CME AVAILABLE FOR THIS ARTICLE AT ACOEM.ORG

Can Health Insurance Improve Employee Health Outcome and Reduce Cost? An Evaluation of Geisinger’s Employee Health and Wellness Program Daniel D. Maeng, PhD, James M. Pitcavage, MSPH, Janet Tomcavage, RN, MSN, and Steven R. Steinhubl, MD

Objective: To evaluate the impact of a health plan–driven employee health and wellness program (known as MyHealth Rewards) on health outcomes (stroke and myocardial infarction) and cost of care. Methods: A cohort of Geisinger Health Plan members who were Geisinger Health System (GHS) employees throughout the study period (2007 to 2011) was compared with a comparison group consisting of Geisinger Health Plan members who were non-GHS employees. Result: The GHS employee cohort experienced a stroke or myocardial infarction later than the non-GHS comparison group (hazard ratios of 0.73 and 0.56; P < 0.01). There was also a 10% to 13% cost reduction (P < 0.05) during the second and third years of the program. The cumulative return on investment was approximately 1.6. Conclusion: Health plan–driven employee health and wellness programs similarly designed as MyHealth Rewards can potentially have a desirable impact on employee health and cost.

F

or several years, employers have seen significant increases in the cost of providing health insurance benefits for their employees.1 Consequently, employees have seen steady increases in their premiums and out-of-pocket cost in the form of higher copayments and deductibles. These rising costs are threatening the viability of the current employer-based private health insurance market. Health plans have used traditional methods to manage populations, including medical management strategies and high-cost case management. Such traditional methods, however, entail two important limitations: First, these are often reactive rather than proactive; that is, health plans are unable to identify and manage high-risk patients until some kind of a trigger event (eg, hospitalization) takes place. Second, these offer no incentive for moderate- to high-risk patients to engage in their own health and alter their health behaviors. In response, employers and health plans are evaluating methods to improve the delivery of health care benefits to reduce costs and improve health outcomes.2 This has resulted in an increase in employee wellness programs.3,4 To date, numerous innovative employee health and wellness programs have been implemented in various settings,5 but limited evidence exists to show long-term impact on population health outcome and cost of care.6 The challenge of workplace promotion of health and wellness is likely to be emphasized in near future because of the Patient Protection and Affordable Care

Geisinger Health System (Drs Maeng and Steinhubl and Mr Pitcavage) and Geisinger Health Plan (Ms Tomcavage), Danville, Pa. All authors are employees of Geisinger Health System, and the work was done as a part of their employment with Geisinger Health System. The authors declare no conflicts of interest. Authos Maeng, Pitcavage, Tomcavage, and Steinhubl have no relationships/ conditions/circumstances that present potential conflict of interest. The JOEM Editorial Board and planners have no financial information related to this research. Address correspondence to: Daniel Dukjae Maeng, PhD, Geisinger Health System, 100 N Academy Ave, MC 44-00, Danville, PA 17822 (ddmaeng@geisinger .edu). C 2013 by American College of Occupational and Environmental Copyright  Medicine DOI: 10.1097/JOM.0000000000000009

Learning Objectives

r Summarize the components of MyHealth Rewards, the prir r

vate health plan-driven employee health and wellness program evaluated by the current study. Outline the effects of MyHealth Rewards on the cardiovascular event rate and health costs. Discuss the implications for employee health and wellness programs, including the effects of participation rate on return on investment.

Act that calls for an expansion of employer-based employee wellness programs as a part of the health care reform implementation.3,7 In this study, we examine a version of a private health plan– driven employee health and wellness program known as MyHealth Rewards (MHR). MyHealth Rewards was developed by Geisinger Health Plan (GHP) in collaboration with Geisinger Health System (GHS) and implemented among employees of GHS for the purposes of improving health outcomes and reducing the cost of care. We hypothesize that MHR is associated with improved long-term health outcomes, specifically stroke and myocardial infarction (MI), as well as lower cost of care, and test this hypothesis by comparing 5 years of claims data between GHS and non-GHS employee cohorts.

BACKGROUND Beginning in 2007, GHS has implemented a health and wellness program for its employee population referred to as MHR in collaboration with GHP. Geisinger Health Plan provides health coverage for about 65% of GHS employees. Geisinger Health System is an integrated health system located in Central and Northeast Pennsylvania, serving about 2.6 million residents. Geisinger Health System currently employs approximately 16,000 employees, about 10% of whom are physicians and 17% are nurses. Geisinger Health Plan is a full-service regional health insurer that is a subsidiary of GHS, covering roughly a quarter million lives in its service areas that currently include parts of Pennsylvania, West Virginia, and Maine. MyHealth Rewards is characterized by the following key features: 1) Establishment of a Geisinger Wellness group. Starting in July 2006, GHS started the employee wellness department, hiring a team that includes health educators, dieticians, and project managers to coordinate employee wellness programs. 2) Health risk assessment. Geisinger Health Plan used a third-party Web-based questionnaire to help employees identify areas of opportunity for self- improvement and develop health and wellness goals on their own or in conjunction with their health care team. 3) Medications for hypertension, high cholesterol, and diabetes, with $0 copay to the employee. Geisinger Health System employees with GHP coverage were eligible to receive free medications from a list of approximately 200 generic and brand name drugs designated for high blood pressure, cholesterol, and diabetes

JOEM r Volume 55, Number 11, November 2013 Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

1271

JOEM r Volume 55, Number 11, November 2013

Maeng et al

management after meeting the benefit deductible. The number of drugs included in the zero-copay drug list modestly increased over time. For instance, in 2007, there were 12 generic medications and 30 brand name drugs included in the zero-copay drug list for diabetes. By 2011, there were 13 generic medications and 33 brand name drugs included for the same condition. 4) Structured disease management programs coupled with financial incentives to participate. Employees with a confirmed diagnosis of diabetes, hypertension, coronary artery disease, obstructive pulmonary disease, chronic kidney disease, and/or heart failure were eligible to participate in evidenced-based disease management programs with nurses hired and supervised by GHP. All disease management programs focused on several strategies, including self-management training, nutrition and physical activity interventions, medication management in collaboration with the employee’s primary care providers, and acute exacerbation management. Eligible employees could elect to receive a $200 enrollment incentive payment to participate in the health management program, an additional $200 at 6-month point with goal attainment, and yet another $200 after 1 year of goal maintenance. Between 2007 and 2011, participation in MHR was strictly voluntary. All GHS employees and their dependents covered by GHP were eligible to participate in the health risk assessment programs. Nevertheless, as described earlier, only those who had one or more of the selected chronic conditions were eligible to participate in the zero-copay program and the health management/financial incentive program. The study sample, therefore, included employees as well as their dependents enrolled in GHP during the study period. There are two related questions that need to be answered in this context: First, what was the impact of offering MHR to GHS employees in comparison with non-GHS employees who were never offered the program (ie, intention to treat)? Second, what was the impact of participating in MHR relative to nonparticipation (ie, treatment effect)? In this article, we focus on answering the first question for two reasons: First, we believe it is the more relevant question from the perspective of employers potentially interested in adopting a similarly designed intervention. Because each employee’s participation in MHR was strictly voluntary, the intervention that can be replicated by other employers is offering of the program to all its employees. Second, it is possible that the impact of offering MHR was not strictly limited to just those who participated in MHR. To the extent that even nonparticipating employees were exposed to the wellness efforts and potential behavioral changes of their peers, there might have been a “spillover” effect of MHR even among the nonparticipants.

DATA We obtained claims data from GHP, covering a 7-year period from 2005 through 2011. Our data, therefore, contained 2 years of per-member-per-month claims experience before the implementation of MHR (2005 and 2006) and 5 years of claims experience after the implementation (2007 to 2011). Because MHR was available only to GHS employees, employees from non-GHS employers served as the comparison group. The potential limitation of this choice of comparison group is that the GHS employees may be systematically different from non-GHS employees; that is, GHS employee population may have different patterns of health care use and health status compared with non-GHS employees. Furthermore, this potential bias may be exacerbated by the fact that, over time, some non-GHS employees might have become GHS employees at some point, possibly attracted by the health insurance benefits offered by GHS. Therefore, to minimize this potential bias, we applied the following inclusion criteria to our sample: (1) did not change employment between GHS and non-GHS employers throughout the study period; (2) appeared in all 7 years of this study period as having one 1272

of GHP’s commercial, non–Medicare Advantage Plan types (to eliminate confounding variables due to coverage gaps and/or medicare coverage); and (3) received primary care from one of Geisingerowned primary care clinics (to control for any confounding provider effect) throughout the study period. In addition, as described hereafter, we used a propensity score–matching method on the basis of a set of baseline characteristics to capture differences in other observable characteristics. The main outcome variables of interest were the following: (1) time to first claims related to stroke or MI since the inception of MHR in 2007 (measured in months) and (2) total cost of care. We did not consider death as an outcome, because we are unable to determine the cause of death from our claims data and, thus, unable to exclude deaths due to unrelated events, such as trauma. Moreover, our inclusion criterion that only those who remained GHP members throughout the study period were to be included in the study sample precluded using death as an outcome variable. If a member was deceased during the study period, he or she would be considered “disenrolled” from GHP and, therefore, be excluded from the sample. This, therefore, ensured that all individuals in this study sample were alive at least during the study period. Total medical cost of care was defined as the “allowed” amounts—the sum of all payments made by GHP to all providers plus the member out-of-pocket expenses, such as copays and deductibles—incurred in each given 3-month period (quarter). We calculated total cost on a per-member-per-quarter basis as opposed to a per-member-per-month basis, because quarterly total costs showed less variability and, therefore, improved the precision of our estimates. In addition, we considered only medical cost—that is, the cost of prescription drugs was not included in calculating the total cost. Because one of the key features of MHR was the provision of zero-copay prescription drugs for qualifying chronic conditions, it was anticipated that MHR would increase drug adherence, thereby increasing the overall drug costs. By excluding the drug cost from our analysis, therefore, we focused on examining whether MHR was associated with cost reduction due to lower use of medical care and services, such as inpatient admissions, emergency department visits, specialty visits.

METHODS We hypothesized that, compared with non-GHS employees who were never exposed to MHR, GHS employees would experience delayed incidence of stroke or MI and incur lower total cost of care. As mentioned earlier, propensity score matching was used to stratify individuals in our sample into mutually exclusive strata based on the following set of baseline characteristics: gender, age, and total cost of care as defined earlier during the preintervention period. This method yielded seven strata; within each stratum, the intervention group (GHS employees) and comparison group (nonGHS employees) were balanced in terms of the propensity scores and the covariates. As described hereafter, this allowed for stratified analyses that accounted for the baseline differences across individuals in our sample. To estimate the differences in time to first claims related to stroke or MI between GHS and non-GHS employees, Cox proportional hazard model was used. The key explanatory variable was the binary indicator variable for whether each individual was a GHS employee or not. Other covariates for the Cox model included age, gender, and indicator variables for whether the individual was under GHP’s disease or case management programs (offered to both GHS and non-GHS employees) in a given period, whether he or she was in one of Geisinger’s medical homes, and whether the individual had one or more of the following chronic conditions: chronic kidney disease, coronary artery disease, chronic obstructive pulmonary disease, hypertension, diabetes, cancer, and depression.

 C 2013 American College of Occupational and Environmental Medicine

Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

JOEM r Volume 55, Number 11, November 2013

Employee Wellness Program Impact on Health and Cost of Care

TABLE 1. MyHealth Rewards Participation 2007–2011

Year

% Participating in MyHealth Rewards

% Receiving Incentive Payments*

% Receiving Free Drug*

2007 2008 2009 2010 2011

12.0 15.7 16.7 16.3 14.8

96.9 78.2 76.1 77.0 74.5

82.9 83.2 85.5 87.3 88.2

*Among Geisinger Health System employees who participated in MyHealth Rewards for the year.

Moreover, we allow the baseline hazard function to differ, depending on each individual’s baseline characteristics, by stratifying the estimation by the seven strata obtained via the propensity scores mentioned earlier. This approach relaxes the assumption that the baseline hazard function is constant across all individuals in the sample, which is unlikely to be a reasonable assumption in this particular context because of the nonrandom selection of individuals into the intervention group (GHS employees) and the control group (non-GHS employees) before the intervention period. This implies that individuals in the sample may have different baseline hazard functions that depend on their preintervention characteristics. Our model, therefore, explicitly recognizes this possibility by allowing individuals to have different hazard functions across the strata, but within each stratum, the individuals are assumed to have the same hazard function. To estimate the differences in total cost of care between GHS and non-GHS employees, we used a two-part model in which the expected total cost conditional on nonzero claims and the probability of incurring nonzero claims in a given quarter were modeled and estimated separately.10 This two-part model was necessary because about 40% of the observations had zero cost, thereby leading to a skewed distribution of the cost data. In the first part, the probability of incurring nonzero cost (total cost greater than $0) during a given quarter was estimated using a logistic regression model. In the second

TABLE 2. Baseline Characteristics* Baseline Characteristics (2005–2006)

Non-GHS (n = 12,077)

GHS (n = 4,895)

Cost ($ per-member-per-quarter) Had myocardial infarction Had stroke Female Age as of 2006, yr Enrolled in disease/case management Has chronic kidney disease Has coronary artery disease Has chronic obstructive pulmonary disease Has hypertension Has diabetes Has asthma Has cancer Has depression

995 (2,763) 0.8% 1.6% 50.7% 45 (9) 1.7% 0.2% 2.6% 1.0% 18.0% 5.5% 5.0% 1.9% 7.4%

1,118 (3,309) 1.0% 1.4% 58.3% 44 (9) 1.9% 0.2% 2.3% 0.6% 15.1% 4.4% 5.8% 2.2% 7.0%

Standard deviations shown in parentheses. GHS, Geisinger Health System.

FIGURE 1. Estimated survival function: stroke. part, the expected total cost conditional on incurring nonzero cost was estimated using a generalized linear model with log link function and gamma distribution. The expected total cost was then obtained as the product of the estimated probability and the expected total cost. The covariates included in this two-part cost model were essentially identical to those included in the Cox model, except that a set of indicator variables for the calendar year (to capture yearly trends) and the propensity score strata variables (to account for the baseline differences) were also included. To allow the effect of MHR to change over time, a set of interaction effects between the GHS employee indicator variable and each of the calendar year indicator variables was included as well. Finally, to estimate the expected difference in total cost between the GHS and non-GHS groups in dollar terms, we calculated regression-adjusted observed cost for each member of the GHS employee cohort in each period and the corresponding regression-adjusted expected cost with the GHS employee indicator variable and all its interaction terms set to zero. The latter represented the hypothetical counterfactual in which the GHS employee cohort had never been exposed to MHR (ie, the nonGHS comparison group during the study period). We then obtained the average difference across the GHS employee cohorts. The corresponding standard error and the 95% confidence intervals were obtained using a bootstrap method with 100 replications. On the basis of the estimated cost savings, we also estimated return on investment (ROI) between 2007 and 2011. Return on investment was calculated as a ratio between the cost savings associated with MHR (return or the numerator) and the total cost of implementing MHR (investment or the denominator) in each year. The cost of implementing MHR included the foregone copay (ie, copayments that GHS employees would have had to pay in the absence of MHR), drug acquisition cost, incentive payment amounts, personnel cost, and the cost of on-line health management tools. A ROI greater than one would indicate that the returns from the MHR implementation had exceeded the cost of implementing it.

FIGURE 2. Estimated survival function: myocardial infarction.

 C 2013 American College of Occupational and Environmental Medicine

Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

1273

JOEM r Volume 55, Number 11, November 2013

Maeng et al

TABLE 3. Estimated Cost Impact and Return on Investmenta Year 2007 2008 2009 2010 2011 All Years

GHS ($ PMPQ)

Non-GHS ($ PMPQ)

Difference

% Difference

Return on Investment

1,036 1,315 1,599 1,967 2,251 1,634

1,004 1,461 1,839 2,003 2,209 1,703

32 (−42 to 107) − 146* (−254 to −37) − 240* (−386 to −94) − 36 (−211 to 139) 42 (−184 to 267) − 70 (−215 to 76)

3.21 (−3.97 to 11.23) − 9.98* (−16.39 to −2.72) − 13.05* (−19.65 to −5.49) − 1.8 (−9.82 to 7.46) 1.89 (−7.72 to 13.12) − 3.94 (−11.51 to 4.72)

− 0.91 (−2.99 to 1.18) 3.47 (0.88 to 6.05) 5.25b (2.06 to 8.44) 0.82 (−3.18 to 4.83) − 1.05 (−6.71 to 4.61) 1.52 (−1.99 to 5.02)

*P < 0.05. a The values given in parentheses are 95% confidence intervals. b Return on investment is statistically greater than 1 at 5% level. GHS, Geisinger Health System; PMPQ, per-member-per-quarter.

RESULTS Table 1 shows the proportion of the GHS employee cohort in each calendar year since 2007 who participated in MHR as well as the proportions of those participating in MHR who received any of the zero-copay drugs and any incentive payments in each year. The proportion of the GHS employee cohort in this study participating in MHR during the first year (2007) was 12%; then, it reached its peak in 2008 at 17%, dropping to about 15% by 2011. In addition, it shows that while the proportion of those MHR participants who received any zero-copay drugs increased slightly from around 83% to 89% over time, the proportion of those receiving any financial incentives declined during the same period, from 97% in 2007 to about 75%. One likely reason for the steady increase in the proportion of MHR participants receiving zero-copay drugs is expansion of the list of zero-copay drugs over time. Table 2 shows the baseline difference between the GHS employee cohort and the non-GHS employee comparison group, using their claims data from 2005 and 2006. Note that even after applying a rather stringent set of inclusion criteria, the final sample size in each group remained substantial, with 4898 GHS employees and 12,077 non-GHS employees. Table 2 indicates that the GHS employee cohort had a slightly larger proportion of women than the non-GHS employee group (58.3% vs 50.7%) and slightly higher per-memberper-quarter total cost during the baseline period ($1118 vs $995). The non-GHS employee group, on the contrary, seems to have had a slightly higher prevalence of hypertension (18.0% vs 15.1%) and diabetes (5.5% vs 4.4%). Yet, the two cohorts are quite similar in terms of the age distribution. Recall that a critical hypothesis proposed earlier in this article was focused on positive health outcomes from the MHR program design. Figures 1 and 2 show the estimated survival functions for the GHS and the non-GHS employee groups on the basis of the Cox model parameter estimates. As hypothesized, GHS employees have a consistently higher probability of an event-free outcome (ie, not experiencing an adverse event) at each period of time since the start of the MHR program in 2007 than the non-GHS employee comparison group, either in terms of stroke or MI. The estimated hazard ratios associated with GHS employee status were 0.73 (95% confidence interval: 0.60 to 0.90; P = 0.002) for stroke and 0.56 (95% confidence interval: 0.40 to 0.79; P = 0.001) for MI. Table 3 compares the regression-adjusted expected total costs in each year since the start of MHR between the GHP and nonGHP employee groups. This suggests that the magnitude of the observed reductions in cost of care changed over time. In particular, during the second and third years since the inception MHR (2008 and 2009), there are statistically significant cost savings of 10% and 13%, respectively, associated with the GHS employee cohort, relative to the non-GHS employee group. Note that these significant cost savings were observed despite the fact that the GHS employee cohort seemed 1274

to have incurred higher total cost at the baseline, as shown in Table 1. Nevertheless, starting in 2009, these cost savings seem to have eroded in subsequent years. Table 3 also shows the estimated ROI in each year since MHR’s initial implementation in 2007. Only in 2009, the yearly ROI estimate was statistically greater than one. Overall, the cumulative ROI observed during the entire study period was 1.6, although it was not statistically different from one.

DISCUSSION The earlier-mentioned results are the first such results that we are aware of that demonstrate both positive health outcomes and cost benefit of a health plan–driven employee health and wellness program in a large employee cohort. By providing employees the opportunity to receive proven preventative medications without a copay requirement and increasing access to evidence-based health management services backed with a financial incentive to participate, MHR is shown to be associated with significantly lower incidences of stroke and MI over the next 5 years in that population. In addition, our cost analysis indicates that the cost of care was significantly reduced by approximately 10% relative to matched controls in 2 of the 5 years of the program, leading to an overall cumulative ROI of about 1.6 during the entire study period. It is important to note that these population-wide benefits were realized despite only a minority of patients actively participating in the program. Our data suggest that never more than 17% of GHS employees voluntarily participated in MHR annually. This, therefore, implies that MHR is likely to have impacted the most appropriate individuals—that is, employees at highest risk of adverse health outcomes and, thus, account for a significant portion of the total cost of care. This is consistent with findings in previous studies,8,9 which suggest that focused care strategies on the higher risk segments of a population can reduce total cost for a whole population. Why the cost savings did not continue in later years is not clear. It is interesting, however, that the greatest cost saving was in the year with highest enrollment in MHR. There are two possible explanations: First, the decreasing enrollment in the MHR and, thus, in the disease management program might have led to the erosion in the cost savings. Alternatively, there may have been increases in the unit cost of services and care rendered by certain providers because of changing negotiated contract arrangements between GHP and the providers. Unfortunately, our data do not allow us to account for such unit cost changes over time. In addition, it is not clear why the proportion of individuals receiving any financial incentives dropped from 95% in the first year to 75% of the participants 4 years later. Because the program design had remained unchanged during this period, except for the expansion of the list of drugs eligible for zero-copay, one possible explanation is that some of those individuals who had signed up to receive the bonus payments during the early years of MHR chose not to continue

 C 2013 American College of Occupational and Environmental Medicine

Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

JOEM r Volume 55, Number 11, November 2013

in subsequent years. This may have happened either because they had already achieved their goal and maintained it, thus avoiding the need for additional interventions, or they had tried the program once and realized that the program did not meet their needs and, therefore, saw no need to continue in the program. Further research is necessary to explore these possibilities. Despite the statistical adjustments to account for the confounding factors, our findings need be interpreted with caution because they were based on observational data and, therefore, might still be subject to unobserved biases. In particular, it is possible that GHS employees are systematically different from their non-GHS counterparts in terms of their skills and knowledge for self-management, as well as preference for healthy behavior and lifestyles, which we were unable to capture from our data. A recent white paper suggests that workers in the health care sector may be generally more likely to be sicker and be higher users of health care than average workers.11 This is consistent with our baseline comparison between GHS and non-GHS employees, as shown in Table 1. As such, it is possible that health care workers such as GHS employees may be more amenable to workplace wellness programs and interventions than average workers. Although there is no evidence to suggest that such bias exists in our current data, this is a potential limitation of this study.

REFERENCES 1. Yu H, Dick AW. Impacts of rising health care costs on families with employment-based private insurance: a national analysis with state fixed effects. Health Serv Res. 2012;47:2012–2030. 2. Choudhry NK, Rosenthal MB, Milstein A. Assessing the evidence for valuebased insurance design. Health Aff (Millwood). 2010;29:1988–1994.

Employee Wellness Program Impact on Health and Cost of Care

3. Health Policy Brief: Workplace Wellness Programs. Health Aff (Millwood). Updated December 4, 2012. Available at: http://www.healthaffairs. org/healthpolicybriefs/brief.php?brief id=93. Accessed October 16, 2013. 4. Fendrick AM, Edlin ML. Value-Based Insurance Design Landscape Digest. Washington, DC: National Pharmaceutical Council; 2009. Available at: http://www.npcnow.org/system/files/research/download/ValueBasedInsurance-Design-Landscape-Digest-7-2009_0.pdf. Accessed October 16, 2013. 5. Pelletier K. A review and analysis of the clinical and cost-effectiveness studies of comprehensive health promotion and disease management programs at the worksite: update VII 2004–2008. J Occup Environ Med. 2011;53: 1310–1331. 6. Abraham JM, Nyman JA, Feldman R, Barleen N. The effect of participation in a fitness rewards program on medical care expenditures in an employee population. J Occup Environ Med. 2012;54:280–285. 7. de Souza JA, Ratain MJ, Fendrick AM. Value-based insurance design: aligning incentives, benefits, and evidence in oncology. J Natl Compr Canc Netw. 2012;10:18–23. 8. Pelletier K. A review and analysis of the clinical- and cost-effectiveness studies of comprehensive health promotion and disease management programs at the worksite: 1998–2000 update. Am J Health Promot. 2001;16:107– 116. 9. Pelletier K. A review and analysis of the clinical and cost-effectiveness studies of comprehensive health promotion and disease management programs at the worksite: update VI 2000–2004. J Occup Environ Med. 2005;47:1051– 1058. 10. Buntin MB, Zaslavsky AM. Too much ado about two-part models and transformation? Comparing methods of modeling medicare expenditures. J Health Econ. 2004;23:525–542. 11. Taylor M, Bithoney WG. 10 Steps to Developing a Culture of Health for Hospital and Health System Employers. White Paper. Truven Health Analytics; 2012. Available at: http://www.nghealthcaresummit.com/media/whitepapers/ Truven - 10 Steps To Culture of Health.pdf. Accessed October 17, 2013.

 C 2013 American College of Occupational and Environmental Medicine

Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

1275

Can health insurance improve employee health outcome and reduce cost? An evaluation of Geisinger's employee health and wellness program.

To evaluate the impact of a health plan-driven employee health and wellness program (known as MyHealth Rewards) on health outcomes (stroke and myocard...
159KB Sizes 0 Downloads 0 Views