Attitudes towards chiropractic: a repeated cross-sectional survey of Canadian family physicians

With the assistance of epidemiologists and content experts, and reference to the previous literature [8, 10, 21‐23], we developed a 50-item, English and French-language questionnaire to examine family physicians’ attitudes towards chiropractic (Additional file 1). The final questionnaire provided response options as checkboxes, as a previous report has shown that closed-ended questions result in fewer incomplete questionnaires than open-ended formats [24].

We pre-tested our survey with three family physicians, two clinicians with both medical and chiropractic training, and two chiropractors, to evaluate if the questionnaire adequately measured attitudes towards chiropractic, and if the individual questions adequately reflected the domains of formation of attitudes, referral practices, and impressions towards chiropractic. The pre-test participants also commented on the clarity and comprehensiveness of the questionnaire and the time required for completion.

We used the 2009 Scott’s Canadian Medical Directory [25] to acquire a random sample of 1000 Canadian family physicians with a random-number generator. The Scott’s Directory contains telephone and fax numbers for physicians, but email addresses are infrequently provided. Between October and December 2010, all physicians’ offices were called to establish if they were in active practice, confirm a working fax number, and inquire if an English or French-language survey was preferred. Eligible physicians (those in active practice and for whom a working fax number was identified) were sent a survey by fax. Recipients were provided with a disclosure letter detailing the intent of the survey and explicit instructions that, should they choose not to complete the survey, they could provide this decision by fax or email to avoid further requests. Therefore, informed consent was implied if physicians provided a completed survey.

At 4 and 8 weeks following the initial survey, we re-faxed the questionnaire to all non-responders (i.e., those fax numbers from which we did not receive a completed survey) unless they indicated they did not wish to participate. We telephoned each office that received a 3rd (final) survey prior to faxing to encourage completion of the instrument, which has been shown to improve response rates [26]. Our survey was performed in accordance with the Declaration of Helsinki and approved by the Hamilton Integrated Research Ethics Board (project no. 10–305), and all methods were performed in accordance with relevant guidelines and regulations [27].

We subsequently used the 2019 Scott’s Canadian Medical Directory [22] to acquire a random sample of 2996 Canadian family physicians selected using a computer-based random number generator. From September to November 2019, we administered the same 50-item survey to physicians in this sample who were in active practice and for whom we confirmed a working fax number, in the same manner as in 2010. The Hamilton Integrated Research Ethics Board granted approval for re-administration of our survey (project no. 7355).

Members of our study team transferred information from surveys with single-key entry, as they were received, to an electronic database (SPSS) on a password-protected computer. Data was checked by a second team member for inconsistencies or unusual answers (e.g., age > 100). Once data was entered and verified, all paper surveys were shredded and disposed of.

We generated frequencies for all collected data and, for purposes of presentation, collapsed responses to individual CAQ items into agree (strongly agree + agree), undecided, and disagree (strongly disagree + disagree). We reported categorical data as proportions and continuous data as means and standard deviations (SDs) if normally distributed and as medians and interquartile ranges (IQRs) if not. To reduce the risk of spurious associations due to multiple testing, we identified any individual question within the CAQ in which the proportion of respondents who agreed or disagreed changed by ≥10% between the 2010 and 2019 survey administrations and used an independent samples Mann-Whitney U test to explore for statistical significance.

Based on previous surveys [28‐30], we hypothesized, a priori, the following associations of respondents’ attitudes towards chiropractic: (1) older physicians would hold more negative attitudes; (2) more positive attitudes if they saw a greater proportion of patients with musculoskeletal complaints; (3) physician’s endorsing patient feedback, a relationship with a specific chiropractor, personal treatment experience, or feedback from family and friends as sources of information on chiropractic would hold more positive attitudes; and (4) physician’s endorsing the scientific literature, professors, the media, or lectures during medical school as sources of information on chiropractic would hold more negative attitudes. We also hypothesized that re-administration of the survey in 2019 would show more positive attitudes versus the original administration in 2010. These variables were entered into a generalized linear model. The dependent variable, attitude towards chiropractic, was defined as the aggregate score of the CAQ. We calculated that we would require at least 110 completed surveys to ensure that our regression model was reliable (10 respondents for each independent variable considered) [31].

All comparisons were 2-tailed and an independent factor was considered statistically significant if it had a p-value < 0.05 in the final multivariable model. We report the unstandardized regression coefficient and 95% confidence interval (CI) for each variable in our regression model. The value of the unstandardized regression coefficient represents the change in response score on the 80-point CAQ. Multicollinearity was deemed concerning if the variance inflation factor for any independent variable was greater than five [32]. We performed all analyses using IBM SPSS 26.0 statistical software (Armonk, NY: IBM Corp).

In 2010, among 685 of 1000 family physicians who were in active practice and for whom we confirmed a working fax number and sent our survey, 251 returned a completed questionnaire (37% response rate; Additional file 1: Fig. 1). Among 2429 eligible family physicians identified in 2019, 162 provided a completed survey for a 7% response rate (Additional file 1: Fig. 2).

The mean age of respondents was 50 (SD 10) and 56% were men, although there was a higher prevalence of women in the more recent survey (40% in 2010 and 49% in 2019). Most respondents had been active clinically for > 20 years and worked in a community-based practice focused on general family medicine. Most physicians attended to patient populations of which > 30% presented with musculoskeletal complaints (Table 1).

Respondents endorsed multiple sources of information regarding chiropractic, but feedback from their patients was the most common. Seventy-one percent of family physicians reported referring patients for chiropractic care and, among these, most referred ≤25 patients per year. Referrals were usually prompted by patient request (57%; 237 of 413) or non-response to medical treatment (40%; 166 of 413) (Table 2).

Only 13% of physicians (53 of 413) worked in a multidisciplinary environment where chiropractic care was available, and 40% (165 of 413) had sought chiropractic care for themselves. Most had not received information on chiropractic during their medical training, and the majority (80%) felt their education should (52%; 214 of 413) or possibly should (28%; 115 of 413) include such information. Most respondents’ opinions on chiropractic were formed after medical school (82%; 337 of 413), and most (51%; 209 of 413) described themselves as a little knowledgeable. In 2010, most respondents (52%) felt that adverse events were uncommon with chiropractic care, and in 2019 most physicians believed that adverse events were common but serious events were rare (47%). In 2010, most respondents (46%) were very comfortable discussing chiropractic with their patients, whereas in 2019 most (41%) were only somewhat comfortable (Additional file 1: Table 1).

Fifteen percent (62 of 413) of physicians felt that chiropractic care should be available in multidisciplinary settings (29% were unsure), and 25% felt that chiropractic should be available in hospitals, either with (17%; 69 of 413) or without (8%; 34 of 413) physician referral. Respondents varied on whether chiropractic care should be offset by government funding: 35% agreed, 33% were unsure, and 27% disagreed. Forty-three percent of family physicians definitely (17%) or somewhat (26%) perceived chiropractors as primary care providers, and most (81%; 335 of 413) wanted consultation notes from chiropractors who attended their patients. Seventy-five percent of respondents had received requests from chiropractors to refer patients for imaging studies. Most physicians (59%; 245 of 413) believed that practice diversity within the chiropractic profession was a barrier to interprofessional collaboration. (Additional file 1: Table 2).

Forty-eight percent of family physicians (198 of 413) endorsed a positive impression of chiropractic, 27% were unsure, and 25% held negative views. Respondents endorsing a positive view had an average CAQ score of 50.2 out of 80 (SD 7.5), undecided respondents had an average CAQ score of 39.4 (SD 5.8), and physicians with negative impressions had a mean CAQ score of 24.9 (SD 9.1). An important change in continuous outcome measures can be estimated as half a SD of the aggregate score for a given population [33], and by this standard, a 6-point difference on the CAQ would be considered meaningful.

Responses to individual items on the CAQ are provided in Table 3. Most physicians felt that chiropractors provide effective management for some musculoskeletal disorders (84%), that chiropractic was a useful supplement to medical care (65%), and chiropractors could reduce patient overload for family physicians (52%). Many physicians endorsed that chiropractors provide a patient-centred approach (45%) and use approaches from which medicine could benefit (43%). Alternately, most respondents disagreed that chiropractic was effective for non-musculoskeletal conditions (77%) and were unsure whether chiropractors treat in accordance with evidence-based practices (52%). Many felt that chiropractic manipulation of the neck was unsafe (47%) and 37% agreed that chiropractors provide patients with misinformation regarding vaccination.

There were 5 items on the CAQ in which the proportion of respondents who agreed or disagreed shifted by ≥10% between administrations in 2010 and 2019, of which three were statistically significant. Canadian family physicians surveyed in 2019 were more likely to: (1) disagree that chiropractors promote unnecessary treatment plans (28% in 2019 vs. 18% in 2010; p < 0.001), (2) agree that chiropractors provide effective care for post-surgical rehabilitation (24% vs. 14%; p = 0.05), and (3) agree that chiropractors treat in accordance with evidence-based practices (26% vs. 14%; p = 0.05).

In our adjusted regression model, overall impressions towards chiropractic showed trivial improvement from 2010 to 2019 (0.31 points on the 80-point CAQ; 95%CI 0.001 to 0.62). More negative attitudes were associated with older age (− 1.55 points for each incremental decade from age 28; 95%CI − 2.67 to − 0.44), belief that adverse events are common with chiropractic care (− 1.41 points; 95% CI − 2.59 to − 0.23) and reported use of the research literature (− 6.04 points; 95% CI − 8.47 to − 3.61) or medical school (− 5.03 points; 95% CI  − 7.89 to − 2.18) as a source of knowledge on chiropractic. More positive attitudes were associated with endorsing a relationship with a specific chiropractor (5.24 points; 95% CI 2.85 to 7.64), family and friends (4.06 points; 95% CI 1.53 to 6.60), or personal treatment experience (4.63 points; 95% CI 2.14 to 7.11) as sources of information regarding chiropractic. (Table 4) The variance inflation factor was less than 2 for each independent variable, suggesting no issues with multicollinearity. Our model explained approximately 26% of the variation (adjusted R² = 0.26) in family physician’s attitudes toward chiropractic.

Strengths of our study include random sampling of all Canadian family physicians, and survey design and administration consistent with best practices [27]. Our assessment of attitudes towards chiropractic was based on the CAQ, which has been validated among other groups of Canadian healthcare providers [28‐30]. Our study does have limitations, including an overall response rate of 13%, which was lower for the re-administration of the survey. Non-responders may have differed systematically from responders, and the generalizability of our findings to family physicians practicing outside of Canada is uncertain. Our model explained 26% of the variation in respondent’s attitudes toward chiropractic, indicating that there remain additional variables of importance that our survey did not capture.

In August 2018, the Canadian Chiropractic Association (CCA) published a statement emphasizing a focus on promoting interprofessional collaboration [34], and the CCA advocates for integration of chiropractors into interprofessional health teams [35]. We found that although most Canadian family physicians endorse chiropractic care as a useful supplement to conventional medicine, only one in eight physicians reported working with a chiropractor, and practice diversity within the chiropractic profession was perceived as a barrier to interprofessional collaboration.

Most family physicians disagreed that chiropractic care was effective for non-musculoskeletal conditions, and systematic reviews on this topic have not found evidence to challenge this assertion [36‐39]. Most respondents agreed that chiropractic care is effective for certain musculoskeletal complaints, and spinal manipulation, which is commonly provided by chiropractors, has received support for management of axial complaints from recent systematic reviews [40‐45] and clinical practice guidelines [46‐48]. Paradoxically, support from the scientific literature was a common reason given for referring patients for chiropractic care, while reliance on research literature for information on chiropractic was associated with more negative attitudes. Reasons for this disconnect are uncertain.

Close to half of respondents disagreed that chiropractic manipulation of the cervical spine was generally safe; however, although some observational studies have suggested a rare association with stroke [49‐51], studies with greater methodologic safeguards against bias have failed to confirm either an association between greater utilization of chiropractic and risk of stroke [52], or an association between chiropractic care and an increased risk of stroke compared to care by primary care physicians [53, 54]. The associations reported in some studies between chiropractic care and stroke may be due to patients with early dissection-related symptoms seeking care prior to developing their strokes [55‐57].

Musculoskeletal complaints, particularly low back pain, are common in primary care [58]. Our findings suggest that most Canadian family physicians believe chiropractors can provide effective care for some musculoskeletal complaints; however, many physicians are uncertain whether chiropractors treat in accordance with evidence-based practices and have concerns regarding the safety of cervical manipulation. The chiropractic profession may help address such concerns by continuing to assess the concordance between evidence and practice [59‐62] and promoting greater standardization of care where important variability exists. Further research on the benefits and harms of cervical manipulation is needed to establish the appropriate role of this modality [63, 64].

Despite the many challenges that exist, there are good reasons to pursue improved relations between chiropractors and family physicians; interprofessional collaboration among healthcare providers is associated with improved patient satisfaction and outcomes [65, 66]. Moreover, preliminary evidence suggests that collaboration between chiropractors and physicians for shared patients may reduce use of prescription medication, including opioids, unnecessary imaging studies, and inappropriate referrals for surgical consultation [67, 68]. Efforts to improve relations may benefit from increased opportunities for family physicians and chiropractors to work together and learn from each other [7, 69, 70].