Diagnostic and decision-making abilities of Swiss physiotherapists in a simulated direct access setting

ABSTRACT Background Direct access to physiotherapy requires physiotherapists to recognize red flags and determine adequate management plans. Purpose Investigate Swiss physiotherapists’ ability to diagnose and triage patients in a simulated direct access setting and whether their characteristics were associated with correct diagnoses and decision-making. Methods We conducted a national online survey using a questionnaire containing 12 first-contact case scenarios. A linear mixed model estimated scores for correct diagnoses and management decisions, differences between musculoskeletal (MSK), non-critical medical (non-CrM), and critical medical (CrM) case scenarios, and the estimated effects of physiotherapists’ characteristics. Results The linear mixed model of data from 1492 participants estimated 55.0% correct diagnoses (62.7% for non-CrM, 61.7% for MSK, and 40.5% for CrM scenarios) and 71.2% correct management decisions (78.1% for non-CrM, 73.0% for MSK, and 62.3% for CrM scenarios). For correct diagnoses, the ‘academic education/continuing education’ variable showed significant estimated effects for the MSK and CrM scenarios, as did ‘professional experience’ for the non-CrM scenarios, and the ‘≥ 50% musculoskeletal patients in consultations’ variable for all scenario groups. For correct management decisions, ‘academic education/continuing education’ variable showed significant estimated effects in CrM scenarios, as did ‘professional experience’ in non-CrM and CrM scenarios, and the ‘≥ 50% musculoskeletal patients in consultations’ variable in MSK scenarios. Conclusion The estimated effects of physiotherapists’ characteristics on correct diagnoses and management decisions showed substantial heterogeneity. Improving Swiss physiotherapists’ screening abilities remains important. Further research is required to develop innovative educational concepts and improve training for screening for red flags.


Introduction
Patients in Switzerland do not have direct access to physiotherapy, and health insurance reimbursement of that treatment requires a referral to a physiotherapist (PT) prescribed by a physician.Direct access would enable a person to access physiotherapy treatment and be reimbursed by their health insurance company without a physician's referral (Ojha, Snyder, and Davenport, 2014).A direct access care model might shorten pretreatment waiting times, reduce physicians' high workloads, increase patient satisfaction, achieve better treatment results, and reduce costs (Demont, Bourmaud, Kechichian, and Desmeules, 2021;Piano et al., 2017;Piscitelli et al., 2018).A recent review showed that direct access to physiotherapy within the US civilian health services was associated with reductions in total physiotherapy costs, total healthcare costs, and numbers of physiotherapy visits, as well as better functional outcomes than physician-first systems (Hon, Ritter, and Allen, 2021).A study on a large US military base found similar results, with direct access associated with more returns to duty and less overall longterm disability than the referral-based system (Szymanek, Jones, Shutt-Hoblet, and Halle, 2022).These findings are highly relevant to the emerging challenges to healthcare systems worldwide.Aging populations and sedentary lifestyles are leading to a higher prevalence of chronic diseases and multimorbidity (World Health Organization, 2016, 2018), developments that aggravate the healthcare pinch points resulting from worldwide shortages of healthcare professionals (World Health Organization, 2014).Thus, ensuring the provision of high-quality, cost-effective primary care will necessitate more efficient treatment pathways and a redistribution of healthcare system responsibilities (Gagnon et al., 2021;World Physiotherapy, 2017).In many countries, direct patient access to physiotherapy has been available for several years (e.g. in the USA in all states, but with provisions in some, Australia, the Netherlands, and Sweden) (Bury and Stokes, 2013;Hon, Ritter, and Allen, 2021) with the twin goals of optimizing musculoskeletal care and strengthening the profession's autonomy (Sandstrom, 2007).
In direct access settings, a PT's ability to recognize clinical danger signs (i.e.screening for red flags) and understand when patients should be referred to a physician is vital to patient safety (Welch, 2011).PTs have been reported to diagnose as accurately as orthopedic surgeons, sports medicine physicians (Decary et al., 2017), and emergency department physicians in Canada (Matifat et al., 2019) and as accurately as general practitioners in the UK (Downie, McRitchie, Monteith, and Turner, 2019).
In Switzerland and neighboring German-speaking countries, direct access has been the subject of intense political debate in recent years (Klein, Rottler, and von Wietersheim, 2018;Konrad, Konrad, and Geraedts, 2017;Nast et al., 2013).Before 2006, PTs received an Advanced Federal Diploma of Higher Education on completing their training in colleges of further education (Schämann, 2019).Since 2006, academic education for PTs occurs in Switzerland's universities of applied sciences in its German, French, and Italian-speaking regions (i.e. at Berner Fachhochschule BFH; Haute École Specialisée de Suisse occidentale HES-SO; Scuola Universitaria Professionale della Sivzzera Italiana SUPSI; and Zürcher Hochschule für Angewandte Wissenschaften ZHAW).Today a prospective PT graduate with a Bachelor of Science (BSc) degree will have gained at least 40 European Credit Transfer System points through practical training earned during clinical internships (Bundesamt für Gesundheit, 2019).Hence, students are required to complete clinical training during their study programs.However, some universities of applied sciences also require more hours of practical training, and one of the largest institutions in German-speaking Switzerland calls for 34 weeks of additional internships (Zurich University of Applied Sciences, 2021).
Switzerland's universities of applied sciences are obliged to offer continuing education in addition to entry-level education.They offer Master of Science (MSc) and PhD graduate programs as well as continuing education programs for qualified PTs (i.e.Master of Advanced Studies, Diploma of Advanced Studies, and Certificate of Advanced Studies) (Swiss Federal Department of Education Research and Innovation, 2004).PTs who hold an Advanced Federal Diploma of Higher Education in physiotherapy awarded before 2006 can subsequently acquire their BSc degree by completing additional scientific courses (Swiss Federal Department of Education Research and Innovation, 2012).
Making clinical decisions based on a physiotherapeutic examination is a specific ability taught during the entrylevel BSc in physiotherapy.Therefore, all new PTs are trained in assessing whether physiotherapy treatment is indicated (Ledergerber, Mondoux, and Sottas, 2009;World Physiotherapy, 2011a).However, one significant Swiss health policy concern is that direct access could lead to greater quantities of physiotherapy services delivered and thus higher costs (Die Bundesversammlung -Das Schweizer Parlament, 2016).Furthermore, critical questions are still being raised about whether the levels of physiotherapy education and continuing education are sufficient to endow PTs with the competencies to screen for red flags and make informed decisions on further management requirements or the need for referral to a physician (Konrad, Konrad, and Geraedts, 2017;Lackenbauer, Janssen, Roddam, and Selfe, 2017).It is relevant that these reflections not only consider PTs' professional experience and abilities but also the extent to which they treat patients with musculoskeletal (MSK) problems since these health disorders represent one of the largest groups of non-communicable diseases (Bundesamt für Gesundheit, 2016).The present study examined: 1) how successfully Swiss PTs made correct diagnoses and determined appropriate management decisions in three case scenario groups and overall; and 2) whether PTs' characteristics significantly influenced their diagnostic and decision-making abilities, i.e. in terms of their level of academic education/continuing education, professional experience (adjusted to a full-time equivalent level), and the percentage of MSK patients in their consultations.

Study design
A national, cross-sectional, online survey of PTs working in Switzerland was conducted in the three national languages (German, French, and Italian) using QuestBack' Unipark survey software, Cologne, Germany (https://www.unipark.com).A clarification of responsibility was obtained from the Cantonal Ethics Committee in Zurich (BASEC-No.: Req-2020-00742).

Sample selection and recruitment
A broad recruitment strategy was adopted to reach as many PTs as possible from all clinical settings, including those who had graduated or were employed in Switzerland, whether or not they were currently working (e.g.due to maternity leave).In October 2019, a short text was placed in the online newsletter of Physioswiss, along with a link to the survey.Physioswiss, with more than 10,000 members, is the largest Swiss physiotherapy association.The cantonal associations of Physioswiss, other specialist groups, outpatient practices, the universities of applied sciences, and other associations were also contacted by e-mail a few days later.A cover letter was sent to these stakeholders providing detailed information about the study and asking them to forward the online survey to their members, staff, students, or course participants.After four weeks, a follow-up text was published in the November 2019 newsletter, and after six weeks, another reminder was emailed to all the stakeholders.All graduate, licensed PTs able to read and write in German, French, or Italian were invited to participate in the survey.No other inclusion or exclusion criteria were set.

Questionnaire
The survey was based on a questionnaire developed by Jette, Ardleigh, Chandler, and McShea (2006).Twelve case scenarios and 12 supplementary questions inquired about the PT's clinical practice and patient management.Each scenario briefly described a patient history along with their associated signs and symptoms.Firstly, participants were asked to select the most likely diagnosis from a list of 18 possible diagnoses and then choose their recommended management approach from three options.These were: 'provide a physiotherapy intervention without referral to a physician,' 'provide a physiotherapy intervention and refer to a physician,' or 'refer to a physician before a physiotherapy intervention.' Jette, Ardleigh, Chandler, and McShea (2006) developed case scenarios in three different groups: 1) a musculoskeletal (MSK) group, with cases treatable by a PT (five scenarios); 2) a non-critical medical (non-CrM) group with red flags necessitating a medical referral but with no time urgency (four scenarios); and 3) a critical medical (CrM) group including red flags indicating the necessity for an immediate medical consultation but no physiotherapy intervention (three scenarios).This categorization was not provided to the participants.Case scenarios were developed from the literature on the symptoms, risk factors, and differential diagnoses of MSK problems and disorders resembling them (Jette, Ardleigh, Chandler, and McShea, 2006).A review by physiotherapy experts strengthened the scenarios' case content validity.No detailed information was available on reliability.
Linguistically and culturally-validated versions of the 12 case scenarios translated by Scheermesser et al. (2011) were used in Switzerland's German-and French-speaking regions.Translation into Italian and validation was performed according to the best practice guidelines for cross-cultural surveys (Coulthard, 2013;Survey Research Center Institute for Social Research, 2016).Nineteen supplementary questions were developed for the survey in German and translated into French and Italian requesting more detailed information on PTs' education, working conditions, and direct access care setting.The resulting questionnaire contained the 12 case scenarios and 31 supplementary questions.Five participants from each linguistic region undertook a pretest before final adjustments to ensure comprehensibility.The online survey questions were not randomized, and participants were able to modify their answers by pressing a 'back' button.The questionnaire is available in a Supplementary File.

Data analysis
Data were exported from Unipark to the Statistical Package for the Social Sciences (SPSS), version 26.0, for cleaning.Each case scenario's correct diagnosis and management decision answer was coded according to Jette, Ardleigh, Chandler, and McShea (2006) solution key (Table 1).The primary outcome 'score' was calculated as the mean percentage of: a) correct diagnosis; and b) correct management decision answers over all 12 case scenarios and for each case scenario group (MSK, non-CrM, CrM).To compare the percentages of correct diagnoses and management decisions between different case scenario groups, a missing diagnosis or management decision was designated as an incorrect answer.
The 'level of academic education/continuing education' variable was constructed by combining each participant's answers to questions 25 (highest educational level in physiotherapy) and 30 (continuing education).This resulted in the following groups (Table 2): Group 1 -MSc or PhD; Group 2 -BSc or Advanced Federal Diploma of Higher Education with the subsequent acquisition of a BSc (BSc-A) and continuing education; Group 3 -BSc or Advanced Federal Diploma of Higher Education with the subsequent acquisition of a BSc (BSc-A) but no continuing education; and Group 4 -Diploma (no BSc-A and no continuing education).
The 'professional experience' variable was calculated from data on 'years of work experience' and 'mean work percentage' (question 27).Participants initially selected their years of work experience from the categories: '< 3 years'; '3-5 years'; '6-10 years'; or '11-20 years.'The mid-point from their selected category was used in calculations (e.g.category 6-10 years = an estimated 8 years of work experience).For participants choosing the '> 20 years' category, the mean age at which the respondents started working after graduating from their entrylevel education was calculated.This calculation was based on data from the four other categories: each participant's estimated number of years of work experience (e.g. 8 years) was subtracted from their age (e.g.32 years old minus 8 years of experience implies an age at entry-level graduation of 24 years).The overall mean age of physiotherapists entering the profession from all four categories was calculated as 24.5 years old.This mean age was subtracted from the age of each participant in the '> 20 years' category (e.g.54 years old -24.5 years = 29.5 years of professional experience).In a second step, 'estimated years of work experience' was multiplied by the 'mean work percentage' which each respondent gave as the percentage of a full-time  equivalent position to adjust for part-time employment and potential breaks during professional careers (e.g. 8 years of 'estimated years of work experience' at a mean 80% work percentage results in an adjusted professional experience of 6.4 years).Since all the PTs indicated consulting either 0% or ≥ 50% of MSK patients in their daily work (question 18), the 'percentage of MSK patients in consultations' variable was calculated using the two groups of 0% or ≥ 50%.
Initial data analysis was conducted using SPSS software for descriptive statistics (frequency) to characterize the PTs and their practice settings based on: 1) current work settings; 2) highest level of education in physiotherapy; 3) professional experience (specifically in an outpatient setting and generally); and 4) continuing education.Secondly, a linear mixed model (LMM) was fitted to the data using the lme4 package (Bates, Mächler, Bolker, andWalker, 2015) in R software, version 4.0.4 (2020) to analyze associations between PTs' characteristics and the dependent variables' 'scores' for correct diagnoses and management decisions.The fixed effects included were: 1) the between-subject covariates of 'level of academic education/continuing education,' 'professional experience,' and 'percentage of MSK patients in consultations'; and 2) the within-subject covariates of 'group' (MSK, non-CrM, CrM) and 'dimension' (diagnosis/decision). Two-way and three-way interaction effects between each between-subject covariate and the within-subject covariates were included, as were random intercepts for participants and participants' interactions with the within-subject variables, to account for the correlation structure of the withinsubject design.Model comparisons, using likelihood ratio tests, revealed that three-way and two-way interactions could not be removed from the model.The confidence level was set at 0.95, and the results report them at levels of p < .05,p < .01,p < .001.

Results
The survey ran from 1 October to December 2, 2019.In total, 4,328 persons clicked on the survey link, 3,200 completed the first page (choice of language) and 1,504 completed the questionnaire to the end and returned it (34.75%response rate).Submissions from entry-level students (n = 4) and questionnaires with no answers (n = 8) were excluded.Some questionnaires contained missing values (e.g. a missing diagnosis in a case scenario).The number of missing in each case scenario is depicted in Table 3. Questionnaires with few missing values were included in the analysis, resulting in a final sample of 1,492 eligible questionnaires.
For correct management decisions, the model showed significantly lower estimated scores for CrM case scenarios than for non-CrM case scenarios (−15.78%,95% CI: −18.09 to −13.46, p < .001)and for

Effect of 'level of academic education/continuing education' on diagnoses and decisions
Significant estimated differences (Figure 2 and

Within-case scenario-group effects of 'level of academic education/continuing education
As illustrated in Figure 3 and specified in  For correct management decisions, significant estimated group differences were present in the CrM case scenarios: MSc/PhD degree versus BSc/BSc-A without continuing education (+8.86%, 95% CI: 2.94-14.77,p < .001)and BSc/BSc-A with continuing education versus BSc/BSc-A without continuing education (+5.18%, 95% CI: 0.87-9.48,p < .05).There were no estimated group differences in management decisions for the non-CrM and MSK case scenario groups.

Effects of 'professional experience' on diagnoses and decisions
Years of professional experience showed no significant estimated effect on a correct diagnosis for any of the 12 case scenarios.However, there was a significant estimated effect on management decisions (trend: +0.129%, 95% CI: 0.042%-0.217%,p < .01),as presented in Figure 4.

Within-group effects of 'professional experience'
Years of professional experience had an estimated effect on the correct diagnoses scores in the non-CrM case scenarios (trend: +0.298%, 95% CI: 0.160-0.436,p < .001),but not in the MSK and CrM case scenarios.An estimated effect on correct management decision scores was seen in the CrM (trend: +0.256, 95% CI: 0.105-0.406,p < .001)and non-CrM (trend: 0.236%, 95% CI: 0.098 to 0.374, p < .001)case scenarios, but not in the MSK case scenarios (Figure 5).

Effects of 'percentage of MSK patients in consultations (0 versus ≥ 50%)'
For the 'percentage of MSK patients in consultations' variable (Figure 6 and
In terms of correct management decisions, these estimated group differences were −5.9% for the MSK (95% CI: −8.18 to −3.62, p < .001)and +2.7% for the CrM (95% CI: 0.06 to 5.36, p < .05)case scenario groups.No group differences were observed for the non-CrM case scenarios.

Discussion
This survey of PTs working in Switzerland used 12 case scenarios of first contacts with patients to investigate their ability to diagnose and make treatment management decisions.The fitted linear mixed model estimated a lower overall ability to diagnose accurately (55% correct) than to make accurate management decisions (71% correct).PTs' characteristics had a heterogeneous influence on the ability to diagnose correctly: the 'academic education/continuing education' variable had a significant estimated effect in the MSK and CrM case scenario groups; years of 'professional experience' had a significant estimated effect on the non-CrM case scenarios; and the '≥ 50% MSK patients in consultation' variable affected all three scenario groups.
Regarding management decision-making abilities, PTs with a higher level of academic education/continuing education were estimated to make more correct decisions in the CrM group; those with more years of 'professional experience' made more in the non-CrM and CrM groups; and those with '≥ 50% MSK patients in consultation' made more correct treatment management decisions in the MSK group.
The lower percentage of correct diagnoses than correct treatment management decisions may reflect the fact that, currently, medical diagnoses are primarily made by physicians in the Swiss healthcare system.Making initial medical diagnoses is not part of a PT's BSc curriculum and is outside their daily practice activities.In contrast, the screening process for identifying red flags and making decisions on further management (treat or refer) is already routinely applied (Ledergerber, Mondoux, and Sottas, 2009;Lüdtke, Grauel, and Laube, 2017).Jette, Ardleigh, Chandler, and McShea (2006), Klein, Rottler, andvon Wietersheim (2018), andNast et al. (2013) all used the same questionnaire, but none of them reported results on diagnostic abilities.However, all three studies' management decision-making outcomes can be compared with those from this study.Indeed, results were consistent across the three case scenario groups: participants made the best management decisions for the non-CrM group, followed by the MSK and CrM groups.The present study's percentages of correct management decisions (i73.0%and 62.3% in the MSK and CrM case scenario groups, respectively) were within the ranges of the three prior studies (71.3%-87.3%for MSK and 56.9%-79.0%for CrM case scenarios).However, our 78.1% rate of correct management decisions for non-CrM case scenarios was below their range of 81.3%-89.5%.A possible explanation for this difference is that the number of correctly solved non-CrM case scenarios was positively associated with increasing years of professional experience.Therefore, the difference may be due to our sample including fewer PTs with over 10 years of professional experience than that of Jette, Ardleigh, Chandler, and McShea (2006).However, percentages of correct management decisions were notably lower in all our groups than in the Jette, Ardleigh, Chandler, and McShea (2006) study which included a sample of American PTs, and in the Klein, Rottler, and von Wietersheim (2018) study on a sample of Dutch and German physiotherapy students.This is significant since, at the time of the study, PTs in 48 US states were allowed to examine and evaluate patients without a referral from a physician (Jette, Ardleigh, Chandler, and McShea, 2006).In the Netherlands, direct access was introduced in 2006 (Leemrijse, Swinkels, and Veenhof, 2008), and its physiotherapy curriculum appears to be more aligned to the abilities required in this environment than Switzerland's (World Physiotherapy, 2013).In addition, we hypothesize that student PTs are more used to working on hypothetical patient cases than are graduate PTs.Case scenarios are a regular part of their education (e.g. in problem-based learning courses) (Gwee, 2009).Furthermore, students are familiar with written tests including clinical case scenarios.This might explain the higher scores obtained by the students who participated in the study by Klein, Rottler, and von Wietersheim (2018).

Influence of PTs' Characteristics
Our data showed that the effects of PTs' characteristics on correctly-solved case scenarios were heterogeneous for both the diagnosis and decision-making dimensions and within the three case scenario groups.

Level of academic education/continuing education
For patient safety in clinical practice, it is essential that PTs are able to recognize red flags (e.g.CrM cases) and make appropriate management decisions (e.g.referral to a physician when necessary).The LMM estimated that  only 62% of the Swiss PTs participating in this study would choose the correct management decisions in CrM cases.This implies that recognizing red flags and appropriate clinical decision-making processes are insufficiently advanced in Switzerland.Nevertheless, a solid foundation for this has been laid through the profession's academic education and continuing education and by the inclusion of clinical decision-making processes in physiotherapy students' curricula (Ledergerber, Mondoux, and Sottas, 2009;Schämann, 2019).Our findings revealed a significant estimated difference between different educational levels in the CrM case scenarios (e.g. between participants holding an MSc/PhD and those holding a BSc/BSc-A degree without continuing education).However, comparing this with the results of Klein, Rottler, and von Wietersheim (2018) and Jette, Ardleigh, Chandler, and McShea (2006), it is evident that the different educational backgrounds of PTs in various countries, and the physiotherapy profession's status within its healthcare system, also contribute to PTs' varying abilities and expertise in screening and clinical decision-making (Lackenbauer, Janssen, Roddam, and Selfe, 2017) only consider the results from our MSc/PhD group, their percentage of correct decisions was lower than that found in the studies from the US and the Netherlands, including all educational levels.This clearly emphasizes the need to adapt physiotherapy education and training in Switzerland.

Professional experience
Our results implied that with greater professional experience, PTs were more likely to make correct management decisions, particularly in non-CrM and CrM case scenarios.Nast et al. (2013) also reported a significantly greater number of correct management decisions in CrM case scenarios among Swiss PTs with five or more years of professional experience.In contrast, Jette, Ardleigh, Chandler, and McShea (2006) found that PTs with over ten years' experience were no more likely to make correct clinical decisions than those with less than ten years in all three case scenario groups.This might be explained by more novice PTs in the US having higher levels of professional education and training.Further evidence associates greater experience with improved clinical practice.Doody and McAteer (2002) mixed-methods study revealed that both novice (students) and expert (> 10 years of professional experience) PTs used hypothetico-deductive reasoning processes.However, while expert PTs could evaluate the signs and immediately make their hypothesis through pattern recognition, novice PTs faced problems identifying patterns and sometimes linked their decisions to signs or unevaluated hypotheses (e.g.guesses).These results were supported by those of Budtz et al. (2021) who used an adapted version of the Jette, Ardleigh, Chandler, and McShea (2006) questionnaire and conducted interviews with PTs.Their quantitative analysis also showed that PTs with greater professional experience were more likely to choose the correct management decision.In interviews, PTs also emphasized that greater experience helped identify medical conditions that shared the signs or symptoms of MSK problems.Forbes, Mandrusiak, Smith, and Russell (2017) reported that more experienced PTs (≥ 11 years) provided more frequently patientcentered education practices (e.g.providing specific information about the patient's condition or exploring patient's ideas and perceptions about their condition) than novice PTs (≤ 5 years), and the latter perceived more patient-related barriers (e.g. a patient's emotional or cognitive status) in their daily work than more experienced PTs.

Percentage of MSK patients in consultation (0 versus ≥ 50%)
In Jette, Ardleigh, Chandler, and McShea (2006) study, PTs with > 50% orthopedic patients had an odds ratio of 2.23 of making a correct management decision in MSK case scenarios.This result supports the present study's findings, where PTs with '≥ 50% of MSK patients in consultation' were estimated to make more correct management decisions.This contrasts with the PTs with '0% MSK patients in consultation,' who were estimated to make better management decisions in CrM cases.These findings suggest that specialist consultations or work settings with high numbers of patients suffering from similar health problems may be conducive to making more appropriate treatment management decisions.This was also observed by Ladeira (2018) in the USA who studied PTs and different case scenarios of low back pain.These results supported Jette, Ardleigh, Chandler, and McShea (2006) who reported that PTs specialized in clinical orthopedics or manual therapy were more likely to make correct management decisions in CrM case scenarios.However, Nast et al. (2013) found no differences between PTs with an MSK specialization and those with other specializations.Nevertheless, PTs in the study by Budtz et al. (2021) described how specializations affected clinical reasoning processes as they allowed more structured approaches to patient examinations.In summary, evidence on how PTs characteristics predicted their diagnostic and treatment management decision-making abilities were inconsistent.In Switzerland, the development of PTs' key competencies has been based on the CanMEDS framework (Frank and Danoff, 2007), which describes seven different roles that professional PTs must perform (Ledergerber, Mondoux, and Sottas, 2009).The model emphasizes that the provision of high-quality physiotherapy treatment is dependent on PTs developing multi-layered competencies.This requires not only highlevel knowledge and abilities (e.g. the role of expert) but also other more personal skills (e.g.roles of scholar or manager).Compared with Nast et al. (2013) results, the present study found a lower percentage of correct diagnoses and management decisions even though the average PT had undergone more years of academic training.The data implied that raising PTs' abilities in simulated direct access situations would require enhancing multiple competencies, not just an improvement in one factor (e.g.their level of education).
According to Scheermesser et al. (2011) in 2010 over 84% of Swiss PTs felt confident that they could practice in direct-access situations, perhaps because this is viewed as a key practice in a PT's daily work.Although currently, patients can only access physiotherapy services after a referral by a physician, an initial physiotherapy consultation includes an anamnesis and a physical examination resulting in a physiotherapeutic diagnosis (Bundesamt für Gesundheit, 2019).Furthermore, a physician's diagnosis is often insufficient or missing.A retrospective examination of over 1,000 physiotherapy prescriptions in the USA by Riley, Tafuto, and Brismee (2016) found that over 29% contained nonspecific diagnoses (e.g.lumbago), and almost 60% included nonspecific treatment measures.Scheermesser et al. (2011) reported, that more than 60% of participants in their study thought that further education in screening would be needed before implementing direct access.The present study's results reflected this, particularly in the CrM case scenarios.Based on the literature and our research, we recommend combining four approaches to increase the screening abilities of PTs in Switzerland.Firstly, entry level curricula and continuing education should be adapted based on World Physiotherapy guidelines (World Physiotherapy, 2011a, 2011b).Secondly, students should have more opportunities to improve their clinical decision-making processes by gaining more experience in outpatient settings during their entry-level education.There are few internships on offer in Switzerland's clinical outpatient practiceswhere direct access would take place-due to a lack of financial incentives and difficulties in getting reimbursed for treatments carried out by students.According to McMahon, Cusack, and O'Donoghue (2014) the greatest obstacles to training students in outpatient practices were a shortage of trainers and a lack of financial incentives and tradition.Schämann (2019) reported that 17% of Swiss PTs' work was with inpatients versus 83% with outpatients.The author considered that comprehensive preparation for direct access settings would be of central importance to the future of the physiotherapy profession.This could be achieved by further expanding treatment simulation programs or creating outpatient clinics for education and practice, where students could learn and practice under supervision in a professional and interprofessional setting.The first centers of this kind are currently being established in Switzerland's universities of applied sciences.The third approach would be the provision of nationally approved (academic) continuing education in screening for red flags and using appropriate methodologies, thus supporting the development of PTs from novice to expert and maintaining high levels of performance (Ericsson, 2015;Faucher, 2016).The fourth approach would be the promotion of interdisciplinary networks and professional exchange.

Methodological considerations and limitations
Physioswiss's 2019 statistics report its members as 74% German-speaking, 21% French-speaking, and 5% Italian-speaking (Physioswiss, 2020), suggesting an over-representation of German-speaking participants and an under-representation of French-speaking ones in our study sample.However, our study sample did correspond to Physioswiss's member distribution of 70.3% female and 29.7% male.Overall, our study sample was representative of Swiss PTs.There was nevertheless an evident high non-response bias since only 1,504 of over 10,000 Swiss PTs contacted took part in the survey.According to Sax, Gilmartin, and Bryant (2003) a fastmoving culture can lead to fewer survey participants, and our lengthy questionnaire may have influenced participants' motivation to participate.
In this study, we decided to include data from participants with missing values and to consider missing values as wrong answers.The questionnaire simulates an exam situation, where only correct answers are scored.However, we did not reveal to the participants, that missing answers will be treated as wrong answers in the analysis, and a minor underestimation of the number of correct answers in a forced-choice situation is likely.
The random effects in our LMM account for the correlation structure of the within-subject design.In addition, potential confounding was controlled for since 'level of education' and 'professional experience' (among other variable)s were put into the model simultaneously.However, interpreting the results about participants' characteristics requires considering these variables' attributes since some variables (e.g.'level of education') can be influenced by the PT, whereas others cannot (e.g.'professional experience').Despite the study accounting for part-time work and possible breaks from work, it must be acknowledged that the 'professional experience' variable remains an estimation of years of professional practice with the potential for measurement bias.Further study limitations result from reliability and validity issues.First, no specific thresholds in the questionnaire categorize the performance of Swiss PTs into good, medium, or bad, for example.Categorizing PTs' abilities would require comparison with other studies.Secondly, the appropriateness of using a questionnaire to reflect PTs' diagnostic and decision-making abilities is questionable because physical examination and a patient's general presentation are such essential components of a PTs' clinical decision-making.Thirdly, leaving the case scenarios in their original form helped compare our results with other studies, but it could also have led to data falsification.According to current guidelines, some case scenarios should be treated differently (e.g.case scenario 5: According to National Institute for Health and Care Excellence (2018) physiotherapy could be started instead of an immediate referral).Fourthly, it was impossible to control for overrandomness in the responses.

Conclusion
In conclusion, improving the ability of Switzerland's physiotherapists to screen for red flags remains important.While entry level education may provide a foundation to implement direct access consultations, more attention should be paid to physiotherapists' ability to make differential diagnoses.Adapting the curricula for entry-level students, enabling them to gain practical experience in outpatient settings, and providing continuing education courses for graduate physiotherapists nationwide will be crucial steps toward ensuring that Switzerland's physiotherapists have the superior abilities required to ensure direct-access physiotherapy services.Further research is required to investigate innovative concepts for educating physiotherapists, training them to screen for red flags, and implementing their skills in clinical practice.

Figure 2 .
Figure 2. Effect of 'level of academic education/continuing education' on diagnoses and decisions.MSc/PhD = Doctoral or Master of Science degree; BSc/BSc-A (with continEdu) = Bachelor of Science or Diploma with the subsequent acquisition of BSc (BSc-A) and continuing education; BSc/BSc-A (no continEdu) = Bachelor of Science or Diploma with subsequent acquisition of BSc (BSc-A) but no continuing education; Diploma = no BSc-A, no continuing education Table5.Estimated marginal means: effect of 'level of academic education and/or continuing education' on diagnoses and decisions.

Figure 1 .
Figure 1.Between-group effects for diagnosis and decision.MSK = musculoskeletal case scenario group; non-CrM = noncritical medical case scenario group; CrM = critical medical case scenario group

Figure 3 .
Figure 3. Within-group effects of 'level of academic education/continuing education.'MSK = musculoskeletal case scenario group; non-CrM = non-critical medical case scenario group; CrM = critical medical case scenario group; MSc/PhD = Doctoral or Master of Science degree; BSc/BSc-A (with continEdu) = Bachelor of Science or Diploma with the subsequent acquisition of BSc (BSc-A) and continuing education; BSc/BSc-A (no continEdu) = Bachelor of Science or Diploma with subsequent acquisition of BSc (BSc-A) but no continuing education; Diploma = no BSc-A, no continuing education
CS = case scenario 1-12; No. = number; * A missing diagnosis or management decision in a case scenario scored no points.

Table 2 .
Categories in the 'level of academic education/continuing education' variable.
a entry-level degree before 2006; b entry-level degree since 2006

Table 3 .
Number of missing values per case scenario (out of N = 1,492).

Table 5 .
Estimated marginal means: effect of 'level of academic education and/or continuing education' on diagnoses and decisions.

Table 7
) group differences were estimated between 'no MSK patients in consultations' and '≥ 50 MSK patients in consultations' for diagnosis (−4.73%, 95% CI: −6.28 to −3.19, p < .001),but no estimated significant group difference was shown for management decisions.

Table 6 .
Estimated marginal means: within-group effects of 'level of academic education/continuing education.'

Table 7 .
Estimated marginal means: Effect of 'percentage of MSK patients in consultations' on diagnoses and decisions.