M. A. Chaudhary1, W. Jiang1, S. Lipsitz1, Z. Hashmi1, T. Koehlmoos2, P. Learn2, A. J. Schoenfeld1, A. H. Haider1 1Brigham And Women’s Hospital,Center For Surgery And Public Health,Boston, MA, USA 2Uniformed University Of Health Sciences,Bethesda, MD, USA
Introduction:
Thirty-day complications are reported by National Surgical Quality Improvement Program (NSQIP) and widely used in the surgical literature as a quality indicator for benchmarking surgical care. However, there is little evidence to suggest that the 30-day time point is the optimum length of surveillance to capture complications after surgery. The objective of this study was to determine the optimum surveillance period for complications in a national sample of high volume surgical patients.
Methods:
The TRICARE insurance database (2007-2014), with 9 million enrollees (military personnel and their dependents), was queried for adult (age 18-64 years) patients who underwent 1 of 11 high volume surgical procedures (including, general surgery, neurosurgery, orthopedic, cardiothoracic and urological procedures). Kaplan Meier (KM) curves were constructed to visualize the inflection point in the proportion of patients with a complication (wound, infectious, neurological, cardiopulmonary, vascular and genitourinary complications) at each incremental follow-up day. Multiple linear spline regression modeling, based on observed survival at each incremental follow-up day, was performed and adjusted R-squared values calculated. Optimum length of surveillance was defined as the follow-up day for which the model had the highest R-squared value. Bootstrapping (non-parametric random resampling of the data) with 300 repetitions was performed to generate a 95% confidence interval around the optimal length of surveillance estimate.
Results:
Of the 100,098 patients included in the analysis, 21.8% had at least one complication within 90 days from the day of procedure. 49% of complications were captured within the first 15 days while 74% were captured in 30 days. Visual inspection of the KM curve (Figure) exhibited a demonstrable change in slope before the 20-day mark. In the spline model, day 15(C.I.: 14-15) had the highest R-squared value (0.98), indicating an inflection point.
Conclusion:
The data demonstrates that 75% of complications occur within 30 days. However, the majority of complications (nearly 2/3rd) actually occur within 15 days after surgery. Thus, a shorter follow up period for complications may be acceptable.
