Physicians' Beliefs About Faith-Based Treatments for Alcoholism

Between September 2009 and June 2010, we mailed a confidential self-administered questionnaire to a stratified random sample consisting of 1,504 U.S. primary care physicians and 512 U.S. psychiatrists age 65 or younger. The sample was generated from the American Medical Association Physician Masterfile, a database intended to include all practicing U.S. physicians. To increase minority religious group representation in the primary care sample, we used validated surname lists to create four strata and oversampled from these strata (11–13). We sampled 121 primary care physicians with typical south Asian surnames, 171 primary care physicians with typical Arabic surnames, 86 primary care physicians with typical Jewish surnames, and 1,126 additional primary care physicians (from all those whose surnames were not on one of these ethnic lists). The psychiatrist sample was not sufficiently large to warrant oversampling by ethnic surname. Physicians received up to three separate mailings of the questionnaire. The first mailing included a twenty-dollar bill, and the third offered an additional $30 for participation. Data were double-keyed, cross-compared, and corrected against the original questionnaires. The study was approved by the University of Chicago Institutional Review Board.

We presented the following vignette in the survey materials: “A 47-year-old man is admitted to the hospital with acute alcohol poisoning. After a medical detox, the patient says he has been drinking heavily for years and wants to get help. He has never been hospitalized or gone through rehabilitation before.” The survey included a split-ballot experiment that varied whether the vignette concluded with “He identifies himself as a Christian but says he has not been to church in months” or with “He identifies himself as a Christian but has never been a church-goer.” Each potential respondent was randomly chosen to receive one of these two versions. Identifying the vignette patient as Christian allowed more direct comparison with other literature that focuses on Christian faith-based organizations (4,5).

Physicians were asked about the likely effectiveness of three treatments: participation in a local chapter of Alcoholics Anonymous, pharmacological therapy by a physician who specializes in the treatment of addiction, and completion of a residential rehabilitation program. Response options included very effective, somewhat effective, not very effective, and not at all effective (these responses were dichotomized for multivariable analyses as very effective or less effective).

Physicians were then asked, “To the best of your knowledge, are there any explicitly faith-based alcoholism treatment programs (not including AA) in your area to which you could potentially refer this patient?” Response options were yes or no. If respondents answered yes, they were asked, “How likely would you be to refer this patient to one of those programs?” If respondents answered no (they do not know of a faith-based program nearby), the follow-up question was, “Assuming there were, how likely would you be to refer this patient to one of those programs?” The four response options were very, somewhat, not very, and not at all likely, which were dichotomized as very likely or less likely. Finally, we asked, “To what extent do you agree with the following statement: ‘An emphasis on spirituality is critical to the success of 12-step programs.’ ” Responses were dichotomized as agree or disagree.

Primary predictors were physicians' clinical specialty (primary care versus psychiatry) and their religious characteristics. Religious affiliation was categorized by self-report as non-Evangelical Protestant, Evangelical Protestant, Catholic (Roman Catholic and Eastern Orthodox), Muslim, Jewish, Hindu, other religion (including 24 Buddhists), and none or no religious affiliation. The importance of religion was assessed by asking, “How important would you say your religion is in your own life?” Response options were dichotomized as “not very important in my life/fairly important” or “very important/the most important part of my life.”

Stratum weights for the primary care physician sample were calculated to account for oversampling from the ethnic surname strata (the design weight). We also created a poststratification adjustment weight to correct for higher response rates among U.S. medical school graduates (versus international medical school graduates) and among physicians whose roles are primarily teaching or “other” (versus office-based, hospital-based, research, or administrative roles or roles identified as unclassified because of missing data). Weights were the inverse probability that a person with the relevant characteristic would be in the final data set. The final weight for each case or respondent was the product of the design weight and the poststratification adjustment weight. This method of weighting—widely used in population-based research (14)—enabled us to adjust for sample stratification and variable response rates in order to generate estimates for the population of U.S. primary care physicians. Weights were not calculated for the psychiatrist sample because no disproportionate sampling by name strata was performed and because response rates by background variables available from the Masterfile did not differ significantly.

We used the chi square test to see if the experimental variable—whether the patient in the vignette had ever been a church-goer—was associated with any of the primary care physicians' or psychiatrists' responses. We then pooled data from both experimental arms and generated population estimates for psychiatrists' and primary care physicians' responses to each survey item measure. Next, we used multivariable logistic regression to test the primary hypotheses. Model 1 examined the effect of physician specialty, after adjustment for religious affiliation, importance of religion, sex, race, age, and region. Model 2 examined the effect of religion (affiliation or importance) after adjustment for sex, race, age, and region (primary care physicians and psychiatrists were analyzed separately). All analyses were conducted with the survey-design-adjusted commands of Stata SE statistical software, version 11.0.

More psychiatrists (64%) than primary care physicians (57%) thought the vignette patient's participation in Alcoholics Anonymous would be very effective (Table 2) (odds ratio [OR]=1.2, 95% confidence interval [CI]=1.1–1.4, model 1). Primary care physicians' beliefs did not vary with religious affiliation or the importance of religion (Table 3; model 2). Psychiatrists' beliefs showed some variation by religious affiliation. Whereas 76% of non-Evangelical Protestants (referent) believed that participation in Alcoholics Anonymous would be very effective, this belief was less common among Jewish psychiatrists (44%; OR=.3) and Catholic psychiatrists (57%; OR=.4) (Table 4; model 2).

Psychiatrists (31%) were more likely than primary care physicians (22%) to believe that pharmacological therapy by an addiction specialist would be very effective (Table 2) (OR=1.3, CI=1.1–1.6; model 1). Primary care physicians' beliefs varied with religious affiliation. Although 16% of non-Evangelical Protestants (referent) thought pharmacological therapy with a specialist would be very effective, the belief was more common among Catholics (23%, OR=1.9, CI=1.1–3.3), Muslims (44%, OR=4.6, CI=1.8–11.7), and those practicing some “other” religion (32%, OR=2.7, CI=1.2–6.5) (model 2). Psychiatrists were not analyzed by religious affiliation because of low cell counts. Their beliefs about pharmacological therapy did not vary with the importance of religion (model 2).

More psychiatrists (47%) than primary care physicians (38%) believed that a residential rehabilitation program would be very effective (Table 2) (OR=1.2, CI=1.1–1.4; model 1). Primary care physicians' and psychiatrists' beliefs about residential rehabilitation did not vary with religious affiliation or the importance of religion.

More psychiatrists (40%) than primary care physicians (29%) were aware of faith-based treatment programs nearby (Table 2) (OR=1.4, CI=1.2–1.6; model 1). Among primary care physicians, Evangelical Protestants were significantly more likely than non-Evangelical Protestants to know of a faith-based treatment program in their area (50% and 31%, respectively; OR=2.3; Table 3, model 2). Among psychiatrists, more Evangelical Protestants than non-Evangelical Protestants knew of a faith-based treatment program in their area (70% versus 39%, respectively; OR=4.1; Table 4, model 2).

Psychiatrists (29% very likely) and primary care physicians (40% very likely) did not significantly differ in their likelihood of referring individuals to faith-based programs (Table 2) (OR=.9, CI=.8–1.1; model 1). Among primary care physicians, 40% of non-Evangelical Protestants were very likely to refer the vignette patient to a faith-based program (referent). Significantly more Evangelical Protestants were very likely to refer the vignette patient (76%, OR=5.1), whereas smaller proportions of Jewish physicians (16%, OR=.4) and unaffiliated physicians (20%, OR=.4) were very likely to refer (Table 3, model 2).

Among psychiatrists, 35% of non-Evangelical Protestants were very likely to refer patients to a faith-based program. However, a smaller proportion of psychiatrists with no religious affiliation were very likely to refer patients to such programs (16%, OR=.3; Table 4, model 2).

The importance of religion also played a role. Compared with physicians who consider religion unimportant or not very important, more physicians who considered religion very important or the most important part of their lives were very likely to refer the patient to a faith-based program. This pattern was observed for primary care physicians (55% versus 28%, OR=1.8; Table 3) and psychiatrists (48% versus 19%, OR=1.9; Table 4, model 2).

Of note, physicians who knew of a faith-based program nearby were more likely than physicians who did not know of any nearby program to refer the patient to a faith-based program. This was true for primary care physicians (55% versus 32% very likely, OR=2.1, CI=1.4–3.1) and psychiatrists (41% versus 21% very likely, OR=2.9, CI=1.5–5.4) and was significant after adjustment for religious affiliation, importance of religion, sex, race, age, and region.

Most respondents agreed somewhat or strongly that an emphasis on spirituality is critical to the success of 12-step programs (85% of primary care physicians and 81% of psychiatrists) (Table 2, model 1). Among primary care physicians, 90% of non-Evangelical Protestants (referent) believed spirituality is critical. However, a smaller proportion of Jews (73%, OR=.4) and of physicians with no religious affiliation (57%, OR=.1) agreed (model 2).

Compared with those who considered religion unimportant or fairly important, physicians who considered it very important or the most important part of their lives were more likely to consider spirituality critical to 12-step program success; this was true for primary care physicians (95% versus 78%, OR=2.3; Table 3, model 2) and psychiatrists (94% versus 75%, OR=2.3; Table 4, model 2).