Home hazard assessment and environmental modification to prevent falls in older people: the OTIS trial

Study design

We conducted OTIS,⁹ a pragmatic, multicentre, two-arm, parallel-group, modified cohort randomised controlled trial, (cRCT)¹⁰ between October 2016 and August 2019. The protocol was approved by West of Scotland Research Ethics Committee 3 (REC reference 16/WS/0154) and has been published.⁸

Setting and participants

We recruited participants from eight NHS trusts in England based in primary and secondary care. Participants were initially recruited into an observational cohort, after being informed about the embedded trial and the possibility of being offered an OT home assessment visit. Potential participants were identified using: (i) cohorts held at the Yorkshire Health Study¹¹ and York Trials Unit (YTU);^12-14 (ii) a database search of General Practitioner (GP) patients; (iii) advertising (e.g. in faith magazines and GP surgeries); and (iv) opportunistic screening by podiatrists and GPs. Once identified, potential participants were sent trial information including a screening questionnaire and consent form to complete and return to YTU if they wished to take part.

Participants were eligible for the cohort if they: (i) were aged 65 years or over; (ii) were community dwelling, i.e. not living in a nursing/residential home; and (iii) reported a fear of falling or a fall in the previous 12 months. Participants were excluded if they were unable to: (i) give informed consent (e.g. due to dementia or Alzheimer’s disease); (ii) speak English and had no relative or friend to translate/interpret for them; (iii) walk 10 feet even using a walking aid; or (iv) if they had had an OT assessment for falls prevention in the past 12 months or were waiting for an assessment.

Eligible participants were sent a baseline questionnaire and monthly falls calendars to record when they fell and the number of times, to post back to YTU. Participants were eligible to be randomised once they had returned a completed baseline questionnaire and at least one falls calendar in the preceding three months.

Randomisation and blinding

Participants were randomised using the YTU’s secure, remote, web-based randomisation service, stratified by centre. When OTs had capacity to deliver the intervention visits, a member of the trial team at YTU randomised a batch of eligible participants from that centre in a single block. The size was determined by the number of participants available to randomise and the number the centre had capacity to deliver the intervention to within a reasonable timeframe. The allocation sequence was generated by an independent data systems manager, who was not involved in recruitment. To reduce costs, unequal allocation was used in favour of the usual care group, usually 2:1; however, on occasion, other allocation ratios were used. For example, towards the end of recruitment, some centres had capacity to see more, or fewer, than a third of the remaining eligible participants who were pending randomisation, so ratios ranging from 1:1 to 9.7:1 (usual care:intervention) were used to allow as many eligible participants to be randomised as possible. It was not possible to blind trial participants, the research team actively involved in the day-to-day running of the study or the statistician to treatment group, due to the unequal allocation ratio. However, data entry staff were blinded.

Intervention and comparator

Both groups were sent a falls prevention leaflet produced by Age UK (‘Staying Steady’, published in June 2015) and continued to receive their usual healthcare from their GP or other healthcare professionals irrespective of the trial. In addition, the intervention group were offered one OT home hazard assessment and environmental modification. The intervention was delivered by Health and Care Professions Council registered OTs, who were trained by the study team to use the Westmead Home Safety Assessment Form (WeHSA).

The assessment was guided by the validated WeHSA form.¹⁵ This validated tool, consisting of 57 items, split into 15 domains was developed in Australia for older people. The OTs used this assessment tool to help identify potential falls hazards and risk-taking behaviours when walking through the participant’s home. A list of recommendations was agreed and a record of any suggested equipment made. This included: steps; outdoor lights; external rails to the outside of the property; ramp; wheelchair; grab rails/bannister; mobility aids; furniture raisers; shower rails/bath safety bars; bath lift; removable bath board; raised toilet seat; toilet frame; step; safety aids; sensor-operated or remove control lights; emergency alarms; assistive technology devices; light bulbs; bed hoist; bed; key safe; ferrules; walking aid parking devices; ladders; carpet glue/reflective anti-slip tape; alterations to house; or other miscellaneous items. The visits lasted approximately one hour during which recommendations on appropriate home modifications such as improving lighting were made.

Treatment fidelity was explored by observations of the home visits, audits of training methods and case report forms completed by OTs, and semi-structured interviews.

Outcome measures

The primary outcome was the rate of falls sustained per participant over the 12 months from randomisation. A fall was defined as an unexpected event in which the participant comes to rest on the ground, floor or lower level.¹⁶ Falls were reported via the monthly falls calendars and we also asked about number of falls in the past four months on questionnaires posted to participants at four, eight and 12 months post-randomisation. Participants who indicated on their falls calendar that they had fallen were telephoned to complete a Falls Data Collection Sheet (FDCS) to ascertain cause, location and consequences of the fall (e.g. superficial injury, fracture, hospital admission). Secondary outcomes:

Participants who did not return falls calendars or questionnaires were telephoned for data and/or posted reminders. The primary source of falls data was the post-randomisation monthly falls calendars, but where none were returned, falls data from the follow-up questionnaires were used, if provided. The economic analysis will be published separately.

Sample size

We proposed to randomise 1299 participants 2:1 to usual care (n = 866) or intervention (n = 433). This allowed for 10% attrition and provided 90% power (two-sided, α = 0.05) to show a difference in the percentage of participants who fall at least once in the 12 months following randomisation from 60% in the usual care group to 50% in the intervention group. The sample size is based on a binary outcome rather that the primary outcome as power calculations for count data require greater assumptions and parameter estimates and, at the start of this trial, we felt there was insufficient data on which to reliably base this calculation.

Statistical methods

Statistical analyses were performed using Stata v15¹⁸ on an intention-to-treat (ITT) basis, using two-sided tests at the 5% significance level. Stata (RRID: SCR_012763) is a proprietary software; an open-access alternative that can perform an equivalent function to Stata for analysis is R, a free software environment for statistical computing and graphics (RRID: SCR_001905).

For the primary analysis, mixed-effects negative binomial regression was used adjusting for sex, age at randomisation, history of falling (number of falls in 12 months prior to recruitment: <2/≥2) and allocation ratio as fixed effects, and centre as a random effect. The model included an exposure variable for the number of months that the participant provided falls data. Where no falls data were provided at all, we assumed zero falls over a negligible follow-up time of 0.1 months to achieve a strict ITT analysis. The adjusted incidence rate ratio (IRR), 95% confidence interval (CI) and p-value are presented.

A Complier Average Causal Effect (CACE) analysis^19-21 to assess the impact of receiving the OT home assessment visit within 12 months of randomisation on the primary estimate was undertaken, using a two-stage instrumental variable (IV) regression approach with randomised group as the IV.

A post hoc sensitivity analysis, including Parkinson’s disease as a covariate in the primary model, was undertaken due to a chance baseline imbalance in the proportion of participants in the two groups with this disease. Further sensitivity analyses considered the impact of missing data and clustering by OT. Pre-specified subgroup analyses assessed differential effects of the intervention based on whether a participant received hospital care as a result of a fall in the four months before baseline.

Fear of falling was analysed using a mixed effects, covariance pattern model incorporating all post-randomisation time points and adjusting for baseline fear of falling, sex, age, history of falling, allocation ratio, treatment group, time and a treatment-by-time interaction, with participant and centre as random effects. Logistic regression models explored the likelihood of reporting at least one fall, more than one fall, and a fear of falling (at least “some of the time”) at 12 months. Time between falls was analysed using the Andersen and Gill method for repeated events, via a Cox Proportional Hazards regression model with robust standard errors to account for dependent observations by participant. These models adjusted as for the primary outcome.

Participant flow and characteristics

Between October 2016 and August 2018, we mailed out to 19308 potential participants (Figure 1); a total of 3100 screening forms were returned and reviewed for eligibility. We randomised 1331 participants to the intervention (n = 430) or usual care group (n = 901), with 381 (88.6%) intervention participants receiving an OTIS intervention visit within 12 months of randomisation (Figure 2).

Figure 1. Pre-randomisation flow chart.

Figure 2. Post-randomisation flow chart.

Baseline characteristics were comparable between the groups (Tables 1 and 2), apart from a chance imbalance in the proportion of participants with Parkinson’s disease (intervention n = 14, 3.3%; usual care n = 9, 1.0%).

Outcomes

Overall, 1303 (97.9%) participants returned at least one falls calendar following randomisation (intervention n = 419, 97.4%; usual care n = 884, 98.1%); a further five (four usual care, one intervention) provided falls data via a participant questionnaire. Questionnaire data were used for analysis for these participants. In total, 2260 falls were reported: 826 in the intervention group (mean 1.9 per participant, SD 5.5, median 1, range 0 to 94) over a mean of 338 days (median 365), and 1434 in the usual care group (mean 1.6, SD 3.0, median 1, range 0 to 41) over a mean of 345 days (median 365). The adjusted negative binomial model indicated an increase in fall rate in the intervention group relative to usual care (IRR 1.17, 95% CI 0.99 to 1.38; p = 0.07); this was not statistically significant. This result was robust to sensitivity analyses (Table 3).

About a tenth of participants (intervention group n = 41, 9.5%; usual care group n = 83, 9.2%) reported that they attended hospital for a fall in the four months before baseline. There appeared to be a qualitative interaction between this factor and treatment group, with those in this subgroup benefiting from the intervention (IRR 0.86, 95% CI 0.50 to 1.47) compared with those who did not attend hospital for a fall in the four months before baseline (IRR 1.21, 95% CI 1.01 to 1.44); however, the interaction term between this factor and treatment group included in the primary model was not statistically significant (p = 0.24).

There was no evidence of a difference in any of the secondary outcomes or other data (Table 3 and Table 4) and no serious related adverse events reported.

The data suggested that the OTs received adequate training and delivered the intervention, mainly as intended. OTs commented that some trial participants did not reflect those seen in their usual clinical care as they were higher functioning. Recommendations were followed to varying degrees and depended on whether they were provided by health or social care services. A full account of the assessment of treatment fidelity will be published separately.

Underlying data

Full underlying (non-aggregated) data cannot be made publicly available since the ethics approval of this study does not cover openly publishing non-aggregated data.

In order to access this data, it must be requested from the corresponding author. Data requestors will have to provide: i) written description and legally binding confirmation that their data use is within the scope of the study; ii) detailed written description and legally binding confirmation of their actions to be taken to protect the data (e.g. with regard to transfer, storage, back-up, destruction, misuse, and use by other parties), as legally required and to current national and international standards (data protection concept); and iii) legally binding and written confirmation and description that their use of this data is in line with all applicable national and international laws (e.g. the General Data Protection Regulation of the EU).

Reporting guidelines

Open Science Framework: CONSORT checklist for ‘Home hazard assessment and environmental modification to prevent falls in older people: the OTIS trial’, https://doi.org/10.17605/OSF.IO/P8AU2.²²

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

Consent

All participants will give written informed consent prior to entry to the study.