Waist Circumference and Body Mass Index Variability and Incident Diabetic Microvascular Complications: A Post Hoc Analysis of ACCORD Trial

Study participants and data collection

ACCORD was a randomized clinical trial of 10,251 participants with T2DM who were followed to assess the health effects of intensive glycemic, lipid, and BP control as against standard control [21,22]. The design and main results of the ACCORD study have been published previously [22]. Briefly, ACCORD had three study arms: (1) glycaemia trial (glycosylated hemoglobin [HbA1c] <6.0% vs. 7.0%< HbA1c <7.9%); (2) lipid trial (fenofibrate vs. placebo); and (3) BP trial (systolic blood pressure [SBP] <120 mm Hg vs. SBP <140 mm Hg), with all participants involved in the glycaemia trial [23]. Recruitment of participants into the study began in January 2001 through to October 2005 from 77 clinical sites across Northern America (i.e., USA and Canada). Participants were followed up until June 2009 when the ACCORD study was ended [24]. Ethical approval for the ACCORD study was granted by Institutional Review Boards (IRB) of each clinical site and written informed consent was obtained from all recruited participants (trial registration: ClinicalTrials.gov., no. NCT00000620) [25]. It is noteworthy to mention that the authors of this current study were not involved in the ACCORD trial project but only conducted a post hoc analysis of the ACCORD dataset.

For this study, participants were excluded if they had less than two recorded anthropometric measurements (n=460). Then, for the analysis of each microvascular outcome, individuals with the very prevalent microvascular outcome at baseline were excluded. For neuropathy outcome, 2,587 participants with prevalent neuropathy were excluded; for nephropathy outcome, 1,037 participants with prevalent macroalbuminuria and 34 persons with serum creatinine >3.3 mg/dL were excluded; and for retinopathy, 1,003 patients with retinopathy, 2,184 patients with eye disease and 387 patients who have had eye surgeries were excluded from the analysis. The analytic samples for neuropathy, nephropathy, and retinopathy outcomes were 7,204, 8,720, and 6,217, respectively.

Measures of variabilities in WC and BMI

Exposure of interest was anthropometric measurements obtained at baseline and then annually, including WC and BMI. WC was measured at the smallest point between the 10th rib and the iliac crest over bare skin using steel or non-stretching plastic tape. Height was measured using either a steel tape measure marked in centimeters which was hung vertically on the wall with a headboard at a right angle or commercial stadiometers. Participants were asked to position themselves against the wall having their heels, buttocks, and/or shoulder blades touching the board (tape), and their eyes looking straight, feet together with ankles touching or as close as possible. Bare feet were preferred while light or nylon socks were also allowed during the measurement. Finally, the weight of participants was measured using high-quality scales in clinical practice; during the weighing procedure, participants were asked to wear as little clothing as possible, removing shoes, and other external accessories. All values were recorded to the nearest decimal place. BMI was calculated as weight (kg) divided by squared height (m).

Variabilities in WC and BMI were defined as intra-individual variabilities between measurements using the following four variability measures: (1) average successive variability (ASV), which was defined as the average absolute difference between successive visits; (2) standard deviation (SD); (3) coefficient of variation (CV); and (4) variability independent of the mean (VIM). VIM was calculated as 100×SD/mean^β, where β is the regression coefficient, based on the natural logarithm of the SD on the natural logarithm of the mean. The uncorrected VIM was corrected using this formula: (VIM uncorrected×mean of CV)/mean of VIM uncorrected [26]. Because the results from all variability measures were similar (with exception of the results between variability measures and the relationship with retinopathy in our primary analysis), the corrected variability independent of the mean (cVIM) was used as the primary variability measure in this study whiles results from the other variability measures are reported for comparative purposes [16,27,28].

Study outcomes and follow-up

The endpoints considered in our analysis were the predefined ACCORD microvascular outcomes defined as neuropathy: a composite of (1) new score of >2.0 on the Michigan Neuropathy Screening Instrument (MNSI), (2) new loss of vibratory sensation (tested using 128 Hz tuning fork), (3) new loss of ankle jerk during Jendrassik maneuver, or (4) new loss of light touch (as measured by 10 g force monofilament test); nephropathy: a composite of (1) development of macroalbuminuria, (2) development of renal failure, or (3) doubling of baseline serum creatinine or more than 20 mL/min/1.73 m2 decrease in estimated glomerular filtration rate (eGFR); and retinopathy: a composite of (1) retinal photocoagulation or vitrectomy to treat retinopathy, (2) eye surgery for cataract extraction, (3) three‐line change in visual acuity (as measured using Log MAR visual acuity chart), or (4) severe vision loss (as measured by Snellen fraction <20/200). Patients who experienced any one of the predefined microvascular events that comprised the composite neuropathy, nephropathy, and retinopathy outcome were considered to have experienced that specific disease. Supplementary Table 1 shows the three predefined ACCORD microvascular endpoints and their frequency of assessment.

Statistical analysis

Participants’ characteristics were described using mean±SD or median (interquartile range) for continuous variables, depending on whether the data distribution was normal (assessed by the Shapiro-Wilk test); categorical variables were described by frequency and percentage. The association between WC and BMI variability (as measured by corrected variability independent of mean) and the risk of outcomes was evaluated with the use of WC and BMI variability as a categorical variable.

Cox proportional hazards regression model was used to estimate the hazard ratio (HR) and 95% confidence interval (CI), using WC and BMI variability as a time-dependent covariate. The crude models and fully adjusted models were used. With the exception of baseline age, sex, and race, all baseline characteristics with a P<0.10 in the univariate analysis were included in the multivariable models for respective microvascular outcomes. In Kaplan-Meier analysis for the outcomes, the highest quartile (Q4) category was compared with the lower three quartiles (Q1–Q3) as a reference category. The differences between categories were assessed using the log-rank test. Additionally, to account for WC and BMI variability as a categorical variable, patients were divided into quartiles, using the lowest quartiles as the reference.

The multivariable models comprised baseline age (continuous), sex (male or female), race (white or non-white), diabetes duration (continuous), BP versus lipid treatment arm (standard BP, intensive BP, fibrate, and placebo), SBP (continuous), diastolic blood pressure (DBP; continuous), low density lipoprotein (LDL; continuous), high density lipoprotein (HDL; continuous), HbA1c (continuous), baseline BMI (continuous) (or baseline WC) in addition to allocation to glycemia treatment arm (intensive and standard), insulin use (yes or no), alcohol consumption (yes or no) for neuropathy (model 2¹); antihypertensive medication (yes or no), cardiovascular disease (CVD) history (yes or no), eGFR (continuous), and UACR (continuous) for nephropathy (model 2²); allocation to glycemia treatment arm (intensive and standard) and insulin use (yes or no) for retinopathy (model 2³).

Further exploration was carried out to examine possible interactions between each SD increase in WC and BMI variability (cVIM), and the following baseline characteristics of the participants with respect to each microvascular outcome: age (≤65 or >65 years), sex, diabetes duration (≤9 or >9 years), glycemia arm (intensive or standard), BP versus lipid treatment arm (standard BP, intensive BP, lipid placebo, and lipid fenofibrate), and SBP (≤139 or >139 mm Hg). The statistical significance of the interactions was assessed by adding a multiplicative term to the Cox models. Sensitivity analyses were also performed by excluding participants who had the respective events within the first year of follow-up and participants with any serious adverse event (attributed to study medication and not related to hypoglycemia) that occurred in the first 18 months of follow-up.

All statistical analyses were 2-sided, and we considered a P value of <0.05 to be statistically significant. All analyses were performed using Stata version 14.2 (StataCorp., College Station, TX, USA).

Ethics statement

The studies involving human participants were reviewed and approved by ACCORD Protocol Review Committee appointed by National Heart, Lung, and Blood Institute (NHLBI) & local IRBs of participating clinical sites. This study has been reviewed and found the study exempted from IRB approval (No. 2022-0407). The patients/participants provided their written informed consent to participate in this study.

Data availability statement

Publicly available dataset was analysed in this study. This data can be found at: https://biolincc.nhlbi.nih.gov/studies/accord/ (Repository: Biologic Specimen and Data Repository).

WC and BMI variability and risk of the outcome

The Kaplan-Meier curves for the relationship between obesity variability measures and neuropathic, nephropathic, and retinopathic events are shown in Supplementary Figs. 1-3. The highest quartile in cVIM of WC showed a higher risk for neuropathy and nephropathy compared to lower quartiles (Supplementary Figs. 1 and 2), while the highest quartile in cVIM of BMI showed a higher risk for nephropathy, compared to lower quartiles (Supplementary Fig. 2).

Fig. 1 shows that the rates of any neuropathic, nephropathic, and retinopathic event increased in increasing quartiles of mean WC and BMI variability as measured by cVIM. Compared with the lowest quartile, the hazard ratio (HR) (95% confidence interval [CI]) for the highest quartiles of WC and BMI variability measured by cVIM in the fully-adjusted model were 1.21 (95% CI, 1.05 to 1.40) and 1.16 (95% CI, 1.00 to 1.33) for any neuropathic event, respectively (Table 2). A similar trend of association of observed when ASV, CV, and SD were used to evaluate the relationship of WC and BMI variability with neuropathy risk (Table 2). For the relationship of WC and BMI variability with the nephropathic outcome, higher quartiles of BMI variability were also associated with increased risk of nephropathic events but not WC variability. The fully-adjusted HR (95% CI) for the 4th quartile of BMI variability measured cVIM was 1.31 (95% CI, 1.18 to 1.46), compared with the 1st quartile (Table 3). A similar trend of association was observed when other variability measures (ASV, CV, and SD) were used in the analysis (Table 3). Meanwhile, for the retinopathic outcome, the results were null for both WC and BMI variability using cVIM and ASV as the measures of variability (Table 4); but, when CV and SD were used as the measures of variability, a significant association was observed between BMI variability and retinopathic risk (Table 4).

Fig. 1.

Rates of any microvascular event in quartiles of (A, B, C) waist circumference (WC) and (D, E, F) body mass index (BMI) variability as measured by corrected variability independent of mean (cVIM).

Table 2.

Quartiles of WC and BMI variability measures and risk of neuropathy

Table 3.

Quartiles of WC and BMI variability measures and risk of nephropathy

Table 4.

Quartiles of WC and BMI variability measures and risk of retinopathy

Subgroup and sensitivity analysis

Supplementary Table 2 shows the HR (95% CI) for the subgroup analysis of the links between each SD increase in WC and BMI variability and various outcomes. For nephropathic outcome, we found sex to significantly modify the relation between each 1SD increase in the WC and BMI and the risk of incident nephropathy (both P_interaction=0.03). However, the link between WC variability and diabetic nephropathy risk was similar for both men and women (Supplementary Table 2). No significant modifying effects were observed for the other factors.

Sensitivity analyses were conducted to check the robustness of our results. Namely, results remained unchanged after excluding participants who had the respective microvascular events within the first year of follow-up and excluding participants with any serious adverse event (attributed to study medication and not related to hypoglycemia) that occurred in the first 18 months of follow-up for neuropathic and nephropathic outcomes (Supplementary Tables 3 and 4). Meanwhile, for the retinopathic outcome, all measures of variability to evaluate the relationship with both WC and BMI became null in the sensitivity analysis (Supplementary Table 5).