a sociology essay

By John P. Caloyeras, Hangsheng Liu, Ellen Exum, Megan Broderick, and Soeren Mattke

Managing Manifest Diseases, But Not Health Risks, Saved PepsiCo Money Over Seven Years

ABSTRACT Workplace wellness programs are increasingly popular. Employers expect them to improve employee health and well-being, lower medical costs, increase productivity, and reduce absenteeism. To test whether such expectations are warranted, we evaluated the cost impact of the lifestyle and disease management components of PepsiCo’s wellness program, Healthy Living. We found that seven years of continuous participation in one or both components was associated with an average reduction of $30 in health care cost per member per month. When we looked at each component individually, we found that the disease management component was associated with lower costs and that the lifestyle management component was not. We estimate disease management to reduce health care costs by $136 per member per month, driven by a 29 percent reduction in hospital admissions. Workplace wellness programs may reduce health risks, delay or avoid the onset of chronic diseases, and lower health care costs for employees with manifest chronic disease. But employers and policy makers should not take for granted that the lifestyle management component of such programs can reduce health care costs or even lead to net savings.

W orkplace health and wellness programs are becoming an increasingly common work- place benefit in the United States. The recently pub-

lished RAND Workplace Wellness Programs Study found that about half of employers with at least 50 employees and more than 90 percent of those with more than 50,000 employees of- fered a wellness program in 2012.1 In general, wellness programs screen employees and some- times their dependents to identify health risks, provide interventions to address health risks and manifest disease, and promote healthy lifestyles. A wellness program’s specific program compo- nents (for example, lifestyle management to pro- mote healthy living habits or disease manage- ment to help employees manage a chronic condition or illness) and interventions (for in-

stance, on-site exercise classes) vary across em- ployers, with larger employers more likely than smaller employers to offer more-elaborate pro- grams that combine a variety of components and interventions.2

The popularity of wellness programs is driven by employers’ expectation that the programs im- prove employee health and well-being, lower medical costs, increase productivity, and reduce absenteeism. For instance, a 2011 Automatic Data Processing (ADP) survey of employers with at least 1,000 employees found the four most commonly cited reasons for offering a wellness program to be “improve employee health,” “con- trol health care costs,” “increase productivity,” and “reduce absenteeism” (cited by 78 percent, 71 percent, 42 percent, and 43 percent of employ- ers, respectively).3 Furthermore, 43 percent of employers responding to a 2012 Deloitte survey

doi: 10.1377/hlthaff.2013.0625 HEALTH AFFAIRS 33, NO. 1 (2014): 124–131 ©2014 Project HOPE— The People-to-People Health Foundation, Inc.

John P. Caloyeras is a doctoral fellow at the Pardee RAND Graduate School and an assistant policy analyst at the RAND Corporation in Santa Monica, California.

Hangsheng Liu is a policy researcher at the RAND Corporation in Boston, Massachusetts.

Ellen Exum is the director of global wellness at PepsiCo, in Purchase, New York.

Megan Broderick is the senior director of health and welfare benefits at PepsiCo, in Purchase, New York.

Soeren Mattke (mattke@rand .org) is a senior scientist at the RAND Corporation and the managing director of RAND Health Advisory Services, RAND Health’s consulting practice, in Boston, Massachusetts.

124 Health Affairs January 2014 33:1

Employee Wellness Program

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

said they believed that investments in wellness programs offered high levels of value to the over- all health care system per dollar spent on them.4

The popularity of wellness programs is ex- pected to continue to grow. A 2011 Aon Hewitt survey found that among employers without a health improvement or wellness program, 47 percent planned to add such a program in 2012, with an additional 47 percent reporting they may add such a program in the next three to five years.5 Employers with programs want to get more employees participating in them: 70 percent of employers in the Aon Hewitt survey identifiedincreasinguse of wellnessprogramsas a top priority.5 The Affordable Care Act also has several provisions to promote workplace well- ness. For example, section 4303 of the act estab- lishes a technical assistance role for the Centers for Disease Control and Prevention to provide tools and resources to assist employers with planning, implementing, and evaluating well- ness programs. Employers’ optimism regarding the benefits of

wellness programs is driven by countless success stories in the popular press and trade publica- tions and by studies in the peer-reviewed litera- ture that have largely concluded that wellness programs save money and are a good bet for employers looking to lower health care costs. The evidence for the prevailing wisdom to-

day—that wellness programs can reduce health care costs and absenteeism in excess of program costs—has been established by several reviews.6,7

Those reviews’ findings were further reinforced by a recent meta-analysis by Katherine Baicker and colleagues that stated that health care costs fall by $3.27 and absenteeism costs fall by $2.73 for every dollar invested in a wellness program.8

By contrast, the recently released RAND Work- place Wellness Programs Study, which pooled 362,136 employees from five employers, found that lifestyle management programs can achieve improvements in risk factors, such as reductions in smoking, and increases in healthy behavior, such as exercise. Thestudy, however,did notfind that lifestyle management programs achieve sta- tistically significant reductions in health care costs.1 Although a 2012 three-year evaluation of the University of Minnesota’s wellness pro- gram also found no evidence that lifestyle man- agement lowers health care costs, the study did find the disease management component of the program to do so.9 Neither lifestyle management nor disease management were found to reduce absenteeism.9

Reconciling these seemingly contradictory findings requires not only asking, “Do wellness programs work?” but also,“Whichprogramcom- ponents have which effects under which condi-

tions?” Such an approach is particularly impor- tant given the heterogeneity of offerings that can be subsumed under the label “workplace well- ness” and the variety of settings in which these programs are implemented. Against this background, we assess the impact

of two common wellness program components— disease management to support employees with chronic conditions and lifestyle management to reduce employees’ health risks—on health care cost, use, and absenteeism by individual compo- nent and for both components together. Our study uses two baseline years and seven

program years of data from PepsiCo’s Healthy Living program and builds upon two prior eval- uations at three years into the program that showed the overall program, but not its lifestyle management component, was associated with lower health care costs.10,11

We undertook our current study to determine whether overall cost reductions were sustainable over a longer time period and whether the life- style management component would begin to contribute to the savings. To our knowledge, our study is one of the longest evaluations of a comprehensive wellness program to date in the published literature.

Study Data And Methods The PepsiCo Program PepsiCo introduced in 2003 what evolved into their Healthy Living pro- gram. Healthy Living is a wellness program made up of numerous components that include health risk assessments, on-site wellness events, life- style management, disease management, com- plex care management, a 24=7 nurse advice line, and maternity management. All PepsiCo employ- ees and their dependents can participate in the Healthy Living program, except for the fewer than 10 percent of employees who are enrolled in a health maintenance organization or receive their health coverage through their union. Health risk assessments help employees and

their dependents understand their health status and health risks, directing those with risks, such as obesity and smoking, to lifestyle management interventions consisting of mailed educational materials, online programs, and telephonic coaching for those with higher risk levels. In 2011 there were five distinct lifestyle manage- ment programs: weight management, nutrition management, fitness, stress management, and smoking cessation. Completion of a telephonic lifestyle management program involves a series of calls with a wellness coach over a six-month period. Disease management is offered to employees

with at least one of ten chronic conditions and

January 2014 33:1 Health Affairs 125

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

focuses on improving medication adherence and patient self-care knowledge and abilities. The ten conditions covered by the disease management program were asthma, coronary artery disease, atrial fibrillation, congestive heart failure, stroke, hyperlipidemia, hypertension, diabetes, low back pain, and chronic obstructive pulmo- nary disease. Completion of a disease manage- ment program typically requires six to nine months, during which participants have a series of calls with a nurse that average fifteen to twenty-five minutes per call. Completion of a program occurs when the participant is success- fully managing his or her condition. Study Sample We selected our sample from a

pool of 67,541 unique members who were eligi- ble for disease management or lifestyle manage- ment, or both, representing 400,657 member- years of data. We required participants to have at least two full years of health plan and program data as well as one year of data prior to partici- pation. Our sample consists of 14,555 partici- pants in disease management, 22,880 in lifestyle management, and 9,324 in both disease manage- ment and lifestyle management, among whom 2,610, 17,432, and 2,162 were successfully matched to a similar eligible nonparticipant, re- spectively. There are a total of 22,204 matched pairs, representing 238,724 member years. The matched pairs in the final analytic sample have on average 6.4 years of data. Data We combined PepsiCo’s health and phar-

macy plan claims data with Healthy Living eligi- bility and participation data for all employees and dependents for the period between Septem- ber 2002 and August 2011. These data cover two baseline years (September 2002 to August 2004) and seven program years (September 2004 to August 2011) for the lifestyle management pro- gram and one baseline year (September 2002 to August 2003) and eight program years (Septem- ber 2003 to August 2011) for the disease man- agement program. Our analytic sample was restricted to employees and dependents ages 18–64 with at least two full years of enrollment in a PepsiCo health plan and one full year of data prior to the year in which they first participated. Following common practice, we removed pro- gram years involving pregnancy-related care12

as well as individuals eligible for complex care management, which targets complex, high-cost conditions, such as terminal cancer and organ transplants, because the course of such condi- tions cannot be expected to be influenced by lifestyle or disease management. Program costs include the vendor’s per partic-

ipant per year fees for lifestyle and disease man- agement and the health risk assessment fee per completed survey.

Analytic Approach Our analytic sample con- sisted of all employees and dependents who were invited to participate in the lifestyle or disease management components of Healthy Living.We used those who decided to join the program as the intervention group and those who declined as the comparison group. Differential changes between the groups over time were used to esti- mate program impact, a so-called difference-in- differences design. To adjust for differences between participants

and nonparticipants, we used propensity score matching based on baseline data to balance ob- servable variables. Propensity scores were gen- erated based on a multinomial probit model, using the first year of data available for each member. The dependent variable was participa- tion in lifestyle management or disease manage- ment, or both, and the independent variables included age, sex, being an employee, geograph- ic region, calendar year, health plan enrollment tenure, total health care costs, emergency de- partment visits, hospital admissions, and co- morbidities.13

One-to-one propensity score matching was conducted with replacement, because there were fewer nonparticipants than participants, and was stratified by year and by program eligi- bility for lifestyle management or disease man- agement, or both. After matching, regression models were used to estimate the effects of par- ticipation in difference-in-differences on the out- comes of interest. The models included lagged variables of program participation, adjustments for time-varying covariates as appropriate (age, geographic region, calendar year, and co- morbidities) and individual-level fixed effects to control for unobservable characteristics. Measures Program eligibility represented

whether an employee was eligible and invited for a program via a telephone invitation and mailed letters during a given program year. Par- ticipation represented whether the eligible em- ployee subsequently chose to participate, as reflected in a monthly participation status indi-

Employers’ optimism regarding the benefits of wellness programs is driven by countless success stories.

Employee Wellness Program

126 Health Affairs January 2014 33:1

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

cator variable.We used two approaches to define program participation. Aggregate participation was specified as an indicator for any participa- tion in lifestyle management or disease manage- ment, or both. Component-specific participation indicators were used to capture the individual impact of the lifestyle management and disease management components on our outcomes. Our outcomes of interest were defined as health care cost per member per month, adjusted to 2012 US dollars using the Consumer Price Index14 and hospital admissions and emergency department visits per 1,000 employee-years. Absenteeism was measured based on individuals’ answers to a question on the health risk assessment about work days lost because of illness or injuries for the preceding twelve months.Work days lost was monetized by multiplying hours lost by the aver- age hourly wage for private, goods-producing industries ($34.14) from the Bureau of Labor Statistics.15

Return on investment (ROI) was calculated as the ratio of estimated reductions of health care and absenteeism costs to program costs as out- lined above over the entire seven-year interven- tion period.

Limitations As with all observational designs, our study may have produced results that suffer from bias because of unobservable differences between intervention- and comparison-group members. These may include differential moti- vation to improve health, health plan or wellness program literacy, or work schedule issues that make participation difficult. To minimize poten- tial bias, we used propensity score matching to account for observable differences between non- participants and participants, such as age, sex, comorbidities, and prior health care use. We in- cluded individual-level fixed effects in our re- gression analyses to account for unobservable differences that are constant over time. Thus, to attribute our estimates to bias, one would have to assume the existence of unobservable charac- teristics that vary over time and are associated with our endpoints. Because of a limited pool of nonparticipants,

particularly among people eligible for disease management and those eligible for both disease management and lifestyle management, our pro- pensity score matching did not balance all mem- ber characteristics between our participant and nonparticipant groups (Appendix Exhibit 1A).16

However, we controlled for such unbalanced differences in our regression models. PepsiCo is a large employer, and its experience

with disease management and lifestyle manage- ment programs might not be generalizable to other organizations, particularly smaller ones. Employers considering adopting a wellness pro-

gram should proceed with caution. Even if the program they implement is very similar to Pep- siCo’s lifestyle management and disease man- agement components, key differences in pro- gram implementation, design, and promotion to employees may affect results. For example, differences in program design and implementa- tion might affect participation and dropout rates and intervention effects. We did not examine health behavior outcomes,

such as exercise frequency or medication adher- ence.We also did not investigate program effects on more granular endpoints, such as wellness- sensitive hospitalizations, as proposed by Gautam Gowrisankaran and colleagues, which may capture program effects with greater ac- curacy.17

Lastly, we may have overstated the true ROI because our estimates of program component costs were confined to vendor fees. Specifically, we did not have information for the following cost items that affect ROI: the cost of PepsiCo’s program staff, the cost of employees’ time re- quired for program participation, and any costs generated by false positives through extended screening.

Study Results Looking at the lifestyle management and disease management components as a whole, we found participation to be associated with lower health care costs. Exhibit 1 compares the cost trends of

Exhibit 1

Aggregate Impact Of Lifestyle Management And Disease Management On Per Member Per Month Health Care Costs At PepsiCo, 2004–11

Participants

Nonparticipants

SOURCE Authors’ analysis of PepsiCo health plan and Healthy Living program data. NOTES Cost es- timates are adjusted by demographics, comorbidities, and calendar years based on propensity score matching and regression analyses. This exhibit assumes that members participated continuously dur- ing 2004–11; 2003 is the baseline year.

January 2014 33:1 Health Affairs 127

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

participants with those of statistically matched nonparticipants over seven years. After the third year of participation, differences in health care costs become statistically significant. We found that seven years of continuous participation was associated with an average reduction of $30 per member per month, or $360 annually (p < 0:01)

(Exhibit 2). However, when we broke down the effect on

health care cost by program component, we found disease management but not lifestyle management to be associated with lower costs (Exhibit 2).We estimate disease management to reduce costs among participants by $136 per member per month, or $1,632 annually, driven by a 29 percent reduction in hospital admis- sions (p < 0:01). When looking at the subset of participants that

had joined both the lifestyle management and the disease management components of the pro- gram, we estimate a reduction in health care costs of $160 per member per month, or about $1,920 per year (p < 0:01), and a 66 percent re- duction in hospital admissions (p < 0:05). Looking again at the lifestyle management and

disease management components as a whole, we found participation to be associated with a re- duction in self-reported absenteeism of 0.1 day, or forty-eight minutes (in an eight-hour work- day), per year (p < 0:01). This effect is driven by lifestyle management participation, which is as- sociated with a reduction of 0.13 day, or sixty-two minutes (in an eight-hour work day), per year (p < 0:01). The monetized impact of 0.10 and 0.13 day is estimated to be $28 and $35, respec- tively. No significant effect on absenteeism was observed among disease management partic- ipants. Based on our analyses, we estimate that the

lifestyle management and disease management components returned an average of $0.48 and $3.78, respectively, for every dollar invested when both health care and absenteeism impacts were included (Exhibit 3). Together, they re- turned $1.46 for every dollar invested. As shown in Exhibit 3, themain driver of the positive ROI is the reduction in health care costs associated with disease management participation.

Discussion We estimate the impact of a disease and lifestyle management program and find that disease management is associated with decreased health care costs and net savings after seven years—a result that confirms our previous analysis of this program after three years.10 Participation in life- style management interventions is associated with a small decrease in absenteeism but has no statistically significant effect on health care costs. These findings are not necessarily surpris- ing: As with any preventive intervention, it is often easier to achieve cost savings in people with higher baseline spending, as we found to be the case among disease management partic- ipants. Interestingly, the disease management

Exhibit 2

Per Member Per Month Cost Savings At PepsiCo, By Healthy Living Program Component, 2004–11

Any program**Lifestyle management Disease management**

SOURCE Authors’ analysis of PepsiCo health plan and Healthy Living program data. NOTE All savings are difference-in-differences estimates. “Lifestyle management” is the lifestyle management com- ponent; “disease management” is the disease management component; “any program” represents our measure of aggregate participation (participation in either lifestyle management or disease manage- ment, or both). Intervals for each estimate represent 95 percent confidence intervals. ** p < 0:05

Exhibit 3

Return On Investment For PepsiCo’s Healthy Living Program, By Program Component, 2011

Any programLifestyle management Disease management

Health care Absenteeism Overall

SOURCE Authors’ analysis of PepsiCo health plan and Healthy Living program data. NOTES Program effects are difference-in-differences estimates based on 2004–11 participation data; program costs represent those incurred in 2011. Return on investment denotes savings for each dollar spent. For descriptions of program components, see Exhibit 2 notes.

Employee Wellness Program

128 Health Affairs January 2014 33:1

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

participants who also joined the lifestyle man- agement program experienced significantly higher savings, which suggests that proper tar- geting can improve the financial performance of lifestyle management programs. Our findings are consistent with two recent

publications. The RAND Workplace Wellness Programs Study, which is the largest evaluation of workplace wellness programs conducted to date, found lifestyle management participation to produce no statistically significant reduction in health care costs.1 A 2012 evaluation con- ducted by John Nyman and colleagues of the lifestyle management and disease management components of the University of Minnesota’s wellness program used an approach similar to that of our study and found the two components together to generate an overall ROI of 1.76— quite similar to our ROI of 1.46.9 Additionally, the authors found the savings to be driven en- tirely by the program’s disease management component, with none generated by the lifestyle management component, further mirroring our results.9 Collectively, these findings cast doubt on thewidely held beliefin a strong business case for lifestyle management that is often supported by the above meta-analysis of Baicker and col- leagues.8

To investigate why several recent studies came to a different conclusion than those of the well- ness programs meta-analysis, we closely re- viewed the seven papers that Baicker and col- leagues analyzed. First, five papers looked at programs that operated more than twenty years ago,18–22 a time in which smoking was permitted in offices23 and statins were just emerging.24

These factors make it likely that the gains from lifestyle management interventions were higher twenty years ago than they are today. Second, the studies have a variety of methodological weak- nesses, such as a lack of statistical controls for health status;18–22,25 a lack of adjustment for con- comitant participation in disease manage- ment;26 and data limitations, such as imputation of costs from self-reported use.22 Lastly, the included populations are not easily generaliz- able—one study was of retirees21 and another was of 1,000 small-town city employees.18

Because of the long latency between reduction of risk factors and avoided onset of chronic dis- ease, it is possible that longer follow-up is re- quired to detect savings, but it appears unlikely that the lifestyle management component of Healthy Living will ever be able to fully offset its cost, particularly when all net program costs incurred to date areconsidered. A recentanalysis by Howard Bolnick and colleagues estimated that lowering modifiable risk factors to their theoretical minima would reduce health care

costsof anaverageworking-ageadultby18.4 per- cent.27 In other words, under perfect conditions, a lifestyle management program could save $876 per person per year, on average, using the 2012 average cost of coverage in the United States.28–30

However, even effective programs obviously can- not achieve the complete elimination of avoid- able health risks. As data from the RAND Work- place Wellness Programs Study show, programs managed to keep a quarter of smokers off nico- tine and increased the share of normal-weight participants from 21 percent to 33 percent after three years.1 From these numbers, we estimate that programs can realize about 10–25 percent of the theoretically possible cost savings, or $88– $219 per year and participant, which roughly corresponds to the $157 average annual savings estimate from the RAND study. Average annual cost per lifestyle management program partici- pant for PepsiCo’s program was in line with the $144 reported by Baicker and colleagues.1,8 To- gether, these estimates suggest that well-execut- ed lifestyle management programs may be ap- proximately cost-neutral. A lack of financial return does not imply that

lifestyle management cannot create value. Our study finds a significant effect on absenteeism, and both the RAND Workplace Wellness Pro- grams Study and a recent systematic review showed statistically significant and clinically meaningful improvements in certain health risks among program participants, even though one needs to caution that most of the evidence so far has been generated from a limited set of com- mitted employers and may have limited gener- alizability.31 At the same time, wellnessprograms may have unintended consequences in the form of overdiagnosis and overtreatment.32–34

Further, ADP employer survey data suggest that economic outcomes are not the only reason employers offer wellness programs: The most common reason is to improve employee health, with “attract and retain talent” and “maintain or increase benefit offerings” also offered as key reasons.3

Our and other recent results have several im- plications for employers, researchers, and policy makers. First, the current evidence suggests that blanket claims of “wellness saves money” are not warranted, and it underscores that any program evaluation needs to be scrutinized to understand its results. Readers should ask which features the program offered, what their respective contribu- tion to the outcomes was, whether an appropri- ate comparison strategy was used, and how the results should be interpreted in light of the com- parison strategy. For example, the most recent evaluation of Johnson & Johnson’s Live for Life program compared the medical cost experience

◀

$1.46 Return on investment Together, the lifestyle management and disease management components of Healthy Living returned an average of $1.46 for every dollar invested.

January 2014 33:1 Health Affairs 129

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

of Johnson & Johnson employees with the expe- rience of statistically matched employees in sim- ilar firms and found annual increases to be 3.7 percent lower among Johnson & Johnson employees.12 This evaluation’s design does imply that Johnson & Johnson’s overall health and wellness strategy is successful, but its compari- son strategy compares Johnson & Johnson em- ployees to similar employees in other firms. In contrast to our analysis, which compares pro- gram participants and statistically matched non- participants within one firm, the Johnson & Johnson evaluation cannot parse out whether individual components of the company’s well- ness program or other company characteristics, or both, such as benefits design, hiring, work- placepolicies,35,36 and corporateculture,aredriv- ing the results, which makes it difficult for read- ers to use such results for wellness program decision making. Second, employers need to align program con-

figuration with their objectives. If the primary objective is cost control, they should focus on interventions for higher-risk employees, such as those with multiple risk factors or manifest chronic disease. Conversely, if the objective is to improve workforce health, investment in evi- dence-based lifestyle management programs may be warranted.1

Third, employers need to carefully consider the total cost of a program before deciding what to offer and which vendor to use. For example, theCenters for Disease Control and Prevention is advocating greater use of awareness campaigns and wellness events as more cost-effective ap- proaches than individual coaching.37 Recently issued federal rules38 allow employers to tie sub- stantial incentives to wellness program partici-

pation and control of risk factors.39 Given their magnitude, such incentives can quickly change an employer’s cost-benefit calculation. As a mat- ter of public policy, we will need to understand what proportions of the savings to employers stem from true improvements in health and what proportions are the result of cost shifting to em- ployees with health risks.40

Conclusion Workplace wellness programs have the potential to reduce health risks and to delay or avoid the onset of chronic diseases as well as to reduce health care cost in employees with manifest chronic disease. But employers and policy mak- ers should not take for granted that the lifestyle management component of such programs can reduce health care costs or even lead to net sav- ings. ▪

The findings in this article were previously presented at the AcademyHealth 2013 Annual Research Meeting, Baltimore, Maryland, June 23– 25, 2013. Funding was provided by PepsiCo.

NOTES

1 Mattke S, Liu H, Caloyeras JP, Huang CY, Van Busum KR, Khodyakov D, et al. Workplace Wellness Programs Study. Santa Monica (CA): RAND Corporation; 2013. (Pub. No. RR-254-DOL).

2 Program components can include health promotional materials, health screening or biometric testing, smoking cessation or weight loss interventions, access to a nurse help line, and web-based health coaching.

3 ADP Research Institute. Why you should care about wellness programs

[Internet]. Roseland (NJ): ADP Re- search Institute; 2012 [cited 2013 Dec 11]. Available from: http:// www.adp.com/tools-and-resources/ case-studies-white-papers/~/media/ White%20Papers/WellnessFinal 22112.ashx

4 Deloitte Center for Health Solutions, Deloitte Consulting. 2012 Deloitte survey of US employers: opinions about the US health care system and plans for employee health benefits [Internet]. New York (NY): Deloitte; 2012 Jul [cited 2013 Dec 11]. Avail-

able from: http://www.deloitte.com/ assets/Dcom-UnitedStates/Local %20Assets/Documents/us_dchs_ employee_survey_072512.pdf

5 Aon Hewitt. 2012 health care survey [Internet]. Chicago (IL): Aon Hewitt; 2012 [cited 2013 Dec 11]. Available from: http://www.aon.com/ attachments/human-capital- consulting/2012_Health_Care_ Survey_final.pdf

6 Chapman LS. Meta-evaluation of worksite health promotion econom- ic return studies: 2005 update. Am J

Employers need to carefully consider the total cost of a program before deciding what to offer and which vendor to use.

Employee Wellness Program

130 Health Affairs January 2014 33:1

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

Health Promot. 2005;19(6):1–11. 7 Pelletier KR. A review and analysis of

the clinical and cost-effectiveness studies of comprehensive health promotion and disease management programs at the worksite: update VIII 2008 to 2010. J Occup Environ Med. 2011;53(11):1310–31.

8 Baicker K, Cutler D, Song Z. Work- place wellness programs can gener- ate savings. Health Aff (Millwood). 2010;29(2):304–11.

9 Nyman JA, Abraham JM, Jeffery MM, Barleen NA. The effectiveness of a health promotion program after 3 years: evidence from the University of Minnesota. Med Care. 2012; 50(9):772–8.

10 Liu H, Mattke S, Harris K, Weinberger S, Serxner S, Caloyeras JP, et al. Do workplace wellness programs reduce medical costs? Ev- idence from a Fortune 500 company. Inquiry. Forthcoming 2014.

11 Liu H, Harris KM, Weinberger S, Serxner S, Mattke S, Exum E. Effect of an employer-sponsored health and wellness program on medical cost and utilization. Popul Health Manag. 2013;16(1):1–6.

12 Henke RM, Goetzel RZ, McHugh J, Isaac F. Recent experience in health promotion at Johnson & Johnson: lower health spending, strong return on investment. Health Aff (Mill- wood). 2011;30(3):490–9.

13 Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27.

14 Bureau of Labor Statistics. Consum- er Price Index. Washington (DC): BLS; 2013 Mar.

15 Bureau of Labor Statistics. Employer costs for employee compensation summary. Washington (DC): BLS; 2013 Mar 12.

16 To access the Appendix, click on the Appendix link to the right of the article online.

17 Gowrisankaran G, Norberg K, Kymes S, Chernew ME, Stwalley D, Kemper L, et al. A hospital system’s wellness program linked to health plan en- rollment cut hospitalizations but not overall costs. Health Aff (Millwood). 2013;32(3):477–85.

18 Aldana SG, Jacobson BH, Harris CJ, Kelley PL, Stone WJ. Influence of a mobile worksite health promotion program on health care costs. Am J Prev Med. 1993;9(6):378–83.

19 Bly JL, Jones RC, Richardson JE. Impact of worksite health promotion

on health care costs and utilization. Evaluation of Johnson & Johnson’s Live for Life program. JAMA. 1986; 256(23):3235–40.

20 Fries JF, Harrington H, Edwards R, Kent LA, Richardson N. Randomized controlled trial of cost reductions from a health education program: the California Public Employees’ Retirement System (PERS) study. Am J Health Promot. 1994;8(3): 216–23.

21 Leigh JP, Richardson N, Beck R, Kerr C, Harrington H, Parcell CL, et al. Randomized controlled study of a retiree health promotion program: the Bank of America study. Arch Intern Med. 1992;152(6):1201.

22 Shi L. Health promotion, medical care use, and costs in a sample of worksite employees. Eval Rev. 1993;17(5):475–87.

23 The first state restriction on smoking in the workplace was enacted by California in 1994 as Assembly Bill 13 and became law in 1995 (Labor Code 6404.5).

24 The first statin (lovastatin) for the lowering of cholesterol was ap- proved by the Food and Drug Ad- ministration in 1987.

25 Ozminkowski RJ, Dunn RL, Goetzel RZ, Cantor RI, Murnane J, Harrison M. A return on investment evalua- tion of the Citibank, N.A., health management program. Am J Health Promot. 1999;14(1):31–43.

26 Naydeck BL, Pearson JA, Ozminkowski RJ, Day BT, Goetzel RZ. The impact of the Highmark employee wellness programs on 4- year healthcare costs. J Occup Envi- ron Med. 2008;50(2):146–56.

27 Bolnick H, Millard F, Dugas JP. Medical care savings from workplace wellness programs: what is a realistic savings potential? J Occup Environ Med. 2013;55(1):4–9.

28 The minimum medical loss ratio al- lowed by the Affordable Care Act, section 158.210, is 0.85.

29 Kaiser Family Foundation. Employer health benefits: 2012 annual survey [Internet]. Menlo Park (CA): KFF; 2012 [cited 2013 Dec 11]. Available from: http://kaiserfamily foundation.files.wordpress.com/ 2013/03/8345-employer-health- benefits-annual-survey-full-report- 0912.pdf

30 Obtained by applying the 18.4 per- cent savings estimate and the mini- mum medical loss ratio of 85 percent permitted by the Affordable Care Act

for the large-group market to the average annual premium for single coverage in employer-sponsored health insurance ($5,615).

31 Osilla KC, Van Busum K, Schnyer C, Larkin JW, Eibner C, Mattke S. Sys- tematic review of the impact of worksite wellness programs. Am J Manag Care. 2012;18(2):e68–81.

32 Cassels A. Seeking sickness: medical screening and the misguided hunt for disease. Vancouver (BC): Greystone Books; 2012.

33 Krogsbøll LT, Jørgensen KJ, Gøtzsche PC. General health checks in adults for reducing morbidity and mortality from disease. JAMA. 2013;309(23):2489–90.

34 Sox HC. The health checkup: was it ever effective? Could it be effective? JAMA. 2013;309(23):2496–7.

35 Asch DA, Muller RW, Volpp KG. Conflicts and compromises in not hiring smokers. N Engl J Med. 2013;368(15):1371–3.

36 Schmidt H, Voigt K, Emanuel EJ. The ethics of not hiring smokers. N Engl J Med. 2013;368(15):1369–71.

37 Centers for Disease Control and Prevention. Workplace health pro- motion: workplace health model [Internet]. Atlanta (GA): CDC; [cited 2013 Dec 11]. Available from: http:// www.cdc.gov/workplacehealth promotion/model/index.html

38 Department of Health and Human Services. Incentives for nondiscrim- inatory wellness programs in group health plans; final rule. Federal Register [serial on the Internet]. 2013;78(106):33157–92. Available from: http://www.gpo.gov/fdsys/ pkg/FR-2013-06-03/pdf/2013- 12916.pdf

39 Health-contingent wellness pro- grams may use incentives up to 30 percent of the cost of employee- only coverage or up to 50 percent if the program is designed to prevent or reduce tobacco use. Using the average annual premium for single coverage, employer-sponsored health insurance ($5,615), the max- imum average incentive allowed under the Affordable Care Act is about $1,680, or $2,800 if also tar- getting tobacco use.

40 Horwitz JR, Kelly BD, DiNardo JE. Wellness incentives in the work- place: cost savings through cost shifting to unhealthy workers. Health Aff (Millwood). 2013;32(3): 468–76.

January 2014 33:1 Health Affairs 131

by Rachel McCartney on January 9, 2014Health Affairs by content.healthaffairs.orgDownloaded from

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.