Background
Hip fractures are among the most sinister outcomes of falls in elderly populations due to detrimental effects on functional capacity [
1] independence [
2] and mortality [
3]. The vast majority of hip fractures are preceded by a fall, so predisposing risk factors for falls and hip fractures are in many cases similar [
4-
8]. While predisposing risk factors such as high age and comorbidities indicate susceptibility to hip fracture, they do not explain the ways in which the falls which precede hip fractures occur. To understand the circumstances surrounding fall-related hip fractures the precipitating factors for these falls need to be examined and evidence for these is sparse in the literature.
Previous studies which provide the most detailed accounts of fall circumstances have been based in residential care settings [
9-
13]. Based on a research teams analysis of video-captured falls, incorrect shifting of body weight, followed by trips and stumbles were reported as the primary causes of imbalance [
10]. When resident clinicians have, on the other hand, analysed falls not captured on video, symptoms of acute disease have been cited as the primary precipitating fall risk factors [
11]. Although third party observations of falls in residential care settings provide valuable insights into fall mechanisms, they do not account for subjectively experienced factors such as dizziness or pain, or provide contextual health details of the person who fell [
9,
13]. Furthermore, these studies are not generalizable to all patients with hip fracture since the majority of hip fractures occur unobserved in community settings [
14]. Findings from interview studies which have explored older people’s perspectives on falling, on the other hand, describe a diversity of fall perceptions involving both individual [
15,
16] and environmental or accidental precipitants [
17,
18]. Explorative studies however, rarely place these perceptions in the context of the health of the faller [
17,
19-
21], which may explain these reported diversities in fall perceptions [
22].
Characteristics of poor health and function are known to predispose to falls and hip fractures. Physical function is an indicator of health among older people [
23] and is commonly measured using self-reported performance of mobility, personal activities of daily living (P-ADL), balance, and physical activity (PA) [
24-
26]. A non-linear relationship is reported between gait speed and falls, with those at lower and higher levels at greatest risk [
27], a relationship possibly mediated by exposure to environmental hazards [
28]. Impaired balance [
29,
30] and muscle weakness, particular of the lower limb [
31], are also associated with increased risk for falls and hip fractures. Additionally, impairments in physical function may be the manifestation of underlying chronic disease, or the side effects of drugs which treat disease, thus comorbidity and medication use are important factors for consideration in fall investigations [
32,
33]. Hip fracture is, in almost all instances, preceded by a fall and treated in hospital. By approaching all patients with incident hip fracture, it is therefore possible to identify a population based sample of persons who have experienced a fall.
Fall circumstances and the predisposing health characteristics of those who fall have, to date, mostly been investigated in separate studies. Placing fall descriptions within the health context of the person who falls may help identify important patterns, which can inform preventive efforts. The main purpose of this study was to investigate the circumstances surrounding fall-related hip fractures and to describe these circumstances in relation to participants’ health and functional characteristics. More specifically, we aimed to explore the fall descriptions of patients, without cognitive impairment, who were admitted to a Swedish hospital with hip fracture during a 10-month period. Based on qualitative content analysis, categories were created and we then aimed to explore whether health and functional characteristics varied across these categories and whether there were identifiable patterns in the data.
Methods
Design
This descriptive study was explorative in nature and used a concurrent triangulation mixed methods design [
34]. With this design, qualitative and quantitative data are collected concurrently, analysed separately and then merged. The two forms of data were merged in the results section using a joint display of the data [
35] and then interpreted in the discussion section. The rationale for this approach is that of complementarity whereby fall descriptions will explain one aspect of fall circumstances and quantitative health data are collected to elaborate and enhance these descriptions [
36].
Participants
All patients aged 50 years and over, with radiographically confirmed hip fracture (S72.0–S72.2 according to International Classification of Diseases (ICD-10)), admitted to a Swedish hospital during the period August 2009 − June 2010, were considered for inclusion in the study. The hospital is responsible for all acute care and surgery in the catchment area of Uppsala County, which enabled us to identify all hip fractures in this population during the study period. Participants were enrolled consecutively during hospital stay, if they fitted the following inclusion criteria; memory of the fall and verbal ability to recount its details and a Mini-mental State Examination (MMSE) score ≥24 points. Those with reduced consciousness, post-operative confusion or medical instability were not considered for participation, in this otherwise population-based setting. The interviews were performed during hospital stay within a one week period of when the fall event had occurred.
Of the 350 patients with hip fracture approached during the 10-month period, 192 people were unfit for interview due to; diagnosed dementia, communication difficulties or postoperative complications such as reduced consciousness, confusion, or medical instability. A further 29 patients were excluded following MMSE (score <24 points), leaving 129 participants eligible for interview. Following initial interviews analysis four participants were excluded due to fall recall uncertainty. Pre-fracture characteristics of the 125 men and women aged 55–96 years included in the present study are presented in Table
1.
Table 1
Characteristics of 125 interviewed participants
Demographics
| | | |
Age, years (mean(SD)) | 77.6 (9.1) | 79.7 (9.5) | 79.1 (9.4) |
Residential status | | | |
Community-dwelling | 34 (94.4) | 83 (93.2) | 116 (92.8) |
Serviced apartment | 1 (2.8) | 2 (2.3) | 3 (2.4) |
Residential care | 1 (2.8) | 4 (4.5) | 5 (4.0) |
BMI (n = 109)
| | | |
Underweight | 2 (6.3) | 11 (14.3) | 13 (11.9) |
Normal weight | 16 (50.0) | 44 (57.1) | 60 (55.1) |
Overweight/Obese | 14 (43.7) | 22 (28.6) | 36 (33.0) |
Mobility
1
| | | |
Low2 | 10 (27.8) | 26 (29.2) | 36 (28.8) |
Moderate3 | 5 (13.9) | 26 (29.2) | 31 (24.8) |
High4 | 21 (58.3) | 37 (41.6) | 58 (46.4) |
P-ADL Participation
1
| | | |
Dependent in ≥1 activity of personal care | 5 (13.9) | 8 (8.9) | 13 (10.4) |
Balance
1
| | | |
Self-rated balance (bad), (n = 114) | 20 (62.5) | 39 (47.6) | 59 (51.75) |
Fear of falling (yes), (n = 122) | 9 (26.5) | 27 (30.7) | 36 (29.5) |
Previous falls
| | | |
≥1 fall previous year, (n = 122) | 23 (63.9) | 37 (43.0) | 60 (49.2) |
Physical activity
1
(n = 118) | | | |
Sedentary | 8 (22.2) | 14 (15.7) | 22 (17.6) |
Light exercise | 26 (72.2) | 67 (75.3) | 93 (74.4) |
Hard physical training | 1 (2.8) | 2 (2.3) | 3 (2.4) |
Grip strength (n = 107) | | | |
Normal (≥10th percentile) | 17 (51.5) | 43 (58.1) | 60 (56.1) |
Low (5th-10th percentile) | 7 (21.2) | 16 (21.6) | 23 (21.5) |
Abnormally low (<5th percentile) | 9 (27.3) | 15 (20.3) | 24 (22.4) |
Number of chronic diseases
5
| | | |
0 | 16 (44.4) | 51 (57.3) | 67 (53.6) |
1 | 13 (33.3) | 29 (32.6) | 41 (32.8) |
≥2 | 8 (22.2) | 9 (10.1) | 17 (13.6) |
Fall-risk-increasing drugs (FRIDs)
| | | |
0 FRIDs | 6 (16.7) | 18 (20.2) | 24 (19.2) |
Cardiovascular (Cvd) FRIDs | 13 (36.1) | 35 (39.3) | 48 (38.4) |
Psychotropic (Psy) FRIDs | 5 (13.9) | 8 (8.9) | 13 (10.4) |
Concomitant Cvd & PsyFRIDs | 12 (33.3) | 28 (31.5) | 40 (32.0) |
Participants received written information about study details and were given time to read this information before choosing whether to participate or not. All participants gave their verbal and written consent for inclusion in the study. The study was approved by the Regional Ethics Committee in Uppsala, Sweden.
Data collection
Fall circumstances
Qualitative interviews explored circumstances of the first two phases of falling, as outlined by Noury et al. [
37] firstly the ‘pre-fall phase’ involves the last activity carried out before falling; secondly, the ‘critical phase’, involves the sudden movement of the body towards the ground, until contact is made with the ground or an obstacle. The final two of Noury et al.’s falling phases (the post-fall and recovery phase) were not in focus in the study. During semi-structured interviews, performed by the same investigator, participants were encouraged to speak freely while describing fall events. The interview opened with the question ‘Can you describe for me what happened when you fell and broke your hip?’ Follow-up questions also aimed to investigate the eventual influence of other specific environmental, individual or situational factors which may have played a role in the fall occurrence. Interviews varied in length from 8 to 25 minutes and were recorded and transcribed verbatim. Each interview was followed by an interview-administered questionnaire containing closed-answer questions concerning more specific details of the fall, e.g. time of the fall and use of mobility aids.
Data concerning pre-fracture functional status were collected by interview-administered questionnaire. Questions related to prior performance of P-ADL, mobility, balance, previous falls and PA participation and were categorized as described below. Hand grip strength was tested, by the same investigator, using the ‘Baseline’ hydraulic hand dynamometer (Fabrication Enterprises Inc.) with the elbow flexed and supported at 90 degrees and the bed backrest elevated to achieve a high-lying position. Participants performed three maximal grip contractions of the dominant hand and the amount of time to maintain the grip was not specified. The mean value of the 3 automatically recorded contractions (measured in kg force) was calculated and is considered a valid and reliable method of measuring muscle strength [
38], in turn an indicator of functional capacity [
39]. Data concerning pre-fracture comorbidity and medication were retrieved from medical records. Comorbidity was categorized as 0, 1 and ≥ 2 chronic diseases, based on Charlson’s unweighted comorbidity score, calculated from ICD-10 diagnoses [
40,
41]. Charlson’s comorbidity index predicts the 10 year mortality for a person who may have a range of comorbid conditions. Fall-risk-increasing drugs (FRIDs), i.e. drugs with the strongest evidence for increasing fall risk such as, for example, benzodiazepines, antidepressants or anticonvulsants [
42,
43], were categorized as: no FRIDs, psychotropic FRIDs (PsyFRID; Anatomical Therapeutic Chemical (ATC) classification system codes N02A, N03A, N04A-B, N05A-C, N06A), cardiovascular FRIDs (CvdFRID; ATC codes C01A, C01BA, C01D, C02, C03, C07-C09, G04CA) and concomitant use of PsyFRID and CvdFRID.
Data analysis
Analysis of fall circumstances from interviews
All interviews were recorded and transcribed and then systematically analysed using qualitative content analysis, which is a replicable and valid method for making specific inferences from text [
44]. The analysis focused on the manifest content of the data with the aim to produce descriptive categories, a process which involves interpretation at lower levels of abstraction [
45]. The analysis was initiated by repeated reading of each interview to get a sense of the entirety [
46]. Meaning units of analysis were then identified and constituted sentences in the text describing i) physical activities in the pre-fall phase and ii) circumstances and precipitating factors during the critical phase. These text units were then condensed and grouped on the basis of similarity to form sub-categories and then categories for both ‘Activity at the time of the fall (pre-fall phase)’ and ‘Nature of the fall (critical phase)’. Categories were labelled to reflect participant descriptions and in consideration of the literature, so as to enable a comparison of findings. To ensure validity, this manifest analysis was constantly checked and compared with the original interview data. The first author was responsible for the primary coding of all interviews and to ensure trustworthiness of the analysis, emergent categories were adjusted and refined during repeated peer-debriefing sessions between the first and last authors [
47]. In these sessions, the last author acted as a peer to the first author through a process of review and discussion of the credibility of categories in the data. Interview analysis preceded and was performed independently of the analysis of additional fall details, such as time of fall and use of a walking aid, which were collected by questionnaire and summarized by means of percentage, frequency.
Analysis of health data from questionnaires, medical records and dynamometry
Pre-fracture mobility was divided into the categories ‘Low’, ‘Moderate’ or ‘High’ which were defined as follows: Low (required a walking aid indoors); Moderate (required a walking aid outdoors only); High (no walking aid required). P-ADL participation was categorized as independent/dependent in ≥ 1 activity of personal care, whereby for example participants who required assistance/supervision during bathing were considered dependent in one activity. With regards to PA participation, participants answering ‘hardly any exercise’ to the question ‘How much exercise did you engage in prior to the fracture?’ were classified as ‘Sedentary’, with other divisions including ‘Light exercise’ (Lighter exercise such as regular walks or gardening) or ‘Hard physical exercise’ (Do you engage in hard physical training or sport?). This question was based on a questionnaire, created in collaboration with the Swedish national institute of health which measures lifetime physical activity [
48], and has been used in previous cohort studies of older Swedish populations [
49,
50]. Absolute grip strength values were compared with published normative grip strength data [
51], and stratified according to age, gender and height. ‘Low’ grip strength incorporated values between the 10th and the 5th percentile of normative data and ‘Abnormally low’ grip strength as those under the 5th percentile of normative data. In this calculation, 7 individuals (1 man and 6 women) lacked height data so for these we used the median height in the sex-specific age group in our sample (women > =65y: 164 cm; men > =65y: 175 cm). This imputation was only used for the definition of low grip strength.
Merging of the fall circumstance and health data
Following separate analysis of the interview and health-related data, both data sets were interrelated. To facilitate the integration of the data sets, qualitative fall circumstance categories were treated as categorical variables. We organized the data in relation to the fall circumstance categories which were viewed in relation to variables of health and function using cross tabulation and descriptive statistics. This triangulation of data occurred during both the analysis and interpretation phases with the purpose of interrelating and identifying patterns in the data [
52].
Statistical analysis
Descriptive statistics are presented as mean (SD) for continuous variables and as number (proportion) for categorical variables. The descriptive tables directly display relations among the study categories [
53] and numerical comparisons are presented. In addition, statistical comparisons between groups of fall circumstances were performed by using analysis of variance for age and Fisher’s exact test for categorical variables. Data management and statistical analyses were performed using Stata version 12 (StataCorp, College Station, TX, USA).
Discussion
Being descriptive in nature, the present study aimed to identify patterns between fall descriptions and health characteristics of 125 cognitively unimpaired people with hip fracture, who were included from a population-based sample of patients admitted during a 10-month period. Firstly, those who fractured during positional change had the poorest functional status (greatest mobility and P-ADL limitations, poorest self-rated balance, greatest fear of falling and previous falls, and lowest physical activity participation). The majority of these falls were also described in terms of physiological precipitants. Secondly, participants describing indoor falls of environmental nature had a higher reported and observed physical function (fewer mobility and P-ADL limitations, better self-rated balance and hand grip strength) than those describing falls involving physiological factors, but had nonetheless, an equally high prevalence of comorbidities and fall-risk-increasing drug use and previous falls. Thirdly, the health and functional characteristics of those fracturing outdoors in snow-free environments were more similar to those fracturing indoors than to other outdoor groups.
Most previous investigations of fall circumstances have reported falls in general and not specifically those resulting in hip fracture [
54-
57], which limits the comparability of our findings. We observed however, similar proportions of hip fractures occurring while walking as well as a greater proportion of men who fractured outdoors as previously reported [
57].
Falls during positional change amongst those with poor physical function
Previous studies which have dichotomized falls into indoor or outdoor falls have established that people who fall indoors have poorer health characteristics than those falling outdoors [
58-
60]. By further division of falls according to activity at the time, our findings go on to suggest that those whose fall occurred during positional change had characteristics implying poorer function than other indoor fallers. Possible explanatory mechanisms for these observed tendencies may be found in the literature. Studies which have examined movement transitions among older people have, for example, demonstrated that those with fear of falling perform sit-to-walk in a way which threatens postural stability [
61,
62]. Incorrect shifting of bodyweight has also been reported as a major contributor to falls among nursing home dwellers, a population also characterized by poor health status [
10]. Practical implications of these findings could involve a focus on task-specific training of muscular strength, postural control and compensatory strategies during chair and bed rises for older people who have fallen during positional change. Additionally, in consideration that many (15/24) of those fracturing during positional change in the current study, described physiological symptoms such as dizziness or disequilibrium, factors such as postural, or orthostatic hypotension may also have precipitated the fall. The diagnosis of orthostatic hypotension, which commonly presents asymptomatically, is not fully defined and it is recognized that good clinical judgment is of importance when investigating and managing this condition [
63,
64]. Evaluation of orthostatic hypotension may therefore be especially indicated as part of a comprehensive falls assessment among those presenting with falls and fall-related hip fractures following positional change. Further investigation is required in relation to participants who are unable to describe the nature of their falls for two reasons. Firstly, although we observed a tendency for this group to have a poor overall health status, we were unable to draw conclusions due to the small number of these participants (n = 10). Secondly, others report that people who are unable to describe the reason for their fall tend also to restrict their post fall activities and environments [
20], which further implies that such individuals are at higher risk for poor outcomes following the fracture.
Falls of environmental nature amongst those with high levels of comorbidity and medication use
Those who described indoor falls of environmental nature appeared to have higher levels of physical function when compared to those describing falls of physiological nature. Evidence for the association between higher levels of mobility and environmental hazards can be found in the literature [
65,
66]. However, we also observed equally high levels of comorbidity, FRID use and previous falls among these two groups of indoor fallers. Despite the frequency by which older adults attribute environmental factors for their falls, evidence linking environmental hazards and falls appears weak [
28,
67,
68] and it is sometimes proposed that older people attribute falls to the environment as a strategy to deflect from failing health or feelings of vulnerability [
10,
69]. Our findings concerning the levels of disease and medication usage among people describing falls of environmental nature may offer further explanation to this pattern. One possible interpretation could be that whilst stronger beliefs in mobility performance predispose these individuals to hazardous or challenging environments, they may, on the other hand, underestimate the adverse effects of comorbidities and medications on their postural control and protective reactions in these situations. Nonetheless, the high prevalence of previous falls among this group highlights the importance that clinicians pose questions concerning fall history and health factors to those presenting with falls, regardless of perceived cause, as medication revision and fall preventive advice may be required. Interestingly, no participant described drug side-effects as a fall precipitant, despite strong evidence in the literature for the association between certain drugs and falls and fractures [
42,
70,
71].
Health variations according to the nature of outdoor falls
Outdoor falls are generally associated with vigorous elderly people but previous investigations have analysed outdoor falls as a homogenous group [
22,
58-
60]. Our findings suggest that those fracturing outdoors in snow free environments had a poorer health status than other outdoor groups. This implies that not all older people fracturing outdoors can be considered vigorous and those falling under less challenging circumstances may therefore require extra measures to ensure safe mobility and offset future fear of falling or outdoor activity restriction, both commonly arising consequences of hip fracture [
72-
74]. The role played by the physical environment in falls causation is thought to be mediated by health and function, which in turn increases/decreases an older person’s exposure to environmental hazards or hazardous situations [
28,
75]. Nevertheless, despite poorer observed general health, no participant who fractured outdoors in snow-free conditions in the current study described their fall in terms of health factors. This further highlights the disparity between older people’s and experts’ accounts of the nature of falls and related fractures. However, due to the small numbers of participants involved in the outdoors groups it is difficult to draw conclusions regarding health patterns and these findings must be regarded as preliminary. Further study is thus required regarding interactions between environmental, health and behavioral factors in relation to hip fracture occurrence and whether perceptions of fall circumstances are associated with the adoption of fall prevention strategies.
Strengths and limitations
This is the largest study thus far to integrate qualitative data concerning fall descriptions with quantitative data outlining health characteristics and thereby yields a multi-faceted analysis of fall-related hip fractures, which are by nature complex multifactorial events. To ensure data dependability, all interviews were performed by the same investigator who also compared fall accounts with medical records and, when possible, with observer accounts of the fall event. However, falls are often emotive occurrences and it is not possible to fully ensure the accuracy of the fall descriptions. Self-reported functional performance may be influenced by personal and health characteristics and cannot be solely considered a reflection of performance capability. Pre-fracture objective and more detailed measurements of health patterns and functional performance including mobility, physical activity performance, perceived balance, disease severity and timing of medication would have been optimal but were in this study setting unobtainable. We also lacked data concerning bone mineral density, visual capacity and time spent indoors and outdoors. Being a descriptive and explorative study, the statistical analysis is kept simple with cross tabulations of relevant information. To minimize the time elapsed between the fall occurrence and data collection, interviews were performed during hospital stay. This approach may however have negatively affected the measurement of health-related variables such as grip strength. It is also interesting to note that less than one third of all patients with hip fracture, approached during the 10-month study period, were capable of verbally recounting details of their fall. Therefore, our findings cannot be transferred to those with cognitive impairment or others fracturing in institutional care. The health characteristics of our sample correspond, on the other hand, well with community-dwelling patients with hip fracture in the population-based cohort from which they were included [
14]. By the descriptive and explorative nature of our study we strived for an inductive approach to the text analysis, not involving testing of a theory or hypothesis. To display relations between study categories, we have therefore limited the statistical analysis to simple cross tabulations [
76]. In addition, several of the categories involved in our analysis of patterns were small in number and in many cases the differences observed between the groups were not statistically significant. However, in consideration of the fact that inferential statistics and therefore p-values are dependent on both variance and sample size we are conscious of the fact that that large p-values do not necessarily mean that there is no difference between the groups in our study. Patterns however need to be verified in other populations and our findings may inform and be built upon in further research using qualitative or quantitative methods.
Conclusions
Our findings indicate that patterns exist in relation to the falls circumstances and health characteristics of people with hip fracture. These patterns, when verified, can provide useful information as to the ways in which fall prevention strategies can be tailored to individuals of varying levels of health and function, who are at risk for falls and hip fracture. Examples of such tailored fall preventive efforts could include: the focus on task-specific training during transfers among frail elderly people, medication review and fall risk education among elderly with a history of falls despite moderate physical function and efforts to enable continued outdoor mobility among less vigorous older people whose falls occur outdoors.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BL: Study design, data collection and analysis, wrote the first draft of the manuscript. ACÅ: Supervised study design, data collection, analysis of the qualitative and quantitative data and interpretation of the results, critical revision of manuscript drafts. LB: Supervised study design, analysis of the quantitative data and interpretation of the results and critical revision of manuscript drafts. HM: Study design, interpretation of results, and critical revision of manuscript drafts. KM: Study design, supervision of data collection, interpretation of the results and critical revision of manuscript drafts. All authors read and approved the final manuscript.