A Clinical Measure of Trunk Neuromuscular Function Predicts Falling in Older Adults with Chronic Low Back Pain

Abstract

Background and Purpose:

Older adults with low back pain (LBP) are at risk for falling, but condition-specific mechanisms are unknown. Trunk neuromuscular function is critical for maintaining balance during mobility tasks, and is often impaired in older adults with LBP. The purpose of this study was to assess if aberrant lumbopelvic movements (or aberrant movements), a clinical index of trunk neuromuscular function, were associated with increased fall risk among older adults with chronic LBP over a 12-month follow-up period.

Methods:

This study analyzed data from a prospective cohort study of 250 community-dwelling older adults with chronic LBP. Participants were screened for four aberrant movements during three trials of forward flexion from a standing position: instability catch, painful arc, altered lumbopelvic rhythm, and Gower’s sign. Aberrant movements were totaled to yield a summary score (i.e., 0–4). Prospective falls were monitored via monthly fall calendars for 12 months. A generalized linear model with Poisson distribution and log link function was used to evaluate the association between aberrant movements and prospective fall risk. Age, sex, body mass index, LBP intensity, dynamic balance performance, prior falls, anxiolytic medication usage, and hip osteoarthritis characteristics were included as covariates in the model.

Results:

Baseline aberrant movements were independently associated with greater fall risk (risk ratio = 1.249, 95% CI = 1.047–1.491, p=0.014); each 1-unit increase in aberrant movement score imparted a 24.9% increase in the risk of falling.

Conclusions:

Aberrant movements increased the risk of falling among older adults with chronic LBP over a 1-year span.

INTRODUCTION

Low back pain (LBP) is more responsible for years lived with disability than any other health condition.¹ LBP prevalence increases with age,¹ and older adults with LBP are particularly susceptible to the burden of disability.^2,3 The older adult population in the US is projected to increase from 48 million in 2015 to 88 million in 2050⁴; aging of the population is likely to compound the current public health crisis posed by LBP.⁵ To mitigate disability and lessen public health burden, it is essential to identify and intervene upon modifiable LBP impairments that predispose older adults to events that expedite health decline (i.e., falling).

Falling is common among older adults, and can lead to adverse health outcomes such as injury, functional decline, hospitalization, institutionalization, and death.⁶ Risk factors for falling in older adults are multifactorial, and may include neuromuscular or cognitive impairments, fear of falling, balance dysfunction, self-efficacy, and/or medication usage.⁷ More recently, literature has established that pain increases the risk of sustaining a fall.⁸ This association has been demonstrated in older adults with chronic pain^9,10 and specifically among older adults with LBP,^11,12 prompting a need to investigate the underpinnings of falls in older adults with chronic LBP.

Biomechanical evidence suggests that trunk neuromuscular function (i.e., voluntary or reactive muscle activity to elicit movement or maintain stability) contributes to postural control in both resting and perturbed standing conditions.^13–16 Aging and LBP appear to negatively influence trunk neuromuscular function, and may lead to impaired postural control during anticipated and reactive perturbations due to decreased limits of stability.^17–19 Since postural control has been linked to prospective fall risk,²⁰ impaired trunk neuromuscular function may predispose older adults with chronic LBP to falling, and may also restrict their ability to recover while falling.

Quantification of trunk neuromuscular function is difficult in the clinical setting; a qualitative alternative is to approximate trunk neuromuscular function by observing lumbopelvic coordination during a forward bending task. Prior research demonstrates that impaired trunk neuromuscular function and altered lumbopelvic coordination occur in parallel during trunk flexion and return from flexion among adults with LBP.^21,22 Clinical research has demonstrated that observable aberrant lumbopelvic movements during forward bending are associated with impaired trunk neuromuscular function (i.e., construct validity)²³ and moderately agreeable with altered lumbopelvic coordination during kinematic analysis (i.e., criterion validity).²⁴ Among adults with LBP, observation of aberrant lumbopelvic movements has previously established fair to excellent interrater reliability.^25,26 Further, aberrant lumbopelvic movements have recently been associated with prospective functional decline among older adults with chronic LBP.²⁷ As a clinical marker of trunk coordination (and by proxy neuromuscular function), the number of aberrant lumbopelvic movements (i.e., an aberrant movement summary score)²⁵ observed at baseline may predict falling in this patient population.

The purpose of this study is to evaluate whether clinically observable aberrant lumbopelvic movements predict fall risk over a 1-year follow-up period in older adults with chronic LBP. We hypothesize that a higher aberrant movement summary score will be associated with increased fall risk in this population, after controlling for well-established risk factors for falls.

METHODS

We performed a secondary analysis of data collected from March 2013 through December 2016 in a prospective cohort study of 250 community-dwelling older adults (i.e., 60–85 years), the details of which have been published elsewhere.²⁸ Older adults with chronic LBP that negatively impacted their daily function²⁹ were recruited via newspaper advertisements, posted fliers, and visits to local senior centers. Chronic LBP was defined as LBP intensity >3/10 interfering with daily life for ≥3 months and ≥4 days per week. Potential participants were excluded if they had visual impairments limiting their ability to fully complete the assessments, mobility impairments requiring more assistance than a single-point cane, medical red flags of non-mechanical LBP (e.g., fever accompanying LBP), progressive neurological disorders (e.g., Parkinson disease), current fractures in their back or hip, pain in the lower limbs that was greater than the pain in the back, inflammatory spinal conditions (e.g., ankylosing spondylitis), or a terminal illness. Additionally, people were excluded if they scored less than 18/30 on the Mini-Mental State Examination, due to concerns regarding their ability to fully participate in the consent process.³⁰

Licensed physical therapists (average experience: 10.6 years) were trained to conduct standardized hip, lumbar, and performance assessments at four time points (baseline and 3-, 6-, & 12-months post-baseline) in a Clinical Research Laboratory at the University of Delaware; measures for this study were collected at each time point. The study was approved by the Institutional Review Board at the University of Delaware and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from each participant. Participants received monetary compensation for partaking in the study.

Participant Characteristics

At baseline, participants reported their age, sex, race, and one-year fall history; height (cm) and weight (kg) were collected to calculate body mass index (BMI) in kg/m². Pain thermometers, a valid and reliable method for capturing pain intensity,³¹ were used at each assessment time point. Participants identified their current LBP intensity (0=no pain; 10=worst pain imaginable), as well as the best and worst LBP intensity in the 24 hours prior to the assessment. The three measures (current, best, and worst) were averaged to create a composite pain intensity.³² Participants brought their medications and supplements to each assessment time point, and completed a medication sheet detailing dosage, route, and frequency of all medications and supplements; the therapist performing the first part of the assessment cross-referenced the medication bottles with the log entries to ensure accuracy. Medications were sub-categorized and coded according to their drug class; those medications falling into the benzodiazepine drug class were coded as ‘anxiolytic medications’, which are an established risk factor for falling.³³

Hip Osteoarthritis (OA) Signs and Symptoms

The American College of Rheumatology (ACR) criteria for the classification of osteoarthritis (OA) of the hip were used to assess hip OA signs and symptoms at baseline.³⁴ Prior to the hip examination, licensed physical therapists assessed the presence of hip pain by asking, “Do you have pain in your groin, buttock, or thigh?”. They also asked, “Do you have morning stiffness in your hip that lasts less than or equal to 60 minutes?”. Hip pain and morning stiffness were coded as dichotomous variables (no=0; yes=1). During the hip examination, therapists measured bilateral hip internal rotation (IR) range of motion (ROM) with an inclinometer with the patient in prone^35,36; scores were given if hip IR ROM ≥15⁰ (0=absent; unilateral=1; bilateral=2). The presence of pain with hip IR was assessed simultaneously with ROM by asking, “does this motion cause any pain in the groin, buttock, or thigh?”; responses were coded in the same manner as ROM (0=absent; unilateral=1; bilateral=2). Criteria were summed to create a burden variable with scores from 0–6, with higher scores indicating higher likelihood of hip OA. Prior work established that higher hip OA symptomatology is associated with increased fall risk.³⁷

Aberrant Lumbopelvic Movements

Participants were asked to stand and perform three trials of forward trunk flexion under the supervision of a licensed physical therapist to screen for four aberrant lumbopelvic movements: instability catch (i.e., deviation from sagittal plane with forward flexion), painful arc of motion (i.e., pain during forward flexion or with return to upright stance from flexion), reversal of lumbopelvic rhythm (i.e., trunk extension prior to hip and pelvis extension during return to upright from forward flexion), and Gower’s sign (i.e., use of upper extremities on the thighs to attain upright stance).³⁸ If the therapist observed an aberrant movement during any of the three trials, that movement was noted as ‘present’ and was scored a ‘1’. If the movement was not observed, it was noted as ‘absent’ and was scored a ‘0’. Each participant was given a summary score, with a minimum of ‘0’ (i.e., no aberrant movements noted) and a maximum of ‘4’ (i.e., all four aberrant movements noted).²⁵

Four-Square Step Test

The Four-Square Step Test (FSST) is a valid and reliable test of dynamic balance and fall risk in the older adult population.^39,40 Participants were asked to step in a specific sequence over a device used to mark out four squares on the floor. After one practice trial, they performed two timed trials. For each trial, participants stepped clockwise through four quadrants, then counterclockwise through the same quadrants; timing started with movement initiation and stopped when participants’ feet contacted the floor in the last quadrant in the series. Trials were marked as invalid if the participant lost their balance, touched the marker device, or altered the sequence while stepping. The better of the two trials was used in the analysis. Prior literature suggests scores >15s can discriminate multiple fallers from non-multiple fallers.³⁹

Falls

The number of falls over the course of the year was collected via falls calendars, the optimal method of capturing this information.^41,42 At each assessment, participants were given calendars to complete monthly, with return envelopes to help facilitate compliance. When distributing the calendars, members of the study staff defined falls for each participant as “any accidental or unintentional event in which contact was made with a lower surface or the ground”.⁴³ Participants were asked to mark each calendar day with an “F” for fall or an “N” for no-fall; they were not asked to keep track of the number of falls per day. A member of the study staff called each participant monthly to remind them to complete and return fall calendars. A tally of falls was kept via review of falls calendars and information gleaned from monthly phone calls; if falls calendars were missing and/or phone calls were unsuccessful, falls data for that participant was excluded. Falls data was also excluded if participants withdrew, were lost to follow-up, or expired prior to completion of the study.

Statistical Analysis

All analyses were completed using IBM SPSS Statistics software version 28 (SPSS, Inc. Armonk, NY), with significance set to P< 0.05. Test-specific data assumptions (e.g., normality) were assessed in parallel with analyses. For descriptive purposes, participants were categorized as non-fallers, single fallers, and multiple fallers (i.e., sustaining ≥2 falls over the course of the year). Chi-square and multivariate analysis of variance tests (or MANOVA, with planned Tukey post-hoc comparisons) were utilized to examine differences in baseline characteristics across levels of aberrant movement summary scores. Bivariate correlation matrices and variance inflation factors were utilized to assess for multicollinearity among regression model variables. The relationship between the number of falls and aberrant movements was performed with Generalized Linear Models adjusting for 8 covariates: age, sex, BMI, composite LBP intensity, number of prior falls, anxiolytic use, number of ACR criteria for hip OA signs and symptoms, and dynamic balance performance (FSST).^{9,11,12,33,37,39,44,45} Both negative binomial and Poisson distributions with log link functions were tested; findings were consistent between models, and the Poisson model was chosen in order to maintain consistency with a prior publication on this cohort.³⁷ The observed power for the analysis is β=0.99 based on sample size, as well as standard deviations and the degree of association between the independent and dependent variables, respectively.

RESULTS

Of the 432 people who were screened, 250 were enrolled. From the original cohort, participants with incomplete falls (n = 38) or balance data (n = 16) were excluded from analysis (n = 196); two outliers in fall counts were also removed, leaving 194 individuals for analysis (FIGURE 1). The cohort was 49% female, and participants were an average of 69.5±6.5 years old with BMIs of 29.4±5.6. Comparisons for baseline characteristics across levels of the aberrant movement summary score are detailed in TABLE 1; MANOVA testing revealed significant differences in participant characteristics based on aberrant movement summary score (F (24, 555) = 1.839, P = .009; Pillai’s trace = 0.221, partial η² = .074). Main effects were significant for composite LBP intensity (F (3) = 2.841, P = .039, partial η² = .043) and FSST balance performance (F (3) = 4.290, P = .006, partial η² = .063); composite LBP intensity was not significantly different between levels of aberrant movement summary scores after taking multiple comparisons into account, while FSST performance was worse in participants with aberrant movement summary scores of 2 compared to participants without aberrant movements (mean difference = 1.6, P = 0.009, 95% confidence interval = 0.3–2.9).

FIGURE 1.

Consolidated standards of reporting trials (CONSORT) study flow diagram

Table 1:

Comparison of baseline participant characteristics across levels of aberrant movement summary scores.

	Entire Sample (n=194) (mean ± SD or %)	Aberrant movement Score of 0 (n=72) (mean ± SD or %)	Aberrant movement Score of 1 (n=74) (mean ± SD or %)	Aberrant movement Score of 2 (n=42) (mean ± SD or %)	Aberrant movement Score of 3 (n=6) (mean ± SD or %)	Main effect P - value	Post-hoc P - value‡
Age (years)†	69.6±6.5	68.2±5.8	70.6±6.8	70.1±7.2	69.2±5.0	0.164	-
Sex (female)	49%	50%	51.3%	42.9%	50%	0.843	-
BMI (kg/cm2)	29.4±5.6	29.3±5.6	29.2±5.8	30.2±5.4	28.0±3.8	0.739	-
Composite LBP intensity (0–10)	2.9±1.3	2.8±1.2	2.7±1.2	3.2±1.3	4.0±2.3	0.039*	-
Number of prior falls	1.0±2.6	1.2±2.4	0.7±1.7	1.0±3.9	1.3±2.4	0.738	-
Number of anxiolytics	0.1±0.3	0.03±0.2	0.1±0.3	0.1±0.4	0.0±0.0	0.171	-
Number of hip OA signs and symptoms†	3.0±1.3	2.9±1.2	3.1±1.2	3.1±1.6	3.5±2.1	0.704	-
FSST (seconds)	10.2±2.7	9.6±2.3	10.1±2.4	11.2±3.3	11.9±3.5	0.006*	0.009*a
Number of prospective falls	1.0±1.8	0.7±1.2	1.2±2.3	0.9±1.4	2.7±2.6	0.027*	0.042*b

Abbreviations: BMI = body mass index; FSST = Four-Square Step Test; LBP = low back pain; OA = osteoarthritis; SD = standard deviation.

^†= violated Levene’s test based on median;

^*= p<0.05;

^‡= post-hoc p-values of significant differences;

^*a= significant difference between aberrant movement summary scores of 2 and 0;

^*b= significant difference between aberrant movement summary scores of 3 and 0

FIGURE 2 depicts the distribution of prospective falls categories in the study sample, delineated by presence or absence of aberrant movements. Over the course of one year, falls were sustained by 40.7% of participants in this secondary analysis (n = 194). Of the individuals who fell (n = 79), participants with aberrant movements accounted for 61.8% of single fallers and 71.1% of multiple fallers. Among participants with aberrant movements (n = 122), 43.4% fell at least once; of these participants, 60.4% fell multiple times. Conversely, 36.1% of participants without aberrant movements fell (n = 72), of which 50% fell multiple times. MANOVA revealed significant main effects for prospective fall total (F (3) = 3.120, P = .027, partial η² = .047). Post-hoc testing revealed that participants with aberrant movement summary scores of 3 fell significantly more than participants without aberrant movements (mean difference = 2.0, P = 0.042, 95% confidence interval = 0.1–3.9).

FIGURE 2.

Delineation of fall categories based on presence or absence of aberrant lumbopelvic Movements

In Poisson regression models, multicollinearity does not appear to be present between aberrant movement summary scores and covariates based upon variance inflation factors (all between 1.062–1.370) and bivariate correlations (all r ≤ 0.338). TABLE 2 contains risk ratios (RR) for each independent variable. The overall model was statistically significant (χ² = 61.4(9), P < 0.001). Baseline aberrant movement summary scores were independently associated with falling, such that fall risk accrues by 24.9% for each additional aberrant movement present (RR = 1.249, 95% CI = 1.047–1.491, P = 0.014). Anxiolytic medication usage (P = <0.001), number of prior falls (P < 0.001), and number of hip OA signs and symptoms (P = 0.036) were also positively associated with falling. Neither composite LBP intensity nor FSST performance were associated with falling (P > 0.05).

Table 2:

Poisson regression model risk ratios for the association between aberrant lumbopelvic movements and falling (n = 194).

Variable	RR (95% CI)	P - value
Age (years)	1.020 (0.995–1.045)	0.112
Sex	1.091 (0.774–1.538)	0.618
BMI (kg/cm2)	1.000 (0.973–1.028)	0.991
Composite LBP intensity (0–10)	0.981 (0.878–1.097)	0.740
Number of prior falls	1.086 (1.053–1.120)	<0.001*
Number of anxiolytics	2.320 (1.628–3.305)	<0.001*
Number of hip OA signs and symptoms	1.141 (1.009–1.292)	0.036*
FSST (seconds)	0.975 (0.915–1.038)	0.424
Aberrant movements	1.249 (1.047–1.491)	0.014*

Abbreviations: BMI = body mass index; CI = confidence interval; FSST = Four-square step test; LBP = low back pain; OA = osteoarthritis; RR = risk ratio.

^*= p<0.05

DISCUSSION

The purpose of this study was to evaluate if clinically observable aberrant lumbopelvic movements predicted fall risk over 1-year follow-up in older adults with chronic LBP. After adjusting for well-known risk factors for falling in this cohort,³⁷ baseline aberrant movement summary scores were positively associated with falling; each additional aberrant movement present corresponded to an increase in fall risk by 24.9% during follow-up. To our knowledge, this study is the first to demonstrate a link between observable lumbopelvic incoordination (i.e., aberrant movements) and heightened fall risk in a chronic LBP population.

Our longitudinal findings are in accordance with cross-sectional biomechanical evidence which established that trunk neuromuscular function (and resultant coordination) are important components of stationary, anticipated, and reactive postural control and can be impacted by both aging and LBP.^13–19 Impaired trunk neuromuscular function, suggested by the presence of aberrant movements, may impact postural control through several neuromuscular pathways: hindering stationary postural sway, thereby increasing the likelihood of falling;^13,20 reducing the efficiency of anticipatory postural control, thereby complicating expected balance tasks;^17,46 and decreasing reactive postural control, thereby impeding compensatory strategies to avoid falling.^{14–16,18,19} The net result of these potential postural control deficiencies may be an inability to maintain (or return) the center of mass within the base of support. At baseline, participants with aberrant movements performed worse during an unperturbed dynamic balance task that incorporates multidirectional stepping; as such, aberrant movements may be related to anticipatory postural control. By controlling for dynamic balance performance, our statistical model partially accounts for anticipatory postural control; therefore, the independent association between aberrant movements and prospective fall risk may suggest that aberrant movements could be linked to reactive postural control through altered trunk neuromuscular function. To better understand the link between aberrant movements and falling, future studies are needed to elucidate the relationship between underlying trunk impairments (e.g., reduced muscle activity), aberrant movements, and indices of anticipatory and reactive postural control.

The present findings converge with evidence that demonstrates a link between trunk impairments, poor balance, and falls. Trunk muscle characteristics such as strength, endurance, size, and composition are well-established contributors to balance performance among older adults with and without LBP.^47,48 The relationship between trunk characteristics and balance is extended by evidence showing that among patients with hip fractures, a group of known fallers, trunk musculature is atrophied compared to controls.⁴⁹ A recent report established that higher aberrant movement summary scores are independently associated with longitudinal decline in a latent construct of physical function among older adults with chronic LBP.²⁷ Importantly, this latent construct was representative of several physical function tests that are predictive of falling (e.g., repeated chair rise, gait speed, and stair climbing).⁵⁰ Given the congruence between literature on trunk muscle characteristics and aberrant movements, it is reasonable to suggest that aberrant movements may be surrogate signs of morphological or functional changes to trunk musculature that accompany aging and chronic LBP. Clinicians should consider screening for aberrant movements in order to inform prognosis about functional trajectories in older adults with chronic LBP.

Our findings align with available literature on the implications of aberrant movements. Aberrant lumbopelvic movements are one component of a clinical prediction rule used to identify individuals who may respond to trunk stabilization exercises.⁵¹ A systematic review corroborated that trunk stabilization exercise may decrease fall risk in older adults.⁵² In fact, our previous work in older adults with chronic LBP preliminarily demonstrated that trunk muscle stabilization training, augmented with neuromuscular electrical stimulation, can lead to significant improvements in performance-based and self-reported measures of physical function.⁵³ Therefore, trunk stabilization exercise may normalize impairments that underlie aberrant movements, thereby lessening fall risk among older adults with chronic LBP who demonstrate aberrant movements; future research is needed to more directly address this research question.

Our study results should be considered in light of several limitations. In particular, the data was culled from a prospective observational cohort study on older adults with chronic LBP. The observational nature of the study precludes the ability to make causative claims about the associations between aberrant lumbopelvic movements and fall risk in this population. Additionally, post-hoc testing revealed that participants with missing data had significantly worse composite LBP intensity (p=0.007), poorer FSST balance performance (p=0.029), and greater aberrant movement summary scores (p=0.040) compared to participants with complete data; as such, it is possible that this loss-to-follow-up biased our findings towards the null hypothesis.

Our study has notable strengths. First, our Poisson regression model was statistically significant and accounted for a range of well-reported risk factors in this specific cohort³⁷ as well as gerontology literature in general.^{7–12,33,44,45} Second, our cumulative fall totals were calculated from monthly fall calendars, which are recognized as an optimal method of data collection in older adults^41,42; moreover, our attrition rate for falls data is comparable to other longitudinal fall investigations among older adults with LBP.¹¹ Third, our results were consistent across negative binomial and Poisson regression models, suggesting that our results are not biased by variable distribution. Finally, the parent cohort study was a prospective, longitudinal study of 250 older adults with chronic LBP that provided a robust data set for analysis.

CONCLUSION

Aberrant lumbopelvic movements are predictive of falls in older adults with chronic LBP, beyond well-known risk factors for falling. This finding provides clinical context for biomechanical evidence that previously established the relationship between trunk neuromuscular function, trunk coordination, and maintenance of postural control across various experimental conditions. Clinicians should consider screening for aberrant movements in older adults with chronic LBP in order to develop plans of care that may mitigate downstream consequences of poor trunk neuromuscular function and coordination. Future studies are needed to elucidate which trunk impairments contribute to aberrant movements in this patient population, as well as to determine the efficacy of trunk stabilization exercises to reduce fall risk.

CLINICAL IMPLICATIONS.

• Older adults with LBP are more likely to fall than pain-free counterparts, so it is essential to identify risk factors for falling in this population.

• Aberrant lumbopelvic movements during forward bending are associated with heightened fall risk in older adults with chronic LBP above and beyond well-established risk factors for falling.

• To optimize the quality of care for this patient population, clinicians should consider screening for aberrant lumbopelvic movements.