The Adult Deformity Surgery Complexity Index (ADSCI): a valid tool to quantify the complexity of posterior adult spinal deformity surgery and predict postoperative complications.

BACKGROUND CONTEXT: In 2008, Mirza et al. designed and validated the first and only index capable of quantifying the complexity of spine surgery. However, this index is not fully applicable to adult spinal deformity (ASD) surgery as it does not include the surgical techniques most commonly used and most strongly associated with perioperative complications in patients with ASD.

PURPOSE: The objective of this study is to develop an "Adult Deformity Surgery Complexity Index" (ADSCI) to quantify objectively the complexity of the ASD posterior surgery.

STUDY DESIGN/SETTING: This is an expert consensus (Delphi process) and retrospective analysis of prospectively collected data using multiple regression models.

PATIENT SAMPLE: Patients were prospectively enrolled in two comparable multicenter ASD databases sharing the same inclusion criteria.

OUTCOME MEASURES: The ADSCI was internally and externally validated using R2 for intraoperative bleeding and length of surgery. Receiver operating characteristics (ROC) and area under the curve (AUC) analysis was used to assess the sensitivity and specificity of ADSCI.

METHODS: The development and validation of ADSCI was based on the construction and comparison of three different tools: ADSCI score was constructed by three rounds of expert consensus (ADSCI-Delphi) and two multiple regression models (ADSCI-RM-Simple and ADSCI-RM-Mixed). Their predictive capability was compared by means of R2 values in the same subrogated of surgical complexity as in the Mirza index validation: intraoperative bleeding and duration of surgery. Sensitivity and specificity were evaluated using ROC curves and AUC analysis. The external validity was also examined by evaluating its predictive capability in another multicenter ASD database of comparable patients in the United States.

RESULTS: At the time of the study, the database included 1,325 patients. Four hundred seventy-five patients were eligible for the study, having been treated surgically using a posterior approach only (52.2 years standard deviation 20; 77.7% women; 85.4% American Society of Anesthesiologists I/II). Fifty-one international experts participated in the Delphi consensus process. The surgical variables selected by consensus and included in the equation were divided into actions and factors. Actions selected were number of fused segments, decompressions, interbody fusions, and cemented levels; number and type of posterior osteotomies; and use of pelvic fixation. The factors included were implant density, revision surgery, and team experience. ADSCI-RM-Mixed (regression model with Delphi formula interactions) provided the best estimates and predictive value, well above Mirza's invasiveness index. The ADSCI-RM-Mixed, with greater AUCs (>0.70), was also the most sensitive and specific for both of the dependent variables studied and for complication prediction. ADSCI-RM-Mixed obtained also the highest R2 value in the validation cohort in predicting blood loss (R2 =0.34) and surgical time (R2 =0.26) with effect sizes similar to those for the derivation cohort.

CONCLUSIONS: The ADSCI is the first tool to be specifically developed for the preoperative assessment of the complexity of ASD surgery. This study confirms its validity, specificity, and sensitivity, and shows that it has greater predictive capability than the more generic Mirza invasiveness index. The ADSCI should be useful for quantitatively estimating the increased risk associated with more invasive surgery and adjusting for surgical case-mix when making safety comparisons in ASDS.