Free
Correspondence  |   June 2016
In Reply
Author Notes
  • University of California, San Francisco, San Francisco, California (E.L.W.). elizabeth.whitlock@ucsf.edu
  • (Accepted for publication February 24, 2016.)
    (Accepted for publication February 24, 2016.)×
Article Information
Correspondence
Correspondence   |   June 2016
In Reply
Anesthesiology 6 2016, Vol.124, 1419. doi:10.1097/ALN.0000000000001102
Anesthesiology 6 2016, Vol.124, 1419. doi:10.1097/ALN.0000000000001102
We thank Dr. Stucke and colleagues for bringing attention to this important issue and for providing an illuminating example from their own institution. Failure to correct for multiple procedures in the same patient may bias estimates unpredictably. Unfortunately, no single-patient identifier is available through the National Anesthesia Clinical Outcomes Registry (NACOR).
After we concluded our analysis,1  NACOR began releasing more granular date-of-procedure information; the data set now includes year, month, and day of individual procedures. We merged procedure dates from the newer files with cases from the older file used in our manuscript. Of 944 cases reporting perioperative death, 50 cases were identified as occurring in the same facility on the same year, month, and day in patients with identical age and sex, implying 24 unique patients with more than 1 procedure on the same date (23 patients with 2 procedures; 1 patient with 4), who ultimately died. Reassuringly, eliminating those patients’ “second procedure” does not impact our findings in an informal post hoc sensitivity analysis we performed to answer this letter. We have not attempted to identify procedures beyond same-day cases, however, because this methodology is already quite crude.
Simple tricks like the above may be helpful in sensitivity analyses, but we emphatically agree that a unique patient number would be vastly preferable. A “hashed” identifier (anonymous, unable to be decrypted, and based on static unique patient identifiers) was proposed in detail by the Multicenter Perioperative Outcomes Group not only to identify duplicate cases, but also to link patient-level data from multiple sources.2  This could include multiple institutional databases, insurance payor files, surgical databases (e.g., American College of Surgeons’ National Surgical Quality Improvement Program), and national files (e.g., Social Security Death Index).
The Anesthesia Quality Institute is currently reorganizing its data collection structure and will reopen this question in the near future. Commentary on limitations stemming from the inability to identify multiple procedures in the same patient, like the letter from Dr. Stucke and his colleagues, serves to raise awareness of this important issue. We echo their call for careful consideration about the inclusion of an anonymized patient identifier, which we believe would further NACOR’s mission of improving the quality of anesthesia care.
Competing Interests
The authors declare no competing interests.
Elizabeth L. Whitlock, M.D., M.Sc., John R. Feiner, M.D., Lee-lynn Chen, M.D. University of California, San Francisco, San Francisco, California (E.L.W.). elizabeth.whitlock@ucsf.edu
References
Whitlock, EL, Feiner, JR, Chen, LL Perioperative mortality, 2010 to 2014: A Retrospective cohort study using the National Anesthesia Clinical Outcomes Registry.. Anesthesiology. (2015). 123 1312–21 [Article] [PubMed]
Kheterpal, S Clinical research using an information system: The multicenter perioperative outcomes group.. Anesthesiol Clin. (2011). 29 377–88 [Article] [PubMed]