OBJECTIVE: To determine whether curve magnitude of scoliosis at presentation correlates with BMI.
METHODS: Retrospective chart review of 180 patients presenting with scoliosis was performed. Curve pattern and magnitude, Risser status, occurrence of surgery, zip code, height and weight, race, and insurance status were recorded. Relationships were examined by Spearman rank and Pearson correlations, and logistic regression analysis was used to determine odds ratios.
RESULTS: For both thoracic and lumbar curve patterns, there was a correlation between BMI and curve magnitude. Spearman rank correlation was 0.19 for thoracic (P = .03) and 0.24 for lumbar curves (P = .02). Overweight or obese patients were not more likely, however, to present with curves at higher risk of progression or more likely to have surgical intervention. With respect to potential confounding socioeconomic variables, thoracic curve magnitude was negatively correlated with median family income (Spearman rank correlation –0.17, P = .04). Curve magnitude was not correlated with race, distance, or insurance payer.
CONCLUSIONS: Patients with high BMI and scoliosis are more likely to present with larger curves, but not more likely to require surgery. This is concerning because of the national trend of increasing childhood obesity and because scoliosis treatment may be more complicated in larger curves. Socioeconomic factors may also be barriers to access.
What’s Known on This Subject:
Early detection of scoliosis facilitates treatment. For detection, topographic features, such as truncal asymmetry or rib hump, are used.
What This Study Adds:
We show a correlation between curve magnitude at presentation and BMI. Obesity may obscure physical examination findings.
There is an epidemic of childhood obesity in the United States and other industrialized nations. In the NHANES data collected from 2003 to 2007,1 the prevalence of obesity was 31.6% nationally and 36.0% in the study population’s home state. Therefore, ∼1 in 3 patients examined by pediatricians either is obese or is considered to be at high risk for developing this health problem. Health problems associated with obesity include psychosocial consequences, earlier maturation and advanced bone age, hyperlipidemia, glucose intolerance, hepatic steatosis, and cholelithiasis.2–4 Orthopedic complications also are common in overweight and obese children and adolescents. In addition to conditions commonly thought to be related to obesity, such as adolescent tibia vara5 and slipped capital femoral epiphysis,6,7 obese children have a higher prevalence of fracture, reported musculoskeletal pain, and impaired mobility.4
Screening programs for scoliosis in children and adolescents rely on standard physical examination of surface topography, asymmetry of the Adam forward bend test, waist or shoulder asymmetry, or trunk shift. Although we found no publication evaluating the diagnostic accuracy of physical examination findings as relates to body habitus, we suspect that these findings may be obscured by obesity, making it more difficult to detect scoliosis (Fig 1). Therefore, we hypothesize that obesity (increased BMI) would be associated with increased curve magnitude at initial presentation for evaluation of scoliosis. Because socioeconomic factors may be associated with obesity and also may result in barriers to seeking care, we also hypothesized that they may correlate with curve magnitude at presentation.
After obtaining approval of the institutional review board, a retrospective chart review was performed. Medical records were identified by querying our billing database for a diagnosis code of scoliosis with an evaluation and management code for a new patient visit or new consultation. Inclusion criteria were the following: (1) an initial clinic visit at the Children’s Hospital of Alabama for evaluation of scoliosis that had been previously untreated, (2) height and weight data from the initial clinic visit, (3) a diagnosis from the attending physician of adolescent idiopathic scoliosis (AIS), and (4) radiographic measurements of Cobb angles from the initial set of spine radiographs. Curves were categorized as having their apex in the thoracic spine (T2–T11/12 disc), thoracolumbar spine (T12–L1), or lumbar spine (L1/2 disc–L4) and the magnitude in degrees of the greatest curve was noted (Fig 2). Risser status was recorded if stated in the patient record by the orthopedist or radiologist. Occurrence of surgical intervention also was recorded. Patient records were excluded if a diagnosis other than AIS was conferred (eg, congenital, neuromuscular, or infantile). A total of 180 patients, (146 girls and 34 boys) met the criteria and were included in the analysis. Of those excluded, 130 were found not to have scoliosis, 300 had scoliosis other than AIS, 71 had incomplete records, and 67 were tertiary referrals and did not contain a record of curve magnitude at initial presentation. BMI was calculated by weight (kilograms)/height (meters)2. BMI percentile was found by using the Centers for Disease Control and Prevention BMI Percentile Calculator for Child and Teen Metric Version (http://nccd.cdc.gov/dnpabmi/Calculator.aspx?CalculatorType=Metric). We then specifically stratified patients as presenting with low, moderate, or high risk of curve progression based on curve magnitude and maturity at presentation (Table 1).
We examined the effects of potential barriers to care, including distance, income, and insurance status. Distance was calculated as distance from home zip code to our facility by using an XL plug-in (CDXZipStream; CDX Technologies, Randolph, NJ). Income was approximated by assigning the median family income for the patient’s reported zip code (based on US Census data 2000). Insurance status was recorded based on billing records from initial visit and recorded as private insurance, government insurance (Medicaid), and no insurance. The number of patients with no insurance was small, and these patients were grouped with Medicaid patients for analysis.
BMI and socioeconomic variables did not have a normal distribution, so nonparametric statistics were used for analyses. Relationships between 2 continuous variables such as BMI and magnitude of curvature were examined with Spearman rank correlations. Comparisons between 2 groups, such as gender, were made with the Mann-Whitney U test.
A total of 84 patients were identified by school screening, 75 by primary care physicians, 10 each by parents and incidental finding (eg, chest radiograph), and 1 by massage therapist. Patient characteristics are summarized in Table 2. There were 121 white patients, 58 black patients, and 1 Hispanic patient. There were 136 thoracic curves, 102 lumbar curves, and 29 thoracolumbar curves with mean curve magnitude at presentation of 27.5, 24.6, and 25.2 degrees, respectively. The mean largest curve was 27.5 degrees. The mean BMI of the 180 patients included in the analysis was 20.3 (range 11.3–42.4, SD 4.4) with skewness 1.336. The mean BMI percentile was 56.3 (range <5 to >95) with an SD of 30.2 and skewness of –0.302 (Fig 3). Fifteen patients were obese (BMI percentile ≥95) and 25 were overweight (BMI percentile ≥85 and <95).
BMI was plotted against curve magnitude for the largest curve at presentation (Fig 4), and for the various curve locations (Table 3). Spearman rank correlation was 0.19 for thoracic curves (P = .03) and 0.24 for lumbar curves (P = .02). There was not a significant association between BMI and thoracolumbar curve magnitude. Based on this comparison, there is a statistically significant association between BMI and magnitude of curvature at presentation, although the linear correlation coefficient is not strong. There was no difference in the proportion of patients who presented with curves of a surgical magnitude (>45 degrees) (14% normal-weight patients versus 12% overweight and obese patients) or patients who ultimately underwent surgical intervention (17% normal weight versus 10% overweight and obese). We then evaluated the risk stratification of patients who were overweight and obese compared with normal weight. We found no statistically significant differences in the distribution of patients presenting at low, moderate, or high risk of progression (Table 4). So, although BMI correlates with curve magnitude at presentation, overweight and obese patients do not appear to be significantly more likely to present with curves at high risk of progression (ie, those who might need bracing) or with a curve of surgical magnitude.
We also evaluated whether other barriers to care may have been confounding variables that affected curve magnitude at presentation by examining socioeconomic and geographic factors. Payer status was private insurance for 118, and Medicaid, self-pay, or other government insurance for 62 patients. Mean distance to the treating facility was 39 miles (range 0.8–217.2, SD 35.9). Mean zip code family median income was $43 289 (range $10 489–$118 275, SD $16 979). There were no differences in mean curve magnitude at presentation between black and white patients or between those with private insurance and those with Medicaid or self-pay. Distance from home zip code to the referral center did not correlate with curve magnitude at presentation (Table 2, plotted in Fig 5).
Thoracic curve magnitude was negatively correlated with median family income (Spearman rank correlation –0.17, P = .04), and there was a similar trend for largest curve magnitude (Spearman rank correlation –0.13, P = .08) (Table 2, plotted in Fig 6). Finally, we evaluated whether BMI was associated with the socioeconomic factors studied, and found no differences in BMI or BMI percentile based on race, gender, or insurance status.
In the current study, we have documented an association between obesity and magnitude of scoliosis at presentation. Of interest is that this association was statistically significant for thoracic and lumbar curves but not for thoracolumbar curves or when curve magnitude for all locations of curves is considered together. One previous study from Poland also demonstrated a correlation among BMI z-score, body composition, and magnitude of scoliosis. Their number of patients was smaller, with a lower proportion of obese patients and their study group included both juvenile onset and AIS.8 Another study evaluated the body morphometric parameters of patients with and without scoliosis in a large survey of Greek school children, but did not attempt to correlate the parameters with curve magnitude. They found no significant differences between scoliotic and nonscoliotic children.9
Our study does not examine the cause for increased curve magnitude at presentation. It would seem logical that screening by physical examination is less sensitive in obese patients, but this has not been previously reported. This theory may be supported by the impact of curve location in the current study. Increased BMI was associated with greater curve magnitude for both thoracic and lumbar curves. Thoracic curves are typically most easily appreciated by the rib prominence on forward bend and lumbar curves are often noted by waist asymmetry. Both of these physical examination findings might be obscured by a thicker soft tissue envelope. In contrast, BMI and curve magnitude did not correlate with thoracolumbar curves, which are noteworthy for causing a trunk shift (where the upper torso and neck appear displaced to one side compared with the midline of the pelvis), which we postulate is less likely to be obscured by girth. To avoid possible delays in detection due to obscuration of body contours by increased BMI, scoliosis screeners may wish to emphasize elements of the physical examination less obscured by girth, such as shoulder height asymmetry or the use of a scoliometer. In addition to the traditional scoliometer, applications are now commercially available to use the inclinometer in many mobile devices, such as smart phones as scoliometers. Furthermore, scoliosis screeners could emphasize family history as a risk factor.
An alternate explanation for the increased curve magnitude at presentation would be that obesity is associated with greater magnitude of scoliosis. An interaction between adipose tissue/leptin and the autonomic nervous system has been proposed as a possible etiology of AIS.10 Another possible explanation would be that obesity is associated with an additional, confounding variable that delays presentation. To examine that possibility, we evaluated socioeconomic factors that also may present barriers to presentation, including geographic distance, insurance status, and median family income in our study population. We found a negative correlation between median family income and curve magnitude. In many studies, obesity has been associated with low income, although that association was not confirmed in the current study. One previous study addressed potential disparity in scoliosis treatment, but addressed only inpatient treatment, not initial presentation.11
Many studies have documented the relationship between curve magnitude at presentation and curve progression12,13; therefore, detection at a smaller curve magnitude is desirable. Despite an association between BMI and curve magnitude in the current study, we did not find a higher proportion of curves at high risk of progression or a higher rate of curves of a surgical magnitude. The current study was not longitudinal, and so curve progression in these patients is not known. A larger, prospective, study would be needed to determine whether overweight and obese patients are at increased risk of curve progression or surgical intervention.
Limitations of the current study include the retrospective design, which suffers from missing or possibly inaccurate data and inability to collect further information on socioeconomic or other factors that may be barriers to accessing specialized care. Multivariate analysis would be potentially illuminating to determine the relative contributions of these potential confounders, but was not possible with our data set because of the nonparametric distribution of data. Prospective evaluation could provide additional information on the sensitivity of physical examination in obese patients and detailed information regarding socioeconomic factors.
We have demonstrated that patients with scoliosis who are obese present with curves of greater magnitude. Hailing from a lower-income zip code also may be associated with higher magnitude at presentation. Providers performing screening for scoliosis may wish to consider adjuncts to the Adams forward bend test.
- Accepted March 24, 2015.
- Address correspondence to Shawn R. Gilbert, MD, Division of Orthopaedic Surgery, University of Alabama at Birmingham, Lowder Bldg, Se 316, 1600 7th Ave South, Birmingham, AL 35233. E-mail:
Dr Savage was responsible for data collection and the drafting of the initial manuscript; Dr Whitesell was responsible for data collection and analysis; Dr Conklin was responsible for reviewing data and manuscript editing; Dr Fineberg was responsible for reviewing data, performing statistical analysis, and drafting the portion of the manuscript describing analysis; and Dr Gilbert conceptionalized and designed the study, and supervised data collection, review, and editing.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- ↵Data Resource Center for Child and Adolescent Health. Childhood Obesity State Report Cards – Alabama. Available at: www.childhealthdata.org/docs/nsch-docs/alabama-pdf.pdf. Accessed July 5, 2013
- Taylor ED,
- Theim KR,
- Mirch MC,
- et al
- Burwell RG,
- Aujla RK,
- Kirby AS,
- et al
- ↵Tan KJ, Moe MM, Vaithinathan R, Wong HK. Curve progression in idiopathic scoliosis: follow-up study to skeletal maturity. Spine. 2009;34(7):697–700.
- Lonstein JE,
- Carlson JM
- Copyright © 2015 by the American Academy of Pediatrics