Introduction

In 2002 there were 315 million people worldwide classified as obese by the World Health Organization [1]. As of 2012, 34.9 % of adults in the United States were reported to be obese [2]. Obesity is associated with disk degeneration, low back pain, and increased incidence of neurologic symptoms with concurrent lumbar disease [3–5]. Obesity is also thought to exacerbate lumbar spine disease, however, its specific impact on validated patient-reported outcomes (PROs) following lumbar surgery remains controversial [6–13]. Previous studies consist of small retrospective cohorts [10, 14] and short-term analyses utilizing administrative databases [8, 15]. A high quality post hoc analysis was performed utilizing data from the Spine Patient Outcomes Research Trial (SPORT) [16] that showed the potentially negative impact of obesity on outcomes [7, 17, 18]. However, there remains a paucity of information on the precise long-term impact of obesity on patient-reported outcome following elective surgery for degenerative lumbar conditions.

The purpose of this study was to investigate the impact of obesity on patient-reported outcomes following elective degenerative lumbar spine surgery using a prospective longitudinal registry at a high-volume academic center. We hypothesized that obesity would have no impact on PROs after surgery for degenerative lumbar conditions when controlling for clinically relevant co-variables.

Methods

Patients

A prospective longitudinal spine registry database was used to evaluate consecutive patients undergoing elective lumbar laminectomy or laminectomy with fusion for spinal stenosis, disk herniation, or spondylolisthesis from October 2010 to November 2014 at a single academic institution. Enrollment into this registry received institutional review board exemption. The overall follow-up rates for the outcomes registry at our institution is 86 % at 3 months and 78 % at 12 months. Inclusion criteria for the present analysis were: age 18 years or greater, failure of 3 months of conservative care or progressive neurological deficit, history and physical exam with correlative imaging, and a minimum follow-up of 12 months. Exclusion criteria were cases of trauma, tumor, infection, urgent/emergent surgery, deformity, and pseudarthrosis. We did not include patients undergoing an anterior lumbar fusion or microdiskectomy procedures (Table 1).

Patient-reported measures

PROs were collected in person or by telephone interview at baseline and at 3 and 12-month postoperatively. The Oswestry Disability Index (ODI, version 2.1a), Short-Form 12 (SF-12) physical and mental component scales (PCS and MCS), EuroQol-5D (EQ-5D), Numerical Rating Scale for pain (NRS), Zung Depression Scale (ZDS), and Modified Somatic Perception Questionnaire (MSPQ) measures were utilized. Each of these PRO measurement tools is validated and widely used in the spine literature [19–29]. In addition we utilized the NASS patient satisfaction scale to compare the rate of patient satisfaction between groups. This scale is scored 1 through 4 with 1 indicating the surgery met the patient’s expectations, 2 indicating improvement was less than desired, but the patient would still undergo the same procedure for the same result, 3 indicating some improvement, but dissatisfaction with the result and unwillingness to undergo the same procedure for the same result, and 4 indicating the patient feels their condition is the same or worse than prior to surgery [30]. Patient who scored a 1 or 2 on this scale were considered to have achieved the minimum clinically important difference in satisfaction.

Minimum clinically important difference (MCID) values used were: ODI 14.9 percentage points, BP 2.1 points, leg pain 2.8 points, and EQ-5D 0.46. These particular values for MCID were calculated by Parker et al. [21] in patients undergoing lumbar fusion procedures; we chose these MCID values because the majority (70 % or more) of patients in this study underwent a fusion procedure. Patients were defined as “obese” for BMI greater than or equal to 35 based on the World Health Organization (WHO) definition of class II obesity [31].

Statistical analysis

Descriptive statistics were used to examine mean scores, standard deviations, and frequency of demographic and clinical characteristics of the study population. Chi-square tests and student t test assessed the association of obesity with disability, as well as percent achieving MCID for ODI and satisfaction at 12 months. Multivariable step-wise linear regression analysis was performed with 12-month ODI scores as the outcome variable. The variables included in the analysis were age, sex, smoking status, preoperative narcotic use, preoperative unemployed, depression, anxiety, history of diabetes, revision surgery, duration of symptoms, presence or absence of neurogenic claudication, number of levels involved and arthrodesis performed. A subgroup analysis was also performed on patients with a BMI ≥40. A p value of 0.05 was considered statistically significant.

Results

Prospective registry data were available for 1868 patients undergoing lumbar procedures during the study period. Of these, 602 patients met the inclusion criteria and were therefore analyzed (Table 1). There was no significant difference in type of procedure performed in the obese versus non-obese groups, with 29 % of obese and 26 % of non-obese undergoing laminectomy alone, and 71 and 74 % undergoing laminectomy plus fusion. Six percent of the non-obese underwent non-instrumented fusion, none of the obese group had a non-instrumented fusion (Table 1).

Table 1 Demographic characteristics in obese versus non-obese (N = 602)
Full size table

All PROs improved significantly over 12 months in both BMI groups (Fig. 1, Table 2). With univariate analysis, BMI ≥35 was associated with increased ODI at baseline (50.6 vs. 47.2 %, p = 0.01) and 12 months (30.6 vs. 25.6 %, p < 0.01). Obesity was also associated with worse EQ-5D, SF-12 PCS, and SF-12 MCS at baseline. At 12 months, obesity was associated with lower SF-12 PCS, and there was no difference seen in EQ-5D (Table 3). There was no difference in pre- to post-operative change scores in ODI (21.2 vs. 19.4, p = 0.32) (Table 4). These improvements in ODI exceeded the threshold for MCID (14.9 %) and substantial clinical benefit (18.8 %) for lumbar fusion [21, 32]. In addition, there was no significant difference in percentage of patients achieving minimum clinically important difference in ODI (59.6 vs. 64 %), NASS satisfaction score of 2 or better (78.9 vs. 80.5 %), and both leg and back pain scores (Table 4). There was no difference in rate of 90-day complications or 90-day readmission between BMI groups (Table 4). With multivariate analysis, BMI ≥35 was not predictive of worse disability at 12 months as measured by ODI (correlation coefficient 1.23, 95 % confidence interval −0.225 to 2.676) (Table 5). In a subgroup analysis with groups separated into BMI <25, 25–30, 30–40, and over 40, there were no significant differences in ODI change scores between all BMI groups (22.8, 21.75, 20.11, 21.61 %, respectively, p = 0.6). Additionally there were no differences in change scores in EQ-5D, SF-12 PCS, and SF-12 MCS across WHO BMI categories (Table 6).

Fig. 1
figure 1

Graphical demonstration of improved outcomes over time in all study patients

Full size image
Table 2 Patient-reported outcomes for all study patients
Full size table
Table 3 Patient-reported outcomes in obese versus non-obese
Full size table
Table 4 Change scores in obese vs. non-obese
Full size table
Table 5 Multivariable linear regression analysis for ODI
Full size table
Table 6 Change scores across WHO BMI categories
Full size table

Discussion

In this study we utilized prospective registry data to investigate the association between obesity and patient-reported outcomes following lumbar surgery for degenerative conditions, while controlling for important co-variables. BMI ≥35 was associated with worse absolute scores in PROs and disability both at baseline and at 12-month postoperatively; however, there was no difference in change scores for PROs between BMI groups. Importantly, there was also no difference in percentage of patients achieving MCID and satisfaction at 1 year. In addition, no difference in 90-day complications or 90-day readmission was observed. These findings suggest that obese patients may achieve equally meaningful improvement in PROs compared to non-obese patients.

Previous studies have presented varying results regarding the impact of obesity on patient-reported outcomes in lumbar surgery. A study from the Swedish Spine Registry reported that higher BMI was associated with greater dissatisfaction and inferior results on ODI and EQ-5D postoperatively in patients undergoing surgery for lumbar stenosis [6]. While absolute scores were lower, there was no direct comparison of change scores in this study, thus their findings show the negative impact of obesity on absolute PROs, but do not show how BMI directly affects surgical outcomes. Similar to the findings of the present study, Djurasovic et al. reported no significant difference in PRO improvement based on BMI in patients undergoing lumbar fusion [13]. Rihn et al. analyzed data from the Spine Patient Outcomes Research Trial (SPORT) [16] and showed that BMI ≥30 was associated with less treatment effect on PROs for those undergoing surgery for disk herniation [7]. In a separate review of spinal stenosis (SpS) and degenerative spondylolisthesis (DS) patients, Rihn et al. demonstrated that a BMI ≥30 was not associated with less improvement in PROs following surgery for these conditions [17]. Reinforcing these findings, another subgroup analysis of the SPORT data showed that BMI ≥35 did not impact the treatment effect of surgery for SpS and DS; while inferior change scores were seen in those with a BMI ≥35 undergoing surgery for disk herniation [18].

The findings presented here support the hypothesis that obesity is not associated with a lesser treatment effect following surgery for degenerative lumbar conditions. To our knowledge this is the first large scale study to utilize in-depth prospective registry data to analyze the association between obesity and the treatment effect of surgery for degenerative lumbar conditions while controlling for confounding variables. Additionally, in this analysis we found significant associations between poor PROs and the following patient characteristics: Anxiety, unemployment, depression, revision surgery, and diabetes (Table 5). These findings have also been reported elsewhere in the literature [33–35]. The strength of this study is the in-depth nature of our institution’s registry data that allowed us to carefully control for those variables that are known to have an impact on outcomes in spine: narcotic use, smoking status, payer status, depression and anxiety, diabetes, and revision surgery [34–38]. Additionally, our study is unique in the relatively high number of patients with BMI ≥35 surgically treated at a single institution. There is little dispute that obesity adds complexity to surgical care; and our findings suggest that obese patients can achieve an equally meaningful improvement in surgical outcomes when cared for at a high-volume center that is equipped to manage additional complexity.

This study is not without its limitations. The PROs reported here were obtained 12 months following surgery. Certainly longer term follow-up is needed to characterize the impact of obesity on durability. However, numerous studies have revealed that the majority of patients undergoing spine surgery achieve maximal improvement by 3–6 months; thus the benefit of surgery is likely to be realized at the 12 month follow-up [39–41]. Another potential limitation is the combining of patients undergoing decompression only with those undergoing a fusion procedure. The MCID used in our study was one calculated for patients undergoing lumbar fusions; therefore, this MCID would be overly stringent on patients undergoing less invasive procedures (i.e., decompression only) and therefore underestimate the benefit to those patients. The finding that these patients achieved MCID despite the use of this more rigorous threshold may add confidence that clinically meaningful benefit was attained. Future studies are needed to better characterize the contribution of obesity to outcomes for specific diagnoses and procedures.

Conclusion

In this analysis of prospective data from patients undergoing elective lumbar surgery, BMI ≥ 35 is associated with worse baseline and 12-month PROs, however, no differences in PRO change scores were observed across BMI groups. Additionally, no difference in 90-day complications was seen between groups. After controlling for clinically important covariates, BMI ≥35 was not an independent predictor of worse PROs at 12 months. Obese patients may therefore achieve meaningful improvement following surgery for degenerative lumbar conditions.