Date of Award

January 2015

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Jonathan N. Grauer

Subject Area(s)



In the setting of the recent increase in the use of nationwide databases to study rates of adverse events in orthopaedic surgery, the purposes of this thesis are three-fold. (1) To demonstrate how nationwide databases can be optimally used to answer research questions in orthopaedics. (2) To determine how the two most commonly used databases in orthopaedic research compare in terms of documented rates of comorbidities and adverse events. (3) To provide an alternative outcome to the currently dominant composite adverse event outcomes for use in orthopaedic database research.

The following methods were used. (1) Patients undergoing intertrochanteric hip fracture surgery were identified and their perioperative outcomes were compared between treatment with intramedullary and extramedullary implants using the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database. (2) Patients undergoing lumbar fusion surgery were identified in the ACS-NSQIP and the Nationwide Inpatient Sample (NIS) and documented rates of comorbidities and adverse events were compared between databases. (3) Faculty at two large orthopaedic institutions were surveyed regarding their perceptions of severity of adverse events in orthopaedics, and their responses were used to generate a severity-weighting scheme that can be applied to orthopaedic patients in databases and used as an outcome in future database research.

Findings were as follows. (1) The rates of adverse events did not differ by implant type. Postoperative length of stay was shorter with intramedullary than extramedullary implants (5.4 versus 6.5 days; p<0.001). (2) There were large differences between databases in the rates at which specific comorbidities and adverse events were documented, with relative risks ranging from 0.25 to 8.47. (3) Of the 50 orthopaedic faculty who were invited to participate, 47 (94%) completed the exercise. Generated severity weights were highly disparate, ranging from 0.23% of death for urinary tract infection (least severe) to 15.14% of death for coma (most severe).

Conclusions are as follows. (1) The finding of increased length of stay among patients undergoing treatment with intramedullary implants may negate or reverse the excess cost perceived to be associated with intramedullary treatment. When the optimal dataset and statistical methods are used to answer appropriate research questions in orthopaedics, databases can be a powerful research tool. (2) The differences between databases are likely the result of the very different mechanisms through which the databases collect their comorbidity and adverse event data. Findings highlight concerns regarding the validity of administratively coded data. (3) The presentation of this severity-weighted outcome score fills an important gap in orthopaedic database research methodology. Future studies may benefit from using the severity-weighted outcome score presented here.


This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.