DNB Thesis

Significance of waist to height ratio as an early predictor in developing metabolic syndrome in children of age group 5-12 years in a tertiary care centre in Trichy: Part II

Dissertation Submitted to the National Board of Examinations, New Delhi

Dr. Nimisha PV*

Department of Pediatrics, KMC Speciality Hospital, Trichy, Tamil Nadu

*Correspondence: [email protected]

Aim

  1. To study the significance of waist to height ratio as an early predictor in children prone to developing metabolic syndrome, in the age group 5-12 years, in a tertiary care centre in Trichy.

Objectives

  1. To identify the cut off value of waist to height ratio, above which intervention is needed.
  2. To compare the association of metabolic syndrome with BMI and Waist to height ratio.

Materials and Methods

Study design

Cross sectional study

Study setting

Kauvery Speciality Hospital, a private tertiary care centre in Trichy

Study duration

A total period of 24 months from January 2019 to December 2020.

Sample size: 170

Sample size for frequency in a population
Population size(N): 750
Proportion (p): 84%
Confidence limits as %(d): 5%
Design effect (DEFF): 1
Sample Size(n)
Confidence Level (%)- 95% Sample Size- 170

Equation

Sample size n = [DEFF*Np(1-p)]/ [(d2/Z21-α/2*(N-1)+p*(1-p)]

(OR)

dnb-formalue

where,

n = Sample Size

Deff = Design Effect

N= Population Size

 = the estimated proportion of the population

=  1-

d = desired absolute precision (or) absolute level of precision

Statistical analysis

Statistical analysis was performed using SPSS, (Version 20.0). The continuous variables are expressed as Mean and Standard deviation. Categorical variables are expressed as frequency and percentage. Independent ‘t’-test is used to find the significance difference between groups. Chi- square test and Fisher’s exact test are used to find out association between the categorical variables.  P<0.05 is considered as statistically significant.

The diagnostic yield of the study was calculated by the following variables.

Reference

TP True Positive
TN True Negative
PPV Positive Predictive Value
NPV Negative Predictive Value

(1). Sensitivity: Sensitivity or true positive rate is a conditional probability of correctly identifying a diseased subject, given by the formula

Sensitivity = TP/(TP + FN)

(2). Specificity: Specificity or true negative rate is a conditional probability of correctly identifying the non- diseased subjects, given by

Specificity = TN/(TN + FP)

(3). Positive predictive value: It is the probability that subjects with a positive screening test truly have the disease, given by

PPV =TP/(TP + FP)

(4). Negative predictive value: It is the probability that subjects with a negative screening test truly don’t have the disease, given by

NPV = TN/(TN + FN)

(5). Accuracy: In simple terms, given a set of points from repeated measurements of the same quantity, the set can be said to be accurate if their average is close to the true value of the quantity being measured.

Accuracy = (TN + TP)/(TP + FP + TN + FN)

Inclusion Criteria

  1. Overweight and obese children in the age group 5 years to 12 years from both OP and IP department according to the current definition.

Exclusion Criteria

  1. Syndromes that may cause obesity.
  2. Children with endocrine abnormalities.

METHODOLOGY

Measure Weight / height / BMI / Waist circumference of children of this age group

 

 

Identifying the children who fall in over weight / obese category

 

 

Applying exclusion criteria

 

 

Proceed with BP measurement and check Fasting blood sugar, Fasting lipid profile.

A detailed physical examination was done to rule out any systemic disorder. Waist circumference measurement was taken from the midway between iliac crest and costal margin using inch tape. Height was taken following the accepted method using stadiometer. Blood samples were collected in fasting state in heparinized glass tubes. These were processed almost immediately, for further evaluation of total cholesterol (TC), serum triglycerides (TG), High density lipoprotein cholesterol (HDL-C) by using specific enzymatic methods. LDL by direct calculation and FBS done by hexokinase method.

A total of 188 children met inclusion criteria in which 12 children’s parents were not willing for blood investigations, 4 were hypothyroid and 2 were syndromic. Excluding them, total of 170 sample were collected. Data was then analysed.

Operational definitions of study variables

Metabolic syndrome is defined as the presence of any 3 or more of the following according to their age related values.

  1. Elevated fasting blood sugar (> 100mg/dL)
  2. High Blood pressure (systolic/diastolic) (> 90th centile for age).
  3. Elevated fasting triglycerides (age specific cut-off values)
  4. Low HDL (< 35mg/dL)
  5. Abdominal adiposity (WC > 90th centile)

All the reference values taken from previous studies. BMI was calculated and plotted in IAP growth chart. Waist circumference to height ratio centile chart for south Indian children was referred from the previous study.19 Age specific reference values for blood lipids were taken from International journal of obesity.32 BP centiles were taken from previous data of BP distribution in Indian children.20 Systolic or diastolic BP more than 90th centile was taken as high BP.

Results

A total of 11,401 children got admitted/ had out-patient visit during the study period, of which 188 met inclusion criteria. 12 parents were not willing for blood investigations and 4 children were excluded by applying exclusion criteria. 170 children were then recruited for the study.

Table 1. Gender

Gender %
Male 92 54.1
Female 78 45.9
Total 170 100.0

As shown in the Table 1, 54 % were boys. The mean age of study participants was 8.78 + 2.2. Participants were either obese or overweight according to the current IAP BMI definition. Blood investigations were conducted in all.

Table 2. Age vs Gender comparison (%)

Age Variables

(years)

Gender Total %
Male % Female %
5 – 7 16 17.4% 35 44.9% 51 30.0%
8 – 10 48 52.2% 25 32.1% 73 42.9%
> 10 28 30.4% 18 23.1% 46 27.1%
Over All 92 78 170
DNB-1

Fig. 1. Age vs Gender

Sample population contained more children in the age group 8-10 years. In girls, 44.9% of the girls in the study population were in the 5-7 years age group. Total of 77.2% of the boys were obese according to IAP definition (>27th adult equivalent) and 22.8% were overweight (>23rd equivalent). Among girls, 66.7% were obese and 33.3% were overweight. Of the 170, 123 children were obese and 47 were overweight as per current IAP definition.

Table 3: BMI vs Gender (%)

BMI Gender Total % p Value
Male % Female %
OBESE 71 77.2% 52 66.7% 123 72.4% χ2= 0.127
OVERWEIGHT 21 22.8% 26 33.3% 47 27.6%
Over All 92 78 170

Table 3 shows that 77.2% of boys and 66.7 % of girls were obese. No significant correlation between gender and BMI was identified.

All the 5 variables namely FBS, HDL, WC, BP, TG were compared in obese and overweight children recruited. Additional variable LDL was also compared in the study group. Later comparison was made to identify association of these parameters to waist to height ratio above and below 95th centile, according to the age. All 170 children had age specific waist circumference more than 90th centile. 150 had a waist circumference to height ratio of more than 95th centile.

Table 4: Descriptive Statistics – Sample Size 170

Parameter Minimum Maximum Mean ± Std.Dev
AGE 5 12 8.78 ± 2.25
HEIGHT 103 164 133.52 ± 14.30
WEIGHT 21.0 82.0 40.23 ± 12.81
WC 60 100 77.04 ± 9.47
BMI 15.61 30.49 21.93 ± 3.02
WC/HT .49 .69 0.57 ± 0.03
SBP 80 130 106.46 ± 11.82
DBP 50 88 70.07 ± 9.07
FBS 65 134 92.19 ± 15.10
HDL 20 65 39.87 ± 8.51
LDL 27 156 90.04 ± 18.42
TRIGLYCERIDES 45 457 145.53 ± 47.76

Table 4 shows the general characteristics of the study population of 170.

COMPARISON BETWEEN BMI AND 5 VARIABLES STUDIED

Table 5a. BP vs BMI

Variables BMI Total % p Value
obese % overweight %
BP High 44 35.8% 13 27.7% 57 33.5% χ2= 0.316
Normal 79 64.2% 34 66.5% 113 66.5%
Over All 123 47 170

Table 5a shows 35.8 % children with high BP for age were obese as per BMI classification, 27.7 % were overweight. The association was not statistically significant.

Table 5b. TG vs BMI

Variables BMI Total % p Value
obese % overweight %
TG High 109 88.6% 38 80.9% 147 86.5% χ2= 0.185
Normal 14 11.4% 9 19.1% 23 13.5%
Over All 123 47 170

147 (86.5%) children had abnormal triglycerides, of which 38 were overweight. No significant association was found as seen in table 5 b.

Table 5c. HDL vs BMI

Variables BMI Total % p Value
obese % overweight %
HDL Normal 56 45.5% 27 57.4% 83 48.8% χ2= 0.164
Low 67 54.5% 20 42.6% 87 51.2%
Over All 123 47 170

Table 5c. shows that half the children from the study population had low HDL. 56 children were obese and rest were in the overweight category.

Table 5d. LDL vs BMI

Variables BMI Total % p Value
obese % overweight %
LDL Normal 87 70.7% 37 78.7% 124 72.9% χ2= 0.294
High 36 29.3% 10 21.3% 46 27.1%
Over All 123 47 170

Table 5d. shows no significant correlation of LDL cholesterol value and BMI. 72.9% children had normal LDL level.

Table 5e. FBS vs BMI

Variables BMI Total % p Value
Obese % Overweight %
FBS Normal 83 67.5% 37 78.7% 120 70.6% χ2= 0.150
High 40 32.5% 10 21.3% 50 29.4%
Over All 123 47 170

As seen in table 5.e, 120 children had abnormal FBS value which constitutes around 70.6% of study population. Among them 83 were obese and 37 were overweight.

All the biochemical parameters and BP were then compared to the WHtR Centile. Reference values for WHtR was taken from previous studies done in Indian children.19 Cut off value of 95th centile was taken and its significance was compared with metabolic parameters. Of the 170 children, 150 had waist to height ratio more than 95th centile. The results are shown in the following tables.

Table 6a. BP vs WHtR Centile

BP WHtR Centile Total % p Value
< 95th % > 95th %
Yes 7 35.0% 50 33.3% 57 33.5% χ2= 0.882
No 13 65.0% 100 66.7% 113 66.5%
Over All 20 150 170

57 children had high BP for age in the study group. As seen in Table 6a, WHtR was more than 95th centile in 50 of them. However, there was no statistical significance.

Table 6b. TG vs WHtR Centile

TG WHtR Centile Total %  p Value
< 95th % > 95th %
High 15 75.0% 132 88.0% 147 86.5% χ2= 0.241
Normal 5 25.0% 18 12.0% 23 13.5%
Over All 20 150 170

As seen in Table 6b, 88% children with WHtR > 95th centile had abnormal triglycerides in the serum, but is not significant as the p value is more than 0.05.

Table 6c. HDL vs WHtR Centile

HDL WHtR Centile Total %  p Value
< 95th % > 95th %
Normal 12 60.0% 71 47.3% 83 48.8% χ2= 0.287
Low 8 40.0% 79 52.7% 87 51.2%
Over All 20 150 170

As shown in the table 6.c, 79 out of 87 children with high WHtR centile had abnormally low HDL cholesterol. 8 children with WHtR <95th centile also had low HDL. There was no statistical significance.

Table 6d. LDL vs WHtR Centile

LDL WHtR Centile Total % p Value
< 95th % > 95th %
Normal 14 70.0% 110 73.3% 124 72.9% χ2= 0.127
High 6 30.0% 40 26.7% 46 27.1%
Over All 20 150 170

Table 6d shows that 110 children with normal LDL cholesterol level had high waist to height ratio. But the association was not significant.

Table 6e. FBS vs WHtR Centile

FBS WHtR Centile Total % p Value
< 95th % > 95th %
Normal 14 70.0% 106 70.7% 120 70.6% χ2=0.951
High 6 30.0% 44 29.3% 50 29.4%
Over All 20 150 170

Out of the 50 children with high FBS, 95% children had a high waist to height ratio. 106 children with normal FBS also had high waist to height ratio. Table 6e shows no significant association.

In summary, among the study group 29.4% children had high FBS, 86% children had high triglycerides, 51.2% had low HDL, 27.1% had high LDL and 33.5% children had high BP.

Table 7. MetS in comparison with BMI

BMI MetS Total % p Value
Yes % No %
Obese 56 80.0% 67 67.0% 123 72.4% χ2= 0.062
Overweight 14 20.0% 33 33.0% 47 27.6%
Over All 70 100 170

By applying the described criteria for diagnosing MetS, out of 170 children studied ,70 children satisfied the criteria for metabolic syndrome. Only 56 children with metabolic syndrome had BMI more than 27th adult equivalent. 14 children had BMI in the overweight range.

Table 7 shows BMI classification identified 80% children with metabolic syndrome. 20% children with BMI in less than obese centile had features of MetS.This association was not statistically significant.

DNB-2

Fig. 2. Association of BMI and MetS

Fig. 2 shows that, 45% of obese and 30% of overweight children had MetS.

Table 8. MetS in comparison with WHtR Centile

MetS WHtR Centile Total % p Value
< 95th % > 95th %
Yes 1 5.0% 69 46.0% 70 41.2% χ2=0.000
No 19 95.0% 81 54.0% 100 58.8%
Over All 20 150 170

In the 70 children with metabolic syndrome 98.5% had waist to height ratio more than 95th centile and there was significant association between these two parameters, p value 0.00, as seen in table 8. Only 1 child with Waist to height ratio less than 95th centile had MetS.

DNB-3

Fig. 3. Association of WHtR Centile and MetS

Fig 3 shows that MetS was found in 46% of those children who had a high waist to height ratio, as against 5% in those whose waist to height ratio was less than 95th centile.

DNB-4

Fig. 4. WHtR Centile in children with MetS

Fig 4 shows that, out of 70 children with MetS, 69 had waist to height ratio more than 95th centile for their age, which is statistically significant.

With the above data, statistical analysis was done. Sensitivity, specificity, positive predictive value, negative predictive value of both BMI chart and WHtR chart were calculated. To recap the following statistical method was used.

Statistical Methods

The diagnostic yield of the study was calculated by the following variables.

Reference

TP True Positive
TN True Negative
PPV Positive Predictive Value
NPV Negative Predictive Value

(1). Sensitivity: Sensitivity or true positive rate is a conditional probability of correctly identifying a diseased subject, given by the formula

Sensitivity = TP/(TP + FN)

(2). Specificity: Specificity or true negative rate is a conditional probability of correctly identifying the non- diseased subjects, given by

Specificity = TN/(TN + FP)

(3). Positive predictive value: It is the probability that subjects with a positive screening test truly have the disease, given by

PPV =TP/(TP + FP)

(4). Negative predictive value: It is the probability that subjects with a negative screening test truly don’t have the disease, given by

NPV = TN/(TN + FN)

(5). Accuracy: In simple terms, given a set of points from repeated measurements of the same quantity, the set can be said to be accurate if their average is close to the true value of the quantity being measured.

Accuracy = (TN + TP)/(TP + FP + TN + FN)

Table 9. BMI vs MetS

BMI vs MetS
Parameter Estimate %
Sensitivity 80.0%
Specificity 33.0%
Positive predictive value 45.5%
Negative predictive value 70.2%
Diagnostic accuracy 52%

Table 10. WHtR Centile vs MetS

METS vs WHtR Centile
Parameter Estimate %
Sensitivity 46.0%
Specificity 95.0%
Positive predictive value 98.6%
Negative predictive value 19.0%
Diagnostic Accuracy 52%

From Tables 9 and 10 we came to the conclusion that waist to height ratio more than 95th centile had 95% specificity and 98.6% positive predictive value. Diagnostic accuracy was comparable to that of BMI.

Table 11. Descriptive statistics of 70 children with METS

Parameter Minimum Maximum Mean ± Std.Dev
AGE 5 12 9.9 ± 1.95
HEIGHT 113 164 140.14 ± 13.04
WEIGHT 23 82 46.63 ± 12.68
WC 60 100 82.6 ± 8.40
BMI 17 30 23.21 ± 2.81
WC/HT 0.53 0.68 0.59 ± 0.03
SBP 90 130 113.11 ± 11.30
DBP 60 88 75.28 ± 8.39
FBS 65 134 100.52 ± 15.38
HDL 20 56 36.42 ± 7.25
LDL 65 156 99.28 ± 19.96
TRIGLYCERIDES 67 256 150.58 ± 39.74

Table 11 shows the descriptive variables in children with metabolic syndrome.

Following tables are used to highlight the association between 5 variables studied with BMI and WHtR among the 70 detected to have MetS.

Table 12aAssociation of FBS with BMI in those with MetS

FBS Obese BMI Total % p Value
Yes % No %
Normal 22 39.30% 9 64.30% 31 44.30% χ2= 0.092
High 34 60.70% 5 35.70% 39 55.70%
Over All 56 14 70

Table 12a shows total of 34 children out of 56 with obese BMI and MetS had high FBS. Five children with overweight BMI also had high FBS showing no statistical correlation.

Table 12b. Association of FBS with WHtR Centile in those with MetS

FBS WHtR Centile Total % p Value
< 95th % > 95th %
Normal 1 100.0% 30 43.5% 31 44.3% χ2=0.259
High 0 0.0% 39 56.5% 39 55.7%
Over All 1 69 70

As seen in table 12b, fasting blood sugar was high in 56% and normal in 44% among those with a high WHtR and MetS.

Table 13a. Association of BP with BMI in those with MetS

BP Obese BMI Total % p Value
Yes % No %
High 37 66.10% 8 57.10% 45 64.30% χ2= 0.533
Normal 19 33.90% 6 42.90% 25 35.70%
Over All 56 14 70

Table 13b. Association of BP with WHtR in those with MetS

BP WHtR Centile Total % p Value
 < 95th % > 95th %
Yes 1 100.0% 44 63.8% 45 64.3% χ2= 0.453
No 0 0.0% 25 36.2% 25 35.7%
Over All 1 69 70

Tables 13 a and b shows no significant association between metabolic syndrome and BP as a single parameter.

Table 14a. Association of HDL with BMI in those with MetS

HDL Obese BMI Total % p Value
Yes % No %
Normal 14 25.00% 4 28.60% 18 25.70% χ2= 0.784
Low 42 75.00% 10 71.40% 52 74.30%
Over All 56 14 70

Table 14b. Association of HDL with WHtR in those with MetS

HDL WHtR Centile Total % p Value
< 95th % > 95th %
Normal 1 100.0% 17 24.6% 18 25.7% χ2= 0.087
Low 0 0.0% 52 75.4% 52 74.3%
Over All 1 69 70

p value for the association of HDL as seen in the Table 14a and b were 0.7, 0.08 respectively, thus showing no statistical significance between serum HDL and MetS.

Table 15a. Association of LDL with BMI and WHtR in those with MetS

LDL Obese BMI Total %   p Value
Yes % No %
Normal 26 46.40% 7 50.00% 33 47.10% χ2= 0.811
High 30 53.60% 7 50.00% 37 52.90%
Over All 56 14 70

Table 15b. Association of LDL with WHtR in those with MetS

LDL WHtR Centile Total % p Value
< 95th % > 95th %
Normal 0 0.0% 33 47.8% 33 47.1% χ2= 0.341
High 1 100.0% 36 52.2% 37 52.9%
Over All 1 69 70

Table 15 a and b show no significant association seen in children with MetS and their LDL values against BMI centile and WHtR.

Table 16a. Association of Triglycerides with BMI in those with MetS

TG Obese BMI Total % p Value
Yes % No %
High 53 94.60% 14 100.00% 67 95.70% χ2= 0.376
Normal 3 5.40% 0 0.00% 3 4.30%
Over All 56 14 70

Above Table 16a shows no association with BMI centiles to children with MetS.

Table 16b. Association of Triglycerides with WHtR in those with MetS

TG WHtR Centile Total % p Value
< 95th % > 95th %
High 0 0.0% 67 97.1% 67 95.7% χ2= 0.000
Normal 1 100.0% 2 2.9% 3 4.3%
Over All 1 69 70

On comparing WHtR and TG value as shown in Table 16b, p value was found to be 0.00 (<0.05), which denotes significant association.

Table 17 Association of Gender and MetS

Gender MetS Total % p Value
Yes % No %
Male 42 60.0% 50 50.0% 92 54.1% χ2= 0.198
Female 28 40.0% 50 50.0% 78 45.9%
Over All 70 100 170

Table 17 shows no significant association of MetS and gender.

DNB-5

Fig. 5. Association of Gender and MetS.

Table 17 and Fig. 5 show no significant association between gender and MetS. As shown in Fig. 5, in the 70 children with MetS, 60% are boys, 40% are girls.

With the above results ROC curve was plotted.

Table 18. Detailed report of sensitivity and specificity

Cut off point Sensitivity (%) Specificity (%) Classified (%) LR+ LR-
(≥ .49 ) 100.00 0.00 41.18 1.0000
(≥ .5 ) 100.00 2.00 42.35 1.0204 0.0000
(≥ .51 ) 100.00 4.00 43.53 1.0417 0.0000
(≥ .52 ) 100.00 6.00 44.71 1.0638 0.0000
(≥ .53 ) 100.00 11.00 47.65 1.1236 0.0000
(≥ .54 ) 97.14 17.00 50.00 1.1704 0.1681
(≥ .55 ) 92.86 27.00 54.12 1.2720 0.2646
(≥ .56 ) 85.71 40.00 58.82 1.4286 0.3571
(≥ .57 ) 74.29 50.00 60.00 1.4857 0.5143
(≥ .58 ) 62.86 62.00 62.35 1.6541 0.5991
(≥ .59 ) 50.00 70.00 61.76 1.6667 0.7143
(≥ .6 ) 41.43 78.00 62.94 1.8831 0.7509
(≥ .61 ) 28.57 84.00 61.18 1.7857 0.8503
(≥ .62 ) 22.86 91.00 62.94 2.5397 0.8477
(≥ .63 ) 14.29 93.00 60.59 2.0408 0.9217
(≥ .64 ) 10.00 96.00 60.59 2.5000 0.9375
(≥ .65 ) 5.71 98.00 60.00 2.8571 0.9621
(≥ .66 ) 1.43 98.00 58.24 0.7143 1.0058
(≥ .68 ) 1.43 99.00 58.82 1.4286 0.9957
(≥ .69 ) 0.00 99.00 58.24 0.0000 1.0101
( >  .69 ) 0.00 100.00 58.82 1.0000

Fig. 6. ROC CURVE

DNB-6
Dr.-Nimisha-PV

Dr. Nimisha PV

DNB PG

Kauvery Hospital