did not meet their deductible. Using a common definition of underinsurance— a deductible exceeding 5% or medical expenses exceeding 10% of family income; or if low-income, medical expenses exceeding 5% of family income&mdash;82% of the sample was underinsured.15 Because our data do not include expenses for other family members, the underinsurance rate is likely even higher.

In focus group discussions and during survey interviews, participants described how medical spending caused them considerable financial hardship and emotional stress. Twenty-two percent reported medical debt, with 25% owing $5000 or more and 6% $20,000 or more. They told of delaying or forgoing preventive services, diagnostic testing, elective surgeries, and medical equipment and employing strategies to stretch medication supplies, such as skipping pills, selectively not filling prescriptions, and continuing to use old medications after new ones had been prescribed.

They also reported enrolling in plans with high deductibles because these were the only ones for which they could afford the premiums.

The primary instrument for measuring health status was the SF-12v2, a scale derived from the SF-36.

We also administered the World Health Organization Quality of Life brief form (WHOQOL-BREF), the World Health Organization Health and Work Performance Questionnaires (HPQs), standardized ADL-IADL measures,

 and asked for self-reported diagnoses and health conditions. All instruments were administered by telephone at baseline and at 8-month intervals over the 41-month study period (5 rounds). SF-12v2 data are reported below for the first 4 rounds (baseline plus 3) corresponding with the first 2 years of the study. Claims data for case mix analysis are summarized for the corresponding 2-year period.

To measure comorbidity burden (case mix), we used the ACG, Version 8.2 with DX-PM model and prior cost for a non-elderly population.

We did not include prescription data because a preliminary analysis indicated it did not significantly affect outcomes for this study population. The ACG maps each beneficiary’s age, gender, and diagnostic codes to a single actuarial cell, or ACG, representing their estimated resource use. ACG reference unscaled concurrent weights, or case-mix weights, are ratios comparing estimated resource use for a given ACG with the average resource use of a nationally representative sample of insured individuals. The average case-mix weight for the national population is set at 1. Thus, someone with a case-mix weight of 2 has an estimated resource use twice the national average. Although ACG case-mix weights are measures of estimated resource use, they are also proxy measures of disease burden. Generally, weights greater than 1.0 indicate a population is sicker than the national average, and weights of less than 1.0 indicate it is healthier.

ACG software also produces other measures of disease burden, including Aggregated Diagnostic Groups (ADGs), Expanded Diagnostic Clusters (EDCs), counts

of major EDCs (chronic condition counts), and resource utilization bands (RUBs). Recent versions have also included a predictive model algorithm for forecasting cost.

This article demonstrates a threat to validity when using claims-based risk tools with chronically ill and/or underinsured populations whose underutilization masks actual risk or comorbidity.

The Validity of Claims-Based Risk Estimation in Underinsured Populations

Accountable care organizations (ACOs) are considered by many to be a key component of healthcare delivery system improvement. One expectation is that the structural elements of the ACO model, including clinical integration and financial accountability, will lead to better coordination of care for patients. But, while structure and incentives may facilitate the delivery of coordinated care, they will not necessarily ensure that care coordination is done well. For that, physicians and other healthcare providers within ACOs must possess and utilize specific skills, particularly in the areas of collaboration, communication, and teamwork. In this article, we present strategies in 3 domains—training, support tools, and organizational culture&mdash;that ACOs can implement to foster the development of these skills and support their use in clinical practice.

While clinical integration and financial incentives should facilitate care coordination in accountable care organizations (ACOs), they will not necessarily ensure that care coordination is done well. In response, ACOs will have to develop and support professional skills, particularly in the areas of collaboration, communication, and teamwork. Strategies for ACOs to build these skills include:

		Establish training programs designed to enhance clinicians’ care coordination skills.

		Implement support tools that enable the use of these skills in clinical practice.

		 Create an organizational culture that recognizes and values the use of these skills.

Better coordination of care for patients is an essential component of the broad effort to improve healthcare quality and efficiency in the United States.

 But coordinating care across the various settings in which patients are treated is a daunting task.

 One possible solution to the care coordination challenge is accountable care organizations (ACOs). The great promise of ACOs is that by establishing a structure conducive to clinical integration and a payment system that demands accountability, the care that is delivered will be well coordinated. As a result, our healthcare system would be less fragmented, more fiscally responsible, and higher quality.

However, coordinating care requires more than a shared organizational structure and financial incentives.

 Individual physicians and other healthcare providers within ACOs must also know how to coordinate care. This requires specific professional skills, especially in the areas of collaboration, communication, and teamwork. These skills allow clinicians to integrate actions and expertise, negotiate differences in judgment, and determine shared priorities for patients’ care. In order to deliver coordinated care, ACOs will need to implement organizational strategies that develop and promote care coordination skills among their staff. Many of these strategies have a precedent in medicine or other professions and can be categorized into the following 3 key domains: training, support tools, and organizational culture.

 ACOs should establish training programs specifically designed to enhance clinicians’ care coordination skills. Interpersonal communication is fundamental to coordinating care, yet almost one-half of  physicians surveyed in 2009 acknowledged the need to improve communication processes.

 Some medical schools, physician groups, and nursing organizations have begun to offer training in this area, but ACOs will have to make these efforts more systematic. For instance, at Kaiser Permanente (a system with many characteristics of an ACO), physicians serving as “communication consultants” run workshops on communication between providers. Training should also include education on the roles of staff members within the care team. Understanding who is responsible for what is critical for effective collaboration in any organization, but even more so in an ACO setting, where traditional clinical roles are likely to evolve and staff from a variety of disciplines and specialties will be working together to coordinate patient care.

To build teamwork skills, ACOs can look to programs that have borrowed principles from other fields and applied them in medicine.

 For example, the US Air Force’s Medical Team Training Program was developed with an eye toward using human-factors concepts to reduce medical errors. Medical personnel are trained through a combination of coursework, observation in the work environment, and feedback. Unfortunately, evidence about the impact of this program, and others like it, is limited. Plus, the teamwork skills needed to effectively collaborate in a large, cross-disciplinary clinical setting like an ACO may differ from the emergency department or operating room settings where many of these programs have been focused. As a result, new training programs in ACOs should undergo continuous evaluation, and their success (or failure) should be disseminated so that other ACOs can learn from the experience.

The use of information systems does not obviate the need for this type of training. A study from 4 Kaiser Permanente medical centers showed that while the electronic medical record improved coordination of diabetes care, coordination suffered when providers had discordant views about team member roles.

 Likewise, nurse care managers should not be relied upon to coordinate the care of patients without also cultivating these skills among the rest of the care team. Dedicated support through additional personnel may be an important component of the overall strategy to improve care coordination. However, care coordination should be an organizationwide priority and a valued service in ACOs, and therefore everyone involved in patient care must be fully prepared to participate in the effort. This applies especially to physicians, who will always remain engaged in some of the most complex and important communications, regardless of changes in the healthcare delivery system.

While training will equip clinicians with care coordination skills, tools are necessary to support their use. Support tools provide structure to the content of care coordination activities and help integrate these activities into routine practice. For example, the “Situation, Background, Assessment, Recommendation” (SBAR) technique is a communication tool that has been used effectively to standardize information exchanged by members of the healthcare team, particularly between nurses and physicians.

 Support tools like this must be used thoughtfully, since the information that needs to be standardized varies based on the type of communication (eg, specialist consultation vs hospital unit transfer).

Another support tool that could help communication is a checklist. In the construction industry, communication tasks are put into checklists, ensuring that experts from different fields coordinate their efforts before the project can proceed.

Similarly, ACOs could add a checklist for communication tasks to the “plan” section of the electronic medical note template, which might compel clinicians to make communication with each other part of their plan of care.

Support tools—in particular, ones now made possible through new technology&mdash;can also enhance the efficiency and quality of collaboration.

 For example, privacy-sensitive text messaging and e-mail allow clinicians to rapidly communicate with each other. In addition, electronic health records can allow them to easily identify patients with particular care coordination needs, such as diabetics overdue for ophthalmic exams. But high-tech support for care coordination could also be taken further through tools with social networking features. For instance, adding a “wiki” to a patient’s electronic medical record would enable physicians and other clinicians to maintain an updated history and comprehensive assessment in a single document (with links to more detailed data) and to share an integrated care plan.

Another potentially useful tool, particularly for medically complex patients, is activity streams. Fed by alerts on medication changes, transfers in care, physician appointments, and laboratory data, activity streams can be quickly scanned for relevant and important information, allowing clinicians to stay updated and connecting them with others participating in their patients’ care. Use of tools like these must be accompanied by training and evaluation in order to maximize their impact.

ACOs already have the “why” of their mission solidly defined: to provide high-quality, wellcoordinated care for their patients. Leaders within ACOs will need to take this mission and shape an organizational culture that supports the “how” for frontline clinicians. One way to signal the value of care coordination is by protecting time in the workday for care coordination activities and accounting for these demands in clinical scheduling. In addition, ACOs could formally schedule multi-disciplinary meetings to coordinate care for complex patients. Physicians and other healthcare providers should also be given a forum in which to share with each other innovative strategies for care coordination.

ACOs should consider including care coordination skills among the professional standards conveyed to current and prospective clinical staff. Just as good communication with patients is considered to be important, so too should good communication with other healthcare clinicians. Adding these skills to the definition of a high-quality provider and incorporating both patient and peer feedback on this metric into performance reviews will require a culture shift. But leaders of ACOs are uniquely positioned to spearhead this change because of the central role of care coordination in their mission.

There are many new ACOs and organizations similar to ACOs already working to provide coordinated care to patients. These groups will amass a wealth of experience and knowledge, and opportunities to share their best practices will be essential for continued improvement.

 A key component of such learning collaboratives should be the domains that we have outlined—training clinicians, providing support tools, and changing organizational culture&mdash;to help build the professional skills necessary for good care coordination. Making the development of these skills a priority will help ACOs live up to their potential to ameliorate the fragmentation and discontinuity that plague our healthcare system.

The authors wish to acknowledge Lawrence P. Casalino, MD, PhD (Weill Cornell Medical College), for comments on a draft of this article. They also wish to thank Peter B. Bach, MD (Memorial Sloan-Kettering Cancer Center), and Joseph B. Press, PhD (Deloitte Consulting), for conversations that informed the section on new technology.

From Departments of Public Health and Medicine (MJP), Weill Cornell Medical College, New York, NY; Weill Cornell Medical College (MDM), New York, NY; New York University Wagner Graduate School of Public Service (LHM), New York, NY.

 The authors (MJP, MDM, LHM) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (MJP, MDM); acquisition of data (MJP, MDM, LHM); analysis and interpretation of data (MJP, MDM, LHM); drafting of the manuscript (MJP, MDM); critical revision of the manuscript for important intellectual content (MJP, MDM, LHM); administrative, technical, or logistic support (MJP); and supervision (MJP).

Matthew J. Press, MD, MSc, Department of Public Health, Weill Cornell Medical College, 402 E 67th St, New York, NY 10065. E-mail: map9069@med.cornell.edu.1. National Priorities Partnership. National Priorities and Goals: Aligning Our Efforts to Transform America’s Healthcare. Washington, DC: National Quality Forum; November, 2008.

2. Pham HH, O’Malley AS, Bach PB, Saiontz-Martinez C, Schrag D. Primary care physicians’ links to other physicians through Medicare patients: the scope of care coordination. Ann Intern Med. 2009;150(4): 236-242.

3. Walker J, McKethan A. Achieving accountable care—“It’s not about the bike.” N Engl J Med. 2012;366(2):e4.

4. Hess BJ, Lynn LA, Holmboe ES, Lipner RS. Toward better care coordination through improved communication with referring physicians. Acad Med. 2009;84(10 suppl):S109-S112.

5. McDonald KM, Schultz E, Albin L, et al. Care Coordination Atlas Version 3 (prepared by Stanford University under subcontract to Battelle on Contract No: 290-04-0020). Rockville, MD: Agency for Healthcare Research and Quality; November 2010.

6. Baker DP, Gustafson S, Beaubien J, Sala E, Barach P. Medical teamwork and patient safety: the evidence-based relation. Rockville, MD: Agency for Healthcare Research and Quality; April 2005. Publication No: 05-0053.

7. MacPhail LH, Neuwirth EB, Bellows J. Coordination of diabetes care in four delivery models using an electronic health record. Med Care. 2009;47(9):993-999.

8. Leonard M, Graham S, Bonacum D. The human factor: the critical importance of effective teamwork and communication in providing safe care. Qual Saf Health Care. 2004;13(suppl 1):i85-i90.

9. Horwitz LI, Detsky AS. Physician communication in the 21st century: to talk or to text? JAMA. 2011;305(11):1128-1129.

10. Gawande A. The Checklist Manifesto: How To Get Things Right. New York: Metropolitan Books, Henry Holt and Company, LLC; 2009.

11. Walker JM, Carayon P. From tasks to processes: the case for changing health information technology to improve health care. Health Aff (Millwood). 2009;28(2):467-477.

12. O’Malley AS. Tapping the unmet potential of health information technology. N Engl J Med. 2011;364(12):1090-1091.

13. Fisher E, Shortell SM. ACOs: Making sure we learn from experience: the Commonwealth Fund blog. http://www.commonwealthfund.org/Blog/2012/Apr/ACOs-Making-Sure-We-Learn-from-Experience.aspx. Published April 2012. Accessed June 11, 2012.

COMMENTARY

In order to live up to their potential, ACOs should ensure that clinical staff possess the professional skills necessary to effectively coordinate care.

Care Coordination in Accountable Care Organizations: Moving Beyond Structure and Incentives

Cost-effectiveness analyses compare the value of different treatment strategies, given that one strategy is more effective but also more costly than the other. The results are often used as a pharmacoeconomic basis to support a particular treatment strategy.

 To determine whether studies that reported favorable cost-effectiveness results in the antiviral treatment of hepatitis B and hepatitis

 A systematic literature search was performed to identify all original cost-effectiveness studies published from 1987 through 2011 regarding antiviral treatment of hepatitis B or hepatitis C. Data were abstracted independently by 2 reviewers. Corresponding authors were also contacted.

Of the 77 original cost-effectiveness analyses, 55 (71.4%) utilized a cost-effectiveness ratio below $50,000 per quality-adjusted life-year (QALY) in reaching a main conclusion for the study. A total of 46 analyses (59.74%) utilized ratios below $20,000/QALY. Most analyses (71 [92.2%]) reported favorable cost-effectiveness results. Countries outside of the United States and Europe tended to report lower cost-effectiveness ratios. Academic-funded studies were less likely to report cost-effectiveness ratios below $20,000/QALY. Most of the corresponding authors stated that their cost-effectiveness studies were published.

Most published cost-effectiveness analyses regarding antiviral treatment of hepatitis B or hepatitis C reported favorable results. These results may suggest that researchers only attempt cost-effectiveness studies when there is a high likelihood of positive results. Regardless of the explicit cause of the bias, more awareness and scrutiny are needed when utilizing cost-effectiveness studies in decision making.

Cost-effective studies are often used to support a particular change in treatment strategy, however:

		 Most cost-effective results are favorable in the literature.

		 Most cost-effective studies are supported by pharmaceutical industry and there is lack of financial disclosure.

		Potential bias may exist at the author, funding source, reviewer, and/or editor level.

Over the past 10 to 15 years, there have been an increasing number of treatment strategies for hepatitis B and C. Overall, every generational succession in viral hepatitis treatment has been associated with incremental increase in efficacy and occasionally costs.

Regardless of the cause of bias, more awareness and scrutiny are needed when utilizing cost-effectiveness studies for healthcare decision making.

December 2012

Personalized Preventive Care Leads to Significant Reductions in Hospital Utilization

Association Between Different Types of Social Support and Medication Adherence

The Validity of Claims-Based Risk Estimation in Underinsured Populations

Care Coordination in Accountable Care Organizations: Moving Beyond Structure and Incentives

Trends in Viral Hepatitis Cost-Effectiveness Studies

Communication About Diabetes Risk Factors During Between-Visit Encounters

Population-Based Breast Cancer Screening in a Primary Care Network

Prioritized Post-Discharge Telephonic Outreach Reduces Hospital Readmissions for Select High-Risk Patients

Impact of MMRV Combination Vaccine on Childhood Vaccination Compliance

Impact of Education Program on Influenza Vaccination Rates in Spain