Recent fall and high imminent risk of fracture in older men and women

Abstract

Background

despite fall history being a well-known risk factor for falls and fractures, the association between very recent falls and imminent fracture risk is not clearly elucidated.

Objective

to study the very recent (<4 months) fall-related absolute risk of fractures in the following year.

Methods

two large prospective cohort studies of women (Study of Osteoporotic Fractures [SOF]) and men (Osteoporotic Fractures in Men Study [MrOS]) aged 65 years or older were included. Data on falls were collected every 4 months, and the primary outcomes were any non-spine and hip fractures in the next 12 months.

Results

a total of 9,704 women contributed 419,149, and 5,994 men contributed 223,885 four-monthly periods of observations during the 14.8-year SOF and 12.6-year MrOS follow-up. Falls within 4 months indicated a high risk of non-spine and hip fractures in the following year for both sexes; in women, a recent fall indicated an 8.1% absolute risk of a non-spine fracture within 1 year, a 2.5-fold higher risk than that in women without falls, a 2.5% absolute risk of hip fracture, and a 3.1-fold increased risk. Falls increased the risk of fractures regardless of whether a fracture occurred or not. Men had similar risk patterns, albeit with a lower absolute risk of fracture.

Conclusions

in older people, a fall within 4 months indicates a high risk of fracture in the next year, regardless of fracture occurrence. A recent fall warrants urgent evaluation and consideration of treatments to reduce the imminent risk of fractures.

Key Points

• A very recent fall that occurs within 4 months predicts an imminent risk of fracture in the following year.

• Recent falls within 4 months confer a similar increased risk as that of a recent fracture for fractures in the following year.

• Older patients with a recent fall should be assessed and treated similarly as that of a patient with a recent fracture.

• A recent fracture indicates a high imminent risk of fracture that warrants urgent evaluation and prevention.

Introduction

Fractures are a common cause of chronic pain, disability, loss of independence and mortality, and confers a substantial economic burden [1–3]. A recent fracture indicates a high fracture risk in the following year that decreases with time [4, 5], and current guidelines recommend urgent evaluation and consideration of pharmacologic treatment of patients with a recent fracture [6]. However, recent falls are not yet an indication of similar urgent assessment and treatment to reduce fracture risk [7–9].

Falling is one of the most common events in older people, occurring in nearly one-third of people aged >65 years [10]. A history of falls implies a higher risk of another imminent fall [11], and particularly one within 12 months indicates an increased risk of future fractures [12–14]. Current guidelines for fall prevention recommend the annual evaluation of older patients for the incidence of falls and initiation of appropriate fall-prevention measures based on additional risk factors [9]. Although the recommendations to prevent falls and their consequences include an evaluation of osteoporosis, no pharmacologic intervention for osteoporosis and fractures is indicated [15, 16]. As falls are a risk factor for future fractures, several fracture risk-prediction tools consider a history of falls as a risk factor. However, even tools that include fall history do not specify when the fall occurred or specify a fall within 1 year [17, 18].

The risk of falls can be reduced by exercise programs, therapies to improve balance, and modification of risk factors for falls [19–21]. In addition, some treatments for osteoporosis can quickly improve bone strength and reduce the fracture risk within 1 year [22]. Therefore, if a very recent fall indicates a high fracture risk in the following year, this could be a sufficient, independent, sole indication for treatment of osteoporosis, and patients with recent falls may benefit from urgent evaluation and interventions to reduce the risk of imminent fracture.

We hypothesised that a recent fall within 4 months indicates a high risk of fracture in the following year and used data from two large cohorts of older men and women to test this hypothesis. Short-term risk of hip and non-vertebral fractures associated with falls within the last 4 months was compared with fracture risk associated with falls within 12 months.

Methods

Study design and participants

Data from the Study of Osteoporotic Fractures (SOF) and Osteoporotic Fractures in Men Study (MrOS) were used for this analysis. The SOF included 9,704 community-dwelling Caucasian women older than 65 years, who were ambulatory, and had no prior bilateral hip replacement from 1986 to 1988 at four clinical sites in USA (Baltimore, MD; Minneapolis, MN; Monongahela Valley, Pittsburgh, PA and Portland, OR) [23]. In-person examinations were conducted every 2–4 years with a mean follow-up of 14.8 years. Community-dwelling 5,994 men aged ≥65 years without hip arthroplasty who were able to walk without assistance were included in the MrOS and underwent baseline evaluations from 2000 to 2002 at six study sites (Birmingham, AL; Minneapolis, MN; Monongahela Valley, Pittsburgh, PA; Palo Alto, CA; Portland, OR and San Diego, CA) [24], with four in-person examinations and a mean follow-up of 12.6 years. Further information on SOF and MrOS can be found at https://sof.ucsf.edu and https://mrosonline.ucsf.edu, respectively. Participants from both cohorts were followed from the baseline visit to either date of death/termination or the last contact (December 2009 for SOF and August 2019 for MrOS), whichever occurred first. The institutional review board at each site approved the study protocol, and written informed consent was obtained from all participants.

Falls and fractures

In both SOF and MrOS, participants were queried every 4 months by postcard or telephone whether they had a fall or fracture within the last 4 months, and the number of falls was recorded. All self-reported fractures were centrally adjudicated by a physician’s review of the medical records and/or radiologic reports.

Clinical assessments

Data on height and weight were collected at every clinic visit, and body mass index (BMI) was calculated. The total hip bone mineral density (BMD) was measured using dual-energy X-ray absorptiometry on a Hologic QDR 1000 or 2000 densitometer for SOF and on a QDR 4500-W densitometer for MrOS (Hologic, Waltham, MA, USA) at five clinic visits in SOF and four clinic visits in MrOS.

Statistical analyses

History of falls was classified by whether the participant fell in the preceding 4 months or 1 year (three postcards or calls). Falls within 4 months or 1 year were further divided into two subgroups by the presence of fractures: falls but no fractures, or falls resulting in fractures. All adjudicated fractures were included as outcomes and prevalent fracture covariates in the analysis. Risks of non-spine and hip fractures in the following year and in the second and third years were estimated for participants stratified by a history of falls within 4 months or 1 year. Associations between fall history and the risk of future fracture was assessed in a pooled logistic regression model with multiple records. The multivariate model for fractures included a fall history and any adjudicated fracture in the preceding 4 months, age, BMI and total hip BMD to evaluate the independent predictive values. Using a repeated-measures design, time-dependent covariates (including BMI and total hip BMD) used in the multivariable models were ascertained from the most recent visit close to the date of postcard collection, except that the baseline BMD value for SOF was obtained during the second clinic visit. The age in each postcard observation was calculated from the baseline age and the duration from baseline to postcard collection. The absolute fracture risks (%) by fall history and adjudicated fractures in the preceding 1 year were estimated by dividing the number of future fractures in a group by the total number of intragroup observations. Data from the SOF and MrOS were analysed separately using Stata/MP version 16 (StataCorp, College Station, TX, USA); a two-sided P < 0.05 was considered significant.

Results

A total of 9,704 women from the SOF and 5,994 men from the MrOS were included in the analyses. The baseline mean age of the female and male participants was 71.9 and 73.7 years, respectively (Table 1). Approximately 90% of the survivors in the SOF and MrOS were followed up for 14.8 and 12.6 years, respectively. The female participants underwent 419,149 four-monthly observations, while the male participants underwent 223,885 observations.

Table 1.

Baseline clinical characteristics of the participants of the SOF and MrOS

	SOF (N = 9,704)	MrOS (N = 5,994)
Age, years	71.9 (5.3)	73.7 (5.9)
BMI, kg/m2	26.4 (4.6)	27.4 (3.8)
Average follow-up time, years	14.8 (6.1)	12.6 (5.3)
Total hip BMD, g/cm2	0.75 (0.13)	0.96 (0.14)
Total hip T-score, at baseline	1.0 ± 0.1	0.7 ± 0.1
Non-spine fracture, N (%)	4,257 (43.9%)	1,280 (21.4%)
Hip fracture, N (%)	1,444 (14.8%)	340 (5.7%)

Data are shown as mean (SD) or N (%). BMI, body mass index; BMD, bone mineral density.

At the 4-months interval observations, approximately 10% of the participants reported falls within the last 4-month period, resulting in a total of 64,038 incident falls in women and 38,047 falls in men during the follow-up time (Table 2). In women, 4,257 (43.9%) had non-spine fractures, and 1,444 (14.8%) had at least one hip fracture. In men, 1,280 (21.4%) had non-spine fractures, and 340 (5.7%) had hip fractures.

Table 2.

Multivariate analyses of the risks of non-spine fractures and hip fractures in the following year

	(%)	Non-spine fractures			Hip fractures
		OR	95% CI	P-value	OR	95% CI	P-value
SOF
In the previous 4 months
No fall and no fracture	53.5%	1.0	Reference		1.0	Reference
At least one fall, but no fracture	9.2%	2.4	(2.3–2.5)	<0.001	2.8	(2.6–3.0)	<0.001
At least one fall and fracture	0.5%	2.1	(1.9–2.4)	<0.001	2.0	(1.6–2.7)	<0.001
At least one fracture, but no fall	0.4%	1.9	(1.6–2.3)	<0.001	1.5	(1.1–2.1)	<0.001
Age (per 10-year increase)		1.1	(1.1–1.2)	<0.001	2.2	(2.0–2.3)	<0.001
BMI (per SD decrease)		1.0	(0.9–1.0)	0.1	1.1	(1.0–1.2)	<0.001
Total hip BMD (per SD decrease)		1.5	(1.4–1.5)	<0.001	1.9	(1.8–2.0)	<0.001
MrOS
In the previous 4 months
No fall and no fracture	54.3%	1.0	Reference		1.0	Reference
At least one fall, but no fracture	10.8%	2.2	(2.0–2.3)	<0.001	2.3	(2.0–2.7)	<0.001
At least one fall and fracture	0.3%	2.5	(1.9–3.3)	<0.001	2.2	(1.3–3.8)	<0.001
At least one fracture, but no fall	0.1%	2.0	(1.2–3.1)	<0.001	1.9	(0.8–4.5)	0.2
Age (per 10-year increase)		1.4	(1.3–1.0)	<0.001	2.2	(1.9–2.6)	<0.001
BMI (per SD decrease)		0.9	(0.8–1.0)	<0.001	0.8	(0.7–0.9)	<0.001
Total hip BMD (per SD decrease)		1.7	(1.6–1.8)	<0.001	2.8	(2.4–3.1)	<0.001

BMI, body mass index; SD, standard deviation; BMD, bone mineral density; OR, odds ratio.

A fall within 4 months was associated with a substantially increased risk of fracture in the following year, and the risk was much greater in participants who experienced falls within 4 months than in those with falls within 1 year (Figure 1). The increased risks of non-spine and hip fractures after falls diminished in the second and third years after the fall, but remained higher than those without any incidence of falls (Figure 1). In addition, a history of two or more falls indicated a further increased risk of non-spine and hip fractures (Figure 2).

Figure 1.

The possibility of hip and non-spine fracture in the following 3 years based on the history of falls.

Figure 2.

Odds ratios of fall and recurrent falls in the preceding year for the risk of fractures in the next 3 years.

Notably, the incidence of falls without fractures was commoner than that of fractures during the 4-month period (9.2% in women and 10.8% in men; 0.9% and 0.4%, respectively; Table 2). In women, a recent fall but no fracture in the same period indicated an 8.1% (95% confidence interval [CI] 7.9–8.3%) and 2.5% (95% CI 2.4–2.7%) absolute risk of non-spine and hip fractures, which was 2.4-fold and 2.8-fold greater than that in women without a fall or a fracture, respectively (Table 2 and Figure 3). This pattern of increased risk of fracture within 1 year was similar in men, although the absolute risks (3.8% for non-spine and 1.0% for hip fracture) were lower than those in women (Figure 3).

Figure 3.

Absolute risk (%) of non-spine and hip fractures based on the history of falls in the preceding 4 months or 1 year.

The risk of non-spine and hip fractures for falls in the previous 5–12 months were 4.3% and 1.2% respectively for women and 2.5% and 0.7% for men, substantially lower than for falls within 4 months (data not shown). The sensitivity of falls within 4 months to fractures in the following year exceeded 0.8, which was higher than that of falls within 12 months of 0.66, although the specificity was lower than that of falls at 12 months (Supplementary Table 1 is available in Age and Ageing online).

Discussion

The results of this study show that a recent fall indicates a high imminent (<12-month) risk of fracture; the results also challenge the clinical guidelines that only recommend evaluation of falls within the past year, do not indicate treatment to reduce the fracture risk, and do not emphasise urgency for implementing measures to prevent imminent fractures. A recent fracture indicates a high imminent risk of fracture that warrants urgent evaluation and prevention [25]. A recent fall carries the same high imminent risk as a recent fracture. Therefore, an older adult with a recent fall should undergo a similar urgent evaluation and treatment to prevent fractures in the next 12 months and beyond. Importantly, falling without a fracture is a commoner event than fracture in older adults. Thus, interventions to reduce fractures in patients with recent falls would reduce the risk of fractures in a population than merely targeting patients with recent fractures.

Several algorithms, including FRAX, QFracture and Garvan, are used to identify individuals with a higher fracture risk [26]. Garvan considers history of falls but without any specific time frame and QFracture includes fall history in the previous year [15, 18]. FRAX does not yet include a history of falls. Most of the previous studies that the guidelines are based on used data that were retrospectively collected, typically on falls in the previous year [17, 27]. In this respect, the present study is unique because SOF and MrOS collected fall history every 4 months. The ORs of falls for the future risk of fractures in the present study were greater than that observed in the previous studies that evaluated the fall history in the previous year or without of specific time frame [28–30]. Falls within 4 months carry a substantially greater risk of fracture than to falls in the previous year, thus our study indicates that the incidence of falls should be screened more often than the currently recommended 12-month period. Since the sensitivity of a history of falls within 4 months was much better in both older men and women, then a history of falls within 4 months would identify more incidents of fracture than a history of falls within 12 months.

The increased risk of fractures in the first year following a recent fall decreased in the second and third years. The reason for this transient increase in fracture risk remains unclear and might reflect the development or worsening of conditions or physical functions that increase the risk of falls. Fear of additional falls might contribute to further impairment in postural balance and functional status immediately after falls that may diminish over time [12, 31].

Existing guidelines for fall assessment and prevention recommend annual screening for the risk of falls and fall incidence in the previous year, with evaluations for predisposing factors in older adults with two or more falls, emergency department visits for a fall, or an abnormal balance or gait problem [9, 15]. However, they do not include recommendations about pharmacotherapy [32]. Our results suggest that the guidelines should be revised to recommend more frequent assessment of falls in older patients. Moreover, patients presenting with a fall or with a fall within the last 4 months should undergo a similar evaluation and receive similar interventions to prevent falls and consideration of pharmacotherapy to prevent fractures as those who had a recent fracture.

Lower BMD was a strong risk factor for imminent fractures following a fall, which emphasises the importance of bone strength as a determinant of fractures in patients at risk of falls. For a patient who has a recent fall, measuring the BMD may be invaluable for deciding whether the fracture risk warrants pharmacotherapy to prevent fractures.

The analysis indicates that in an older woman with a recent fall, no additional risk is conferred by a history of fracture. However, in men, the report of a fracture with a fall added to the risk of imminent fracture. Men have a lower absolute risk of fracture than women. Therefore, a recent fall in an older man may prompt the measurement of BMD and other risk factors of fracture to determine whether pharmacologic treatment should be administered to reduce the imminent risk of fractures.

This study has several strengths. Information on thousands of collection periods, falls, and fractures from two large cohorts provide a high degree of certainty with very narrow CIs around the estimates of risk. The patterns of risk were very similar in both large cohorts of women and men from several communities in the United States. This validates the results in each cohort and indicates that the findings are generalizable to both women and men. Unlike most of the other studies that collect the history of falls without any specific time fracture or in the previous 1 year, a history of falls in this study was collected every 4 months, and over 90% of the participants had a complete follow-up. Furthermore, all fractures were confirmed by central adjudication of medical records or X-ray reports. However, there were several limitations in this study. Men and women were older than 65 years, and the vast majority of participants were Caucasians; thus, generalization to younger and non-white populations may be limited. In addition, the history of falls was based on the participants’ recall, which has limited accuracy; a more accurate method should be used or a clinical history should be obtained to strengthen the associations between recent falls and the risk of fracture [33]. A very small number of participants in this study reported fractures, but no falls during the same period. These might represent limited recall or inaccurate reporting of the fall. A few might have had pathologic fractures due to malignancy. However, our cohort studies did not include that degree of clinical detail, thus we could not further characterise those fractures. Data on the history of falls collected in these cohort studies did not include consequences or circumstances of falls, which might improve the prediction of future fractures. Furthermore, the study aimed to evaluate the risk of fractures related to recent falls and to compare it with that of recent fractures. Thus, data on the individual types of fractures were not separated from those of previous fractures. Moreover, the models were not adjusted for other covariates. However, the purpose of the analysis was the prediction of fracture risk related to the recent falls, not to test whether falls cause fractures. Therefore, we do not consider adjustments for potential confounders in the model.

In conclusion, patients who have a recent fall within 4 months have an increased risk, similar to that following a recent fracture, of fractures in the following year. Older patients with a recent fall should be assessed and treated similarly as patients with a recent fracture. As the occurrence of recent falls is commoner than that of fractures, targeting falls may potentially prevent other incidence of fractures.

Declaration of Conflicts of Interest

Dr Kim and Dr Cummings receive investigator-initiated research funding from Amgen. Dr Cummings has been a consultant to Amgen.

Declaration of Sources of Funding

The Study of Osteoporotic Fractures (SOF) is supported by National Institutes of Health funding. The National Institute on Aging (NIA) provides support under the following grant numbers: R01 AG005407, R01 AR35582, R01 AR35583, R01 AR35584, R01 AG005394, R01 AG027574 and R01 AG027576. The Osteoporotic Fractures in Men (MrOS) Study is supported by the National Institute on Aging, the National Institute of Arthritis and Musculoskeletal and Skin Diseases, the National Center for Advancing Translational Sciences, and the National Institutes of Health (NIH) Roadmap for Medical Research (AG027810, AG042124, AG042139, AG042140, AG042143, AG042145, AG042168, AR066160 and TR000128).