Slip-resistant footwear to reduce slips among health-care workers: the SSHeW RCT

Article history

The research reported in this issue of the journal was funded by the PHR programme as project number 15/05/28. The contractual start date was in October 2016. The final report began editorial review in December 2019 and was accepted for publication in April 2020. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PHR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2021 Cockayne et al. This work was produced by Cockayne et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2021 The authors

Slips in the workplace and the role of slip-resistant footwear

Injury-reporting systems rarely differentiate between slips, trips and other causes of falls on the same level, but it is thought that slipping is a major contributor. Courtney et al. 12 reviewed six reporting systems in the USA, the UK and Sweden, and found that just three (in the USA and Sweden) provided sufficient information to attempt to isolate the contribution of slipping to the overall burden of STFL. The authors12 suggested that slipping contributed to between 40% and 50% of STFL-related injuries.

Workplace slips and their resulting falls happen for a number of reasons. Factors that can contribute to slips in the workplace include the floor surface, the presence and characteristics of any contamination of the floor (wet or dry), the cleaning regime, the level and type of pedestrian activity, footwear choice, the working environment (e.g. lighting or noise levels) and human factors (how people interact with their working environment). 13 A multitude of factors may be present at the same time and so it is rarely just one factor that causes a specific slip incident.

In the UK, employers have a duty to assess risks (including slip risk) and, when necessary, take action to address them. This is done through a risk assessment, when the employer considers what workplace risks may lead to slip injuries and decide what suitable and effective control measures could prevent them. In many instances, straightforward measures can readily control risks (e.g. ensuring that the floor surface is kept clean and dry or by installing a floor surface that is sufficiently slip resistant when contaminated). However, when it is not practicable to prevent the floor surface becoming slippery, employers may consider the use of slip-resistant footwear.

The choice of slip-resistant footwear is important and the selection of inappropriate footwear (i.e. with poor slip resistance) could potentially add to the problem rather than provide a solution. Choosing appropriate footwear can be difficult, as there is a lack of robust testing and reliable information to help guide the decision-making process.

It has been suggested that testing footwear under more lifelike conditions will provide a more accurate assessment of the slip-resistant properties of the footwear. 14,15 In response to this, and to help guide the selection process, the HSE developed the ‘GRIP’ rating scheme. 16 This scheme utilises a test regime designed to reflect the biomechanics of walking, and categorises footwear in accordance with the level of slip resistance it provides on a smooth surface contaminated with (1) water and (2) a 75% volume per volume glycerol solution (aqueous). ‘GRIP’ star ratings, ranging from one to five stars, are assigned to the footwear soling units on the basis of the results obtained, with five stars being awarded to those with the highest level of slip resistance. The risk assessment process can help guide the employer on which star rating is required for their workplace.

Current evidence relating to the use of slip-resistant footwear

A number of publications have described studies investigating the influence of footwear on slip risk using a range of laboratory tests. The evidence suggests that the level of slip resistance provided by different types of footwear varies considerably and that the effectiveness of footwear at preventing slips and falls depends on factors such as the footwear material, the tread design, the type of contamination and the characteristics of the floor. 17–19

The HSE has developed tests to accurately assess the slip resistance of footwear and has helped to inform the footwear selected by some companies that have subsequently seen a reduction in accidents and personal liability claims. For example, one case study20 showed how the test method could be used to remove inadequate footwear from use and identified alternatives. As a consequence, the risk of workplace accidents was reduced. 20 However, these are individual, non-peer-reviewed before-and-after case studies and the findings are not in the context of a randomised controlled trial (RCT).

There is, however, some evidence to support this suggestion in studies that have measured a reduction in slips occurring in the workplace by the use of appropriate slip-resistant footwear. Bell et al. 21 evaluated a comprehensive slip, trip and fall prevention programme for hospital employees in the USA. The programme included slip-resistant footwear for certain employee subgroups, but also included many other prevention strategies, such as hazard assessments, changes to housekeeping, general awareness campaigns and flooring changes. The study suggested that the programme reduced workers’ compensation claims for slips, trips and falls by 58%, but it was not possible to disentangle the specific effect of slip-resistant footwear from the other preventative interventions. 21

An uncontrolled before-and-after study among fishermen in Denmark suggested that slip-resistant boots led to a reduction in self-reported slips, trips and falls. 22 In contrast, a cross-sectional study23 of 125 fast-food restaurant workers in the USA did not find an association between slipping and use of footwear perceived to be slip resistant. The authors suggested that the null finding could have been due to the method of classifying footwear as slip resistant (through visual inspection of the tread), possibly leading to misclassification and the potential for reverse causality (people who had experienced a slip or fall in the past being more likely to wear slip-resistant footwear).

More recently, in the USA, a prospective cohort study24 of 475 fast-food restaurant workers found that the use of slip-resistant footwear was associated with a 54% reduction in the reported rate of slipping. A case-crossover study nested within the same cohort found that rushing and walking increased the rate of slipping, but that use of slip-resistant footwear reduced the effects of rushing and walking on slip risk when on a contaminated floor. 25 Further analysis of the cohort indicated that slip-resistant shoes worn for < 6 months were marginally more effective at preventing slips than slip-resistant shoes worn for > 6 months. 26

We are aware of only one RCT27 assessing the effectiveness of slip-resistant footwear. This cluster randomised trial of food service workers from 226 school districts in the USA found that the probability of a slipping injury fell by 67% in the intervention group, who received slip-resistant footwear. By contrast, the control group (who wore slip-resistant footwear that they purchased themselves in accordance with company policy) saw a slight increase in the probability of slipping injury during the follow-up period.

Overall, there seems to be promising evidence that slip-resistant footwear can substantially reduce the burden of accidents at work, but much of this evidence has been gathered for footwear with slip-resistant properties that are ill-defined. It was therefore considered important to test the clinical effectiveness and cost-effectiveness of appropriately specified slip-resistant footwear in a large pragmatic trial within a UK setting.

Trial setting

It is estimated that the health and social care sector employs 1 in 10 of the working population28 and has one of the highest numbers of non-fatal STFL compared with other industries in GB. It is estimated that every year around 14,000 workers in this industry sector are injured as a result of STFL [Health and Safety Executive. Estimated Incidence of All Self-Reported Workplace Non-Fatal Injury Resulting in a Slip, Trip or Fall on the Same Level, Sustained in Current or Most Recent Job, By Industry, for People Working in the Last 12 Months, Averaged 2014/15–2016/17 (Unpublished data). London: Labour Force Survey; 2018].

The NHS is the biggest employer in the sector and the fifth biggest employer in the world. 29 NHS sites, particularly hospitals, present challenging working environments with respect to slip risk. For example, the floor is generally smooth and not carpeted to aid regular cleaning, there are multiple sources and types of contamination, there is a relatively high level of pedestrian activity and of varied types (e.g. walking, pushing and pulling) and people frequently walk at speed, often multitasking or focused on a more critical objective than the environment around them. Many large hospital sites are a conglomeration of buildings and so people often switch between indoor and outdoor environments, and community staff are mainly offsite with no control over their surrounding environment. In such circumstances, it would seem potentially beneficial for staff to consider wearing slip-resistant footwear.

It is for these reasons that NHS staff were considered to be a good population on which to base research into the potential effectiveness of appropriately specified (in this study, 5-star, GRIP-rated) slip-resistant footwear. In addition, the NHS includes different working subpopulations (e.g. nursing staff, kitchen staff, porters and people who work in the community). This will help to generalise any research to other industries that share many of the same risk factors, such as retail, hospitality, education and manufacturing, and maximise the impact of the research. One other potential benefit of undertaking the trial within the NHS is that its staff are generally used to the research environment although it is usually patients rather than the staff who take part in research. However, it seemed likely that, for this reason, participation in this research would be higher among NHS staff than among those of other employers.

Objectives

Study design

The Stopping Slips among Health-care Workers (SSHeW) trial was a pragmatic, two-arm, open RCT with an internal pilot trial, economic evaluation and qualitative study. Participants were randomised to either the intervention group and were offered a pair of 5-star, GRIP-rated, slip-resistant shoes to wear at work (free of charge to themselves), or to the waiting list control group and were asked to wear their usual work footwear for the trial period and offered a pair of the slip-resistant shoes at the end of their participation in the trial (free of charge to themselves). The trial included an economic evaluation (see Chapter 4) and a nested qualitative study to explore the experiences of participants in the trial (see Chapter 5). The SSHeW trial also served as a host trial for an embedded study within a trial (SWAT) to evaluate the effectiveness, on response rate, of enclosing a pen with the 14-week post-randomisation postal follow-up questionnaire.

Ethics approval and research governance

Ethics approval for the study was obtained from the University of York, Department of Health Sciences Research Governance Committee (DHSRGC), on 28 November 2016. The study was approved by the Health Research Authority (HRA) on 7 March 2017 (HRA reference number 17/HRA/0435). The HRA did not require Research Ethics Committee approval for the study, as the participants were NHS staff (as opposed to NHS patients). Approval and ‘confirmation of capacity and capability’ was obtained for each participating trust prior to commencement of the trial at that site (see Appendix 1). Substantial amendments to the protocol were submitted to the DHSRGC, the HRA and each site’s research and development office during the course of this study.

The trial was registered with the International Standard Randomised Controlled Trial Number registry as ISRCTN33051393. The trial protocol is published. 1

Setting

Recruitment of participants took place in seven NHS trusts in England. The participating trusts ranged from large acute hospital trusts to smaller trusts, and included district general hospitals serving local communities, a trust that provided mental health services and a community trust.

Recruitment procedure

Recruitment of participants into the study took place in multiple sites within participating NHS trusts, facilitated by members of the research and development teams, and assisted by members of the HSE team and University of York researchers in some sites. Sites used a variety of recruitment strategies to recruit participants, which were developed during the pilot phase of the study. These included holding ‘shoe shops’ (where interested staff could learn more about the study, and see and try on samples of 5-star, GRIP-rated, slip-resistant footwear) and stands promoting the trial in convenient hospital locations, such as reception foyers or the canteen, or at staff conferences or induction days. Other recruitment activities included promotion of the trial via posters, information on the staff intranet, screensavers and weekly staff bulletins. Direct staff recruitment occurred by members of the research team attending staff meetings and ward areas, particularly at shift handover time, and via word of mouth as participants discussed the trial with colleagues who then approached research staff to discuss participation.

At recruitment, research staff provided information about the study and demonstrated the 5-star, GRIP-rated, slip-resistant shoes being tested in the trial. Sample styles and sizes were available for interested staff to try on. Research staff briefly assessed eligibility of interested staff by asking if they were employed by the recruiting trust, their area of work and the average number of hours they worked per week, before giving them a recruitment pack. The recruitment pack contained an invitation letter from the principal investigator at the NHS trust (see Report Supplementary Material 1), a participant information sheet, which included contact details for the research team should they have had any queries about the study (see Report Supplementary Material 2), a flier with pictures of the shoe styles being offered at the trust (see Appendix 2 for an example of this document offered at one trust), a baseline questionnaire including more detail on the eligibility criteria (see Appendix 3), a consent form (see Report Supplementary Material 3) and a prepaid envelope for returning the completed paperwork.

Consent process

Participation in the SSHeW trial was voluntary. Participants who wished to take part in the trial after receiving information about the trial from a member of the research team and/or reading the participant information sheet provided in the recruitment pack were asked to complete the consent form by initialling that they had read the items listed, and signing and dating the form.

The consent form also asked if participants were willing to take part in a telephone interview about the study (for the qualitative component of the study, further consent was obtained from the participant by post prior to undertaking a telephone interview). Participation in the SSHeW trial was not affected if the member of staff did not consent to participate in a telephone interview.

Baseline assessment

On receipt of completed written consent, researchers at the York Trials Unit (YTU) assessed the participants’ responses to the baseline questionnaires for eligibility in accordance with the following criteria.

Participant eligibility

Pre-randomisation phase

After the welcome message was sent, all eligible participants were sent a text message each Sunday at 18.30 for 4 weeks, unless they requested that these be stopped. We chose to send text messages at this time as we anticipated that fewer participants would be at work, which would lead to higher response rates. This message read:

SSHeW trial: How many slips did you have at work between [DD/MM/20YY] and [DD/MM/20YY]? Please provide a single number (e.g. 2) or 0 if you did not slip. Thank you.

The dates were populated with the previous Monday and the current date. Those who provided a valid response (a numerical value or written number, rather than a written text response or other message, such as a symbol) to at least two of the slip data collection text messages were randomised into the trial.

It became apparent during the trial pilot phase that some eligible participants had misunderstood the trial process. Instead of providing slip data in response to their text messages, they reported not having received their trial shoes as they mistakenly thought that they needed to report slip data only once they had received their trial shoes. Therefore, the protocol was amended (substantial amendment 6, approved 20 June 2018) to allow two additional text messages to be sent to participants who had not responded to their pre-randomisation text messages, along with a reminder that they needed to reply to these messages if they wished to be included in the trial.

Randomisation

Participants who fulfilled the eligibility criteria, provided written consent to take part in the study, completed a baseline questionnaire and returned at least two of the weekly text messages providing slip data were eligible for randomisation. Participants were randomised using the YTU’s secure web-based randomisation system based on an allocation sequence generated by an independent data systems manager at the YTU. The systems manager was not involved in the recruitment of participants. Block randomisation, stratified by trust, was used with variable block sizes. Participants were randomised at a particular site in batches of 1 : 1 between the intervention and control group (depending on when sites had capacity to order and deliver footwear). The block size was equal to the number of participants to be randomised in each ‘batch’. Participants were notified of their group allocation by letter from the YTU.

Blinding of trial participants to group allocation was not possible because of the nature of the intervention. Members of the study team, including the statistician, health economist and those involved in the administration of the study, were not blinded to group allocation.

Group allocation

Intervention

Individuals allocated to the intervention arm were offered a pair of 5-star, GRIP-rated, slip-resistant shoes. The footwear was provided by a single manufacturer [Shoes For Crews (Europe) Ltd] and the cost covered by the NHS trust. A range of intervention shoes were offered as part of the trial. The shoes included a variety of styles, to suit different people and job roles, including dress shoes, casual styles and ‘trainer’-type designs. The shoes had either a water-resistant leather or synthetic mesh upper (the synthetic materials being vegan). Each individual trust reviewed the footwear available in terms of style; to offer a range of footwear representative of that already worn by staff; health and safety issues, for example potential for needlestick injuries; potential infection control issues; and cost. They then produced a list of footwear which was deemed suitable for staff at their trust to wear. All participants were asked to indicate on their baseline questionnaire, prior to randomisation, which style and size of trial shoe they would like to wear. This facilitated swift ordering of the shoes by the trust for participants subsequently randomised to the intervention group. The indicated shoes were ordered for the participant, who was then provided with instructions about how to collect their trial shoes. Further information was provided to participants about how to exchange their shoes if required. (Participants were able to exchange their shoes if they were found to be unsuitable, for example incorrect size, as long as the shoes were unworn and returned in their original box.) We had hoped that this process would facilitate shoes being available for intervention participants to collect no later than 2 weeks after randomisation; however, we know that in some sites, and at certain times, there were delays in ordering and/or delivery of the trial shoes. In addition, we know that not all intervention participants collected their shoes if they were delivered to a ‘collection point’ at the hospital, and those that did may not have done so in a timely manner. More information on receipt, and timing of receipt, of trial footwear is provided in Chapter 3, Receipt of shoes.

The shoes used in the trial had been assigned a 5-star GRIP rating from the HSE. The GRIP rating scheme was developed by the HSE to allow footwear users to identify slip-resistant footwear to help prevent slipping accidents. During GRIP-rating assessment, the soling of footwear is tested under very challenging conditions (i.e. a smooth surface contaminated first with water and then with a glycerol solution). Slip resistance of shoes is rated one to five stars on the basis of the results, with five stars being awarded to the highest performing footwear. The shoes used as the intervention in the trial had a rubber sole with an intricate tread pattern that provided higher levels of slip resistance than most other types of footwear on surfaces that are contaminated with non-clogging substances (Figure 1). This level of slip resistance is achieved by the soles providing a relatively large surface contact area, while also effectively dispersing any surface contamination. When walking surfaces are clean and dry the trial footwear would be expected to behave like any other footwear; however, their slip resistance is not compromised to the same extent as other footwear when walking on surfaces that are contaminated.

FIGURE 1.

Example of sole of intervention footwear. Photograph received and reproduced with permission from Shoes for Crews (Europe) Ltd.

Photograph received and reproduced with permission from Shoes for Crews Ltd.

Participant follow-up

All participants in the SSHeW trial were sent weekly text messages for 14 weeks post randomisation to collect slip data (i.e. the number of slips the participant had experienced at work in the past week). As with the pre-randomisation messages, these were sent on a Sunday evening (18.30) and the wording was the same, with dates inserted relative to the past week:

SSHeW trial. How many slips did you have at work between [DD/MM/20YY] and [DD/MM/20YY]? Please provide a single number (e.g. 2) or 0 if you did not slip. Thank you.

These messages started, for all participants, the first Sunday after they were randomised, regardless of whether or not the intervention participants had received their shoes.

Non-responders did not receive reminder text messages, as there was limited time to send these because of the short time frame of 1 week before the next text message was sent. Participants were also sent a group-specific postal questionnaire at 14 weeks post randomisation to collect data on slips, falls, injuries, time off work, health-care resource use, if participants were worried about slipping and falling, how often they had worn the trial shoes while at work and if participants had any views on trial shoes (intervention group only) (see Appendices 5 and 6). Participants who did not return this questionnaire within 3 weeks were sent a postal reminder. Participants who reported a slip in their weekly text response were sent a questionnaire for further details regarding their first reported slip, including whether or not they were wearing the trial shoes when they slipped, any resultant injury, health service use and time off work. Furthermore, participants reporting an injury were sent an injury follow-up questionnaire.

Wear testing

The soles of all footwear degrade over time and so it was important to have a concept of how long the slip resistance of the trial footwear lasted in a real-world setting. This information could help determine any replacement schedule, and results from this fed into the health economic analysis for the trial.

To assess the serviceable life of the intervention footwear, we asked eligible participants from three trusts who had continued to wear their trial shoes beyond the trial period to return them for assessment. Participants from the intervention group who had been in possession of the trial footwear for 6, 9 or 12 months since randomisation, and had reported wearing the footwear either all of the time or most of the time in response to the 14-week questionnaire, were contacted to establish if they had continued to wear the footwear. Those who had were then asked if they would be willing to return their trial footwear in exchange for a new pair. The exchange of footwear was co-ordinated by the participating trusts.

The soles of each pair of footwear were cleaned, rinsed and allowed to dry prior to the assessment. A visual inspection was carried out to assess the condition of the footwear and identify any defects. Each pair of footwear was inspected by one member of the HSE research team and a HSE colleague who was not involved in the study to provide layperson corroboration. The condition of the upper was rated as good, reasonable or poor, and any defects were recorded. The depth of tread was measured to determine the approximate proportion of sole area that had tread depth below the 2 mm minimum recommended by the manufacturer.

The slip resistance of the worn footwear was then tested on the HSE’s simulated slip test31 and the results compared with those generated for the same styles of footwear when new. Each pair was tested in line with the conditions set out in the HSE GRIP scheme handbook. 32

Outcomes

Sample size

Parts of this text have been reproduced with permission from Cockayne et al. 2 This is an Open Access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

There were limited published data on which to base a sample size calculation for this trial. A prospective cohort study24 found that 49 of 422 (11.6%) workers in a restaurant setting in the USA reported at least one ‘major’ (i.e. resulting in a fall and/or injury) slip over a 12-week follow-up period. The exact number who experienced any type of slip was not reported in that study, but could reasonably be expected to be higher than this. For our sample size calculation, we required an estimate of the proportion of individuals in the control group who would experience at least one slip over a 14-week follow-up period; we conservatively assumed a proportion of 10%. We proposed to randomise 4400 participants using 1 : 1 allocation (i.e. 2200 per group) to have 90% power to show a 30% relative reduction in the proportion of participants who report at least one slip over a 14-week period (a 3-percentage-point absolute reduction from 10% to 7%), allowing for 20% attrition. This sample size would also give us 80% power to see an absolute reduction of 2 percentage points in the risk of falls, from 5.5% to 3.5%, allowing for 20% attrition. Although we based the sample size calculation on detecting a difference in proportions, the primary outcome is the incidence rate of slips over the 14 weeks and so we used a mixed-effects negative binomial regression model to compare this outcome between the two groups. As this analysis uses more information than a simple binary outcome, we expected it to still be adequately powered.

Trial completion and exit

Participants completed the trial if they completed the 14-week trial period post randomisation. Participants exited the trial if they had completed the 14-week follow-up period, had fully withdrawn from the trial (no further follow-up by text or post), were lost to follow-up or had died.

Participant change of status and participant withdrawal

Participants could withdraw from the trial at any time without giving a reason. If, however, a reason was provided, then it was recorded. If we were informed that a participant wanted to withdraw from the trial, a member of the research team would attempt to clarify to what extent they wished to withdraw (i.e. fully or partially). Participants could partially withdraw from the trial in two ways; first, by withdrawing from the intervention only, meaning that they would not wear the trial shoes but continue to provide text and questionnaire data; and, second, participants could withdraw from data collection either by declining to respond to text messages or by declining to complete the follow-up questionnaires. Data were retained for withdrawn participants unless a participant requested specifically that their data be removed. A change of circumstance form was completed by a member of the research team for any participant changing status during the trial (see Report Supplementary Material 4).

Adverse events

Adverse events (AEs) were reported by participants to the research team at the YTU via telephone or text message or by writing free-text comments in the postal questionnaires. AEs could also be reported through the local research teams at the participating sites. Details of AEs were recorded on a SSHeW AE form (see Report Supplementary Material 5). The AE reporting period started at the point when the participant gave their consent to be in the trial and ended at 14 weeks after they were randomised.

Serious AEs that were related to being in the study and were unexpected were recorded. Non-serious AEs were not recorded, unless they were related to being in the study or were related to the intervention.

A serious AE was defined as any untoward occurrence that:

• resulted in death

• was life-threatening

• required hospitalisation or prolongation of existing hospitalisation

• consisted of a congenital anomaly or birth defect

• was otherwise considered medically significant by the investigator.

The SSHeW internal pilot trial

An internal pilot trial was conducted during the first 6 months of the study (March–September 2017). The objectives of the pilot trial were to:

• test and refine recruitment strategies for the study

• check the sample size calculation assumptions by reviewing the proportion of participants in the control group who experienced a slip

• check the attrition rate

• explore and address any issues regarding footwear compliance.

To progress to the main SSHeW trial, the pilot phase had to meet the following criteria:

• at least 400 participants recruited in 6 months

• 80% of participants contributing at least 50% of the requested follow-up text data (i.e. responding to 7 of the 14 weekly post-randomisation text messages)

• 90% of participants responding to at least one post-randomisation text message

• a slip rate in the control group of at least 7%.

Data analysis

Primary analysis

The difference in slip rate between the intervention and control groups over 14 weeks was analysed using a mixed-effect negative binomial regression model, adjusting for gender (three categories: male; female; prefer not to say or missing), age at enrolment (continuous) and job role [four categories: administration and information technology (administrator/receptionist/secretarial, ward clerk); facilities workers (catering, porter); direct patient care (doctor/consultant, qualified nurse/midwife, physiotherapist, occupational therapist, health-care assistant, pharmacist/pharmacy technician, social worker, support worker, podiatrist, other qualified staff/health-care professional); other/missing (imaging, other)], and baseline weekly slip rate ascertained from the pre-randomisation text message responses as fixed effects. Trial site was included as a random effect to account for potential clustering by recruitment site. The total number of hours that participants worked in the weeks for which they provided slip data was calculated (by multiplying the number of weeks the participant provided valid post-randomisation text message slip data by the number of hours they indicated they worked, on average, per week on the baseline questionnaire). This amount of time was accounted for in the negative binomial model (using the exposure option in Stata).

The incidence rate ratio (IRR) and associated 95% CI and p-value for the treatment effect are presented.

This analysis primarily included slip data from the weekly text message responses. When no post-randomisation responses were returned, data from the 14-week questionnaire were used for a participant, when available.

The primary analysis included all participants who had full baseline covariate data. The model could accommodate participants who did not provide a valid response to any weekly text messages or the 14-week questionnaire by considering that they reported zero slips over a negligible exposure time of 0.1 hours.

The primary analysis was checked by a second statistician (GF), who reviewed the analysis syntax, the derivation of variables required for the analysis, and the reporting and interpretation of the results.

Sensitivity analysis

Secondary analysis

All secondary outcomes and other important collected data (including data from the SDCSs) are summarised descriptively overall and by trial arm.

The IRR of falls (both resulting and not resulting from a slip) over 14 weeks was analysed in the same way as described for the primary outcome of slips (not including the exposure option in the model as the follow-up is time fixed).

The following outcomes were analysed using mixed-effects logistic regression, adjusting for the same fixed-effect covariates as the primary analysis, with NHS trust as a random effect. The first outcome was the proportion of participants who slipped at least once over 14 weeks (two separate analyses: one according to weekly slip text message data or 14-week questionnaire if no text message data were provided, and one using only 14-week questionnaire data), and the second was the proportion of participants who fell at least once during the 14 weeks. The number of participants who reported a fracture was very small and so no formal analyses of this outcome were undertaken (see Chapter 4). Odds ratios (ORs) and their associated 95% CIs are provided.

Time to first slip and the time to first fall were calculated. Participants who did not report a slip were treated as censored at their date of trial exit (completion of follow-up or withdrawal). The proportion of participants yet to experience a slip is summarised by a Kaplan–Meier survival curve for each group. Time to first slip was analysed using Cox proportional hazards regression with shared centre frailty and adjusting for the same covariates as in the primary analysis model. Hazard ratios (HRs) and their associated 95% CIs are provided. The proportional hazards assumption was evaluated using Schoenfeld residuals. Time to first fall was not analysed via a Cox proportional hazards regression model for two reasons: (1) relatively few falls were reported and (2) the date of the fall was poorly reported for any falls that were reported.

Adverse events

Adverse events are summarised descriptively by treatment arm.

Patient and public involvement in research

The SSHeW trial was informed by the involvement of NHS staff from diverse roles. These included nurses, health-care assistants, radiographers, physiotherapists, catering staff, and maintenance and housekeeping staff of different ages (age range 20–71 years) and genders. The study design was discussed with ward managers at their ward management task and finish group meeting and among NHS staff. The staff provided feedback about the rationale for the trial, the range of shoe styles to be offered in the trial, the use of text messages to collect data, the use of a slip diary and the length of the follow-up period. We had originally planned to follow-up participants for 12 months. However, the patient and public involvement (PPI) group felt that 12 months was too onerous and that participants would not provide outcome data for this length of time. We therefore reduced this to 3 months (12 weeks), but added an additional 2 weeks to allow for the ordering and delivery of shoes. During the trial, further views on the footwear, footwear buying habits and testing of staff’s usual shoes to determine slip resistance took place. In total, 30 pairs of shoes were tested.

Changes to the trial protocol

Minor changes to the trial protocol were submitted during the trial. These are listed in Appendix 10.

Study within a trial: pen substudy

To add to the body of evidence relating to making trial processes more efficient, a SWAT33 was conducted. The aim of this SWAT was to evaluate the effectiveness of enclosing a YTU, University of York pen with the 14-week postal questionnaire, on the response rates to the questionnaire.

Participants in the SSHeW trial who were to receive their 14-week questionnaire were randomised using simple randomisation in the ratio of 1 : 1 (receive a pen or not receive a pen). Generation of the allocation sequence was conducted by the trial statistician, who was not involved with sending the questionnaires.

The pen substudy was embedded into the trial part-way through recruitment. All participants in the main trial who were to be sent their 14-week questionnaire between 4 July 2018 and 12 February 2019 were included in this SWAT (unless they had withdrawn from follow-up). Participants in the main trial who had already been sent their 14-week questionnaire before the SWAT was introduced were excluded. No pens were included with questionnaires sent after 12 February 2019.

The primary outcome measure for the pen substudy was the proportion of participants in each group who returned the questionnaire. Secondary outcome measures included length of time taken to respond to the questionnaire, number of items completed and whether or not a reminder notice to return the questionnaire was required. As is usual with an embedded trial, no formal power calculation was undertaken, as the sample size was constrained by the number of participants to be sent the 14-week questionnaire. Binary data were compared using logistic regression, time to response by a Cox proportional hazards model and number of items completed by a linear regression model. All models were adjusted for the main trial allocation.

Pilot phase

The findings from the pilot phase of the trial evaluating the progression criteria are presented in Table 1. As all of the progression criteria were met, the trial continued seamlessly.

Recruitment

Participants were enrolled into the SSHeW trial from seven NHS trusts in England: (1) Cheshire and Wirral Partnership NHS Trust (CWP), (2) Leeds Teaching Hospitals NHS Trust, (3) York Teaching Hospital NHS Foundation Trust, (4) Lancashire Care NHS Foundation Trust, (5) Nottingham University Hospitals NHS Trust, (6) Harrogate and District NHS Foundation Trust and (7) University Hospitals of Derby and Burton NHS Foundation Trust. York Teaching Hospital NHS Foundation Trust was split into two geographical locations. Theses were considered as two distinct populations (Scarborough and Bridlington areas, and York and Malton areas) and formed eight trial sites. A total of 8524 recruitment packs were estimated to have been handed out in the eight sites between March 2017 and November 2018 (Figure 2), with follow-up occurring until June 2019. Varying numbers of recruitment packs were sent to the sites from YTU to be handed out to potential participants, depending on the size of the site and its capacity. The number of unused packs was reported to the YTU at the end of the trial. The number of packs handed out was estimated at 8524 by subtracting the number reported unused from the number originally sent to the site, on the assumption that the rest were distributed (CWP, n = 2029; Leeds, n = 1686; York, n = 642; Scarborough, n = 775; Lancashire, n = 1251; Nottingham, n = 1240; Harrogate, n = 401; and Derby, n = 500).

FIGURE 2.

A CONSORT flow diagram of participants in the SSHeW trial. a, n = 1930 (85%) received trial shoes within the 14-week trial follow-up. Reproduced with permission from Cockayne et al. 2 This is an Open Access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Cockayne et al. 2 This is an Open Access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

A completed baseline form was received at YTU from 5309 (62.3%) potential participants, of whom 498 (9.4%) were ineligible. Reasons for ineligibility were not mutually exclusive. One of the most common reasons was not working enough hours, on average, per week. Initially, the inclusion criteria specified that potential participants had to work, on average, at least 80% WTE (30 hours/week, assuming 37.5 working hours/week as full time). When it was observed that this was a predominant reason for ineligibility, it was, with the agreement of the TSC, DMEC and the funder, decided to lower this to 60% WTE (22.5 hours/week).

A total of 4811 participants were eligible to be sent the pre-randomisation text messages. Three participants withdrew before these text messages were started (one participant moved out of the trust, one participant was discovered to be a duplicate and one participant had an issue with their mobile telephone number that was never resolved), and 254 participants did not provide a valid response to at least two text messages. Therefore, 4554 participants were randomised (a slight increase of 154 participants on the 4400 planned). One participant was discovered to be ineligible after randomisation and so was immediately withdrawn (their data are not included in summaries of randomised participants). The flow of participants is illustrated in a CONSORT diagram (see Figure 2).

Pre-randomisation text message response rates

A total of 4808 participants were deemed eligible for the trial and were sent at least one pre-randomisation text message. Participants could request that the text messages be stopped at any time and so not all were sent at least four messages. Send and response rates to the initial four text messages were all > 86% (Table 2). A total of 4605 (95.8%) participants responded to at least one text message. A total of 260 participants were sent a fifth text (108 replied, 41.5%) and 245 were sent a sixth text (103 replied, 42.0%). Overall, 87 of 245 (36%) replied to both messages.

All but one of the 4555 participants who replied to at least two pre-randomisation text messages (either two of the first four messages, or both the fifth and sixth messages) were randomised into the trial (one participant withdrew prior to randomisation).

Randomisation

Our required sample size was 4400 participants, which we aimed to achieve by the end of April 2018. However, recruitment was slightly slower than anticipated and we agreed with the trial sponsor, funder and HRA to extend recruitment to late 2018 to meet the target.

The first participants were randomised on 1 June 2017 and the last on 10 January 2019. Over this 20-month period, 4554 participants were randomised into the trial (2276 to the intervention group and 2278 to the control group); however, one participant (intervention group) was randomised in error, so their data have not been used post randomisation. A median of 529 (range 299–1014) participants were recruited from each site. Figure 3 presents the target compared with actual recruitment.

FIGURE 3.

Target vs. actual participant recruitment.

Baseline data

Baseline data are presented by randomised group and overall in Tables 3–6. One participant, allocated to the intervention group, withdrew shortly after they were randomised and requested that all their data be removed; they are included in the denominators in Tables 3–6, but all their data are missing.

The recruited participants were predominantly female (n = 3869, 85.0%) and the average age was 42.7 (range 18–74) years. Participants worked a median of 37.5 hours per week and qualified nurse or midwife was the most represented job role (n = 1937, 42.5%). One participant who reported working 15 hours per week (less than the required amount) was randomised in error into the trial. Options to manage this participant were discussed with the TSC and DMEC and it was agreed that this participant should be retained in the trial and their data included in the analyses (as the impact of this one participant would be negligible).

Just over one-third of participants reported experiencing a slip at work in the previous 12 months (median of two slips), of whom 11% had suffered an injury as a result of one of these slips.

Via the pre-randomisation slip text messages, a total of 3891 slips were reported (2035 from participants who were subsequently randomised to the intervention group and 1856 from participants subsequently randomised to the control group). Both groups provided pre-randomisation text message slip data for an average of 3.4 (SD 0.8, median 4, range 2–4) weeks. Overall, 1627 (35.3%) participants (of the 4605 participants who responded to at least one text) reported at least one slip in the pre-randomisation text messages. The average number of reported slips was 0.89 (SD 2.00, median 0, range 0–22) in the intervention group and 0.81 (SD 1.76, median 0, range 0–26) in the control group. The baseline slip rate was calculated as the average weekly number of slips: 0.27 (SD 0.59, median 0, range 0–7) in the intervention group and 0.24 (SD 0.51, median 0, range 0–6.5) in the control group.

The treatment groups, as randomised, appear to be comparable on all measured baseline data.

Post-randomisation text message response rates

Following randomisation, participants were sent up to 14 weekly slip text messages. At any time, participants could request not to be sent any further messages. Response rates to the post-randomisation text messages are presented in Table 7 by group allocation. Of the 4553 participants sent at least one post-randomisation text message, a valid response to at least one message was received from 4494 (98.7%). Participants in the intervention group responded to an average of 12.3 (SD 3.1, median 14) text messages and participants in the control group to an average of 12.6 (SD 3.0, median 14) text messages.

Withdrawals

There were 35 (0.8%) full, post-randomisation withdrawals [n = 26 (1.1%) from the intervention group, n = 9 (0.4%) from the control group]. The majority (n = 28) were requested during the 14-week follow-up period (median of 35 days after randomisation, range 6–93 days). Reasons for these withdrawals included participant left the trust or was on long-term absence (e.g. sickness, adoption leave) (n = 15); did not want to take part, issue with trial shoes (n = 7); unhappy with sizing and/or appearance of shoe (n = 5); shoes were causing trips and blisters (n = 1); saw colleague trip on carpet in trial shoes and participant works in similar area and concerned about tripping (n = 1); and unknown reason (n = 6). Seven of the full withdrawals were requested after the 14-week questionnaire was sent and none of the questionnaires was returned. Six of these withdrawals were in the intervention group, with four participants reporting issues with the shoes as a reason for not completing the 14-week questionnaire and choosing to formally withdraw. Issues reported included being unable to select suitable footwear (n = 1), did not wear shoes as they did not fit (n = 1), shoes caused trips (n = 1) or shoes were unsuitable after a change in job role (n = 1). The other two withdrawals did not provide a reason for withdrawing. One further participant withdrew from postal follow-up 84 days into their participation in the trial (no reason recorded) and so was not sent a 14-week questionnaire, but continued to receive text messages.

There were two reported deaths [one in each group (0.04%); both deaths were reported after sending the 14-week questionnaire, which was not returned, and in both cases the date of death was unknown].

Final (14-week) questionnaire response rates

The first set of 14-week questionnaires was sent out on 7 September 2017. A total of 4524 (99.4%) questionnaires were sent out [intervention group, 2254/2275 (99.1%); control group, 2270/2278 (99.7%)], of which 2839 [62.8%; intervention group, 1383/2254 (61.4%); control group, 1456/2270 (64.1%)] were returned. Overall, a questionnaire was received from a slightly higher proportion of randomised participants in the control group (63.9%) than the intervention group (60.8%). The median time between the questionnaire being sent and completed was 6 days in the control group (range 1–140 days) and 7 days in the intervention group (range 1–146 days).

Primary outcome

Raw data

In total, 4494 (98.7%) trial participants provided a valid response to at least one slip text message following randomisation [intervention group, n = 2254 (99.1%); control group, n = 2240 (98.3%)], with 2689 (59.1%) participants returning a complete 14 weeks’ worth of responses [intervention group, n = 1289 (56.7%); control group, n = 1400 (61.5%)].

Of the 59 participants who did not provide a response to at least one post-randomisation text message, 11 provided data for number of slips experienced on the 14-week questionnaire. These data were used for these participants in the analysis and therefore post-randomisation slip data were available for 4505 randomised participants [98.9%; intervention group, n = 2258 (99.3%); control group, n = 2247 (98.6%)]. In total, 6743 slips were reported: 2633 slips in the intervention group (mean of 1.17 per participant, SD 3.0, median 0, range 0–36) and 4110 in the control group (mean of 1.83 per participant, SD 4.7, median 0, range 0–83) (Figure 4).

FIGURE 4.

Histogram displaying the distribution of the number of reported slips, by randomised group (participants with at least 1 week of slip text message data). (a) Intervention; and (b) control.

For the analysis, participants who did not provide any follow-up slip data were ‘censored’ immediately following randomisation (i.e. they were recorded as having reported zero slips over a minimal follow-up time of 0.1 hours). Therefore, all randomised participants were accounted for in the primary analysis model except those missing relevant covariate data (summary statistics for slips per group in the analysis: intervention group – mean of 1.16 per participant, SD 2.9, median 0, range 0–36; control group – mean of 1.80 per participant, SD 4.6, median 0, range 0–83).

Sensitivity analysis

Subgroup analyses

Secondary outcomes

Time to first slip

Figure 5 presents the Kaplan–Meier survival curves for time to first slip for the two groups. The two lines appear to follow one another reasonably closely until about 2 or 3 weeks, when they start to diverge. At this point, about one-quarter of participants in each group have had an event. Indeed, the time for one-quarter of participants to report a slip following randomisation was 18 (95% CI 13 to 19) days in the control group and 20 (95% CI 18 to 25) days in the intervention group. The time for one-third of participants to report a slip following randomisation was 27 (95% CI 25 to 32) days in the control group and 46 (95% CI 34 to 60) days in the intervention group. The intervention group was found to have a lower risk of slipping than control group (HR 0.73, 95% CI 0.67 to 0.80; p < 0.001).

FIGURE 5.

Kaplan–Meier survival plot for time to first slip, by randomised group.

Adverse events

SSHeW trial footwear testing

Study within a trial: pen substudy results

A total of 1453 participants were included in this substudy: 725 participants were allocated to the control group (no pen) and 728 participants to the intervention group, who were sent a pen with their 14-week questionnaire.

Introduction

As stated in the Scientific summary and Chapter 1, the primary objective of this research was to assess whether or not the offer of slip-resistant footwear to NHS staff would lead to a reduction in the incidence rate of self-reported slips over 14 weeks. A secondary objective was to assess whether or not the provision of the footwear would be cost-effective.

To address this, an economic evaluation was conducted alongside the SSHeW trial. The aim of this economic analysis was to help decision-making in determining whether or not the intervention represents a cost-effective means of mitigating the risk of slips among health-care workers in the UK NHS.

Methods

Costing framework

The conceptual framework applied in the CEA (Figure 6) is based on the HSE’s costs of illness model,44 which is used to produce annual national statistics on the costs of workplace injuries and illnesses in GB.

FIGURE 6.

Summary of cost-effectiveness model.

Intervention costs

NHS trusts implementing the footwear intervention will primarily incur two types of costs: (1) the cost of the footwear and (2) resources expended in distributing the footwear to staff.

Footwear purchase costs

Data from the trial on the number and style of footwear purchased, along with the purchase cost from the trial footwear supplier, were used to estimate a weighted average purchase cost per pair. This was then multiplied by the number of employees in the trial intervention group to estimate the purchase cost from the NHS perspective. From the societal perspective, the provision of the intervention footwear displaces footwear that would otherwise be purchased and worn (or, if own footwear is worn occasionally, as observed in the trial, then own footwear is replaced less frequently because of reduced wear). If prior to the intervention employees purchase their own footwear, the intervention represents a transfer from the NHS employer to the member of staff (which, depending on the relative purchase prices, may entirely cancel out). Data on the costs and longevity of footwear usually worn were gathered from 311 trial participants at two trusts during recruitment to the trial to estimate the costs differential between intervention and baseline footwear.

Distribution costs

The trial protocol proposed not to quantify the costs of distributing footwear to staff, on the basis that the NHS has existing dress requirements and provides staff uniform, and any staff time incurred in distributing the footwear to staff in addition to existing processes for uniform would be negligible.

Discussions with a small number of staff at participating trusts involved in distributing footwear during the trial indicated that this had, in some cases, been burdensome, because of the nature of a large number of orders being created and handled in a short period of time. Other, more cost-effective, methods are likely to be applied in practice. The ordering and distribution of shoes could be staggered to minimise logistical costs. An alternative model, confirmed by the footwear supplier as operating in other sectors [T Swinney, Shoes For Crews (Europe) Ltd, 19 September 2019, personal communication], would be for NHS staff to obtain the shoes directly from the supplier, using a unique code to pass the charge on to the NHS, with the footwear delivered to the employee’s home address or other chosen location. The supplier would then charge the NHS on a periodic basis for all shoes supplied. Delivery and returns would be free of charge (as confirmed by the supplier).

Therefore, distribution costs are treated as negligible and accounted for in the existing footwear purchase price.

Health-care resource use

Health-care resource use data were collected via participant-reported questionnaires at 14 weeks and, for injuries unresolved after 14 weeks, during the monthly injury follow-up questionnaire (collected until either the participant reported that the injury had fully resolved or the trial data collection ended).

Participants were asked to provide information on their use of the following services: (1) primary and community care [e.g. general practitioner (GP) and nurse attendance, physiotherapy and occupational therapist visits] and (2) secondary and hospital care (e.g. inpatient, accident and emergency, day case and outpatient attendances, podiatry care). Reported episodes of care were valued using the relevant NHS reference costs45 and Personal Social Services Research Unit unit costs of health and social care,36 summarised in Table 14.

TABLE 14 - Unit costs of health-care resource use per episode of care

A&E, accident and emergency.

Supplementary table (p. 18). Inflated from 2016/17 to 2017/18 prices using the Hospital and Community Health Service Index.

Episode of care	Unit cost (£)	Source
GP at a general practice or at home	43.40	a Unit Costs of Health and Social Care 201836
Nurse at a general practice or at home	12.61	a Unit Costs of Health and Social Care 201836
Occupational therapist	74.87	a Unit Costs of Health and Social Care 201836
Physiotherapist	50.99	a Unit Costs of Health and Social Care 201836
Podiatrist	44.48	a Unit Costs of Health and Social Care 201836
Clinic outpatient visit	125.00	NHS Reference Costs 2017/18 45
A&E visit	160.00	NHS Reference Costs 2017/18 45
Hospital day case	742.00	NHS Reference Costs 2017/18 45
Non-elective inpatient	1603.00	NHS Reference Costs 2017/18 45

To derive the unit health-care costs per injury, the total costs of health-care resource use observed in the trial were divided by the relevant number of trial injuries. For the base case, the calculation included health-care use relating to injuries resulting from slips and the denominator was the total number of injuries resulting from slips reported on the 14-week questionnaire. The estimated unit health-care costs per injury were multiplied by the expected change in the number of injuries due to the intervention across the trial population, to estimate the total expected avoided health-care costs.

Absenteeism costs

From the NHS perspective, worker absence because of injury gives rise to costs associated with temporarily replacing the injured worker and sickness payments made to the injured worker (in lieu of the worker’s wages).

Data on time off work were collected on the SDCS, 14-week questionnaire and at monthly follow-ups after this period. To avoid double counting, time off work data from the SDCS were not included, as this period was covered by the 14-week questionnaire.

Hourly pay rates, including on-wage costs (pensions and National Insurance contributions), were obtained from one NHS trust involved in the trial for each of the occupation types and checked against published NHS earnings data.

Sickness payments were modelled based on the NHS Terms and Conditions of Service Handbook. 46 The rate and duration of sickness pay depends on length of service. Given that 69% of hours lost in the trial occurred within the first month of absence, no absences were > 4 months, and data on the duration of service for trial participants were not available, we have assumed for simplicity that all participants receive sickness payment equivalent to full pay for the duration of their absence.

The NHS can take a range of actions to ensure continuity of service in the event of worker absence due to injury, including hiring replacement bank or agency staff, seeking overtime cover from existing staff or absorbing the absence if staffing levels are adequate. A range of scenarios were considered; however, modelling these would add considerable complexity to the analysis. Therefore, we made a further simplifying assumption that all absences were covered by overtime, at the standard NHS rate of 1.5 times pay, which should provide a reasonable approximation of the likely costs. 46

At the societal level, the principal economic cost of worker absenteeism caused by injury is the loss of output or production due to one less worker (temporarily or permanently) in the labour force. We apply the conventional assumption that the worker’s marginal productivity is at least equal to their costs of employment. 47 This is equal to (1) the worker’s wage plus (2) on-wage costs (National Insurance and pensions contributions). This assumes that the replacement worker would otherwise be doing some other productive activity (i.e. is not involuntarily unemployed). This is reasonable given (1) the current historically high levels of employment and (2) the high vacancy rate in the NHS relative to other public sector occupations.

The analysis did not account for presenteeism costs; that is, instances of when staff returned to work following injury, but before they were fully recovered. This can result in staff being less productive for some period, because of reduced strength or mobility. Quantifying the incidence of presenteeism, and its effects on productivity, is challenging,48 and was not undertaken to avoid adding complexity and uncertainty to the data collection and analysis.

Compensation costs

Workers who suffer an injury caused by a slip or fall at work may submit a claim for compensation. NHS Resolution handles these claims on behalf of NHS trusts, via the Liabilities to Third Parties Scheme. The Liabilities to Third Parties Scheme claims are subject to excesses, with NHS trusts responsible for funding below-excess claims. However, after discussions with NHS Resolution and with a NHS claims manager who was working in one of the participating trusts, it appears that, in practice, NHS Resolution handles the vast majority of claims. Damages paid for settled claims represent a transfer from the NHS to the claimant worker, with additional costs arising from legal resources expended on behalf of the claimant and defendant.

Data on total non-clinical claims and associated claims cost for injuries with ‘slip or trip’ as the primary reason code for NHS employees between 2007/8 and 2016/17 were obtained from NHS Resolution via a freedom of information request (Table 15). Total costs to the NHS are the sum of claimants’ recoverable legal costs (see Table 15, column 2), total damages paid (see Table 15, column 3) and total defence costs (see Table 15, column 4). From this, the compensation cost per claim can readily be calculated; there were on average 528 settled claims per year over the period, resulting in an average annual total claim cost of £12.7M (or £24,000 per claim). However, this is likely to differ from the cost per injury, as many injuries will not result in a claim.

TABLE 15 - NHS Resolution data on successful non-clinical claims closed from 2007/8 to 2016/17 with a primary cause of a ‘slip or trip’

Source: reproduced with permission from NHS Resolution freedom of information request (FOI_3312) (NHS Resolution, 28 February 2018, personal communication).

Year of notification	Number of settled claims closed	Total damages paid (£)	Total defence costs (£)	Total claimant costs paid (£)	Sum of total paid (£)
2007/8	608	5,908,829	1,585,870	5,846,194	13,340,893
2008/9	603	5,826,471	1,569,863	5,612,495	13,008,829
2009/10	459	4,645,895	1,131,722	4,276,821	10,054,437
2010/11	482	4,720,081	891,907	4,306,962	9,918,950
2011/12	801	8,732,507	1,369,108	7,260,387	17,362,002
2012/13	588	7,027,166	1,149,515	6,094,007	14,270,688
2013/14	552	7,270,611	1,207,061	6,425,411	14,903,084
2014/15	444	6,807,880	1,062,790	5,535,280	13,405,950
2015/16	410	5,190,604	867,058	4,917,494	10,975,156
2016/17	332	4,582,955	790,836	3,909,156	9,282,947
Total	5279	60,712,999	11,625,731	54,184,207	126,522,937
Average annual	528	6,071,300	1,162,573	5,418,421	12,652,294

Claim costs per injury were derived by dividing the average annual total claim cost by the total annual injuries due to slips across the NHS. The latter was estimated by applying the estimated baseline annual injury rate from the trial (of 4.06 per 100 workers, see Baseline injuries) to the 1.22 million NHS employees,49 giving an estimated 49,800 annual slip-related injuries. Using this as the denominator results in an estimated compensation claim cost per injury of £254.

At the societal level, the damages paid were excluded as a transfer, leaving average annual costs of £6.66M (total defence costs plus total claimant costs), or £132 per injury.

Results

Qualitative methods

Qualitative results

Summary

Although participants had not considered slips and falls to be a particular issue for staff in the workplace, they identified many potential hazards (e.g. wet floors, clutter from equipment and rushing). Some participants had experienced slips and described the impact of these for both themselves and others (e.g. in terms of staffing levels). Staff interviewed reported little training in slips, trips and falls prevention and felt that any health and safety training received focused more on manual handling safety to prevent back injury.

Participants took several factors into account when choosing work shoes. The main factor was comfort; however, appearance was also considered important, as was the shoes meeting dress code requirements. The consideration of slip resistance was not a predominant thought, although some participants did make an assessment of the sole. Many trial participants were happy with the slip-resistant shoes provided to wear in the trial, finding them comfortable, hard-wearing and smart. A few participants, however, reported problems with them and were unable to wear them. Participants noticed a difference when walking in the trial shoes (associated with the grip) and some reported slips being prevented.

Participants generally had a positive experience of being in the trial. The trial was well promoted within the trusts, making signing up easy. The predominant reason given for taking part was the prospect of receiving a free pair of shoes. Participants found the methods of data collection acceptable and particularly liked the use of text messages to collect data.

Interviews were undertaken via the telephone. This was more practical and allowed easy communication with dispersed participants at convenient and flexible times. This method also minimised costs. We acknowledge that conducting the interviews via the telephone would have prevented us from picking up on non-verbal cues.

Summary of main findings

Cost-effectiveness

Under the base-case scenario, the cost per QALY from the NHS perspective was £38,900 (–£5900 when applying the higher baseline injury rate). From the societal perspective, the trial footwear displaces usual footwear, meaning that the intervention was unequivocally cost-effective at –£60,400 cost per QALY (–£48,400 when applying a higher baseline injury rate).

Sensitivity analysis indicates that the intervention would be cost-effective from the NHS perspective at a £30,000 cost-per-QALY threshold if the cost savings per averted injury were roughly one-fifth higher. This is plausible, given that the costing method is likely to underestimate the cost savings attributable to the intervention and that the intervention could also result in net cost savings to the NHS (i.e. a positive return on investment) within plausible levels of effect in reducing injuries.

The CEA is based on limited data: the number of injuries due to slips observed in the trial was small (n = 79), with the majority of slips being minor and not requiring health-care treatment or time off work. To mitigate a high level of random variability between treatment groups and maximise available data, the costing approach aggregated injury outcome data across both groups to derive mean values per injury averted. If the intervention reduced both the frequency and severity of injuries due to slips, the costing approach would underestimate the resulting reduction in injury costs. A direct comparison of costs in the intervention and control groups was undertaken in sensitivity analysis and indicated an improved incremental cost per QALY from the NHS perspective of –£99,700 (i.e. cost saving), suggesting that costs in the base-case analysis may be underestimated. Improved cost-effectiveness should also be observed in practice if higher compliance with the footwear is achieved (e.g. if it is compulsory) with all else equal.

Qualitative findings

The qualitative study explored participants’ thoughts about slips and falls in their workplace and their experience of being in the trial, particularly with respect to their views of the slip-resistant shoes provided.

Through exploration of participants’ job roles and working environments, we were able to understand the risks for slips and falls that these health professionals were aware of when undertaking their work. Slip and fall hazards were encountered from cleaning activities and through clinical care, as well as from busy, often cluttered, work spaces. Participants described many different types of flooring in their workplaces and reported that they would often traverse several types in a working day. Although most of their time was spent on typical ‘hard, shiny’ hospital flooring, some participants crossed onto carpet in certain hospital areas, such as their offices. Furthermore, community staff encountered carpets in patients’ homes. A few participants described some difficulties due to the grip of the trial shoes in carpeted areas.

Many participants went outside in the course of their work, either in community work or in moving from one part of the hospital to another. This external environment was seen to present slip risks that were unavoidable because of the weather conditions and the slip-resistant shoes were seen to be of benefit in these areas.

Participants described many features they sought when choosing shoes for work and the slip-resistant shoes offered in the trial met many of these requirements: they were comfortable, practical and complied with the trusts’ dress codes. Participants noticed the increased grip when wearing the slip-resistant shoes and reported instances of slip avoidance. Most participants were happy with the slip-resistant shoes and continued to wear them beyond the 14-week intervention period. However, the shoes were not suitable for all participants and some would have preferred a wider range which included smarter styles.

Participants described that the trial was well promoted and received in their NHS trusts, and found trial processes for recruitment and data collection straightforward and not too burdensome. The collection of slip data by text messaging was particularly well received.

The qualitative sample was purposefully selected to include participants from several different job roles and working in different health-care environments. The sample also included participants who had reported some issues with the trial shoes, so we could explore this in greater detail. Owing to modification to the recruitment strategy involving the research teams at site, the resultant qualitative sample included several health-care workers who worked in the research field. These participants may have had a better understanding of the trial processes, including familiarity with questionnaires and the importance of accurate data.

Comparison with other studies

Our results corroborate the limited evidence to support the potential effectiveness of appropriately specified slip-resistant footwear to prevent slips in the workplace. A longitudinal study by Bell et al. 21 indicated a 58% reduction in the number of workers’ compensation claims when a comprehensive slip, trip and fall prevention programme (which included slip-resistant footwear among other strategies) for hospital employees was introduced in the USA. A before-and-after study22 in Denmark suggested that new slip-resistant boots reduced slips, trips and falls compared with older boots by around 80% among fishermen. Additionally, a prospective cohort study of 475 fast food restaurant workers found that the use of slip-resistant footwear was associated with a 54% reduction in the reported rate of slipping. 24 To the best of knowledge, there has been only one other RCT27 of slip-resistant footwear. In this cluster trial, undertaken in the USA, 226 school districts were randomised to receive appropriately specified slip-resistant footwear or continue with the usual company policy, which generally required staff to purchase slip-resistant footwear but did not specify performance requirements. The participants in the study were food service workers and a 67% reduction in slipping injuries was observed (adjusted OR 0.33, 95% CI 0.17 to 0.63). Our population was different to this trial in that only 1.5% of our population included food service workers. In contrast, a cross-sectional study of 125 fast food workers in the USA23 did not find an association between slipping and use of footwear perceived to be slip resistant. However, this finding may have been as a result of the way in which footwear was classified as slip resistant, which was by visual inspection of the tread only (as opposed to any scientific evaluation of slip resistance).

Strengths and limitations of the study

To the best of our knowledge, this is the largest pragmatic trial evaluating the clinical effectiveness and cost-effectiveness of 5-star, GRIP-rated slip-resistant shoes conducted to date, randomising a total of 4554 participants. The trial was well powered. We had anticipated that the proportion of participants in the control group who slipped would be 10% and powered the trial at 90% to detect a 30% relative reduction, to 7%, in the intervention group. In the event, of the 4505 participants who provided slip data, 36% in the intervention group and 45% in the control group reported at least one slip, so the power for this outcome was increased.

The trial design had several strengths: a short data collection period (14 weeks), use of text messages to collect slip and compliance outcomes, and use of a run-in period with outcome data collection, which could have reduced the incidence of post-randomisation attrition. The use of text messages to collect primary outcome and compliance data was acceptable to trial participants. Response rates to the primary outcome were high. Both groups responded in a similar way, with the intervention group responding to an average of 12.3 of the possible 14 (SD 3.1, median 14) text messages and the control group responding to an average of 12.6 (SD 3.0, median 14) text messages. Participants reported that it was a quick way to report outcomes with minimal effort. Text messages were sent weekly to minimise the risk of recall bias.

There were some limitations to the study. Exposure to the intervention was perhaps lower than intended. It was hoped that most intervention participants would receive their footwear within 2 weeks of randomisation; however, the actual median time to receipt of shoes was approximately 4 weeks. The logistics of ordering and distributing large numbers of shoes to staff working varying schedules and over geographically dispersed areas was greater than trusts had anticipated, leading to delays. In addition, despite their best efforts, the trusts could not always get every intervention participant to collect their shoes in a timely manner, if at all. Further delays were experienced if participants needed to exchange their shoes, despite the shoe manufacturer amending their usual returns policy to accommodate swift exchanges.

Once the trial shoes were received, around 50% of intervention participants reported wearing them all of the time when at work, which means that about 50% of participants wore the shoes some, or even none, of the time. This is not necessarily a problem for the trial, as we were primarily testing the offer and provision of slip-resistant footwear without mandating and checking that participants were wearing the footwear. However, it does affect the generalisability of results to other work environments where use of appropriate footwear may be more closely monitored. The primary analysis was based on intention to treat and so may underestimate the true benefit of the slip-resistant footwear. Secondary analysis indicated that intervention participants who showed greater compliance with wearing the slip-resistant footwear had a lower incidence of slipping in the workplace than participants with less or no exposure to the trial shoes. However, we cannot necessarily conclude that a larger effect size would have been produced if all intervention participants wore the shoes all the time when at work. Within the trial population, those with greater ‘compliance’ are likely to be different from those with lower ‘compliance’ (e.g. have a higher or lower slip risk).

Another potential issue is the possibility of contamination. The control group could have seen colleagues wearing their trial shoes and purchased shoes from the same manufacturer, or other 5-star, GRIP-rated shoes, for themselves. Although we do not have data on contamination rates, we used a waiting list control group design whereby control participants were informed that they would receive their shoes soon after their participation in the study ended. Anecdotally, we know from correspondence with control participants that several participants delayed purchasing new footwear as they waited for their trial shoes.

There were some limitations to using text messages. Some NHS staff who would have been eligible for participation could not be recruited as they were unable or unwilling to send and receive text messages. It is difficult to estimate how significant a problem this was. A small number of participants (n = 13) who returned a screening form were ineligible because they explicitly reported that they did not have a mobile phone or were unable or unwilling to send and receive text messages, although this question was asked on an early version of the baseline form only. Following this, we assumed that if participants provided a valid mobile phone number on the contact sheet then they were willing and able to send and receive text messages.

Up to a further 68 screened individuals were ineligible as they had missing or invalid consent or contact details, which may have involved an issue with their mobile phone number. What we cannot know is the number of potential participants in the hospital who did not even express an interest in the trial because they would be unable or unwilling to send and receive text messages, or the number that were ‘screened out’ before being handed a recruitment pack to complete. We tried initially to encourage sites to retain a screening log to record the reason why potential participants who expressed an interest in the trial were not handed a recruitment pack because they were deemed ineligible. However, owing to the overwhelming interest in the trial, and often frenetic recruitment activities, this proved infeasible.

Another limitation was that the text messages were automated and, because of the timing, some text messages were sent out on Christmas Eve and New Year’s Eve 2017, resulting in some messages not being delivered. There is also the possibility that some participants in the intervention group may have incorrectly responded to their compliance text, believing it to be their slip data text. Although participants were requested to send only a numerical reply in response to their text messages requesting slip data, many participants sent non-numeric responses (e.g. messages to the study team).

Another potential limitation to the study is that the outcome data are participant self-reported, which may have led to inaccuracies in the reported slip data. Although we asked participants to report any slip regardless of how ‘minor’, some participants may have reported only slips that were memorable for some reason (e.g. those resulting in an injury or a fall). This could have resulted in a lower slip number being reported. However, as that the slip rate observed in the trial is higher than we initially thought, it would seem unlikely that recall bias occurred, especially given the short time frame over which the data were collected.

It may also be the case that a very small number of people experienced a major injury as a result of a slip at work and may not have been able or willing to complete the follow-up questionnaires, thereby further reducing the number of reported slips. However, it is likely that the research team would have found out about such injuries because of the trusts’ reporting requirements. Conversely, although participants were provided with a definition of a slip and a fall on the participant information sheet, participant diary and, when possible, at recruitment, some participants may have reported some trips as slips, resulting in an overestimate of the number of slips. At set-up, the study team did consider this potential issue and considered sending participants multiple text messages per week. The first text would ask if the participant had been in work that week. If the participant indicated that they had, three further text messages would be sent asking for the number of slips, trips and falls separately. Unfortunately, the text messaging service was not sophisticated enough to dynamically respond to large numbers of participants in this way and so it was decided to send participants one simple text asking if they had experienced any slips.

This raises another potential issue. We have assumed that a zero response via text message means that the participant had worked at least some hours that week and had not experienced any slips when at work. However, it is feasible that people could have replied zero, but not actually been at work. This could potentially underestimate the slip rate; however, we hope that this issue was minimal (indeed, some people replied to the slip text message to explicitly report that they had not been in work that week) and would therefore have a limited impact on the effect size of the intervention. We asked on the 14-week questionnaire whether or not participants had taken time off work for any reason in the last 14 weeks and, if so, how much time they had off. The amount of time taken off was similar between the two groups and therefore this allays some concerns about selective reporting between the two groups. A further issue which could have affected the accuracy of the slip data is that it was not possible to blind participants to their group allocation. Responses could have been affected by the knowledge that they were or were not wearing potentially slip-resistant footwear.

The rate of response to the week 14 questionnaires, which were posted to participants’ homes, was low, at 62.8%. This was despite reminders being sent to non-responders and a SWAT being implemented in the trial to improve response rates to this questionnaire. The response rate was slightly higher in the control group than in the intervention group (64.1% vs. 61.4%, respectively). A potential reason for this is that the intervention group had already received their shoes by the time the questionnaire was sent to them and so there may have been a level of disengagement in the study by this point. On the other hand, participants in the control group were still to receive their shoes and so perhaps more likely to remain engaged in data collection. In general, a contributing factor to the low response rate overall might be that participants had forgotten to expect a postal, paper questionnaire (between being informed at trial enrolment and 14 weeks post randomisation) and were used to providing outcome data via text messages only. Additionally, this was a busy, working population and, as such, the are likely to struggle to find the time to complete questionnaires. It is possible, although difficult to measure, that there is selection bias introduced in that responders are likely to be different to non-responders.

Generalisability of the results

The SSHeW trial was a pragmatic RCT across seven trusts within England, including small and large acute trusts, and mental health and community trusts. The trial included participants who were undertaking a wide variety of jobs, and included clinical, administration, domestic and catering staff. They worked in diverse environments, such as wards, offices, food preparation areas, laboratories and in the community, and walked on a range of floorings, including lino, carpet, slip-resistant surfaces and outside areas, such as car parks, paths and patients’ homes. We were, however, unable to include all staff working in all of the hospitals. Some staff, such as porters and operating theatre staff, in some trusts had to wear protective shoes provided by the trust, which made them ineligible for the study. The number of catering staff and domestic staff who took part in our study is low. This was because in some of the participating hospitals catering staff and domestic staff were employed by a subsidiary trust and could not be enrolled in the trial as a consequence of indemnity issues.

Eighty-five per cent of participants were female. Certainly, we know that in one of our included trusts their male to female ratio is split at approximately 20% male to 80% female and so our population may be reasonably representative in terms of the gender distribution of NHS staff. In this same trust, approximately 11% of their workforce are aged 16–29 years, 23% are aged 30–39 years, 28% are aged 40–49 years, 30% are aged 50–59 years and 8% are aged > 60 years. In our trial, the corresponding proportions are 17%, 21%, 28%, 29% and 5%, respectively. Therefore, except perhaps being marginally skewed to the youngest ages, our population was representative of NHS staff in terms of age. Our trial population was representative in terms of ethnicity as, on average, the ethnicity in our participating trusts was approximately 87% white or white British.

The findings of our study are applicable to footwear that has been tested using the HSE GRIP rating scheme and is appropriately specified for the workplace and role. Although the footwear used in our trial was awarded the highest 5-star rating, it is important to note that this rating may not be appropriate for all roles and workplaces. It is the friction generated between the footwear and floor surface that prevents slips, and footwear with lower GRIP ratings would be expected to provide similar levels of friction in environments that are less challenging (i.e. where there are fewer contaminants or where the floor surface has a lower slip potential). Some roles are also likely to have a lower slip risk than some of the roles included in our trial. For example, they may involve shorter walking distances, less rushing, and no pushing and pulling. Similar results may therefore be obtained with other types of slip-resistant footwear, provided they have been appropriately specified for the role and workplace environment by undertaking a risk assessment and a meaningful evaluation of the footwear’s performance. The results may not be applicable to other footwear that claim to be ‘slip resistant’, but have not been appropriately evaluated and rated with the GRIP rating scheme.

Implications for health care

The preference would always be to try to minimise or adequately control a potential slip risk for all individuals it could affect (workers and members of the public). This can be done by eliminating, for example, floor surface contamination or by specifying floor surfaces that remain sufficiently slip resistant when they are contaminated. However, this may not always be possible. In this situation, employers or staff may consider the use of slip-resistant footwear. Our results suggest that there is a role for appropriately specified footwear in reducing slips in the workplace. The qualitative work has shown that these shoes would be acceptable to staff.

Further research

Although the staff represented a working group that was very diverse, our findings may not generalise to all workplaces. It may therefore be worth considering replicating the trial in other high-risk environments, such as some manufacturing and food-processing facilities, or in settings which have different surfaces, such as in the construction industry, where surfaces are more likely to include loose gravel or mud.

The use of text messages to collect outcome data, for which a single number response is required, was acceptable to both the participants and the researchers conducting the study. From the participants’ perspective, text messages were an easy and convenient way to provide data. For the researchers, text messages were a cheap way of collecting a large number of data from a wide geographical area. The data were collected quickly, with high response rates which were similar between the two groups. As participants did not find it burdensome to respond to the text messages, text messages could be sent weekly, which minimised the possibility of recall bias being introduced. However, we would recommend caution if using text messages to collect more than one variable. In this study there was the possibility of participants confusing the text messages requesting slip and compliance data, even though the text messages were sent out on different days. In future trials, researchers may wish to consider using other technologies, such as accelerometers, to collect outcome data.

Further qualitative research could be undertaken to explore how the introduction of the slip-resistant footwear could be implemented in different trusts.

In this study we took the opportunity to undertake a SWAT, which contributed to the body of evidence regarding how to run RCTs more efficiently. The SWAT, which evaluated an intervention to improve response rates to postal questionnaires, was cheap and easy to implement. We would recommend researchers consider embedding a SWAT in their research. Guidance about conducting SWATs can be found at the Trial Forge website [URL: www.trialforge.org/trial-forge-centres/ (accessed 26 May 2020)].

Conclusion

The SSHeW trial has shown clear evidence that 5-star, GRIP-rated footwear can reduce slips and falls resulting from slips in the NHS workplace. The results indicate that the intervention could also be cost-effective at the £30,000 per QALY threshold from NHS perspective and cost saving from a societal perspective if the footwear is distributed to the employee’s home address rather than at NHS premises. Participants found the shoes to be acceptable. In cases when it is not practicable to prevent floor surfaces becoming slippery in the workplace, employers or staff may consider the use of appropriately specified slip-resistant footwear.

Contributions of authors

Sarah Cockayne (https://orcid.org/0000-0002-1288-5497) (Research Fellow, Health Sciences) was co-chief investigator and the SSHeW trial manager. She contributed to the overall study design and was the lead for study management. She undertook some of the qualitative interviews and analysis, and was involved in writing the initial version of the report. She was responsible for co-ordinating the compilation, formatting, proofreading of the final report and gave final approval of the manuscript.

Caroline Fairhurst (https://orcid.org/0000-0003-0547-462X) (Statistician, Health Sciences) was a co-investigator, contributed to the overall study design, wrote the statistical analysis plan, conducted the statistical analysis, contributed to writing the initial version of the report and gave final approval of the manuscript.

Michael Zand (https://orcid.org/0000-0001-9340-7958) (Economic Adviser, HSE) was a co-investigator, contributed to the overall study design, wrote the health economics analysis plan, led the health economics analysis, wrote the health economics section of the report and gave final approval of the manuscript.

Gillian Frost (https://orcid.org/0000-0003-3339-8898) (Senior Epidemiologist) was a co-investigator, contributed to the overall study design, undertook PPI work, recruited participants, checked the primary statistical analysis, contributed to writing the initial version of the report and gave final approval of the manuscript.

Mark Liddle (https://orcid.org/0000-0001-9332-129X) (risk and human factor) was a co-investigator, contributed to the overall study design, undertook PPI work, recruited participants, co-ordinated the footwear assessments, liaised with the shoe manufacturer, contributed to writing the initial version of the report and gave final approval of the manuscript.

Rachel Cunningham-Burley (https://orcid.org/0000-0001-8724-8276) (Research Fellow, Health Sciences) was the trial co-ordinator, assisted with the day-to-day management of the study, undertook qualitative interviews and analysis, and was involved in writing the initial version of the report and gave final approval of the manuscript.

Catherine Hewitt (https://orcid.org/0000-0002-0415-3536) (Professor in Statistics, Health Sciences) was a co-investigator, contributed to the overall study design and implementation, supervised the statistical analyses and gave final approval of the manuscript.

Heather Iles-Smith (https://orcid.org/0000-0002-0520-2694) (Leeds Teaching Hospitals NHS Trust) was a co-investigator, contributed to the overall study design, co-ordinated recruitment at the trust and gave final approval of the manuscript.

Lorraine Green (https://orcid.org/0000-0003-1620-2443) (Research Podiatrist) was a co-investigator, contributed to the overall study design and gave final approval of the manuscript.

Emily Bain (https://orcid.org/0000-0002-7173-797X) (Assistant Health Economist, HSE) assisted with the health economic analysis and gave final approval of the manuscript.

Misbah Mogradia (https://orcid.org/0000-0003-2792-6455) (Student Health Economist, HSE) assisted with the health economic analysis and gave final approval of the manuscript.

David J Torgerson (https://orcid.org/0000-0002-1667-4275) (Professor, Director of YTU, University of York) was the lead applicant and chief investigator for the SSHeW trial. He had overall responsibility for the design and implementation of the trial, was responsible for the writing of the report and had final approval of the report submission.

Publications

Cockayne S, Fairhurst C, Frost G, Hewitt C, Liddle M, Zand M, et al. SSHeW study protocol: does slip resistant footwear reduce slips among healthcare workers? A randomised controlled trial. BMJ Open 2018;8:e026023.

Cockayne S, Fairhurst C, Frost G, Liddle M, Cunningham-Burley R, Zand M, et al. Slip-resistant footwear reduces slips among National Health Service workers in England: a randomised controlled trial [published online ahead of print January 15 2021]. Occup Environ Med 2021.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the PHR programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the PHR programme or the Department of Health and Social Care.

SSHEW trial qualitative topic guide: interviews with participants as part of the SSHEW trial