Value Drivers: An Approach for Estimating Health and Disease Management Program Savings

Abstract

Health and disease management (HDM) programs have faced challenges in documenting savings related to their implementation. The objective of this eliminate study was to describe OptumHealth's (Optum) methods for estimating anticipated savings from HDM programs using Value Drivers. Optum's general methodology was reviewed, along with details of 5 high-use Value Drivers. The results showed that the Value Driver approach offers an innovative method for estimating savings associated with HDM programs. The authors demonstrated how real-time savings can be estimated for 5 Value Drivers commonly used in HDM programs: (1) use of beta-blockers in treatment of heart disease, (2) discharge planning for high-risk patients, (3) decision support related to chronic low back pain, (4) obesity management, and (5) securing transportation for primary care. The validity of savings estimates is dependent on the type of evidence used to gauge the intervention effect, generating changes in utilization and, ultimately, costs. The savings estimates derived from the Value Driver method are generally reasonable to conservative and provide a valuable framework for estimating financial impacts from evidence-based interventions. (Population Health Management 2013;16:356–363)

Introduction

In the last decade, disease management has grown to include health promotion and wellness.¹ Newly coined health and disease management (HDM) programs involve a range of services that address gaps in medical care necessary to improve health outcomes and reduce medical costs. Common components include patient education with a significant emphasis on self-care, ongoing patient monitoring, care coordination, and promotion of evidence-based protocols.^2,3 Programs can target a health plan population or individuals within the plan who have manifest symptoms or who are at risk of developing certain conditions such as asthma, diabetes, or congestive heart failure (CHF).⁴

HDM programs can take many forms. They can be classified as self-managed, nurse-managed, or team-managed. They can be integrated into a health care practice, operate as a stand-alone, or be a mix of the 2.³ HDM programs can be proactive or passive in targeting a population for intervention. Information can be presented to consumers through face-to-face meetings, telephonic coaching, or case management, mail, e-mail, and/or other forms of social media. Many different methods may be used to “engage” the consumer. These design features have a direct impact on a program's effectiveness.

HDM programs can improve clinical outcomes and care quality⁵; however, evidence of the impact that HDM programs have on health care costs is mixed. There are examples of individual programs that have led to cost savings in high-risk populations and in those with certain conditions, such as CHF, but other evaluations of HDM have failed to find consistent evidence of cost savings.⁶

An analysis by the Congressional Budget Office (CBO) in 2004 found insufficient evidence to support the claim that chronic disease management programs generally reduce health care costs, particularly when the costs of the intervention are taken into account.⁷ Critics charged that the report was limited to programs targeted at patients with chronic conditions only.⁸ Other concerns were that the CBO report did not perform a formal meta-analysis, omitted some evidence favorable to disease management programs, and did not distinguish programs by structural type, an important predictor of success.^9,10

Contributing to the discussion, Goetzel et al conducted a review of 44 studies that investigated the return on investment (ROI) for HDM for various chronic conditions. The authors found a positive ROI for programs addressing CHF and multiple illnesses, but mixed evidence for programs for asthma and diabetes.¹¹

A recent analysis of the Medicare Health Support Pilot Program further challenges claims of savings.¹² The Program used nurse case managers (NCMs) in stand-alone call centers to educate patients and assist with care coordination. The findings from this randomized controlled trial (RCT) involving eight HDM companies indicated that this effort had “limited success” in improving processes of care and produced no significant savings for Medicare in terms of reduced emergency room (ER) visits or hospital admissions. Net management fees ranged from approximately 4% to 10% per beneficiary per month, or the cost of 2 to 3 physician office visits of moderate complexity.

The authors acknowledged several limitations to their findings. The study emphasis was on managing a very difficult population, namely high-risk elderly patients with multiple comorbidities including psychosocial problems. The programs lacked the real-time information available in many other HDM programs, and in many cases could not identify candidates for participation until weeks or months after an acute event occurred. Also, the analysis likely suffered from regression to the mean.¹³ Furthermore, the stand-alone coordination model investigated is not the only HDM structural approach. Industry standards strongly support the role of the physician and strong linkages between the NCM and physicians of record.¹⁴

Documenting value is critical to purchasers of HDM programs. Methodological issues abound in determining an HDM program's ability to alter utilization and overall costs. The level of consumer engagement in a given HDM program to date has been an important industry metric. Engagement has served as a proxy for program effectiveness when effectiveness has not been “proven” through an RCT or other scientifically sound analyses.¹⁵ Current industry efforts are now focused, in part, on creating a standard for direct documentation of program effectiveness without having to resort to the expense and time of conducting an RCT (or similar experiment) of a given program.¹⁴

OptumHealth (Optum), an industry leader, provides population health management services that address the physical, mental, and financial health care needs of organizations and individuals. In line with this mission, Optum developed Value Drivers to provide estimates of economic value from identifying and proactively managing patients with chronic or other health conditions. Value Drivers employ the best evidence available to estimate the effect on utilization from HDM interventions and to identify the savings associated with the effect among compliant patients. This article describes in detail the methodology for calculating Value Drivers, the process used in the external review of savings estimates, and 5 Drivers for illustration of the method. The contributions of this project to the HDM field are then discussed.

Optum Value Driver Approach

In response to internal management and industry concerns, Optum developed Value Drivers that estimate medical savings associated with compliance with specific clinical and lifestyle interventions. Interventions apply evidence-based protocols, use real-time patient information, and emphasize professional peer-to-peer interactions among NCMs and physicians and a team-based approach to care with the team integrated into clinical practice. Interventions are proactive and wide-ranging and include educating and counseling patients, monitoring biomarkers, communicating with physicians about member needs and concerns, arranging patient transportation, negotiating benefits, and planning for hospital discharge to prevent readmissions, among other interventions.

The Value Driver approach grew out of a traditional HDM program that focused primarily on implementing clinical guidelines, such as those for diabetes care. In developing the approach, Optum combined claims, pharmacy, biometric, laboratory, and self-reported data to identify patients affected by or at risk for certain conditions. It also began to profile provider and health plan performance against industry standards, such as Healthcare Effectiveness Data and Information Set measures, to identify opportunities to improve care delivery. Additionally, Optum formalized NCM activities into standardized protocols. Nurses and health advocates began to use real-time information to prioritize high-value and high-volume conditions and to work interactively with patients and providers.

By identifying interventions that could maximize potential health care savings, Optum focused on creating Value Drivers that met 3 general criteria: (1) evidence that interventions, particularly those leading to changes in patient behavior, could provide clinical benefit; (2) conditions for which expenditures were relatively large because of high prevalence or high cost per case; and (3) high likelihood that NCMs could convince patients and providers to carry out the interventions so that savings were achievable relatively quickly within a commercial population. Interventions that met clinical and case management criteria led to the development of Value Drivers, the new foundation of Optum's HDM model.

Value Driver Savings Method

Optum assembled a multidisciplinary team, headed by a medical director and including epidemiologists, statisticians, and clinical experts, to assess possible savings attributable to each HDM intervention. First, staff from the relevant clinical team provided a detailed description for review of the intervention and its intended outcome. Based on the strongest evidence available, a specific utilization effect was attributed to successful completion of the intervention. The evidence base was derived from a range of sources—from results of published studies employing rigorous RCT methods to Optum staff experts' opinions.

For example, 1 Value Driver focused on securing compliance with taking beta-blockers among patients with heart failure. Data from a meta-analysis of RCTs involving patients with the condition showed that taking a beta-blocker as prescribed reduced the rate of hospital admissions by 4% annually.¹⁶ This value became the utilization effect associated with the intervention. Optum staff estimated savings among compliant patients by multiplying the change in utilization by the average cost per hospitalization for heart failure.

Optum established formal guidelines to calculate savings for all Value Drivers. Analyses accounted for direct costs to the health plan and patient, but not indirect costs such as travel or time lost from work. All relevant categories of utilization, including pharmacy, were counted. All claims-based Value Drivers were monetized using Allowed Amounts, which included health plan liability (net paid amount) and patient liability (coinsurance and deductible amounts), and using a UnitedHealthcare (UHC) fully insured (FI) commercial population as the reference group. If an intervention averted a full episode of care, Ingenix Episode Treatment Groups (ETG) data were used to calculate savings.¹⁷ The analytic time frame for each Value Driver was 1 year and all costs were adjusted for inflation and denominated in the same base year. Any effects of the intervention on mortality were not included.

An important assumption in the calculation of savings was that patients successfully followed nurse or health coach directives for each intervention (ie, they “closed the gap” in care). All Value Drivers were assigned a defined goal, such as “the member becomes medication compliant,” and success was measurable. A successful or closed case occurred when: (1) a NCM or health advocate interacted with the member eligible for the intervention, (2) a NCM completed the intervention with the member, and (3) the member followed the advice of the NCM, health advocate, or physician and a reduction in utilization occurred. Costs associated with delivering the intervention, including setup and/or operating costs, were not included in monetization models, and analyses did not take into account any probabilities associated with closing a case. Thus, the savings estimates were derived from what may be described as an ideal scenario in which the member who would benefit from the intervention was properly identified and successfully contacted. This scenario required that the member complied with the advice or treatment offered and health care utilization was affected leading to a reduction in medical costs. The closed case approach focuses on obtainable savings within 1 year. A critical issue, then, is how reliable the savings estimates are.

Against this backdrop and with the intent of being transparent about its methodology, Optum documented in detail the sources from which it assigned a utilization effect and then subsequently calculated savings. Based on the type of evidence for the utilization effect, Optum rated the drivers as strong, moderate, or weak. These ratings do not reflect the clinical value associated with the intervention. For further validation, Optum contracted with faculty of Emory University to perform an external review of the Value Driver savings methodology.

Review of Value Drivers

The authors evaluated a convenience sample of 52 Value Drivers, selected by Optum to represent high-value/high-volume Value Drivers and the types of calculations that had been undertaken. The general review process was as follows. Regarding the evidence for the utilization effect, the authors assessed the quality of the analysis (ie, RCT or not), on which it was based and then assigned a rating to it (strong, moderate, or weak). The categories for ranking evidence are shown in Table 1.¹⁸ The authors next evaluated how the effect was combined with cost data and rated whether the resulting savings estimate was likely to be: (1) an overestimate, (2) an underestimate, (3) reflective of savings, or (4) a situation in which possible bias could not be determined. This assessment was based on the nature of the evidence for the utilization effect and its applicability to the patient population involved in the intervention.

Table 1.

Description of Evidence Rankings

□ Level 1 (Strong)

□ Meta-analyses of peer-reviewed studies □ Experimental peer-reviewed studies

□ Quasi-experimental studies with good comparison groups and good control for confounders

□ Observational study with good comparison groups and good control for confounders

□ Other (specify): _____________________________________

□ Level 2 (Moderate)

□ Quasi-experimental studies with poor comparison groups and poor controls for confounders

□ Observational study with poor comparison groups and poor control for confounders

□ Pre-post experimental design □ Other (specify): _______________________________

□ Level 3 (Weak)

□ Expert estimate □ Indirect evidence (ie, studies examining similar but not identical interventions)

□ Other (specify): ______________________________________

The rating process itself involved several steps. First, in order to establish interrater reliability, the 3 reviewers (ie, the authors) assessed 5 random drivers separately, then compared and calibrated their findings. For the remaining analyses, each Value Driver was assigned a primary and secondary reviewer who conducted the initial assessment and conferred. When discrepancies existed, input from the remaining reviewer was obtained. The reviewers then presented their reports during biweekly conference calls in which all reviewers, Ron Goetzel, PhD, and Optum staff participated. Following these discussions, final ratings were determined by the authors alone.

Table 2 provides a complete list of the Value Drivers reviewed. Value Driver topics ranged from promoting hospice care to biomonitoring for obesity. Most (60%) dealt with condition management such as heart disease or asthma, 31% with complex conditions such as cancer, and 6% with wellness or health promotion. The authors rated 16% of the Value Drivers as providing “strong” or level 1 evidence, 50% as providing “moderate” or level 2 evidence, and 34% as providing “weak” or level 3 evidence for a utilization effect associated with the Driver. Six percent of Value Drivers were projected to increase costs; 20% generated savings above $10,000. Four presented savings within ranges. The authors found 6% of savings values to be overestimates, 20% to be underestimates, and 42% to be reflective of likely savings; in 32%, the direction of possible bias could not be determined.

Table 2.

Value Drivers by Rating Level from the External Review

Description of Value Driver	Rating of evidence by level	Assessment of bias
ACE inhibitor/ARB medication adhered to	1	cannot be determined
ACE inhibitor/ARB medication added to treatment plan	1	overestimate
ACE/ARB added to treatment plan (diabetes with microalbuminuria)	1	reflective of savings
Antiplatelet medication added (aspirin/dipyridamole)	1	reflective of savings
Bariatric surgery facility: out-of-network to in-network Center Of Excellence	1	underestimate
Beta-blocker medication added after MI to treatment plan	1	reflective of savings
Beta-blocker medication added to treatment plan	1	reflective of savings
Beta-blocker medication adhered to	1	reflective of savings
Blood pressure managed to target	1	reflective of savings
Cholesterol is optimally managed (No CC/MCC)	1	underestimate
Decision support for treatments for low back pain	1	reflective of savings
Discharge planning to prevent readmission	1	reflective of savings
Essential hypertension	1	reflective of savings
Fluid control achieved	1	underestimate
HbA1c improved by 1%–1.9%	1	reflective of savings
Heart failure program participation for patients with inpatient admission in the preceding 24 months because of heart failure	1	underestimate
Hypertension optimally managed—CKD	1	reflective of savings
Hypertension optimally managed—diabetes	1	cannot be determined
Infertility clinic shift: in-network to Center of Excellence switch	1	cannot be determined
Infertility clinic shift: out-of-network to in-network switch	1	reflective of savings
In-network non-Center of Excellence to in-network Center of Excellence facility switch	1	underestimate
LDL cholesterol optimally managed	1	reflective of savings
Out-of-network to in-network facility switch	1	reflective of savings
Statin medication added	1	reflective of savings
Switch to specialty infertility medications vendor	1	reflective of savings
Weight loss of 6%> to 10%< of body weight achieved among an overweight population	1	reflective of savings
AV fistual	2	cannot be determined
Bariatric surgery avoided	2	overestimate
Decision support for treatments for morbid obesity	2	reflective of savings
Decision support for patient considering knee replacement	2	reflective of savings
Immunization completed: flu (patients ages 50–64)	2	reflective of savings
Inhaled corticosteroids medication adhered to	2	cannot be determined
Medication added: 17 alpha-hydroxyprogesterone caproate for patients with previous preterm birth between 20–37 weeks	2	cannot be determined
Out-of-network to in-network provider shift: ancillary dialysis provider	2	reflective of savings
Acute symptoms/complications optimally managed for low back pain	3	cannot be determined
Assistance arranged: caregiver for patient with serious chronic disorder	3	cannot be determined
Assistance arranged: support for caregiver who is unable to cope or needs respite care	3	cannot be determined
Assistance arranged: transportation for medical appointments	3	cannot be determined
Asthma: action plan adhered to	3	cannot be determined
Cancer specialist added to treatment team	3	cannot be determined
Cancer: treatment plan adhered to	3	overestimate
Comorbidities of pregnancy optimally managed	3	underestimate
Condition rating (asthma): patients rate their condition based on impact to productivity at time of engagement and then reevaluate at program close	3	cannot be determined
Gestation diabetes treatment plan adhered to	3	underestimate
Hospice program enrollment completed	3	underestimate
Infertility counseling for adoption	3	cannot be determined
Initial education provided on pregnancy	3	Not Applicable
Inpatient admissions because of drug–drug interactions for patients taking both statins and cyclosporine	3	reflective of savings
Pain optimally managed or minimized	3	cannot be determined
Preemptive kidney transplant completed for patient with imminent need for dialysis	3	underestimate
Skilled nursing benefit extension to patients in institutional setting so they can transition to community	3	cannot be determined
Weight gain/stabilization achieved for seniors	3	underestimate

ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; AV, arteriovenous; CC, comorbid conditions; CKD, chronic kidney disease; LDL, low-density lipoprotein; MCC, multiple comorbid conditions; MI, myocardial infarction.

Five drivers, described in the following sections, were selected to illustrate Optum's approach in terms of types of interventions and sources of evidence. The first 2, adding a beta-blocker and case management post hospital discharge, represent the most common clinical service lines among the Value Drivers reviewed: cardiovascular health and general medical. The third involves shared decision making with a patient for whom Optum presented a savings range rather than a specific value. The final 2 address savings from a wellness program and from a community intervention to secure transportation for patients with transportation deficits. Each Value Driver reviewed takes a different approach in calculating savings.

Value Driver Examples

Adding a beta-blocker

The beta-blocker therapy (BBT) Value Driver aims to secure patient adherence to medication for those suffering from CHF. BBT is widely recognized and accepted as the standard of care for patients suffering from left ventricular dysfunction.¹⁹ A meta-analysis of 22 RCTs found that appropriate use of BBT can reduce rates of hospital admissions for CHF by 4%.¹⁶ Optum used this research to establish the link between the NCM intervention and utilization.

Optum's BBT Value Driver targets patients with CHF not currently on BBT. NCMs identify them by reviewing clinical and pharmaceutical claims data. The NCM then reviews medications with the patient, checks for contraindications, addresses prior problems with beta-blockers, educates the patient about the importance of medication compliance, and assists the patient with securing a physician appointment for evaluation and obtaining a prescription for the medication. The intervention also involves NCM feedback to physicians about this gap in care.

Optum estimated the net savings for successful BBT as the savings from avoiding an episode of care due to CHF, less the costs of medication and the physician visit associated with initiating therapy. ETG data for 4769 CHF patients were the basis for the episode cost. The ETG rate includes 5 cost areas: patient management, surgery, facility, ancillary, and pharmacy. The ETG estimate is $46,728 (2009 $) per CHF episode. This value was multiplied by 4% (the likelihood an episode is avoided), resulting in $1869 gross annual savings.

Offsetting these savings were the additional medication and physician visit costs. Internal pharmaceutical claims data from Optum indicated that BBT costs $0.65 per day on average with patient adherence estimated at 90% (Stephanie Shulman, MD, unpublished data, January 21, 2011). Additional annual medication cost $214 and the annual cardiologist visit added $187 making the total additional costs equal to $401. The resulting difference between the avoided hospitalizations ($1869) and the additional costs ($401) netted $1468 and represented the potential savings per CHF patient from BBT.

For this Value Driver, the utilization evidence was rated “strong” or level 1 (Table 1). Even though the time period of the study, 1966 to July 2000, is not current, it was judged to represent the best data available. The financial data reflected a large sample and the relevant costs offsetting savings were accounted for; thus, the savings estimate appeared to be reflective of likely medical cost savings.

While strong, the estimate raised some concerns. First, most participants in the meta-analysis were also receiving concurrent treatment with angiotensin-converting enzyme (ACE) inhibitors. The Value Driver did not attribute any of the potential improvement in patient outcomes to ACE treatment or address the possible need for concurrent ACE treatment to achieve this outcome. Also, the financial estimates reflected data from the UHC commercial population, which may be younger and have different comorbidities than older heart failure patients. Nevertheless, the authors accepted the estimate as representative, given the large sample size.

Discharge planning for high-risk patients

The Transitional Case Management (TCM) Value Driver addresses gaps in care for patients at high risk for readmission following a hospital stay. Readmission rates are known to be high among patients with certain conditions such as heart failure, or with certain characteristics such as having little home support. Research on preventing readmission, however, is limited. One small RCT has shown that reductions in rehospitalizations within 30, 60, and 90 days are possible through an intensive postdischarge program that includes home visits.²⁰ Through the Optum TCM program, NCMs contact patients who have screened to be at high risk of readmission. Nurses review discharge plans with patients and encourage them to be compliant with postdischarge instructions. NCMs also may contact providers to ensure that a comprehensive plan is in place. NCM activities may include assisting with securing prescriptions, setting up follow-up appointments, and arranging for home health services.

As external evidence of effectiveness for this intervention was lacking, Optum conducted an internal evaluation of the TCM program. It compared rehospitalization rates for a high-risk, matched, nonparticipant population to those for TCM participants. Analysts found that nonparticipants had a significantly higher risk of a 30-day readmission and that the TCM program reduced the rate of hospitalization by 5.6% from a baseline of 17.85% for high-risk patients. Based on an all-cause hospital cost estimate of $20,138, the reduced readmissions saved $1,133.

Because of the strength of the internal study, the authors rated the study evidence as “strong” and the savings likely to be reflective. Any care that takes place as a result of the NCM intervention, such as a recommended follow-up visit, likely would encourage visits sooner in time rather than prevent the need for such visits altogether and thus generate no additional costs.

Counseling consumers with chronic low back pain who are considering surgery

Low back pain (LBP) is a condition that can be treated in very divergent ways, from surgery to medication. In a wide range of cases of LBP, nonsurgical interventions, such as physical therapy, are known to be as effective as surgical intervention.²¹ The LBP Value Driver targets consumers primarily seeking surgery as a first option and attempts to steer them to less invasive options. It falls into a group of Value Drivers labeled treatment decision support. For this group the treatment options for the conditions are highly variable and can range from nonsurgical to surgical interventions.

For LBP, the NCM outlines the benefits and risks of back surgery and discusses the likely outcomes of back surgery compared to measures that are more conservative. As part of the Value Driver protocol, the nurse and patient may discuss the option of getting a second opinion and how the patient might discuss alternative plans with his/her treating physician.

To generate a net savings estimate, Optum compared the costs of possible patient treatment choices pre-counseling to the cost of possible treatment choices post-counseling, with a patient acting as his/her own control. Treatment options include medication, lifestyle changes, physical or exercise therapy, chiropractic care, injections, and surgery. Counseling can produce shifts from more to less invasive options, no change, and shifts from less to more invasive care.

Optum calculated the net savings of all possible pre/post intent choices. For example, based on UHC-FI median claims, a patient who initially planned to have back surgery but then switched to physical therapy for treatment of back pain after NCM counseling saved $34,334—the cost of surgery ($35,055) less the costs of physical therapy ($721). The figures were then adjusted for inflation. The authors rated the net savings, represented as a range, as “likely to be reflective” of possible savings. They rated the evidence as level 2 or “moderate” because the treatment choices were assumed to be equally likely and not weighted, such as more patients moving from intensive to less intensive treatment rather than vice versa. An alternative approach would be for Optum to weight combinations based on past utilization data to give an indication of the more likely choice pairs. However, not applying weightings did not bias the results.

Promoting weight reduction of 6%–10% in obese people

Optum's weight reduction Value Driver targets patients with a body mass index (BMI) of greater than 25 who are participating in a wellness program. As part of the wellness program, NCMs contact patients and provide information about the health risks of excess weight, the benefits of a proper diet, exercise strategies, lifestyle modification advice, and the benefits and risks of weight loss medications. NCMs also help establish weight loss targets, provide motivational messages, and refer patients to commercial weight loss programs, when appropriate.

For this Value Driver, Optum needed to document that the intervention produced weight loss and that the weight loss led to savings. To project plausible ranges of weight loss for program participants, Optum analyzed its prior experiences with telephonic weight loss counseling programs among UHC enrollees in Florida. Optum identified 7689 participants (about 23% of the total) who completed at least 1 phone call and lost weight. Of the 7689 participants, 6618 lost between 0% and 6% of their original body weight, 817 lost between 6% and 10%, and 254 lost more than 10%. The Value Driver estimated the cost savings associated with a 6% to 10% weight loss, which was equivalent to a 2.6 unit reduction in BMI units among the 817 participants who lost that amount.

Optum needed additional information to translate the weight loss from the intervention into savings. Wang et al. examined the association between BMI and health care spending among active and retired General Motors employees and spouses, including unionized enrollees.²² Unlike most other studies of the relationship between BMI and costs, which measure BMI categorically, Wang and associates reported the dollar value of each BMI unit change. They found an approximately $200 increase in costs for each unit increase in BMI.

Based on the Wang et al data, Optum estimated that annual costs would decline by $822 among overweight or obese enrollees who lost between 6% and 10% of their body weight. The analysis was a reasonable attempt to translate weight reductions into cost savings and received a “moderate” or level 2 rating. Like other analyses of its kind, the temporality of the relationship between weight and health and costs is unknown. Weight loss does not always lead to immediate improvements in health or decreases in costs.

Transportation assistance added

The transportation assistance added Value Driver aims to ensure that patients, such as seniors, have transportation available to them for timely medical care. Patients with transportation deficits may face barriers to accessing routine and preventive services and such delays may lead to avoidable ER use and/or hospitalizations. Securing adequate transportation for consumers may result in patients receiving more physician care and reductions in other types of health care use.

With this intervention, the NCM identifies patients who lack transportation to medical appointments and refers them to a social worker who helps them identify transportation options within their community. The Value Driver is closed when the NCM confirms that the patient has transportation options in place. For transportation assistance added, Optum estimated the net savings associated with securing appointment transport for those with obstacles based on the reduction in ER visits and ambulance use, offset by the cost of the increase in physician care.

The savings estimate was calculated in several steps using data from the literature and expert estimate. All dollar values were based on internal claims data. Ambulance savings were: average trip cost multiplied by the percentage of ambulance use estimated as unnecessary, multiplied again by the percentage of ambulance use related to a lack of transportation. The percentages were drawn from literature comprised primarily of chart reviews of or questionnaires from convenience samples.^23,24

The savings from the decline in ER visits was average ER visit cost multiplied by number of visits avoided because of having alternate transport, an expert estimate, and the percentage of ER visits estimated to be unnecessary, estimated via a case-control study.²⁵ The increase in physician costs was the price of an outpatient visit multiplied by the increase in visit numbers, also an expert estimate. The ambulance and ER savings ($1554) less the increase in physician visit costs ($149) was multiplied by the likelihood that a patient was willing to accept alternate transportation (80%).²⁶ This resulted in a final savings of $1124.

The concept behind the Driver is sound: eliminating transportation deficits may change patient behavior, direct them toward more appropriate primary care, and avoid costly ambulance and ER use. However, the analysis relies heavily on expert estimates, particularly regarding the increase in physician visits and number of ER visits avoided because of elimination of a transportation deficit, with no other supporting documentation. Also, success of this intervention depends on the willingness of individuals to accept alternative non-ambulance transportation, and care not provided in the ER. The Driver uses existing survey literature to assess the number of patients willing to make changes, but these estimates are based on what patients say they are willing to do rather than any evidence of them actually making changes. For these reasons, the Driver was rated as weak in terms of documenting a utilization effect.

Also, although claims data are an appropriate basis for the cost estimates, the data are based on the commercial population. It is not possible to identify the subset of this group who experience transportation deficits and it is unclear if this group would have higher/lower costs than the average beneficiary. In this case the direction of bias estimate was rated as unknown.

Discussion

This article describes the process used by Optum to develop a monetized Value Driver as an HDM tool and reviewed the approach by examining 5 Drivers in detail. Based on this work, including an external review, the authors find that the Value Driver methodology offers a sound approach to estimating the savings from a wide array of NCM interventions when these interventions are successfully implemented. The data here suggest that savings estimates derived from Value Drivers are generally reasonable to conservative. Also, the effect evidence accompanying each Driver is relatively strong, as the authors rated that evidence as Level 2 (moderate evidence of effectiveness) or above in 66% of the cases.

Major strengths of the approach are as follows. The savings estimated by Value Drivers are associated with specific interventions that delineate the necessary actions needed by clinicians and patients to generate positive health impacts. Extensive documentation supports the clinical validity of the intervention, as well as its effect on utilization. The Value Driver method accommodates assessment of a wide range of activities, from decision support related to hospice enrollment to altering insurance benefits to permit community-based services as an alternative to institutional care for certain patients. The standardized methodology also allows for relative rankings of Drivers and prioritization based on savings estimates.

The Value Driver approach also has limitations. The time frame for each analysis is 1 year. Although some interventions may show immediate effects, such as prenatal education, others, such as weight reduction, may not affect health care utilization and costs in the short run. The calculations do not address the impact on consumer productivity, including absenteeism, as a result of changes in health nor do they address the benefit of reducing mortality in the short or medium term. Also, data from UHC's commercial population were the primary source of claims data. Certain populations, such as seniors or those in the Medicaid Program, may have cost patterns that differ systematically from this group.

Importantly, Value Driver savings estimates do not take into account uncertainty related to their “real-world” adoption and successful implementation. Generally, savings are presented as point estimates with no indication of variability. In addition, estimates of savings from Value Drivers assume that all parties involved have carried out the necessary steps described by the intervention to bring about the health effect. The parties include, but are not limited to, physicians, nurses, and patients. The steps can be numerous and challenging, such as correctly identifying patients for the intervention, securing patient compliance with the intervention, carrying out timely follow-up, and maintaining contact with patients.

Specificity, while a strength of Value Drivers, is also a limitation. As is common in the literature, Value Drivers evaluate single interventions under well-defined circumstances. They do not assess interventions in combination. In reality, patients may present with multiple conditions, such as obesity and high blood pressure, and may be candidates for more than 1 intervention. Certain interventions may be synergistic. No evidence exists to assess to what degree, if any, the savings from Value Drivers are additive.

A clear contribution to the HDM literature, the methodology described here provides a valuable framework for estimating financial impacts of evidence-based interventions that are founded on credible research and insurance claims data. Moving from the “ideal scenario” to one that involves “real-world” providers and patients is necessary to field-test these assumptions. Further research is needed to determine the validity of the Value Driver approach as a leading indicator of cost savings when compared to actual medical cost savings for the programs when implemented. Until then, the Value Driver methodology provides a useful tool for estimating the value delivered from interventions that are evidence based and lead to closing gaps in care.

Footnotes

Acknowledgments

The opinions expressed in this work are solely the authors' and do not necessarily represent the opinions of Emory University or Optum. The authors thank Ron Goetzel, PhD and Enid Chung Roemer, PhD of the Emory University Institute for Health and Productivity Studies for their support and contributions to this article.

Author Disclosure Statement

Drs. Phillips, Becker, and Howard declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors received the following financial support for the research, authorship, and/or publication of this article: This study was supported with funding from Optum.

References

Aldana

, Merrill

, Price

, Hardy

, Hager

. Financial impact of a comprehensive multisite workplace health promotion program. Prev Med, 2005, 40:131–137.

Care Continuum Alliance. Care Continuum Alliance definition of disease management. Available at: http://www.carecontinuumalliance.org/dm_definition.asp. Accessed January 21, 2013 .

Krause

. Economic effectiveness of disease management programs: A meta-analysis. Dis Manag, 2005; 8:114–134.

Kahn

, Robertson

, Smith

, Eddy

. The impact of prevention on reducing the burden of cardiovascular disease. Circulation, 2008; 118:576–585.

Weingartern

, Henning

, Badamgarav

, et al. Interventions used in disease management programmes for patients with chronic illness. Which ones work? Meta-analysis of published reports. BMJ, 2002:325:925–939.

Ofman

, Badamgarav

, Villagra

, Duffy

. Does disease management improve clinical and economic outcomes in patients with chronic disease? A systematic review. Am J Med, 2004; 117:182–192.

Congressional Budget Office. An analysis of the literature on disease management programs. Washington, DC: Congressional Budget Office; 2004.

Diamond

. DM's cost-effectiveness doubted in CBO report. Manag Care, 2004; 12:31–32, 35.

Linden

. What will it take for disease management to demonstrate a return on investment? New perspectives of an old theme. Am J Manag Care, 2006; 12:217–222.

10.

Linden

, Adams

, Roberts

. An assessment of the total population approach for evaluating disease management program effectiveness. Dis Manag, 2003; 6:93–102.

11.

Goetzel

, Ozminkowski

, Villagra

, Duffy

. Return on investment in disease management: A review. Health Care Financ Rev, 2005; 26:1–19.

12.

McCall

, Cromwell

. Results of the Medicare health support disease management pilot program. N Engl J Med, 2011; 365:1704–1712.

13.

Wilson

. Evaluating ROI in state disease management programs. State Coverage Initiat Issue Brief, 2003; 4:1–6.

14.

Care Continuum Alliance. Outcomes Guidelines Report, Volume 5. Washington, DC: Care Continuum Alliance; 2010.

15.

Lynch

, Chen

, Bender

, Edington

. Documenting participation in an employer-sponsored disease management program: Selection, exclusion, attrition, and active engagement as possible metrics. J Occup Environ Med, 2006; 48:447–454.

16.

Brophy

, Joseph

, Rouleau

. Beta-blockers in congestive heart failure. A Bayesian meta-analysis. Ann Intern Med, 2001; 134:550–560.

17.

Ingenix, Inc. Episode Treatment Groups. Decatur, GA: Ingenix, Inc.; 2008.

18.

Ryan

, McPeck

, Chapman

. Evidence-based programming: A practitioner's guide. Am J Health Promot, 2011; 25:TAHP1–TAHP9.

19.

Hunt

; American College of Cardiology; American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). ACA/AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). J Am Coll Cardiol, 2005:46:e1–e82.

20.

Coleman

, Parry

, Chalmers

, Min

. The care transitions intervention: Results of a randomized controlled trial. Arch Int Med, 2006; 166:1822–1828.

21.

Fourney

, Dettori

, Hall

, Hartl

, McGirt

, Daubs

. Systematic review of clinical pathways of lower back pain and introduction of the Saskatchewan spine pathway. Spine (Phila Pa 1976), 2011; 36:S164–S171.

22.

Wang

, McDonald

, Bender

, Reffitt

, Miller

, Edington

. Association of healthcare costs with per unit body mass index increase. J Occup Environ Med, 2006; 48:668–674.

23.

Billittier

, Moscati

, Janicke

, Lerner

, Seymour

, Olsson

. A multisite survey of factors contributing to medically unnecessary ambulance transports. Acad Emerg Med, 1996; 3:1046–1052.

24.

Camasso-Richardson

, Wilde

, Petrack

. Medically unnecessary pediatric ambulance transports: A medical taxi service?. Acad Emerg Med, 1997; 4:1137–1141.

25.

Grumbach

, Keane

, Bindman

. Primary care and public emergency department overcrowding. Am J Public Health, 1993; 83:372–378.

26.

Yarris

, Moreno

, Schmidt

, Adams

, Brooks

. Reasons why patients choose an ambulance and willingness to consider alternatives. Acad Emerg Med, 2006; 13:401–405.