| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the Center for Informatics and Perioperative Management, Department of Anesthesiology, Perioperative Medicine and Pain Management, University of Miami/Jackson Memorial Hospital, Miami, Florida.
Address correspondence and reprint requests to Michael M. Vigoda, MD, MBA, Department of Anesthesiology, Perioperative Medicine and Pain Management, 1611 NW 12th Ave. (C-300), Miami, FL 33136. Address e-mail to mvigoda{at}med.miami.edu.
Abstract
BACKGROUND: Previous anesthesia information management systems-based studies have focused on intraoperative data analysis. Reviewing preoperative data could provide insight into the outpatient treatment of patients presenting for surgical procedures. As gender-based disparities have been demonstrated in the treatment of patients with cardiac disease, we hypothesized that there would be gender disparities in the outpatient pharmacologic management of patients with coronary artery disease (CAD) scheduled for elective noncardiac surgery.
METHODS: We analyzed electronic medical records of ambulatory patients with CAD (prior myocardial infarction [MI], coronary artery bypass surgery, and angioplasty with or without stenting, angina) presenting for elective noncardiac surgery between 1/2004 and 6/2006 (30 mo) at an inner city hospital.
RESULTS: Of 21,039 ambulatory patients seen in the preanesthesia clinic, 6.4% (1346) had CAD. Patients with CAD: Men were more likely to be taking β-blockers (P < 0.002), statins (P < 0.0001), aspirin (P < 0.0001), and antiplatelet medications (P < 0.04), although there was a trend of increased use of aspirin (P < 0.01) by women over the course of the study. Patients with history of prior MI: Men with a prior MI were more likely to be taking β-blockers (P < 0.0001) and statins (P < 0.02), although there was a trend of increased use of β-blockers (P < 0.0005) and aspirin (P < 0.03) by women over the course of the study. Quarterly prevalence rates for outpatient medication use were greatest for β-blockers and least for aspirin. Patients were more likely to be taking a statin, aspirin, or oral antiplatelet medication if they were receiving chronic β-blocker therapy (P < 0.0001 for each medication).
CONCLUSION: Aggregating anesthesia management information systems data provides an epidemiological perspective of community care of patients presenting for surgery. We found that gender disparities in outpatient medical treatment of patients with CAD, which previously favored men, have diminished primarily as a result of increased use of these medications in women. Nonetheless, despite evidence supporting the use of risk-reduction strategies, our patients are undertreated with standard medical therapies.
We describe a novel use for an anesthesia information management system (AIMS) as a screening tool for a referral population.
Previous AIMS-based studies have focused on intraoperative data. Retrospective analysis is useful in determining reference limits for vital signs during anesthesia (1), as well as characterizing predictors of hypotension after induction of general anesthesia (2), difficult and impossible mask ventilation (3), and the need for antiemetic rescue treatment in the postanesthesia care unit (4). Use of real-time intraoperative data analysis can reduce deficiencies in billing documentation (5–7) and improve compliance with guidelines for prophylactic antibiotics (8,9).
Our approach, using preoperative AIMS data, was designed to test the hypothesis that there are gender differences in the medical management of patients with coronary artery disease (CAD) presenting for surgical procedures. The methodology is based on literature that demonstrated gender disparities in the acute management of myocardial infarctions (MIs) (10–14), post acute MI care (15), and the use of statin and antiplatelet therapies (16).
We retrospectively reviewed a large urban academic medical center's preoperative AIMS data to assess patterns of use of β-blockers, aspirin, statins, and antiplatelet medications in patients with CAD.
METHODS
We reviewed data stored in our preanesthesia database system (version 6.3 until May 2004 then 7.1, Picis, Wakefield MA 01880) for ambulatory patients with CAD who were evaluated at our preanesthesia clinic and underwent elective noncardiac surgery from January 2004 to June 2006. The data were gathered as part of the hospital's routine quality assurance/quality improvement process and did not contain any personal health information. Our IRB determined that such retrospective database analysis did not require their review.
Patients were considered to have CAD if they had previously had coronary artery bypass grafting (CABG), angioplasty with or without stenting, MI, or if they currently had angina of definite cardiac origin or atypical angina with a positive stress test. Data was obtained either by patient's self reporting or from existing medical records.
Each patient encounter (visit to the preanesthesia clinic) was considered to be a separate event. Preoperative data included gender, age, preexisting cardiac diagnoses, use of chronic preoperative β-blocker therapy, statins, aspirin, and antiplatelet medications.
Preoperative medications were classified according to their family as follows: β-blockers (acebutolol, atenolol, bisoprolol, carvedilol, labetalol, metoprolol, nadolol, pindolol, propranolol, sotalol, timolol); statins (atorvastatin, fluvastatin, lovastatin, pravastatin, rosuvastatin, simvastatin); aspirin, and oral antiplatelets (clopidogrel, ticlopidine).
To detect possible trends in medication use during the 30-mo period, data were also analyzed on a longitudinal basis.
Data Entry
The preanesthesia clinic is staffed by full-time Advanced Registered Nurse Practitioners who enter data while interviewing the patients. Users select from a drop-down menu, which lists the majority of commonly used medications. Users can also enter the medication name in free text (in a field termed "OTHER") in the rare event that a medication is not listed or if they do not use the drop-down menu.
Records in which a medication was listed in the OTHER field were reviewed manually for misspellings in the event that the medication could have been (but was not) selected from the drop-down menu. Where appropriate, a medication that appeared in the OTHER field was categorized in one of the drug families (i.e., β-blockers, statins, aspirin, and oral antiplatelets).
Statistical Analysis
The primary data for analysis were the difference between men and women in the rate at which the medications β-blockers, statins, aspirin, and oral antiplatelets were used as chronic therapy in patients presenting to the preanesthesia clinic. These data for all patients and by various subgroups are presented as percents.
In the major statistical analysis, observed differences were adjusted for the hypertension, diabetes, elevated cholesterol, smoking history, heart rate, and time of observation in Logistic Regression analyses. For these analyses, heart rate was dichotomized as above and below the median, whereas the time of observation was dichotomized into two equal periods. The analytic strategy was first to investigate possible gender interactions with the other variables. Main effects and interaction terms of gender with the other variables was entered into a stepwise selection procedure with gender forced into the model. If any interaction term was significant for a certain variable, for instance, diabetes, then diabetes was replaced in the model by the terms "male within diabetes" and "female within diabetes." This was done for all variables showing interaction with gender and the stepwise analysis repeated as before.
This approach is based on the well-known property in the Analysis of Deviance for Maximum Likelihood estimation in Logistic Regression that the sum of the "within" effects is algebraically identical to the sum of "main" plus "interaction" effects. The virtue of this approach is that it allows the simultaneous estimation of within-gender effects in the presence of the other adjustment variables. All models retained gender as a main effect regardless of whether there were significant gender differences, but other model terms are reported only if they were significant at P < 0.05. These analyses were repeated for subgroups of patients.
Since time period effects were found in the overall analyses, the effect of time of observation was investigated more fully by aggregating the data into calendar quarters and performing linear regression on the prevalence of β-blocker, statin, and aspirin use as a function of gender and time. The analyses are presented as slope ± se and P values overall, by gender, and gender difference for all subjects and subjects who had a prior MI. Finally, the associations among the medications were investigated by 
2 tests.
RESULTS
Between January 2004 and June 2006, there were 21,039 patient visits in which ambulatory patients were seen in the preanesthesia clinic before their scheduled elective surgery. Of this group, 1545 patients had a diagnosis of CAD (defined in Methods section). We excluded 16 patients who were <21 yr of age, 2 patients with no age (or birth date) recorded, 103 patients scheduled for cardiac surgery, and 78 patients for whom there were no surgical data.
After manually reviewing the remaining 1346 records, we found 35 cases in which the recorded gender was "unknown" and 5 in which the gender was entered incorrectly. Gender was recorded after cross-referencing data from the hospital's information system.
Overview
After excluding all encounters with missing data, there were 6.4% (or 1346) of patients with CAD, consisting of 62% (835) men, and 38% (511) women. Men were more likely to have had a previous MI, CABG, angioplasty with or without stenting, whereas women were more likely to have angina (Table 1).
|
Results of multivariate analysis that were statistically significant are displayed in Tables 2–6. We highlight those that pertain to gender differences and longitudinal changes.
|
|
|
|
|
Patients with CAD were more likely to be taking β-blockers (P < 0.03; odds ratio = 1.31) and antiplatelet medications (P < 0.003; odds ratio = 1.58) during the second half of the study period (Table 2). There was increased use of aspirin (P < 0.01; odds ratio = 1.72) in women with CAD in the second half of the study period (Fig. 1).
|
As seen in Figure 1, quarterly prevalence rates for outpatient medication use in patients with CAD were greater for β-blockers (M:80% vs F:64%) than statins (M:70% vs F:58%) than aspirin (M:57% vs F:50%).
Comparing Patients by Diagnosis and Preoperative Medication
Prior MI
During the entire study period, men with a prior MI were more likely to be taking β-blockers (P < 0.0001; odds ratio = 3.13 [1/0.32]), and statins (P < 0.02; odds ratio = 1.89 [1/0.53]).
In the second half of the study, there was increased use of antiplatelet medications (P < 0.04, odds ratio = 1.59) for the entire group and increased use of β-blockers (P < 0.0005; odds ratio = 3.44) and aspirin (P < 0.03, odds ratio = 2.13) in women (Table 3).
As seen in Figure 2, quarterly prevalence rates for outpatient medication use in patients with CAD were greater for β-blockers (M:83% vs F:93%) than statins (M:84% vs F:70%) and aspirin (M:60% vs F:68%).
|
Angina
During the entire study period, men with a history of angina were more likely to be taking statins (P < 0.03; odds ratio = 1.85 [1/0.54]). There was an increased use of antiplatelet medications in women with a history of angina (P < 0.03; odds ratio = 2.63) in the latter half of the study (Table 4).
Angioplasty With or Without Stenting
During the entire study period, women with a history of angioplasty were more likely to be taking antiplatelet medications (P < 0.005; odds ratio = 2.11) (Table 5).
CABG
For patients with a history of CABG, there was an increased use of aspirin in women (P < 0.04; odds ratio = 3.47) and in the use of antiplatelets for all patients (P < 0.006; odds ratio = 2.26) in the latter half of the study (Table 6).
Strong associations were noted between the use of β-blockers and use of statins, aspirin, and oral antiplatelet medications (Table 7).
|
DISCUSSION
If the preanesthesia clinic is a common pathway for patients presenting for surgery, then an AIMS can provide useful monitoring data on the epidemiology of community care. This retrospective study, based on 30 mo of AIMS data, describes gender-based disparities of preoperative medical therapy in ambulatory patients with CAD undergoing noncardiac surgery. We found that men were more likely to be taking β-blockers, statins, aspirin, and antiplatelet medications. Moreover, both men and women at risk for myocardial events were being undertreated.
The gender disparity of ambulatory medical management of patients with CAD is noteworthy from two perspectives. First, at the beginning of the 30-mo study period there was a pronounced difference in the use of β-blockers, statins, and aspirin. Second, for patients with CAD, this gap has been narrowing over time, with improvement primarily seen in women, perhaps the result of media attention focused on women's health issues. For patients with a prior MI, the gender difference disappeared for β-blockers, increased for statins (favoring men), and slightly increased for aspirin (favoring women).
The persistence of undertreatment throughout the study period is troubling as there were recommendations for secondary prevention of cardiac events in patients with CAD, particularly those with a prior MI, before the study period (16–18). However, others have found similar rates of noncompliance.
In an evaluation of statin therapy for secondary prevention, compliance was frequently <50% after 6–12 mo (19). One study (20) found that at 1 yr after discharge, only 50% of patients were still taking aspirin, a β-blocker, and a statin. Moreover, 12% of patients discharged with three medications (aspirin, β-blockers, and statins) discontinued using all of them within 1 mo after discharge from hospitalization for MI. Patients who discontinued use of all medications at 1 mo had lower 1-yr survival rates (88.5% vs 97.7%) compared with patients who continued to take one or more medication(s). Moreover, discontinuation of any single medication was independently associated with higher mortality. It is therefore possible that some perioperative cardiac morbidity may simply be associated with the natural history of a patient's condition and be unrelated to surgery, and better community care would impact perioperative complication rates.
With specific regard to β-blockers, Fleisher (21) noted in a recent editorial that "there are large groups of patients currently not taking β-blockers but who have Class I indications for β-blocker independent of noncardiac surgery." Our data support this assertion.
Personnel in an anesthesia preoperative clinic may be in a unique position to affect both the gender bias and the undertreatment, given the large number of at risk patients from multiple medical practices that pass through their care. A similar role has been proposed for encouraging smoking cessation in patients (20,21).
As these clinics can be very busy and multiple sources of data need to be aggregated in a short period of time, using AIMS in preanesthesia clinics may address deficiencies in medical treatment by standardizing care. As we have previously shown, the use of an electronic medical record can be a very useful analytic tool for not only administrative oversight (22), but oversight of clinical care (23,24). With proper programming, these systems may provide additional support to the anesthesia preoperative clinic care providers to inform both patient and referring physician about the need for additional long-term therapy. This is a simple process that can easily be instituted (25).
Our study was limited to ambulatory patients having elective noncardiac surgery. Inpatients were not evaluated in the preoperative clinic and our results may not reflect their experience.
Given the unique characteristics of our institution (over 50% minority representation and approximately 40% of our patient population is uninsured), we may not be able to extrapolate our results to other health care environments. Nonetheless, others have found that there is increasing evidence of treatment disparities in cardiovascular disease in the population subgroups (26) seen at our preanesthesia clinic (i.e., African-Americans, Hispanics/Mexican Americans, and persons with low socioeconomic status). The recent multiethnic study of atherosclerosis highlights the importance of race and gender relative to cardiac risk (27).
Data entry may have resulted in some errors (i.e., omission of medications, listing of incorrect medication), although our nurse practitioners have had more than 2 yr experience with the electronic medical record. In addition, patients may have forgotten to either bring or recall the medications that they were taking.
CONCLUSIONS
We identified gender disparities in the pharmacologic treatment of ambulatory patients with cardiac risk factors presenting for noncardiac surgery. Although there are gender disparities in the preoperative use of these medications (β-blocker, statins, aspirin), the disparity seems to be lessening over time, primarily due to increased use by women. The accumulation of patients from many medical practices across a community into a single anesthesia preoperative clinic may provide a means to identify those receiving suboptimal care. Implementation of decision support within the context of an anesthesia electronic medical record may assist physicians in optimizing the delivery of care.
ACKNOWLEDGMENTS
The authors thank Frank Gencorelli and Joan Leonard for their assistance with preparing and proofreading the manuscript.
Footnotes
Accepted for publication September 4, 2007.
This work was done at the University of Miami/Jackson Memorial Hospital and was not supported by any grants related directly to this project.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  K. Gelb and A. W. Gelb Sex and Gender in the Perioperative Period: Wake Up to Reality Anesth. Analg., July 1, 2008; 107(1): 1 - 3. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
