SEARCH

SEARCH BY CITATION

Keywords:

  • body-mass index;
  • body weight;
  • type 2 diabetes;
  • utility

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Background:  Weight gain is a common side effect of many therapies for type 2 diabetes (T2DM). Selecting utility values for incorporation into cost-utility analyses (CUAs) of T2DM therapies is difficult because of variations in methodologies to elicit utilities and other study limitations.

Methods:  A review of the medical literature was conducted to identify studies assessing the impact of body weight on patient utility.

Results:  Eighteen articles presented either: 1) utility values by body-mass index (BMI) or body weight, or 2) the change in utility scores or quality-adjusted life-years based on unit changes in BMI or body weight. Regardless of the study population or methodology used to elicit utility scores, all studies reviewed found that as body weight increased, patient utility decreased. Utility scores obtained using standard gamble were generally higher than those using time trade-off(TTO) or the EQ-5D. Most studies reported utility scores stratified by BMI and used regression analyses to attribute the difference in utility scores to differences in weight while controlling for other factors. Studies generally assumed a constant change in utility occurs with a one unit change in BMI. Recent studies, however, demonstrate the magnitude of changes in utility may vary depending on 1) valuing weight loss versus weight gain; 2) valuing a small or large change in body weight; and 3) baseline BMI.

Conclusions:  Various utility values associated with body weight using different methodologies have been published. Careful consideration should be given to determine the most appropriate utility values to use in CUAs of T2DM therapies.


Introduction

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Obesity has reached epidemic proportions worldwide as more than 1 billion adults are overweight with at least 300 million adults identified as obese [1]. An adult classified as overweight has a body-mass index (BMI) between 25.0 and 29.9 kg/m2, while an individual who is obese has a BMI of at least 30 kg/m2[2].

It is well established that the prevalence of many conditions such as type 2 diabetes mellitus (T2DM), high blood pressure, dyslipidemia, and cardiovascular disease increases with an increase in body weight [2–4]. In addition, being overweight or obese contributes to decreases in life expectancy [5,6] and leads to decreases in health-related quality of life (HRQOL) [7–9].

Being overweight or obese is clinically important in T2DM because it can exacerbate metabolic abnormalities, leading to further decreases in glycemic control and worsening of diabetes symptoms [10,11]. In fact, lifestyle interventions to promote weight loss are generally considered first-line therapy for T2DM and should be included as part of diabetes management for all patients [12].

Despite the need for weight loss in T2DM, weight gain is a common side effect of many therapies for T2DM and presents a challenge to clinicians because patients with T2DM are generally overweight before starting any type of antidiabetic medication [11,13]. Upon initiation of therapy, the average amount of weight gain seen per patient varies among antidiabetic agents. Weight gain with various insulins can range from approximately 2 to 4 kg per patient, and is likely proportional to the correction of glycemia [12]. While metformin generally leads to no change in body weight, and agents, such as exenatide, may lead to a reduction in body weight [12,14–18].

Because weight gain is common among most therapies to treat T2DM, when examining the value of drug therapies it should be important to assess not only the potential impact of weight gain on clinical outcomes, but also the potential impact on patient HRQOL, preferences, and utility. This review was undertaken to evaluate the availability of data in the published medical literature assessing the impact of body weight on utility scores and to determine what impact, if any, different methodologies for eliciting utility scores may have on the resulting utility values. Specific research questions included:

  • • 
    What is the directional relationship between weight change and utility in individuals with or without type 2 diabetes?
  • • 
    What is the magnitude of utility change based on body weight or BMI in patients with or without type 2 diabetes?
  • • 
    What are the similarities and differences among published utility/disutility values associated with change in body weight?

Methods

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

A review of the medical literature from 1996 to present was conducted using MEDLINE, EMBASE, and PsycINFO to identify studies assessing the impact of body weight on patient utility. Additionally, abstracts from relevant professional conferences from January 2004 to May 2006 were searched. Keywords and MeSH terms included: body weight, body-mass index, BMI, obesity, diabetes, weight reduction, weight loss, quality of life, utilities, health status, health state preferences, cost-utility, quality-adjusted life-year, cost-effectiveness, rimonabant, orlistat, sibutramine, and all combinations of these terms. References listed in relevant journal articles were also reviewed. Because the focus was to evaluate utility scores based on a change in body weight for utilization of antidiabetic therapies, utility scores associated with specific surgical interventions or dietary modification programs were excluded from this review. The literature search was current as of May 2006.

Studies were included in the review if they were:

  • • 
    conducted in adults (18 years old or less);
  • • 
    written in English;
  • • 
    used a validated rating scale or methodology (e.g., standard gamble [SG] or time trade-off [TTO]) to elicit utility scores; and
  • • 
    provided a clear definition of utility score anchors of 0 to 1.

Studies were excluded from the review if they:

  • • 
    did not present utility scores or change in utility scores by BMI or body weight;
  • • 
    focused on utility scores obtained from evaluations of surgical treatments or dietary modification for obesity rather than pharmacological treatments;
  • • 
    were conducted in children (less than 18 years old);
  • • 
    only included utility estimates derived from another study; and
  • • 
    did not provide a clear methodology for eliciting utility scores.

Results

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

General Findings

Sixty articles fitting the search criteria above were identified. Of these, 42 were eliminated because they did not provide specific data on patient utility, were conducted during the evaluation of a surgical weight loss procedure, or did not use validated instruments to evaluate patient utility. Eighteen articles meeting the inclusion criteria described above were identified and presented either: 1) utility values by BMI or body weight, or 2) changes in utility scores or quality-adjusted life-years (QALYs) based on unit changes in BMI or body weight. In one case, more than one article was published based on the findings from one study. Bagust and Beale [19] present data for the entire Cost of Diabetes in Europe—Type 2 (CODE-2) study population in five European countries, while Redekop et al. [20] present a subset of CODE-2 data from The Netherlands. These articles present different aspects of utility data and are included as separate entries in this review.

Table 1 presents the articles included in this review alphabetically by author and grouped by study population (e.g., diabetics, general population, hospital inpatients). The main differences between studies included variations in sample size, study population, country included in the study, methods used to elicit utility scores, and main study outcomes (e.g., utility score by BMI category or change in utility score associated with a change in BMI measurement).

Table 1.  Comparison of studies evaluating the impact of body weight on patient utility
SourcenStudy populationCountryUtility methodsMain outcome examined by study
  • *

    This study is a cost-utility analysis of sibutramine and uses quality of life estimates from four clinical trials [37] (n = 854) and from the German Sibutramine Adiposity Therapy Trial, which is unpublished. The sample size for the German trial was not provided. The quality of life estimates were transformed into utility values by Warren et al. [34] for use in a UK-specific cost-utility analysis.

  • BMI, body-mass index; CODE-2, cost of diabetes type 2 in Europe; EQ-5D, EuroQoL 5-D instrument; QWB, Quality of Well-Being Index; QWB-SA, Quality of Well-Being Index Self-Administered; SF-36, Medical Outcomes Study Short-Form 36; SF-12, Medical Outcomes Study Short-Form 12; SF-6D, a summary preference-based health measure derived from the SF-36; SG, standard gamble; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; TTO, time trade-off; VAS, visual analog scale.

[19]4,641T2DMBelgium, Italy, The Netherlands, Spain, SwedenTTO, VASChange in utility score with a change in BMI
[21]2,048T1DM or T2DMUSQWB-SAUtility score by BMI category, gender, and diabetes therapy
[22]402Obese patients with and without hyperglycemiaUSTTO, VASChange in utility score with a decrease in BMI Utility score by BMI category
[23]27,249(2,575 with diabetes)Hospital inpatients and outpatientsUKEQ-5D indexChange in utility score with a change in BMI and utility score by BMI category
[24]129T2DMScotland and EnglandVAS, SGChange in utility score with a 3% or 5% gain or loss of body weight
[25]14,775Hospital inpatients and outpatients with a subset of T1DM or T2DM patientsUKEQ-5D indexUtility score by BMI category
[20]1,348T2DMThe NetherlandsEQ-5D Index and VASUtility score by BMI category
[26]18,223Hospital inpatients and outpatientsUKEQ-5DChange in utility score with a change in BMI
[27]13,152Hospital inpatients and outpatientsUKEQ-5D, SF-6DChange in utility score with a change in BMI
[28]1,326General populationUSQWBUtility score by BMI category
[29]13,646General populationUSSF-12, EQ-5D Index, EQ-5D VASChange in utility score by BMI category
[30]12,661General populationAustraliaSF-36, SF-6DChange in utility score with a change in BMI and utility score by BMI category
[31]31,397General populationUSHALexUtility score by BMI category and gender
[32]1,865General practice patients ≥45 years oldUKEQ-5D Index, EQ-5D VAS, SF-6DUtility score by BMI category
[33]38,151General populationCanadaHUI-3Utility score by BMI category
[34]*Individuals with a BMI between 27 and 40 kg/m2*SF-36Change in utility score by kg of weight lost
[35]365Hospital-based primary care patientsUSSG, TTOUtility score at current weight, a BMI = 25 kg/m2 or with a 5%, 10%, 20% weight loss
[36]General populationUKEQ-5D IndexUtility score by BMI category and gender

Four articles reported the impact of weight loss on utility scores and, as expected, utility increased as body weight decreased [22,24,34,35]. Seven papers reported studies conducted in patients with diabetes or had a subset of patients with diabetes from which utility values specific to diabetic patients could be derived [19–25]. Two articles were based on obese or overweight patients, or included a subset of obese or overweight individuals [22,34]. Five articles included hospital inpatients or outpatients [23,25–27,35] and six studies were conducted in individuals from the general population of various countries [28–31,33,36]. Table 2 shows studies whose main outcomes present utility scores by BMI level or body weight. Table 3 includes studies in which the main outcome is to present the change in utility score associated with specific increases or decreases in BMI or body weight.

Table 2.  Utility values by body-mass index or body weight from the published medical literature
SourceMethod(s)Study populationMain outcome
  • *

    Data are estimates obtained from a graph presented in the article. Actual values were not presented.

  • BMI, body-mass index; EQ-5D, EuroQoL 5-D instrument; HALex, Health and Activities Limitation Index; HUI-3, Health Utilities Index—Version 3; QWB, Quality of Well-Being Index; QWB-SA, Quality of Well-Being Index Self-Administered; SF-36, Medical Outcomes Study Short-Form 36; SF-12, Medical Outcomes Study Short-Form 12; SF-6D, a summary preference-based health measure derived from the SF-36; SG, standard gamble; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; TTO, time trade-off; VAS, visual analog scale.

[22]TTO, VASOverweight and obese patients with and without hyperglycemia TTOVAS
Overweight (25–29.9 kg/m2)0.880.77
Obese I (30.0–34.9 kg/m2)0.850.74
Obese II (35–39.9 kg/m2)0.790.67
Obese III (≥40 kg/m2)0.810.69
[23]*EQ-5D indexHospital inpatients and outpatientsNormal weight (<25 kg/m2) 0.73
Overweight (25–29.9 kg/m2) 0.69
Obese (30–39.9 kg/m2) 0.62
Extremely obese (≥40 kg/m2) 0.47
T2DM subsetNormal weight (<25 kg/m2) 0.62
Overweight (25–29.9 kg/m2) 0.59
Obese (30–39.9 kg/m2) 0.51
Extremely obese (≥40 kg/m2) 0.40
[25]EQ-5D IndexHospital inpatients and outpatients with a subset of T1DM or T2DM patientsBMINo diabetesT1DMT2DM
220.730.790.60
240.720.780.57
260.690.820.53
280.670.750.58
300.650.580.55
320.600.480.55
340.560.580.47
360.590.330.44
[20]EQ-5D index and VAST2DMAll T2DM patients: 0.74 (Index); 0.68 (VAS)
Obese: 0.70 (Index); 0.66 (VAS)
Not obese: 0.77 (Index); 0.69 (VAS)
[28]QWBGeneral populationUnderweight (<20 kg/m2) 0.698
Normal (20–24.9 kg/m2) 0.709
Overweight (25–29.9 kg/m2) 0.695
Obese (≥30.0 kg/m2) 0.663
[31]HALexGeneral population MenWomen
Normal (18.5–24.9 kg/m2)0.860.86
Overweight (25–29.9 kg/m2)0.870.82
Obese class 1 (30–34.9 kg/m2)0.830.79
Obese class 2 (35–39.9 kg/m2)0.790.75
Seriously obese (40–49.9 kg/m2)0.760.71
Superobese (50–90 kg/m2)0.680.60
[32]EQ-5D index, EQ-5D, VAS, SF-6DGeneral practice patients ≥45 years old IndexEQ-VASSF-6D
Normal0.8077.80.78
Preobese0.7875.40.77
Obese I0.7069.80.72
Obese II0.6866.80.67
Obese III0.6258.50.67
[33]HUI-3General populationNormal (19–24.9 kg/m2) 0.93
Overweight (25–29.9 kg/m2) 0.93
Obese (30.0–34.9 kg/m2) 0.91
Morbidly obese (≥35 kg/m2) 0.89
[36]EQ-5D IndexGeneral population MenWomen
Normal (18.5–24.9 kg/m2)0.8770.879
Overweight (25–29.9 kg/m2)0.8940.871
Obese (≥30 kg/m2)0.8580.812
Table 3.  Change in utility values associated with a change in body-mass index or body weight
SourceMethod(s)Study populationMain outcome
  • *

    Data are estimates obtained from a graph presented in the article. Actual values were not presented.

  • BMI, body-mass index; EQ-5D, EuroQoL 5-D instrument; HALex, Health and Activities Limitation Index; HUI-3, Health Utilities Index—Version 3; QWB, Quality of Well-Being Index; QWB-SA, Quality of Well-Being Index Self-Administered; SF-36, Medical Outcomes Study Short-Form 36; SF-12, Medical Outcomes Study Short-Form 12; SF-6D, a summary preference-based health measure derived from the SF-36; SG, standard gamble; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; TTO, time trade-off; VAS, visual analog scale.

[19]TTO, VAST2DMFor each one unit increase in BMI more than 25, utility decreases by 0.0061 (via TTO).
For each one unit increase in BMI more than 25, utility decreases by 0.29 (via VAS)
[21]QWB-SAT1DM or T2DMUtility scores decreased by 0.016 units for obese patients vs. normal patients with T1DM
Utility scores decreased by 0.021 units for obese patients vs. normal patients with T2DM
Overweight and obese patientsFor each unit decrease in BMI, utility increases by 0.0166 (via TTO) and 0.0203 (via VAS)
[22]TTO, VASOverweight and obese patients with hyperglycemia (BMI ≥ 28)For each unit decrease in BMI, utility increases by 0.0285 (via TTO) and 0.036 (via VAS)
[23]*EQ-5D indexHospital inpatients and outpatientsIncreasing BMI by one unit decreases utility by 0.0079
T2DM subsetIncreasing BMI by one unit decreases utility by 0.01
[24]VAS, SGT2DM VASSG
Current weight64.40.89
3% Weight gain47.80.85
5% Weight gain39.80.83
3% Weight loss71.60.91
5% Weight loss77.20.92
[20]EQ-5D index and VAST2DMUtility scores decreased by 0.067 units for obese patients vs. nonobese patients
[26]EQ-5DHospital inpatients and outpatientsUtility decreases at a rate of 0.0133 for BMI = 29–30 and 0.0325 for BMI = 34–35
[27]EQ-5D, SF-6DHospital inpatients and outpatientsFor BMI values >25 kg/m2, for each unit increase in BMI, utility decreases by 0.0168 (EQ-5D) and 0.00704 (SF-6D)
[30]SF-36, SF-6DGeneral populationFor each unit increase in BMI, utility decreases by 0.0024 (in men) and 0.0034 (in women)
[32]EQ-5D index, EQ-5D VAS, SF-6DGeneral practice patients ≥45 years oldChange in utility for normal vs. obese (classes I-III):
Index −0.040
VAS −5.0
SF-6D −0.038
[34]SF-36Individuals with a BMI 27–40Utility gain per kg of weight lost was 0.00142 (in placebo-treated subjects) and 0.00185(in sibutramine-treated subjects)
[35]SG, TTOHospital-based primary care patientsNormal weight patients: No change with varying amounts of weight loss
 OverweightObese
Current Weight (SG)0.950.88
 (TTO)0.990.97
5% weight loss (SG)0.960.91
(TTO)0.990.97
10% weight loss (SG)0.980.93
(TTO)0.990.98
20% weight loss (SG)0.980.96
(TTO)1.000.99
BMI of 25 kg/m2 (SG)0.980.98
(TTO)1.001.00

Regardless of the study population or methodology used to elicit utility scores, all studies reviewed generally found that as body weight increased, patient utility decreased. In two studies [22,25], an increase in utility was observed for some higher BMI subgroups rather than the expected decrease in utility. Reasons for these unexpected results were not clearly explained by the study authors. Overall, however, the structure of the data for studies included in this review indicated an inverse relationship between obesity or higher BMI and health utility.

Main Differences and Limitations of Studies Evaluating Body Weight and Utility

There are several limitations in the studies evaluating the impact of body weight on patient utility from the medical literature. In addition to general study limitations, such as small sample sizes or differences in BMI thresholds, there are several issues that may compromise the use of current utility estimates from the published literature in CUAs of therapies for T2DM. Caution should be used when comparing the utility values from each study because the studies identified differ in four main ways, including differences in: 1) study population; 2) methods for eliciting utility scores; 3) expression of study outcomes; and 4) underlying assumptions used in determining the change in utility scores seen with a change in body weight or BMI (Table 1).

Study population.  The impact of body weight or BMI on patient utility was evaluated using several different study populations, including individuals with diabetes, obese individuals, hospital inpatients or outpatients, general practice patients, and the general population. Studies were conducted in Australia, Europe, and North America (Table 1).

In general, utility scores obtained from patients with diabetes were lower across all levels of BMI than scores obtained from the general population (Table 2). Utility scores for individuals of normal weight without diabetes (BMI of less than 25 kg/m2) ranged from a low of 0.71 [28] to a high of 0.93 [33]. For obese individuals without diabetes, utility scores ranged from a low of 0.60 [31] to a high of 0.91 [33]. For patients with diabetes who are of normal weight (BMI of less than 25 kg/m2), utility scores ranged from 0.57 [25] to 0.77 [20]. Utility scores decreased for individuals with diabetes considered obese (BMI of 30 kg/m2 or more) and ranged from a low of 0.33 [25] to a high of 0.70 [20].

When evaluating utility, the choice of study population is particularly important because the impact of a change in weight varies across different study populations. For example, weight management is a cornerstone of diabetes management and patients with T2DM who are overweight or obese are likely to have been told to lose weight as part of the therapeutic approach to managing their disease. Thus, a change in body weight may have more impact in these patients and could have a greater effect on utility compared with the general population. Although statistical analyses were not conducted, the findings from this review are consistent with this hypothesis, in that patients with diabetes had slightly larger changes in utility scores with a one unit change in BMI compared with individuals who did not have diabetes [22,23] (Table 3).

Methodologies to elicit utility scores.  Of the 18 articles evaluating the impact of body weight on patient utility, seven used more than one methodology to elicit utility scores. Multiattribute health status classification systems, such as the EQ-5D, were the most frequently used method to elicit utility scores in studies evaluating the impact of body weight on patient utility (Fig. 1). Only two studies used SG and three studies used TTO to elicit utility scores, which are considered the gold standard methods for obtaining utility scores. As seen in other studies [38–40], utility scores obtained using SG techniques were generally higher than those using other methods.

image

Figure 1. Eighteen articles evaluated the impact of body weight on patient utility. Several articles used more than one methodology to elicit utility scores. Multiattribute health status classification systems, such as the EQ-5D, were the most frequently used method to elicit utility scores in studies evaluating the impact of body weight on patient utility. MA, multiattribute health status classification system; SG, standard gamble; TTO, time trade-off; VAS, visual analog scale.

Download figure to PowerPoint

Wee et al. [35] used an SG approach to evaluate the impact of a 5%, 10%, or 20% weight loss in patients in a primary care setting in the United States. Utility scores increased as the magnitude of weight loss increased, regardless of baseline weight. Similar improvements in utility were seen in patients with type 2 diabetes in the United Kingdom [24]. When type 2 diabetes patients evaluated a 5% weight loss, utility increased from 0.89 to 0.92 (a 0.03 increase in utility, which is similar to the findings of Wee et al. [35]). The study by Secnik et al. was the only article to evaluate the impact of a specific amount of weight gain on utility. When patients with type 2 diabetes evaluated a 5% weight gain, utility decreased from 0.89 to 0.83 [24].

Generally, existing studies have not examined the utility or disutility of small differences in body weight from the patient's perspective. Many published studies have found that individuals with a higher BMI generally report lower utility compared with individuals with lower BMI levels [20,23,30–32]. Other than the studies by Secnik et al. [24] and Wee et al. [35], however, no studies have directly assessed patients' preferences, and the associated utilities or disutilities, for a weight higher or lower than his or her own.

Expression of study outcomes.  Published studies mainly used multiattribute health status classification systems or visual analog scales (VAS) to evaluate utility values in a population and then stratified by BMI. These studies present utility values by BMI and attribute the difference in utility scores to differences in BMI oftentimes controlling for observed factors using regression analyses. Twelve studies presented the change in utility associated with a specific change in body weight, BMI, or obesity classification (e.g., obese I, obese III, obese III) (Table 3). The change in utility ranged from a low of 0.0061 to a high of 0.29 per one unit change in BMI. Both the highest and lowest changes in utility values, however, came from the same study by Bagust and Beale [19] and showed the wide variation in utility changes obtained using TTO versus VAS.

Of the studies presenting a change in utility score associated with a change in BMI in individuals without diabetes, four present the utility change attributable to a one unit change in BMI [22,23,27,30] and one presents the utility change per kilogram of weight lost [34]. There is a large variation in the utility change seen in studies conducted in individuals without diabetes. With a one unit increase in BMI, utility scores decreased by a low of 0.0024 obtained using the SF-36 [30] to a high of 0.0168 obtained using the EQ-5D [27]. Hakim et al. [22] found that in obese patients, for each unit decrease in BMI, utility increased by approximately 0.02 via TTO or VAS.

These studies do not specifically evaluate how a change in weight will affect patients, but rather they evaluate the HRQOL of patients at various BMI levels. This is, however, not the same as evaluating the impact of a specific amount of weight change. Only two studies specifically evaluated the change in utility scores by asking patients to evaluate the impact of specific amounts of weight gain or loss [24,35].

Underlying assumptions.  Most studies presented utility scores stratified by BMI categories and used regression analyses to attribute the difference in utility scores to differences in weight while controlling for other factors. These studies generally assumed that a constant change in utility occurs with a one unit change in BMI. Two recent studies, however, demonstrate that the magnitude of changes in utility scores may vary depending on 1) whether a patient is valuing weight loss or gain; 2) whether a smaller or larger change in body weight is being evaluated; and 3) baseline BMI [24,26].

In studies that present the change in utility score associated with a change in BMI or body weight, it is difficult to tell whether the relative change would be the same if a patient lost weight versus gained weight. Only one study [24] evaluated equivalent magnitudes of weight loss and weight gain within a study population. Secnik et al. [24] evaluated the impact of a 3% and 5% weight loss and a 3% and 5% weight gain within the same population of patients with T2DM. When patients evaluated the impact of weight gain, the decreases in utility scores from baseline (patient's current weight) were larger than the increases in utility scores from baseline seen when patients evaluated weight loss (Table 3).

Conclusions

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Although several studies have evaluated the impact of being overweight or obese on HRQOL, the impact of changes in body weight on patient utility is less clear. A thorough review of published estimates of utility related to obesity was conducted by Dixon et al. [27] and provided insight into the importance of including the impact of obesity on HRQOL in economic analyses fully to evaluate therapeutic interventions. The current study adds to the existing literature in several important ways: 1) this review includes several recent publications evaluating the impact of changes in body weight on patient utility; 2) the impacts of both weight gain and weight loss are included in this review; and 3) studies conducted in individuals with and without diabetes are included, which is particularly important because obesity plays a key role in the development and progression of diabetes. Additionally, because antidiabetic agents have different effects on body weight, it is important to quantify the impact of different changes in weight on patient preferences fully to understand the value of each therapy.

There were several limitations to this review of the medical literature. First, the literature search was limited to articles published in English, which could potentially be a source of bias and limits the generalizability of the findings of this review. Second, the literature search excluded utility values obtained from studies evaluating surgical procedures for weight loss; however, utility values obtained from studies evaluating pharmaceutical therapies for weight loss were included. Because surgical procedures may lead to a greater magnitude of weight loss compared with pharmaceutical therapies, the impact on patient utility may be much larger and may not be appropriate for inclusion in CUAs to evaluate the value of pharmaceutical therapies for conditions, such as T2DM. Third, to allow for comparison between studies, all of the studies included in this review evaluated the change in body weight using BMI levels or categories or magnitude of weight lost or gained. Studies using other clinically relevant measures of weight loss (i.e., waist to chest ratio) were not included in this literature review. To our knowledge, however, the majority of studies evaluating the impact of body weight on patient utility were conducted using BMI or body weight. Other measures of weight loss have not been used to evaluate the impact of body weight on patient utility.

This literature review has identified several recent studies describing the impact of body weight, BMI, and obesity on patient utility scores in patients with and without diabetes. Although utility scores varied by study, regardless of the study population or methodology used to elicit utility scores, each study found an inverse relationship between body weight and patient utility. This finding is consistent with that of Dixon et al. [27], who evaluated the relationship between patient utility and obesity.

Many studies evaluating the impact of body weight on utility have primarily used a multiattribute approach with generic questionnaires or a VAS. The majority of published studies present utility scores by BMI levels and do not actually ask the individual to evaluate how a change in body weight will affect his or her preference for different magnitudes of weight loss. Additionally, some published utility studies assume that the impact of weight on utility scores is a linear function by assuming that the change in utility score is the same for any one unit change in BMI, regardless of the starting or ending BMI. Additional research fully to evaluate this assumption is needed.

When conducting CUAs for antidiabetic agents, the impact of a change in weight may play an important role in determining patient preferences for alternative therapies when estimating cost per QALY. Consistent with conclusions from Tucker et al. [41], there is a lack of suitable utility estimates for economic modeling of antidiabetic agents that account for the effects of changes in BMI on health state preferences. Various utility values associated with body weight using different methodologies have been published. Careful consideration should be given to determine the most appropriate utility values to use in CUAs of T2DM therapies.

Because there is wide variation in the impact of BMI on utility values, care should be taken to choose utility values for the base case of a CUA that most closely reflect the potential study population and drug attributes being evaluated. Conducting sensitivity analyses by using estimates from different studies in the published literature is strongly recommended.

Source of financial support: Funding for this study was provided by Eli Lilly and Company.

References

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References