Incidence and Risk Factors for Native Vertebral Osteomyelitis: A Retrospective Cohort Study Using a National Claims Database.

Abstract

Native vertebral osteomyelitis (NVO) presents a serious clinical challenge because of its nonspecific presentation, diagnostic delays, and associated morbidity and mortality. Existing literature on vertebral osteomyelitis (VO) emphasizes its general clinical features and mortality factors, yet critical gaps remain regarding the specific risk factors that drive NVO. To quantify the incidence and identify risk factors for NVO in the United States using a longitudinal national claims database. This was a retrospective study of patient healthcare records sourced from a national claims database. Patient records were obtained from the PearlDiver database, a comprehensive U.S. dataset encompassing patient records from governmental and commercial insurance providers, as well as the National Inpatient Sample. From this database, a random sample of 1 million patients was selected in 2010 and monitored over a 10-year period to identify new cases of VO. Patients who underwent spinal surgery within 6 months prior to infection were excluded to focus on risks of NVO. Diagnosis of NVO using International Classification of Diseases, 10^th revision (ICD-10) and International Classification of Diseases, 9^th revision (ICD-9) codes from claims data, with subsequent risk factor analysis. Demographic and clinical variables included age, gender, alcohol abuse, chronic steroid use, osteoporosis, diabetes, tobacco use, cannabis use, stimulant use, rheumatoid arthritis, chronic liver disease, and the Charlson Comorbidity Index (CCI) score. Logistic regression was performed to identify independent risk factors for NVO. Cox proportional hazards modeling was applied to estimate the impact of covariates on infection over time. The study analyzed 1 million patients in the PearlDiver database, identifying 1615 cases of VO over a 10-year period. After excluding 146 cases due to recent spinal surgery, 1469 patients were considered to have NVO. The incidence over the 10-year study period was 14.69 per 100,000 patients for 2010-2020. Patients with NVO were older than patients without NVO (61.8 years vs. 50.8 years, p<0.0001) with a slight male predominance (55.75%, n=819). Common comorbidities for patients in the NVO group included tobacco use (53.44%), diabetes (53.10%), obesity (48.47%) and chronic liver disease (24.78%). Multivariable logistic regression and Cox proportional hazards modeling were conducted. Stimulant use was strongly associated with increased odds of NVO (OR 4.03, 95% CI 3.09-5.21, p<0.0001). Additional significant risk factors included rheumatoid arthritis (OR 2.05, 95% CI 1.74-2.41, p<0.0001), diabetes (OR 1.88, 95% CI 1.69-2.11, p<0.0001), cannabis use (OR 1.72, CI 1.35-2.18, p<0.0001), obesity (OR 1.72, 95% CI 1.54-1.92, p<0.0001), tobacco use (OR 1.68, 95% CI 1.50-1.88, p<0.0001), chronic liver disease (OR 1.63, 95% CI 1.44-1.85, p<0.0001), male gender (OR 1.63, 95% CI 1.46-1.82, p<0.0001), steroid use (OR 1.56, 95% CI 1.21-1.97, p=0.0003), alcohol abuse (OR 1.52, 95% CI 1.30-1.77, p<0.0001), and osteoporosis (OR 1.22, 95% CI 1.03-1.43, p=0.020185). Cox proportional hazards modeling revealed that higher CCI scores increased the hazard of NVO by 17% (95% CI 1.06-1.28, p=0.001). Compared to previously reported rates, our findings suggest the incidence of NVO in the United States has doubled over the past decade. Notably, cannabis use emerged as a novel and significant risk factor for developing this infection. Additional risk factors, including stimulant use, steroid use, tobacco use, and diabetes, were also strongly associated with NVO. The implementation of a diagnostic algorithm could assist clinicians in the early detection of NVO, potentially reducing the morbidity and mortality associated with delayed diagnosis.