Current practices of physiotherapists in Switzerland regarding fall risk-assessment for community-dwelling older adults: A national cross-sectional survey

Study design

A cross-sectional survey of physiotherapists currently practising in Switzerland was undertaken between the 21st of November and the 31st of December 2020.

Participants

Registered physiotherapists working in Switzerland and providing care to patients over the age of 65 in their daily practice were eligible to participate. Two mandatory screening questions were used to confirm eligibility: i) Are you currently (or in the last 12 months) working as a physiotherapist in Switzerland? ii) Are you managing patients over 65, regardless of their initial pathology?

Recruitment strategy

In the absence of a federal register comprising all currently practising physiotherapists in Switzerland, a comprehensive recruitment strategy was developed by the research team. It was designed to optimise the results generalisability and sample representativeness. Therefore, a range of organisations was asked to assist with study recruitment by sending the questionnaire to their members, which maintained confidentiality. First, all Swiss cantonal physiotherapy associations were invited to participate. Sending confirmations were received from Bern, Valais, Basel, Neuchâtel, and Aargau organisations. Additionally, the Swiss Association of Independent Physiotherapists (ASPI-SVFP) and the Swiss Association of Sports Physiotherapy (Sportfisio) also agreed to send the study questionnaire to their members. To broaden the sampling frame, Master’s students in physiotherapy at the Bern University of Applied Sciences (BFH) and the Zürich University of Applied Sciences (ZHAW), as well as ALUMNI physiotherapists from the HES-SO Valais-Wallis, were invited to participate. Moreover, lists of physiotherapy practices (n=2000) for each of the 26 Swiss cantons were created manually with online research on local.ch. Finally, the questionnaire was sent to all physiotherapists working on the Swiss CHEF Trial project. This ongoing national randomized controlled trial compares three home-based exercises programs aiming at preventing falls in older people of Switzerland (Mittaz Hager et al., 2019).

Non-monetary incentives

To maximise responses, all the participants who took part in the survey had the opportunity to participate in a prize draw (Andrews et al., 2003; Edwards et al., 2009; Tuten et al., 2000). Three prizes were offered, consisting of an overnight hotel stay. To ensure the anonymity of the survey responses, a separate webpage was used to collect and store the email addresses. There was no link between survey responses and email addresses stored for the prize draw.

Survey development

A draft questionnaire was developed using an iterative process with the research team, which included six physiotherapists with several years of experience in fall prevention and geriatrics. Permission to use clinical practice questionnaires developed for similar studies was obtained from two authors (Ackerman et al., 2019; Gaboreau et al., 2016). Ackerman et al. targeted fall prevention by older people with osteoarthritis and Gaboreau et al. focussed their study on general practitioners’ (GPs) routines linked to this topic. Their questionnaires were partially adapted to align with the research question and Swiss healthcare context. To maximise study quality and research rigour, the Checklist for Reporting Results of Internet E-Surveys (CHERRIES, see extended data “CHERRIES Checklist” (Duc et al., 2022)) (Eysenbach, 2004), results of different guidelines, systematic reviews and studies (Andrews et al., 2003; Artino et al., 2018; Edwards et al., 2009; Kelley et al., 2003; Lumsden & Morgan, 2005) as well as websites (Creative Research Systems, 2016; Deckers, 2017; Parrott, 2020) were followed. See extended data “Designing Tools” for brief descriptions of each of the tools mentioned above (Duc et al., 2022).

The semi-structured online questionnaire consisted of 52 short questions with several branching logics (adaptive questioning). This process allowed some questions to be displayed conditionally based on the answers to prior items, intentionally reducing the responder burden. The survey included a mix of open-ended questions and multiple response options covering first participants characteristics. As no identifying information was collected, the anonymity of the participants was preserved. Workplace, assessed only with the first two postal code numbers, was used to observe the presence of falls prevention programmes by geographical area. Perceived ability to assess and manage older patients was evaluated on two purposed-designed Likert scales ranging from 0 “none” to 100 “excellent”. The responsibility of nine health care professions in fall risk assessment was also questioned on Likert scales ranging from 0 “not at all concerned” to 10 “very concerned”. The way physiotherapists screen for fall risk included questions about the situations leading to testing the patient, the tests used, and the interventions undertaken. Risk factors and the way they are measured were also evaluated. One question specifically targeted the reasons why some therapists never assess fall risk to understand barriers to fall risk assessment. How physiotherapists quantified the risk of falling, reassessed patients and the elements needed to facilitate a more systematic risk assessment were asked at the end of the questionnaire. Three hypothetical patient cases (vignettes) from the musculoskeletal, respiratory, and neurological fields were also presented to respondents and their management related to fall risk assessment evaluated.

The questionnaire was first developed in French and English. It was translated entirely into three of the four country's main languages, French, German and Italian, to send it throughout Switzerland. As Romansh speakers represent only 0.5% of the total population (FSO, 2019a) and are primarily fluent in German or Italian (FSO, 2020), it was estimated that they could use one of those versions. The forwards translation was completed by the authors with the help of two specialised companies. The survey is also supplied in English in the extended data (see “Survey”) (Duc et al., 2022).

For ease of reading, no more than three questions were presented per page. At any time, respondents could review and change their answers using the “previous” button functionality. In most cases, the “other” option allowed participants to indicate a missing response option if necessary. Physiotherapists were free to terminate their participation definitively or momentarily at any time. They could proceed to the next page or skip questions. Except for the two screening questions, only one question was mandatory: “Do you usually use fall risk screening tools when assessing your patients over 65 years of age?”. By submitting their email address, the server automatically generated a link allowing participants to receive a partially completed questionnaire as is so that they could complete it later. See extended data “Survey Development” for a short description of the development stages of this questionnaire (Duc et al., 2022).

Survey validation

Ace Validity Testing and Preliminary Pilot Testing

We searched the literature and contacted experts to define and conceptualise relevant topics for the survey item generation. The different steps of survey validation were conducted with the help of 21 physiotherapists who were experienced in the treatment of older adults and measurement properties.

Questionnaire feasibility was assessed, as well as content and face validity and comprehensibility. Dillman developed a four-step methodological pre-testing process adapted and applied for this study to assess validity according to Artino et al. recommendations (Artino et al., 2018; Dillman, 2000). Refinements were made at each stage as required. In extended data “Survey Pre-testing and Validation” summary tables of Dillman’s pre-testing steps and their application in this study are presented. Questions used for survey comprehensibility assessment are provided in extended data “Comprehensibility Assessment” (Duc et al., 2022). The answers collected during the pre-testing and validation process were deleted before the questionnaire went online.

Assessment of reliability

Reliability and validity are essential scores to consider when elaborating a questionnaire (Artino et al., 2018). Validity assessment was undertaken following the pre-test process of Dillman (Dillman, 2000), and modifications were undertaken if necessary. To meet the 31st of December 2020 deadline for completion of data collection, this step was first not evaluated. However, test-retest reliability could be assessed between November 2021 and January 2022 for the publication of this work. The assessment of reliability was the final step in the survey validation. The questionnaire was sent to seventeen French and German speaking physiotherapists who took part twice to assess the test-retest reliability. Intraclass Correlation Coefficient (ICC) was measured for non-dichotomous data using ICC_2,1 and Cohen's unweighted kappa was calculated to estimate test-retest reliability of dichotomous items (Streiner et al., 2015). The lower limit values indicating moderate reliability were 0.41 for the kappa (Bland, 1991) and 0.7 for the ICC (Streiner et al., 2015).

A priori hypotheses

A set of 18 a priori hypotheses was established before the start of the data collection. These are presented in extended data “A-priori Hypotheses” (Duc et al., 2022). Hypotheses that emerged after the end of data collection were specified as explorative.

Project website

Consistent with the study’s main objectives, a website was specially designed. The survey was available in the three translation languages (French, German and Italian) and contained the key elements of the study organised by subthemes. An anonymous chat provided by Crisp allowed participants to ask questions and obtain almost instantaneous help. No IP tracking or cookies have either been used to ensure the anonymity of individuals visiting this website. The final report of this work will be published on this website for interested participants to access.

Data collection

On the 21st of November 2020, physiotherapists received an email with a link to the online survey hosted on the REDCap server of the HES-SO. Physiotherapists could contact the research team at any time for further information, either by email or anonymously via a chat on the project website. To avoid transcription errors, responses were immediately sent to the project’s password-protected REDCap database (Andrews et al., 2003). One reminder was sent to all participants three weeks after the initial mailing (i.e. 08.12.2020) to increase response rates (Andrews et al., 2003; Edwards et al., 2009; Lau, 2017). Data collection was completed by the 31st of December 2020. All details regarding survey distribution and follow-up are described in extended data “Survey Send Out” (Duc et al., 2022).

The cover letter gave an overview of the study's main objectives and efforts to preserve participant confidentiality and anonymity (see extended data “Cover Letter” (Duc et al., 2022)). It was specified that by submitting the two eligibility assessment questions, informed consent was implied. Eligible participants were led to the survey, and those considered ineligible were informed that they could not take part. Participation could involve any electronic device, and the estimated duration did not exceed 25 minutes. At no time were an IP tracking or cookies used. No other techniques to analyse the log file for the identification of multiple entries were used. The link to the online survey was posted on the project website but no other advertisement was done.

Ethical approval

An approval of Swiss Ethics was obtained via the “Clarification of responsibility” form on the portal of the Business Administration System for Ethics Committees (BASEC) in May 2020. As the survey was anonymous and no health-related data were asked, this study did not fall under the Swiss Human Research Act (CER-VD Req-2020-00515).

Data analysis

Data were analysed with R, version 4.0.5 (R: The R Project for Statistical Computing, 2020), RStudio, version 1.4.1106 (RStudio|Open Source & Professional Software for Data Science Teams, 2020) and Stata (Version 17, StataCorp, College Station, Texas) (Stata: Software for Statistics and Data Science, 2020).

Handling of variable’s grouping

When evaluating how physiotherapists assess the fall risk of their patients, of the 938 participants, only 8 (0,1%) reported that they never assess the fall risk. Therefore, it was decided not to use this item to test the hypotheses. Instead, a binary variable “use standardised assessment” was created from the four response options: i) By a subjective evaluation (observation of the patient, discussion…), ii) By a standardised evaluation (scale, functional tests, questionnaires, timed tests…) iii) By instrumental assessment (inertial sensors, force platforms, EMG…) iv) I do not perform an assessment. We dichotomised them as follows: “use standardised assessment”: yes = ii) & iii), no = i) & iv).

We categorised the year of completion of the highest education level as follows: i) “Before 1990”: < 1990, ii) “1990 to 1999”: > 1990 & < 2000, iii) “2000 to 2009”: > 2000 & < 2010, iii) “2010 or later”: > 2010.

A binary variable “Clinic or Private Practice” was created from the different work settings: i) Hospital, ii) Rehabilitation clinic, iii) Private practice as employee, iv) Private practice as self-employed person, v) Retirement home, vi) Home-based physiotherapist, vi) Other(s). We dichotomised them as follows: “Institution” = i) & ii) & v), “Private practice” = iii) & iv) & v) & vi).

Vignettes of patients with musculoskeletal, neurological, or respiratory problems were presented: i) A 65-year-old man consults for rheumatological problems (e.g., Osteoarthritis). His physiotherapy referral mentions "anti-inflammatory analgesia, improvement of joint and muscle function". What do you do?

ii) A 72-year-old woman consults because of neurological problems (e.g., Parkinson's disease). Her physiotherapy referral mentions "improvement of joint and muscle function, proprioception/coordination". What do you do?

iii) A 68-year-old man consults because of respiratory disease (e.g., Chronic Obstructive Pulmonary Disease). His physiotherapy referral mentions "improvement of cardiopulmonary function". What do you do?

The response options for the three vignettes were: i) I complete my assessment by using a specific fall risk assessment tool, ii) I do not complete my assessment by using a specific fall risk assessment tool, but I assess the main factors such as muscle strength, static and dynamic balance…, iii) As the patient does not consult for a fall risk problem, I do not assess either his risk or his risk factors for falling, iv) I do not treat this type of pathology in my daily practice. For some analyses, we dichotomised them as follows: “does fall risk assessment”= i) & ii), “does no fall risk assessment” = iii). The last answer option (iv) was excluded from this analysis.

Statistics

The distributions of the continuous variables were analysed visually, and median, interquartile range, minimal and maximal values were reported. Categorical variables were summarised with absolute and relative frequencies. 95% confidence intervals were reported to document the statistical precision of the estimates. The significance level was set at p<0.05 for all tests.

Differences between participants with more than 10% missing values and those with fewer missing values were compared with non-parametric tests for continuous variables and with Chi-squared tests for frequency tables. Cohen’s d effect size was calculated for continuous variables. An effect size of 0.2 was considered a slight difference, 0.5 a moderate difference and 0.8 a significant difference. Cramer’s V effect sizes were calculated for categorical variables, where 0.1 was considered small, 0.3 moderate and 0.5 large.

The hypotheses were tested using multivariable logistic regression for binary dependent variables and Chi-squared tests for categorical dependent variables. The logistical and linear regressions were adjusted for a set of potential confounders. Variables for the multivariable models were selected using Directed Acyclic Graphs (DAGs) according to theoretical considerations (VanderWeele, 2019). Nonparametric Cochran-Armitage statistics was used to test trend (i.e., hypothesis 2, increasing proportion of physiotherapists assessing with increasing education level). Contrast after an ANOVA was used to test whether the responsibility for screening was higher for physiotherapists as for other professions (hypothesis 5).

An additional exploratory analysis was performed. i.e., it has been clearly stated that no a priori hypothesis was formulated and that this result needed to be interpreted with caution.

The textual content of the semi-open questions (n=13) was analysed through a content analysis using an inductive approach (Elo & Kyngäs, 2008; Graneheim & Lundman, 2004). This analysis approach was the most feasible due to the extensive data and the human resources available. Each free text response (meaning unit) was closely reviewed and summarised into main concepts (condensed meaning units). Those were then grouped into emerging themes making clinical or theoretical sense. To illustrate them, some participants' responses are provided verbatim in their original language.

Finally, because there was some evidence of differences regarding fall risk activities and admission patterns to nursing home according to Swiss cantons (Merçay, 2017), it was decided to present the descriptive results stratified per language.

Response ratio

According to the Swiss Federal Statistical Office (FSO), it was impossible to know the official number of physiotherapists currently practising in Switzerland. Because third parties sent invitation emails, we do not know the number of emails delivered. Therefore, the response ratio could not be calculated.

Sample size considerations

The members of the national physiotherapy association, Physioswiss (n=10’652 in December 2019), and the Swiss Association of Independent Physiotherapists, ASPI-SVFP (n=500 in December 2019) were added to estimate the Swiss population of working physiotherapists, acknowledging that not all physiotherapists are members of one of these associations or that a potential overlap might occur. With an estimated precision of 5% (i.e. 95% confidence interval span of 10%) and an estimated population of 11'155 individuals, complete answers of 371 therapists were required (The Survey System). This calculation required the assumption of random sampling, i.e., that our sample is representative of all physiotherapists working in Switzerland. However, we were not able to test this assumption.

Atypical timestamps

The timestamps associated with questionnaire completion was analysed for each complete questionnaire (submitted or the last question answered) and partially complete questionnaires (not submitted or the last question not answered), all having a completeness ratio ≥ 90%. Data from participants who responded too quickly (i.e., < 5 minutes) were excluded from the analyses. Indeed, as the average time to take part in the questionnaire was estimated at 25 minutes, 5 minutes was not considered a serious attempt (Huang et al., 2012).

Completion ratio and missing values

The survey completion ratio was calculated by dividing the number of respondents who submitted the survey or completed the last question per the number of those who completed the informed consent.

Participant could skip questions if they did not want to respond. Only the inclusion criteria and the screening frequency were set as mandatory. The proportion of missing values per question, and the proportion of missing responses per participant, were calculated considering the branching logics (adaptive questioning). For sensitivity analyses, we analysed the participants with more than 10% missing responses and with less than five minutes completion time separately and compared the results to the other participants with statistical tests and effect sizes (Cohen’s d for continuous variables and Cramer’s V for categorical variables).

Participation ratio

It was intended that study participants should remain anonymous. Therefore, no IP tracking or cookies were used. Hence, it was impossible to know who clicked on the survey invitation link. In addition, due to data protection reasons and the anonymous survey, it was impossible to identify a unique visitor or to test whether a person responded twice. Consequently, the participation ratio (unique participants responding to the survey divided by those clicking on the invitation link) cannot be calculated.

Reliability

The median ICC (for the ordinal and continuous variables) was 0.86, with only one item below 0.7. The median Kappa value for the dichotomous items was 0.65, with 15% of items with Kappa values below 0.41. The items with lower than moderate Kappa values (i.e. <0.41) were items with many multiple choice response options (e.g. the type of falls screening tool used, the advice given regarding falls risk the type of balance assessments used). Furthermore, the responses for the question about which professionals are responsible for the screening of falls risk, the results for the nurses and occupational therapist had too low Kappa values.

Survey responses

A total of 981 individuals completed the two screening questions to assess their eligibility. Among them, 938 (95.6%) were eligible to participate. 715 (76.2%) physiotherapists (PTs) responded to at least 90% of the questions, 714 (76.1%) to the last question and 224 (23.9%) excited the survey prematurely, only providing partial data. All questionnaire data, even partially completed, were imported, analysed, and the attrition ratio (number of missing values per item) taken into consideration. The participants flow chart is presented in Figure 1.

Figure 1. Survey flow chart.

PTs = physiotherapists, n = number. ? = Number unknown because i) associations did not reveal the number of emails sent, ii) overlap of emails sent (physiotherapists may be in different mailing lists). For the percentages: 100% corresponds to the 938 eligible physiotherapists, except for the percentages in the box of the eligible participants, where 100% corresponds to the 981 participants who submitted the responses on the inclusion and exclusion criteria.

Atypical timestamps

The median participation time was 17 minutes (IQR 11 to 27), and no participant data were excluded because of atypical time required to complete the questionnaire.

Completion ratio and missing values

The completion ratio for this study was 76.1%. The percentage of missing responses per items ranged from 0 to 24% (see extended data “Missing Values” (Duc et al., 2022)).

The sensitivity analyses with the comparison of those with more than 10% missing responses to those with fewer missing responses showed several differences between both groups. Participants with more missing responses were three years older (Cohen’s d of 0.24, p<0.05) and less frequently had an education with a bachelor’s degree or higher (Cramer’s V effect size 0.13, p<0.05). Moreover, they also had 10% more participants who received their diploma before 1990 (p<0.05), 12% more who worked in private practice as employees (Cramer’s V effect size 0.13, p<0.05), and 7% less who worked in hospitals (Cramer’s V effect size 0.09, p<0.05). Furthermore, those with more missing values reported less confidence in the assessment of the fall risk (Cohen’s d 0.25, p<0.05) and less confidence in the management of people with an increased fall risk (Cohen’s d 0.19, p<0.05), compared to those with less missing values. Because of these systematic differences, it was decided not to exclude those with more than 10% missing as it would induce a sampling bias.

Participants’ socio-demographic characteristics

The characteristics of the respondents are shown in Table 1.

Most PTs were women (n=635, 69.2%), and the median age was 44 years (IQR 33 to 56). Participants came from all three language regions of Switzerland (see Figure 2). Most of the respondents were either graduates of a specialised school (n=321, 35.9%) or had a bachelor’s degree (n=310, 34.7%). Moreover, the most frequent formal training was a Master of Sciences in Physiotherapy (n=110, 12.3%). Several participants also had specific training in geriatrics (n=274, 30.6%). Certificates attesting to the highest education level were obtained after 2010 for 396 PTs (45.3%).

Figure 2. Number of participants per two-digits zip-code zones.

The numbers correspond to the number of participants in each two-digits zip-code zone.

Figure 3. Distribution of mutually exclusives categories of assessments used by the physiotherapists.

Percentage of physiotherapists using a specific category of assessments.

The participants worked in different settings with a high representation of those in private practice as self-employee (n=525, 56%). Of the 349 (39.1%) respondents working in a team, 198 (57.4%) were in a multidisciplinary team, including other health professionals. Most participants (n=536, 61%) worked part-time with a median of 60% (IQR 50 to 80).

The median number of patients over 65 years of age treated per week by the respondents was 14 (IQR 8 to 20). Within 12 months, about 53 % (n=348) of PTs received less than 5 referrals to reduce the fall risk, 21% (n=137) got between 6 and 10 referrals and 19.2% (n=126) more than 16. Finally, on a Likert scale ranging from 0 “none” and 100 “excellent”, the participant’s median confidence level in assessing the risk of falls in older adults was 75 (IQR 60 to 83) and 79 (IQR 65 to 87) in managing a patient at risk.

A priori hypotheses analysis

Of the 18 a priori hypotheses formulated before the questionnaire was sent to the participants, two were merged during the analyses because of their similarity (17 and 18). The set of hypotheses as initially drafted is presented in extended data “A-priori Hypotheses” (Duc et al., 2022). The summarised results of each hypothesis are provided in Table 3. Details of the descriptive statistics of the variables related to these hypotheses can be found in Tables 4 to 11.

Figure 4. Boxplots of the rating of how much professions should be involved in fall risk assessment.

Figure 5. Elements that lead to screening.

Multiple responses were possible, therefore percentages to not add up to 100%.

Figure 6. Frequency of reasons not to systematic assess fall risk.

Multiple responses were possible, therefore percentages to not add up to 100%.

Figure 7. Fall risk programmes awareness according to region.

The numbers in the legend correspond to the percentage of physiotherapists knowing at least one fall prevention programme (per two-digit zip region).

Characteristics of physiotherapists assessing their patients

First, the main characteristics related to the practice of fall risk assessment were investigated. Hypothesis 1a (H1a): Standardised tests or instruments are reported to be used by 580 (62%) PTs to evaluate fall risk and subjective assessment was used by 729 (77.7%) PTs (see Table 4). When participants were split into mutually exclusive categories regarding their risk assessment approach, it was observed that 49% performed a combination of subjective and standardised fall risk evaluation while 25% evaluated the risk only with a subjective assessment. Less than 5% used an instrumented evaluation (see Figure 3).

The odds of using a standardised assessment were 1.93 times higher for participants working in a clinic or hospital setting (institutional context) compared to those working in private practice (adjusted OR 1.93, 95% CI 1.37 to 2.7). H1b: 96 (36%) PTs working in an institutional setting and 129 (21%) of those working in private practice used a standardised instrument for 50% to 94% of their older patients. Furthermore, only 38 (14%) of those working in a clinic or hospital setting and 25 (4%) in private practice used it systematically, i.e., in at least 95% of their patients (χ² 70.8662, df 3, p <0.001). Results are presented in Table 2.

H1c: Among the 700 PTs treating older patients with musculoskeletal problems, 640 (91.43%) reported usually assessing them for fall risk or fall risk factors, even if they were not referred for a fall-related problem. The odds of assessing musculoskeletal patients for fall risk was 2.02 time higher for PTs engaged in an institutional setting (adjusted OR 2.02, 95% CI 1.01 to 4.04). H1d: Among the 666 PTs treating older patients with musculoskeletal problems, 660 (99%) reported usually assessing them for fall risk or fall risk factors, even if they were not referred for a fall-related problem. The odds of assessing neurological patients for fall risk was 1.02 time higher in PTs engaged in a clinic or hospital setting (adjusted OR 1.02, 95% CI 0.86 to 1.20). However, the 95% confidence interval range [0.86; 1.20], which was consistent with the null hypothesis that there is no difference in the fall risk assessment between PTs in institutional or private practice. H1e: Finally, among the 580 PTs treating older patients for respiratory diseases, 428 (74%) would usually assess the fall risk or fall risk factors in those patients even without a referral to reduce their risk of falls. PTs working in an institution did not assess these patients more often than those working in private practice (OR 1.51, 95% CI 0.98 to 2.33).

H2: There was a statistically significant increase in the proportion of PTs using a standardised fall risk assessment with a higher education level (test for trend, p<0.001). H3: Part-time PTs did not assess less than full-time PTs (OR 1.18, 95% CI 0.48 to 1.65). H4: PTs working alone used a standardised assessment less frequently than PTs working in a team (adjusted OR 0.52, 95% CI 0.37 to 0.72). H5: Finally, participants were asked to rate on a scale of 0 (not at all concerned) to 10 (very concerned) the responsibility for screening of nine health care professions. PTs practising fall risk assessment considered that this role was mainly incumbent on their profession rather than other professions (p<0.001, see Figure 4).

Current practices in fall risk assessment

In addition, the current practices related to the assessment of the risk of falling were considered. H7: The two most frequently factors leading PTs to assess their patients were the history of falls (n=469, 87.7%) and the evocation of a fear of falling (n=469, 87.7%). Receiving a specific referral to reduce the risk of falls was next (n=430, 80.4%), which only partially confirms the a priori hypothesis (see Figure 5 and Table 6).

H9: PTs who regularly performed fall risk assessment used a specific tool more often than those who evaluated it less often. Significant between-group differences were not always observed, but for those with p<0.05, the physiotherapist reporting evaluating fall risk “always” chose these tests more often compared to the physiotherapist assessing less often. There was one exception, the Timed-Up and Go (TUG), where therapists reporting assessing fall risk "often" (n=158, 70.2%) chose it more frequently than those who assess "always" (n=44, 69.8%). The standardised assessments most frequently employed by respondents were the Berg Balance Scale (BBS, 57.5%), the Timed-Up and Go (TUG, 56.5%), and the Tinetti Balance Assessment tool (Tinetti/POMA, 46.7%). Only a few PTs mentioned tools directly targeting fall risk such as the Stopping Elderly Accident, Death, and Injuries Algorithm (STEADI, 0.2%), the Fall Risk for Older People in the Community Assessment Tool (FROP-Com, 0%), or the AGS-BGS algorithm (0%) (see Table 8). H10: When assessing the risk factors of falls, PTs focused mainly on fall history (n=655, 88.3%), home hazards (n=617, 84.1%), perceived functional capacity (n=595, 81%) and joint mobility (n=576, 77.9%). All individual risk factors were assessed by at least 57% of the PTs except for blood pressure, which was assessed by 37.7% of PTs (see Table 9). H14: Among PTs using a standardised tool, the odds of quantifying risk was 4.24 higher than those not using it (adjusted OR 4.24, 95% CI 2.70 to 6.65). H15: The most common way participants did this was by classifying patient’s risk as “No risk /low risk /moderate risk/high risk” (n=430, 59.1%) or by dichotomisation “At-risk/not at risk” (n=134, 18.4%) (see Table 4). This does not support our a priori hypothesis. H11: Regarding the reassessment of the risk of falling, PTs who regularly assessed their patients over 65 years old also conducted re-evaluations more often than others (X² 8.9432, df 2, p = 0.0114). H12: In addition, PTs who systematically assessed their patients' risk of falling did 4.5 more reassessments per year than their colleagues who only “sometimes” assessed the risk (95% CI 0.33 to 8.65, adjusted regression).

Barriers affecting systematic fall risk assessment

H6: The main barriers to systematic fall risk assessment were the lack of technical resources (n=88, 39.8%), lack of theoretical knowledge (n=67, 30.3%), lack of time (n=49, 22.2%) and lack of financial recognition, i.e., reimbursement (n=35, 15.8%) (see Figure 6 and Table 5). Other barriers identified by PTs against fall risk assessment are presented in section 3.8 Content analysis (see Barriers against fall risk assessment and Barriers against systematic fall risk assessment).

Interventions to reduce risk of falling

PTs may undertake interventions linked with fall risk assessment. H8: PTs who assessed more often also provided more advice on risk factors. Indeed, whatever the risk factor for which advice was given, a higher value was observed among participants who "always" assessed their patients. However, statistically significant differences were not always observed (see Table 7). H13: Additionally, if a patient was found to be at risk, PTs informed them about the situation (n=500, 83.6%) and gave them exercises to do at home (n=486, 81.3%) or during physiotherapy sessions (n=486, 81.3%). In a similar situation, only a small percentage of PTs treated the patient just for the initial reason for consultation (n=4, 0.7%) (see Table 10).

Assumptions on the implementation of guidelines recommendations

H16: The opinion of the PTs regarding the implementation of a systematic assessment, i.e., of every new patient over 65 years of age, was questioned. Only 35% of the PTs agreed with this proposition (see Table 4). H17: A non-mandatory open-ended question was used to test this final hypothesis. The frequency and percentage of each extracted themes were calculated. Among the respondents, 181 (19.3%) desired a quick and easy-to-use tool, 49 (5.22%) more time for their initial assessments and 38 (4.05%) a checklist listing all the risk factors for falls to guide them in their assessment of patients over 65 years old. Moreover, 36 PTs (3.84%) would like to see the establishment of uniform and standardised procedures across the different health care providers in Switzerland (see Table 11).

Geographical distribution of fall prevention programme awareness

Figure 7 shows the percentages of PTs aware of at least one programme per two-digit zip-code region. The list of known programmes mentioned, and their respective websites is presented in extended data “Fall-risk Prevention Programmes” (Duc et al., 2022).

Exploratory analyses

A post-hoc hypothesis concerning the influence of the initial pathology on the assessment of the risk of falls was made during the statistical analysis. This result should be treated with caution.

To address the hypothesis that the assessment rate could be different in patients with musculoskeletal, neurological or respiratory pathologies, three hypothetical patient vignettes were used. In patients with musculoskeletal problems, 91% of the PTs assessed fall risk factors (99% of PTs in patients with neurological, and 74% in patients with respiratory pathologies). For the use of a standardised fall risk tool, only 11.3% of the PTs would have assessed patients with musculoskeletal problems, 48.4% of the PTs patients with neurological and 10% PTs patients with a respiratory disease). Thus, our post-hoc hypothesis that patients with neurological conditions are more frequently assessed for fall risk than those with other pathologies was confirmed.

Content analysis

Open-ended questions (n=13) allowed participants to express themselves freely on different aspects related to fall risk assessment. The ideas underlying some of the themes are developed below, supported by verbatim quotes in their original language. All themes extracted from the analysis of these open-ended questions are presented in extended data “Extracted Themes” (Duc et al., 2022).

Other situations leading to fall risk assessment

Other situations than those mentioned in the questionnaire may lead PTs to assess their patients' risk of falling. If this is not done during the initial assessment, the therapist might observe that the patient has balance problems when moving, exercising, or using walking aids. This may lead to a more detailed risk assessment. Return home after hospitalisation may also prompt PTs to assess the risk of falls of their patients.

Other standardised assessments used

Other assessments closely or remotely related to the risk of falls used by the PTs interviewed were: i) Functional Independence Measure (FIM), ii) Hierarchical Assessment of Balance and Mobility (HABAM), iii) Limit of stability test (LOS), iv) Modified Romberg Test (mRomberg), v) Multiple Sclerosis Questionnaire for Physiotherapists (MSQPT), vi) Romberg Test, and vii) The de Morton Mobility Index (DEMMI).

Static balance assessment

Static balance was assessed through various tests such as the One-Legged stance Test (OLST) or Single Leg Stance (SLS), or the 4-Stage Balance Scale Test. Some tests used such as the Berg Balance Scale (BBS) or the Balance Evaluation Systems Test (BESTest) examine both static and dynamic balance. PTs also mentioned testing balance on an unstable surface or with eyes closed without specifying whether this was done in a standardised way or not. Finally, the use of devices that also assess dynamic balance, such as the Huber 360 Evolution^® or the Galileo^®, was also stated. Participants who did not assess static balance gave reasons such as lack of knowledge, of specific tests, or of time during the initial assessment. For others, testing did not contribute to diagnosis and management and was therefore not considered necessary. A focus was placed on the reason for the initial consultation, thus favouring treatment over assessment.

Dynamic balance assessment

Dynamic balance was assessed using walking tests (6 Minutes or 10 Meter Walk Tests), mixed assessments such as the Tinetti Balance Assessment Tool (Tinetti/POMA) or the Short Physical Performance Battery (SPPB), and assessments whose standardisation remains uncertain such as walking on a line, backwards, or sideways for example. Devices such as the GaitUp System^® or the Dividat Senso^® were also mentioned. The reasons for not assessing dynamic balance were similar to those reported for static balance. However, the lack of space to carry out the assessment was also reported.

Muscle strength assessment

Muscle strength was assessed by Manual Muscle Testing (according to the Medical Research Council, Kendall, Janda, Daniels), using devices such as Legpress or dynamometers or by testing some muscle groups individually such as quadriceps, triceps surae, abdominal muscles, abductors, or gluteus. Some limiting factors to muscle strength assessment were the lack of suitable assessments or devices to quantify this. Some respondents said they did not trust the 0-5 manual muscle testing accuracy, which does not assess, for example, intra-muscular coordination.

Other risk quantification

Another way of quantifying the risk of falls was to follow the classification proposed by the test used. However, some participants pointed out that a fixed quantification of risk was impossible. Indeed, this one could depend on the use or not of walking aids, the fluctuation of the cognitive state, of the time of day, or of the environment.

Other advice provided in relation to the risk of falls

Advice regarding concentration during activities of daily living (ADL), foot sensitivity in diabetes, moving around at night, sedentary lifestyle and vestibular pathologies was also provided by PTs regarding fall risk.

Other measures undertaken if the patient is at risk

If a patient consulted for a reason other than rehabilitation to reduce the risk of falling and the assessment revealed that they were at risk, an additional measure undertaken was to encourage the patient to discuss this directly with their general practitioner. This helps to empower the patient while allowing them freedom to determine what measures they want to take (or not take) regarding this risk.

Barriers against fall risk assessment

Some barriers hindered some PTs from assess fall risk of their patients aged 65 and over for the risk of falling. One of these was the lack of sense in this approach. Indeed, a well-taken anamnesis, observation of the patient during exercises, during movements in the practice (waiting room - treatment room), when undressing or dressing, during functional exercises for example, or simply experience was according to them sufficient to assess the risk of falls.

«Through history and functional tests and inspection in the assessment, the risk is identifiable. » (Translated from German to English by the authors.)

«I believe that my experience is sufficient for an evaluation.” (Translated from French to English by the authors.)

In addition, many patients were deemed not to be at risk because their gait seemed safe (observation), or they practiced a sport in their leisure time and might therefore find it questionable to assess their risk of falling. This led some therapists to focus on the initial reason for consultation mainly.

Barriers against systematic fall risk assessment

Barriers given against systematic assessment of patients over 65 are various. First, the age criterion alone was not sufficient to justify a systematic risk assessment according to some participants. 75-80 years old would be more appropriate for a so-called "systematic" assessment. In addition, this supplementary evaluation was perceived to be too time-consuming.

"As I find the age limit for a systematic assessment of every patient of 65 years old too low, and we do not have the time and resources to clarify the risk of falls in all these patients regardless of the diagnoses and the A-Z. A large part of the patients I experience as still very fit and active at the age of 65." (Translated from German to English by the authors.)

For some, a case-by-case approach would therefore be preferred in the first instance to a systematic procedure. The patient should be assessed according to his level of activity, his pathologies, his anamnesis, his antecedents, the subjective and objective assessment elements, of the general impression for example. However, other participants considered that everyone should be assessed because "prevention is better than cure". Patients are often referred to PTs too late when the situation is already complex (generalised deconditioning, significant and/or disabling joint damage and cognitive disorders…).

Clinical implications

Despite the multitude of tests available, many participants would like a quick and easy-to-use assessment tool or a checklist listing all the risk factors for falls to guide them in assessing patients over 65 years old. The implementation in Switzerland of a tool directly targeting fall risk assessment such as the STEADI algorithm, could be relevant. Indeed, it was specially created to help caregivers to perform fall risk assessments and treat patients appropriately (Stevens & Phelan, 2013). Physiotherapists working in private practice, either alone or part-time, should particularly be supported to use standardised assessments. Improvements in basic education, continuing education courses, online materials, financial incentives (i.e., remuneration for assessments), and interprofessional prevention campaigns could also potentially increase the implementation of established best practices in fall prevention.

In addition, many therapists called for the standardisation of current practices between different health care providers. This would enable the adoption of a "common language" in terms of testing, risk quantification and interventions. According to Gaboreau et al. 65.3% of French GPs considered the implementation of an annual falls risk assessment useful. Unfortunately, only 28.8% of them performed it each year (Gaboreau et al., 2016). To optimise practices related to falls prevention, it is therefore essential to increase the awareness and involvement of each health professional. More generally, there is still room for improvement in the quality of care, coordination between the treating physician and the specialist, and follow-up care after discharge from hospital in Switzerland (Merçay, 2017).

Future studies

Future studies are needed to develop a fall risk assessment adapted to the needs and constraints of physiotherapist’s practice in Switzerland. This tool could be based on the STEADI, which was proven to reduce falls-related hospital admissions in older people and associated health care costs (Johnston et al., 2019). Then, local implementation studies including in-depth qualitative interviews with sub-groups of physiotherapists to identify potential barriers and facilitators could be planned. This tool could thus serve as a basis for discussion and joint decision-making among health professionals.

Limitations

Participant’s recruitment

One limitation was, that we could no address all, or a random sampling of all physiotherapists working in Switzerland. In the absence of a federal register comprising all currently practising physiotherapists in Switzerland, a recruitment strategy was developed to maximise the results generalisability and sample representativeness. The research team sent more than 3600 emails and the partner associations and schools to set up our sample. However, this strategy did not avoid the phenomenon of overlapping; obviously, some physiotherapists have been contacted in several ways while others were omitted. Moreover, physiotherapists who took part may have been more sensitive to this topic than non-respondents, potentially influencing the results. The generalisation of cross-sectional survey findings is often discussed in the literature. For some authors, sources of disparity within the population studied do not allow the results to be generalised between the individuals contacted and the others (Andrews et al., 2003). For others, due to the large number of individuals contacted, this design is more likely than others to acquire data from a representative sample and thus allow extrapolation (Kelley et al., 2003). We assumed that our sample is representative of all physiotherapists working in Switzerland but are not able to test this and therefore these findings should be interpreted with caution. Nonetheless, this study provides a snapshot of the current global situation in Switzerland regarding fall risk assessment, considering that the physiotherapist's practices and knowledge are in constant evolution.

A further limitation was, that few questions showed low reliability. However, these were multiple choice items with many response options. For example, the question that targeted different fall risk assessment tools had 81 different response options.

Open-ended questions analysis

A quantitative design was chosen to answer questions based on the current practices of physiotherapists in Switzerland in fall prevention. Without pretending to fulfil the criteria of a mixed design, but to allow participants to specify or clarify their answers, twelve open-ended questions were also asked. The analysis of the participants' answers was limited to a content analysis based on the inductive approach proposed by Elo and Kyngäs and Graneheim and Lundman (Elo & Kyngäs, 2008; Graneheim & Lundman, 2004). The lead author (MD) was responsible for summarising each response and then classifying them according to general recurring themes. These were then counted to determine their frequency. This work, although long and tedious, was fascinating. It raised relevant aspects concerning the needs of physiotherapists and the future reflections to be carried out to prevent falls in our country. However, it would have been relevant to carry out this work with the help of a third person to ensure that the answers were well understood and that the classification by themes was adequate and unanimous.

Strengths

First, to our knowledge, it is the first study to explore this topic in Switzerland. Fall prevention is of high importance to improve the care of our older patients. Guidelines have underlined the effectiveness of multifactorial assessment to determine an individual’s risk of falling and implement suitable prevention strategies (American Geriatrics Society & British Geriatrics Society, 2011; Beauchet et al., 2011; Feder et al., 2000; Moreland et al., 2003; National Institute for Health and Care Excellence, 2013). It was, therefore, necessary to review current practices and to propose measures according to physiotherapists needs. Furthermore, our recruitment strategy and the development of a questionnaire in three languages (French, German, Italian) enabled nationwide participation, with the final sample size exceeding our initial expectations. In addition, this study led some participants to reflect on their current practices.

Underlying data

Open Science Framework: Current Practices of Physiotherapists in Switzerland Regarding Fall Risk-Assessment for Community-Dwelling Older Adults: A National Cross-Sectional Survey, https://doi.org/10.17605/OSF.IO/MP9U4 (Duc et al., 2022).

This project contains the following underlying data:

Extended data

Open Science Framework: Current Practices of Physiotherapists in Switzerland Regarding Fall Risk-Assessment for Community-Dwelling Older Adults: A National Cross-Sectional Survey, https://doi.org/10.17605/OSF.IO/MP9U4 (Duc et al., 2022).

This project contains the following extended data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).