Attempting to explore chiropractors and their clinical choices: an examination of a failed study

Clinician decisions among first contact health practitioners have been described as being sometimes difficult to comprehend, because they do not always appear to follow logical pathways of reasoning [5‐7]. Poor chiropractic clinical decisions have been identified as consequences of belief systems [8]. Some chiropractors align themselves with a vitalist philosophy, described by some as ‘unorthodox’ [9], and perceive that the ‘subluxation’ is a cause of disease that can be remedied with a spinal manipulation or ‘adjustment’ [9]. This group has demonstrated a predilection for non-guideline use of X-rays, non-evidence based treatment choices, and a lower likelihood of engaging in inter-professional collaboration [9].

Clinical decision-making has recently been studied in chiropractic students [10, 11]. These students appear to struggle recognising non-indicated care [11] (Goncalves et al., in press) and this difficulty was also extremely strongly associated with unorthodox beliefs (Goncalves et al., in press).

One of these studies of chiropractic students has shown a relationship between decision making and psychological factors such as intolerance of uncertainty (IU) [10]. IU refers to a dispositional characteristic that reflects a set of negative beliefs about uncertainty and represents an underlying fear of the unknown [12, 13]. These negative beliefs result in a desire for predictability [14, 15] and are associated with lower levels of confidence with decision-making [13, 16]. For medical practitioners, IU has been shown to be associated with a range of less than desirable outcomes such as lower compliance with evidence-based guidelines, [17] and generally increased resource use in the health care system [18]. The role of IU has not been explored in practising chiropractors’ clinical decisions.

This recent study exploring IU in chiropractic students found that the vast majority of students had the desire to learn a prescriptive system that would indicate ‘where the problem is’ and ‘how to treat it’, rather than accepting the grey shades of clinical reality [10]. However, unexpectedly, IU was not found to be a significant predictor of this desire and this was thought to be due to a lack of clinical experience or that the IU failed to capture a relevant psychological profile. The authors of that study suggested that further research with practicing chiropractors adding broader psychological measures and questions with respect to technique adoption might bring more information. The “Big 5” psychological traits / dimensions, also known as the Five Factor Model, is a widely accepted and commonly used model of personality [19, 20] and, as such, may be an appropriate broader measure of personality to further study this unexpected result.

While practitioner uncertainty and anxiety may impact negatively on clinical decisions [21, 22] over self-confidence has been shown to be equally detrimental in medical students and physicians [23]. This also has not been studied in chiropractors.

Thus, we wanted to understand better how the psychological profiles of chiropractors might impact on their health care decisions. To this end, and based on previous research findings, we proposed four different frameworks of conceptualizing how a chiropractor may be influenced, when making clinical decisions by examining:

Our objectives were to determine if categorising chiropractors according to these frameworks would reveal differing responses on the following aspects of health care decisions:

In March 2018 our questionnaire was distributed via an e-mail invitation to 700 chiropractors who were members of the Australian Chiropractic Research Network (ACORN) practitioner cohort [24]. This was a quantitative cross-sectional descriptive study using an anonymous on-line survey platform (Survey Monkey).

Before distribution, the survey was first pilot tested on a small number of practitioners and academic staff for errors and to measure the time taken for completion. This resulted in some minor sentence structure changes. The initial invitation e-mail contained details of the study, the involved researchers’ names as well as information that the likely completion time was 25 to 30 min. A follow-up reminder e-mail was sent 4 weeks after the initial distribution. This was the maximum number of reminders allowed by ACORN.

The ACORN consists of a practitioner cohort of 1680 chiropractors, which represented approximately 36% of the licensed chiropractors in Australia, who were recruited from a national survey of all chiropractors in 2015. These chiropractors have since been approached in several surveys (but not all yet published) and to not fatigue the ACORN participants, questionnaires are sent out only to a randomly selected proportion of the cohort for each study.

Human Research Ethics approval was granted by Murdoch University (Project No 2017/157).

Information on the demographic details of the participants and their clinics was sought (sex, age, years of practice, number of patients, and X-ray usage).

Two basic cases with several potential scenarios were presented (Additional file 1) seeking chiropractors’ clinical decisions in relation to a patient with neck pain and another with low back pain. The neck and low back pain (LBP) questionnaires have been previously used to assess chiropractors’ clinical decision-making profiles [25‐28].

From this survey, variables shown to be associated with undesirable chiropractic practice characteristics were selected from previous research (as described above), in addition to the demographic and practice data:

Data were to be summarised using descriptive statistics and reported as means, standards deviations or medians and IQRs, depending on normality for continuous data and frequency distributions for categorical data. Data would thereafter be transformed according to the previous descriptions.

Data were to be analysed using SPSS v.24 (IBM Corp, Armonk NY, USA).

Of the 700 contacted chiropractors, 141 responded to the on-line survey (20% response). The average time taken to complete the survey was 23 min.

However, there were 33 incomplete sets of data that resulted in a final 108 complete responses (15.4% response rate) of which 66% were males with a mean age of 44.3 (SD 11) years (Table 1). The 33 incomplete questionnaires all lacked the responses for practice characteristics, and the measures of IU and Big-5. Of these 33, only 4 responded to the neck and LBP scenarios although they all responded to questions on chiropractic philosophy, subluxation, and technique use.

This low response rate indicates that the study sample is unlikely to be representative of its study population. Nevertheless, the profile of the responders is presented in Tables 1 and 2 and summarized below.

The chiropractors in this study worked on average 27 h (SD 10.3) per week. They treated 91 (SD 58.7) patients per week and had been in practice for 17.9 (SD 10.4) years. These responses approximated those of the larger ACORN project participants whose characteristics are also shown in Table 1.

Almost 70% of the participants in this study regarded chiropractic philosophy as important or very important for what they do in practice (see Table 2). Half of the subjects rated subluxation as very important or important in guiding what they do in practice. Almost 75% of the chiropractors rated their professional level as a bit above average or above average.

There were too few responders to make association testing appropriate.

This appears to be the first attempted study seeking to investigate chiropractors’ clinical reasoning. It sought to progress findings from previous studies in chiropractic student populations by proffering four different ‘psychological’ conceptual and belief frameworks that might influence clinical decisions that were, in turn, to be linked with a number of different aspects of chiropractic practice used as indicators of the quality of health care delivered.

However, only 141 chiropractors participated in this study and of these only 108 provided complete data; the final response rate therefore being only 15.4%. Although there were similarities with the larger ACORN population, this low response rate means it was not possible to generalize the findings of this study to the ACORN population and certainly not to the entire chiropractic population of Australia. The low number of responders made it unsuitable to perform tests of association.

There is limited literature to guide researchers in the practical aspects of recruiting professionals for research studies. The magnitude of this problem has previously been demonstrated with a review of the National Institute for Health inventory of trials showing that only 34% of trials ever reached their projected sample size [37].

Nevertheless, previous studies of chiropractors in other countries have resulted in better response rates. For example; Denmark 72% [38], Finland 88% [39], Norway 61% [40], Sweden 60% [25] and 77% [41]. These studies were all conducted in Scandinavia. The low response rate in the present study could therefore be a cohort problem, particular to Australia, as only 43% of all registered chiropractors originally consented to become members of the ACORN ‘panel’. Of these 2005 practitioners, only 83% (1680) did thereafter agree to join. Despite this acceptance, the one other published study from this ‘panel’ reported a response rate of only 33% [42].

Even so, the response rate in our study is discordant with that previous ACORN study. There would therefore appear to be something that triggered a general non-response. The pilot testing did not identify any obstacles that can explain this low response rate.

A potential explanation for the difference in the willingness to participate in surveys could be that the Scandinavian studies might have been considered clinically ‘helpful’ as they related to clinical questions such as predictors of outcome and defining maintenance care. It is possible that the previous and present Australian studies were felt to be uninteresting and less meaningful. The additional level of disinterest in the present study could also have been in protest, if chiropractors felt the topic to be politically ‘dangerous’, dealing with practice behaviour, and perhaps too personal regarding personality traits.

Other explanations for potential participants choosing not to respond to such a degree that the study has ‘failed’ could include a lack of belief that a relationship existed between the variables being researched [43], an uncomfortableness with being seen as unable to manage the technical aspects of the study [44], the likely benefit of the study was not worth the time taken to respond [45], or that the research question was not ‘worthwhile’ [46]. It is not possible to definitively know which one, or which combination of these explanations – if any - impacted on this study.

Another significant impediment to response is likely to be the inability to remind chiropractors to respond. Best practice for enhancing response rates in surveys is to initially send a “herald” notice that the survey is coming and important, then a despatch of the survey and at least two follow-up reminders at spaced intervals [47]. Unfortunately, ACORN will not allow this type of reminder system and this could impair response rates in future surveys of ACORN participants.

Examination of the 33 incomplete response sets identified that absence of replies to the intolerance of uncertainty questionnaire (12 items) and the Big-5 inventory (60 items) was common to all of these partial responders. It is possible that non-responders failed to participate for the same or similar reason, namely, that this study gives the appearance of practice behaviour being associated with psychopathology or deviant behaviour patterns. As such, it could be negatively perceived as an ‘attack’ on the chiropractic profession. The offer of anonymity on the patient information recruitment email does not appear to have alleviated concerns.

Studies exploring psychological profiles and belief systems have been conducted in other health professions and have provided valuable insight into factors that impact on clinical decision making and have the potential to improve health economics and patient safety [17, 18, 23, 48]. These factors may also be important in chiropractic care. This study suggests that this is a ‘sensitive area’ and that future efforts to gather data to examine such issues may be inherently problematic. It is possible that repeated similar attempts will only result in accentuating this type of non-response behaviour and ‘drive it further underground’. One possibility to improve the situation may be to conduct a qualitative study by a known and trusted interviewer(s) to identify the reasons for this reluctance to participate. This method may also seek information from interviewees on ways to engage this population better. Once the drivers for this type of behaviour are identified it may be possible to better articulate the aims, wording and design of the study to potential participants and thus improve response rates. Past research has shown, with appropriate funding, the better known methods are monetary (preferably prepaid) and non-monetary incentives (lotteries, gifts) [49], SMS reminders [50], and personal contact to achieve this end [51]. Future studies may also benefit from adding locally known and trusted network leaders to the target population [52]. In addition, ACORN should review its restrictions on reminders to enhance response rates.