Cleveland Clinic, Cleveland, Ohio. kempenp@ccf.orgTo the Editor: A recent article presented for continuing medical education credit studied perioperative statin therapy in elective aortic surgery during 2001–2009, a period of significant advances in aortic surgery, statin use, and perioperative management.1Although propensity scoring (PS) was used to balance some variables related to the choice of statin exposure versus outcomes, validity demands that all relevant parameters be subjected to validate the methodology.2Although the authors specifically stated: “The frequency of treatment (statins) according to the year of surgery increased significantly with time” and “chronic statin therapy is used in association with other cardiovascular medications,” the year of surgery as a variable was not considered in the PS adjustment relative patient outcome.How did advances and trends toward endograft treatments (annual rates of endograft vs. open abdominal aortic aneurysm, did any patients receive endograft or were converted to open procedures?), variable aneurysm type and diameter, abdominal versus retroperitoneal approach, suprarenal cross-clamping incidence and time, surgeon experience, therapy options, blood salvage techniques, American College of Cardiology/American Heart Association guideline introduction and updates including statin and repeated revision of β-blockade recommendations (as well as Perioperative Ischemic Evaluation (POISE) study results affecting β-blockade utilization), affect any care delivered and outcomes in 2001 (before American Heart Association/American College of Cardiology guidelines) versus 2006 versus 2009! The duration and indication of preoperative statin use, time from aortic disease diagnosis to surgery, smoker versus nonsmoker (baseline carbon monoxide-hemoglobin concentration), preoperative lipid and C-reactive protein levels, as well as the frequency and quality of primary medical care before surgery, would have been additional important data to consider in applying PS. Although “all patients were screened in accordance of American College of Cardiology/American Heart Association guidelines,” the findings of Polderman's very astute Erasmus group's care raised serious questions concerning just how these guidelines are/were actually implemented, or conversely, specifically resulted in any observed increased incidence of statin-treated patients in later years.3,4The year of surgery is specifically worthy of inclusion in the study's PS analysis, given the author's own clear observation. Many other factors not analyzed would be expected to have greater effect than statin therapy itself. Are findings valid without these multiple listed considerations, yet alone those differences specifically noted but not considered by the authors?The presence of statins and other drug therapies may specifically indicate superior presurgical care states, medical and patient education advances over a period of years, or simply modern patient compliance and concern. Active patient-directed pharmaceutical advertisements are commonplace in the United States and resulted in high statin usage; is this also typical of France and Europe, leading to the very high usage today versus 2001? The 21 patient characteristics included 12 of 21 parameters, which demonstrated statistically significant differences between the groups, “equalized” by PS. What are the really important available parameters? Was PS meticulously used and how does this reflect on results, conclusions, reader's interpretations, and continuing medical education value/emphasis of the paper?“First of all, as propensity score can only remove overt (known) bias but unlike randomization it cannot be expected to remove hidden (unmeasured) bias, results from its application should be considered with caution. Interpretation will depend on the quality and amount of information about the efficacy of the treatments under evaluation.”2It may be time to introduce ongoing education regarding statistical analysis as a necessary component of the education section of this journal and a component of all continuing medical education review articles, given the important ongoing changes in statistics and importance to medical decisions and ultimately, patient care. PS should/can ultimately/only lead to randomized trials to confirm a PS suggested observation , whenever possible.5
Read full abstract