Background

Fall-related injuries sustained by older adults have worse outcomes than in younger individuals [1, 2]. Due to complications of ageing, older adults who sustain an injury are more likely to suffer from decreased function, disability, chronic pain, long stays in hospital, and increased mortality [1,2,3]. Falls and fall-related injuries have been noted to be increasing after adjusting for the increasing age of the global population, and this increase in falls is likely associated with an increase in fall-related injuries such as hip fractures [4,5,6]. Due to the seriousness of these injuries, the New Zealand government has made it a priority to improve health outcomes associated with injurious falls [7].

Standardised assessments have been mandated in New Zealand since 2012 to identify home health care needs of community-dwelling older adults. New Zealand uses interRAI-home care (HC) assessments to identify what areas of health care an individual requires. The interRAI-HC assessment contains 236 questions across 20 domains, including cognition, psychosocial factors, health conditions, self-reported falls history, and skin conditions. A copy of the interRAI-HC form can be obtained from the interRAI Home Care Assessment Form and User’s Manual [8]. All New Zealanders who require publicly funded home care services must undergo an interRAI-HC assessment [9].

Fall-related injuries in older adults are a global concern and have been studied widely within the literature. Risk factors identified from previous research include recent hospitalisations, previous history of falls, stroke, Parkinson’s disease, lower limb amputations, cognitive impairment, arthritis, and diabetes [10,11,12,13,14,15,16,17,18]. Additionally, there have been other studies in New Zealand examining hip fracture incidence rates [19,20,21,22]. However, hip fractures are not the only injuries from falls. Few studies have identified risk factors for other significant fall-related injuries such as pelvic fractures, fractures of the distal radius, or proximal humerus fractures. A 2017 study found that falling onto low-impact flooring reduced the number of fall-related injuries compared with falling onto standard vinyl flooring [23]. Elley et al. conducted a trial on community-based older adults who had a previous fall, and they found there was no difference in the number of falls an individual had, after nurse-led interventions were put in place [24]. However, an earlier study conducted by Campbell et al. found that falls and fall-related injuries were reduced in women aged 80 years and older who took part in a tailored exercise programme [25].

Work on reducing falls has been conducted, but it is also important to work towards reducing fall-related injuries [26]. One way to achieve this is by identifying risk factors associated with fall-related injuries and to see if they are different from those that predict hip fractures. The primary objective of our study was to determine risk factors for trauma related injuries in community-dwelling older people receiving home care services in New Zealand. Our secondary objective was to identify risk factors for the non-hip fracture injuries.

Methods

Study design

We conducted a retrospective time-to-event study from a national cohort.

Participants

Participants included New Zealand resident adults aged 65 years and older who had an interRAI-HC assessment between 1 July 2012 and 1 June 2018. All participants were living in the community at the time of their assessment and consented for their information to be used for planning and research purposes. Where an individual had multiple assessments, the first interRAI-HC assessment was utilised. We randomly selected 75,484 individuals for the development cohort.

Instruments/variables

The interRAI-HC version 9.1 (© interRAI corporation, Washington, D. C., 1994–2009) assessment tool is a comprehensive geriatric assessment that consists of 236 questions across 20 domains including Mood and Behaviour, Functional Status, Disease Diagnoses, and Oral and Nutritional Status [27, 28]. In New Zealand it has been mandated that this assessment tool be used for all older adults who require publicly funded home care services or for people entering aged residential care. Health practitioners refer individuals for an interRAI-HC to assess their health care needs. Each assessment is carried out by a trained interRAI assessor who has undergone rigorous training. The assessment will usually be carried out in the individual’s home and the assessor uses a variety of sources to complete assessments; including, checking the patient’s medical records, interviewing the individual and sometimes their family members [29]. The assessment data are entered into an electronic database and the data are stored by New Zealand’s Technical Advisory Services (TAS).

Approximately 93% of people undergoing an interRAI-HC assessment consent for their information to be used for planning and research purposes [27]. The assessments can be linked to other health datasets such as mortality and hospital admissions using a National Health Index (NHI) number. An NHI number is a unique identifier allocated to anyone who receives health care services in New Zealand [30].

Within the interRAI-HC assessment, individuals may choose up to three ethnicities; for analysis purposes we employed priority coding according to New Zealand’s Ministry of Health (MoH) guidelines [31]. Where an individual indicated more than one ethnicity priority was given to Māori, then Pasifika, and then Asian ethnicities.

Hospital admissions data for the injuries of interest were obtained from the National Minimum Dataset (NMDS) [32], released from the MoH with encrypted NHI numbers for all consenting interRAI-HC participants. Injuries chosen for this study were identified as common fall-related injuries within the literature and only those injuries that led to the individual being admitted to hospital. For instance, people who sustain a wrist fracture are less likely to be admitted to hospital. The injuries included for analysis were fracture of the distal radius, hip fracture, pelvic fracture, proximal humerus fracture, subarachnoid haemorrhage, traumatic subdural haematoma, and vertebral fracture. Injuries were identified using ICD-10-AM (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification) diagnostic codes I60 I62.0 S22 S32 S52.2 S42.3 S49.0 S720 S721 S722 S723 S724 S728 and S729. Ecodes (External cause of injury codes) were unavailable for all hospital admissions and were unable to be utilised. The first hospital admission after an individual’s assessment was used. The hospital admission dataset contains up to 20 different diagnosis codes. For cases where the individual sustained multiple injuries of interest, the first instance of one of the injuries listed above was noted down. When censoring individuals with hip fracture if there was any instance of hip fracture in any of the 20 diagnostic codes, the individual was censored from analysis. Mortality data were provided by the MoH from their Mortality Collection (MORT) [33] and included encrypted NHIs so they could be matched to the interRAI-HC assessments.

Statistical analysis

Reporting of the results adhered to RECORD (REporting of Studies Conducted using Observational Routinely collected Data) guidelines [34] to ensure this study reports results accurately and clearly. Basic frequency distributions of each variable of interest were reported. Multicollinearity was tested for by examining the variance inflation factor of each variable of interest. Most questions within the interRAI-HC are mandatory and therefore there was very little missing data. Body Mass Index (BMI) had a high percentage of missing values and these were incorporated into an ‘Unknown’ category. All other missing variables were included in analysis due to their low numbers. Risk factors for injuries were determined using competing risk regression models with the Fine and Grey method [35], injurious falls were the failure event and death was the competing event. Unadjusted models of each variable of interest were utilised and an adjusted model including all variables and adjusting for age, sex, and ethnicity were conducted. Subhazard ratios (SHRs) and 95% confidence intervals (CIs) were reported for each variable in the model. A competing risk regression model treating hip fracture as a censored event was also conducted to identify whether there was a difference between the risk factors for hip fracture and other injuries of interest. IBM SPSS version 26 [36] was used for descriptive analyses and data cleaning, and Stata IC version 16 [37] was used to run competing risk regressions, and α = 0.05 defined statistical significance.

Ethics

Ethics permission for this study was granted by the Ministry of Health’s Health and Disability Ethics Committees (14/STH/140). Only those individuals who consented to their data to be used for research purposes were included in this study.

Results

Participants

After exclusions were applied, there were a total of 75,484 participants. Figure 1 below, details the exclusion criteria.

Fig. 1
figure 1

Exclusion criteria for interRAI HC assessments

Full size image

Demographics

Participant’s mean age was 82.1 years (range: 65 to 107 years), with 44,939 (59.5%) females. Approximately 87.5% (66080) of people identified as European ethnicity and 5.8% (4348) people identified as Māori. Approximately 61.9% (46733) of people required an assistive device such as a cane, walker, or pushing wheelchair. Over half of individuals reported minimal to moderate levels of fatigue (58.2%).

Injuries and deaths

Within the cohort over the 6-year period 7414 (9.8%) sustained an injury, 36,212 (48.0%) died and the remaining 31,578 (41.8%) had not been admitted to hospital for a fall-related injury or died by the end of the study period. Most injuries sustained were hip fractures (4735 63.9%), followed by fractures of the proximal humerus (735 9.9%). A full list of injuries and their frequencies can be found in Table 1. Median follow-up time after the interRAI-HC assessment was 21.5 months (25th percentile 8.5 months, 75th percentile 39.0 months) from assessment, with a total person-time of 157,134 years and the rate of injurious falls was 47 per 1000 person-years.

Table 1 Distribution of injury type sustained
Full size table

Unadjusted and adjusted analyses

The results of the unadjusted and adjusted analyses for the demographic variables are displayed in Table 2. People who were living with others were less likely to sustain an injury than those who lived alone (SHR 0.94 95% CI: 0.89–0.99). The risk of sustaining a fall-related injury increased for those aged 65 to 94 years, but those who were aged 95 years and older were slightly less likely to sustain an injury than those aged between 85 and 94 years. Females were more likely to sustain an injury than males (SHR: 1.31 95% CI: 1.24–1.38).

Table 2 Demographic frequencies for each event and unadjusted and adjusted competing risk regression results
Full size table

Table 3 presents frequency distributions and the unadjusted and adjusted SHRs for each variable of interest. Risk factors for injuries were sex, age, living alone, Parkinson’s disease, stroke/CVA, falls, unsteady gait, tobacco use, and being underweight. Cancer, dyspnoea, high BMI, and a decrease in the amount of food or fluid usually consumed, were associated with a reduced risk of sustaining an injury.

Table 3 Variables of interest frequencies for each event and unadjusted and adjusted competing risk regression results
Full size table

Injuries excluding hip fracture

Table 4 presents the adjusted SHRs after removing hip fractures from the analyses. After removing hip fractures from the injuries sustained, female sex, age, previous falls, tobacco use, and being underweight were associated with an increased risk of injury. High BMI and dyspnoea were associated with a reduced risk of injury. Females were 1.3 times more likely to sustain an injury than males (SHR 1.30 95% CI: 1.19–1.42). Compared to people who had no dyspnoea, individuals who had dyspnoea at rest were the least likely to sustain an injury (SHR: 0.54 95% CI: 0.44–0.67).

Table 4 Adjusted competing risk regression with hip fractures censored from analysis
Full size table

Discussion

Key findings

There were a total of 7414 (9.8%) people who sustained a falls-related injury. We identified female sex, older age, living alone, a diagnosis of Parkinson’s disease, stroke/CVA, previous falls, unsteady gait, tobacco use, and being underweight were associated with an increased risk of sustaining an injury. In contrast, a reduced risk of injury was associated with a diagnosis of cancer, dyspnoea, high BMI, and a decrease in the amount of food or fluid usually consumed. Two mobility-related risk factors had a non-monotonic association with injury (Primary Mode of Locomotion and Timed 4 Metre walk) both suggesting that those unable to walk were not significantly associated with fall-related injury.

Hip fractures comprised nearly two-thirds (63.9%) of all injuries sustained by older people, and when these were removed from analysis there were fewer factors associated with injury. The factors that were associated after hip fractures were censored were sex, age, previous falls, dyspnoea, tobacco use, and BMI.

Findings within the literature

Co-morbidities were associated with fall-related injuries. Stroke/CVA and Parkinson’s disease were both associated with an increased risk of injury. Previous studies have shown that stroke/CVA is associated with an increase in falls risk [38, 39]. Additionally, several studies have identified that individual’s with Parkinson’s disease are at an elevated risk of falling and sustaining an injury [40]. Our study found that cancer was associated with a reduced risk of injury. However, many studies exploring the risk of falls and fall-related injuries in cancer patients have identified that individual’s with cancer were at an elevated risk of falling and sustaining an injury [41, 42]. Schluter et al. included cancer in an adjusted model for identifying if incontinence was associated with falls within an interRAI-HC cohort. Their model showed that people with cancer had a reduced risk of falls [43]. In an interRAI-HC setting individuals with cancer appear to be at a reduced risk of falls and fall-related injuries. This is possibly due to behavioural differences with cancer patients being less likely to take part in activities such as walking that can lead to falls and falls-related injuries. This could also explain the lower risk in individuals who experience dyspnoea, as they are less likely to undertake extraneous activities and therefore be at a reduced risk of falling.

BMI was associated with fall-related injuries. Those who were underweight were 1.2 times more likely to sustain a fall-related injury than people with a normal BMI. Previous research has suggested that those with a low BMI are more likely to be at risk of multiple types of fractures [44]. It has also been noted in the literature that those who are overweight and obese are at a reduced risk of sustaining an injury [44, 45].

Environmental factors such as disrepair of the home, and limited access to rooms were not associated with fall-related injuries in the interRAI-HC. This may be due to the limited number of people who have such problems with their home environment. Previous studies have identified items common to homes such as hard-flooring, loose rugs, and inadequate lighting as risk factors for falls and fall-related injuries [46, 47].

Limitations of the study

Osteoporosis is a well-known risk factor for many bone fractures and this was unable to be included as a variable. The New Zealand interRAI-HC version does not contain osteoporosis as a diagnosis; therefore, we were unable to identify anyone who had a diagnosis of osteoporosis within the cohort. Additionally, the BMI data for many individuals were missing with approximately 35.7% of individuals having no recorded BMI. This is most likely because it is difficult to measure height and weight data for frail individuals, particularly as many people in this cohort have mobility issues and stepping up onto a scale may be challenging. The results show BMI, was significantly associated with injuries suggesting that BMI is an important risk factor and may have clinical implications.

Ecodes were unavailable for a number of hospital admissions information. This means that some of the injuries sustained may not have occurred due to a fall. However, the injuries chosen are common falls-related injuries, so we anticipate that most injuries included were sustained from a fall.

Our study used New Zealand data of older people who require health care services and may not be generalizable to a larger international audience or to a healthy population of New Zealand older adults. However, the results may be generalizable to other home care cohorts internationally.

Summary of the implications of the work for practice and research

The New Zealand MoH has implemented the New Zealand healthy ageing strategy to provide support for people with high and complex needs, including those who require home care services, so they can live as independently as possible [48, 49]. The interRAI-HC is a useful tool for determining the appropriate care programme for each individual. Fall-related injuries lead to reduced quality of life and significant disability for older adults. While falls prevention programmes are being implemented, it is also important to reduce the risk of serious injury after a fall [23]. Determining risk factors for a range of fall-related injuries can allow for targeted interventions to help reduce injuries sustained. Additional work will also be conducted to develop an injury risk score to identify those who may be at an elevated risk of sustaining a falls-related injury. While there are already well established injury prediction models such as the FRAX and the Garvan, an injury score developed for a frail community-dwelling cohort will more precisely predict who is at risk of injury [50,51,52]. Additionally, the FRAX and Garvan scores predict the 5- and 10-year risk of hip fracture, whereas the home care cohort may be unlikely to live that long, so a 1- or 2-year risk of injury would be better utilised for this population.

Conclusions

Risk factors for injuries were female sex, older age, living alone, Parkinson’s disease, stroke/CVA, falls, unsteady gait, tobacco use, and being underweight. Cancer, dyspnoea, high BMI, and a decrease in the amount of food or fluid usually consumed, were associated with a reduced risk of sustaining an injury. While it is important to reduce the risk of falls, it is also important to reduce the risk of falls-related injuries. Knowing risk factors associated with these types of injuries can help to develop focused intervention programmes. Additionally, development of a predictive model to identify those who would benefit from intervention programmes would be beneficial.