Comparing excess costs across multiple corporate populations.

The purpose of this study was to examine the relationship of health risk level to charged medical costs and determine the excess cost of higher risk individuals compared to low risk. Two years of medical claims from six corporations were used to determine costs of health risk assessment (HRA) participants and nonparticipants. A total of 165,770 employees, 21,124 of which took an HRA, were used for the study. Costs increased as risk level increased. There were no significant differences within a risk level between companies for the cost ratio. Percent of medical costs due to excess risk ranged from 15.0-30.8% for HRA participants and 23.8-38.3% for the study population. Cost patterns were consistent across companies. Excess cost as the result of increased risk level accounted for a substantial portion of the cost at each company. These results can be used to justify the need for a health-promotion program and to estimate potential savings as the result of excess risk. Even without the use of an HRA, health practitioners should feel confident stating that excess risk accounts for at least 25% to 30% of medical costs per year across a wide variety of companies, regardless of industry or demographics. The numbers can be used as a realistic estimate for any health promotion program financial proposal.

Application of a prediction model for identification of individuals at diabetic risk.

As a follow-up to our preceding paper, we attempted to extract features of health risk progression for diabetes in Sequential Multi Layered Perceptron (SMLP) via inverse processing of the learned structure. The time-varying risk progress was assessed with risk trajectory and conditional mixture model. Overall risk cut along with the prediction was stable over time and high body mass index (BMI) tops the health behavioral risks predicting the onset of diabetes. For the initial prediction, high BMI (obesity), high blood pressure (BP), high cholesterol, and diet in fatty food were significant. Over time, variations in trajectory were due to changes in BMI, stress, BP, cholesterol, and fatty food intake. We tested the effectiveness of identifying prediabetics by the SMLP by applying the implemented SMLP to a test population of employees from a large manufacturing company, where an early worksite health promotion was initiated (1984). This resulted in a potential sensitivity (71.4%) although there were issues like mapping corresponding risks and large time lags. A secondary test on the similar population as in the previous paper showed a promising sensitivity (86.5%) over 3 years. When combining with targeted screening such as impaired glucose tolerance test only for those predicted to be diabetics, the presented prediction model and extracted features can be used in implementing an effective disease prevention and management program.

Association of health risks with the cost of time away from work.

The purpose of this study was to combine absences, short-term disability, and workers' compensation into a sum of the cost of time away from work (TAW) and compare it with health risk status and individual health risks of 6220 hourly workers at Steelcase Inc. The study used 3 years (1998 to 2000) of TAW and health risk appraisal data. Higher TAW costs were associated with illness days, drug/medication use, the individual's lower perception of physical health, job dissatisfaction, high stress, life dissatisfaction, and physical inactivity. More high-risk individuals (80.6%) had a TAW occurrence than medium- (72.8%) and low-risk (61.1%) individuals. High-risk individuals had higher TAW costs than medium- and low-risk individuals. Of the total TAW costs, 36.2% was attributed to the excess risks of the medium- and high-risk individuals or nonparticipants compared with low-risk participants. If TAW costs follow risk reduction, a potential annual savings of $1.7 million could be achieved.

A sequential neural network model for diabetes prediction.

This paper presents a neural network (NN) model to evaluate an existing Health Risk Appraisal (HRA) for diabetes prediction over 3 years (1996-1998) based on a simulated learning algorithm on individual prognostic process, using the repeatedly measured HRAs of 6142 participants. The approach uses a sequential multi-layered perceptron (SMLP) with backpropagation learning, and an explicit model of time-varying inputs along with the sequentially obtained prediction probability, which was obtained by embedding a multivariate logistic function for consecutive years. The study captures the time-sensitive feature of associating risk factors as predictors to the occurrence of diabetes in the corresponding period. This approach outperforms the baseline classification and regression models in terms of gains (average profit: 0.18) and sensitivity (86.04%) for a test data. The result enables a time-sensitive disease prevention and management program as a prospective effort.

Changes in health risks among the participants in the United Auto Workers--General Motors LifeSteps Health Promotion Program.

PURPOSE: To examine health risk changes among participants of a multicomponent worksite health promotion program. DESIGN: A study using health risk changes among health risk appraisal (HRA) participants linked to program participation records. Baseline risk and participation in multiple programs were examined in relation to risk change in multivariate models. SETTING: Worksite health promotion programming sponsored by the United Auto Workers (UAW) and General Motors (GM). SUBJECTS: Active employees (12,984) who voluntarily participated in an HRA in each of two program years. INTERVENTION: The nationwide program was a mailed HRA and a 1-800 nurse line. A pilot program (implemented in two cities) added screening, wellness programs, a materials resource, and, for high risk participants, health coaching and vouchers for medical office visits. MEASURES: Using 13 selected health risk factors from the HRA, changes in overall health risks were measured as program outcomes in three ways: one-directional, net, and risk status change. RESULTS: A greater decrease in the number of health risks was observed with increased program participation. The decrease was significantly related to the number of baseline risk factors and eligibility for high risk programs. Associated with program participation, the number of people at low risk status increased from 70.1% to 71.3% at year 2 among nationwide participants and from 52.4% to 58.9% among pilot participants. CONCLUSIONS: Participation was associated with a significant impact on health risk. Baseline risk of participants and eligibility for high risk programs were necessary factors to control when measuring program effects on health risk changes.

Emerging research: a view from one research center.

The Health Management Research Center at the University of Michigan has assembled a database on health risks, medical care costs, an in some cases, productivity measures for over 2,000,000 individuals. For employees of its corporate consortium members, the database contains seven to eighteen years of data. Working with this data, the research team has observed a number of emerging trends. These trends have been stable in this data set for a number of years, but some of them are yet to be subjected to rigorous external peer review. The trends are summarized below. 1) Annual participation rates of 20% to 30% in Health Risk Appraisal are typical; over 10 years, 80% participate at least once, 60% at least twice and 40% at least three times. 2) Among the employers in the data base, excess risk factors account for 21% to 31% of medical care costs, with a mean of 25%. 3) Medical care costs increase as the number risk factors and age increase. As risk factors increase, medical costs increase; as risk factors decrease, medical care costs decrease. The mean cost increase per risk factor increased ($350) may be more than double the mean cost decrease per risk factor decreased ($150). 4) Cost savings greatest among those who participate in programs multiple times. 5) Absenteeism seems to be higher and other measures of productivity lower for those with health risk factors. 6) Programs designed to keep healthy people healthy in addition to reducing the risks of those with multiple risks will probably provide the greatest return to the employers. 7) Best results may be achieved by focusing efforts on employees who have clusters of risk factors associated with low perceived health status. 8) A corporate wellness score which combines risk factor levels and participation rates may provide a "corporate wellness score" which can be used to compare health status across employer. 9) Increased use of longitudinal data sets, fuzzy cut points for data categories and data mining techniques may allow breakthroughs in future analysis efforts.

Internal associations among health-risk factors and risk prevalence.

OBJECTIVE: To explore risk associations between health variables and to compare these associations with the prevalence of each risk. METHODS: Manufacturing-company employees (N=16,879) completed a self-reported health risk appraisal with on-site biometric screening. RESULTS: Risk prevalence and risk association findings differed by gender, age, and overall risk level. Risks that were most highly associated with other risks were different from risks that were most prevalent. CONCLUSIONS: The findings suggest that in addition to risk prevalence, individual characteristics and the level of association between risks are also important to consider when planning health interventions.

The association of health risks with workers' compensation costs.

The purpose of this study was to investigate the association between health risks and workers' compensation (WC) costs. The 4-year study used Health Risk Appraisal data and focused on 1996-to-1999 WC costs among Xerox Corporation's long-term employees. High WC costs were related to individual health risks, especially Health Age Index (a measure of controllable risks), smoking, poor physical health, physical inactivity, and life dissatisfaction. WC costs increased with increasing health risk status (low-risk to medium-risk to high-risk). Low-risk employees had the lowest costs. In this population, 85% of WC costs could be attributed to excess risks (medium- or high-risk) or non-participation. Among those with claims, a savings of $1238 per person per year was associated with Health Risk Appraisal participation. Addressing WC costs by focusing on employee health status provides an important additional strategy for health promotion programs.

A case study of 10-year health risk appraisal participation patterns in a comprehensive health promotion program.

Between 1990 and 1999, cumulative participation in a health risk appraisal (HRA) reached 79%, and participation in any health promotion program reached 94% of employees employed continuously by the Progressive Corporation during that time period. Participation rates in the HRA per year ranged from 14% to 36%. First time participants reached no higher than 2.6% after the sixth year. Non-participants tended to have higher mean but lower median costs.

Bank One's worksite-based asthma disease management program.

Non-occupational asthma is a common disease affecting approximately 6% of the US population. Asthma accounts for an estimated 3 million workdays lost each year in the United States and for reduced employee job productivity. Although asthma disease management programs are relatively common in managed care organizations, they generally have not been offered at the workplace. A total of 168 employees with asthma were identified, and 76 participated in a worksite asthma disease management program. Of them, 41 (53.9%) completed an Asthma Therapy Assessment Questionnaire (ATAQ) at baseline, post-program (2 months), 4 months, and 12 months. The total ATAQ Index declined from 3.59 at baseline to 2.76 at 12 months (p < 0.01). After the program, significantly more employees reported using controller medications (the desired behavior) rather than reliever medications. A worksite-based disease management program for asthma should reduce medical care costs and absenteeism and improve worker productivity.

The impact of allergies and allergy treatment on worker productivity.

Allergic disorders are a chronic and highly prevalent condition in the general population and the workforce. Their effect on workers and corporate costs go beyond the direct cost of treatment, as the condition can lower a worker's productivity. Previous research includes estimates of the decrease in productivity associated with allergic disorders. None of these studies, however, offered an objective measure of how worker productivity is affected by allergic disorders. In the present study, the productivity of telephone customer service representatives suffering from allergic disorders is examined before, during, and after the ragweed pollen season. In addition, these workers were surveyed as to the type of medication they used in response to their condition. A significant correlation was observed between an increase in pollen counts and a decrease in productivity for workers with allergies. Compared with workers without allergies, employees with allergies who reported using no medication showed a 10% decrease in productivity. No differences were observed among workers with allergies using different types of medications, although the medication groups had significantly higher productivity than the no-medication group. The expected lowered productivity of those workers with allergies who used sedating antihistamines may have been offset by their relatively lower level of symptom severity and by the nature of the job and the productivity measures used.

Self-reported utilization of preventive health services by retired employees age 65 and older.

OBJECTIVES: Increased utilization of preventive services among the aging has been associated with improved health status and decreased medical costs. We sought to examine the use of the Health Risk Appraisal (HRA) in benchmarking compliance and characterizing those retired employees who met preventive service guidelines. DESIGN: A retrospective cohort study of retired employees age 65 and older. SETTING: Nation-wide health promotion program offered by General Motors Corporation. PARTICIPANTS: 59,670 retired General Motors employees age 65 and older who participated in a nationwide mailed HRA health promotion program. MEASUREMENTS: Preventive health services compliance was measured using selected HRA questions. Gender, HRA participation patterns, overall health risk status, medical plan selection and disease status were examined as predictors of increased compliance. Multivariate logistic regression models were developed to test the relative contributions of participant characteristics to increased utilization. RESULTS: The self-reported HRA data indicated that compliance levels were higher than national averages. The Healthy People 2000 goals for the preventive services studied were met and exceeded (with the exception of tetanus immunization). Higher compliance was associated with being male, younger than 70 years, multiple-year HRA participation, overall low risk status and HMO insurance plan selection. CONCLUSION: The results from the HRA indicated that this population participated at a higher level than a comparable national sample exceeding goals set by Healthy People 2000.

The role of health risk factors and disease on worker productivity.

The costs attributed to employee health problems are usually measured by employers in terms of direct health care costs, such as medical plan claims. Although it has been understood that employee health problems also produce indirect costs for employers, their measurement has been far less frequent. At best, studies have recorded one component of indirect health costs: the time lost to employee absenteeism and disability. The study presented here includes a measure of the actual decrease in the productivity of employees while they are on the job, in addition to measures of absenteeism and disability. These three measurements were combined to produce a Worker Productivity Index (WPI). The WPIs of 564 telephone customer-service agents were correlated with the employees' number and type of health risks, as measured by a Health Risk Appraisal. Additionally, the WPI was also examined across different disease states in the same population of employees. As the number of health risks increased, an employee's productivity decreased. The nature of the health risk may also differentially affect the pattern of the decrease. Finally, disease states were also associated with different patterns of productivity reduction.

The costs of body mass index levels in an employed population.

We studied 3,066 First Chicago NBD employees by using health risk appraisals and personnel data to determine the direct and indirect costs (in 1996 dollars) associated with varying levels of body mass index (BMI). The BMI is the most widely used measure of weight adjusted by height. We found that as BMI increases, so do the number of sick days, medical claims and health care costs and that the mean annual health care costs for the BMI "at risk" population (men with a BMI of > or = 27.8 kg/m2, women > or = 27.3 kg/m2) was $2,274 versus $1,499 for the "not at risk" group. Major differences in health care costs were observed for employees 45 years old and older, particularly among the women in this age group. Analysis was completed for those employees with and without a risk for BMI. A "J-shaped" curve between medical claims costs and BMI exists, with the low point occurring at about 25 to 27 kg/m2. We conclude that indirect and direct costs to an employer increase with increasing BMI. Employers may benefit from helping employees achieve a healthier weight. The initial target population should be those who are at highest risk of complications from obesity.

The economic costs associated with body mass index in a workplace.

This study was undertaken to determine if a progressive correlation exists between body mass index (BMI), health care costs, and absenteeism and to identify an economically optimal BMI. We studied 3,066 First Chicago NBD employees by using health risk appraisals and personnel data. Analysis was completed for those employees with and without a risk for BMI. People at risk for BMI are more likely to have additional health risks, short-term disability and illness absence, and higher health care costs than those not at risk for BMI. A "J-shaped" curve between health care costs and BMI exists, with the low point occurring at about 25 to 27 kg/m2. We concluded that indirect and direct costs to an employer increase with increasing BMI. Employers may benefit from helping employees achieve a healthy weight. The initial target population should be those who are at highest risk of complications from obesity.

The financial impact of changes in personal health practices.

The impact of changes in health practices on medical claims costs between 1985-1987 and 1988-1990 was examined among 796 employees. Employees completed a health risk appraisal in 1985 and 1988 and were categorized into health risk levels. High-cost status and high-risk status were significantly associated at both times. The percentage of employees with high-risk status decreased from 31.8% to 25.3% between 1985 and 1988 (P < 0.01). Changes in average costs and the percentage with high-cost status followed the risk change. The largest increase in average costs occurred in employees who moved from low-risk to high-risk status. The greatest reduction in average costs occurred in employees who moved from high-risk to low-risk status. Median costs were not as sensitive to risk change. The findings provide strong evidence that improving individual health status is associated with financial benefits.

The worksite environment as a cue to smoking reduction.

With a modified version of the Health Belief Model as the conceptual framework, the hypothesis that exposure to a worksite health-promoting environment acted as a cue to smoking reduction among 310 smokers was tested with a quasi-experimental design. Using path analysis (LISREL), the total effects on posttest smoking were decomposed. Results showed exposure to the health-promoting environment had statistically significant direct and indirect effects on posttest smoking. The indirect effect was through the posttest perceived barriers variable. The LISREL model explained 74% of the variance in smoking reduction and fit the data satisfactorily. The importance of developing the worksite as a health-promoting force is discussed.

Corporate medical claim cost distributions and factors associated with high-cost status.

Medical claims cost distributions and factors that relate to these costs were studied among 7796 employees who selected an indemnity medical plan for at least one year during 1985 to 1990. Descriptions for medical claims costs were presented for both single- and multiple-year time periods. Factors that associated with high cost status were studied by using multiple logistic regression models. Employees at or above the top tenth cost percentile were accounted for approximately 80%, 65%, and 58% of the total employees' medical costs to the employer in single-year, 3-year, and 6-year periods, respectively. Bivariate analyses indicated tht six of the seven selected demographic variables were significantly related to cost status. When the multivariate models excluded health risk measures, four of the demographic variables and the frequency of health risk appraisal completion were significantly associated with cost status. When the multivariate models included health risk measures, the health risk measures became the dominant predictors of the high-cost status.