Overview of longitudinal database findings

This study is the first to present a large cohort of spinal fusion surgical patients for whom complete data have been collected prospectively, although analyzed retrospectively. In the current analysis, we used large longitudinal commercial, multistate Medicare and Medicaid databases (1) to determine the true incidence of SSI, together with comorbid risk factors and (2) to determine the real-world financial burden to payers from SSIs after spinal fusion surgeries. This study differed from prior analyses because we relied on patient-level claims data and did not use surrogate data based on hospital episode statistics. Such data can seriously underestimate the incidence and payments because these data only capture inpatient episodes and hospital-based care. Also, both the CDC-NHSN and American College of Surgeons NSQIP surveillance criteria are used to benchmark infection rates at 30 days. In device-related infections, which are often biofilm mediated, microbial growth occurs over days, weeks, and months, followed by dispersion into the adjacent tissues. Under these conditions, it is not unusual to document a superficial surgical site infection beyond 30 days. ^{Reference Maron, Neifert and Ranson28} Finally, an economic analysis of this type has policy implications and is helpful in facilitating evidence-based decision making.

The use of the cohort records from the IBM MarketScan commercial and Medicaid-Medicare supplemental databases in this study highlights the payer cost of SSI following SFS to US payers. In the studied population of >200,000 patients, the risk of SSI 6 months postoperatively was 6.6%; of these, 75% were deep-incisional SSIs. The surgical approach was associated with higher risk of deep incisional SSI. Specifically, the posterior approach showed a 2-fold higher risk of deep-incisional SSI compared with the anterior approach. Also, emergent procedures similarly documented a higher risk of SSI compared with elective procedures. The costs of treating SSIs in the commercial and Medicare patient populations were substantial and continued to increase over a 24-month postoperative period of follow-up. The cost of treating deep incisional SSIs ranged from $48,662 to $93,741, whereas the cost for superficial incisional SSIs ranged from $11,044 to $32,821.

The rationale to include 12- and 24-months of follow-up is to document that the fiscal burden to payers is not limited to the first 30 days after SFS but rather continues to increase over a 24-month postoperative period. Furthermore, as noted earlier, the detection and management of device-related postoperative infection often occurs beyond the 30-day NHSN (CDC) or NSQIP (ACS) threshold period.

Comparison of database findings to earlier publications

Although the rate of superficial SSI in the present analysis aligns with some of the findings from previous publications, the overall fiscal burden shown in the current analysis is higher than previously reported. ^{Reference Horn, Segreto and Alas10,Reference Zhou, Wang, Huo, Xiong, Kang and Xue11,Reference Kuhns, Lubelski and Alvin30–Reference Tomov, Mitsunaga, Durbin-Johnson, Nallur and Roberto32} Posterior spinal-approach procedures had higher deep and superficial incisional SSI rates than anterior-approach procedures, which agrees with previous findings. ^{Reference Olsen, Mayfield and Lauryssen33} Risk factors, identified to have a significant impact on the risk of deep-incisional SSI, were also aligned with the current analysis. ^{Reference Pesenti, Pannu and Andres-Bergos8} However, the rate of deep-incisional SSI after SFS was much higher than estimates in earlier studies. ^{Reference Zhou, Wang, Huo, Xiong, Kang and Xue11} Possible explanations for this finding may be related to the 30-day postdischarge SSI surveillance time horizon reported in many of these other studies. The NSQIP database, for example, specifically utilizes a surveillance period of 30 days after discharge, whereas the current study documented SSIs up to 6 months postoperatively, a wider catchment window. For many surgical procedures, especially those involving device-related implantation, a 30-day postdischarge time horizon may be inadequate, leading to an underestimation of infection risk. ^{Reference Tanner, Kiernan, Leaper and Baggott34}

Published estimates of the payment and cost of SSI after spinal surgery have ranged from $16,242 to $37,009 per episode of care. ^{Reference Featherall, Miller and Bennett1,Reference Pennington, Sundar, Lubelski, Alvin, Benzel and Mroz5,Reference Blumberg, Woelber, Bellabarba, Bransford and Spina12} Surrogate approaches to derive the cost of SSIs have been widely used for economic analyses, which are based on hospital cost records to calculate amounts based on increased length of stay. ^{Reference Coello, Charlett, Wilson, Ward, Pearson and Borriello35,Reference Jenks, Laurent, McQuarry and Watkins36} By comparison the current study takes a payer perspective, using a large database with payment data that includes all reimbursed inpatient and outpatient claims. The current study used regression models to calculate differences in the payments between patients with and without infection, so visits that may be due to the infection but were not coded as such were identified in this analysis. In addition, a longer follow-up period (up to 24 months) suggests that healthcare protocols need to address infections that may extend beyond the conventional SSI surveillance periods (Table 3). In this manner, the amounts reimbursed by payers for services to healthcare providers would reflect the true cost of the total episode of care (surgery and follow-up). The findings of the current study emphasize the need for further research as well as implementation of robust evidence-based infection prevention surgical care bundles (and compliance with them) to mitigate the risk of infection after SFS. ^{Reference Aleem, Tan, Nassr and Riew4}

One study reported an infection prevention bundle that included 9 evidence-based components: (1) screening for Staphylococcus aureus nasal colonization and decolonization with mupirocin, (2) self-preparation bath with 4% chlorhexidine gluconate (CHG), (3) self-preparation with 2% chlorhexidine gluconate (CHG) wipes, (4) storage optimization of operating room supplies, (5) preoperative antibiotic administration algorithm, (6) staff training on betadine skin antiseptic preparation, (7) intrawound vancomycin in instrumented cases, (8) postoperative early patient mobilization, and (9) wound checks at 2 and 6 weeks postoperatively. ^{Reference Featherall, Miller and Bennett1} In total, 1,770 patients were included in the study from 2012 to 2013. Also, 40 infections were observed in the preintervention cohort, whereas 16 were observed in the intervention cohort (4.12% vs 2.00%; risk ratio, 0.48; 95% CI, 0.27–0.86; P = .01). These researchers concluded that implementation of an infection prevention bundle was associated with a 50% reduction in SSIs and an $866 per capita reduction in the surgical episode of care cost. Another cohort study that documented a significantly decreased SSI rate and associated cost reduction over a 10-year period also showed significantly decreased SSI rates and associated cost reduction after SFS with the implementation of an infection prevention bundle and increased physician awareness. ^{Reference Agarwal, Agarwal and Querry15}

This study had several limitations. The results were limited to the information captured by the IBM MarketScan databases. All information within the IBM MarketScan commercial and Medicaid-Medicare Supplemental databases are provided by individual healthcare settings and are subject to errors in incomplete hospital reporting, coding errors, or misclassification of patients. It was not possible to control for potentially important factors including physical function, socioeconomic status, clinical practice regarding postoperative wound care, and nutritional status. Additionally, due to coding limitations, important factors (eg, the number of spinal levels fused) could not be analyzed because this information only started to be captured in ICD-10-CM procedures. The exclusion of these and other potential predictive factors could impair the accuracy of the model estimates. The occurrence of SSIs was identified based on ICD-9-CM and ICD-10-CM diagnostic codes, without the availability of laboratory confirmation, although diagnosis of an SSI often reflects clinical decision. Also, some infections managed in the outpatient arena may not have been identified. Although this may be possible, recent analyses of both colorectal infections and infections following abdominal hysterectomy would suggest that the probability is low. ^{Reference Leaper, Holy and Spencer24,Reference Edmiston, Bond-Smith, Chitnis, Holy, Spencer, Chen and Leaper37} Future prospective studies are warranted to supplement the results of the current analysis.

In conclusion, the results of this study highlight the clinical and economic burden associated with SSI following SFS. The incidence and costs of an SSI found in this longitudinal study are considerably higher than those reported in studies that did not incorporate extensive postdischarge follow-up or those reported in studies that used surrogate studies of cost. Our analysis also showed that SFS SSIs often occur after the conventional 30-day postoperative surveillance period. These findings should be factored into future studies assessing the risk of postoperative infection in the spinal surgical patient population.