Obesity/Obesity with Associated Co-morbidities
The most widely used and accepted metric for identifying obesity is having a body mass index (BMI) greater than 30. Waist circumference is also being recognized as an important factor in assessing obesity. Men with a waist circumference of 40 inches or greater, and women with a waist circumference of 35 inches or greater, are considered obese.
OBESITY WITH ASSOCIATED CO-MORBIDITIES
Higher body weights are associated with an increase in mortality from all causes. Obese individuals with co-morbidities are those who are at the highest risk because they tend to have multiple risk factors. Being overweight or obese substantially increases the risk of chronic conditions and illnesses such as hypertension, dyslipidemia, type 2 diabetes, coronary artery disease, stroke, gallbladder disease, osteoarthritis, and sleep apnea and respiratory problems, as well as cancers of the endometrium, breast, prostate, and colon.
The prevalence of overweight and obesity is increasing rapidly in the United States. Sixty-eight percent of the U.S. population aged 20 and over is overweight; thirty-four percent of adults in the United States are obese. Moreover, 15 percent of school age children are overweight and the proportions are even higher among some ethnic groups. All in all, an estimated 97 million adults in the United States are overweight or obese. Many of these individuals exhibit prediabetes and other co-morbid conditions.
The concept of obesity and obesity with associated co-morbidities as manageable, chronic conditions is emerging. Robust epidemiological and scientific evidence clearly demonstrates that obesity should be considered in the context of chronic disease. Population health management offers a new model of care that shifts treatment toward chronic care and proactively interfaces with other existing chronic illnesses common in obese individuals. By recognizing the central role that obesity plays in the development of these illnesses, better care can result.
New thinking suggests the need to focus chronic care management on those who are at highest risk, are already obese, and have a cluster of risk factors or co-morbid conditions.
The following table from the National Heart, Lung, and Blood Institute (NHLBI) and the National Institutes of Health (NIH) presents categories for overweight and obesity. These are presented as both BMI and waist circumference measurements. Associated risk factors are also shown.
Cut-off points are used to identify increased relative risk for the development of obesity-associated risk factors. While imperfect, these cut-off points indicate the need for management of the clinical issues relating to overweight and obesity to reduce risk factors, improve health overall, and reduce resource consumption.
|Classification of Overweight and Obesity by BMI, Waist Circumference, and Assiated Disease Risk|
|Disease Risk Relative to Normal
Weight and Waits Circumference
|BMI (k2/m2)||Obesity Class||Men ≤ 102cm (≤40in)
Women ≤ 88cm (≤35in)
|> 102cm (40in)
> 88cm (35in)
|Normal||18.5 - 24.9|
|Overweight||25.0 - 29.9||Increased||High|
|Obesity||30.0 - 34.9||I||High||Very High|
|Obesity||35.0 - 39.9||II||Very High||Very High|
|Extreme Obesisty||≥40||III||Extremely High||Extremely High|
Claims data collection systems however, were not designed to collect information with which to assess obesity levels. This deficiency has hindered population health management program development. In the future, other methods of identifying the risk associated with obesity may include assessing for signs of insulin resistance, glucose intolerance, and a proinflammatory and prothrombotic state. These might include testing for inflammation through the measurement of serum C-reactive protein, a prothrombotic state by measuring serum PAI-1.
METRICS FOR IDENTIFYING OBESE POPULATIONS
Most often BMI is used to determine overweight and obesity. Waist circumference is another key measure for identifying obese individuals in chronically ill populations. Both BMI and waist circumference have limitations in assessing obesity and risk. Other methods, which are expensive and not generally feasible in the clinic or home settings— include calipers (skin-fold measurement), underwater weighing, and computerized topography.
BMI: Obesity is commonly assessed by using BMI, defined as the weight in kilograms divided by the square of the height in meters (kg/m2). A BMI of over 25 kg/m2 is defined as overweight, and a BMI of over 30 kg/m2 is defined as obese. These markers provide common benchmarks for assessment, but the risks of disease in all populations can increase progressively from lower BMI levels.
Limitations of BMI: Simple BMI calculations may be misleading. According to the existing definition and calculation of BMI, anyone with a BMI over 25 would be classified as overweight whether their body is composed of fat or muscle. Athletes, for example, may be considered to be overweight even though they may have very little visceral fat. BMI is an imperfect indicator of risk of disease. People with the same BMI but different amounts of visceral fat face different risks of disease. Furthermore, weight is only one among many risk factors.
BMI is calculated the same for adults and children but is interpreted differently for children. The CDC notes that “for children ages 2–20 years, BMI is plotted on a growth chart specific for age and gender.”
Although some authors use categories such as “moderately overweight” for those with a BMI of 25-30, the NHLBI does not designate overweight with such qualifiers. The extreme obesity classification (BMI > 40) is a commonly used cut-off for determining qualification for bariatric surgery. However, BMI cut-off points for obesity vary around the world.
Waist circumference: NHLBI, part of the NIH, has the following definition of waist circumference:
“[The] presence of excess fat in the abdomen out of proportion to total body fat is an independent predictor of risk factors and morbidity. Waist circumference is positively correlated with abdominal fat content. It provides a clinically acceptable measurement for assessing a patient’s abdominal fat content before and during weight loss treatment.”
Limitations of waist circumference: Waist circumference is valuable in assessing risk in the BMI < 35 range and is particularly useful in ethnically diverse groups, where waist-to-hip ratio may be an even better predictor. Waist circumference in individuals with a BMI > 35 generally exceeds the cut-off points noted above. The relative risk faced by individuals within a BMI or waist circumference range can be estimated compared with the risk that individual would face at a normal weight or waist size. These relative risk calculations do not reflect the individual’s absolute risk, which is determined by adding all of his/her risk factors.
Predicting risk is essential to disease management. According to the American Obesity Association, “obesity is associated with more than 30 medical conditions, and scientific evidence has established a strong relationship with at least 15 of those conditions.” The American Heart Association also now recognizes obesity as a risk factor for heart attack.
The prevalence of various medical conditions increases with overweight and obesity for both men and women. The following table from the American Obesity Association shows correlations of medical condition to BMI for men.
|Prevalence of Medical Conditions by Body Mass Index (BMI) for Men|
|Medical Condition||Body Mass Index|
|18.5 to 24.9||25 to 29.9||30 to 34.9||>40|
|Type 2 Diabetes||Prevalence Ratio (%)|
|Coronary Heart Disease||8.84||9.60||16.01||13.97|
|High Blood Pressure||23.47||34.16||48.95||64.53|
|Prevalence of Medical Conditions by Body Mass Index (BMI) for Women|
|Medical Condition||Body Mass Index|
|18.5 to 24.9||25 to 29.9||30 to 34.9||>40|
|Type 2 Diabetes||Prevalence Ratio (%)|
|Coronary Heart Disease||6.87||11.13||12.56||19.22|
|High Blood Pressure||23.26||38.77||47.95||63.16|
Overweight and obesity are associated with insulin resistance and the metabolic syndrome. However, the presence of abdominal obesity is more highly correlated with the metabolic risk factors than is an elevated BMI. Therefore, the simple measure of waist circumference is recommended to identify the body weight component of the metabolic syndrome.
Though the existence of metabolic syndrome is often debated, there does exist a cluster of cardiovascular risk factors often associated with obesity. According to the following ATP III criteria table developed by the National Cholesterol Education Program expert panel, metabolic syndrome can be diagnosed when three of the following five diagnostic criteria are present: abdominal obesity, elevated glucose or triglycerides, reduced HDL cholesterol, and hypertension. Evidence suggests that treating patients once metabolic syndrome has advanced to these diseases may not be cost-effective. In a study of adults with metabolic syndrome, researchers found drug costs were four times higher for these patients than the average annual drug costs for patients without metabolic syndrome. Patient outcomes are also suboptimal in those with metabolic syndrome, but new research suggests that treating the underlying causes of the condition could lead to better outcomes.
|ATP III Criteria|
|Risk Factor||Defining Level|
|Abdominal obesity, given a waist circumference||Men||>102cm (>40in)|
|HDI, cholesterol||Men||<40 mg/dl.|
|Blood pressure||130/85 mm Hg|
|Fasting glucose||110 mg/dl.|
American Obesity Association (AOA). “AOA Fact Sheets: Health Effects of Obesity and Obesity Related Medical Conditions.” 14 Jun. 2006 http://www.obesity.org/subs/fastfacts/Health_Effects.shtml
Benjamin, S., Valdez, R., Geiss, L., Rolka, D., and Narayan, K. “Estimated Number of Adults with Pre-diabetes in the US in 2000: Opportunities for Prevention.” Diabetes Care; 26(3) (March 2003): 645-649.
Fagot-Campagna, A., et al. “Type 2 Diabetes Among North American Children and Adolescents: an Epidemiologic Review and a Public Health Perspective.” Division of Diabetes Translation, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention (2000).
Ford, E., et al. “Prevalence of the Metabolic Syndrome Among US Adults.” Journal of the Americal Medical Association 287(3) (Jan 2002): 359.
Grundy, S., Brewer, H., Cleeman, J., Smith, S., and Lenfant, C. “Definition of Metabolic Syndrome.” National Heart, Lung, and Blood Institute and American Heart Association 109(2004): 433-438.
Mokdad, A., et al. “Prevalence of Obesity, Diabetes and Obesity Related Health Problems.” Journal of the Americal Medical Association 269(1)(Jan 2003): 76.
National Center for Chronic Disease Prevention and Health Promotion. “Overweight and Obesity. Division of Nutrition and Physical Activity.” 14 June 2006 http://www.cdc.gov/nccdphp/dnpa/obesity/index.htm
National Center for Health Statistics. “Fast Stats.” 14 Jun 2006 http://www.cdc.gov/nchs/fastats/overwt.htm
National Cholesterol Education Program. “Third Report on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults.” Circulation 106(2002): 3143–3421.
National Heart, Lung and Blood Institute and National Institutes on Health. “Classification of Overweight and Obesity by BMI, Waist Circumference, and Associated Disease Risks.” 14 June 2006 http://www.nhlbi.nih.gov/health/public/heart/obesity/lose_wt/bmi_dis.htm
National Heart, Lung and Blood Institute. “Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults.” 14 June 2006 http://www.nhlbi.nih.gov/guidelines/obesity/ob_home.htm
Neels, J., and Olefsky, J. “Inflamed Fat: What Starts the Fire? Journal of Clinical Investigation 116(1)(Jan 2006): 33-5.
Wang. Y, et al. “Comparison of Abdominal Adiposity and Overall Obesity to Predicting Risk of Type 2 Diabetes in Men.” American Journal of Clinical Nutrition (2005).
Weisberg, S., et al. “Obesity is Associated with Macrophage Accumulation in Adipose Tissue.” Journal of Clinical Investigation 112(12)(Dec 2003): 1796-1808. World Health Organization. “Obesity and Overweight.” 14 June 2006 http://www.who.int/hpr/NPH/docs/gs_obesity.pdf
Opt-in/Opt-out Enrollment Strategies
Opting (in or out) describes the chronic care management program’s enrollment design and is related to the way participants are engaged. They are either invited to participate and must take action to be included (opt-in) or informed that they are in the program unless they take action to be excluded (opt-out).
An outcome is a study measure that indicates the result of the performance (or nonperformance) of a population health management program. It is the result of an intervention. Outcome measures for population health management or chronic care management evaluation include financial, utilization, economic, quality, and humanistic outcomes (e.g., patient and provider satisfaction).
Todd, W., and Nash, D. Disease Management- A Systems Approach to Improving Patient Outcomes. San Francisco: Jossey-Bass (2001).
Outcome measures are those measures that population health management programs are designed to affect.
Population health management interventions are likely to affect many outcomes simultaneously. Program evaluators might find it advantageous to amass data from multiple areas.
- Clinical outcomes – changes in clinical process and short- and long-term outcome measures (e.g., percent of diabetics with improved HbA1c; percent of diabetics with amputations).
- Utilization (sometimes referred to as economic) measures – nonmonetary measures such as reduced resource consumption or resource utilization.
- Financial outcome measures – typically thought of as return on investment (ROI) or medical cost savings. Both direct and indirect cost measures may be used.
- Humanistic factors – patient quality of life, satisfaction. and impact on loyalty and retention of members.
- Quality measures – indicators of quality care promulgated and publicly reported by respected organizations such as National Committee for Quality Assurance (NCQA), URAC and The Joint Commission.
- Other measures that bring value to some or all stakeholders (e.g., reduced absenteeism, improved productivity).
Whenever possible, the domains outlined above should be reported in the context of numerator (the number of individuals with a successful achievement of an outcome) and denominator (those with the target condition of interest).
Care Continuum Alliance. (2010). Outcomes guidelines report, vol. 5. Washington, DC: Care Continuum Alliance.
An outlier is an observation in a data set that is an extremely high (or low) value. An outlier in a chronic care management program is typically the experience of a patient who incurred extraordinarily high costs, perhaps because (s)he was extremely sick, encountered an unexpected complication, or suffered a random, traumatic event.
The distribution of health expenditures of individuals is highly skewed. In general, a significant percentage of health plan members have no or very small expenditures, and 1-2% of enrollees have extremely high expenditures. Data or policy analysts must first clearly define or identify these outliers and then decide how they should be handled.
Outliers are important because an analysis (such as the estimated savings attributable to a chronic care management program) may be sensitive to outliers. The effect of a single patient’s high resource utilization, in a chronic care management evaluation with a small sample size, can disproportionately affect the results of the analysis. If outliers are completely ignored, the conclusions reached from the study could then be incorrect. In an analysis, outliers should be clearly defined, unambiguously handled, and their effect on the study’s overall conclusions should be discussed.
STATISTICAL DEFINITION OF AN OUTLIER
Two methods are commonly used to identify outliers. In the parametric approach, the sample mean and standard deviation are calculated. Observations that are greater than the mean + 3 standard deviations, or are less than the mean – 3 standard deviations, are outliers. Some analysts use 2.0 or 2.5 standard deviations to define outliers.
The second method is the nonparametric approach. The first and third quartiles (25th percentile and 75th percentile), referred to as Q1 and Q3, are calculated. The interquartile range is Q3 – Q1. Observations that are greater than the Q3 + 1.5 * interquartile range, or are less than Q1 – 1.5 * interquartile range, are outliers. Some analysts use 2.0 instead of 1.50 to define outliers.
Once outliers are flagged, they can be eliminated from the original data set to form what is called a trimmed data set. Ideally, the analysis should be run twice, first using the entire data set and then repeated using the trimmed data set. The results should then be compared.
A second, preferable way to handle outliers, is to reset extreme values to some less extreme value, sometimes called topcoding or winsorizing. In topcoding the actual values for all cases greater than the stop-loss value slv are replaced by slv. Topcoding is analogous to the stop-loss approach familiar to actuaries: a level of claims is determined, set either equal to the individual stop loss insurance attachment point or at a point determined by the variance of the claims distribution. In a small group, such as a small self-insured employer, a stop-loss limit may be $50,000 annually. For a larger group, the limit may be raised to $100,000 or more. In the stop-loss approach, claims above the limit are ignored in the specific calculation.
Some chronic care management savings calculations are performed using utilization data rather than direct dollar costs. Outliers may still occur, and a few patients with very long stays may distort utilization rates. In this situation, it is not possible to apply a stop-loss/topcoding approach, and the patient who experienced an extraordinarily long length of stay may have to be omitted from the analysis.
A histogram of the annual medical expenditures of 200 health plan members with heart failure is shown in Figure 1.
FIGURE 1. DISTRIBUTION OF ANNUAL MEDICAL EXPENDITURES (N=200)
These patients were randomly assigned to a chronic care management program. An alternative way of displaying these data involves first sorting the 200 observations of medical expenditures in ascending order, shown in Figure 2.
FIGURE 2. PLOT OF 200 ENROLLEES’ EXPENDITURES
Suppose we decide to use the second method of defining outliers, which involves using the interquartile range. We can easily calculate that Q1 (the 25th percentile) = $8,831, and Q3 = $16,741. If we use a factor of 1.5 to establish outliers, we find that there are three high outliers, since they exceed $16,741 + 1.5 * ($16,741-$8,831) = $28,606. If we topcoded these outliers, the values for the three outliers would be reset to $28,606.
Notice if we used a factor of 2.0 (instead of 1.5) to establish outliers, the cut-off value is $16,741 + 2 * ($16,741-$8,831) = $32,561, and the above data set would contain only one outlier.
EXAMPLE OF STOP-LOSS APPROACH
Consider the following (typical) distribution of individual member net paid claims:
In this example the average cost per member is $68.81. Assume that the population consists of 100 persons; if we add one more person with the maximum claim ($300,000) the average claim increases 3.5%. If, however, the additional claim is truncated at $100,000 the average claim increases only by 0.4%. Truncating the claim at $200,000 results in an increase in average expenditures per member of 1.9%.
The treatment of outliers will have an impact on a chronic care management savings calculation. Consider the following:
- Baseline population metrics are measured over the period 1/1/2000 to 12/31/2000;
- The intervention period begins 1/1/2001 and the first measurement period runs from 1/1/2001 to 12/31/2001;
- The populations for comparison are selected using exactly the same criteria;
- Baseline cost per member per month is $68.81, and intervention period cost per member per month is $65.00; and
- Trend (derived from an external source) is 5%.
To estimate the savings due to the Intervention, use the following calculations.
- In the absence of outlier or catastrophic claims, savings are:
$68.81 * 1.05 = $72.25 - $65.00 = $7.25
- We investigate the population used to derive Trend and find that there was a catastrophic claim of $300,000 in the projection year. Applying the stop-loss methodology to this catastrophic claim, using a $100,000 limit. reduces the Trend to 3%. (There is no corresponding distortion due to catastrophic claims in the disease management populations.) The estimated savings are now:
$68.81 * 1.03 = $70.87 - $65.00 = $5.87
- We investigate the chronic care management baseline population and find a claim of $300,000; there is no corresponding claim in the intervention period. We adjust the baseline period claims applying the stop-loss approach:
$67.75 * 1.05 = $71.14 - $65.00 = $6.14
Friedman, G. Primer of Epidemiology. 5th ed. New York: McGraw-Hill (2004).
Iezzoni, L. Risk Adjustment for Measuring Health Care Outcomes, 3rd ed. Chicago: Health Administration Press (2003).
Rothman, K. and Greelandad, S. Modern Epidemiology. 2nd ed. Baltimore: Lippincott, Williams & Wilkins (1998).
To purchase goods or subcontract services from an outside supplier or source. For example, an employer may choose to outsource part or all of its employee worksite wellness program by contracting with one or more population health vendors to deliver the services to employees.
outsourcing. (n.d.). Dictionary.com Unabridged. Retrieved December 08, 2010, from Dictionary.com website: http://dictionary.reference.com/browse/outsourcing