Falls Among Community-dwelling Mid- and Older-Adults in New Zealand

Abstract

Background: Falls and subsequent injuries among older adults (65 years and over) are a major public health issue contributing to increased morbidity and mortality. Risk factors for falls among older adults have been examined extensively over the years. However, there is a dearth of literature regarding risk factors for falls among middle-aged adults (50 – 84 years) − the age where the trajectory for fall rates starts to increase.

Aim and Objectives: The main aim of this thesis is to identify the risk factors for any type (injurious and non-injurious), injurious, and recurrent falls among community-dwelling midand older-adults aged 50- 84 years in New Zealand (NZ).

Methods: A systematic review examined the published literature on risk factors for falls among community-dwelling mid- and older-adults aged 50 years and over which informed the design of the main analytical component of the thesis. A secondary analysis was carried out on data collected as part of the Vitamin D Assessment (ViDA) study; a randomised, double-blind, placebo-controlled trial that recruited 5,108 adults aged 50-84 years in Auckland with over three years of follow-up. Data analysed included sociodemographic, lifestyle characteristics, and medical conditions collected at the ViDA baseline assessment. Prescription data came from the NZ Ministry of Health Pharmaceutical Information Database. Follow-up data on self-reported falls was collected from questionnaires mailed initially monthly, and then 4-monthly, to the home of ViDA participants. Descriptive analyses summarised the distribution of baseline characteristics and prescription medications. A univariate analysis utilising t-tests and chi-square tests was conducted to investigate the cumulative fall risk (for any type of fall, injurious falls and recurrent falls) during follow-up for participants according to different exposure levels. For cohort analysis, cox proportional hazards and negative binomial regression (plus mean cumulative function analysis) were used to examine the risk factors for the fall types of interest. A test for interaction analysis was utilised to compare the relationship between factors across the three outcome measures of interest. Directed Acyclic Graphs (DAG) were constructed to assist with interpreting results and identifying interactions between variables and fall outcomes.

Results: 152 studies met the literature review inclusion criteria. Risk factors were classified into five domains: socio-demographic, lifestyle, physical, medical conditions, and pharmaceutical risk factors. The key findings revealed the following as risk factors for falls: age, sex, ethnicity (White), living alone, marital status, poor self-rated health, depression, diabetes, stroke, angina, arthritis, pain, history of fracture, history of falls, respiratory diseases (asthma, emphysema and shortness of breath), benzodiazepines, antiepileptic/anticonvulsant medications, antidepressants, psychotropic medications, analgesics, diuretics, and respiratory medications. While Black/African American was shown to have a protective effect against falls. Cohort analyses of the outcome ‘any type of fall’ were carried out on 5,049 participants after excluding those who did not return a fall questionnaire during follow-up (n=52) or who were missing time to first fall (n=7). Analyses of the outcomes ‘injurious fall’ and ‘recurrent falls’ were caried out on 5,053 participants after excluding those who did not return a fall questionnaire during follow-up (n=52) or were missing time-dependent variables (time to injurious fall/total no. of recurrent falls data, n=3). Factors associated with increased hazard ratios [HR] of having any fall (all types including injurious) were female sex [HR 1.44, p<0.0001], living alone [HR 1.16, p=0.01], stroke and/or transient ischemic attack (TIA) [TIA only; HR 1.28, p=0.03], fall history [HR 1.77, p<0.0001], decreased confidence to do daily activities without falling [Quite Confident: HR 1.33, p<0.0001], arthritis [HR 1.10, p=0.03], previous fracture (or broken bone) [HR 1.10, p=0.03], asthma [HR 1.16, p=0.01], depression [HR 1.34, p<0.0001], antiepileptic medication [HR 1.26, p=0.02], antidepressants [HR 1.16, p=0.03], and anti-Parkinson medication [HR 1.94, p=0.02]. Medication affecting the renin angiotensin system (ARAS) and education level were the only protective factors associated with reduced hazard of any fall [HR 0.85, p=0.001]. For injurious falls, significant risk factors included female sex [HR 1.48, p<0.0001], living alone [HR 1.21, p=0.0004], employment [retired; HR 1.15, p=0.01], stroke and/or angina [TIA only; HR 1.43, p=0.01], fall history [HR 1.76, p<0.0001], decreased confidence to do daily activities without falling [Quite Confident; HR 1.33, p<0.0001], arthritis [HR 1.11, p=0.03], previous fracture (or broken bone) [HR 1.13, p=0.01], asthma [HR 1.15, p=0.02], depression [HR 1.31, p=0.0002], and anti-Parkinson medications [HR 2.09, p=0.003]. Ethnicity (South Asian) [HR 0.63, p=0.004] and current smoking [HR 0.79, p=0.02] were the only factors associated with reduced hazard of injurious falls. Factors associated with increased incidence rate ratios [IRR] of recurrent falls included female sex [IRR 1.17, p=0.0003], living alone [IRR 1.22, p= 0.0001], employment (retired) [IRR 1.24, p<0.0001], heart attack and/or angina [Angina only; IRR 1.25, p=0.02], stroke and/or TIA [TIA only; IRR 1.47, p=0.0003], fall history (last four weeks) [IRR 1.92, p<0.0001], decreased confidence to do daily activities without falling [Quite, IRR 1.49, p<0.0001; Not at all, IRR 2.14, p<0.0001], chronic pain [IRR 1.23, p<0.0001], depression [IRR 1.24, p=0.002], antidepressants [IRR 1.18, p=0.01] and anti-Parkinson’s medications [IRR 4.07, p<0.0001]. Factors associated with reduced incidence rate of recurrent falls were ethnicity (South Asian) [IRR 0.64, p<0.0001] and BMI (underweight and overweight) [IRR 0.40, p=0.01; HR 0.86, p=0.002 respectively]. A test for interaction indicated that the hazard/rate of falls was consistent across all three fall outcomes, and the following are the most significant risk factors for falls among mid and older adults in NZ: sex (female), living alone, employment (retired), angina, TIA, fall history (last four weeks), arthritis, previous fracture, chronic pain, asthma, depression, antiepileptic medication, antidepressants, and anti-Parkinson medication; while ethnicity (South Asian), education (secondary), current smokers, and BMI (under-weight and overweight) were significantly protective against falls across all three fall outcomes. DAG diagrams were created for all variables that were included in the final multivariable models, which provided a clear picture of mediators and confounding factors that may impede the association of certain factors with falls. These variables (i.e., marital status) were then not included in the final multivariable models.

Conclusion: This study indicates that risk factors for falls (any, injurious and recurrent) among mid and older community-dwelling older adults are complex and multifaceted. In addition to traditional risk factors identified, other factors not commonly researched − such as asthma, TIA only, anti-Parkinson’s medications, and antiepileptic drugs − were shown to be associated with significantly increased risk of falls; while ethnicity (South Asian) and current smokers were shown to have a reduced fall risk in this population sub-group. This is the first cohort study to use pharmaceutical dispensing data to examine the association between certain medications such as psychotropic, antipsychotics, laxatives and fall risk. The findings indicate that older adult fall prevention strategies need to commence in middle age, and consideration needs to be given to addressing novel factors in future risk prevention strategies. Additional research is required to reaffirm the association between medical conditions and specific medication classes on fall risk and to understand why there is a lower risk of falls among South Asian population groups.