Economic evaluation of multiple falls prevention interventions among community-residing older adults: we know the ‘value’, but what about ‘value for money’?

Abstract

Introduction

Falls among older adults are a substantial concern, with considerable effects on health, service use and costs. Preventing falls is therefore an important goal. In our previous systematic review and network meta-analysis (NMA), several effective fall prevention interventions were identified, but their cost-effectiveness in Canada remains unknown.

Methods

A microsimulation model was used to conduct a cost-utility analysis from a public healthcare payer perspective and a lifetime time horizon. Data sources included the previous NMA, administrative datasets and published literature. Three interventions (exercise, exercise plus vision assessment [Ex + va], and Ex + va + environmental assessment) were compared to usual care in the primary analysis; a secondary analysis assessed four additional interventions. Uncertainty was characterised in scenario analyses.

Results

Ex + va had the lowest cost ($149 346, 95% confidence interval [CI] 147 985–150 706) and the highest quality-adjusted life years (QALYs) gained (7.35, 95% CI 7.34–7.37), dominating all other interventions. Differences in costs and QALYs between interventions were small (ranges: $149 346–152 691; 7.27–7.35); usual care was the costliest and least effective. Results were unchanged in the scenario analyses.

Discussion

Ex + va dominated all other interventions, and all primary interventions dominated usual care. Adoption of a societal perspective may affect these conclusions, as some costs for vision care and environmental modifications are borne by patients in our setting, which were not included. Uptake of interventions that reduce falls and costs is recommended. These findings demonstrate that in Canada, multiple forms of fall prevention interventions may reduce costs and improve outcomes.

Key Points

• Falls are an increasing and significant problem amongst older adults, leading to substantial health and economic burden.

• Building upon a previous network meta-analysis of fall prevention interventions, this paper assessed the cost-effectiveness of seven fall prevention interventions (previously identified to be effective) for community-dwelling older adults in Canada from a public healthcare system perspective.

• The findings showed that certain interventions to support fall prevention can improve outcomes and reduce costs compared to no intervention at all, highlighting the importance of some types of fall prevention interventions.

• The combination of exercise and vision assessment yielded the greatest health benefit gains at the lowest cost amongst all seven interventions.

Introduction

Falls among older adults have been described as a national concern in Canada [1, 2], where one in three older adults (ages 65 years and older) falls each year [1]. Falls are estimated to lead to 95 000 hospitalisations and 425 000 emergency department (ED) visits annually [3]. Patients hospitalised due to falls spend more time in the hospital than the average inpatient (14.3 vs. 7.5 days) [4]. In 2018, ~5000 deaths were attributed to falls in the older adult population, and the total cost of falls was estimated to be CAD$10 billion, with direct health care costs accounting for $9 billion [3].

There are a wide range of fall prevention interventions, including single and multicomponent strategies. In 2017, members of this research team published a systematic review and network meta-analysis (NMA), which found that exercise alone and various combinations of interventions were associated with a lower risk of injurious falls compared to usual care (no intervention) [5]. World guidelines for falls prevention published in 2022 state that all older adults should be advised on falls prevention and physical activity [6]. These guidelines also state that community-dwelling older adults should be assessed for risk, with those deemed intermediate or high risk being offered additional interventions [6].

Decision-makers need answers to both the effectiveness question: ‘does the intervention work?’, as well as the ‘value-for-money’ question: ‘is the intervention cost-effective?’ The cost-effectiveness of falls prevention interventions in Canada remains unclear. In other jurisdictions, exercise has been shown to be cost-effective, but the nature of the exercise intervention may vary [7, 8]. The cost-effectiveness of falls prevention interventions may not be generalisable across jurisdictions, due to several factors such as differences in intervention components, insurance schemes, target populations and settings [9, 16]. Therefore, this study aimed to estimate the cost-effectiveness of fall prevention interventions in community-residing older adults based on the previously conducted NMA.

Methods and analysis

A study team was assembled, which included health services researchers, health system administrators, primary care providers, geriatricians and patient partners. A cost-utility analysis was conducted in accordance with Canada’s Drug Agency economic evaluation guidelines [9] and was reported according to the Consolidated Health Economics Evaluation Reporting Standards (CHEERS) checklist (Appendix 1) [10]. A Canadian public payer perspective and a lifetime time horizon were adopted. A modified Delphi process was used to inform model design, which has been published previously [11]. The model was developed in TreeAge Pro™ software.

Data sources

Parameters were derived from published literature, census data and Ontario health administrative databases housed at ICES (formerly the Institute for Clinical Evaluative Sciences). ICES is an independent, nonprofit research institute with health care and demographic data in Ontario, Canada [12].

Target population and setting

The target population consisted of Ontario residents aged 65 years and older living in a community setting, including those living in their own home, a retirement home, supportive housing or assisted living [13] but excluding residential aged care [i.e. long-term care (LTC)]. The population age and sex distributions were based on 2016 Ontario census data [14].

Interventions

The modelled interventions were selected based on the 2017 NMA published by Tricco et al. [5] and through consultation with stakeholders. Interventions included in the primary analysis had to be available in the community setting and have a statistically significant impact on the rate of injurious fall. Three interventions met these criteria: exercise (Ex); combined exercise and vision assessment (Ex + va); and combined exercise, vision assessment and environmental assessment (Ex + va + ea). A secondary analysis was conducted including interventions that showed some evidence of effectiveness relative to usual care but were not statistically significant in Tricco et al. Table 1 shows the costs of intervention delivery and odds ratios (ORs) for intervention effectiveness, with further details in the Methods and Analysis subsections.

Table 1.

Interventions included in the falls model, including relative effectiveness and cost

	Description	Abbrev.	Injurious fall reduction OR (95% CI)a	Cost ($) (range)b	Visualisation
Primary interventions; include exercise
1	Exercise alone	Ex	0.51 (0.33–0.79)	149 (112–186)
2	Combined exercise & vision assessment	Ex + va	0.17 (0.07–0.38)	288 (216–360)
3	Combined exercise, vision assessment and environmental assessment	Ex + va + ea	0.30 (0.13–0.70)	469 (352–586)
Secondary interventions; include exercise
4	Combined exercise & environmental assessment	Ex + ea	0.46 (0.20–1.09)	421 (337–505)
5	Combined exercise, electromagnetic field therapy plus whole-body vibration, calcium and vitamin D supplementation	Ex + em + wb + ca + vi-d	0.40 (0.12–1.33)	4767 (3575-5959)
Secondary interventions; do not include exercise
6	Combined vision assessment and environmental assessment	Va + ea	0.46 (0.14–1.53)	320 (240–400)
7	Combined multifactorial assessment and treatment and patient-level QI strategies	Multi QI	0.50 (0.16–1.49)	275 (206–344)
	Usual care	–	1.00	–	–

Abbreviations: Ca, calcium; CI, confidence interval; ea, environmental assessment; em, electromagnetic; Ex, exercise; OR, odds ratio; va, vision assessment; QI, quality improvement; SD, standard deviation; vi-d, vitamin D; Wb, whole body.

All costs are in 2021 CAD.

^aSource: Tricco 2017 NMA [12].

^bCost range is +/−25% of base amount. Micro-costing details of interventions provided in Appendix 4.

Model structure and outcomes

Outcomes were expressed as quality-adjusted life years (QALYs) and 2021 Canadian dollars. The model also reported the mean number of falls per person, number of falls resulting in ED visits and fall-related hospitalisations. Costs were inflation-adjusted to 2021 dollars using the Consumer Price Index [15]. Future costs and health benefits were discounted at an annual rate of 1.5% as per Canadian guidelines [9]. A half-cycle correction was used.

A person-level state-transition model was developed to simulate each hypothetical person’s interactions with the healthcare system. The model states represent settings defined by where individuals reside, as these are strongly associated with exposure to the interventions, risk of future events, costs and quality of life and are inherently mutually exclusive (see Figure 1) [16]. The decision to describe states based on location is a distinct feature of this model [19]. The five states are: community, hospital, rehabilitation hospital, residential aged care (LTC) and death (an absorbing state; it is possible to transition to death from any state). At the start of each simulation, all individuals reside in the community setting. At each two-week cycle, individuals may: (1) remain in the community setting; (2) transition from the community setting to hospital, LTC or death; and (3) transition from LTC to hospital or death. Transition probabilities are affected by whether individuals experience an injurious fall. It was not possible to return to the community setting after entering LTC.

Figure 1.

Markov model depicting falls model structure. ED, emergency department. Note: States are defined by the location of individuals within the model at discrete time intervals. The model limits individual stays in the hospital state to one cycle (2 weeks) and in the rehabilitation hospital to two cycles (4 weeks).

An injurious fall could lead to an ED visit. If an ED visit occurred without hospitalisation, the cost of the ED visit was accrued, but no further effects were modelled. Patients who are hospitalised after an injurious fall have future cost and quality of life implications, accruing additional health care costs for 3 years and experiencing a reduced quality of life for 7 years (described further in subsequent sections). The six modelled injury types leading to hospitalisation were: vertebral fracture, hip fracture, wrist fracture, other fracture, intracranial haemorrhage and other nonfracture injury. The distribution of injury type was based on ICES data (described in the Model Parameters section and shown in Table 2). Some injuries resulted in individuals first going to a rehabilitation (rehab) hospital before discharge. It was assumed that only one injurious fall occurred per cycle, only one injury occurred per fall and that individuals would not fall again while in the hospital or rehab hospital. Individuals may only stay in the hospital for one cycle, and in the rehab hospital for two cycles. This assumption was based on the average lengths of stay for fall injuries [16, 17]. With the exception of injury type being included as a covariate affecting the probability of mortality in the hospital setting, individuals who experienced a fall or a subsequent fall resulting in injury did not have their future risk of death altered in the model. It was also assumed that if a person transitioned to LTC, a spot was readily available for them.

Table 2.

Hospital, emergency department and rehabilitation hospital costs, utilities and discharge probabilities by injury type

Injury type	Costs			Distribution in ED		Probab-ility of discharge back to community	Disutilityb
	Cost per ED visita	Cost per hospital staya	Cost per rehab hospital stayb	Community setting	Residential aged care setting
Vertebral fracture	$686	$15 406	$18 988	2%	1%	0.975ɅɅ	0.767ǂ
Other fracture	$609	$13 442	$14 875	22%	16%	0.975ɅɅ	0.779ǂ
Hip fracture	$471	$12 402	$15 859	7%	19%	0.752Ʌ,**	0.722ǂ
Wrist fracture	$618	$10 973	$13 748	8%	3%	0.975ɅɅ	0.786ǂ
Intracranial haemorrhage	$798	$13 940	$25 636	2%	2%	0.75ɅɅ	0.776ǂǂ
Other nonfracture injury	$528	$12 301	$16 558	59%	58%	0.975ɅɅ	0.981ǂǂ
Source	OCCI [27] OSB [29]	OCCI [27] OSB [29]	National Rehabilitation Reporting System; Wodchis et al. [23]	ICES	ICES	ɅCIHI [27] ɅɅClinical expert opinion	ǂTarride 2016 [19] ǂǂGBD, 2015 [20]

Abbreviation: ED, emergency department.

Assumption for all ED visits: 1 emergency department physician on duty consultation (specialist) for all visits

Note: Rounded to nearest dollar. All costs are in 2021 CAD. All costs and probabilities were varied +/−25% in the probabilistic analysis unless otherwise noted (^**used a standard deviation of 0.096). Costs were assigned a gamma distribution; probabilities and utilities were assigned a beta distribution. Full parameter details provided in Appendix 2.

^aSources: Ontario Case Costing Initiative [27]; Ontario Schedule of Benefits [29]; clinical expert opinion. Numbers of consultations assumed were as follows: Fractures assumed 1 internal medicine, 1 orthopaedic surgery, 1 rehab & occ. medicine & daily most responsible physician for average length of stay (by injury type); intracranial haemorrhage assumed 1 neurosurgery, 1 internal medicine, 1 rehab & occ. medicine & daily most responsible physician for average length of stay; other nonfracture injury assumed 1 internal medicine, 1 rehab & occ. medicine & daily most responsible physician for average length of stay.

^bNote: These values are multiplied by the utilities associated with the individual’s setting.

Model parameters

A complete table of all model parameters is shown in Appendix 2.

Intervention effectiveness

Intervention effectiveness was based on ORs reported in the NMA and applied to the probability of an injurious fall [5]. Interventions occurred once and did not repeat. The reduction in injurious fall probability was only applied in the community setting and lasted 1.5 years (18 months); the mean duration of the included randomised controlled trials (RCTs). This assumption was varied in a scenario analysis.

Transition probabilities

We estimated that a community-dwelling individual had a mean 34% [standard deviation (SD) 5%] probability of having an injurious fall each year, based on Tricco et al. [5]. The probability of visiting the ED after an injurious fall in the community setting was 71% (SD 9%) [18]. ICES data showed that those living in LTC are less likely to visit the ED after an injurious fall, likely due to existing care on site (6%; SD <1%) [18]. The probability of admission to hospital, hospital discharge and hospital mortality were functions of age, sex and type of injury, estimated using coefficients derived from a logistic regression analysis of retrospective cohort studies using ICES data, described in detail in Appendix 3 [18, 19]. Types of injury resulting from an injurious fall, and their impacts on transition to LTC, are shown in Table 2. The baseline probability of entering LTC from the community setting was based on age and sex [18]. Once in LTC, individuals were at a higher risk of falling and dying (see Appendix 2).

Intervention costs

Intervention costs were applied once at the beginning of the model. Appendix 4 shows detailed intervention cost components and their uptake weights, while.

Table 1 shows the final total cost. All interventions, except for exercise and whole-body vibration, began with all individuals receiving a comprehensive assessment by a primary care physician. Other components of the intervention were weighted by adherence reported in the associated RCTs.

The exercise intervention was assumed to include two 60-minute exercise classes per week for 14 weeks led by a trained professional (e.g. physiotherapist) in groups of 9, based on the program descriptions in RCTs in the NMA [5]. Environmental assessment included a home visit by an occupational therapist. Vision assessment costs included cataract surgery, applied to 1% of people receiving an assessment. It is important to note that glasses purchases following a vision assessment are not covered by Ontario’s public health system, nor are home modifications identified during the environmental assessment (e.g. bars added to a bathroom). Currently (as of December 2024), whole-body vibration (18 weeks under the supervision of an occupational therapist) is not publicly reimbursed; however, its costs were included in this analysis. Vitamin D and calcium supplementation are also currently not publicly reimbursed and were not costed.

Emergency department and hospitalisation costs

Hospitalisation costs were obtained from the Ontario Case Costing Initiative (OCCI) [20]. Costs were derived by injury type using ICD-10 codes [21]. Costs for physician services were calculated by multiplying service use by fees in the Ontario Schedule of Benefits [22]. Service use was estimated by clinical experts on the study team [email communication, August 2022]. Hospital and ED costs by injury type are shown in Table 2.

Rehabilitation hospitalisation costs

Rehabilitation hospital costs were estimated by using the method of Wodchis and colleagues [23] (see Table 2). Costs were obtained for each injury type from the National Rehabilitation Reporting System ICES data January 1, 2018, to December 31, 2019 [20].

Posthospitalisation costs

Health service use costs after hospital discharge for physician visits, medication and home care were based on a study by Hopkins et al., which used administrative health data from Manitoba, Canada, to compare resource use between incident and prevalent patients with fracture with controls [24]. Hopkins reported days of home care use by injury type, which was multiplied by the weighted average cost of home care in Poss et al. [25], and unit costs for Ontario home care reported in the Heath Data Branch Web Portal [26]. For injury types not included in Hopkins, we relied on clinical expert opinion [email communication, August 2022] [24].

Utility values

Age- and sex-specific utility weights for the community setting were obtained from a large Canadian study [27]. In LTC, a utility weight value of 0.325 was applied [28]. For cycles spent in hospital or rehab hospital, individuals retained their utility weight from the community or LTC setting, depending on where they were previously. Utilities were obtained for each fall injury type (see Table 2) based on the study by Tarride et al. [29]. Fall-related injury disutility lasted 3 years postinjury [30]. When patients experienced a second fall-related injury within that 3-year period, the disutility associated with the worst of the two injuries was applied.

Mortality

Background mortality rates for the community and rehabilitation settings were taken from Statistics Canada Life Tables using 2017 to 2019 Ontario data [31]. Mortality in LTC was estimated by applying a hazard ratio calculated for nursing homes in the USA to the mortality rate in the community setting [32]. In-hospital mortality was estimated using logistic regression equations developed using ICES data [18] (Appendix 2).

Analysis, uncertainty and validation

Secondary and scenario analyses results were obtained, running the model using 5000 person-level Monte Carlo simulations to account for inherent stochasticity between trials. The primary analysis model was run probabilistically as a two-dimensional simulation, using 1000 second-order Monte Carlo simulations sampling variables from their respective distributions to account for uncertainty in parameter estimates.

For probabilities and utilities, a beta distribution was used, and the range was +/−25% unless otherwise specified in the parameter tables. Costs were assigned a gamma distribution, and similarly, a +/−25% range was used unless otherwise specified. A cost-effectiveness threshold (CET) of $50 000 per QALY gained was adopted based on typical Canadian practices [33].

To further address uncertainty, scenario analyses were planned. The model was run deterministically changing the model time horizon (shortened to 10 years), discount rate (0% and 3%) and the duration of intervention effectiveness (shortened from 18 to 6 months).

The model was examined for face validity and internal validity [34]. To ensure face validity, we consulted subject matter experts on the model structure, input parameters and data sources [34, 35]. Internal validity was assessed by testing with extreme parameter values.

Results

The validity review found that this model reported a high number of mean falls per person (5.68 over the lifetime under usual care); this was believed to be driven by the high number of falls reported in trials included in the NMA [12].

In the base case, combined exercise and vision assessment (Ex + va) had the lowest cost ($149 346, 95% confidence Interval [CI] $147 985–150 706) and the highest QALYs (7.35, 95% CI 7.34–7.37), dominating all other interventions (meaning it was both more effective and less costly). The other interventions also dominated usual care. The base case economic and health outcomes are shown in Table 3, with the interventions ordered by ascending costs. Overall, differences among the interventions were small. In general, the total costs associated with each intervention were around $150 000 ($149 346–152 691), with QALYs in the range of 7.3 years (7.27–7.35). A cost-effectiveness acceptability curve for the primary analysis is shown in Appendix 5. Ex + va is the optimal choice at all CETs.

Table 3.

Results of base case analysis assessing the cost-effectiveness of fall prevention interventions in community-dwelling older adults in Ontario, Canada

Strategy		Economic outcomes					Health outcomes (per person) (95% CI)
Name	Visual	Total cost ($)	Total QALYs	Increm-ental costs ($, vs. usual care)	Incremental QALYs (vs. usual care)	ICER vs. usual care ($/QALY)	Mean # of falls	Mean # of falls resulting in ED Visits	Mean # of fall-related hospitalisations
Ex + va		149 346 (147 985-150,706)	7.35 (7.34–7.37)	−3345	0.08	Dominated	5.23 (5.17–5.28)	2.34 (2.31–2.37)	0.45 (0.44–0.45)
Ex + va + ea		150 113 (148 746-151,480)	7.33 (7.31–7.35)	−2577	0.06	Dominated	5.31 (5.26–5.36)	2.39 (2.36–2.42)	0.46 (0.45–0.46)
Ex		150 615 (149 244-151,986)	7.30 (7.29–7.32)	−2076	0.03	Dominated	5.41 (5.35–5.46)	2.45 (2.42–2.48)	0.47 (0.46–0.47)
Usual Care		152 691 (151 300-154,082)	7.27 (7.25–7.29)	Reference	Reference	Not applicable	5.68 (5.63–5.74)	2.63 (2.60–2.66)	0.49 (0.49–0.50)

Abbreviations: CI, confidence interval; ea, environmental assessment; ED, emergency department; Ex, exercise; va vision assessment; ICER = incremental cost-effectiveness ratio; QALY = quality-adjusted life year.

Note: All costs are in 2021 CAD.

Amongst the secondary interventions evaluated, the intervention including supplementation and whole-body vibration was slightly more costly and more effective than usual care, yielding an ICER of $19 950, and would be considered cost-effective at a CET of CAD$50 000/QALY. The other three interventions all dominated usual care (see Table 4). The scenario analyses yielded identical overall rankings as the base case, except for when the duration of intervention effectiveness was shortened (see Table 4). If intervention effectiveness lasted for 6 months (18 months was the base case assumption), Ex yielded the same number of QALYs as usual care (7.28) but was slightly less costly ($155 677 vs. $156 192).

Table 4.

Results of secondary and scenario analyses

Strategy		Total cost ($)	Total QALYs	Incremental cost ($, vs. usual care)	Incremental effectiveness (vs. usual care)	ICER vs. usual care ($/QALY)
Name	Visual
Secondary interventions
Ea + va		153 300	7.32	−2892	0.04	Dominated
Multi + QI		153 563	7.31	−2629	0.03	Dominated
Ex		154 149	7.31	−2043	0.03	Dominated
Ex + em + wb + ca/vit d		156 990	7.32	798	0.04	19,950
Usual care		156 192	7.28	Reference	Reference	Not applicable
10-year time horizon
Ex + va		53 462	5.27	−3445	0.08	Dominated
Ex + va + ea		55 178	5.25	−1729	0.06	Dominated
Ex		55 499	5.23	−1408	0.04	Dominated
Usual care		56 907	5.19	Reference	Reference	Not applicable
6-month intervention effectiveness
Ex + va		154 345	7.30	−1847	0.02	Dominated
Ex		155 677	7.28	−515	0.007	Dominated
Ex + va + ea		156 011	7.29	−181	0.01	Dominated
Usual care		156 192	7.28	Reference	Reference	Not applicable

Abbreviations: Ca/vit D, calcium and vitamin D supplementation; Ea, environmental assessment; ED, emergency department; em, electromagnetic; Ex, exercise; va, vision assessment; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life year; QI, quality improvement; Wb, whole body vibration.

Discussion

This economic evaluation of several fall prevention interventions found that the combination of exercise plus vision assessment was less costly and more effective than all other interventions. Adding environmental assessment to the combination of interventions was the next most attractive intervention, with exercise alone being the third-ranked intervention. All other interventions, except for whole-body vibration, also dominated usual care (no intervention). The dominance of all interventions over usual care supports a policy recommendation to make some form of fall prevention intervention available to older adults.

Exercise is recommended for all older adults by the 2022 World Fall Prevention Guidelines [6]. Exercise is a powerful intervention with potential spillover effects for other health outcomes, which were not quantified in this analysis. Our findings align with other recent publications evaluating the cost-effectiveness of fall prevention interventions, particularly exercise. Pinheiro et al. systematically reviewed economic evaluations of exercise-based fall prevention interventions and found that most report exercise as cost-effective or dominant over usual care, noting that intervention costs are typically not well described in these analyses [8]. In the current paper, the micro-costing of the intervention components has been performed. Kwon et al. conducted a systematic methodological overview of systematic reviews of community-based falls prevention interventions, also finding that exercise was cost-effective, and that there was significant heterogeneity in modelling approaches and costing; it was noted that multifactorial interventions, including home modifications, while cost-effective, may risk exacerbating social inequities of health [7]. Subgroup analyses of research in New Zealand found lower health gains in Maori populations compared to non-Maori, indicating that some interventions worsened existing health inequities between ethnic groups [7]. The mechanism for the differential health impacts has not been characterised. There are unexplored potential equity implications for interventions that may require out-of-pocket spending, such as home modifications, new glasses and vitamin supplementation. Since we adopted a public healthcare payer perspective, out-of-pocket costs were not included. If a societal perspective were taken, overall conclusions could vary. Local factors should be considered when applying these findings in jurisdictions with different insurance schemes.

In Tricco et al., adding environmental assessment to the combination of interventions lessened the effectiveness compared to the combination of exercise and vision assessment alone (Ex + va OR = 0.17; Ex + va + ea OR = 0.30). A subgroup analysis showed that Ex + va + ea was associated with an increased risk of injurious falls among patients who had fallen previously [12]. This may reflect heterogeneity in either the environmental assessment interventions themselves or their uptake in the source studies.

While all interventions reduced the number of falls and health care costs compared to usual care, the differences between them were small. All the interventions were inexpensive (ranging from $149 to 469 per person), except for whole-body vibration and supplementation ($4767). Given that falls typically have significant short and long-term implications, only a small reduction in falls would be needed to offset the initial intervention cost.

We have identified strengths and potential limitations related to this work. This analysis builds upon extensive previous work examining the efficacy of 40 falls interventions. Many model parameters were drawn from ICES databases, a robust data source for the Ontario population (~14 million people) [18]. The work was supported by a diverse team, and a modified Delphi process was used to design the model [11].

Potential limitations arise from assumptions made in the model design and data availability. It was assumed that all community-dwelling adults experienced the same treatment effect; however, the effect size may vary depending on individual characteristics. There may also be differences in the costs and effectiveness of different exercise regimens; Sherrington et al. found that regimens including balance and functional exercises were most effective in preventing falls [36]. The duration of intervention effectiveness was a simplified assumption for the model; longer durations may be overly optimistic, while shorter durations tested in a scenario analysis did not significantly alter the overall conclusions. The 2022 falls guidelines recommend exercise for all, with further interventions tailored to risk assessments [6]. Stratifying the study population and interventions by risk was not performed in this model; this approach may have better aligned with the design of the world fall prevention guidelines. A differential length of time for impact on costs and utilities is not a common modelling choice but was driven by previous research on fall injuries. Intervention costs accounted for uptake as reported in the RCTs; however, if real-world uptake differed, intervention costs would be affected. The assumption that future risk of falls and death were unaffected by fall history likely underestimates the impact of falls [37]. Secondary and scenario analyses were evaluated deterministically, limiting the ability to fully characterise uncertainty around parameter estimates.

As policymakers address the problem of falls in older adults, cost-effectiveness evidence may be useful for informing decision-making processes. These results show that interventions to reduce falls for community-dwelling older adults are likely to be cost-effective or cost-saving compared to usual care. It was found that the combination of exercise and vision assessment led to the greatest QALY gains and lowest costs. Falls are a substantial and growing problem, and their prevention should be a policy priority.

Declaration of Conflicts of Interest

S.S. is an Associate Editor for Age and Ageing but was not involved in the decision-making process for this article.

Declaration of Funding

This work was supported by the Canadian Institutes of Health Research or CIHR (Grant #408,542). The funder played no role in the design, execution, analysis and interpretation of data, nor in the writing of the study.