Kiran Bharthapudi has more than seven years of experience in print, broadcast and new media journalism. He has contributed to several major news agencies, including United Nations radio, BBC online and "Consumer Reports" magazine. His articles specialize in the areas of business, technology and new media. He has a Ph.D. in mass communications.

Anthropometric and Metabolic Phenotype

Table 1. Table 1. Baseline Anthropometric and Metabolic Characteristics of the Study Cohort.

Anthropometric and metabolic data are shown in Table 1. Values for glucose, insulin, insulin resistance, triglycerides, C-reactive protein, interleukin-6, and systolic blood pressure, as well as the prevalence of impaired glucose tolerance, increased significantly with increasing obesity, whereas HDL cholesterol and adiponectin levels decreased with increasing obesity (Table 1). Moderately and severely obese black subjects had lower triglyceride and higher HDL cholesterol levels than similar white and Hispanic subjects. The percentage of subjects with impaired glucose tolerance increased directly with the severity of obesity in subjects in all racial and ethnic groups, a trend that persisted after adjustment for sex, pubertal status, and race or ethnic group. The severity of obesity and the prevalence of the metabolic syndrome were strongly associated after adjustment for race and ethnic group (P=0.009) and for race and ethnic group and sex (P=0.03).

The overall prevalence of the metabolic syndrome was 38.7 percent in moderately obese subjects and 49.7 percent in severely obese subjects; no overweight or nonobese subject met the criteria for the metabolic syndrome. The prevalence of the metabolic syndrome in severely obese black subjects was 39 percent. When we analyzed our data according to the commonly accepted criteria of the National Cholesterol Education Program²⁶ (which are not specific to any race or ethnic group), the prevalence of the metabolic syndrome among severely obese black subjects was only 27 percent.

Factor Analysis

Table 2. Table 2. Pearson Correlation Coefficients of Variables in the Analysis.

Table 3. Table 3. Principal-Factor Analysis and Oblique Analysis of the Whole Cohort of Obese and Overweight Children and Adolescents, According to Risk Factors for the Metabolic Syndrome.

As shown in Table 2 and Table 3, three factors were sufficient to explain correlations between variables — obesity and glucose metabolism, the degree of dyslipidemia, and blood pressure. The three factors explained 58 percent of the total variance in the data (27 percent of the variance was explained by the first factor, an additional 17 percent by the second factor, and another 14 percent by the third factor). The first factor was obesity and glucose metabolism, reflecting strong correlation with the z score for the body-mass index, insulin resistance, and fasting and two-hour plasma glucose levels. The second factor was dyslipidemia, reflecting a positive correlation of insulin resistance with the triglyceride level and a negative correlation of insulin resistance with the HDL cholesterol level. The third factor was blood pressure, reflecting a positive correlation with systolic and diastolic blood pressure. When the C-reactive protein level was incorporated into the analysis (for 293 subjects), it loaded significantly only with the obesity and glucose metabolism factor.

Insulin Resistance

Figure 1. Figure 1. Effect of Insulin Resistance on the Prevalence of the Metabolic Syndrome in White Subjects (Panel A), Hispanic Subjects (Panel B), and Black Subjects (Panel C), According to the Degree of Obesity.

Subjects were grouped into three categories of insulin resistance, with cutoffs at the 33rd and 66th percentiles.

To test the effect of insulin resistance on the prevalence of the metabolic syndrome, we categorized the subjects according to three insulin-resistance categories, using the 33rd and 66th percentiles as cutoffs, and race or ethnic group, with adjustment for the degree of obesity (Figure 1). The prevalence of the metabolic syndrome increased significantly with increasing insulin resistance (P for trend, <0.001) after adjustment for race or ethnic background and obesity group. The prevalence was lower in black subjects than in white subjects (P<0.001) but not than in Hispanic subjects (P=0.20), and it was higher in severely obese subjects than in moderately obese subjects (P=0.03).

Multiple Logistic-Regression Analysis

For the multiple logistic-regression analysis of risk factors associated with the metabolic syndrome in overweight and obese children and adolescents, we incorporated age, sex, z score for BMI, race or ethnic group, and insulin-resistance level into the model. The overall significance of the model was P<0.001. Increasing insulin-resistance levels according to the homeostatic-model assessment were significantly related to the risk of the metabolic syndrome (odds ratio for each increase of one unit, 1.12; 95 percent confidence interval, 1.07 to 1.18). Each half-unit increase in the z score for the body-mass index (one half of 1 SD) was associated with a significant increase in the risk of the metabolic syndrome (odds ratio, 1.55; 95 percent confidence interval, 1.16 to 2.08). White subjects had a higher risk of the metabolic syndrome than black subjects (odds ratio, 2.20; 95 percent confidence interval, 1.35 to 3.59); there was no significant difference in risk between Hispanic subjects and black subjects. Girls were at lower risk for the metabolic syndrome than boys (odds ratio, 0.59; 95 percent confidence interval, 0.39 to 0.89). When the z score for the body-mass index was excluded, the odds ratios associated with each unit of increase in insulin resistance, female sex, and white race as compared with black race did not change significantly.

Proinflammatory and Antiinflammatory Markers and Insulin Resistance

Figure 2. Figure 2. C-Reactive Protein and Adiponectin Levels According to the Degree of Obesity and the Insulin-Resistance Category.

Panel A shows C-reactive protein levels. P<0.001 for the association with the obesity group, and P=0.12 for the association with insulin-resistance category. P=0.64 for the interaction between the obesity group and the insulin-resistance category. Panel B shows adiponectin levels. P=0.04 for the association with the obesity group, and P=0.005 for the association with the insulin-resistance category. P=0.07 for the interaction between the obesity group and the insulin-resistance category. After stratification according to the obesity group, the effect of the insulin-resistance category was evident in moderately obese subjects; those in the highest category of insulin resistance had significantly lower adiponectin levels than those in the middle and low categories (P=0.04 and P=0.002, respectively, with Holm's adjustment).

C-reactive protein levels (Figure 2A ) were significantly related to the degree of obesity (P<0.001) but not to the level of insulin resistance (P=0.12). The levels tended to rise with the number of components of the metabolic syndrome in this cohort, but the trend did not reach statistical significance.

Adiponectin levels decreased with increasing obesity (Table 1). When the subjects were stratified according to obesity group and insulin-resistance category (Figure 2B), the adiponectin levels were significantly associated with the obesity category (P=0.04) and insulin-resistance category (P=0.005); the adiponectin levels were lowest in subjects with the highest level of insulin resistance. There was an interaction between obesity and insulin resistance, but it was not statistically significant (P=0.07). After stratification according to obesity group, the effect of insulin-resistance category was evident in the moderately obese group; subjects in the highest category of insulin resistance had significantly lower adiponectin levels than those in the middle and low categories (P=0.04 and P=0.002, respectively, with Holm's adjustment). In contrast, adiponectin levels in the severely obese group did not vary significantly according to the insulin-resistance category. Adiponectin levels were negatively correlated with C-reactive protein levels (R=–0.18, P=0.005).

Interleukin-6 levels rose significantly with the degree of obesity (Table 1) and were correlated with C-reactive protein levels (R=0.37, P<0.001) but not with the degree of insulin resistance. The relation between interleukin-6 and C-reactive protein levels persisted after adjustment for the z score for the body-mass index (R=0.29, P<0.001).

The Metabolic Syndrome Phenotype after Two Years of Follow-up

Seventy-seven subjects underwent a second comprehensive assessment after a mean (±SD) interval of 21.5±10.5 months. Twenty-four of the 34 subjects in this group who had met the criteria for the metabolic syndrome initially met these criteria at the time of the second evaluation as well. The 10 who did not meet the criteria on follow-up were among the subjects who had a lower BMI initially (z score, 2.42±0.07 vs. 2.62±0.06; P=0.06), had gained less weight (3.74±2.6 kg vs. 11.93±2.9 kg, P=0.05), and tended to have decreased insulin resistance (a reduction from 9.68±1.14 to 7.54±0.82, P=0.07). The syndrome developed over time in 16 of 43 children who did not have the metabolic syndrome at the time of the first evaluation. The baseline z score for the body-mass index in these 16 subjects was similar to that in the 10 subjects who had improvement during follow-up (2.39±0.11 and 2.42±0.07, respectively; P=0.86), yet they gained significantly more weight (16.91±4.4 kg vs. 3.74±2.6 kg, P=0.02). In eight subjects, all of whom had impaired glucose tolerance at the first evaluation, type 2 diabetes developed during follow-up.

Patients

We recruited patients requiring treatment of varicose veins in 11 vascular surgery departments in the United Kingdom between November 2008 and October 2012. All patients were assessed by a vascular surgeon and underwent initial duplex ultrasonographic scanning to assess suitability for treatment and entry into the study. Inclusion criteria were an age of 18 years or older, the presence of unilateral or bilateral primary symptomatic varicose veins (grade C₂ or higher according to the clinical, etiologic, anatomical, and pathophysiological [CEAP] classification system, with C₀ indicating no signs of venous disease, C₁ telangiectases or veins ≤3 mm in diameter, C₂ varicose veins >3 mm in diameter, C₃ the presence of edema, C₄ skin and subcutaneous changes, C₅ healed ulcers, and C₆ active ulceration,²²), and reflux of the great or small saphenous veins of more than 1 second on duplex ultrasonography. Exclusion criteria were current deep-vein thrombosis, acute superficial-vein thrombosis, a diameter of the main truncal saphenous vein of less than 3 mm or more than 15 mm, tortuous veins considered to be unsuitable for laser treatment, and contraindications to the use of foam or to general or regional anesthesia.

Randomization and Study Treatment

A computer-generated randomization system was used and was managed by the Centre for Healthcare Randomised Trials, University of Aberdeen, Aberdeen, United Kingdom. Participants underwent randomization with even assignments to all treatment options available at each investigating center and with stratification according to the number of available options (stratum A, eight hospitals offering all three treatment options; and stratum B, three hospitals offering treatment with only foam or surgery). Treatments were assigned with the use of a minimization algorithm that included center, age (<50 years or ≥50 years), sex, reflux of either the great or the small saphenous veins (or both), and the presence or absence of unilateral or bilateral varicose veins.

Details of treatment methods are described in the published protocol ²¹ (available with the full text of this article at NEJM.org). Briefly, surgery consisted of proximal ligation and stripping (of the great saphenous vein only) and concurrent phlebectomies. Foam was produced with the use of the Tessari technique²¹ at a ratio of 0.5 ml of sodium tetradecyl sulfate to 1.5 ml of air (3% sodium tetradecyl sulfate for saphenous veins and 1% for varicosities, with a maximum of 12 ml of foam per session). The use of sodium tetradecyl sulfate is licensed, but the trial involved its off-license use as a foam rather than as its manufactured liquid form. Laser ablation of truncal saphenous veins performed while the patient was under local anesthesia was followed by foam sclerotherapy to residual varicosities at the 6-week follow-up if required, with the exception that one center performed concurrent phlebectomies.

Outcome Measures

Outcomes were assessed at baseline and at 6 weeks and 6 months after treatment. The primary outcome measures were patient-reported disease-specific quality of life, measured with the use of the Aberdeen Varicose Veins Questionnaire (AVVQ), and patient-reported generic (i.e., general) quality of life, measured at 6 months after treatment with the use of the EuroQoL Group 5-Dimension Self-Report Questionnaire (EQ-5D) and the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36). Another prespecified primary outcome for this trial — 5-year estimated cost-effectiveness, measured as cost per quality-adjusted life-year gained — is not reported here.

The AVVQ is an internationally validated change-responsive tool for the assessment of quality of life in patients with varicose veins.^20,23-26 It consists of 12 questions and a set of mannequin legs on which participants are asked to draw their veins. Scores range from 0 to 100, with higher scores indicating a worse quality of life. The EQ-5D is a standardized index valuation for health status; it includes five dimensions (mobility, self-care, usual activities, pain or discomfort, and anxiety or depression; scores range from −0.594 to 1.000, with higher scores indicating a better quality of life) and a visual analogue scale (EQ-5D VAS; scores range from 0 to 100, with higher scores indicating better health).²⁷ The SF-36 is a validated and reliable assessment of quality of life and is widely used for a variety of clinical conditions.²⁸ The 36 questions assess eight domains and yield two summary scores (the physical component and the mental component), with each summary score ranging from 0 to 100 and higher scores indicating greater well-being. For all these measures of quality of life, minimal clinically important differences after treatment for varicose veins are not known.

Secondary outcomes were as follows: clinical success at 6 weeks and 6 months, as measured by the proportion of patients with residual varicose veins (assessed by the participant and the nurse), venous clinical severity score (a score composed of nine categories relating to symptoms or signs of venous disease and one category relating to the use of compression; scores range from 0 [no venous disease] to 30 [most severe venous disease]), and complications (assessed by the surgeon or nurse); quality of life according to the AVVQ, EQ-5D, and SF-36 at 6 weeks; the EQ-5D VAS and the eight SF-36 domains at 6 weeks and 6 months; and ablation rates of the main trunks of the saphenous vein according to duplex ultrasonography at 6 weeks and 6 months, assessed with the use of a standardized technique²² and reporting tool by independent, accredited vascular technologists (with the exception of one center where scanning was performed by a surgeon who had not performed the treatment). Blinding with respect to the treatment used was not feasible.

Study Oversight

The trial was approved by the research ethics committee and the Medicines and Healthcare Products Regulatory Authority. Written informed consent was obtained from all participants. The trial was overseen by a trial steering committee and an independent data and safety monitoring committee. Data analysis was performed by statisticians at the Centre for Healthcare Randomised Trials. The project management group (the first six and the last four authors) takes responsibility for the accuracy and completeness of the data, analyses, and reporting and for the fidelity of the study to the protocol.

Statistical Analysis

An intention-to-treat analysis was performed for the prespecified comparisons of treatment with foam versus surgery, involving participants from strata A and B, and treatment with laser versus surgery, involving participants from stratum A. In addition, we performed a post hoc analysis of laser therapy versus foam sclerotherapy, involving participants from stratum A. The principal analysis of the trial was performed when all participants had completed the 6-month follow-up. Study analyses were conducted according to a prespecified statistical plan (available with the protocol at NEJM.org) with the use of SAS software, version 9.3 (SAS Institute).

To analyze comparisons between groups, we used general linear models with adjustments for covariates used in the minimization algorithm and, where possible, adjustments for baseline scores (AVVQ, EQ-5D, SF-36, and venous clinical severity scores). No adjustment was prespecified for multiple comparisons. However, for the secondary outcome measures presented here, we consider differences to be significant only for P values of less than 0.005. We analyzed the continuous outcomes with mixed-model repeated-measures analysis, with a compound-symmetry covariance matrix and with center fitted as a random effect. Saphenous-vein ablation rates were analyzed with the use of ordinal logistic regression, and rates of complications were analyzed with the use of binary logistic regression. Sensitivity analyses (see Table S1 in the Supplementary Appendix, available at NEJM.org) were carried out in the case of missing AVVQ responses at 6 months.²⁹

The initial planned sample size was 1015 patients, which, at a two-sided 5% significance level, would provide more than 90% power to detect a difference of 0.25 SD in the AVVQ score for the comparison of foam sclerotherapy with surgery,^30,31 and 80% power to detect a difference of 0.25 SD in the AVVQ score for the comparison of laser with surgery. The data and safety monitoring committee and trial steering committee approved a revised recruitment target of 779 patients on the basis of data showing that the correlation between the AVVQ score at baseline and at 6 months was better than originally assumed. Only the data and safety monitoring committee was aware of the outcome data according to group assignment during the trial.