Introduction

Adolescent idiopathic scoliosis (AIS) is the most common type of pediatric spine deformity with a prevalence of up to 5% in adolescents 18 years and younger [1]. Among adolescents with AIS, approximately 10% have progressive AIS that worsen with skeletal growth, requiring treatment with bracing and sometimes invasive spine fusion surgery to prevent continued progression and disability in adulthood [2]. In the United States, approximately 38,000 pediatric patients undergo spine fusion each year for progressive scoliosis [3].

There is little known about the causes and predictors of progressive AIS. There have been few identified modifiable risk factors that have been found to be clinically relevant to prevent progression [4, 5]. Factors such as BMI, vitamin D levels and supplementation, high-risk sports participation, and posture disorders have previously been studied with no clear, strongly established relationships [6–15].

Mixed associations have been found between body mass index (BMI) and AIS in pediatric patients. Some studies have found that lower BMI is associated with scoliosis diagnosis and increased curve severity whereas others have found that overweight patients present with larger scoliosis curves and that higher adiposity is significantly related to more severe scoliosis [6–8]. Studies looking at conservative treatment outcomes also found inconsistent results; one found that overweight patients have greater curve progression despite treatment [9], one found that underweight patients had the highest risk of progressing to severe curves [10], one found that both high and low BMI were more likely to fail brace treatment than mid-BMI patients [11], and still another found no difference between overweight and normal weight patients [12]. Additionally, the biological mechanisms behind the relationship between BMI and AIS are largely unknown [16]. Previous literature includes hypothesized multifactorial mechanisms such as genetic, biomechanical, and environmental factors as well as hormonal and metabolic dysfunctions [16]. However, there remains considerable variability in the literature, and etiopathogenesis of AIS and the relationship and mechanism between BMI and progressive AIS that requires surgical intervention remain largely unknown [16].

The aim of this study was to examine the association between BMI and the incidence of posterior spine fusion among pediatric patients with AIS using a large electronic health record (EHR) dataset.

Materials and methods

Results

A total of 3,739,302 patients were assessed for inclusion. After excluding patients who did not meet the inclusion criteria or who did not have BMI data available, 2,258 patients were included (Table 1) with 2,113 patients (93.6%) who did not undergo spine fusion during the study period and 145 patients (6.4%) who did undergo spine fusion. Most patients in the total sample were female (65.7%), and an even larger proportion of patients who underwent spine fusion were female (82.1%). The mean diagnosis age among the total population was 14 years (standard deviation 2.0 years). Patients most commonly identified as either White (39.6%) or Latinx (25.9%). A majority of patients were covered by private insurance (56.6%). A small proportion of total patients had documented vitamin D supplementation (8.46%) or low vitamin D levels (9.92%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive statistics and bivariate analyses of study population baseline characteristics by spine fusion outcome.

https://doi.org/10.1371/journal.pone.0286001.t001

At baseline, 7.3% of patients were categorized as underweight, 73.2% were healthy weight, 10.2% were overweight, and 9.3% were obese. This BMI distribution seen in the total population did not differ significantly from the proportion of patients in each BMI category in the outcome groups of no spine fusion and spine fusion.

Univariable analysis

Table 2 shows that the crude logistic regression model (Model 1) demonstrated no significant overall association between any BMI category and incident posterior spine fusion (p = 0.679). Specifically, compared to those in the healthy weight group, there was no significant association between surgical intervention and being underweight (odds ratio [OR] 1.40, 95% confidence interval [CI] 0.78–2.50, p = 0.263), being overweight (OR 1.12, 95% CI 0.65–1.93, p = 0.684), or being obese (OR 0.92, 95% CI 0.50–1.70, p = 0.782).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Associations between body mass index and posterior spine fusion with crude and adjusted logistic regression models.

https://doi.org/10.1371/journal.pone.0286001.t002

Sensitivity analyses

The model including variables for the amount of time between diagnosis and BMI measurement as well as total follow-up time per patient demonstrated no notable difference in results compared to the primary adjusted analysis (Table 3, Model 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Sensitivity analyses adjusting for time between scoliosis diagnosis and first available body mass index measurement as well as follow-up time per patient.

https://doi.org/10.1371/journal.pone.0286001.t003

Limiting the crude (Table 3, Model 4) and multivariable logistic regression analyses (Table 3, Model 5) only to patients with available BMI measurements at time of first recorded scoliosis diagnosis, the total sample size was 1,127 patients with 1,037 (92.0%) who did not undergo spine fusion and 90 patients (8.0%) who did receive surgical intervention (Table 3). This demonstrated the same results as the primary analysis for both crude and adjusted logistic regression models.

Discussion

This study found that having a baseline BMI consistent with being underweight, overweight, or obese was not associated with increased odds of undergoing posterior spine fusion compared to being categorized as healthy weight among patients with AIS. This was consistent in both crude and adjusted analyses.

These findings contribute to the current mixed evidence on this topic. The results from this study are in contrast with previous studies that found that lower BMI [6], higher BMI [8, 9], or both [10, 11] were associated with scoliosis curve severity, failure of conservative brace treatment, and risk for surgery. This study’s findings are consistent with prior literature that found that participants who were overweight were not more likely to fail conservative treatment and experience curve progression [12]. This is important to consider, as previous studies have suggested that patient BMI should be taken into account when making treatment decisions [9]. Our findings instead suggest that having a BMI categorized as underweight, overweight, or obese does not have a statistically significant association with undergoing spinal fusion; therefore, BMI perhaps should not be a consideration when deciding whether or not to initiate conservative treatment for AIS. Instead, conservative treatment recommendations should be offered and supported consistent with current bracing evidence for curve magnitude and stage of skeletal maturity regardless of patient BMI at time of scoliosis diagnosis.

In fact, it may be even more important to support patients with BMI ≥ 85th percentile to be adherent to evidence-based bracing and conservative treatment given that previous studies have been consistent in their findings that overweight and obese patients have higher rates of surgical complications [22–25]. This includes intraoperative complications such as increased surgical times and blood loss and postoperative complications including longer hospital stays, surgical site infections, would complications, readmissions, and reoperations [22–25].

Strengths of this study include the large sample size of patients with AIS and the inclusion of multiple statistical models and sensitivity analyses that were consistent in their findings. There are also several limitations to this study that warrant consideration. Because EHR data was used for this study, data quality is limited to the accuracy of clinical data entry in the EHR. For example, date of scoliosis diagnosis for this study was the first date that an ICD-9 or ICD-10 code consistent with AIS was entered into this institution’s EHR; this may not be consistent with the actual first date of diagnosis if scoliosis was initially diagnosed at an outside institution or a different code was initially entered. Information on family history of scoliosis, which is considered a confounder, was not available in the EHR, so it remains possible that our associations were affected by this and other unmeasured variables. Similarly, information regarding baseline scoliosis curve magnitude and conservative treatment with bracing, which are considered mediators, were not available in the deidentified EHR data; omitting these variables resulted in estimations of total effect including any indirect effects from these variables. Future directions to follow up on this study include building models that can further study the direct effect of BMI on need for spine fusion by including important mediators such as curve magnitude at diagnosis and conservative treatment using unstructured clinic and radiology note data. Additionally, approximately 50% of the study sample did not have a BMI measurement available on the same date as the first recorded scoliosis diagnosis. However, sensitivity analyses demonstrated no change in results when adjusting for this delay or limiting the analysis to only patients with available BMI measurement on the same date as first recorded scoliosis diagnosis. This study also assumes that among the study population of patients diagnosed with AIS, if a patient required surgery, they received their fusion internally within the institution. While undergoing spine fusion is an indicator of severe scoliosis, there is oftentimes a delay between surgery indication and surgery itself. This was a single-institution study, so generalizability may be limited.

This study did not find a statistically significant association between underweight, overweight, or obese BMI category and incident posterior spine fusion among patients with AIS. These findings add to the current mixed evidence on the relationship between BMI and surgical risk and may support the recommendation of conservative treatment to patients regardless of BMI. However, given the mixed evidence on the association between BMI and progressive scoliosis, larger studies or meta-analyses are needed to better understand the relationship between BMI and potential need for spine fusion. Further research should be done to determine any potential causal effect and also mechanism of the relationship between BMI and progressive scoliosis.