Impact of Socioeconomic Status on the Risk of Intervertebral Disc Degeneration and Low Back Pain in the European Population and the Mediating Role of Obesity-Related Traits: Evidence from a Mendelian Randomization Study

Introduction

Low back pain is typically characterized by discomfort, muscle tension, or stiffness located below the rib margin and above the lower buttocks, with or without accompanying leg pain.¹ In 2020, low back pain remained the leading global cause of years lived with disability (YLDs), affecting 619 million people.² Although age-standardized rates decreased by about 10% from 1990 to 2020, the burden is projected to rise, with over 800 million cases expected by 2050.² Among the numerous causes of low back pain, symptomatic intervertebral disc degeneration (IVDD) is considered a major contributing factor.³ IVDD is a pathological condition arising from the degeneration and aging of the vertebral body, intervertebral discs, ligaments, and other spinal structures, leading to spinal stenosis or intervertebral disc herniation.⁴ It is characterized by progressive loss of proteoglycans and water in the nucleus pulposus, which initiates spinal changes and contributes to nucleus pulposus protrusion, fibrous ring tears, disc narrowing, and ultimately low back pain.⁵ When intervertebral disc protrusion caused by IVDD compresses the spinal cord or nerve roots, it can lead directly to symptoms of low back pain.⁶

Low back pain and IVDD are major global health burdens, severely reducing quality of life and driving significant clinical and economic costs worldwide. In high-income countries, low back pain results in an annual hospitalization rate of 3.2% and substantial economic burden, with direct and total costs per patient estimated at $9231 and $10,143.1, respectively.⁷ In contrast, in low- and middle-income countries, low back pain leads to significantly higher hospitalization rates of 13.4%–18.7% annually, with an economic burden of $2.2 billion per year and $1226.25 per patient.⁸ It is widely recognized that approximately 80% of people will experience low back pain during their lifetime, often severe enough to require medical intervention and cause a loss of productive workdays.³ Identifying modifiable risk factors influencing IVDD and low back pain is critical for developing targeted interventions and policies aimed at alleviating these global health burden. Several risk factors contribute to low back pain and IVDD, including heavy physical labor,⁹ prolonged sitting,¹⁰ obesity,^11,12 smoking,^13,14 and psychosocial stress such as depression.¹⁵ These factors are more prevalent in lower socioeconomic status (SES) populations due to greater exposure to physical demands, unhealthy lifestyles, and limited healthcare access,¹⁶ underscoring SES as a key determinant of risk.

SES is a multifaceted construct that encompasses an individual’s economic resources, social position, educational attainment, and Townsend deprivation index, which has been widely recognized as a significant determinant of health outcomes across various diseases and conditions.¹⁷ Previous research indicates an association between SES disparities and the prevalence of low back pain and IVDD.^18,19 However, these findings largely stem from observational studies, which are susceptible to confounding and thus unable to clarify causal relationships. Additionally, low SES populations often face challenges such as limited access to healthy food, fewer opportunities for physical activity, and higher levels of stress, all of which contribute to higher rates of obesity.^20–22 Obesity, in turn, is a well-established risk factor for IVDD and low back pain.²³ Excessive weight increases biomechanical stress on the spine and contributes to chronic systemic inflammation, which may accelerate the degeneration of intervertebral discs and exacerbate back pain.²⁴ Given these pathways, obesity may act as a mediator in the causal relationship between SES and IVDD or low back pain. However, the degree to which obesity mediates these effects has not been fully clarified. Randomized controlled trials (RCTs) are considered the gold standard for supporting causal inference in epidemiological studies. However, due to issues such as time, cost, and medical ethics, some RCTs are difficult to implement.

To address these limitations, Mendelian randomization (MR) emerges as a robust genetic epidemiology method that employs genetic variants as instrumental variables (IVs) to explore causal links between exposures and outcomes.²⁵ This approach leverages the random assortment of genetic variants during meiosis to mimic a randomized controlled trial, thereby minimizing confounding factors and reverse causation.²⁶ Single nucleotide polymorphisms (SNPs) are commonly used as IVs in MR studies to proxy diverse exposure levels.²⁷ The widespread availability of summary-level statistics from genome-wide association studies (GWAS) facilitates the implementation of MR analyses.²⁸ This study aimed to conduct an MR analysis to provide robust evidence of the causal influence of SES factors on prevalent spinal conditions, highlighting the practical significance of addressing SES inequalities in guiding targeted interventions, informing healthcare policies, and ultimately mitigating the burden of IVDD and low back pain on a global scale.

Methods

Overall Study Design

Figure 1 illustrates the overall workflow of the present study. This study was conducted within the framework of two-sample MR analysis and two-step mediation MR analysis. Specifically, in this MR study, SES were the exposures, obesity-related traits were the candidate mediators, and low back pain and IVDD were the outcomes. First, the GWAS summary-level statistics for five SES ([i] years of schooling, [ii] occupational attainment, [iii] Townsend deprivation index, [iv] job involves heavy manual or physical work, and [v] average total household income before tax), three obesity-related traits ([i] body mass index [BMI], [ii] waist circumference, and [iii] hip circumference), IVDD, and low back pain were downloaded from public databases. Next, a two-sample MR analysis was conducted to assess (i) the causal effect of SES on IVDD and low back pain, (ii) the causal effect of SES on obesity-related traits, and (iii) the causal effect of obesity-related traits on IVDD and low back pain. Next, various sensitivity tests were conducted to comprehensively assess the causal estimations. Finally, within the framework of a two-step MR analysis, the mediating effect of obesity-related traits on the relationship between SES and IVDD and low back pain was determined. Detailed information for each step is provided in the following subsections. This study was performed based on the “TwoSampleMR” package in R software.

Figure 1 Overall study procedure.

GWAS Summary Level Statistics

The download links for all GWAS summary statistics included in this MR study are shown in Supplementary Table S1. The GWAS summary statistics for years of schooling (n = 766,345) were obtained from the IEU Open GWAS database, derived from a large-scale GWAS meta-analysis by SSGAC consortium.²⁹ The GWAS summary statistics for occupational attainment (n = 248,847) were sourced from a GWAS study by Ko et al.³⁰ The GWAS summary statistics for Townsend deprivation index (n = 462,464), job involving heavy manual or physical work (n = 263,615), and average total household income before tax (n = 397,751) were obtained from the MRC-IEU Consortium’s curated data in the IEU OpenGWAS project database.³¹ The GWAS summary-level statistics for BMI (n = 806,834) were derived from a GWAS meta-analysis by Pulit et al.³² The GWAS summary-level statistics for waist circumference (n = 407,661) and hip circumference (n = 407,662) were obtained from the IEU Open GWAS database, derived from a large GWAS analysis conducted by Mbatchou et al.³³ The GWAS summary-level statistics for low back pain and IVDD were obtained from the FinnGen database’s R11 release.³⁴ The data for low back pain included 37,567 cases and 322,314 controls, while the data for IVDD included 46,205 cases and 322,314 controls. Definitions of all traits included in the present study are detailed in Supplementary Table S1.

Screening Genetic Instrumental Variables

First, the most relevant SNPs were screened from the GWAS summary statistics of the corresponding exposure (based on the genome-wide significance threshold P < 5e-8). Subsequently, to ensure the independence of the IVs, SNPs with linkage disequilibrium were excluded (based on a threshold of r² < 0.001 within 10,000 kb). Next, to minimize the effect of horizontal pleiotropy, SNPs potentially associated with the outcome (P < 0.05) were excluded from the IVs. In addition, palindromic SNPs were excluded from the IVs. Next, the RadialMR R package was used to further remove outlier SNPs. Finally, the F-statistics of each IV were calculated and only IVs with F-statistics > 10 were retained, thus eliminating the bias of weak genetic instruments. Supplementary Tables S2–S11 detail the IVs used for conducting MR analyses to evaluate the causal effects of five of SES on risk of IVDD and low back pain. Supplementary Tables S12–S26 detail the IVs used for conducting MR analyses to evaluate the causal effects of five of SES on three obesity-related traits. Supplementary Tables S27–S32 detail the IVs used for conducting MR analyses to evaluate the causal effects of three obesity-related traits on risk of IVDD and low back pain.

Two-Sample MR Analysis

The primary method for causal inference was Inverse-variance weighted (IVW), which integrates the final causal effect through a meta-analysis of causal inferences based on the Wald ratio method using individual IVs.³⁵ Before performing MR analysis, Cochran’s Q test was conducted to evaluate heterogeneity among causal inferences derived from different individual IVs. If no heterogeneity was found (P > 0.05), the fixed-effects model-based IVW served as the primary approach. In cases of detected heterogeneity (P < 0.05), the random-effects model-based IVW was chosen as the primary method. Additionally, four supplementary methods—MR-Egger regression, weighted median, maximum likelihood, and weighted mode—were employed alongside IVW. Causal inference was reported using odds ratio (OR) and 95% confidence interval (CI). A significant causal association was indicated if the P-value from IVW estimation was less than 0.05 and OR values from the supplementary methods aligned with IVW results.

Sensitivity Tests

The reliability of MR estimation was further evaluated through various sensitivity tests. Horizontal pleiotropy, which refers to the possibility of IVs affecting the outcome through paths other than exposure, was assessed using the MR-Egger intercept test and MR-PRESSO global test. A P-value greater than 0.05 suggested that horizontal pleiotropy did not significantly influence MR analysis. Additionally, the leave-one-out test involved repeated MR analyses after sequentially removing each IV to assess whether any outlier IVs significantly altered the results, thereby examining the stability of estimation.

Two-Step Mediation MR Analysis

A two-step mediation MR analysis further evaluated the obesity-related traits that might have mediated the causal effect of SES on IVDD and low back pain. Specifically, if an obesity-related trait was significantly influenced by a particular SES and, in turn, significantly affected IVDD or low back pain in a matching direction, then that obesity-related trait was considered a potential key mediator. The proportion of mediation effect and the corresponding 95% CI was assessed using the delta method.³⁶ Specifically, β₁ represents the causal effect of SES on the obesity-related trait (with a standard error of SE(β₁)), β₂ represents the causal effect of the obesity-related trait on IVDD/low back pain (with a standard error of SE(β₂)), β₃ represents the causal effect of the SES on IVDD/low back pain (with a standard error of SE(β₃)).;.

Reverse MR Analysis

To comprehensively understand the relationship between low back pain/IVDD and SESs, we further conducted a reverse MR analysis to assess the causal effect of low back pain/IVDD on SESs. The methodology for IV selection and the approach employed for MR analysis were consistent with those utilized in the forward MR analysis.

Results

Causal Impact of SES on IVDD and Low Back Pain

Figure 2 shows the causal effects of five types of SES on risk of IVDD and low back pain evaluated by the IVW method. Years of schooling (OR = 0.75, 95% CI: 0.69–0.82, P = 1.49E-10) were significantly negatively associated with the risk of IVDD, whereas Townsend deprivation index (OR = 1.48, 95% CI: 1.08–2.01, P = 1.36E-02) and job involving heavy manual or physical work (OR = 1.45, 95% CI: 1.06–2.00, P = 2.16E-02) were significantly positively associated with the risk of IVDD. In addition, years of schooling (OR = 0.67, 95% CI: 0.61–0.74, P = 1.95E-15), occupational attainment (OR = 0.90, 95% CI: 0.83–0.98, P = 1.82E-02), and average total household income before tax (OR = 0.80, 95% CI: 0.67–0.94, P = 7.10E-03) were significantly negatively associated with the risk of low back pain, while Townsend deprivation index (OR = 1.89, 95% CI: 1.31–2.72, P = 6.29E-04) and job involving heavy manual or physical work (OR = 1.38, 95% CI: 1.03–1.84, P = 3.19E-02) were significantly positively associated with the risk of low back pain. The direction of MR estimation determined by the four additional MR methods was consistent with IVW (Table 1). Cochran’s Q test indicated no significant heterogeneity among the IVs (P > 0.05) (Figure 2 and Supplementary Table S33). The MR-Egger intercept and MR-PRESSO global test showed that the MR estimates were not significantly influenced by pleiotropy (P > 0.05) (Figure 2 and Supplementary Table S34). Leave-one-out test results indicated that significant causal estimations were not affected by outlier IVs (Supplementary Figures S1–S8).

Table 1 Results of MR Analysis Assessed by Five Diverse MR Methods (Evaluating the Causal Effects of Socioeconomic Status on IVDD and Low Back Pain)

Figure 2 Causal effects of SES on risks of IVDD and low back pain identified by IVW method.

Identifying Obesity-Related Traits Mediating the Causal Effects of SES on IVDD and Low Back Pain

Figure 3A illustrates the causal effects of five types of SES on three obesity-related traits assessed by the IVW method, while Figure 3B illustrates the causal effects of three obesity-related traits on the risk of IVDD and low back pain assessed by the IVW method. Apart from the causal inference of years of schooling on hip circumference and the average total household income before tax on hip circumference, which were significantly affected by heterogeneity (P_{Q_ivw} < 0.05 and P_{Q_Egger} < 0.05) and horizontal pleiotropy (P_PRESSO < 0.05), the other causal inferences were not significantly disturbed by heterogeneity and horizontal pleiotropy (Figure 3A and Supplementary Table S35-S38).

Figure 3 Identification of key obesity-related traits as mediators by two-step mediation MR analysis. (A) Causal effects of SES on obesity-related traits identified by IVW method. (B) Causal effects of obesity-related traits on risks of IVDD and low back pain identified by IVW method. # P Q_ivw/P Q_Egger < 0.05, indicating that this causal association may be influenced by heterogeneity, and therefore excluded from subsequent analysis. † P PRESSO < 0.05, indicating that this causal association may be influenced by horizontal pleiotropy, and therefore excluded from subsequent analysis.

Figure 3A shows that years of schooling has a significant negative causal effect on BMI (β = −0.082, 95% CI: −0.112 - −0.053, P = 4.93E-08) and waist circumference (β = −0.076, 95% CI: −0.105 - −0.047, P = 2.29E-07). Due to the potential influence of horizontal pleiotropy on the causal effect of years of schooling on hip circumference, it is excluded from further analysis. Additionally, the average total household income before tax has a significant negative causal effect on BMI (β = −0.084, 95% CI: −0.138 - −0.030, P = 2.30E-03). Figure 3B shows that all three obesity-related traits have a positive causal association with the risk of IVDD and low back pain (OR > 1, P < 0.05). Tables 2 and 3 indicate that four additional MR methods yielded results consistent with the direction of the IVW method, further strengthening these causal estimates. Furthermore, the results of the leave-one-out test demonstrated that the significant causal estimates remained unaffected by any outlier IVs (Supplementary Figures S9–S17).

Table 2 Results of MR Analysis Assessed by Five Diverse MR Methods (Evaluating the Causal Effects of Socioeconomic Status on Obesity-Related Traits)

Table 3 Results of MR Analysis Assessed by Five Diverse MR Methods (Evaluating the Causal Effects of Obesity-Related Traits on IVDD and Low Back Pain)

Combining these results, a total of five causal associations mediated by obesity-related traits were identified: (i) Years of schooling reduces the risk of IVDD through lowering BMI (mediating proportion = 2.8% [95% CI: 1–4.6%]); (ii) Years of schooling reduces the risk of low back pain through lowering BMI (mediating proportion = 3.3% [95% CI: 1.7–4.9%]); (iii) Years of schooling reduces the risk of IVDD through lowering waist circumference (mediating proportion = 4.5% [95% CI: 1.9–7.1%]); (iv) Years of schooling reduces the risk of low back pain through lowering waist circumference (mediating proportion = 4.8% [95% CI: 2.5–7.1%]); and (v) Average total household income before tax reduces the risk of low back pain through lowering BMI (mediating proportion = 5.9% [95% CI: 1.6–10.2%]). Supplementary Table 39 presents the detailed results of the mediation MR analysis.

Results of Reverse MR Analysis

Figure 4 shows the primary reverse MR results. Interestingly, genetic liability to low back pain was significantly related to reduced average total household income before tax (β = −0.047, 95% CI: −0.085 - −0.008, P = 1.82E-02). Four supplementary MR methods further revealed the parallel findings (β < 0) (Table 4). No further causal relationship was identified by the reverse MR analysis.

Table 4 Results of Reverse MR Analysis Assessed by Five Diverse MR Methods (Evaluating the Causal Effects of IVDD/Low Back Pain on Socioeconomic Status)

Figure 4 Causal effects of IVDD/low back pain on SES identified by IVW method. # MR-PRESSO was not applicable due to number of IVs ≤ 3.

Discussion

This study comprehensively evaluated the causal effects of SES on the risks of IVDD and low back pain a two-sample MR analysis. Additionally, through a two-step mediation MR analysis, obesity-related traits mediating these causal associations were identified.

While previous studies have suggested a potential association between SES and risk of IVDD and low back pain,^18,37–40 the findings have been inconsistent. For instance, a cross-sectional study involving 2876 Chinese army soldiers reported no significant correlation between educational attainment and low back pain.⁴¹ A study of 4771 Danish twins found no clear association between adolescent socioeconomic status and low back pain, though a weak protective effect against persistent pain was statistically unclear.⁴² The inconsistencies in observational study conclusions may result from sample heterogeneity, differences in measurement methods, uncontrolled confounding factors, and variations in statistical analysis. This MR study aligned with the findings of many observational studies, providing a more robust level of evidence. Conducted with five different SES-related indicators, the analysis established a significant causal relationship between higher SES and lower risks of IVDD and low back pain, enhancing the validity of these observational correlations.

The causal pathways linking SES to IVDD and low back pain likely involve multiple interrelated mechanisms. Psychosocial stress from job insecurity, financial instability, and adverse living conditions in lower SES groups can heighten HPA axis activity, leading to chronic inflammation and increased pain sensitivity.^43,44 Occupational factors also play a role, as lower SES individuals are more likely to engage in physically demanding jobs that impose mechanical stress on the spine.⁴⁵ Limited access to healthcare further exacerbates risks by delaying early diagnosis and treatment.⁴⁶ Additionally, unhealthy lifestyle behaviors, such as smoking, physical inactivity, and substance use, are more prevalent in lower SES groups.⁴⁷ Smoking accelerates IVDD by impairing blood flow, reducing nutrient delivery, and inducing oxidative stress, which damages cellular structures and promotes inflammation.^48,49

In addition to these mechanisms, obesity emerges as a particularly important mediator in the relationship between SES and IVDD or low back pain. Previous observational studies have increasingly highlighted a strong correlation between SES and obesity,²⁰ which in turn is recognized as a significant risk factor for IVDD and low back pain.²³ Through mediation MR analyses, we identified BMI and waist circumference as key mediators of years of schooling, and BMI as a key mediator of average total household income before tax. Obesity could increase the risk of IVDD and low back pain by adding mechanical stress to the spine, accelerating disc wear and tear, and altering normal spinal biomechanics.⁵⁰ Additionally, the altered posture and movement patterns in obese individuals could further strain the lumbar region, increasing the risk of IVDD and low back pain.⁵¹ Furthermore, obesity could increase the risk of IVDD and low back pain through chronic low-grade inflammation mediated by adipokines.¹²

The implications of these findings extend beyond the medical and health sciences to inform policy and community health initiatives. It underscores the influence of SES disparities on the incidence of IVDD and low back pain. There is a demonstrated need to enhance educational opportunities, increase economic stability, and improve employment conditions to mitigate the risks associated with IVDD and low back pain. Furthermore, in regions characterized by lower SES, interventions aimed at reducing obesity metrics, particularly BMI and waist circumference, could partially counteract the adverse effects associated with low SES. However, it is crucial to note that the mediators identified only account for a minor portion of SES’s causal effect on IVDD and low back pain (2.8% to 5.9%, as shown in Supplementary Table S39). This indicates that improvements in obesity-related traits may slightly lessen the adverse effects of low SES. Additionally, SES may influence these conditions through other non-genetic factors like access to medical resources and work environment conditions, which are not detectable in GWAS and hence not included in this study.

This study had several strengths. Firstly, the two-sample MR analysis was based on GWAS summary statistics for exposure and outcomes derived from different cohorts, ensuring a large sample size. Secondly, various SES indicators were included to ensure the comprehensiveness of the MR estimation. Lastly, multiple MR methods and sensitivity tests were used to ensure the reliability of the MR analysis.

However, it must be acknowledged that this study had some limitations. Firstly, the GWAS summary statistics used were all derived from European populations, potentially limiting the applicability of the MR results to other ethnic groups. Future research should aim to validate these findings across diverse ethnic backgrounds to enhance the generalizability of the results. Additionally, the reliance on summary-level statistics meant that more detailed subgroup analyses, such as assessing the impact of SES on the risk of IVDD and low back pain in different genders and age groups, could not be performed. Expanding the dataset to include more extensive GWAS summary statistics and potentially individual-level data could provide deeper insights into these associations.

Conclusion

This MR study provided evidence for the causal effect of genetically determined SES on the risk of IVDD and low back pain and revealed the potential mediating effect of obesity-related traits.