Study procedure
A team consisting of the three authors designed the questionnaire and four 4th year students from Murdoch University assisted with the survey administration and data collection.
This cross-sectional study was conducted between October and November in 2016. The chiropractic programs based at two Australian universities (Murdoch University and Macquarie University) were used for data collection. This was a quantitative descriptive study using an anonymous classroom handout questionnaire, as this approach facilitated the collection of a large amount of robust data in a timely and cost-effective manner. The entire student population across both programs was invited to participate and consent was obtained prior to completion of the survey.
Ethics approval was granted by Murdoch University (Project No 2016/118) and was classed as negligible risk research.
The project followed the same protocols in both institutions, consent was obtained from students, data were non-identifiable (anonymous) and permission was obtained from the Head of the Macquarie University chiropractic program to conduct the research. Accordingly, the study met the criteria for classification under the National Statement on Ethical Conduct of Human Research (2007) (Sections 5.1.8 and 5.1.22) as exempt from requiring ethics approval from both Universities.
The questionnaire
The survey contained four sections (see Additional file 1), with the results of some sections to be reported elsewhere. The first section sought demographic details (chiropractic program, sex, year of study) as studies on medical students indicate that levels of IU decrease over the course of training [32]. Further, age and sex have been shown to be independent predictors of IU [33].
In the second section, we devised direct questions to determine attitudes towards chiropractic care and clinical behaviour. The survey also sought students’ attitudes toward their likelihood of giving advice in their own chiropractic clinics and their beliefs on the capacity of spinal manipulation to impact on a number of health issues. It also asked students about their beliefs on the need, once graduated, to adopt a rigid comprehensive chiropractic technique system, which would inform them of the patient’s presenting problem (i.e. a technique system used to reduce uncertainty).
The third section was comprised of two clinical scenarios from previously published case management scenarios with chiropractors [9, 34]. A full description of the rationale for each clinical scenario and the rationale for classification of correct and incorrect choices is attached (see Additional file 2).
In the first case study, five scenarios were presented, beginning with a simple uncomplicated case of neck pain, which gradually progressed through to a scenario requiring immediate medical referral. The neck case consisted of the following general information: “A 28-year old man, tennis player by profession, consults you for a right-sided intense neck pain without any radiating pain. You note an antalgic position of the head, no other musculoskeletal signs (eg., no acute torticollis), no other health problems in particular, normal x-rays for his age, and no signs of alert (red flags).” [34] There was a choice of six answers for each of the five scenarios ranging from treating the patient on his/her own through to not providing treatment and arranging referral.
In a previous study, consensus was demonstrated on the most appropriate management choices across the five scenarios, thus allowing differentiation between chiropractors who select appropriate and inappropriate intervention strategies [34]. The inappropriate choices related to referral out of the chiropractic clinic when the patient should have been treated by the chiropractor and also to continued care when the patient should have been referred out.
The second case described a range of clinical scenarios for a patient with low back pain designed to identify which management strategies the chiropractors preferred to use in response to various outcomes of the initial treatment program [9]. This questionnaire had nine possible outcome scenarios that were briefly described. Six clinical management alternatives were offered for each outcome scenario. The basic facts for this hypothetical patient were: “A 40-year old man consults you for low back pain with no additional spinal or musculoskeletal problems, and with no other health problems. His X-rays are normal for his age. There are no ‘red flags’”.
The nine additional scenarios were constructed in such a way as to include cases that went from uncomplicated to more difficult, including scenarios with no past history of low back pain, those with intermittent low back pain over the past year, and those with several similar events over the past year. In three previous studies, a pattern of self-reported clinical management strategies was demonstrated which allowed identification of those clinicians who did and did not follow ‘clinically logical’ answers (see Additional file 2) [9,10,11].
The fourth section consisted of the Intolerance of Uncertainty Scale. We chose the validated 12-question version (IUS-12) that utilises a 5-point Likert scale with responses ranging from ‘not at all characteristic of me’ to ‘entirely characteristic of me’ [1, 35,36,37,38]. Examples of questions included ‘unforeseen events upset me greatly’ and ‘the smallest doubt can stop me from acting’. The maximum possible score was 60, reflecting high levels of intolerance of uncertainty.
Lastly, we asked students to rate themselves as a chiropractor compared with other chiropractic students in their class.
Procedure
The content and wording of the questionnaire were pilot-tested by a small number of chiropractors. The questionnaire was adjusted in response to their comments and tested on a small number of chiropractic students. This process detected some logical errors in the description of the attitudes and beliefs and resulted in some wording changes, which further improved the content, wording and design of the questionnaire.
Students in both chiropractic programs were informed that participation was voluntary and would be anonymous. Students in years 1 and 2 were not given the case scenarios in their survey, as they were deemed to have inadequate clinical knowledge.
Variables of interest and analysis of data
Data were entered and analysed in SPSS v.22 (IBM Corp, Armonk NY, USA) after identifying and correcting any incomplete or corrupt data. All survey items were dummy variable coded and descriptive statistics generated.
Predictor variable
Visual examination of the distribution of the IU scores suggested a cut-off score at scores of 36 and above for construction of the two IU groups. Receiver operator characteristic (ROC) curves were utilised to evaluate potential high IUS-12 from total IUS-12 score. Sensitivity and specificity of the coordinate points of the resulting ROC curve were used to identify the potential cut-off score for IUS-12. The optimum critical value for the IUS-12 was identified as 35.5. Subsequently scores of 36 and above were allocated into the High-IU group (25% of the students) and those below this score were allocated into a Normal-to-low IU group (75% of the students). This decision was also supported by previous published research with non-clinical samples which used similar values [39, 40].
Outcome variables
Variable 1: Neck Pain Case
From the previously published neck pain case [34], the research team determined which answers were the correct and incorrect treatment choices for each of these five scenarios. These are further explained in Additional file 2.
Students were given one point for selecting the incorrect treatment choice for the first two and last two scenarios, making it possible to obtain a score of between zero and four. For items 1 and 2, all the answers other than treating the patient on his own (response A) were considered incorrect. For items 4 and 5, all the answers other than choosing not to treat the patient but instead, referring the patient out (response E), were considered incorrect.
Variable 2: Low Back Pain Case
A previously published low back pain case was used, on the basis of which students were asked to identify their clinical decisions [9]. The research team determined which answers were correct and incorrect for i) inappropriate selection of premature referral (items 1,2,4; scores ranging from 0 to 3) and ii) inappropriate selection of continuation of care when referral was indicated (items 6,7,8,9; scores ranging from 0 to 4). Thus, high scores were indicative of an inappropriate management choice. These questions and their rationale for the ‘correct’ and ‘incorrect’ answers are described in Additional file 2.
Variable 3: Use of chiropractic technique analysis / evaluation systems
The answers to the question on chiropractic technique were dichotomised into yes (‘yes’ and ‘yes probably’) and no (‘don’t know’, ‘no, probably not’, ‘no’).
Variable 4: Belief in Self
The rating of students’ own future capacity as a chiropractor was grouped into four categories: Below Average (‘below average’ and ‘a bit below average’), Average, Above Average (‘A bit above average’ and ‘above average’), and Don’t Know. Preliminary analysis revealed the Below Average groups to be very small, and accordingly, we combined these respondents with the Don’t Know group, as we considered both types to indicate uncertainty.
Initially, the mean IU was compared for the chiropractic programs, sex, and year of program to see if subgroup analyses would be necessary. For this, tests for significant differences and relationships were carried out with Chi-square, ANOVA and linear regression analyses depending on the type of variable and data distribution.
The associations between the IUS-12 score and the scores for the neck pain and low back pain scenarios were tested in two ways. The score was treated as a continuous variable in a linear regression analysis, as this would make most use of the data. In addition the IUS-12 score was dichotomized into high and normal-to-low IU groups and tabulated with the numbers of correct and incorrect scores on the neck and low back case scenarios, as this would allow for a clinical interpretation of the results. We expected that the two types of analyses would go in the same direction.
Thereafter, associations between Normal-to-low and High IU and the predictor variables were tested for significance using the Chi-square test. Estimates were reported with 95% confidence intervals (CI). Preliminary analyses revealed no links between these variables. For this reason and because of small numbers, no sub analyses by institution, sex, or year of program were carried out in relation to tests of association.
To determine if there was a difference between IU groups in the students’ approaches to the neck pain case and also to the low back pain case (on both the aspect of maintenance care and inappropriate referrals), responses were dichotomised using the threshold that zero incorrect answers were treated as ‘acceptable’ answers and one or more incorrect replies were treated as ‘unacceptable’.