Article Text

Cost–benefit analysis of fall injuries prevented by a programme of home modifications: a cluster randomised controlled trial
1. Michael D Keall1,
2. Nevil Pierse1,
3. Philippa Howden-Chapman1,
5. Chris W Cunningham3,
6. Michael G Baker1
1. 1Department of Public Health, He Kainga Oranga/Housing and Health Research Programme, University of Otago, Wellington, New Zealand
2. 2Economic Consultant Newlands, Wellington, New Zealand
3. 3Te Pūmanawa Hauora Massey University, Wellington, New Zealand
1. Correspondence to Dr Michael D Keall, Department of Public Health, He Kainga Oranga/Housing and Health Research Programme, University of Otago, PO Box 7343, Wellington South, New Zealand; Michael.Keall{at}otago.ac.nz

## Abstract

Background Injuries due to falls in the home impose a huge social and economic cost on society. We have previously found important safety benefits of home modifications such as handrails for steps and stairs, grab rails for bathrooms, outside lighting, edging for outside steps and slip-resistant surfacing for outside areas such as decks. Here we assess the economic benefits of these modifications.

Methods Using a single-blinded cluster randomised controlled trial, we analysed insurance payments for medically treated home fall injuries as recorded by the national injury insurer. The benefits in terms of the value of disability adjusted life years (DALYs) averted and social costs of injuries saved were extrapolated to a national level and compared with the costs of the intervention.

Results An intention-to-treat analysis was carried out. Injury costs per time exposed to the modified homes compared with the unmodified homes showed a reduction in the costs of home fall injuries of 33% (95% CI 5% to 49%). The social benefits of injuries prevented were estimated to be at least six times the costs of the intervention. The benefit–cost ratio can be at least doubled for older people and increased by 60% for those with a prior history of fall injuries.

Conclusions This is the first randomised controlled trial to examine the benefits of home modification for reducing fall injury costs in the general population. The results show a convincing economic justification for undertaking relatively low-cost home repairs and installing safety features to prevent falls.

Trial registration number ACTRN12609000779279.

## Background

Injuries from falls pose a major health burden worldwide. In terms of disability adjusted life years (DALYs), falls have been identified as the 19th most important contributor to the global burden of disease and injury, ranked higher in Australasia (7th) and Western and Central Europe (6th).

Globally, falls have been ranked as the third most important injury cause (after road injury and self-harm).1 Even in a relatively small country like New Zealand, there were over a million medically treated injuries due to falls over the period 2010–2013.2 The costs associated with this injury burden are considerable, in terms of both direct costs and social costs.3 For example, in 2013, the direct medical costs of older adults' falls in the USA were over $34 billion.4 Social costs, which take into account human suffering, sometimes expressed in terms of willingness to pay to avoid death and suffering, tend to dwarf direct costs. The DALY was developed as a measure of population health for setting priorities in resource allocation and to assess the burden of disease.5 DALYs represent disease burden by summing years of life lost (YLL) and years lived with disability within disease or injury categories. For each health condition, severity weights were estimated to represent how disabling a condition is. These weights were set from the judgements of a panel of experts from various countries. Despite its robust construction, the DALY has been criticised for explicitly valuing the life of a disabled person as less than that of a person without a disability, and for not acknowledging the importance of the context (eg, the lack of facilities and support for disabled people in low/middle-income countries, which significantly diminishes life quality for the disabled).6 In developed countries, the value of non-fatal injuries can be considered to be overstated by the DALY, and understated in low/middle-income countries. The value of statistical life (VOSL) is derived from the valuation of changes in the degree of exposure to the risk of death rather than the valuation of the life of a specific individual. In New Zealand, as in many other countries that have estimated the VOSL, an important component is the cost of pain and suffering due to the loss of an unidentified life.7 This cost is estimated from representative population surveys of the amount of money the population in total would be willing to pay for a safety improvement that prevents one premature death. Added to this are more tangible costs, including loss of output (if not already included in the VOSL estimate), medical costs, legal costs and property damage costs (where applicable). A similar approach is used to value the prevention of injuries of various levels of severity.7 Thus the diminishment of life quality from non-fatal injuries is accounted for by the amount the population is willing to pay to reduce the risk of such an injury, which places a value on pain, suffering and loss of physical and/or mental capability. A meta-analysis of stated preference studies conducted for the OECD found little evidence that the VOSL varies with age.8 Two recent studies have estimated injury-related health loss for New Zealand in terms of DALYs. As part of the Global Burden of Disease 2010 project, it was estimated that falls in New Zealand account for 28% of all injury DALYs nationally, a total of approximately 29 500 DALYs due to falls in 2010.9 This number contrasts with estimates from the New Zealand Ministry of Health and the Accident Compensation Corporation (ACC) using 2006 data, which found that falls accounted for 7536 DALYs, around 10% of all injury-related DALYs.10 Differences between these estimates were discussed briefly in the latter Ministry of Health study and were attributed to the data sources used and partly to the exclusion of workplace falls from the latter study. The latter study then made estimates of New Zealand social and economic costs of injuries using monetised DALYs as the basis for the ‘human cost’ component of costs.11 Using the dollar value for a preventable fatality (the VOSL), a value was set for each DALY such that the sum of discounted expected future life years equalled the VOSL.10 A value of$150 000 per DALY ($NZ: June 2008 prices) was proposed.10 Unlike the VOSL approach to monetising injury, the DALY approach produces values for safety that are strongly tied to the age of the victim. Recently, a New Zealand randomised controlled trial found that home modifications were effective in preventing falls for the general population.12 This Home Injury Prevention Intervention (HIPI) study estimated a 26% reduction in medically treated fall injuries.12 A 2011 Cochrane review of the safety effects of home modification found that few preceding studies had demonstrated safety benefits.13 Reasons for the significant finding in the HIPI study, in contrast to many previous studies, may include: a relatively risky home environment in the homes studied,14 for which there was greater potential for environmental safety improvements; a comparatively large sample studied of 1850 people; a long follow-up period of 3 years on average after the intervention; a high participation rate.12 These HIPI data are the basis for the cost–benefit analysis in the current paper. A limited economic analysis has already been conducted to examine the cost-effectiveness of the intervention tested in the HIPI trial.12 The cost of the intervention per discounted DALY prevented over a 20-year period was found to be$14 300, considerably less than NZ GDP per capita, which was $45 769 in 2012,15 classifying the intervention as ‘very’ cost-effective according to WHO criteria.16 ,17 The HIPI trial studied the general population, but did not consider the effectiveness of the intervention for subgroups of the population. This paper has two aims. First, to estimate the benefit–cost ratios of the home modifications tested. Second, to consider the cost-effectiveness of targeting the intervention. Two potential target groups are older people, whose higher risk of falls is well-documented, and people with a prior history of falls, who were estimated to have significantly higher risk in the HIPI study, controlling for age. ## Methods ### Participants and trial design The HIPI study has already been described.12 ,14 Briefly, this was a single-outcome cluster randomised trial in which occupants of modified homes were found to have a lower rate of home fall injuries over a subsequent 3-year period. Homes were all owner-occupied and constructed before 1980, with at least one occupant who was a holder of a Community Services Card. These cards are held by the unemployed, students, pensioners and sickness beneficiaries, providing entitlement for state benefits and subsidies. The outcome measures, counts and costs of medically treated home injuries, were provided by the national no-fault personal injury insurer, the ACC. ACC covers most of the medical costs of injuries (regardless of fault) and 80% of lost earnings except for the first week absent from work. Where the set fee provided by ACC is insufficient to cover treatment costs of doctors, dentists, physiotherapists, specialists, counsellors and other health professionals, the patient pays an additional surcharge. A qualified builder fitted the following modifications, as required: handrails for steps and stairs, some other minor repairs to steps, grab rails for bathrooms and toilets, adequate outside lighting, high-visibility and slip-resistant edging for outside steps, fixing of lifted edges of carpets and mats, non-slip bathmats and slip-resistant surfacing for outside areas such as decks. The average cost per house was around$560 per household (2012 $NZ value—approximately £290). ### Statistical analysis There were differences between the composition of the treatment and control groups as households were the unit of randomisation. A model was fitted to individual-level data, according to our analysis plan, with explanatory variables: age (in prespecified groups), sex, ethnicity (Māori/non-Māori) and injury costs in the year prior to the study. The model fitted used the same study and subjects as the previously reported HIPI results,12 but direct costs to the injury insurer were used as the outcome variable. The covariates included were identical apart from the actual costs of the intervention per household, fitted as a continuous variable and set to zero for the controls. This replaced an indicator variable that was set to 1 for intervention households and 0 for the controls. To allow for over-dispersion and excess zeros relative to the Poisson distribution, zero-inflated negative binomial generalised linear models were fitted. Benefits in terms of injury costs saved were estimated in two different ways: (i) direct costs and social costs calculated according to monetised DALYs and (ii) direct costs and social costs based on the VOSL. These latter values were updated from previous published 2008 values18 to 2012 values using the annual update of social costs produced by the New Zealand Ministry of Transport19 for the transport sector. These account for loss of life and quality of life, loss of output due to temporary incapacitation, medical costs, legal costs and vehicle damage costs. As in a previous study that used these estimates to value all home injuries,18 legal costs and vehicle damage costs were excluded from the costings used here for fall injuries. Medical costs amounted to 0.2%, 3.6% and 5.0% of the resultant costings for fatal, hospitalised and medically treated fall injuries, respectively.19 Virtually all of the medical costs are borne by the New Zealand government via the public hospital system and ACC, the government-owned injury insurer. ### Costing fall injuries by severity level DALYs from falls occurring in all New Zealand settings have been published.9 To estimate the proportion of these occurring in the home, we accessed ACC data on medically treated fall injuries using an online tool20 and analysed Ministry of Health databases on hospital admissions and fatalities. The proportions of injuries of different severities were then weighted to account for different levels of contribution to total DALYs due to falls. To form these weights, we used information on the proportion of fall injury DALYs to which YLL contributed in Australia, using 2003 data.21 This proportion was estimated to be 47% (derived from an estimated 0.623 YLL and 1.327 DALYs due to fall injuries and deaths per 1000 population).21 To estimate the VOSL-based costs of home fall injuries, we used the same data sources on home falls in three levels of severity as shown in table 2, assumed to equate to the levels of severity used for costing transportation injuries.19 The estimated benefits were calculated using an 8% discount rate, which is typical of discount rates used by the New Zealand Treasury when considering public investment.21 ## Results ### Effect of the intervention on injury insurer costs At 36 months, we compared the costs incurred by ACC for fall injuries in the treatment group with those of the control group, controlling for individual-level differences between the groups. Table 1 shows a reduction of 33% in injury costs (a relative cost of 0.67 for the treatment group vs controls). Table 1 Adjusted relative cost estimates from zero-inflated negative binomial model fitted to data at 36-month follow-up ### Cost–benefit analysis Table 2 shows that the home was the setting of 53%, 50% and 39% of falls with defined settings resulting in medical treatment, hospital admission or death, respectively. A given injury will contribute more to the DALY total if the injury is more severe. To estimate the proportion of all fall injury DALYs in the home setting, we therefore used a weighted average of these three proportions. The first two received equal weight (as the severity of the injuries could not be determined without considerable effort and strong assumptions) and the last received 47% of the weight (as described above). The proportion of all fall injury DALYs that occurred in the home setting was estimated to be the resultant weighted average of 46%. Table 2 Annual rate of injuries from falls by severity: medically treated injuries from ACC claims, hospital admissions and deaths (ACC data for 2009–2013, hospital admissions for 2004–2013 and mortality data for 2004–2010) We used a willingness-to-pay based VOSL as applied in the transport sector to derive another estimate of social costs. Table 3 shows costs both for falls generally and for falls in the home. Costs are also shown for carrying out the intervention nationally, assuming the prevalence of fall hazards in the HIPI study is representative. We estimated that it would cost close to a billion dollars ($NZ, 2012 values: third row of table 3) to carry out a programme costing $564 on average per house for 1 734 500 houses nationally.22 Table 3 Using two different bases for social costs: New Zealand social costs of home fall injuries, monetised benefits for 26% reduction and 33% reduction over 20 years and B/C ratios for a national programme of modifications to houses Table 3 shows the total number of DALYs (as estimated by the GBD, 2010)9 valued at$168 712 per DALY (in 2012 $NZ, adjusted for increases in the VOSL from the original value in 2008 of$150 000)11. The 29 500 DALYs estimated to be due to falls annually9 generate a social cost of $4977 million. The$2281 million for falls in the home is derived by taking 46% of this figure, based on an estimate described above. In terms of social costs based on transport sector values, the proportion of all fall injury costs attributable to the home setting was 44%, similar to the proportion of DALYs (and the proportion of costs calculated as a fixed cost per DALY). These two approaches to monetising injury show strongly contrasted results, with much higher values for the VOSL-based approach. The benefit–cost ratios are correspondingly much higher for the VOSL-based benefits, these benefits estimated at 37 times the costs for a 33% reduction. For the DALY approach, the benefits are eight times the costs. Table 3 also shows the benefit–cost ratios for a 26% reduction, which was the estimated reduction in fall injury rates as estimated previously.12

As a further sensitivity analysis (not shown in the table), a different value was used for minor injuries as based on the VOSL. The high prevalence of whiplash injuries in non-hospitalised transportation injuries may lead to greater long-term impairment than fall injuries. Although long-term impairments are likely to feature in some non-hospitalised fall injuries, it was beyond the scope of the current study to examine this. Instead, we have provided a lower bound for these effects by excluding long-term impairment costs altogether for the non-hospitalised falls. This led to a VOSL-based value of the benefits using the injury cost model with an 8% discount rate of $22.1 billion ($NZ 2012 values) and a benefit–cost ratio of 23. These are considerably lower (almost 40% lower) than the equivalent values stated in the last two rows of table 3.

Finally, an average cost per home fall injury prevented, given the estimated effectiveness of the intervention as described, was also calculated. This cost was estimated to be $980 ($NZ 2012 values), using an 8% discount rate for a 33% reduction in fall injury costs. For a 3% discount rate, this break-even cost was estimated to be much lower: $646 ($NZ 2012 values).

## Discussion

This study provides strong evidence that a package of home modifications focused on relatively common home fall hazards is highly cost-beneficial. The benefits in terms of injuries prevented were eight times the cost of the modifications when using a social cost per DALY prevented, and 37 times the cost when injuries were valued using the VOSL. The latter costing is commonly used in transportation and assigns the same value to each life lost or disability acquired as a consequence of home falls, regardless of the age of the victim. Calculation of direct costs due to fall injuries was outside the scope of this study. However, a model fitted to the costs of the fall injuries studied estimated a 33% reduction in fall injury costs per $NZ564 spent on modifications ($564 was the average amount spent for modifications in the treatment group). Given the large injury burden posed by home falls internationally, and the high benefit–cost ratio estimated here, the home environment should be a key focus for prevention efforts.

This is the first randomised controlled study to our knowledge that has analysed injury cost reductions resulting from a safety intervention. The estimated reduction in home fall injury rates of 26% due to the home modifications previously published12 did not consider injury severity. The current analysis accommodated differences in injury severity and consequent injury costs to some extent by modelling costs as borne by the injury insurer (ACC). The costs analysed were nevertheless only a proportion of direct costs (lost income, medical expenses). Nevertheless, it would be expected that changes in costs would provide a better estimate of changes in social costs than the previous analysis of injury rates that do not account for severity.

The higher baseline rates of falls for older people imply higher injury reductions for the given intervention costs. A complicating factor when inferring likely benefit–cost ratios is that the intervention was carried out at the household level. Older people have lower occupancy rates than younger people. In the 2013 NZ census, nationally 8% of people (of all ages) lived in one-person households; for those aged 70 plus, 33% lived in one-person households.23 Analysis of the HIPI study households show that households with at least one person aged 70 plus had an average of 1.5 participants, whereas those without a person aged 70 plus had 2.7 occupants per household. This means that only a little more than half the number of people benefit from the package of modifications when the house includes an older resident. Those aged 70 plus suffer 4.2 times the fall-related DALYs per person per year than the New Zealand population overall.9 Conservatively, the benefit–cost ratio for home modification for people aged 70 plus can be considered at least double the overall ratio shown in table 3.

Likewise, the higher baseline rates of falls for those with a history of falls imply a higher benefit–cost ratio for this group. Those with a history of fall injuries were estimated to have fall costs that are 1.62 times those for people without such a history (see table 1), controlling for age group as well as other factors. If household sizes for this group are similar to the average, the benefit–cost ratio can be considered to be increased by around 60%. Some caution is required when inferring higher injury costs for people with a history of prior falls because of uncertainty around the estimated coefficient (p=0.086).

Although the higher rate of fall injury costs estimated for Māori shown in table 1 similarly implies increased benefit–cost ratios, the estimate of 1.77 has a poor level of statistical confidence (p=0.23). A study (the Māori HIPI study) is currently in progress, which will provide a more precise estimate for Māori. Māori on average have larger household sizes than non-Māori (3.4 residents compared with 2.7 for all ethnic groups in 2001)24. This difference means that a given home modification would benefit more people on average, enhancing the economic rationale for this home safety programme for Māori specifically.

New Zealand housing consists of mainly stand-alone single-storeyed dwellings with outdoor areas, access paths and often steps. Generalising the benefit–cost ratio to other jurisdictions involves considerations of the specific housing stock, local building costs and also the social costs of injury relevant to the society being considered. In addition, the New Zealand VOSL is relatively low internationally,25 so this aspect of the benefit–cost analysis is likely to lead to relatively conservative decisions.

The two approaches to valuing injury (the DALY-based approach and the VOSL approach) described here produce quite different benefit–cost ratios. The DALY-based estimates tend to produce relatively lower values of the costs of fall injuries as fall victims tend to be older than the population average.26 Setting a constant monetary value for a DALY irrespective of the age of the victim has been criticised in the literature.26 A constant value implies that the VOSL would decline with age as older people have fewer remaining life years, but there is no empirical evidence to support a diminishing VOSL.26 Further, if a valuation of injury is being used to justify public expenditure, the VOSL-based estimate has a stronger basis as it is derived from surveys of the public's willingness to pay to reduce injury risk.7 Whichever approach is used, the current analysis shows a compellingly large benefit–cost ratio.

Another avenue for improving the safety of the home environment involves regulation of new construction, undertaken in New Zealand by local government, who enforce national Building Codes.27 Many of the safety modifications retrofitted to existing housing in the HIPI study would not be necessary for new-builds or renovations subject to the current Building Code. Access paths and steps to dwellings are not usually within the scope of the Building Code; however, which focuses on the building itself. As with many other developed countries, the vast majority of New Zealand dwellings are already existent and constructed according to codes or practices that were less stringent: in New Zealand, only around 1.5% of dwellings in 2015 were new builds,28 a relatively high figure in recent history due to ongoing construction following the Christchurch earthquake in 2011. Home safety improvements arising from new construction therefore have little immediate impact, despite yielding likely important longer-term benefits.

With sound economic justification, the package of repairs studied here could feasibly be rolled out nationally, with costs subsidised by government. An example of such a programme to address deficiencies in housing quality was the Warm Up New Zealand: Heat Smart Programme. A strong rationale was provided by two randomised community trials29 ,30 and estimates of cost-effectiveness in terms of health benefits (morbidity and mortality prevented) compared with costs.31 The highly favourable benefit–cost ratio estimated in the current study and the size of the potential injury burden reduction clearly merit a similar effort.

### What is already known on the subject?

• Falls in the home are a major contributor to the health burden of many countries.

• A New Zealand randomised controlled trial showed that medically treated home falls could be reduced by 26% if certain home fall injury hazards were modified.

• By analysing the direct injury insurer costs for the falls included in the above study, we found a 33% reduction (95% CI 5% to 49%) due to the home modifications tested.

• We then extrapolated the benefits of the home modification programme nationally using two different approaches to monetising the injury burden due to home falls.

• Even when conservative assumptions were applied, the programme was found to be highly cost-beneficial.

### Housing assistance and blood lead

The American Journal of Public Health reported that children living in assisted housing had significantly lower blood lead levels than children whose families did not receive housing assistance. This conclusion comes from linking data from the National Health and Nutrition Examination Survey to records from the US Department of Housing and Urban Development (HUD). Comment: It is likely that the involvement of HUD accounts for the seemingly counterintuitive finding.

## Acknowledgments

We acknowledge the support of ACC, who provided matched data for the current study. Thanks also to Jane Zhang of He Kainga Oranga for extracting hospital admissions and mortality data for home fall injuries, and to Arthur Grimes and Anna Hamer-Adams for commenting on an earlier draft.

## Footnotes

• Contributors MDK designed the trial with input from NP, MGB and PH-C. All authors were involved in the conduct of the trial, interpretation of the results and revision and correction of the report, which was drafted by MDK. The analysis was led by MDK. All authors read and approved the final version of this report.

• Funding The trial is funded by the Health Research Council of New Zealand and is registered, number ACTRN12609000779279.

• Competing interests None declared.

• Patient consent Obtained.

• Ethics approval Central Regional Ethics Committee of the Ministry of Health.

• Provenance and peer review Not commissioned; externally peer reviewed.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.