Study population
We conducted a retrospective cohort study within the VA healthcare system during fiscal years 2000-2012. The study protocol was reviewed and approved by the local VA Institutional Review Board of each co-author and conforms to the ethical guidelines of the 1975 Declaration of Helsinki. Because this study included only data routinely collected as part of clinical care, the Institutional Review Boards approved a waiver of informed consent.
The study population has been previously described [8]. This cohort was initially assembled to examine MACE following non-cardiac surgery in patients within 2 years of placement of a coronary stent [8]. Briefly, patients undergoing placement of coronary stents within the VA healthcare system during fiscal years 2000 through 2012 were identified using International Classification of Diseases 9th edition (ICD9) procedure codes 36.06 for bare metal stents and 36.07 for drug eluting stents in VA administrative data. Non-cardiac surgery in the 2 years following stent placement was identified through the VA Surgical Quality Improvement Program (VASQIP) registry. For each patient with a stent, 2 patients without stents undergoing non-cardiac surgery were included, matched on fiscal year of surgery, age within 5 years, work relative value unit (a measure of surgery complexity) within 5 units, surgical specialty, pre-operative creatinine >2 mg/dl, classification of high risk surgery (suprainguinal, intrathoracic, or intraperitoneal operations), and the components that make up the Revised Cardiac Risk Index (coronary artery disease, HF, stroke, chronic kidney disease, and insulin-dependent diabetes) [9]. All of the patients had a history of coronary artery disease. For the current study, we further limited the cohort to 8,823 stent/surgery patients and 17,818 matched surgery-only patients undergoing elective surgeries. Patients could be included in the cohort more than once if they had multiple surgeries. Patients with a history of MI in the 6 months prior to surgery or a history of HF or valvular disease in the past 24 months were considered a particularly high risk population. To examine the association between echocardiography and surgical volume and number of surgery cancellations, we examined a cohort of 126,773 individuals who had coronary stents placed, identified as described above.
Pre-operative echocardiography
Echocardiography within the 3 months prior to surgery was identified by Current Procedure Terminology (CPT) codes (93312, 93315, 93318, 93350, 93351, 93303, 93304, 93306, 93307, 93308, 93320, 93321, 93325) and ICD9 procedure codes (88.72) in administrative data. Because a recent echocardiogram may influence the decision to perform another echocardiogram, we also identified echocardiography in the 3 to 12 months prior to surgery.
Covariates
Covariates of interest were obtained from VASQIP (demographics, social factors, pre-operative comorbidities, and operative characteristics). Additional comorbidities were identified from the administrative data based on ICD9 diagnosis codes. For patients with surgeries in fiscal year 2005 and later, prescriptions for beta-blockers, statins, aspirin, clopidogrel, warfarin, renin-angiotensin system inhibitors, calcium channel blockers, diuretics, amiodarone, digoxin, and nitrates extending into or through the 30 days prior to surgery were identified using VA pharmacy data.
Post-operative MACE
MACE within 30 days following non-cardiac surgery was defined as an MI recorded in VASQIP or administrative data, revascularization recorded in administrative data, or death recorded in VASQIP or administrative data. This definition of MACE has been used previously in this population [8].
Statistical analysis
Patient and operative characteristics were compared by risk group and receipt of pre-operative echocardiography. Tests for differences conducted using Chi-square tests for proportions and Wilcoxon Rank Sum for continuous variables. These analyses were performed using SAS version 9.3.
For hospitals with at least 25 included surgeries, hospital-level echocardiography rates were calculated as the percentage of patients who received an echocardiogram in the 3 months prior to surgery. Of the 146 VA hospitals, we included 117 with at least 25 surgeries in the overall population, 87 with at least 25 surgeries in the population with recent MI, HF, or VHD, and 117 with at least 25 surgeries in the population without recent MI, HF, or VHD. To adjust the hospital-level echocardiography rates for case-mix, we constructed mixed-effects logistic regression models with the outcome of pre-operative echocardiography, random intercepts for the hospitals, and patient and operative characteristics as the independent variables [10]. Values of the random intercepts were calculated using empirical Bayesian estimates [11], and case-mix adjusted echocardiography rates for each hospital were calculated using the intercept for that hospital and the mean values of the patient and operative characteristics across hospitals. Observed and case-mix adjusted hospital-level echocardiography rates were plotted using a density plot. These analyses were repeated stratified by MACE risk group. Plots were developed using the R-program statistical software and GGPLOT2 package [12].
Linear regression was used to model MACE risk differences associated with a 10 % higher hospital-level rate of pre-operative echocardiography and by quartiles of hospital-level rate of pre-operative echocardiography. Models were repeated for case-mix adjusted rate of pre-operative echocardiography. The association between patient-level receipt of pre-operative echocardiography and MACE following surgery was estimated using logistic regression with a random intercept for hospital. Backwards stepwise selection was used to identify covariates from the patient and operative characteristics expected to be associated with echocardiography and/or MACE. All models were constructed using R-program statistical software and the lme4 package [13, 14].
In the cohort of patients who received a coronary stent, we calculated the hospital-level correlation between rate of pre-operative echocardiography use and surgical volume and number of surgery cancellations in the 2 years following coronary stent placement. Additionally, we examined the percentage of planned surgeries which were cancelled in individuals with echocardiography in the 3 months prior to the planned surgery and the percentage in the overall population.