“A difference, to be a difference, must make a difference.”
In this issue of Anesthesia & Analgesia, Hausman et al.1 assessed the benefits of avoiding general anesthesia in surgical patients with chronic obstructive pulmonary disease (COPD). This retrospective, propensity-matched, cohort study examined the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) Participant Use Data File from 2005 to 2010. The cohorts consisted of surgical patients with NSQIP-defined COPD who underwent surgical procedures under either general or regional anesthesia (e.g., spinal, epidural, or peripheral nerve blockade). The authors matched 2644 pairs using propensity scores with all available demographic and comorbidity data, as well as exact matching by primary Current Procedural Terminology (CPT) code, level of dyspnea, and history of bleeding disorder. The authors reported a statistically significant higher risk for postoperative pulmonary infections in the surgical cohort undergoing general anesthesia (primary outcome), as well as prolonged ventilator dependence, unplanned postoperative intubation, and composite morbidity (secondary outcomes) compared with the cohort undergoing regional anesthesia. Thirty-day mortality (a secondary outcome) was similar for patients with COPD undergoing general or regional anesthesia. These authors concluded that regional anesthesia may result in significant improvements in morbidity outcomes in high-risk surgical populations. In a subgroup analyses by regional anesthetic type, the comparative improvement was limited to patients undergoing spinal anesthesia (the largest subgroup in the regional anesthesia cohort). Patients receiving general anesthesia did not have better mortality outcomes compared with patients receiving epidural anesthesia.
There have only been a handful of previous studies looking specifically at anesthesia type for patients with preexisting pulmonary pathology.2–4 Over 40 years ago, our predecessors at Mayo Clinic, Tarhan and researchers,2 first reported the benefits of avoiding general anesthetics in patients with moderate-to-severe chronic lung disease, advocating instead for spinal or epidural anesthesia. In 2011, van Lier et al.3 also used retrospective propensity matching in a study of 541 patients with COPD and found that epidural analgesia in combination with general anesthesia reduced postoperative pulmonary complications, including mortality, compared with general anesthesia alone. Recently, Memtsoudis et al.4–6 examined administrative claims data to provide evidence at the population level regarding whether anesthesia type affects cardiopulmonary morbidity and mortality outcomes in patients undergoing major joint replacement surgery with various comorbidities.
In the age of “big data,” researchers have access to far larger sample sizes than possible in a study at a single institution. However, large data sets are required if the authors intend to account for confounding variables using propensity scores. Even with the 1.3 million patients available in the NSQIP database through 2010, propensity matching reduced the available data to just 2644 pairs of patients. This was adequate to demonstrate the primary outcome, an effect of anesthetic type on pulmonary infection. Yet, Hausman et al.1 did not detect a statistically significant difference between the cohorts for 30-day mortality. Perhaps a larger matched sample of patients may have provided statistically significant differences between regional and general anesthesia groups within the cohort for secondary outcomes among patients with COPD. However, considering the narrow confidence interval in Hausman et al.1 for the outcome of 30-day mortality (“no difference in 30-day mortality [2.7% general versus 3.0% regional, P = 0.6788, difference = 0.3% (−1.12, 0.67)]”), any statistically significant differences resulting from larger sample sizes (more power) may not have been strong enough to represent clinically meaningful differences. Because general anesthesia is the default anesthesia type and hence more common than regional anesthesia at many institutions across the country,7 future analyses might consider a 1:2 match. Although moving from a 1:1 to a 1:2 match may result in slightly more bias between cohorts, the improvement in precision or confidence in the interval representing the range in which the “true” effect is likely to fall based on the data that may be worth the trade-off.8,9 Improvements in precision would reduce the confidence intervals surrounding estimates of differences and better show whether a particular anesthesia type is truly safer or statistically equivalent. If regional anesthesia is shown to be safer for some surgical populations, then perhaps the “default” anesthetic should be changed from general anesthesia to regional anesthesia.
Limitations within “gargantuan” studies and perioperative database research have been the prime focus of previous editorials.10,11 One limitation germaine to the use of propensity score matching, the ACS-NSQIP data set currently does not release institutional identifiers (deidentified or otherwise) in the Participant Use Data file. Thus, outcomes from hospitals where regional anesthesia is available or favored may be different than those where general anesthesia is preferred. In a recent analysis from the Anesthesia Quality Institute, Fleischut et al.7 showed that different regions of the country favor the use of different anesthesia techniques, resulting in wide variability in anesthesia care throughout the United States. Without an ability to control for these unidentified confounding variables, bias cannot be further reduced no matter what the size of the cohort sampled is.
With both administrative and clinical database types available, it is imperative for big data researchers to make informed choices and use the most reliable data set for anesthesia-related research. Unlike most administrative claims databases, ACS-NSQIP is a more trusted resource for surgical procedures and is considered the clinically based standard for evaluating surgical outcomes. Furthermore, ACS-NSQIP has been shown to be a superior database for tracking outcomes in general and vascular surgery,12,13 pediatric surgery,14 and orthopedic surgery.15 Notably, comparative validation studies have been performed confirming high reliability and comparability of the surgical data collected among the ACS-NSQIP participating hospitals.12,16 Future research may be necessary to ensure that ACS-NSQIP provides clean and correctly coded data on anesthesia procedures and anesthesia-related outcomes.
Alternatively, the trustworthiness of surgical administrative databases for accurately reporting anesthesia outcomes has led many in perioperative clinical database research to call for creation of a common coding language, “just the facts ma’am,” specific to anesthesiology. Armed with anesthesia procedures and complications recorded from an anesthesia record, perhaps anesthesia-specific databases such as the American Society of Anesthesiology sponsored Anesthesia Quality Institute17 and the Multicenter Perioperative Outcomes Groupa will become the premier sources of secondary data for our specialty.
Double-blinded randomized-controlled trials (RCTs) are considered the “gold standard” for comparative efficacy. The double-blinded RCT allows for causal inference between the study variable and the primary outcome by equally dividing both known and unknown confounding variables between groups. However, the gold standard RCT has limitations. Sometimes randomization is unethical or impossible. Sometimes an adequately powered prospective trial is prohibitively expensive or would require too long to recruit subjects to provide timely data. RCTs are also ineffective at identifying rare adverse events.
Conversely, propensity score matching is a method of comparing groups and treatments in observational studies. The propensity score methodology attempts to mimic random assignment by controlling for potential confounders through the score matching process. If the matching works as hoped, within each stratum one is comparing patients with matched confounding variables (i.e., matched scores). However, even matched cohorts will have some degree of unmeasured confounding. This greatly limits the power of propensity-matched observational trials when compared with equally sized RCTs.
However, observational databases are not equally sized to RCTs. They are far larger. That is the point. They can provide real world results not constrained by the rigid requirements of the RCT. Big data research can be analyzed for a fraction of the cost and may identify rare events invisible to the RCT. The clinical investigator exchanges the nearly bullet-proof causal inference of the RCT with the protean benefits of propensity adjusted big data analysis. This is why our specialty has recently seen a wave of well-designed large observational studies that have guided safe clinical practice for patients of various ages and comorbidities.4–6,18,19
The findings of Hausman et al.1 have the potential to influence surgeons’ and anesthesiologists’ selection of anesthesia technique for patients with COPD. Given the aforementioned evidence, the decision for surgical patients to elect for regional anesthesia may be made easier but do not breathe a sigh of relief just yet. Arguably, severity of COPD greatly influences the decision for and against general anesthesia with intubation and positive pressure ventilation over a regional technique. For example, there may be research that has yet to define the subset of patients with COPD who may worsen without the controlled ventilation present during general anesthesia, just as there may be patients who require prolonged ventilatory support after surgery as a result of the choice to induce general anesthesia. Work needs to be done comparing safety and quality outcomes of regional and general anesthesia among surgeries where the choice of the anesthetic truly matters. The conclusions provided from this study by Hausman et al.1 are a “breath of fresh air” and of sufficient clinical interest to inspire future studies designed with even larger secondary data samples and/or RCT methodology. Until this future evidence is available, we are still waiting to exhale on the selection of anesthesia type for patients with COPD.
Name: Rebecca L. Johnson, MD.
Contribution: This author helped design the paper and prepare the manuscript.
Attestation: Rebecca L. Johnson approved the final manuscript.
Name: Elizabeth B. Habermann, PhD.
Contribution: This author helped design the paper and prepare the manuscript.
Attestation: Elizabeth B. Habermann approved the final manuscript.
Name: Terese T. Horlocker, MD.
Contribution: This author helped design the paper and prepare the manuscript.
Attestation: Terese T. Horlocker approved the final manuscript.
Terese T. Horlocker is the Section Editor for Regional Anesthesia for the Journal. This manuscript was handled by Dr. Steven L. Shafer, Editor-in-Chief, and Dr. Horlocker was not involved in any way with the editorial process or decision.
a Multicenter Perioperative Outcomes Group. 2014. Available at: https://www.mpogresearch.org/. Accessed November 21, 2014.
1. Hausman MS Jr, Jewell ES, Engoren M. Regional versus general anesthesia in surgical patients with chronic obstructive pulmonary disease: does avoiding general anesthesia reduce the risk of postoperative complications? Anesth Analg. 2015;120:1405–12
2. Tarhan S, Moffitt EA, Sessler AD, Douglas WW, Taylor WF. Risk of anesthesia and surgery in patients with chronic bronchitis and chronic obstructive pulmonary disease. Surgery. 1973;74:720–6
3. van Lier F, van der Geest PJ, Hoeks SE, van Gestel YR, Hol JW, Sin DD, Stolker RJ, Poldermans D. Epidural analgesia is associated with improved health outcomes of surgical patients with chronic obstructive pulmonary disease. Anesthesiology. 2011;115:315–21
4. Memtsoudis SG, Stundner O, Rasul R, Sun X, Chiu YL, Fleischut P, Danninger T, Mazumdar M. Sleep apnea and total joint arthroplasty under various types of anesthesia: a population-based study of perioperative outcomes. Reg Anesth Pain Med. 2013;38:274–81
5. Memtsoudis SG, Sun X, Chiu YL, Stundner O, Liu SS, Banerjee S, Mazumdar M, Sharrock NE. Perioperative comparative effectiveness of anesthetic technique in orthopedic patients. Anesthesiology. 2013;118:1046–58
6. Memtsoudis SG, Rasul R, Suzuki S, Poeran J, Danninger T, Wu C, Mazumdar M, Vougioukas V. Does the impact of the type of anesthesia on outcomes differ by patient age and comorbidity burden? Reg Anesth Pain Med. 2014;39:112–9
7. Fleischut PM, Eskreis-Winkler JM, Gaber-Baylis LK, Giambrone GP, Faggiani SL, Dutton RP, Memtsoudis SG. Variability in anesthetic care for total knee arthroplasty: an analysis from the anesthesia quality institute. Am J Med Qual. 2015;30:172–9
8. Rassen JA, Shelat AA, Myers J, Glynn RJ, Rothman KJ, Schneeweiss S. One-to-many propensity score matching in cohort studies. Pharmacoepidemiol Drug Saf. 2012;21(Suppl 2):69–80
9. Austin PC. Statistical criteria for selecting the optimal number of untreated subjects matched to each treated subject when using many-to-one matching on the propensity score. Am J Epidemiol. 2010;172:1092–7
10. Fleischut PM, Mazumdar M, Memtsoudis SG. Perioperative database research: possibilities and pitfalls. Br J Anaesth. 2013;111:532–4
11. Flood P. Primary versus secondary outcomes in gargantuan studies. Anesthesiology. 2009;111:704–5
12. Cima RR, Lackore KA, Nehring SA, Cassivi SD, Donohue JH, Deschamps C, Vansuch M, Naessens JM. How best to measure surgical quality? Comparison of the Agency for Healthcare Research and Quality Patient Safety Indicators (AHRQ-PSI) and the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) postoperative adverse events at a single institution. Surgery. 2011;150:943–9
13. Lawson EH, Louie R, Zingmond DS, Brook RH, Hall BL, Han L, Rapp M, Ko CY. A comparison of clinical registry versus administrative claims data for reporting of 30-day surgical complications. Ann Surg. 2012;256:973–81
14. Polites SF, Habermann E, Zarroug AE, Wagie AE, Cima RR, Wiskerchen R, Moir CR, Ishitani MB. A comparison of two quality measurement tools in pediatric surgery—The American College of Surgeons National Surgical Quality Improvement Program-Pediatric versus the Agency for Healthcare Research and Quality Pediatric Quality Indicators. J Pediatr Surg. 2015;50:586–90
15. Sebastian AS, Habermann EB, Wagie AE, Kakar S. Current Quality Measurement Tools are Insufficient for Orthopaedic Surgery.American College of Surgeons National Surgical Quality Improvement Project (ACS- NSQIP) ConferenceJuly 2014New York, NY
16. Davis CL, Pierce JR, Henderson W, Spencer CD, Tyler C, Langberg R, Swafford J, Felan GS, Kearns MA, Booker B. Assessment of the reliability of data collected for the Department of Veterans Affairs national surgical quality improvement program. J Am Coll Surg. 2007;204:550–60
17. Hannenberg AA, Warner MA. The registry imperative. Anesthesiology. 2009;111:687–9
18. Memtsoudis SG, Besculides MC. Perioperative comparative effectiveness research. Best Pract Res Clin Anaesthesiol. 2011;25:535–47
19. Memtsoudis SG, Danninger T, Rasul R, Poeran J, Gerner P, Stundner O, Mariano ER, Mazumdar M. Inpatient falls after total knee arthroplasty: the role of anesthesia type and peripheral nerve blocks. Anesthesiology. 2014;120:551–63