Obesity

advertisement
Cut-offs for childhood BMI in prediction of
cardiovascular disease risk factors in adulthood
Leah Li
MRC Centre of Epidemiology for Child Health
UCL Institute of Child Health
Imperial College, 16 Nov 2011
1
Childhood BMI
•
•
Trends in child obesity in recent decades
BMI tracks from childhood to adulthood
(obese children more likely to become obese adults)
•
•
•
Adult obesity linked to increased risks of CVD, type 2
diabetes
This raises issue as to whether individuals with high risk
of adult disease can be identified from their BMI in
childhood
Links between childhood obesity and future health
outcomes?
2
Reference standards for child BMI
• Most based on age- gender-specific BMI distribution
IOTF cut-offs‡ - correspond to percentile curves for 18y BMI
of 25 & 30 kg/m2
Percentile (85th, 90th)† - arbitrary
• Defined within reference population, not directly linked to
adult health outcomes
• Internally derived cut-offs for child BMI (population specific)
(require data on BMI in childhood and disease risks in adulthood)
† Kuczmarski et al CDC growth charts 2001
‡ Cole et al 2000
3
Study sample
1958 British Birth Cohort - born in Great Britain in one week, Mar 1958
(≈17,000), followed up to 50y. Medical assessment at 45y (n=9,377)
Childhood BMI (kg/m2) at 7, 11, 16y
CVD risk factors at 45y:
• Obesity (BMI ≥ 30 kg/m2)
• Abdominal obesity (waist ≥ 102/88 cm)
• High risk of type 2 diabetes (HbA1c 7%; on medication; reported type
2 diabetes; excluding T1 diabetes)
• Hypertension ( ≥ 140/90 mmHg; on medication)
• Low HDL-C ( < 1/1.3 mmol/l)
• High LDL-C ( > 4.13 mmol/l)
• High triglycerides ( ≥ 2.3 mmol/l)
4
CVD risk factors at 45y in 1958 cohort
Overweight
Obese
Males
49.7 %
25.2 %
Females
32.7 %
23.5 %
Abdominal obesity
32.7 %
36.9 %
3.7 %
2.4 %
Hypertension‡
Low HDL
High LDL
34.6 %
9.3 %
24.4 %
16.3 %
19.6 %
14.8 %
High triglycerides
41.5 %
16.0 %
Type 2 diabetes or HbA1c ≥7%
5
Childhood BMI
IOTF cut-offs (%)
Males
Females
Overweight@7
Obese@7
18.2(5.5)
21.1(0.8)
18.0(7.2)
21.0(1.6)
Overweight@11
Obese@11
20.9(6.0)
25.6(0.8)
21.2(7.9)
26.1(1.1)
Overweight@16
Obese@16
23.9(6.3)
28.9(1.4)
24.4(9.3)
29.4(1.2)
Overweight@45
Obese@45
25.0(49.7)
30.0(25.2)
25.0(32.7)
30.0(23.5)
6
Childhood BMI
IOTF cut-offs (%)
Males
Females
Overweight@7
Obese@7
18.2(5.5)
21.1(0.8)
18.0(7.2)
21.0(1.6)
Overweight@11
Obese@11
20.9(6.0)
25.6(0.8)
21.2(7.9)
26.1(1.1)
Overweight@16
Obese@16
23.9(6.3)
28.9(1.4)
24.4(9.3)
29.4(1.2)
Overweight@45
Obese@45
25.0(49.7)
30.0(25.2)
25.0(32.7)
30.0(23.5)
At 7-16y, 6.3-10.5%
classified as overweight
or obese
7
Childhood BMI
IOTF cut-offs (%)
Males
Females
Overweight@7
Obese@7
18.2(5.5)
21.1(0.8)
18.0(7.2)
21.0(1.6)
Overweight@11
Obese@11
20.9(6.0)
25.6(0.8)
21.2(7.9)
26.1(1.1)
Overweight@16
Obese@16
23.9(6.3)
28.9(1.4)
24.4(9.3)
29.4(1.2)
Overweight@45
Obese@45
25.0(49.7)
30.0(25.2)
25.0(32.7)
30.0(23.5)
At 7-16y, 6.3-10.5%
classified as overweight
or obese
HSE06-08: 21% boys,
24% girls (9-11y)
were overweight or
obese
8
Childhood BMI
IOTF cut-offs (%)
Males
Females
Overweight@7
Obese@7
18.2(5.5)
21.1(0.8)
18.0(7.2)
21.0(1.6)
Overweight@11
Obese@11
20.9(6.0)
25.6(0.8)
21.2(7.9)
26.1(1.1)
Overweight@16
Obese@16
23.9(6.3)
28.9(1.4)
24.4(9.3)
29.4(1.2)
Overweight@45
Obese@45
25.0(49.7)
30.0(25.2)
25.0(32.7)
30.0(23.5)
At 7-16y, 6.3-10.5%
classified as overweight
or obese ages.
At 45y, 74.9% men,
56.2% women were
overweight or obese.
9
ROC curves of predicting obesity 45 from child BMI
(Males & females)
ROC analysis (AUC) to assess how well BMI at each age predicts adult obesity
10
ROC curves of predicting obesity 45 from child BMI
(Males & females)
AUC increase with
age
BMI16 was best
predictor for adult
obesity.
Greater increase in
AUC (7-11). Than
(11-16)
ROC analysis (AUC) to assess how well BMI at each age predicts adult obesity
11
Prediction of adult CVD risk factors from child BMI
IOTF cut-offs
Sensitivity Specificity
Obese
7
15.3
94.8
(BMI≥30)
11
19.5
95.8
16
24.8
95.9
Abdominal
7
13.0
95.2
obesity
11
15.7
96.3
16
19.0
96.1
Type 2
7
15.9
92.8
diabetes
11
24.2
92.8
16
31.5
92.0
12
Prediction of adult CVD risk factors from child BMI
IOTF cut-offs
Sensitivity Specificity
Obese
7
15.3
94.8
(BMI≥30)
11
19.5
95.8
16
24.8
95.9
Abdominal
7
13.0
95.2
obesity
11
15.7
96.3
16
19.0
96.1
Type 2
7
15.9
92.8
diabetes
11
24.2
92.8
16
31.5
92.0
Prediction improved with increasing
childhood age
HIGH specificities, but VERY LOW
sensitivities for predicting adult
outcome, only 4.1-5.2% of non-obese
adults were classified as
overweight/obese as a child.
A substantial proportion of obese
adults >75.2% classified as nonoverweight as a child.
Similar for abdominally obese and
high diabetes risk at 45y
13
Prediction of adult CVD risk factors from child BMI
IOTF cut-offs
Sensitivity Specificity
Hypertension
High LDL
Low HDL
High trig
7
8.3
92.7
11
10.6
93.1
16
12.5
92.2
7
7.6
92.5
11
8.2
92.4
16
8.5
91.6
7
10.1
93.0
11
11.8
93.1
16
14.6
92.4
7
7.1
92.4
11
16
8.0
10.0
92.6
92.0
Childhood BMI was a
weak predictor for
adult hypertension,
adverse lipids
14
Internal cut-offs for child BMI
for predicting adult CVD risk factors
Internal cut-offs (%)
BMI @7
15.6-17.6 (15.6-51.9%)
BMI@11
16.2-20.5 (25.0-54.7%)
BMI@16
19.3-24.3 (21.2-55.5%)
15
Internal cut-offs for child BMI
for predicting adult CVD risk factors
Internal cut-offs (%)
BMI @7
15.6-17.6 (15.6-51.9%)
BMI@11
16.2-20.5 (25.0-54.7%)
BMI@16
19.3-24.3 (21.2-55.5%)
Internal cut-offs for 1958 cohort , identifying for each CVD risk factor
to achieve max combination of sensitivity & specificity
Varied according to adult factors, lower than IOTF cut-offs, identified
larger proportions of population in childhood as having increased risk
of adult CVD risks
16
Childhood BMI cut-offs for predicting adult CVD risk factors
IOTF cut-offs (%)
Males
Females
Internal cut-offs (%)
Childhood BMI
Overweight@7
Obese@7
18.2(5.5)
21.1(0.8)
18.0(7.2)
21.0(1.6)
15.6-17.6 (15.6-51.9%)
Overweight@11
Obese@11
20.9(6.0)
25.6(0.8)
21.2(7.9)
26.1(1.1)
16.2-20.5 (25.0-54.7%)
Overweight@16
Obese@16
23.9(6.3)
28.9(1.4)
24.4(9.3)
29.4(1.2)
19.3-24.3 (21.2-55.5%)
17
Prediction of adult CVD risk factors from child BMI
Study specific cut-offs
Sensitivity
Specificity
Obese
7
61.5
61.8
(BMI≥30)
11
66.1
67.7
16
67.3
70.6
Abdominal
7
51.2
65.2
obesity
11
54.7
72.1
16
59.2
68.5
Type 2
7
41.9
76.6
diabetes
11
49.5
73.0
16
60.2
71.6
18
Prediction of adult CVD risk factors from child BMI
Study specific cutoffs
Sensitivity Specificity
Obese
7
61.5
61.8
(BMI≥30)
11
66.1
67.7
16
67.3
70.6
Abdominal
7
51.2
65.2
obesity
11
54.7
72.1
16
59.2
68.5
Type 2
7
41.9
76.6
diabetes
11
49.5
73.0
16
60.2
71.6
Lower specificities than IOTF cut-offs,
improved sensitivities
Obesity: 29.4% non-obese adults
identified as at-risk, 32.7% obese
adults not identified using internal
cut-offs for BMI16.
Diabetes: 23.4-28.4% adults with low
diabetes risk identified as above BMI
cut-offs in childhood, while 39.858.1% adults with high diabetes risks
not identified.
19
Prediction of adult CVD risk factors from child BMI
Study specific cut-offs
Sensitivity Specificity
Hypertension
High LDL
Low HDL
High trig
7
39.0
69.7
11
55.7
56.1
16
44.8
73.9
7
17.3
84.6
11
31.6
76.6
16
25.1
79.8
7
52.1
52.9
11
54.9
54.2
16
39.1
78.0
7
39.6
60.2
11
42.2
67.4
16
24.5
71.2
20
Prediction of adult CVD risk factors from child BMI
Study specific cut-offs
Hypertension
High LDL
Low HDL
High trig
Sensitivity
Specificity
7
39.0
69.7
11
55.7
56.1
16
44.8
73.9
7
17.3
84.6
11
31.6
76.6
16
25.1
79.8
7
52.1
52.9
11
54.9
54.2
16
39.1
78.0
7
39.6
60.2
11
16
42.2
24.5
67.4
71.2
Childhood BMI - weak
predictor for adult
hypertension and adverse
lipids
21
Prediction of adult CVD risk factors from child BMI
IOTF cut-offs
Sensitivity Specificity
Study specific cut-offs
Sensitivity
Specificity
AUC (95% CI)
Obese
7
15.3
94.8
61.5
61.8
0.65(0.64,0.67)
(BMI≥30)
11
19.5
95.8
66.1
67.7
0.72(0.71,0.73)
16
24.8
95.9
67.3
70.6
0.75(0.74,0.77)
Abdominal
7
13.0
95.2
51.2
65.2
0.61(0.59,0.62)
obesity
11
15.7
96.3
54.7
72.1
0.68(0.66,0.69)
16
19.0
96.1
59.2
68.5
0.69(0.68,0.71)
Type 2
7
15.9
92.8
41.9
76.6
0.59(0.54,0.63)
diabetes
11
24.2
92.8
49.5
73.0
0.65(0.60,0.69)
16
31.5
92.0
60.2
71.6
0.68(0.63,0.72)
Li et al. Am J Clin Nutri (2011)
22
Prediction of adult CVD risk factors from child BMI
IOTF cut-offs
Study specific cut-offs
Sensitivity Specificity Sensitivity Specificity
Hypertension
High LDL
Low HDL
High trig
AUC (95% CI)
7
8.3
92.7
39.0
69.7
0.53(0.52,0.55)
11
10.6
93.1
55.7
56.1
0.54(0.52,0.55)
16
12.5
92.2
44.8
73.9
0.54(0.52, .55)
7
7.6
92.5
17.3
84.6
0.50(0.49,0.52)
11
8.2
92.4
31.6
76.6
0.51(0.49,0.53)
16
8.5
91.6
25.1
79.8
0.51(0.49,0.52)
7
10.1
93.0
52.1
52.9
0.54(0.51,0.56)
11
11.8
93.1
54.9
54.2
0.57(0.55,0.59)
16
14.6
92.4
39.1
78.0
0.57(0.55,0.60)
7
7.1
92.4
39.6
60.2
0.52(0.50,0.54)
11
8.0
92.6
42.2
67.4
0.52(0.51,0.54)
16
10.0
92.0
24.5
71.2
0.50(0.49,0.52)
23
Summary
• Prediction of adverse adult CVD risk factors from childhood BMI
was modest:
AUC≤0.75 (adult obesity), ≤0.68 (type 2 diabetes risk),
≤0.57 (hypertension or adverse lipid levels)
• Born in 1958, LOW child overweight/obesity prevalence by IOTF
cut-offs, HIGH in adulthood. Prediction of adult obesity from IOTF
cut-offs was low.
• Internal cut-offs for this population had higher sensitivities than
IOTF cut-offs, but lower specificities.
• Low internal (study specific) cut-offs had limited utility for
identification of children at high risk of adult CVD outcomes
(identified a large proportion of children, differed across adult
outcome).
24
Summary
• Prediction of adult CVD risk factors from childhood
overweight/obesity was modest in a population experiencing
rapid change in obesity prevalence over their lifetime.
• Neither IOTF nor our population-specific cut-offs for childhood
BMI provided adequate diagnostic tools for adult CVD risk
factors in a population experiencing such a rapid change
25
Life-course BMI trajectories
1946 and 1958 British birth cohorts
Males
29
27
46cohort
58cohort
mean BMI
25
1.8kg/m2
23
21
19
17
15
5
10
15
20
*Li et al. Am J of Epidemiol (2009)
25
30
35
40
45
50
55
26
Association between BMI and adult SBP (males)
1946 and 1958 British birth cohorts
27
Life-course BMI trajectories
1946 and 1958 British birth cohorts
Males
29
Correlation (r)
27
58cohort
25
mean BMI
1946
1958
Child slope
0.12
0.20
Adult slope
0.18
0.32
46cohort
1.8kg/m2
23
21
19
17
15
5
10
15
20
25
30
*Li et al. Am J of Epidemiol (2009)
35
40
45
50
55
28
Acknowledgements
Angela Pinot de Moira
Chris power
MRC has funded the 45y survey of the 1958 cohort
Leah Li is funded as a Career Development Award in Biostatistics
29
Prediction of adult CVD risk factors from child BMI
IOTF cut-offs
Sensitivity
Specificity
Study (1958 cohort) specific cut-offs
Cut-offs
%
Sensitivity Specificity
AUC (95% CI)
Obese
7
15.3
94.8
16.0
15.8
43.8
61.5
61.8
0.65(0.64,0.67)
(BMI≥30)
11
19.5
95.8
17.0
17.9
40.5
66.1
67.7
0.72(0.71,0.73)
16
24.8
95.9
20.5
21.3
38.4
67.3
70.6
0.75(0.74,0.77)
Abdominal
7
13.0
95.2
16.0
16.0
40.5
51.2
65.2
0.61(0.59,0.62)
obesity
11
15.7
96.3
17.3
17.9
37.3
54.7
72.1
0.68(0.66,0.69)
16
19.0
96.1
20.5
21.0
41.0
59.2
68.5
0.69(0.68,0.71)
7
15.9
92.8
16.2
17.6
23.9
41.9
76.6
0.59(0.54,0.63)
11
24.2
92.8
17.9
18.4
27.7
49.5
73.0
0.65(0.60,0.69)
16
31.5
92.0
20.4
23.1
29.4
60.2
71.6
0.68(0.63,0.72)
Type 2
diabetes
30
Prediction of adult CVD risk factors from child BMI
IOTF cut-offs
Hypertension
High LDL
Low HDL
High trig
Study (1958 cohort) specific cut-offs
Cut-offs
%
Sensitivity Specificity
AUC (95% CI)
Sensitivity
Specificity
7
8.3
92.7
16.1
16.6
32.5
39.0
69.7
0.53(0.52,0.55)
11
10.6
93.1
16.5
17.7
46.9
55.7
56.1
0.54(0.52,0.55)
16
12.5
92.2
19.8
24.3
30.8
44.8
73.9
0.54(0.52, .55)
7
7.6
92.5
17.1
17.2
15.6
17.3
84.6
0.50(0.49,0.52)
11
8.2
92.4
17.2
20.5
25.0
31.6
76.6
0.51(0.49,0.53)
16
8.5
91.6
21.4
23.0
21.2
25.1
79.8
0.51(0.49,0.52)
7
10.1
93.0
15.7
15.8
47.8
52.1
52.9
0.54(0.51,0.56)
11
11.8
93.1
16.2
18.2
46.6
54.9
54.2
0.57(0.55,0.59)
16
14.6
92.4
23.6
21.2
24.5
39.1
78.0
0.57(0.55,0.60)
7
7.1
92.4
16.3
15.8
39.8
39.6
60.2
0.52(0.50,0.54)
11
8.0
92.6
17.1
18.3
35.4
42.2
67.4
0.52(0.51,0.54)
16
10.0
92.0
22.8
21.0
27.6
24.5
71.2
0.50(0.49,0.52)
31
Prediction of adult CVD risk factors from changes of child BMI
Age
Overwt/obese
Cut-offs for BMI
gain (kg/m2)
% at risk
Sensitivity
Specificity
AUC (95% CI)
7-11y
1.31
1.96
40.2
47.7
73.8
0.61 (0.60, 0.63)
11-16y
3.03
4.41
39.2
45.0
71.8
0.56 (0.54, 0.57)
7-11y
1.26
2.26
37.9
56.3
68.1
0.65 (0.64, 0.67)
11-16y
4.10
4.56
24.5
36.9
79.5
0.59 (0.57, 0.61)
Abdominal
7-11y
1.64
1.51
41.7
56.0
65.9
0.64 (0.62, 0.65)
Obesity
11-16y
4.37
4.56
22.3
29.6
81.7
0.55 (0.54, 0.57)
Type 2 diabetes or
7-11y
1.33
2.60
33.8
59.6
67.0
0.64 (0.60, 0.69)
HbA1c>=7%
11-16y
3.24
5.22
30.2
49.3
70.4
0.60 (0.54, 0.65)
Hypertension
7-11y
0.65
1.69
54.3
63.7
48.8
0.53 (0.51, 0.54)
11-16y
2.65
4.65
41.9
53.5
62.1
0.51 (0.49, 0.53)
High LDL
7-11y
1.74
3.10
24.3
30.9
77.3
0.51 (0.49, 0.53)
(>4.13 mmol/l )
11-16y
2.96
4.42
40.2
44.9
60.9
0.49 (0.46, 0.51)
Low HDL
7-11y
2.66
2.34
22.7
34.2
79.2
0.57 (0.55, 0.59)
(<1.0/1.3 mmol/l)
11-16y
5.60
4.84
14.1
21.3
87.1)
0.52 (0.50, 0.55)
High trig
7-11y
1.57
2.30
32.6
37.5
69.4
0.50 (0.49, 0.52)
(>2.3 mmol/l )
11-16y
1.68
3.77
61.2
73.7
43.8
32
0.49 (0.47, 0.51)
Obese
ROC curves of predicting obesity45 from BMI changes
(7-11, 11-16y, males and females)
33
Download