This Article Abstract Full Text (PDF) Data Supplement Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ochroch, E. A. Articles by Farrar, J. T. Search for Related Content PubMed Articles by Ochroch, E. A. Articles by Farrar, J. T. Related Collections Economics and Health Care Research Pain Medicine Regional Anesthesia Pain

---

ANALGESIA

The Influence of Race and Socioeconomic Factors on Patient Acceptance of Perioperative Epidural Analgesia

Edward Andrew Ochroch, MD, MSCE^*, Andrea B. Troxel, ScD, Jonathan K. Frogel, MD, and John T. Farrar, MD, PhD

From the *Department of Anesthesiology and Critical Care, University of Pennsylvania Health System; Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania; Henry Ford Hospital, Detroit, Michigan; and Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, Pennsylvania.

Address correspondence to E. Andrew Ochroch, MD, MSCE, Department of Anesthesiology and Critical Care, 680 Dulles Building, 3400 Spruce Street, Philadelphia, PA 19104. Address e-mail to ochrocha{at}uphs.upenn.edu.

Abstract

BACKGROUND: Ethnic minorities and patients of lower socioeconomic status may be more averse to the acceptance of epidural analgesia than nonminority counterparts and those of higher socioeconomic status, despite evidence for substantial benefit to the patient.

METHODS: A scripted telephone survey was developed from the 2000 United States Census by a panel of experts. Contact was attempted at least twice for all patients listed for surgery at the Hospital of the University of Pennsylvania over a 4-mo period.

RESULTS: Three thousand seven hundred thirty-nine patients were called and 1265 subjects were successfully contacted and 1193 consented, whereas 72 refused to participate. Seven hundred sixty-two subjects (64%) would accept an epidural if recommended by an anesthesiologist and 425 (36%) would refuse. If the epidural was recommended by both the anesthesiologist and surgeon acceptance increased to 932 (78.5%). The univariate predictor of refusal of perioperative epidural analgesia was African American race. Univariate predictors of acceptance include full- or part-time employment, total household income >$50,001/yr, college graduate, prior epidural treatment, and knowledge of what an epidural is. When the potential confounders of race, total household income, employment, and education were included in a multivariate logistic regression model, African American race predicted refusal (odds ratio [OR], 0.58; P < 0.006; confidence interval [CI], 0.41–0.81) and was the only factor that predicted refusal or acceptance of epidural analgesia.

CONCLUSIONS: Acceptance of perioperative epidural analgesia is strongly affected by race and socioeconomic status. Anesthesiologists need to recognize this potential barrier when trying to maximize patient comfort and outcome.

There are clearly socioeconomic disparities in the availability and utilization of health care in the United States (US). (1,2). Although the causes of these disparities are multifactorial, two important components are patient preference and attitude. The surgical and internal medicine literature suggests that ethnic minorities and patients of lower socioeconomic status may be more averse to the acceptance of invasive interventions than nonethnic counterparts and those of higher socioeconomic status (3–5), despite evidence for substantial benefit. Although anesthesiologists typically treat patients who have already consented to surgery, racial and socioeconomic factors may still play a role in the perioperative management of patients that is independent from the decision to have surgery.

Epidural analgesia has been shown to be a more potent and effective strategy for the management of postoperative pain in thoracic, upper abdominal, and some types of cancer surgery when compared with IV opioids (6–8). Retrospective, prospective, and meta-analysis studies have demonstrated an improvement in surgical outcome through beneficial effects on perioperative pulmonary function, blunting the surgical stress response and improved analgesia. In particular, significant reduction in perioperative cardiac morbidity (approximately 30%), pulmonary infections (approximately 40%), pulmonary embolism (approximately 50%), ileus (approximately 2 days), acute renal failure (approximately 30%), and blood loss (approximately 30%) have been noted (9–11).

Despite evidence of substantial benefit, a subset of patients tend to refuse epidural analgesia for postoperative pain management. Several studies in the obstetrical literature have demonstrated that African American (AA) women and women of lower economic and educational standing are less likely to accept epidural analgesia for labor (12,13). These data, however, may not be applicable to the perioperative setting, because there may be cultural considerations and traditions that influence a woman’s choice of analgesia during childbirth but that may not play a role in planned surgery.

Perioperatively, acceptance of epidural analgesia has only been prospectively examined in the setting of enrollment into clinical trials. AAs of lower socioeconomic and educational status were significantly less likely to enroll in a study if one of the arms included epidural analgesia (14). During our own NIH funded study to examine the impact of epidural analgesia on recovery from hysterectomy, the acceptance of enrollment seemed to follow racial and socioeconomic lines. However, because of the association with a clinical trial, these data may not be applicable in a setting where the epidural placement is recommended by an anesthesiologist as part of necessary perioperative care.

Therefore, to determine if racial, educational, and financial status impacted the willingness to accept epidural analgesia as part of care for elective surgery, we conducted the following study. Our primary hypothesis was that patients with lower socioeconomic status would be less likely to accept epidural analgesia than higher socioeconomic status counterparts, despite identical recommendations by their respective anesthesiologists.

METHODS

After approval by our IRB, data were collected from August 27, 2004 to January 7, 2005 on all eligible outpatients who could be contacted and agreed to participate. Inpatients and patients younger than 18 yr were excluded from the study. Patients who did not speak English were labeled as "not contacted."

Patients scheduled for elective ambulatory or am admit surgery were asked to participate in the study. Questions of race, education, total family income (all income coming into the household via salary, investment income, government support, alimony, child support, or other recurring monies), marital status, and employment were adopted from the US 2000 Census questionnaire. Total household income was defined as income coming into the household via salary of all householders, governmental support, spousal support, investment income, pension payments, settlement payments, and recurring support from friends and family. A 19 question, 5 to 7 min, scripted interview and survey to evaluate the sources of knowledge, experience, and attitudes of epidural analgesia was developed by an expert panel of anesthesiologists, neurologists, and chronic pain physicians. After demographic data, socioeconomic data, chronic pain and anxiety history were collected, patients were asked if they knew what an epidural was and then were read a 132-word explanation (Appendix A available at www.anesthesia-analgesia.org) of an epidural and its risks. Patients were then asked if this was or was not "consistent with what they thought or knew about epidurals." The questionnaire then asked yes/no questions about sources of knowledge, experience, and attitudes regarding epidural analgesia. Finally, the subject was asked if he/she would accept an epidural if recommended by an anesthesiologist and then if recommended by both the surgeon and anesthesiologist.

Every morning at 10:30, our research coordinator received a listing of the next day’s scheduled surgeries, including names and contact phone numbers. For each listed patient, phone contact was attempted at least twice between 11 am and 7 pm to all listed phone numbers (home, work, cell). When available, phone messages were left requesting the patient to call the research coordinator. Before beginning the survey, all subjects were informed that they were in no way obligated to participate, that their responses were completely confidential, and that the survey would have no impact whatsoever on their perioperative care. All interviews were conducted by a single research coordinator.

All data were collected on preprinted TELEFORM® optically readable case report forms and scanned into a Microsoft Access database. All statistics were performed using STATA 9 (STATA Corp) and Splus version 7.0 (Insightful Corp). Categorical variables were summarized by counts and frequencies. ² tests were used to assess group differences for nominal and ordinal characteristics. Fisher’s exact test was used when cell sizes in tables were <5. Univariate analyses were performed for each variable. Logistic regression was used to determine the effect of survey elements on the acceptance of epidural analgesia when recommended by an anesthesiologist in a univariate fashion. Logistic regression models were developed in the following manner: after the univariate modeling, any variable that showed association at a significance level of 0.2 or less was sequentially added to the logistic regression model in order of the strength of its association with the outcome, and at each step the model was tested to determine if previously entered variables were still appropriately included. Confounding was assessed by inspecting odds ratios to determine if addition of a variable caused a change in magnitude of 10% or more, and confounders were retained regardless of significance. Because of the lack of a priori hypotheses concerning effect modification, interaction terms were not assessed. The modeling proceeded in two phases, the first of which considered the demographic variables (sex, race/ethnicity, marital status, education, employment, and income) and the second of which considered knowledge and prior experience. Factor variables were assessed using likelihood ratio tests comparing nested models. A two-sided significance level of 0.05 was used throughout, except as noted for the initial inclusion of variables into the multivariate models.

Primary analyses determined the impact of sex, race, educational status, and total yearly family income on the willingness of the subject to accept epidural analgesia on the recommendation of an anesthesiologist in a fully controlled multivariate logistic regression model. These factors were chosen because this part of the survey was taken from the US 2000 Census and were considered valid and reliable. Secondary analyses considered the source of a subject’s knowledge of epidural analgesia and its impact on acceptance of epidural analgesia, also using logistic regression. These factors were the section of the survey developed by the expert panel and have not been proven reliable and valid and are included for hypothesis generation and the development of future protocols. In addition, race and socioeconomic factors are more readily available to clinicians to influence their approach to a patient. Finally, a predictive model to determine which of all of the survey elements was predictive of acceptance or rejection of epidural analgesia when recommended by an anesthesiologist was developed using a step-wise logistic regression model (Table 1) (15).

We used a classification tree (classification and regression trees [CART]) (16) to confirm the results of the logistic regression procedure. After growing a large tree using binary recursive partitioning, we obtained an optimal tree of between 5 and 10 terminal nodes using split-complexity pruning based on the misclassification error rate (17).

RESULTS

Over the 4-mo period, 3739 patients were called at home the day and/or evening before their scheduled procedures. One thousand two hundred sixty-five subjects were successfully contacted and 1193 consented, whereas 72 refused to participate. Of the 1193 subjects who participated, 149 subjects specifically refused to divulge total family income and 27 additional subjects had data missing in either employment or race. Consequently, there were 1017 complete surveys. The demographics (Table 2) follow the patient mix for our hospital. Forty-five percent were men, 18% AA, and 78.5% Caucasian. Forty percent previously received epidural treatment.

The 72 patients who refused to participate did not vary from the study population by sex: 48% were men (P = 0.37). No other information was analyzable due to Health Insurance Portability and Accountability Act requirements.

Overall, 762 subjects (64%) said they would accept an epidural if it was recommended by an anesthesiologist and 425 (36%) would refuse. If the epidural was recommended by both the subject’s anesthesiologist and surgeon acceptance increased to 932 (78.5%) and 255 (22.5%) would still refuse. The rates of acceptance/refusal are significantly different (P < 0.0001) when a surgeon’s opinion is included.

Predictors of Acceptance of Epidural Analgesia: Univariate Logistic Regression Models
On univariate analysis, race/ethnicity, level of education, employment status, and income level were all significantly associated with acceptance of an epidural (Table 3), whereas sex and marital status were not. Patients with higher incomes, more education, and who were employed either full- or part-time were more likely to accept an epidural. AAs were far less likely than Caucasians to accept an epidural.

Predictor of Acceptance of Epidural Analgesia: Logistic Regression Models: Controlling for Confounding
When controlling for sex, educational level, employment type, income, and marital status (fully controlled model), AA race predicted refusal of epidural analgesia (OR, 0.53; P = 0.003; CI, 0.37–0.76). In this model, no other variable predicted acceptance or rejection of epidural analgesia.

Other Predictors of Acceptance of Epidural Analgesia: Univariate Models
There are many reasons that patients state they do not want an epidural. We chose to explore back problems, chronic pain, anxiety, previous surgery, and previous epidural as they have been previously noted to impact a patient’s acceptance of epidural analgesia (18). "Having problems with your back," "suffering from anxiety," "suffering from chronic pain that has lasted more than 3 mo," and "previous surgery in a hospital" all did not predict acceptance or refusal of epidural analgesia. Previous epidural treatment predicted acceptance of perioperative epidural analgesia (OR, 1.48; P = 0.002; CI, 1.16–1.90); among those with a previous epidural, those who received it for pain relief during childbirth were less likely to accept an epidural (OR, 0.65; P = 0.032; CI, 0.44–0.97), whereas having had one for surgery, pain relief after surgery, or treatment of back pain did not predict acceptance or rejection.

Knowledge of Epidurals
If a subject said that he or she knew what an epidural was, they were likely to accept an epidural (OR, 1.57; P = 0.001; CI, 1.20–2.06). However, the subject’s understanding of an epidural often did not agree with the standard description of an epidural provided in the survey (OR, 0.22; P < 0.001; CI, 0.16–0.30). Consistency with previous understanding was significantly associated with acceptance of epidural analgesia (Table 3). Among subjects who had not had a prior epidural, those who had "heard good things" from any source were more likely to accept an epidural (OR, 1.40; P = 0.037; CI, 1.02–1.91); those who had heard good things from a family member or friend who had an epidural were even more likely to accept an epidural (OR, 2.17; P < 0.0001; CI, 1.54–3.05). Similarly, those who had "heard bad things" from any source were less likely to accept an epidural (OR, 0.34; P < 0.0001; CI, 0.25–0.47); those who had heard bad things from a family member or friend who had an epidural were approximately equivalent (OR, 0.39; P < 0.0001; CI, 0.26–0.57). The strong impact of family member opinion was not altered by controlling for sex, race, educational, and socioeconomic status. Knowledge gained from reading (print and Internet), television, talking with doctors and nurses, and friends or relatives who told subjects what happened to a third party had no significant impact on likelihood to accept or reject epidural analgesia.

Predictive Model of Acceptance/Rejection of Epidural Analgesia
We built a multivariate logistic regression model to determine which knowledge and experience factors were predictive of acceptance or rejection of epidural analgesia when it was recommended by an anesthesiologist. When considering every factor and using a cutoff of P < 0.05 for the regression, income, having had a prior epidural, and having heard good things about epidurals from any source were all positively associated with acceptance of epidural, whereas AA race was negatively associated (Table 1).

CART Analysis
The optimal classification tree is shown in Figure 1, which gives the determination of each split, and the proportion of subjects accepting epidural along with the total number of subjects at each split and node. The first split is based on race/ethnicity; among Caucasians, 70% would accept an epidural whereas among non-Caucasians, 55% would accept an epidural. The results support those obtained via logistic regression, using the same variables of race/ethnicity, income, prior epidural, and employment status.

View larger version (9K): [in this window] [in a new window]   Figure 1. The optimal tree is shown, which gives the determination of each split, and the proportion of subjects accepting epidural along with the total number of subjects at each split and node. Initially 66% of the 1017 subjects with complete records would accept an epidural if recommended by their anesthesiologist. The first split is based on race/ethnicity; among Caucasians, 70% (790 subjects) would accept an epidural whereas among non-Caucasians, 55% (227 subjects) would accept an epidural. The results support those obtained via logistic regression, using the same variables of race/ethnicity, income, prior epidural, and additionally including employment status. Income <$50 k = total household income $50,000 US dollars per year; Income >$50 k = total household income >$50,000 US dollars per year; PT = part-time employment; FT = full-time employment; Ret = retired; HM = homemaker; UE = unemployed; Other = employment status other than the other categories; No Epi = no previous epidural treatment; Prior Epi = prior epidural treatment.

DISCUSSION

We were able to obtain survey information from 32% of potential subjects who did not differ in race or sex from our 2004 perioperative cohort. This response rate is somewhat better than typical phone interviews (19,20). In this cohort, 64% would accept an epidural when it was recommended by an anesthesiologist. Univariate predictors for the acceptance of epidural analgesia included total household income from all sources of more than $50,001. Similarly, full- and part-time employment predicted acceptance. Caucasian race strongly predicted acceptance, whereas AA race predicted refusal. Previous epidural treatment predicted acceptance. Speaking with a friend or family member who had an epidural strongly swayed the subject to accept or refuse depending on the friend’s experience: "never heard anything bad about an epidural from any source" predicted acceptance whereas having "never heard anything good about epidurals from any source" predicted rejection. Current back pain or chronic pain of more than 3-mo duration did not affect acceptance.

The multivariate logistic regression model that included race, total household income, employment, and education indicated that AA race predicted refusal of perioperative epidural analgesia and no other of these factors had a statistically significant impact. This result supports similar findings in surgical procedures (4), invasive cardiology procedures (3,5), and willingness to enroll in clinical trials (21). It remains unclear why decisions fall along racial lines. Because we did not find any confounding of the effect of race by sources of information and no effect modification between race and sources of information, we do not feel that access to/viewing articles, the Internet and television influence the acceptance of epidural analgesia differently between AA and Caucasians. Similarly, there was no differential effect of physicians and nurses being a source of information, and therefore access to health care may not be the source of the racial disparity. There may be other "cultural" sources of information that we did not investigate that cause the racial disparity, and we are concerned that this could be part of the legacy of the Tuskegee syphilis study. However, no racial bias was revealed in a study specifically looking for the presence of knowledge of the Tuskegee study and willingness to enroll in clinical research (18), and we did not specifically address this issue.

Race and socioeconomic status are important factors in determining patient outcome (1,2). For example, AA men with prostate cancer do not seek and receive care similar to Caucasian men. However, when they are given similar treatments, AA men fair just as well with prostate cancer as their Caucasian counterparts (22). Consequently, if all of our patients are going to benefit maximally from our current pain relief regimens, we will need to overcome the racial bias of AA patients against epidural analgesia.

There are several issues with regard to this survey. We chose to develop our survey instrument by first using the 2000 US Census questions. We then employed an expert panel to try to cover the range of issues we were concerned about. We did not use focus groups to determine if potential subjects interpreted our questions as we intended. The section of the instrument dealing with sources of knowledge has not undergone formal reliability and validity testing. The research assistant who performed all of the interviews reported no problems with subjects being unable to understand and answer the questions as posed.

Although we had a large number of subjects, we only were successful in contacting 34% of all potential subjects. Even though our demographics parallel those of our institution, there are potential biases. Our calling time was from 11 am to 7 pm, with multiple attempts for each potential subject. This could bias against enrolling subjects who work two consecutive jobs, have a long commute, or work nontraditional "9 to 5" h. This may be evident in the imbalance between male and female AA who participated. It is unclear in which direction this would potentially bias our results, as there is no necessary correlation between hours worked and income level. We do not feel that using phone interviews biased us against contacting people of lower socioeconomic standing, as all patients had at least one phone number listed, and the rate of disconnected numbers was negligible.

Generalizability of our results to other populations is also a potential limitation. This survey was conducted in the northeastern US and would need to be repeated in other patient populations to determine if there are regional differences.

In conclusion, our study demonstrates that there are factors not easily addressed that seem to interfere with patients agreeing to epidural analgesia, which is known to reduce pain (6–8), reduce postoperative complications (8–11), and improve recovery (8–11). Although recovery from surgery is a complex process, patients are best served by the administration of analgesia that maximizes comfort without increasing risk and delaying recovery. Further research is required to determine the best way to educate the public so that they approach their medical care without a nonmedical bias that might prevent them from getting appropriate therapies. This applies in particular to the potential benefits of epidural analgesia in the AA community.

Footnotes

Accepted for publication August 10, 2007.

This project was supported by NIH grant K23HD40914.

Reprints will not be available from the author.

This article has supplementary material on the Web site: www.anesthesia-analgesia.org.

REFERENCES

1. Keppel KG, Pearcy JN, Wagner DK. Trends in racial and ethnic-specific rates for the health status indicators: United States, 1990–1998. Healthy People 2000 Stat Notes 2002;23:1–16
2. Smedley BD, Stith AY, Nelson AR, eds. Unequal treatment: confronting racial and ethnic disparities in health care. Institute of Medicine. Washington, DC: National Academies Press; 2002
3. Gordon HS, Paterniti DA, Wray NP. Race and patient refusal of invasive cardiac procedures. J Gen Intern Med 2004;19:962–70[Web of Science][Medline]
4. Bosworth HB, Stechuchak KM, Grambow SC, Oddone EZ. Patient risk perceptions for carotid endarterectomy: which patients are strongly averse to surgery? J Vasc Surg 2004;40: 86–91[Web of Science][Medline]
5. Heidenreich PA, Shlipak MG, Geppert J, McClellan M. Racial and sex differences in refusal of coronary angiography. Am J Med 2002;113:200–7[Web of Science][Medline]
6. Jin F, Chung F. Multimodal analgesia for postoperative pain control. J Clin Anesth 2001;13:524–39[Web of Science][Medline]
7. George KA, Wright PM, Chisakuta AM, Rao NV. Thoracic epidural analgesia compared with patient controlled intravenous morphine after upper abdominal surgery. Acta Anaesthesiol Scand 1994;38:808–12[Web of Science][Medline]
8. Leon-Casasola OA, Parker BM, Lema MJ, Groth RI, Orsini-Fuentes J. epidural analgesia versus intravenous patient controlled analgesia. Differences in the postoperative course of cancer patients. Reg Anesth 1994;19:307–15[Web of Science][Medline]
9. Rawal N. Analgesia technique and post-operative morbidity. Eur J Anaesth 1995;10(suppl):47–52
10. Grass JA. The role of epidural anesthesia and analgesia in postoperative outcome. Anesthesiol Clin North America 2000;18:407–28[Medline]
11. Moraca RJ, Sheldon DG, Thirlby RC. The role of epidural anesthesia and analgesia in surgical practice. Ann Surg 2003;238:663–73[Web of Science][Medline]
12. Sheiner E, Sheiner EK, Shoham-Vardi I, Gurman GM, Press F, Mazor M, Katz M. Predictors of recommendation and acceptance of intrapartum epidural analgesia. Anesth Analg 2000;90: 109–13[Abstract/Free Full Text]
13. Hueston WJ, McClaflin RR, Mansfiled CJ, Rudy M. Factors associated with the use of intrapartum epidural analgesia. Obstet Gynecol 1994;84:579–82[Web of Science][Medline]
14. Salomons TV, Wowk AA, Fanning A, Chan VW, Katz J. Factors associated with refusal to enter a clinical trial: epidural anesthesia is a deterrent to participation. Can J Anaesth 2002;49:583–7[Web of Science][Medline]
15. Terrin N, Schmid CH, Griffith JL, D’Agostino RB, Selker HP. External validity of predictive models: a comparison of logistic regression, classification trees, and neural networks. J Clin Epidemiol 2003;56:721–9[Web of Science][Medline]
16. Breiman L, Friedman JH, Olshen RA, and Stone CJ. Classification and regression trees. Belmont CA: Wadsworth International Group; 1984
17. LeBlanc M, Crowley J. Survival trees by goodness of split. J Am Stat Assoc 1988;88:457–67
18. Gaiser RR. Epidural anesthesia. In: Longnecker DE, Tinker JH, Morgan GE. Principles and practice of anesthesiology. 2nd ed. St. Louis, MO: Mosby, 1998:1392–407
19. Miller PV, Cannell CF. Chapter III. Experimental interviewing techniques. Vital & health statistics—series 2: data evaluation & methods research. Natl Center Health Stat 1987;106:20–5
20. Frey JH. Survey research by telephone. New York, Sage, 1988
21. Brown DR, Topcu M. Willingness to participate in clinical treatment research among older African Americans and Whites. Gerontologist 2003;43:62–72[Abstract/Free Full Text]
22. Peters N, Armstrong K. Racial differences in prostate cancer treatment outcomes: a systematic review. Cancer Nurs 2005;28: 108–18[Web of Science][Medline]

This article has been cited by other articles:

				D. N. Wijeysundera, P. C. Austin, W. S. Beattie, J. E. Hux, and A. Laupacis A Population-Based Study of Anesthesia Consultation Before Major Noncardiac Surgery Arch Intern Med, March 23, 2009; 169(6): 595 - 602. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Data Supplement Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ochroch, E. A. Articles by Farrar, J. T. Search for Related Content PubMed Articles by Ochroch, E. A. Articles by Farrar, J. T. Related Collections Economics and Health Care Research Pain Medicine Regional Anesthesia Pain