The study was a mixed methods study with an explanatory sequential design . The study consisted of two phases; a quantitative cross-sectional questionnaire survey and qualitative semi-structured interviews. The study was reported according to the STROBE checklist for cross sectional studies  and the Good reporting of A Mixed Methods Study (GRAMMS) guideline . The qualitative phase was conducted to help explain and elaborate the results from the quantitative phase, the rationale being that the quantitative results provide a general understanding and the qualitative phase refines the results and explores the PTs views and experiences in more depth . In this study, emphasis was given to the quantitative phase.
Integration occurred in two steps; the first step was to use the quantitative results to inform the qualitative interviews and the second to integrate the two sets of connected results and draw integrated conclusions.
Written and oral informed consent to participate in the study was obtained from all participants. The study was approved by the Danish Data Protection Agency (No. 1–16–02-41-19).According to Danish law, this study did not need ethics approval (Act on Research Ethics Review of Health Research Projects, October 2013) .
Participants and setting
In Denmark, GPs act as free of charge gateways to the healthcare system with the overall responsibility for referral to primary and secondary healthcare. Hence, GPs can refer patients to primary care PT with approximately 40% reimbursement. However, direct access to primary care PT is also a possibility (through private healthcare insurance schemes or off-the-street treatment without reimbursement), which does not prerequisite any additional education or certification for the PTs.
A total of 60 primary care PT clinics from Central Denmark Region (one of five Regions in Denmark) were contacted through email and follow-up telephone calls and invited to participate in a questionnaire survey. The clinics were randomly selected with a 1:1 rate on clinic size, meaning 30 small (six PTs or less) and 30 large (more than 6 PTs) were invited. A total of 27 clinics agreed to participate (45%). To ensure high participation rates and avoid selection problems, the decision was made to physically visit the clinics who agreed to participate in the survey. The clinics were all visited from August through October 2020 and during the visit all PTs at the clinic were encouraged to participate in the survey. It was emphasized verbally during the visit, that participation was voluntary and that the PT could decline to participate. The questionnaire was completed electronically by mobile phone. The PTs were not allowed to talk with each other during completion of the survey. Questionnaire data was collected using the RedCap system .
Development and pilot test of questionnaire
The questionnaire comprised two separate sections. The first section included background information on each PT; gender, age, years of clinical experience in private practice, training and education the past 5 years, which type of patients the PT treated and how large a proportion of the patients they treated without referral from the GP. Also, the PTs were asked if they would be willing to participate in a follow-up interview.
The second section of the questionnaire included 12 short clinical vignettes. The use of clinical vignettes is recognized as a valid method for measuring variations in clinical decision making abilities . The vignettes described a hypothetical patients age, gender and the clinical presentations for which the patient sought the PT for assessment and treatment. The vignettes were based on previously developed and validated vignettes . The vignettes described clinical presentations that were either medical conditions, which should not be treated by a PT alone, or musculoskeletal conditions, which would be appropriate for the PT to manage without consulting the patients GP. The medical conditions could be either non-critical or critical based on the urgency for further medical attention. Based on the description, the PTs were asked to make a management decision. There were three possible choices of management; A) to provide physiotherapy intervention, B) to provide physiotherapy intervention while encouraging the patient to contact their GP for further assessment and C) no physiotherapy intervention and a direct referral to the GP.
The original vignettes were developed to describe physiotherapists’ ability to make management decisions on physiotherapy intervention or medical referral in a direct-access setting. In this study, the PTs were not to assess the vignettes in a solely direct-access setting, as Danish primary care physiotherapy embrace both direct-access (without reimbursement) and access through referral (with approximately 40% reimbursement). To ensure the vignettes were appropriate in Danish context, the original vignettes were translated into Danish and additionally six vignettes were developed to ensure a broad range of musculoskeletal conditions were represented. The vignettes were reviewed and revised by an consensus group consisting of 2 GPs and 4 experienced practicing PTs (all had been practicing for more than 10 years) using the Nominal Group Technique . The group was initially presented to 7 musculoskeletal (MS), 6 non-critical medical (NCM) and 5 critical medical (CM) vignettes. First, the group was asked to read the vignettes individually, note any comments on content or unclearness and finally rate each vignette from 1 to 5 in relation to relevance and credibility. Afterwards, the rating was reviewed by the whole group and consensus on which vignettes to include was reached. A total of 5 MS, 4 NCM and 3 CM vignettes were included in the final questionnaire (see Additional file 1).
The final questionnaire was pilot tested in a sample (n = 7) of PTs. The PTs were interviewed using the cognitive techniques think-a-loud interviewing and verbal probing to ensure the questionnaires comprehensibility and comprehensiveness . Only minor revisions were needed after the pilot test.
Correct management decision
Each of the three response categories was dichotomized into correct management decision (yes/no). For musculoskeletal conditions correct management was; A) to provide physiotherapy intervention or B) to provide physiotherapy intervention while encouraging the patient to contact their GP for further assessment, whereas incorrect management decision was C) no physiotherapy intervention and a direct referral to the GP. For non-critical medical conditions correct management was; B) physiotherapy intervention while encouraging the patient to contact their GP for further assessment or C) no physiotherapy intervention and a direct referral to the GP. For critical medical conditions correct management was; C) no physiotherapy intervention and direct referral to the GP.
The main outcome was the PTs ability to correctly identify the correct management decision regarding the group of musculoskeletal, non-critical medical and critical medical conditions. Making a correct management decision is a complex reasoning process, and often the correct answer to the vignettes was debatable, especially in the medical conditions categories. To account for this, different outcomes were defined for the categories of conditions. Hence, correct management decision for the musculoskeletal conditions was defined as five correct answers to the five MS vignettes. We defined correct management decision for the non-critical medical conditions as three correct answers to the four NCM vignettes and finally for critical medical conditions as two correct answers to the three CM vignettes.
The explanatory variables were a priori chosen primarily based on previously conducted studies [14, 17] and all variables were self-reported:
1) Experience; years of experience as a primary care physiotherapist, 2) specialization; defined as completed and certified MDT (McKenzie Method), MT (Musculoskeletal Specialization) or CMP (Certified Mulligan Practitioner) physiotherapists, 3) treating patients without referral; PTs indicated whether or not they already treat patients without referral from the GP and 4) passed quality audit; A nationwide quality audit commenced in 2019, which includes all primary care physiotherapy clinics with the purpose of evaluate and develop the quality of primary care physiotherapy . Among pre-specified indicators of quality is an evaluation of the written patient record including evaluation of red flags (signs or symptoms of serious pathology) .
Based on previously conducted studies [12, 17], it was anticipated that there would be a 20%-point difference between the highly experienced versus the less experienced PTs abilities to make correct management decisions in the critical medical vignettes. To detect a 20%-point difference (power 0.80, alpha 0.05) a total of 194 PTs were needed. The 27 clinics who agreed to participate employed 239 PTs and with an expected participation rate of 85% the needed number would be reached.
The qualitative phase consisted of semi-structured interviews with primary care PTs. We made a purposeful sampling of PTs among those who had consented to participate in an interview. We included PTs who made correct as well as incorrect management decisions. Also, we wanted a broad range of PTs representing the different explanatory variables, e.g. PTs with few as well as many years of experience. This enabled different views and experiences to inform the qualitative phase.
An interview guide was developed by CRB and HRS based on the quantitative results. The guide focused on the PTs reactions to the results and their experience in relation to management decisions (se interview guide in Additional file 2). The PTs were not confronted with their answers to the questionnaire during the interview, they were merely presented with the overall conclusions. The guide covered both open-end and follow-up probe questions. After the first two interviews were performed by CRB, transcripts were made and read through by CRB and HRS to ensure the questions were covering the aspects of the quantitative results as attended. Consequently, minor revisions were made to the interview guide. We invited 24 PTs to participate in the interviews, and 9 agreed to participate (PTs mostly declined due to time pressure). The interviews (lasting 30–45 min each) were performed online via ZOOM and recorded in January 2021 by CRB.
Descriptive statistics (percentages, means) were used to characterize the participants and practice settings. Also, percentages of correct management decisions were calculated for each vignette as well as the three categories of conditions.
We then analyzed the association between correct management and explanatory variables for each of the three conditions using mixed effects logistic regression models, to ensure clustered data was handled correctly . The dependent variable was correct management decision (yes/no) for the specific group of conditions (e.g. 5 out of 5 correct answers in the musculoskeletal conditions). The explanatory variables were clinical experience in private practice (divided into 0–5 years or 5+ years), specialization (yes/no), treating patients without referral (direct access) (yes/no) and if the clinic had passed a quality audit (yes/no). The analyses were presented as Odds Ratio (OR) with 95% Confidence Interval (95%CI) and adjusted for all other explanatory variables. Considerations on the number of explanatory variables to include in the analyses were based on the principle of at least 10 cases per variable.
All statistical analyses were performed using STATA version 16.0 (StataCorp LP, College Station, TX, USA).
The interviews were transcribed verbatim. A directed content analysis was performed with a deductive approach, as described by Hsiu-Fang and Shannon . The quantitative results were used as categories for coding the transcripts, meaning initial coding involved marking the transcripts based on the following five categories; 1) Ability to make correct management decision, 2) experience, 3) specialization, 4) direct access and 5) quality audit. Relevant quotes were presented to illustrate the analyses results.
All analyses were made using QSR International (1999) NVivo Qualitative Data Analysis Software version 12.