Risk of Falls among Patients who Underwent Surgery for Proximal Femur Fracture One Year after Surgery

Abstract

Introduction Falls in the elderly population are a public health problem, becoming more relevant after a proximal femur fracture (PFF). The “timed up and go” (TUG) test has been linked to the risk of falls and is used in various geriatric societies.

Objective To evaluate the risk of falls in patients with PFF after one year of surgery.

Materials and Methods An observational and cross-sectional study was carried out on patients operated by PFF between January 2017 and May 2020. Patients aged 60 to 85 years, with one year or more since their surgery, were included. Individuals with severe and progressive neurological diseases, BMI > 40, advanced cognitive impairment, and with diseases or mechanical factors that could hinder proper evaluation or rehabilitation were excluded. Patients were summoned and underwent clinical, nutritional and radiological evaluations. Subsequently, the quadriceps strength of the lower extremities was measured and the vertical acceleration and TUG tests were applied. The results were analyzed using statistical models, hypothesis tests and machine learning models.

Results In the machine learning models, the variables of greatest importance for classifying patients as low or medium-high risk of falls were the quadriceps strength on the operated side, quadriceps strength on the contralateral side, and vertical acceleration. On the other hand, in linear regression models, BMI and vertical acceleration had a significant positive and negative effect, respectively.

Conclusion Quadriceps extensor strength, vertical acceleration and BMI are related to the risk of falls. This work shows us accessible and low-cost elements to identify these risk factors and focus the resources for their prevention.

Keyword

risk of falls - proximal femur fracture - quadriceps extensor strength - timed up and go

Introduction

Falls occur in 40% of deaths due to trauma in the geriatric population. In this group, one in three people falls at least once a year, making it a public health problem. The cause is multifactorial and can be attributed to external, neurological or mechanical factors. Furthermore, the loss of muscle mass in the elderly can reach up to 0.64% per year, which is accelerated by prolonged periods of rest.[1] This scenario is one to which we are frequently exposed in our practice when following up patients operated on for proximal femur fracture (FFP).

Suffering from a FFP is associated with a 4 times greater risk of suffering a second hip fracture, which in turn has a worse prognosis than the first.[2] On the other hand, studies have shown that only 40% of patients who suffer from FFP manage to recover their previous functionality after one year. This may be due to a deficit and asymmetry in the extensor power of the lower extremities. It has been seen that women with FFP present an accelerated reduction in the fast-twitch fibers of the quadriceps and that one year after surgery they have up to 70% less extensor explosive power compared to the contralateral limb.[3] It seems that this type of contraction is necessary to achieve stability after an erroneous movement or an imbalance of the body, since both extensor weakness and strength asymmetry have been shown to be an independent risk factor for suffering recurrent falls.[4] [5]

The “timed up and go” test (TUG), published in 1991, is a tool developed to measure mobility and frailty in the geriatric population.[6] In various studies it has been related to the risk of falls.[7] It consists of measuring the time it takes for a patient to get up from a chair, walk three meters, turn around, return and sit down again. It is a simple, low-cost test with excellent intra- and interobserver variability and is used in geriatric societies as a screening method to identify patients at risk of falls.[8]

The objective of this work is to evaluate the risk of falls in patients with FFP operated on one year after surgery. To do this, the TUG time will be used as a measurement directly related to this risk and its association with different variables such as quadriceps strength and vertical acceleration.

Materials and methods

An observational and cross-sectional study was carried out by reviewing records of patients operated on in our center for FFP between January 2017 and May 2020. Patients between 60 and 85 years old, with a year or more of evolution since surgery, were included. Patients with severe and progressive neurological diseases, comorbidities that contraindicate rehabilitation (ASA ≥ 3), BMI > 40, advanced cognitive impairment and mechanical factors that prevent adequate evaluation of extensor power were excluded. Selected patients were scheduled for clinical, nutritional and radiological evaluation. The following measurements were made: a) Bilateral quadriceps strength (Kg/F): with the patient sitting at 90°, he was asked to perform maximum knee extension force, recording the maximum value obtained on a dynamometer in 3 attempts; b) Vertical acceleration test (m/s): an accelerometer was set at the level of the L2 vertebra and the patient was asked to stand up from a sitting position, recording the maximum value in 3 attempts and; c) TUG, using the method explained previously ([Fig. 1]). As a cut-off point, TUG ≤ 20 seconds was determined as low risk of falls and TUG > 20 seconds as medium-high risk of falls, based on the results found in the literature.[6] [9] [10] [11]

Fig. 1 A) Quadriceps strength measurement; B) Vertical acceleration test; C) “Timed up and go” test.

From the clinical record, age, implant used in the surgery and date of the intervention were obtained, emphasizing whether the procedure was performed during the pandemic period. Finally, fractures were classified based on preoperative radiographs by a hip surgeon.

The descriptive analysis of the results was carried out using a) hypothesis test: Welch's t to evaluate the difference between two samples; b) statistical model: b.1) LOESS linear regression for continuous variables, calculating 95% confidence intervals and p values using the Wald approximation, b.2) Assumption analysis for ANOVA (Shapiro-Wilk, Krukal-Wallis) and comparison of groups by Tukey's method for detailed analysis of the difference in strength asymmetry and; c) Machine learning models: Clustering model (K-means) and decision tree model (Classification and Regression Trees CART) to create groups with similar characteristics (clusters). The most important variables for creating the clusters were obtained and compared using ANOVA and Tukey's method. The group with the highest risk of falls was determined using Odd ratio.

Results

Of a total of 1093 patients, after filtering out the deceased and those who did not agree to participate, 55 patients met the inclusion criteria, 39 women (70.9%) and 16 men (29.1%). The average age was 75 years (range 66-84) and the average BMI was 27.6 (range 20.4-36.5). The most frequent fractures were lateral hip fractures (62.5%), followed by medial (36.4%), subtrochanteric (7.3%) and basicervical (3.6%). This relationship remained approximately similar when breaking down the patients by sex in the case of women and was reversed in men, where medial fractures were more prevalent. The most frequent classifications found were the femoral neck Garden 3 (20.0%) and the pertrochanteric AO A2.2 (14.5%) ([Table 1]). The quadriceps force on the operated side was lower than the contralateral one (mean 14.1 Kg/F v/s 17.4 Kg/F) and the TUG had a wide range (10.6-92.0 seconds), with values grouped mainly under 30 seconds.

Table 1

FRACTURE	N	PERCENTAGE
Basicervical	2	3.60%
Lateral	29	52.70%
Medial	20	36.40%
Subtrochanteric	4	7.30%
Total	55	100%

CLASIFICATION	N	PERCENTAGE
Fx Neck Garden 1	1	1.80%
Fx Neck Garden 2	3	5.50%
Fx Neck Garden 3	11	20.00%
Fx Neck Garden 4	5	9.10%
Fx Neck stress type B	1	5.50%
Fx Pertrochanteric A1.1	5	1.80%
Fx Pertrochanteric A1.2	5	9.10%
Fx Pertrochanteric A2.1	3	5.50%
Fx Pertrochanteric A2.2	8	14.50%
Fx Pertrochanteric A2.3	5	9.10%
Fx Pertrochanteric A3.1	1	1.80%
Fx Pertrochanteric A3.3	5	9.10%
Fx Subtroc. Seinsheimer III	1	1.80%
Fx Subtroc. Seinsheimer IV	1	1.80%

FRACTURE/SEX	WOMEN	MEN
Basicervical	1 (2.56%)	1 (6.2%)
Lateral	22 (56.4%)	7 (43.7%)
Medial	12 (30.7%)	8 (50%)
Subtrochanteric	4 (10.2%)	0 (0%)
Total	39 (100%)	16 (100%)

Vertical acceleration had a significant negative relationship with TUG (p = 0.003) and BMI at values ≥ 30 resulted in a statistically significant positive relationship (p = 0.013) using the linear regression model. ([Fig. 2])

Fig. 2 Linear regression model between vertical acceleration and BMI versus TUG. Vertical acceleration resulted with a significant negative relationship and BMI with a significant positive relationship.

The variables age (p = 0.168), surgery during the pandemic (p = 0.445), implant used (p = 0.112) and fracture classification (p = 0.322) did not have a significant relationship.

Analyzes of variance (ANOVA) between TUG time groups showed no significant differences in limb strength asymmetry (Shapiro-Wilk: p = 1.015e-06, Kruskal-Wallis p = 0.196), nor when segmenting results into intervals of 5 Kg/F (Tukey p = 0.176 - 0.999). 10 of the 55 patients in the study (18.2%) presented negative asymmetry, that is, greater strength in the operated leg than in the contralateral leg.

Using machine learning models (K-means and Classification and Regression Trees CART), it was possible to differentiate 3 different groups or clusters among them, with the most important variables for classifying patients according to low or medium-high risk: a) vertical acceleration ; b) quadriceps strength on the operated side and; c) quadriceps strength on the contralateral side. These variables resulted in a very high negative correlation. The cluster with the highest risk of falls was 1 (Odd ratio analysis cluster 2: RR 0.18; cluster 3: RR 0.08), with a TUG greater and different from the others (Tukey p = 0.012). This group corresponded to women with a slightly higher average age (>76 years), with shorter time between fracture and date of surgery and with a slight tendency to undergo surgery during a pandemic ([Fig. 3] and [Table 2]). When performing ANOVA between clusters, no significant differences in quadriceps strength asymmetry were found (Shapiro-Wilk: p = 0.205).

Table 2

Cluster	Age	Day surgery - measurement	Day surgery – fractures	Surgery in pandemic	BMI	OP quadriceps strength (Kg/F)	Quadriceps strength CL (Kg/F)	Vertical acceleration	TUG Time	Dif. quadriceps strength	Falling risk	Male sex	Female sex	Medial fracture	Lateral fracture	THP	CMN
1	76.36	722.86	5.5	0.64	27.88	9.89	13.87	1.78	26.51	3.98	0.71	0	1	0	0.79	0	0.5
2	74.31	930.25	7.69	0.44	26.53	14.54	18.98	2.25	19.27	4.43	0.31	0.44	0.56	1	0	1	0
3	74.64	1330.14	20.21	0.07	27.78	16.83	22.49	3.52	14.11	5.66	0.07	0.43	0.57	0	0.79	0	0.71

Fig. 3 Clustering model with the K-means technique. Elements that share similar characteristics are grouped together in the same cluster, separated from other clusters that do not share those characteristics.

According to the decision tree model, the quadriceps strength variable on the operated side is the one that allowed the greatest segmentation and classification of patients, based on their risk of falls ([Fig. 4]).

Fig. 4 The quadriceps strength variable on the operated side turned out to be the one with the greatest capacity to classify and segment patients into clusters, with high accuracy and sensitivity.

Discussion

Our work evaluated the influence of some easily obtained variables on the risk of falls. For this we use the TUG, given its association found in literature.

Vertical acceleration showed a significant negative relationship with the risk of falls, which is expected, since it reflects the extensor capacity and explosive power of the lower extremities, a factor that has been seen to affect mobility and the ability to maintain balance in older adults.[4] BMI, for its part, was positively related to the risk of falls, especially in values greater than 30, a result that makes sense if we consider that these patients must balance relatively more mass than the rest.

Contrary to what we expected, the quadriceps strength asymmetry of the lower extremities, measured with a dynamometer, was not relevant to explain the greater risk of falls in our patients and did not coincide with what was shown in different publications.[4] [12] [13] [14] It is important to note that in these works the asymmetry of the explosive extensor power (LEP: leg extensor power) was evaluated through a potentiometer, a tool that we do not have at the time of making our measurements. According to authors such as Bassey et al.,[5] the LEP of the weaker limb could be the best predictor of speed and body balance when walking on challenging terrain and its magnitude would determine the ability to react to avoid a fall. In addition to this, Skelton et al.,[4] showed a greater asymmetry of the LEP between fallers and non-fallers. This difference was greater than that found when analyzing only the difference in quadriceps extensor strength.

It is striking that in 10 patients we found a negative asymmetry, that is, greater strength on the operated side, which could be due to deterioration of the other limb prior to the fracture, disuse of the non-operated leg or vigorous rehabilitation of the operated limb. Skelton et al.[4] found that the unfractured limb, on average, had a lower LEP than the healthy leg of elderly women of the same age. Within the explanations it is proposed that the subjects studied by these authors were an “elite” selection, which does not represent the common population. It is also suggested that these patients had a lower LEP prior to the fracture, suggesting that the deterioration of the fast-twitch fibers of the quadriceps is an important factor in the incidence of FFP. Thirdly, they consider that the process of trauma, surgery and hospitalization also affects the healthy leg. It has been shown that short periods of bed rest, even in young people, can compromise muscle mass and function.[15] [16] In our study, it must be considered that many patients were excluded, either due to death or because they were not able to follow or comply with the indications in the measurement tests. In addition, there are those who did not want to attend the evaluations for various reasons, therefore, individuals were evaluated in the best conditions, which could have influenced our results.

Despite not being able to measure LEP asymmetry, we found other variables that allowed us to classify patients. Thus, after analysis with machine learning models, we were able to identify three variables capable of grouping and segmenting patients based on the risk of falls: quadriceps strength on the operated side, contralateral quadriceps strength, and vertical acceleration. This is similar to those found in publications, where patients with hip fracture with less quadriceps strength and less LEP have lower walking speed.[17] [18] For the purposes of our work, lower walking speed will translate into higher TUG and therefore higher risk of falls.

The riskiest cluster (cluster 1) was made up of women, with a higher average age (76 years), lower quadriceps strength on both the operated and contralateral sides, and lower vertical acceleration. We believe that these patients, added to those with a BMI greater than 30, should be the main objective of rehabilitation to restore balance and extensor strength, placing greater emphasis on the operated side since this variable was the one that segmented the groups to a greater extent according to the result of the decision tree model.

In this group, a tendency to undergo surgery during the pandemic was also observed, but the relationship is weak and does not allow us to determine the association. On the other hand, they presented with a shorter period between fracture and surgery. We believe that there is a confounding variable that could not be isolated and that must be clarified later, since logic indicates that shorter rest times and the opportunity to start therapy early should lead to favorable results.

Conclusions

Patients with FFP mostly do not recover their original functional capacity. A fall can have disastrous consequences, with a second fracture having a worse prognosis than the first.

In our work we found that quadriceps extensor strength on the operated side, quadriceps extensor strength on the contralateral side, vertical acceleration and BMI are related to the risk of falls in patients operated on by FFP. These risk factors can be obtained easily and at low cost.

Considering the new variables identified, we can suggest using time for rehabilitation in a targeted manner, focusing resources, while a prospective study is prepared to provide us with a clear protocol with management guidelines for these patients.

Conflict of Interest

None.

• Bibliografía

• 1 Rubenstein LZ. Falls in older people: epidemiology, risk factors and strategies for prevention. Age Ageing 2006; 35 (02, Suppl 2): ii37-ii41
• 2 Klotzbuecher CM, Ross PD, Landsman PB, Abbott III TA, Berger M. Patients with prior fractures have an increased risk of future fractures: a summary of the literature and statistical synthesis. J Bone Miner Res 2000; 15 (04) 721-739
• 3 Levy DI, Young A, Skelton DA, Yeo AI. Strength, power and functional ability. In Passeri M, ed. Geriatrics 94 International Association of Gerontology (I.A.G.) European Region Clinical Section Congress. Rome. Italy: CIC Edizioni, Internazionali, 1994: 85-93
• 4 Skelton DA, Kennedy J, Rutherford OM. Explosive power and asymmetry in leg muscle function in frequent fallers and non-fallers aged over 65. Age Ageing 2002; 31 (02) 119-125
• 5 Bassey EJ, Fiatarone MA, O'Neill EF, Kelly M, Evans WJ, Lipsitz LA. Leg extensor power and functional performance in very old men and women. Clin Sci (Lond) 1992; 82 (03) 321-327
• 6 Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc 1991; 39 (02) 142-148
• 7 Ansai JH, Andrade LP, Nakagawa TH, Rebelatto JR. Performances on the Timed Up and Go Test and subtasks between fallers and non-fallers in older adults with cognitive impairment. Arq Neuropsiquiatr 2018; 76 (06) 381-386
• 8 Zhou X, Feng X. Advances in the study of fall risk assessment tools. J Nurs (Luton) 2018; 33 (21) 109-112
• 9 Rydwik E, Bergland A, Forséén L, Fräändin K. Psychometric properties of Timed Up and Go in elderly people: a systematic review. Phys Occup Ther Geriatr 2011; 29 (02) 102-125
• 10 Killough J. Validation of the Timed Up and Go Test to predict falls. J Geriatr Phys Ther 2006; 29 (03) 128-129
• 11 Shumway-Cook A, Brauer S, Woollacott M. Predicting the probability for falls in community-dwelling older adults using the Timed Up & Go Test. Phys Ther 2000; 80 (09) 896-903
• 12 Portegijs E, Sipilä S, Alen M. et al. Leg extension power asymmetry and mobility limitation in healthy older women. Arch Phys Med Rehabil 2005; 86 (09) 1838-1842
• 13 Portegijs E, Sipilä S, Pajala S. et al. Asymmetrical lower extremity power deficit as a risk factor for injurious falls in healthy older women. J Am Geriatr Soc 2006; 54 (03) 551-553
• 14 Sylliaas H, Brovold T, Wyller TB, Bergland A. Progressive strength training in older patients after hip fracture: a randomised controlled trial. Age Ageing 2011; 40 (02) 221-227
• 15 Proceedings of the 1st International Symposium on Inactivity and Health: Effects of bed rest on health. Ada Physiol Scand Suppl 1994 616. 1-114
• 16 Lamb SE, Morse RE, Evans JG. Mobility after proximal femoral fracture: the relevance of leg extensor power, postural sway and other factors. Age Ageing 1995; 24 (04) 308-314
• 17 McGrath R, Blackwell TL, Ensrud KE, Vincent BM, Cawthon PM. The Associations of Handgrip Strength and Leg Extension Power Asymmetry on Incident Recurrent Falls and Fractures in Older Men. J Gerontol A Biol Sci Med Sci 2021; 76 (09) e221-e227
• 18 Madsen OR, Lauridsen UB, Sørensen OH. Quadriceps strength in women with a previous hip fracture: relationships to physical ability and bone mass. Scand J Rehabil Med 2000; 32 (01) 37-40

Publication History

Received: 12 April 2023

Accepted: 14 August 2023

Article published online:
30 October 2023