Validity of self-reported weights following bariatric surgery

Introduction

Obtaining standardized weights in long-term studies can be difficult. Self-reported weights are more easily obtained, but less accurate than those from a calibrated scale and may be inaccurately reported. Previous studies have reported that women tend to under-report their weight more than men with the degree of misreporting related to body mass index (BMI) whereby overweight individuals tend to under-report while under-weight individuals tend to over-report.^1–4 However, in a study of female gastric bypass candidates, self-reported pre-surgical weights averaged 0.3 kg more than measured weights and misreporting was not significantly related to BMI.⁵

This study investigated whether self-reported weights following bariatric surgery differed from weights obtained by study personnel using a standard scale.

Methods

The Longitudinal Assessment of Bariatric Surgery-2 is a ten center observational cohort study of 2458 adults undergoing an initial Roux-en-Y gastric bypass (RYGB), laparoscopic adjustable gastric band (LAGB), or other bariatric procedure.⁶ This report utilizes data collected between April 2010 and November 2012 at annual assessments 1 to 5 years after RYGB or LAGB. Each center had institutional review board approval and all participants provided written informed consent.

Annually, participants were asked to report on mailed questionnaires two postoperative weights and the dates weights were measured: (1) weight from last medical office or weight loss program visit (self-reported medical weight) and (2) last self-weighing (self-reported personal weight). Self-reported weights could be from any scale with or without shoes or bulky clothing. Using standardized data collection protocols, study personnel measured height before surgery using a stadiometer and measured weights before surgery and annually afterwards on a standard scale (Tanita® Body Composition Analyzer, model TBF-310) without shoes and bulky clothing (measured weight).

Participants with postoperative measured weights and self-reported weights from no more than 30 days before the measured weight were included. If both self-reported weights met this criterion then both were included. If participants had self-reported and measured weights meeting this criterion at multiple time points, weights from one randomly selected time point were used.

Statistical significance of weight differences was assessed using t-tests for each type of self-reported weight and normal mixed models for all self-reported weights combined. Analyses were conducted using SAS (version 9.2). Two-sided P-values less than 0.05 are considered statistically significant.

Results

Of the 992 participants with weights meeting the inclusion criteria, 4 were excluded because of suspected recording error. The 988 participants included 164 with a self-reported medical weight, 580 with a self-reported personal weight and 244 with both self-reported weights. Included participants are characterized in the Table.

Table.

Participant characteristics at time of measured weight and validity of self-reported weights following bariatric surgery.

	Women	Men	P-valuea
Participant Characteristics
Number of participants	808	180
Age (years), mean (SD) [range]	49 (11) [21 to 79]	55 (11) [28 to 80]	<0.001
BMI (kg/m2), mean (SD) [range]	33.0 (7.0) [ 19.0 to 63.2]	34.6 (5.8) [23.2 to 52.6]	0.002
Weight change (%), mean (SD) [range]	−28.3 (12.3) [−58.3 to 13.0]	−24.3 (11.2) [−49.5 to 11.3]	<0.001
Bariatric Procedure, n (%)
Roux-en-Y gastric bypass	603 (74.6)	131 (72.8)	0.61
Laparoscopic adjustable gastric band	205 (25.4)	49 (27.2)
All Self-Reported Weights
Number of participants	808	180
Measured weight (kg), mean (SD) [range]	89.6 (20.2) [45.0 to 194.1]	111.6 (20.7) [74.1 to 180.0]
Self-reported weight (kg), mean (SD) [range]	88.9 (20.3) [44.5 to 195.9]	110.6 (21.0) [72.7 to 184.1]
Self-reported weight minus measured weight (kg)
Mean weight difference (SD) [range]	−0.7 (1.9) [−10.9 to 6.8]	−1.0 (2.7) [−10.9 to 11.8]	0.09
95% Confidence interval	(−0.81, −0.56)	(−1.34, −0.59)
P-valueb	<0.001	<0.001
Percent differencec (%)
Mean percent difference (SD) [range]	−0.8 (2.1) [−11.8 to 10.4]	−0.9 (2.3) [−8.9 to 8.7]	0.46
95% Confidence interval	(−0.93, −0.65)	(−1.20, −0.59)
P-valueb	<0.001	<0.001
Self-Reported Medical Weights
Number of participants	329	79
Measured weightd (kg), mean (SD) [range]	90.1 (21.9) [45.0 to 194.1]	115.1 (22.6) [77.3 to 180.0]
Self-reported medical weight (kg), mean (SD) [range]	90.0 (22.0) [44.5 to 195.9]	115.1 (23.1) [76.4 to 184.1]
Self-reported medical weight minus measured weight (kg)
Mean weight difference (SD) [range]	−0.1 (1.8) [−6.8 to 6.8]	0.1 (2.8) [−10.5 to 11.8]	0.67
95% Confidence interval	(−0.26, 0.13)	(−0.56, 0.72)
P-valueb	0.53	0.80
Percent differencec (%)
Mean percent difference (SD) [range]	−0.1 (2.0) [−6.4 to 10.4]	0.0 (2.2) [−6.6 to 8.7]	0.70
95% Confidence interval	(−0.30, 0.13)	(−0.49, 0.52)
P-valueb	0.42	0.95
Self-Reported Personal Weights
Number of participants	673	151
Measured weightd (kg), mean (SD) [range]	89.3 (19.5) [49.1 to 182.7]	110.6 (19.7) [74.1 to 170.0]
Self-reported personal weight (kg), mean (SD) [range]	88.4 (19.5) [48.2 to 181.8]	109.1 (19.8) [72.7 to 172.7]
Self-reported personal weight minus measured weight (kg)
Mean weight difference (SD) [range]	−0.9 (1.8) [−10.9 to 6.8]	−1.5 (2.4) [−10.9 to 3.6]	0.01
95% Confidence interval	(−1.07, −0.79)	(−1.87, −1.08)
P-valueb	<0.001	<0.001
Percent differencec
Mean percent difference (SD) [range]	−1.1 (2.1) [−11.8 to 7.3]	−1.4 (2.2) [−8.9 to 3.3]	0.11
95% Confidence interval	(−1.22, −0.90)	(−1.72, −1.03)
P-valueb	<0.001	<0.001

Discrepancies between the mean weights and the mean differences are due to rounding. SD is the standard deviation.

^aP-values comparing women and men.

^bP-values comparing the mean or percent difference to 0.

^cPercent difference=[(self-reported weight – measured weight) / measured weight] × 100.

^dMeasured weights of those participants with the corresponding self-reported weight.

Across the two types of self-reported weight, women and men under-reported their weight by an average of 1 kg or less (range: 10.9 kg under-report to 11.8 kg over-report) and the degree of under-reporting was not significantly different between women and men (Table). Self-reported medical weights were significantly closer to measured weights than were self-reported personal weights for both women and men. Weight differences did not vary systematically by measured BMI or percent weight change from baseline (Figure).

Figure.

Difference between self-reported and measured weights versus measured BMI and measured percent weight change, by sex. About 95% of the points are expected to fall within two standard deviations (SD) of the mean.

Discussion

Differences between self-reported and measured weights may be due to differences in clothing, inaccurate personal scales, time between measurements, or intentional misrepresentation.

In a general population survey, obese men and women under-reported their weight, on average, by 1.32 kg and 2.99 kg, respectfully.³ We found smaller differences. Self-reported weights after bariatric surgery may be more accurate because participants who undergo surgery to lose weight may be especially attentive to their weight.

A limitation of this study is that it used a convenience sample of participants whose self-reported weights were no more than 30 days before a measured weight. Those participants who anticipated being weighed by study personnel may have been more likely to report accurately.

In conclusion, self-reported weights following bariatric surgery were close to measured weights. This suggests that self-reported weights may not unduly affect study results of surgically-induced weight change and can be used when measured weights are not available.