Health state utilities associated with weight loss in type 2 diabetes and obesity

Figures & data

Table 1. Sample characteristics.

Characteristics	Participants with T2D (N = 202)	Participants without T2D (N = 203)
Age (mean, SD) (years)	58.4 (10.7)	52.3 (12.0)
Gender (n, %)
Male	104 (51.5%)	100 (49.3%)
Female	98 (48.5%)	103 (50.7%)
Ethnicity (n, %)
White/Caucasian	162 (80.2%)	170 (83.7%)
African, Caribbean or Black	19 (9.4%)	7 (3.4%)
Asian	8 (4.0%)	11 (5.4%)
Mixed race^a	9 (4.5%)	5 (2.5%)
Other^b	4 (2.0%)	10 (4.9%)
Marital status (n, %)
Single	60 (29.7%)	81 (39.9%)
Married/Cohabitating/Living with partner	95 (47.0%)	88 (43.3%)
Other	47 (23.3%)	34 (16.7%)
Employment status (n, %)
Full-time work	62 (30.7%)	82 (40.4%)
Part-time work	40 (19.8%)	47 (23.2%)
Other^c	100 (49.5%)	74 (36.5%)
Education level (n, %)
University degree	67 (33.2%)	95 (46.8%)
No university degree	135 (66.8%)	108 (53.2%)
Weight and BMI
Weight (lbs) (mean, SD)	227.9 (37.2)	236.8 (51.5)
BMI (mean, SD)	36.1 (5.2)	37.3 (6.3)
BMI risk class (n, %)
Class 1: low-risk obesity (BMI = 25.0–29.9)	98 (48.5%)	85 (41.9%)
Class 2: moderate-risk obesity (BMI = 30.0–34.9)	65 (32.2%)	71 (35.0%)
Class 3: high-risk obesity (BMI >40.0)	39 (19.3%)	47 (23.2%)
Duration of diabetes (years)
Mean [SD]	9.5 (8.4)	n/a
Median [range]	7.0 [0.1–50.0]	n/a
Current diabetes medication (n, %)
No medication treatment	28 (13.9%)	n/a
Oral medication	123 (60.9%)	n/a
Non-insulin injectable medication	1 (0.5%)	n/a
Insulin	7 (3.5%)	n/a
Oral + non-insulin injectable	14 (6.9%)	n/a
Oral + insulin	23 (11.4%)	n/a
Non-insulin injectable + insulin	1 (0.5%)	n/a
Oral + non-insulin injectable + insulin	5 (2.5%)	n/a

^aMixed race as reported by participants in the sample with T2D includes 1/8 Tamil, 1/4 Irish, 5/8 English (n = 1); Anglo Indian (n = 1); British/Asian (n = 1); Caribbean + White (n = 1); Mixed African race (n = 1); Persian (n = 1); South African, Asian, and Welsh (n = 1); and White + Black Caribbean (n = 1). One participant did not specify. Mixed race as reported by participants in the sample without T2D includes Black British (n = 1); Brazilian/Latin/mixed Portuguese/indigenous people and Black (n = 1); Irish/Caribbean (n = 1); White/Black Caribbean (n = 1); and White Euro-American (n = 1). One participant did not specify.

^bOther ethnicity reported by participants in the sample with T2D includes Colombian (n = 1); Irish (n = 1); Jewish Ashkenazi (n = 1); and Scottish (n = 1). Other ethnicity reported by participants in the sample without T2D includes Arab (n = 1); Chinese (n = 1); Greek Cypriot (n = 1); Latin (n = 1); Latino American (n = 1); NZ Maori (n = 1); Persian (n = 1); Scottish (n = 1); Thai (n = 1); and Irish (n = 1).

^cOther employment status reported by participants includes: sample with T2D – homemaker (n = 5); student (n = 1); unemployed (n = 11); retired (n = 62); disabled (n = 16); carer (n = 2); consultant (n = 1); self-employed (n = 1); and volunteer (n = 1); sample without T2D – homemaker (n = 7); student (n = 1); unemployed (n = 12); retired (n = 38); disabled (n = 10); self-employed (n = 3); and volunteer (n = 3).

Table 2. Health state rankings^a.

Health state rank frequency (n, %)	1 = Most preferred	2	3	4	5	6 = Least preferred
Participants with T2D
A. Current weight	–	1 (0.5%)	1 (0.5%)	–	3 (1.5%)	197 (97.5%)
B. 2.5% less weight	1 (0.5%)	–	3 (1.5%)	10 (5.0%)	188 (93.1%)	–
C. 5% less weight	–	6 (3.0%)	11 (5.4%)	185 (91.6%)	–	–
D. 10% less weight	10 (5.0%)	13 (6.4%)	178 (88.1%)	1 (0.5%)	–	–
E. 15% less weight	22 (10.9%)	173 (85.6%)	2 (1.0%)	3 (1.5%)	2 (1.0%)	–
F. 20% less weight	169 (83.7%)	9 (4.5%)	7 (3.5%)	3 (1.5%)	9 (4.5%)	5 (2.5%)
Participants without T2D
A. Current weight	–	–	1 (0.5%)	–	3 (1.5%)	199 (98.0%)
B. 2.5% less weight	–	1 (0.5%)	2 (1.0%)	6 (3.0%)	194 (95.6%)	–
C. 5% less weight	1 (0.5%)	2 (1.0%)	6 (3.0%)	194 (95.6%)	–	–
D. 10% less weight	7 (3.4%)	9 (4.4%)	187 (92.1%)	–	–	–
E. 15% less weight	21 (10.3%)	178 (87.7%)	–	3 (1.5%)	1 (0.5%)	–
F. 20% less weight	174 (85.7%)	13 (6.4%)	7 (3.4%)	–	5 (2.5%)	4 (2.0%)

^aRankings had a possible range from 1 to 6, with lower numbers indicating more preferred health states.

Table 3. Health state utilities.

	Health state utility^a		Utility increases^b compared to current weight
	Mean	SD	Mean utility increase	SD
Obesity with T2D (N = 202)
A. Current weight	0.797	0.184	–	–
B. 2.5% less weight	0.808	0.176	0.011	0.029
C. 5% less weight	0.820	0.171	0.023	0.050
D. 10% less weight	0.839	0.167	0.042	0.068
E. 15% less weight	0.850	0.158	0.053	0.087
F. 20% less weight	0.857	0.155	0.060	0.093
Obesity without T2D (N = 203)
A. Current weight	0.817	0.191	–	–
B. 2.5% less weight	0.832	0.180	0.015	0.041
C. 5% less weight	0.846	0.173	0.029	0.055
D. 10% less weight	0.866	0.164	0.048	0.077
E. 15% less weight	0.884	0.159	0.067	0.095
F. 20% less weight	0.894	0.153	0.077	0.108

^aTTO scores are on a scale anchored with 0 representing dead and 1 representing full health.

^bCalculated by subtracting the utility of health state A from the utility of each other health state.

Figure 1. Change in utility as a function of percentage change in weight^a,b,c.

^aThese are plots of the two regression models from Table 5 that do not include covariates. Both models are log-linear regressions of change in utility as a function of the natural log of the weight change represented in the health states.

^bModel for participants with diabetes: y = -0.00860970799921508 + 0.0216883437387603*log(x)

^cModel for participants without diabetes: y = -0.00781776466699391 + 0.023685903755767*log(x)

Table 4. Modeling weight change without covariates.

Effect	Estimate	StdErr	d.f.	t Value	p Value
Natural log of weight change as a predictor of utility change (obesity with T2D sample, N = 202)
Intercept	−0.0086	0.0022	201	−3.88	<.001
Natural log of weight change	0.022	0.0023	201	9.25	<.0001
Natural log of weight change as a predictor of utility change (obesity without T2D, N = 203)
Intercept	−0.0078	0.0032	202	−2.42	<.05
Natural log of weight change	0.024	0.0027	202	8.88	<.0001

Abbreviation. d.f., degrees of freedom.

Table 5. Four equations proposed for estimating utility change based on weight decrease.

Intended population	Covariates and interaction terms	Equation for predicting utility change based on percentage weight decrease
With obesity and T2D	None	Utility change = −0.0086 + 0.022 (LogPWD)
With obesity without T2D	None	Utility change = −0.0078 + 0.024 (LogPWD)
With obesity and T2D	Age, BMI, weight change by age interaction, weight change by BMI interaction	Utility change = −0.014 + 0.028 (LogPWD) + 0.00026 (age) − 0.00027 (BMI) − 0.00055 (LogPWD*age) + 0.00069 (LogPWD*BMI)
With obesity without T2D	Age, weight change by age interaction	Utility change = −0.022 + 0.040 (LogPWD) + 0.00027 (age) − 0.00030 (LogPWD*age)

Abbreviation. LogPWD, natural logarithm of weight change quantified as the percentage of weight decrease; BMI, body mass index of each patient.

Table 6. Examples of how the equations can be used to estimate utility increase associated with percentage of weight decrease.

Intended population	Weight decrease (%)	Age	BMI	Equation	Utility increase
With obesity and T2D	7	–	–	Utility change = -0.0086 + 0.022 (ln(7))	0.034
	18	–	–	Utility change = -0.0086 + 0.022 (ln(18))	0.055
With obesity without T2D	7	–	–	Utility change = -0.0078 + 0.024 (ln(7))	0.039
	18	–	–	Utility change = -0.0078 + 0.024 (ln(18))	0.062
With obesity and T2D	7	40	30	Utility change = -0.014 + 0.028 (ln(7)) + 0.00026 (40) – 0.00027 (30) – 0.00055 (ln(7)*40) + 0.00069 (ln(7)*30)	0.040
	7	65	36	Utility change = -0.014 + 0.028 (ln(7)) + 0.00026 (65) – 0.00027 (36) – 0.00055 (ln(7)*65) + 0.00069 (ln(7)*36)	0.026
With obesity without T2D	7	40	–	Utility change = -0.022 + 0.040 (ln(7)) + 0.00027 (40) – 0.00030 (ln(7)*40)	0.043
	7	65	–	Utility change = -0.022 + 0.040 (ln(7)) + 0.00027 (65) – 0.00030 (ln(7)*65)	0.035

Abbreviation. ln, natural logarithm; BMI, body mass index.

Data availability statement

Data can be made available on reasonable request.