Changes in widespread pain after surgical weight loss in racialized adults: A secondary analysis from a two-year longitudinal study

Abstract

Widespread pain is associated with reduced function and disability. Importantly, three-fourths of the approximately 42% of U.S. adults with obesity have widespread pain. Moreover, rates of adult obesity are higher and widespread outcomes are worse in racialized non-Hispanic Black and Hispanic/Latino/a/X groups, potentially exacerbating existing pain disparities. Bariatric surgery significantly reduces weight and improves pain. However, recurrent or unresolved pain after bariatric surgery can hinder weight loss or facilitate weight regain. The current study conducted a secondary analysis of a longitudinal study of predictors and mechanisms of weight loss after bariatric surgery to examine the point prevalence of widespread pain and pain trajectories 24 months post-surgery. Our secondary aim was to examine the association between weight loss and pain characteristics. Our exploratory aim was to longitudinally examine racial differences in pain trajectories after bariatric surgery. Our results showed that point prevalence decreased after bariatric surgery. Additionally, significant improvements in pain trajectories occurred within the first 3 months post-surgery with a pattern of pain reemergence beginning at 12 months post-surgery. Hispanic/Latino/a/X participants reported a higher number of painful anatomical sites before bariatric surgery, and the rate of change in this domain for this group was faster compared to the racialized non-Hispanic Black participants. These findings suggest that pain improvements are most evident during the early stages of surgical weight loss in racialized populations of adults with widespread pain. Thus, clinicians should routinely monitor patients’ weight changes after bariatric surgery as they are likely to correspond to changes in their pain experiences.

Introduction

An estimated 52 million adults in the United States (U.S.) experience chronic pain. Furthermore, 20 million U.S. adults experience high-impact chronic pain, which is characterized by severe pain for 3 months or longer^1,2. The approximate healthcare costs associated with chronic pain in the U.S. are about $560-$635 billion³. Widespread pain (WP) is a pain subtype in which pain is reported at multiple anatomical sites⁴ and is associated with reduced function⁵, increased fall risk⁶, and disability⁷. Importantly, WP is more prevalent than pain localized to a single site⁸. Furthermore, it is a key feature of regional and chronic overlapping pain conditions, such as knee osteoarthritis⁹, chronic non-specific low back pain, and fibromyalgia^10,11. Thus, understanding the burden and symptomatology of WP is clinically important.

Concerns about the growing domestic and global populations of adults with living with obesity have accelerated the interrogation of the relationship between weight and chronic musculoskeletal pain. Concurrently, the elimination of pain disparities and inequities has become a national priority, and, thus, a burgeoning area of pain research¹². Three-fourths of the approximately 42% of U.S. adults with obesity have WP compared to 10-20% of the general population^13-18. Moreover, obesity prevalence is significantly higher among racialized non-Hispanic Black (NHB) and Hispanic/Latino/a/X adults in the U.S. compared to non-Hispanic White (NHW) adults^13,19. The epidemiological surveillance of chronic pain is critical for estimating future chronic pain burden, as well as for determining if there is resolution, maintenance, or exacerbation of chronic pain disparities following intervention²⁰. Emerging scholarship has highlighted the cultural significance of weight-related social identity and its influence on intervention recommendations for chronic pain²¹ and on other health outcomes.^22,23 Thus, understanding WP disparities at the intersection of race and weight is pivotal to the advancement of pain equity research²⁴.

Recent studies suggest that the prevalence, chronicity, and impact on musculoskeletal pain progressively increase with higher BMI. Bariatric surgery significantly reduces weight and improves musculoskeletal pain up to seven years post-surgery^25-27. Despite the documented benefits, previous studies have shown refractory or worsening musculoskeletal pain after bariatric surgery^25,28. Moreover, recurrent or unresolved musculoskeletal pain after bariatric surgery can hinder weight loss or contribute to weight regain²⁸. However, historically, previous studies examining longitudinal pain outcomes following bariatric surgery featured NHW adults. Since an increasing number of NHB and Hispanic/Latino/a/X adults are electing to undergo bariatric surgery²⁹, limited racial and ethnic representation in surgical weight loss studies has significantly hindered the monitoring of chronic pain disparities and the investigation of the intra-group variability in pain trajectories among populations with diverse ethnocultural identities.

The present study aims to address these important literature gaps. Based on a comprehensive parent longitudinal study involving racialized NHB and Hispanic/Latino/a/X adults who underwent bariatric surgery via sleeve gastrectomy, the present study conducted a secondary analysis with the following primary aims: 1) to examine the point prevalence of WP 24 months after surgery in racialized adults, and 2) to examine the trajectories of pain intensity and pain interference ratings and the number and location of painful anatomical sites 24 months after surgery in racialized adults. Our secondary aim was to examine the longitudinal association between weight loss and pain characteristics. Our tertiary, hypothesis-generating exploratory aim was to longitudinally examine racial differences in pain trajectories after bariatric surgery. We hypothesized that: 1) the point prevalence of WP would decrease following bariatric surgery, 2) pain intensity and pain interference ratings as well as the number and location of painful anatomical sites would be reduced post-bariatric surgery, and 3) greater weight loss would be associated with lower pain intensity and pain interference ratings and with a lower number of painful anatomical sites post-bariatric surgery.

Methods

All information is presented as recommended by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Checklist³⁰.

Study Design

This study is a secondary analysis of data from a longitudinal observational cohort study focused on mechanisms and predictors of weight change after bariatric surgery in a stratified cohort of patients who self-identified as NHB and Hispanic/Latino/a/X adults with widespread musculoskeletal pain prior to bariatric surgery. The current study featured a quasi-experimental study design to examine the effect of surgical weight loss on pain without an equivalent control group or intervention³¹. Since study participants serve as their own controls, the study design for the secondary analysis accounted for patient characteristics that may affect weight loss but would not significantly change over a short period of time (e.g. age). The primary aim of the parent study was to determine mechanisms and predictors of weight loss and weight maintenance after bariatric surgery. The parent study featured anthropometric measurements, a biorepository of tissue samples, and administration of health surveys before and over 2 years (24M) after bariatric surgery. For the current study, we analyzed data from participants that had a full complement of pain and anthropometric data to address our primary study aims of 1) determining WP point prevalence, and 2) examining longitudinal changes in pain trajectories up to 24M after surgical weight loss. The primary clinical endpoints were pain intensity and pain interference ratings, as well as the number of painful anatomical sites before surgery (baseline), at 3 months (3M), 12 months (12M), and 24 months (24M) after surgery. Participants were also queried about the location of the most painful anatomical site at all time points. Patients or patient advocates were not involved in study conceptualization or design prior to study enrollment. The parent study was approved by the institutional review board of NYU Grossman School of Medicine (IRB # 16-01995).

Setting and Recruitment

Patients who were scheduled for a sleeve gastrectomy surgical procedure were recruited in person before or after a clinical visit as part of the standard of care from NYU Langone Tisch and NYC H+H/Bellevue Hospitals in New York, NY, from June 2017 to March 2020. Of note, study recruitment and data collection were temporarily halted due to institutional and municipal restrictions in the wake of the COVID-19 pandemic. The parent study is ongoing with scheduled follow-up remote study visits for 10 years. Written informed consent was obtained in English or Spanish at NYU Langone’s Weight Management Program and at the NYC H+H/Bellevue Bariatric Surgery Clinic prior to study enrollment.

Study Participants

The primary inclusion criteria for the parent study were: 1) 18-75 years of age, 2) bariatric surgical candidates undergoing the sleeve gastrectomy procedure, 3) literate and language- and numeracy-fluent in conversational English or Spanish, and 4) able to provide written informed consent. The primary exclusion criteria for the parent study were: 1) previous history of bariatric surgery, and 2) having insulin-dependent diabetes mellitus. Patients are eligible for bariatric surgery if they have a BMI ≥ 35, or patients with type-2 diabetes and BMI ≥ 35, or BMI between 30 – 34.9 who have not achieved substantial or durable weight loss or comorbidity improvement with behavioral lifestyle weight loss interventions³². Patients also underwent psychological, dietary, and social support assessments prior to surgery. For the secondary analysis of the parent study, we applied additional pain-related inclusion criteria to identify patients who had WP prior to bariatric surgery. Study participants were classified as having WP if they met the following criteria based on modified Widespread Pain Index 2019 criteria (WP2019)⁴: 1) self-reported pain at 3 or more sites and pain intensity of 3 out of 10 or higher on Brief Pain Inventory Short Form (BPI-SF) Severity scale over the last 7 days prior to sleeve gastrectomy. Participants who had the full complement of pain survey and clinical weight loss data prior to sleeve gastrectomy were included in the current study. At the time of data collection for the parent study, research personnel were unable to distinguish participants who did or did not have WP. Based on recommendations from the Sex and Gender Equity in Research (SAGER) guidelines³³, the gender identities of participants in the current study were defined based on self-report. Therefore, gender identity was included in all statistical analyses.

Sample size.

For the parent study, 647 patients were enrolled and 355 participants completed baseline assessments. Of these, 300 participants underwent sleeve gastrectomy. The current analysis includes 256 participants at baseline, 192 at 3M, 174 at 12M and 144 at 24M with complete Brief Pain Inventory Short Form (BPI-SF) questionnaire and anthropometric data with and without WP.

Data Collection Procedures

Pain Assessment.

The pain domains being evaluated in these analyses were pain intensity and pain interference ratings, and number and location of painful anatomical sites which were assessed using the Brief Pain Inventory Short Form (BPI-SF)³⁴. The BPI-SF questionnaire was administered to participants in their preferred language, either English or Spanish, using one of three methods: 1) in-person, via tablet computers utilizing NYU Langone's licensed Qualtrics software, or 2) remotely through a secure link via email with their unique embedded de-identification code distributed via Qualtrics. Qualtrics is an online research-oriented web tool that complies with the General Data Protection Regulation (GDPR) and employs secure data encryption. The BPI-SF is a validated 9-item questionnaire that queries pain intensity, anatomical location (Question 2), and the impact of pain on respondents’ daily functioning. The BPI-SF Severity subscale items query pain temporality (“worst”, “least”, “average”, “now”) with anchors of 0 (“No Pain”) to 10 (“Pain As Bad As You Can Imagine”). The calculated mean score of these item responses was used to generate a BPI-SF composite score for the Pain Severity subscale, and individual items were also scored to assess the location of painful anatomical sites as recommended by IMMPACT for the evaluation of pain outcomes in clinical trials^35-37. The BPI-SF Interference subscale queries how much pain interferes with seven domains of daily living activities over the previous 24 hours with anchors of 0 (“Does Not Interfere”) to 10 (“Completely Interferes”). The score for BPI-SF Pain Interference subscale is the calculated mean of the item responses. The BPI-SF has been validated in bariatric surgical populations^38,39 and in Spanish-speaking populations (Cronbach α =0.93)⁴⁰. A 2-point change in BPI Severity is considered clinically significant in widespread pain populations⁴¹. Data collected at baseline, 3M, 12M, and 24M after bariatric surgery were included in all statistical analyses.

Anthropometrics and Weight Loss Measurement.

Height was assessed with precision to the nearest 1 cm using a stadiometer (CECA 213, Seca GmBH & Co. KG, Hamburg, Germany). Body weight was measured while individuals donned light clothing, without footwear, and was recorded to the nearest 0.1 lb. employing a Stow-A-Weigh scale (Scale-Tronix, Welch Allyn, Skaneateles, NY, USA). BMI was calculated using body weight and height measurements (kg/m²). The percentage of total weight loss (%TWL), the ratio of the amount of weight lost compared to presurgical weight, was measured at 3M, 12M, and 24M. The %TWL was calculated as: [(baseline weight – current weight) / (baseline weight) ×100]⁴², and was an exploratory outcome.

Statistical Analysis

Data from study participants who underwent surgery, completed the BPI, and had anthropometric measurements at baseline were analyzed. Demographic, pain outcome, and anthropometric data were analyzed using standardized descriptive statistics. All statistical analyses were performed using R software version 4.3.0. We considered statistical significance to be achieved when p-values were less than 0.05 or lower for multiple comparisons.

Analyses for Primary Aims

Prevalence of WP overtime.

To test the primary hypothesis that WP prevalence would be reduced after bariatric surgery, frequency analyses were performed to determine the point prevalence of WP within the cohort (relative frequency) at all endpoints (baseline, 3M, 12M, 24M). The proportions of participants that met the criteria for WP at each endpoint were calculated.

Changes in pain-related outcome trajectories.

To determine longitudinal patterns of change in the pain domains (e.g., increasing, decreasing, or stable between baseline and 24M after bariatric surgery), we applied trend analyses to record-level raw pain data⁴³. In visualizing individual pain domain trajectories, we observed “broken-arrow” trajectory patterns over time. These patterns showed steeper slopes from baseline to 3M post-bariatric surgery, followed by relatively flatter slopes from 3M to 24M post-surgery. We also computed the absolute frequencies of reported pain in select anatomical locations.

Analyses for Secondary Aim

To assess the association between pain measures and %TWL, we employed linear mixed models (LMMs) that were fitted at three subsequent time points (i.e., 3M, 12M, and 24M) to appropriately capture these time trends. The LMM accommodates varying slopes before and after a specified event, which, in our study, occurred at the 3M timepoint following bariatric surgery. In the full models, adjustments were made for time, age, race, and gender identity, whereas the reduced models focused solely on the time variable. Random effects were employed to account for subject-specific variability of pain measures at baseline. We considered statistical significance to be achieved when p-values were less than 0.05 or lower for multiple comparisons.

Analyses for Exploratory Aim

To conduct a preliminary examination of the difference in pain trajectories between NHB and Hispanic/Latino/a/X individuals, we utilized piecewise LMMs with a knot at the 3M post-bariatric surgery timepoint⁴⁴. The full models included an interaction effect involving race and time while controlling for age, racial classification, and gender identity. The reduced models specifically focused on the effect of time. To determine differences in the estimates of pain characteristics between the two racial groups at each time point, we selected the appropriate statistical test based on the results of the Shapiro-Wilks normality test. If the data met the normality assumptions, we conducted a t-test. In instances where normality assumptions were not met, we conducted the Wilcoxon rank-sum test.

Sensitivity Analyses and Imputation of missing data.

To evaluate the impact of assumptions, potential biases, and alternative analytical strategies on the primary results, we carried out a series of pre-planned sensitivity analyses. Patterns of missing data were evaluated by using Little’s missing completely at random (MCAR) test⁴⁵. It suggested that the data is missing completely at random (p=0.802). To scrutinize the impact of missing data on our statistical analyses, multiple imputation of missing data was performed using the fully conditional specification using “mice” package (3.16.0) in R. For the secondary aim, the same analysis was also conducted on the imputed data. For testing the difference between groups at multiple time points (Exploratory Aim), we applied the False Discovery Rate (FDR) correction method to address the concern of multiple comparisons.

Results

Demographic Characteristics.

Table 1 summarizes the demographic characteristics of the study cohort included in the secondary analysis. At baseline, study participants self-identified as Hispanic/Latino/a/X (77%) or NHB (23%). Women-identifying participants comprised the majority of the study sample at all time points (Table 1). The study cohort was primarily comprised of younger adults at baseline (M = 38.3 years, SD = 11.7) with an average BMI in the WHO Class III obese range at baseline (M = 43.2, SD = 6.9). The average %TWL was 20.0% (SD = 4.3) at 3M, 28.2% (SD = 8.0) at 12M, and 25.4% (SD = 8.3) at 24M post-surgery (Table 1). The majority of the study cohort reported their employment status as family caregivers seeking work (24%), working full time (20%), family caregivers not seeking work (17%), or working part time (14%) (Table 1). Additionally, the study cohort reported having completed graduate or post-professional education (6%), 4 years or less of collegiate education (47%), or having completed their high school education (27%), while others reported not completing their high school education (20%). The majority of the study cohort reported their household income between less than $10,000 - $50,000 USD (88%). There were no significant demographic differences at baseline between the parent (N = 300) and the secondary analysis (N = 256) study cohorts (Supplementary Table S1). There were significant differences at baseline in age, weight, and pain intensity between those who did or did not complete all of the assessments that were administered for the parent study at 12M or 24M post-surgery (Supplementary Table S2). It is important to note that these demographic characteristics were collected prior to bariatric surgery and the implementation of restrictions in the wake of the COVID-19 pandemic. Collecting this data over time may be important for future studies to track and document changes in these characteristics and their potential impact on pain and weight status. There were statistically significant differences in weight and BMI at baseline between the racialized Hispanic/Latino/a/X and NHB groups at all time points (p < 0.001, Supplementary Table S3). There were no significant differences in age or in the percentage of women-identifying participants between racialized groups (Supplementary Table S3).

Table 1.

Demographic characteristics of the study cohort. Baseline: prior to bariatric surgery; 3M: 3 months after bariatric surgery; 12M: 12 months after bariatric surgery; 24M: 24 months after bariatric surgery. *Employment status, education, and income level were collected at baseline.

	Baseline	3M	12M	24M
	n = 256	n = 192	n = 174	n = 144
Race/Nationality, n (%)
Non-Hispanic Black (NHB)	59 (23.0)	43 (22.0)	36 (21.0)	31 (22.0)
Hispanic/Latino/a/X	197 (77.0)	149 (78.0)	139 (79.0)	113 (78.0)
Age, years (mean, SD)	38.3 (11.7)	41.9 (11.7)	42.6(11.7)	43.6 (11.7)
Gender (% women-identifying)	211 (82.4)	159 (82.8)	145 (83.3)	117 (81.2)
Body Mass Index, kg/m2 mean (SD)	43.2 (6.9)	34.6 (5.9)	30.6 (5.6)	31.9 (6.0)
Weight, kg (mean, SD)	116.6 (22.9)	93.4 (19.7)	82.1 (17.7)	85.8 (19.8)
%TWL, mean (SD)	-	20.0 (4.3)	28.2 (8.0)	25.4 (8.3)
Employment status, n (%)
Working full time	52 (20%)
Working part time	36 (14%)
Seeking work	17 (7%)
Going to school	27 (11%)
Caring for family	43 (17%)
Caring for family & seeking work	62 (24%)
Recovering from illness	18 (7%)
Missing	1 (0%)
Education, n (%)
Grad/Prof school	16 (6%)
4 yr. college	52 (20%)
< 4 yr. college	69 (27%)
High School	68 (27%)
Some High School	51 (20%)
Income level, n (%)*
<$10,000	64 (25%)
Between $10,000 - $30,000	114 (45%)
Between $30,000 - $50,000	46 (18%)
>$50,000	32 (12%)
Pain Intensity	4.3 (2.8)	2.6 (2.7)	2.7 (2.7)	3.2 (2.8)
Pain Interference	3.7 (3.1)	1.9 (2.6)	2.1 (2.9)	2.5 (3.0)
Number of Painful Anatomical Sites	6.4 (6.3)	3.7 (5.5)	3.2 (3.6)	3.7 (4.8)
Location of Painful Anatomical Sites
Low back (%)	66/256 (25.8)	29/192 (15.1)	33/174 (19.0)	35/144 (24.3)
Legs, bilateral (%)	48/256 (18.8)	28/192 (14.7)	18/174 (10.3)	16/144 (11.1)
Feet, bilateral (%)	45/256 (17.6)	12/192 (6.3)	6/174 (3.4)	6/144 (4.2)
No Pain (%)	44/256 (17.2)	82/192 (42.3)	69/174 (39.7)	50/144 (34.7)

Primary Aim 1: WP prevalence (Figure 1).

Figure 1. The prevalence of widespread pain (WP) before and after bariatric surgery.

A. Stacked bar graph of the percentage of participants included in the secondary analysis that had WP (pink) or did not have WP (black) at Baseline (pre-surgery, n=256), and at 3M (n = 192), 12M (n = 174), and 24M (n = 144) after bariatric surgery. B. Bar graphs of the percentage of the NHB (black) and Hispanic/Latino/a/X (light brown) study participants that had WP at Baseline (pre-surgery), and at 3M, 12M, and 24M after bariatric surgery.

The point prevalence of WP for the entire study cohort was 58.6% at baseline, 26.6% at 3M, 32.2% at 12M, and 36.1% at 24M post-surgery (Figure 1A). The percentages of racialized Hispanic/Latino/a/X study participants with WP was 60.4% at baseline (n = 119/197 Hispanic/Latino/a/X study participants), 26.2% at 3M (n = 39/149), 31.7% at 12M (n = 44/139), and 38.9% at 24M (n = 44/113). The percentages of racialized NHB study participants with WP was 52.5% at baseline (n = 31/59 NHB study participants), 30.2% at 3M (n = 13/43), 33.3% at 12M (n = 12/36), and 25.8% at 24M (n = 8/31) (Figure 1B). Thus, racialized Hispanic/Latino/a/X study participants had a greater WP burden at baseline and 24M post-surgery while racialized NHB participants had a greater burden at 3M and 12M post-surgery.

Primary Aim 1: Location of painful anatomical sites.

Table 1 and Figure 2 summarize the absolute frequencies (i.e., percentages) of self-reported painful anatomical sites in the trunk and lower extremities before and after bariatric surgery. The low back, bilateral legs, and feet regions were the most frequently reported painful sites. At baseline, the low back region was reported as the most painful anatomical site (25.8%), followed by the bilateral legs (18.8%) and feet (17.6%). Participants continued to report the low back as the most painful anatomical site at 3M (15.1%) with a greater proportion of participants reporting low back pain at 12M (19.0%) and 24M (24.3%) after bariatric surgery. At baseline, 17% of participants did not report the trunk or bilateral lower extremity regions as the most painful anatomical sites (No Pain). The percentage of participants not reporting pain in the trunk or lower extremities was 42.3% at 3M, 39.7% at 12M, and 34.7% at 24M after bariatric surgery (Table 1, Figure 2).

Figure 2. Location of painful anatomical sites before and after bariatric surgery.

Stacked bar graph of the percentage of participants included in the secondary analysis that reported pain in the low back (black), bilateral legs (pink), bilateral feet (turquoise), or did not have pain in these anatomical locations (No Pain, purple). The measured time points were Baseline (pre-surgery, n=256), and at 3M (n=192), 12M (n=174), and 24M (n=144) after bariatric surgery.

Primary Aim 2: Trajectories of pain intensity and interference ratings and the number of painful body sites.

Table 1 and Figure 3 present the results from trend analyses performed to determine the longitudinal trajectories of pain intensity and interference and of the number of painful anatomical sites up to 24M after bariatric surgery for the entire stratified study cohort in the context of %TWL (Figure 3A). The average pain intensity rating for the cohort at baseline was 4.3 (SD = 2.8), pain interference rating was 3.7 (SD = 3.1), and number of painful anatomical sites was 6.4 (SD = 6.3). The average pain intensity rating for the cohort was 2.6 (SD = 2.7) at 3M, 2.7 (SD = 2.7) at 12M, and 3.2 (SD = 2.8) at 24M post-surgery. The average pain interference rating was 1.9 (SD = 2.6) at 3M, 2.1 (SD = 2.9) at 12M, and 2.5 (SD = 3.0) at 24M post-surgery. The average number of painful anatomical sites was 3.7 (SD = 5.5) at 3M, 3.2 (SD = 3.6) at 12M, and 3.7 (SD = 4.8) at 24M post-surgery (Table 1). There was a sharp decrease in the magnitude of change in pain intensity ratings (Figure 3B), pain interference ratings (Figure 3C), and in the number of painful anatomical sites between baseline and 3M post-surgery (Figure 3D). However, there were no differences in the magnitude of change in any of the pain domains between 3M and 24M post-surgery (Figure 3B-D).

Figure 3. Trajectories of weight loss, pain intensity and interference ratings, and the number of painful body sites.

Line graphs illustrating the results of trend analyses of the trajectories of: A. percentage of weight loss (%TWL), B. Pain intensity (BPI Severity scores), C. Pain interference (BPI Interference scores), and D. Number of painful anatomical sites (Number of Pain Sites). The data represents measured time points were Baseline (pre-surgery, n=256), and at 3M (n=192), 12M (n=174), and 24M (n=144) after bariatric surgery for the entire study cohort.

Secondary Aim: Associations between weight loss and pain.

There were statistically significant differences in weight, BMI, and %TWL between the racialized Hispanic/Latino/a/X and NHB groups at 3M, 12M, and 24M post-surgery (p < 0.009) (Figure 4A, Supplementary Table S3). Bivariate correlational analyses between %TWL, pain intensity and interference ratings, and the number of painful anatomical sites show that %TWL was significantly associated with pain intensity and pain interference ratings but not with the number of painful anatomical sites at 3M, 12M, and 24M post-surgery (Supplementary Table S4). Table 2 presents the results from the LMM performed to test the association between %TWL and the pain domains between 3M (when %TWL is initially measured) and 24M after bariatric surgery. For the unadjusted model (Model 1), there were significant inverse associations between %TWL and pain intensity ratings (β = −0.044, p = 0.004), pain interference ratings (β = −0.046, p = 0.006), and the number of painful anatomical sites (β = −0.055, p = 0.049) between baseline and 24M post-surgery (Table 2). After adjustment for time, age, racial classification (Race), and gender identity (Model 2), %TWL remained inversely associated with pain intensity ratings (β = −0.033, p = 0.029) and pain interference ratings (β =−0.034, p = 0.042), but not with the number of painful anatomical sites (β =−0.049, p = 0.090). The findings for pain intensity and interference ratings were consistent after employing multiple imputation (Model 3), but the %TWL was inversely associated with number of painful anatomical sites and remained significant after adjustment for age. Thus, a greater %TWL was associated with lower pain intensity and lower pain interference ratings following bariatric surgery.

Figure 4. Racial group differences in the trajectories of weight loss, pain intensity and interference ratings, and the number of painful body sites.

Line graph of the trajectories of the percentage of total weight loss (%TWL) (A), pain intensity ratings (B), pain interference ratings (C), and number of painful anatomical sites (D) between Baseline (pre-surgery) and 24M post-surgery. The red lines represent data from participants in the racialized Hispanic/Latino/a/X group. The blue lines represent data from participants in the racialized non-Hispanic Black (NHB) group. Data were analyzed to test for differences between racialized groups at each time point.*p < 0.05, ***p<0.001; ns: not statistically significant.

Table 2.

Results from linear mixed model (LMM) analyses to determine longitudinal correlations between pain and percentage of total weight loss (%TWL) between Baseline (pre-surgery) and 3M after bariatric surgery. Model 1 is unadjusted for covariates. Model 2 is adjusted for covariates (age, racial classification (Race), and gender identity). Model 3 is adjusted for covariates using imputed data. Fixed effects: proportion of variation explained by the fixed effects in the LMM. Bold font indicates statistical significance at p < 0.05.

	Primary IVs	Model 1 Unadjusted		Model 2 Adjusted for covariates: Age, Race, Gender		Model 3 (imputed data) Adjusted for covariates: Age, Race, Gender
		β	p	β	p	β	p
Pain Intensity	%TWL	−0.044	0.004	−0.033	0.029	−0.058	<0.001
	Time (month)	0.037	<0.001	0.033	0.002	0.043	<0.001
	Age	-	-	0.075	<0.001	0.071	<0.001
	Race	-	-	−0.058	0.876	−0.011	0.973
	Gender	-	-	−0.555	0.171	−0.679	0.050
	R2
	Fixed effects	0.051		0.154		0.138
		β	p	β	p	β	p
Pain Interference	%TWL	−0.046	0.006	−0.034	0.042	−0.060	<0.001
	Time (month)	0.033	0.005	0.029	0.015	0.038	0.005
	Age	-	-	0.056	<0.001	0.052	<0.001
	Race	-	-	0.225	0.558	0.123	0.711
	Gender	-	-	−0.423	0.310	−0.519	0.144
	R2
	Fixed effects	0.045		0.100		0.094
		β	p	β	p	β	p
Number of Painful Anatomical Sites	%TWL	−0.055	0.049	−0.049	0.090	−0.099	0.004
	Time (month)	0.013	0.524	0.010	0.625	0.021	0.408
	Age	-	-	0.044	0.063	0.043	0.045
	Race	-	-	−0.395	0.549	−0.695	0.254
	Gender	-	-	−0.873	0.221	−0.979	0.125
		β	p	β	p	β	p
Pain Intensity	%TWL	−0.044	0.004	−0.033	0.029	−0.058	<0.001
	Time (month)	0.037	<0.001	0.033	0.002	0.043	<0.001
	Age	-	-	0.075	<0.001	0.071	<0.001
	Race	-	-	−0.058	0.876	−0.011	0.973
	Gender	-	-	−0.555	0.171	−0.679	0.050
	R2
	Fixed effects	0.051		0.154		0.138
	R2
	Fixed effects	0.018		0.035		0.070

Exploratory Aim Findings: Racial group differences in pain trajectories before and after bariatric surgery.

Table 3 summarizes the results of the piecewise LMMs used to examine racial differences in pain domain trajectories. Figure 4 illustrates pain trajectories for the racialized groups. We observed significant racial group differences in the number and rate of reduction in the number of painful anatomical sites before and after bariatric surgery which were not observed for pain intensity and interference ratings.

Table 3.

Results from the piecewise linear mixed model (LMM) analyses to test the difference in pain trajectories between the non-Hispanic Black (NHB) and Hispanic/Latino/a/x racialized groups with a knot at the 3M post-bariatric surgery timepoint. Model 1 was a reduced model that featured the main effect of Time and the interaction effect between racial classification (Race) x Time. Model 2 is adjusted for covariates (age, racial classification (Race), and gender identity). Fixed effects: proportion of variation explained by the fixed effects in the LMM. Bold font indicates statistical significance at p < 0.05.

	Primary IVs	Model 1 Unadjusted		Model 2 Adjusted for covariates: Age, Race, Gender
		β	p	β	p
Pain Intensity	Time_1 (BL – 3M)	−0.598	<0.001	−0.625	<0.001
	Time_2 (3M - 24M)	0.624	<0.001	0.651	<0.001
	Age	-	-	0.073	<0.001
	Race (NHB)	-	-	−0.150	0.701
	Gender	-	-	−0.730	0.047
	Time_1 x Race (NHB)	-	-	0.080	0.551
	Time_2 x Race (NHB)	-	-	−0.086	0.561
	R2
	Fixed effects	0.063		0.159
		β	p	β	p
Pain Interference	Time_1 (BL – 3M)	−0.608	<0.001	−0.680	<0.001
	Time_2 (3M - 24M)	0.631	<0.001	0.707	<0.001
	Age	-	-	0.056	<0.001
	Race (NHB)	-	-	−0.416	0.329
	Gender	-	-	−0.583	0.132
Pain Intensity	Time_1 (BL – 3M)	−0.598	<0.001	−0.625	<0.001
	Time_2 (3M - 24M)	0.624	<0.001	0.651	<0.001
	Age	-	-	0.073	<0.001
	Race (NHB)	-	-	−0.150	0.701
	Gender	-	-	−0.730	0.047
	Time_1 x Race (NHB)	-	-	0.080	0.551
	Time_2 x Race (NHB)	-	-	−0.086	0.561
	R2
	Fixed effects	0.063		0.159
	Time_1 x Race (NHB)	-	-	−0.284	0.070
	Time_2 x Race (NHB)	-	-	−0.304	0.077
	R2
	Fixed effects	0.064		0.116
		β	p	β	p
Number of Painful Anatomical Sites	Time_1 (BL – 3M)	−0.960	<0.001	−1.149	<0.001
	Time_2 (3M - 24M)	0.957	<0.001	1.146	<0.001
	Age	-	-	0.051	0.026
	Race (NHB)	-	-	−2.480	0.001
	Gender	-	-	−1.582	0.024
	Time_1 x Race (NHB)	-	-	−0.789	0.006
	Time_2 x Race (NHB)	-	-	−0.798	0.011
	R2
	Fixed effects	0.060		0.091

The racialized Hispanic/Latino/a/X and NHB groups did not have significant differences in pain intensity or pain interference ratings at baseline (Figure 4B-C). Further, the racialized Hispanic/Latino/a/X group did not show a significantly faster rate of reduction in pain intensity (β = −0.080, p = 0.551) or pain interference ratings (β = −0.284, p = 0.070) between baseline and 3M or between 3M and 24M after bariatric surgery compared to the racialized NHB group (Table 3). The Hispanic/Latino/a/X racialized group reported more painful anatomical sites at baseline (β = −2.480, p = 0.001), and demonstrated a faster rate of reduction in the number of painful anatomical sites between baseline and 3M (β = −0.789, p = 0.006) and between 3M and 24M after bariatric surgery (β = −0.798, p = 0.011) compared to the racialized NHB group (Figure 4D, Table 3). However, there were no significant racial group differences in the rates of reduction in pain intensity or pain interference ratings between 3M and 24M after bariatric surgery. Additionally, there were no significant differences in pain intensity ratings or pain interference ratings between the racialized groups at baseline or at 3M, 12M, or 24M after bariatric surgery (Figure 4A-B, Supplementary Table S3).

Discussion

This secondary analysis examined the point prevalence of WP and pain trajectories of Hispanic/Latino/a/X and NHB adults up to 24 months after bariatric surgery. We hypothesized that WP point prevalence would decrease after bariatric surgery, which was confirmed. We also hypothesized that there would be improvements in pain trajectories after bariatric surgery, which was partially supported. As a secondary aim, we hypothesized that greater weight loss would be associated with lower pain intensity and interference ratings as well as with a reduction in the number of painful anatomical sites after bariatric surgery, which was partially supported. Results from exploratory analyses showed racial group differences in the number and trajectories of painful anatomical sites between Hispanic/Latino/a/X participants and NHB study participants. Interestingly, there were no group differences in the rates of change in pain intensity or pain interference ratings. To our knowledge, this is the first study to specifically investigate: 1) changes in the point prevalence of WP after bariatric surgery, and 2) the trajectories of multiple pain domains associated with WP after bariatric surgery in a large cohort of racialized NHB and Hispanic/Latino/a/X adults.

WP burden after bariatric surgery

The Department of Health and Human Services (HHS) National Pain Strategy recommends collecting population-level data on the prevalence, outcomes, and progression of chronic pain conditions². Much of the scientific knowledge of WP prevalence relies on cross-sectional or aggregated population data from single points. Few contemporary epidemiological studies have specifically investigated the burden of the WP in racialized groups^46,47 or in adults with BMIs consistent with obesity^15,16,48,49. The current study addresses these important research gaps by evaluating short-term changes in WP burden after surgical weight loss in a large cohort of racialized Hispanic/Latino/a/X and NHB adults. Our findings reveal that nearly 60% of the study cohort met WP criteria before bariatric surgery, and more than one-third met the criteria 24 months post-surgery. This is consistent with reported prevalence of self-reported pain intensity and interference in a cohort of ethnoculturally diverse adults living with obesity enrolled in a lifestyle weight management intervention⁵⁰, and significantly higher than the general population^17,18. Current estimates of weighted population-level prevalence of WP range from 11% to 29% among racialized groups in the U.S.^51,52. Although these findings revealed racial disparities in WP burden, there are often profound disparities for populations at the identity intersection of race and weight status. Thus, there is a greater likelihood of erroneous estimations of the present and future burden of WP in medically underserved populations. Consequently, researchers and clinicians may miss crucial opportunities to monitor and document disparities in WP burden among racialized populations with obesity.

WP trajectories after bariatric surgery: implications for clinical practice

Globally, sleeve gastrectomy has become the most frequently performed surgical weight loss procedure⁵³ with an increasing number of racialized Hispanic/Latino/a/X and NHB adults opting to undergo it²⁹. Additionally, prospective surgical candidates cite pain remission and functional improvement as motivators for undergoing bariatric surgery⁵⁴. The current study shows that the sharpest pain reduction occurred within the first 3 months of post-surgical weight loss. Further, there was a notable deflection point in the pain intensity and pain interference ratings where pain worsened in racialized NHB participants between 12 and 24 months post-surgery though not statistically significant. Our findings are consistent with prior work that reported a similar magnitude and rate of reduction in pain intensity and interference within the first 12 months after bariatric surgery in predominantly NHW patient cohorts^25,28,55. However, the current study findings suggest an earlier pattern of pain reemergence, particularly among NHB participants, whereas previous studies have reported that pain reemerges between 24 and 36 months post-surgery when weight regain is common^25,28. These findings should be interpreted with the understanding that several hypothesized mechanisms underlying pain changes, including mechanical joint loading reduction^56,57, impaired nociceptive modulation^58-61, and an upregulation of biomarkers associated with nociception and pain, stress, and obesity^62,63 are likely active between 12 and 36 months post-surgery^64-66. It is possible, however, that other salient features of the pain experience in ethnoculturally diverse adults such as pain coping strategies^67,68, ‘perceived’ stress⁶⁹, and repeated exposure to multiple forms of discrimination⁵⁰, contributed to the earlier deterioration of pain improvements via the dysregulation of the physiological systems active in chronic pain and obesity in response to repeated exposure to race-related stress^70,71. Our results suggest opportunities to optimize the timing of pain interventions for linguistically diverse adults that specifically address pain intensity and interference - if these symptoms are endorsed - after 12 months post-surgery when weight regain may occur.

The current results show that low back and lower extremities were the most frequently reported painful anatomical sites which improved after bariatric surgery. These findings are consistent with previous studies that have documented the co-occurrence of WP and low back pain^72-74, and improvements in chronic knee and low back pain after bariatric surgery^55,75,76. Notably, nearly 25% of participants in the current study continued to report the low back as the most painful anatomical site 24 months after bariatric surgery, which suggests that the low back pain may be less responsive to surgical weight loss alone compared to other lower extremity sites. Collectively, these findings highlight the need to evaluate changes in spatial pain distribution of pain as part of the standard of care in surgical weight management. Doing so will allow for more precise integration of pain assessments and prescription of rehabilitative interventions that are most efficacious for improving chronic low back and lower extremity pain.

The current study shows significant associations between weight loss and pain intensity and pain interference ratings. Interestingly, weight loss was not significantly associated with the number of painful anatomical sites over time (unimputed data), suggesting a heterogenous response to surgical weight loss among pain domains. Furthermore, our study cohort continued to meet Class I and II WHO obesity criteria (BMI ≥ 30) and WP criteria up to 24 months after bariatric surgery. Despite noted limitations of BMI as an obesity metric⁷⁷, this underscores the potential for continued susceptibility to WP even after significant weight loss. Current results also show differences in weight loss between the racialized groups which is similar to previous study findings in patients who underwent the Roux-en-Y gastric bypass or adjustable gastric band (AGB) procedures⁷⁸. These differences may be attributable to a combination of sociocultural and environmental factors such as dietary customs and disparate access to weight management support⁷⁹.

Racial differences in pain outcomes and the clinical and research implications

Racial differences in pain outcomes have long been studied^24,80,81 but have not been extensively examined in the context of weight loss. The current study shows that pain intensity and interference ratings were similar between NHB and Hispanic/Latino/a/X study participants, consistent with findings from our previous studies⁵⁰. The similarities in pain ratings were consistent despite persistent racial group differences in %TWL. However, racialized Hispanic/Latino/a/X participants reported a higher number of painful anatomical sites prior to bariatric surgery, and the rate of reduction was faster compared to the racialized NHB participants. Previous work has shown that select groups within the Hispanic/Latino/a/X diaspora report numerous painful anatomical sites⁸². Widespread pain is often indicative of central nociceptive processing dysfunction^83-85. Further, racialized NHB and Hispanic/Latino/a/X adults demonstrate differences in central nociceptive processing compared with non-Hispanic White adults⁵⁸. However, the impact of significant weight loss on central nociceptive processing in these racialized groups may differ, although this relationship remains speculative. Intersectionality science features analytical methodologies and conceptual frameworks that critically examine how identity characteristics converge to create unique modes of discrimination that affect health^86-88. Therefore, the intersectionality science may illuminate how the convergence of race, weight, and other identities impacts pain responses to weight loss interventions. This knowledge could accelerate the discovery of unique factors that contribute to vulnerability and resilience to WP symptomatology.

Limitations and recommendations for future studies.

Study attrition over 24 months of follow-up approached 50%. However, work schedule inflexibility, long commutes, and less discretionary income to support long-term participation are significant contributors to study attrition among these populations in weight loss studies^89,90. Also, COVID-19 likely impacted retention. Future studies should consider time flexibility for data collection and providing transportation to enhance study retention. Additionally, other racialized populations were underrepresented in the current study, and we strongly suggest a focused interrogation of the impact of weight loss and regain on these populations given their vulnerability to pain inequities and disparities^51,80. We did not track pain or other medical diagnoses or interventions though evidence suggests significant relationships between obesity and psychological comorbidities.⁹¹. Finally, we did not specifically examine sex and gender differences in pain outcomes primarily due to an underrepresentation of cis- or transgender men in the current study.

Conclusions.

Findings from this study suggest that reductions in WP prevalence and pain improvement are most evident during the early stages of surgical weight loss in racialized populations of adults with WP. Thus, clinicians should evaluate changes in the magnitude and spatial distribution of pain over time, particularly if patients report significant weight change, so that pain interventions can be prescribed with greater precision during and after weight loss.

Perspective.

This article presents the prevalence and pain trajectories of racialized adults with widespread pain (WP) after surgical weight loss. Clinicians should evaluate changes in the magnitude and spatial distribution of pain after significant weight change in these populations so pain interventions can be prescribed with greater precision.

Highlights.

• Surgical weight loss yields early improvement in widespread pain burden

• Pain improvement with surgical weight loss becomes attenuated after one year

• Pain trajectories after surgical weight loss differ between racialized groups