Racial and Social Determinants of Health Disparities in Spine Surgery Affect Preoperative Morbidity and Postoperative Patient Reported Outcomes

Retrospective Observational Study

Sarthak Mohanty, BS; Jenna Harowitz, BA; Meeki K. Lad, MPH; Armaun D. Rouhi, BA; David Casper, MD; Comron Saifi, MD


Spine. 2022;47(11):781-791. 

In This Article


Descriptive Analysis and Demographic Comparisons

Of 1012 patients initially screened, 854 met inclusion criteria; 16 and 121 patients did not complete the PROMIS survey at presentation or follow-up, respectively, and the remaining 21 patients underwent excluded procedures. Demographic characteristics of excluded and included patients did not differ significantly (not shown). Most (73.5%) study subjects identified as White (Table 1). Average age was 61.3 years, average Charleson comorbidity score was 1.67, and average BMI was 28.69. At time of presentation, average pain score was 5.84 [95% CI: 5.66–6.03], median GMH score was 12 [IQR: 9–15], and median GPH score was 11 [IQR: 8–13]. There was no difference in preoperative pain score (P ANOVA = 0.9749) among patients who underwent cervical d/f (5.88, N = 149), cervical/thoracic laminotomy (5.79, N = 80), lumbar laminotomy (5.91, N = 283), LIF (5.75, N = 237), and deformity arthrodesis (5.87, N = 105). Though presenting GMH and GPH scores varied significantly across procedure subtypes (P ANOVA[GMH] = 0.0167; P ANOVA[GPH] = 0.0071), the only clinically meaningful pairwise difference noted was that patients indicated for cervical/thoracic laminotomy had lower presenting GMH than those indicated for cervical d/f (11 vs. 13, P = 0.024). At POD90, there were no significant differences in GMH (P ANOVA = 0.0919) and GPH Scores (P ANOVA = 0.7225) across different surgeries.

Racial Disparities Among Physical Functionality, Pain, and Mental Distress at the Point of Presentation

Though Black/AA patients had higher median presenting pain than White patients across procedure types, these differences were only clinically relevant for cervical/thoracic laminotomy (P = 0.0005), lumbar laminectomy (P < 0.0001), and deformity arthrodesis (P = 0.0012), where Black/AA patients reported 60% higher pain scores (VASBlack = 8 vs. VASWhite = 5; Table 2). Among all procedures, Black/AA patients also had lower median GPH scores, with meaningful differences observed for cervical d/f (12 vs. 9, P = 0.0024), cervical/thoracic laminotomy (11 vs. 6.5, P = 0.0024), and lumbar laminectomy (11 vs. 9, P < 0.0001). Further, among cervical/thoracic laminotomy candidates, Black/AA patients presented with lower GMH scores (12 vs. 6.5, P = 0.0012). Finally, Black/AA patients had clinically worse POD90 and 12M postoperative GPH scores for all procedures (see Table 2).

The Effect of Race in Addition to SES on Physical Functionality, Pain, and Mental Distress at Time of Presentation

To delineate the effect of race on healthcare disparities beyond its relationship with SES, the cohort was stratified into median local household income quartiles. Within these cohorts, the only clinically relevant difference detected between racial groups was in pain scores in the lowest quartile (median household income <$53,498; Table 3); here, Black/AA patients reported 33% higher pain scores (VASBlack = 8 vs. VASWhite = 6, P = 0.0382).

Conversely, to delineate the effect of SES on healthcare disparities, patients in the highest and lowest quartile SES patients within each race were compared. There were no intra-racial differences in VAS pain score when comparing the highest-SES patients with the lowest-SES patients (P Black/AA = 0.7361; P White >0.999). However, there were significant and meaningful differences in both GMH and GPH scores, suggesting that patients of lower SES report increased physical impairment and mental distress at time of presentation. Among White patients, the highest-SES patients had 21.4% higher median GMH scores (14 vs. 11, P < 0.0001) and 16.7% higher median GPH scores (12 vs. 10, P = 0.0107) than their lowest-SES counterparts. Among Black/AA patients, the highest-SES patients had 30% higher median GMH scores (15 vs. 10.5, P = 0.0049) and 35.7% higher median GPH scores (14 vs. 9, P = 0.0002) than their lowest-SES counterparts.

Next, the overall cohort was stratified into presenting GPH score quartiles. Compared with the highest quartile (GPH ≥14), patients with GPH scores in the lowest quartile (GPH = 1–8) reported worse physical and mental distress at both POD90 and POD12M: at POD90, patients had 36.8% worse GMH (12 vs. 19, P < 0.0001) and 37.5% worse GPH (15 vs. 24, P < 0.0001); at POD12M, patients had 33.9% worse GMH (19.5 vs. 29.5, P < 0.0001) and 26.7% worse GPH (22 vs. 30, P < 0.0001). Interestingly, subsequent analysis of the improvement in PROMIS metrics revealed patients with the worst presenting GPH scores had greater improvement in both mental health (90D: P = 0.0233; 12M: P < 0.0001) and physical functionality (90D: P = 0.0005; 12M: P < 0.0001) than patients with the best initial GPH scores.

Notably, when compared with the highest quartile of GPH scores, patients with the lowest GPH scores resided in communities where median household income was 26% lower ($64,117 vs. $86,656; P < 0.0001), median household value was 28% lower ($216,134 vs. $300,136; P < 0.0001), and ADI was 45.4% higher (71.37 vs. 38.94; P < 0.0001; Table 4). These patients' neighborhoods experienced an 8.8% higher obesity prevalence (36.09 vs. 33.19; P = 0.0015), 9.8% higher type II DM prevalence (9.87 vs. 9.01; P < 0.0001), 10.7% higher population per PCP (1.34 vs. 1.21; P = 0.0031), and 19.8% greater percentage of residents without a college degree (66.5% vs. 55.4%; P < 0.0001). These communities were also less physically active (48.0% vs. 52.2% meeting CDC criteria; P = 0.0012).

Regression Analysis

On univariate analysis (Table 5), significant sociodemographic predictors of a lower presenting GPH score were: Black/AA race, neighborhood health (proxied by obesity and diabetes prevalence), population:PCP ratio, percentage of families under the FPL, lower percentage of households with six figure incomes, lower median household value, lower median household income, unemployment rate, and population density. In subsequent multivariable linear regression analysis with stepwise selection, the following were independently associated with clinically meaningful and worse GPH scores:

  • Neighborhood characteristics (non-economic): Each increase in 1000 residents per PCP conferred a 1.616 point decrease in GPH (P < 0.0001). Consistently, areas with population densities exceeding 325.1 people/mi2 showed a subclinical 1.1 to 1.3 point decrease in GPH when compared with those with lower densities (P = 0.0005–0.0023).

  • Neighborhood characteristics (economic): Patients in neighborhoods with median household values less than $169,693 had GPH scores that were 1.504 points lower than those with median household values more than $351,705.5 (P = 0.001).

  • Surgical Indication (clinical): Compared with cervical laminectomy patients, patients eventually indicated for cervical d/f demonstrated GPH scores that were 1.42 points higher (P = 0.0039).