The association of peripheral neuropathy detected by monofilament testing with risk of falls and fractures in older adults

Abstract

Background:

In persons with diabetes, annual screening for peripheral neuropathy (PN) using monofilament testing is the standard of care. However, PN detected by monofilament testing is common in older adults, even in the absence of diabetes. We aimed to assess the association of PN with risk of falls and fractures in older adults.

Methods:

We included participants in the Atherosclerosis Risk in Communities (ARIC) Study who underwent monofilament testing at visit 6 (2016–2017). Incident falls and fractures were identified based on ICD-9 and ICD-10 codes from active surveillance of all hospitalizations and linkage to Medicare claims. We used Cox models to assess the association of PN with falls and fractures (combined and as separate outcomes) after adjusting for demographics and risk factors for falls.

Results:

There were 3617 ARIC participants (mean age 79.4 [SD 4.7] years, 40.8% male, and 21.4% Black adults), of whom 1242 (34.3%) had PN based on monofilament testing. During a median follow-up of 2.5 years, 371 participants had a documented fall, and 475 participants had a documented fracture. The incidence rate (per 1000 person-years) for falls or fractures for participants with PN versus those without PN was 111.1 versus 74.3 (p < 0.001). The age-, sex-, and race-adjusted 3-year cumulative incidence of incident fall or fracture was significantly higher for participants with PN versus those without PN (26.5% vs. 18.4%, p < 0.001). After adjusting for demographics, PN remained independently associated with falls and fractures (HR 1.48, 95% CI 1.26, 1.74). Results were similar for models including traditional risk factors for falls, when falls and fractures were analyzed as separate outcomes, and after adjustment for competing risk of death.

Conclusions:

PN, as measured by monofilament testing, is common in older adults and associated with risk of falls and fracture. Screening with monofilament testing may be warranted to identify older adults at high risk for falls.

Editor’s Note

Us “older” physicians who teach students and residents know that the art and science, as well as the diagnostic value, of the traditional history and physical exam is being lost by the ready availability of technology for making a diagnosis. In this study, an enhanced component of the physical examination (monofilament testing with a specified protocol), was used as a diagnostic tool rather than a more technologic and expensive method for detecting peripheral neuropathy (PN). Using this simple physical examination technique, the authors found a prevalence of PN of 34%. During a median follow up of 2.5 years in this population of 3,617 participants in a primary study of atherosclerotic risk with mean age 79, 59% female, 21% black, and 31% diabetic, 371 had an incident fall, and 475 had an incident fracture. The crude incidence rate (per 1000 person-years) for falls or fractures was 111.1 (95% CI 98.8, 125.0) in participants with PN versus 74.3 (95% CI 67.2, 82.8) in persons without PN. After adjusting for demographics, the risk of incident fall or fracture was significantly higher for participants with PN (HR 1.48, 95% CI 1.26, 1.74).

I guess I am an old-time geriatric geriatrician—I do not use monofilament testing. But almost every patient I see who is diabetic, has had one or more falls, gait instability, or complaints of discomfort in the legs, gets what I call a “screening neurologic exam.” This includes, among other assessments: postural vital signs (performed by me or a member of the team); a carefully observed, untimed stand-up test using no hands if possible, a Romberg test, a 10-foot walk and turn around evaluation; heel tapping and heel to shin testing for cerebellar function; and basic testing of position sense, touch using a Q-tip, and pain sensation using one of my sharp cocktail olive toothpicks. I carry these tools in my white coat. And of course, I look carefully at the feet with socks off for painful deformities, skin integrity, and condition of the toenails, as all geriatricians should.²⁶ This screening examination takes me a grand total of approximately 3–5 min.

My point is, whether it is monofilament testing or your own “screening neurological examination,” the physical exam of older people to detect PN and foot abnormalities is a critical component of geriatric assessment, one that can predict and possibly prevent a future devastating fall or fracture.

-Joseph G. Ouslander, MD

INTRODUCTION

Peripheral neuropathy (PN) contributes to substantial morbidity, including pain, foot ulcers, and lower extremity amputation.¹ PN, as defined by monofilament testing, is specific for loss of protective sensation and typically corresponds to an advanced form of PN.²

In persons with diabetes, annual screening for PN using monofilament testing is the standard of care.² However, decreased sensation in the foot detected by monofilament testing is common in older adults, even in the absence of diabetes.³ The sequelae related to PN in the absence of diabetes are not well understood. We have previously shown a strong link between PN and mortality in the general population⁴ and associations with impairments in hearing, vision,⁵ and cognition⁶ in older adults.

Falls are a major source of morbidity among older adults.⁷ There is evidence that PN in adults with diabetes is associated with gait imbalance and risk of falls, and PN has been associated with self-reported falls in adults with diabetes.⁸ However, the association of PN with risk of falls and fractures has not been rigorously investigated in a general population, including adults both with and without diabetes. The aim of this study was to assess the association of PN (assessed by monofilament testing) with incident falls and fractures in a community-based population of older adults.

METHODS

We included all Atherosclerosis Risk in Communities (ARIC) Study participants who underwent monofilament testing at visit 6 (data collected between 2016 and 2017). ARIC is a community-based study from four US communities (Forsyth County, North Carolina; Jackson, Mississippi; suburban Minneapolis, Minnesota; and Washington County, Maryland). ARIC participants have been followed with serial visits and active surveillance for all hospitalizations.

Among the 4003 participants who attended ARIC visit 6, we excluded participants who were missing monofilament testing (n = 386). The ARIC study was approved by the Institutional Review Boards at each study site and informed consent was obtained from all participants.

PN assessed using monofilament testing

PN was assessed at ARIC visit 6 using Semmes-Weinstein 10 g monofilament testing of three sites on each foot: the plantar-hallux, the plantar-first metatarsal head, and the plantar-fifth metatarsal head. Each site was tested three times by a certified technician.⁹ If two of the three responses for a site were incorrect or indeterminate, the response was considered insensate at that site. We defined PN as having at least one insensate site.

Covariates

Other covariates of interest included age, sex, race, education, body mass index (BMI), smoking status, alcohol consumption, physical activity, cognitive status, diabetes, hypertension, hypercholesterolemia, history of cardiovascular disease, lower extremity peripheral artery disease (PAD), chronic kidney disease (CKD), and cancer (Data S1).

We also conducted analyses adjusting for medication use, including diuretics and psychotropic medications (antidepressants, sedatives, hypnotics, antipsychotic medications, anti-epileptic medications, and anticholinergic medications) associated with fall risk.¹⁰ Medication information was captured based on the medication inventory performed at ARIC visit 6. History of a prior fall at or before visit 6 was defined by hospitalization records or Medicare Claims.

Falls and fractures

Falls and fractures were identified using the International Classification of Diseases (ICD) ICD-9 (prior to 2015), ICD-10 (2015 and later), and ICD-Clinical Modification E codes (Table S1). The use of ICD-9 and E codes for falls and fractures has been previously validated in ARIC.¹¹ We mapped ICD-9 codes to ICD-10 codes for claims after 2014. The primary analysis focused on participant’s first fall or fracture captured (i.e., incident fall/fracture). We performed a sensitivity analysis additionally including self-reported incident falls after visit 6, collected at semi-annual telephone calls using the question “In the past 12 months, did you fall?”

Statistical analysis

We evaluated baseline characteristics of participants according to PN status. We calculated incidence rates of falls and fractures and generated adjusted cumulative incidence curves according to PN status. We used Cox regression to assess the association of PN with incident falls or fractures. Model 1 included age, sex, and race. Model 2 included all variables in Model 1 plus education, BMI, smoking status, drinking status, physical activity, hypertension, hypercholesterolemia, prevalent CVD, prevalent PAD, CKD, cancer, diabetes, cognitive status, antidepressant use, sedative use, hypnotic use, antipsychotic use, diuretic use, antiepileptic use, anticholinergic use, and history of prior falls.

We performed a sensitivity analysis accounting for the competing risk of death using Fine and Gray’s method. We also performed sensitivity analyses additionally including self-reported falls, and assessed the association of PN with rates of hospitalized fall or fracture (counts) using Poisson regression.

We performed subgroup analyses by age (<80 years vs. ≥80 years), sex, and diabetes. In the analyses stratified by diabetes status, we included diabetes duration (<10 years vs. ≥10 years) as a covariate among adults with diabetes.

We performed all analyses using Stata version 17.0 (StataCorp), with two-tailed p < 0.05 denoting statistical significance.

RESULTS

We included 3617 ARIC participants at visit 6, of whom 1242 (34.3%) had PN based on monofilament testing. The mean age was 79.4 (SD 4.7), 40.8% were male, and 21.4% were Black adults. Diabetes was present in 30.6% of participants, and was more prevalent among participants with PN (34.3%) compared to participants without PN (28.6%) (Table 1).

Table 1.

Baseline characteristics of study participants overall and according to peripheral neuropathy (PN) status, ARIC (2016–2017).

	Overall	No PN	PN
N	3617	2375	1242
Age, mean (SD), years	79.4 (4.7)	78.7 (4.4)	80.5 (5.0)
Male sex	40.8%	32.9%	56.0%
Black race	21.4%	21.1%	21.8%
Education
Some college and above	47.6%	48.1%	46.6%
High school or vocational school	41.5%	42.7%	39.1%
Did not complete high school	10.9%	9.2%	14.3%
Body mass index, kg/m2
<25.0	29.5%	30.8%	26.9%
25.0–29.99	38.3%	39.4%	36.3%
≥30.0	32.2%	29.8%	36.8%
Smoking status
Never	45.2%	46.4%	43.0%
Former	48.3%	47.4%	50.0%
Current	6.5%	6.2%	7.0%
Drinking status
Current moderate drinker	46.7%	47.0%	46.1%
Current heavy drinker	4.6%	5.2%	3.6%
Former drinker	28.5%	27.4%	30.7%
Never drinker	20.1%	20.4%	19.6%
Activity index, mean (SD)	2.2 (0.6)	2.2 (0.6)	2.2 (0.6)
Cognitive status
Normal	75.9%	79.4%	69.1%
Mild cognitive impairment	18.7%	16.5%	22.8%
Dementia	5.4%	4.0%	8.1%
Diabetes	30.6%	28.6%	34.3%
Hypertension	84.6%	83.1%	87.5%
Hypercholesterolemia	60.7%	60.3%	61.5%
Prevalent cardiovascular disease	20.9%	18.4%	25.8%
Peripheral artery disease	4.0%	4.2%	3.5%
Chronic kidney disease	39.6%	36.2%	46.2%
Cancer	15.0%	13.3%	18.3%
Medication use
Antidepressants	12.7%	12.9%	12.3%
Sedatives	0.2%	0.2%	0.2%
Hypnotics	5.1%	5.1%	5.2%
Antipsychotics	0.2%	0.3%	0.2%
Diuretics	27.3%	25.6%	30.6%
Anticholinergics	3.8%	3.5%	4.3%
History of fall	21.5%	19.3%	25.7%

During a median follow-up of 2.5 years (maximum 3.5 years), 371 participants had an incident fall and 475 participants had an incident fracture. The crude incidence rate (per 1000 person-years) for falls or fractures was 111.1 (95% CI 98.8, 125.0) in participants with PN versus 74.3 (95% CI 67.2, 82.8) in persons without PN (p < 0.001; Table 2). The age-, sex-, and race-adjusted 3-year cumulative incidence of incident fall or fracture was significantly higher for participants with versus without PN (26.5% vs.18.4%, p < 0.001; Figure 1).

Table 2.

Incidence rates (IR, 95% CI) and hazard ratios (HR, 95% CI) of fall or fracture stratified by peripheral neuropathy (PN) status, ARIC participants 2016 to 2019.

	n event/N	IR (per 1000 person-years) IR (95% CI)	Incidence rate ratio, IRR (95% CI)	Model 1 HR (95% CI)	Model 2 HR (95% CI)
Fall or Fracture
No PN	374/2375	74.3 (67.2, 82.2)	1 (ref)	1 (ref)	1 (ref)
PN	279/1242	111.1 (98.8, 125.0)	1.50 (1.28, 1.75)	1.48 (1.26, 1.74)	1.37 (1.16, 1.61)
Fall
No PN	203/2375	38.2 (33.3, 43.9)	1 (ref)	1 (ref)	1 (ref)
PN	168/1242	62.6 (53.8, 72.9)	1.64 (1.34, 2.01)	1.53 (1.24, 1.90)	1.36 (1.09, 1.69)
Fracture
No PN	281/2375	54.3 (48.3, 61.0)	1 (ref)	1 (ref)	1 (ref)
PN	194/1242	73.9 (64.2, 85.0)	1.36 (1.13, 1.63)	1.40 (1.16, 1.70)	1.32 (1.09, 1.61)

Note: Model 1: Adjusted for age, sex and race. Model 2: Adjusted for covariates in model 1 plus education, BMI, smoking status, drinking status, physical activity, hypertension, hypercholesterolemia, cardiovascular disease, peripheral artery disease, chronic kidney disease, cancer, diabetes status, cognitive status, antidepressant use, sedative use, hypnotic use, antipsychotic use, diuretic use, anticholinergic use, and history of fall.

FIGURE 1.

Adjusted* cumulative incidence of incident fall or fracture according to peripheral neuropathy (PN) status, ARIC 2016–2019. PN is denoted by the red line. No PN is denoted by the blue line. *Adjusted for age, sex, and race.

After adjusting for demographics, the risk of incident fall or fracture was significantly higher for participants with PN (HR 1.48, 95% CI 1.26, 1.74). This association remained significant after further adjustment (Model 2: HR 1.37, 95% CI 1.16, 1.61). Our results were similar when falls and fractures were analyzed separately (Table 2), in analyses accounting for the competing risk of death (Table S2), examining a composite outcome of either incident fall or self-reported fall (Table S3), and when we examined rates of hospitalized falls or fractures (Table S4).

The association of PN with incident fall or fracture was significant among adults <80 years of age (HR 1.54, 95% CI 1.21, 1.96), but attenuated in adults ≥80 years of age (HR 1.17, 95% CI 0.93, 1.48) although the interaction by age was not statistically significant (P-interaction = 0.16). The association of PN with incident fall or fracture was significant for participants without diabetes (HR 1.60, 95% CI 1.30, 1.95) but not for participants with diabetes (p-interaction = .01) (Figure S1).

DISCUSSION

In a community-based population of older adults, PN, as assessed by monofilament testing, was associated with incident falls and fractures. Our findings persisted after adjusting for demographics and risk factors for falls, including medications associated with risk of falls.¹⁰ Associations were stronger among adults without diabetes. Loss of foot sensation is common in older adults, and our findings of a high risk of falls and fractures in this population suggest a possible role for screening with monofilament testing to identify older adults at high risk for falls.

The risk of fall or fracture was significantly higher in older adults with versus without PN, with a 3-year cumulative incidence of 26.5% for older adults with evidence of decreased sensation on monofilament testing. This may be an underrepresentation of the actual fall rates in this population, as our main analysis focused on injurious falls or fractures. Approximately 3 million older adults are treated in emergency departments for a fall injury each year, and one-fifth of those falls results in an injury such as a fracture or head trauma.¹² In 2018, 27.5% of adults aged ≥65 years reported falling at least once in the past year, and 10.2% reported a fall-related injury.¹³ Risk factors for falls include hypertension, balance and gait impairment, cognitive impairment, and use of sedative or hypnotic medications.^14,15 Falls are preventable, and screening older adults for fall risk and assessing modifiable risk factors is important for reducing fall risk.¹⁶

The association of PN with falls has been previously demonstrated in adults with diabetes.⁸ Gait disturbances that alter balance in the setting of loss of protective sensation may play an important role.¹⁷ However, we did not find a significant association of PN with incident injurious falls or fractures in adults with diabetes in our study. It is possible this reflects survival bias, as participants with severe diabetes and related complications were less likely to return to study visits and may not have been captured in our analysis. It is also possible that older adults with diabetes in our study are receiving regular foot screening and appropriate counseling around loss of protective sensation that helped mitigate fall risk. The American Diabetes Association recommends annual foot screening in adults with diabetes that includes monofilament testing for loss of protective sensation.² Those adults who test positive on one or more screening tests during a foot exam should be provided with education about general risk factors, foot self-care practices, consideration of a podiatry referral, and the importance of appropriate footwear and footwear behaviors at home to reduce fall risk.^18,19 A number of previous studies have shown the benefit of wearing athletic footwear for reducing fall risk, particularly compared to stocking feet or bare feet.^19,20 Notably, these recommendations do not extend to adults without diabetes.

Our findings are novel because the association of PN with falls or fractures has not been rigorously explored in older adults without diabetes. The high prevalence of PN in adults without diabetes and its consequences are not well recognized. Guidelines for foot screening for PN are limited to adults with diabetes; there are currently no guidelines that recommend foot screening or monofilament testing in adults without diabetes. The American Geriatrics Society recommends screening older patients at least annually for fall risk by asking whether they have fallen within the previous year and whether they have difficulties with walking or balance.²¹ However, practical clinical screening tests for fall risks are lacking, and falls risk screening in older adults are inconsistent.²² Our results suggest that monofilament testing may be a useful screening approach, regardless of diabetes status, to help assess risk of falls. Although there are other more sensitive and specific tests for PN,²³ monofilament testing is useful as a clinical tool for detecting severe neuropathy and is straightforward, inexpensive, and quick to perform.²⁴ It is possible that a positive monofilament test represents a heterogenous disease that causes falls, but the association of reduced sensation in the foot with incident falls and fractures remains clinically significant regardless of the etiology.

The limitations of our study include the use of monofilament testing at a single time point; we did not have information on duration or severity of PN. Falls and fractures were identified using administrative codes. While the ICD codes for falls and fractures have been previously validated in ARIC,¹¹ there is the possibility of misclassification using claims codes, which may limit the sensitivity of fall and fracture diagnoses. We performed a sensitivity analysis defining falls using both claims codes and self-report, and our overall results were similar. There remains the possibility of residual confounding; we did not account for all potential risk factors for falls and fractures in older adults, including neurologic conditions that may play a role.²⁵ The gold standard for diagnosis of PN is through electromyography studies.²³ The monofilament testing protocol used here for PN⁹ is specific to loss of protective sensation but may not be sensitive for less severe forms of PN. Strengths of our study included the prospective study design in community-dwelling older adults, large sample size, standardized monofilament testing protocol for PN, surveillance for falls and fractures, and assessment of covariates by trained personnel using standardized protocols.

In conclusion, the burden of PN as assessed by monofilament testing is high in older adults and is associated with risk of falls and fractures. Our results suggest that screening for PN with monofilament testing outside the setting of diabetes may be warranted to identify older adults at high risk for falls.

Supplementary Material

Supplemental Tables and Figures

Supplementary Table S1. International Classification of Diseases (ICD) codes and E codes used to identify falls and fractures.

Supplementary Table S2. Association (HR, 95% CI) of peripheral neuropathy (PN) with incident fall or fracture after accounting for competing risk of death (Fine and Gray method), ARIC participants 2016 to 2019.

Supplementary Table S3. Incidence rates (IR, 95% CI) and hazard ratios (HR, 95% CI) of peripheral neuropathy (PN) with incident fall or fracture including incident self-reported falls, ARIC participants 2016 to 2019.

Supplementary Table S4. Adjusted association, incidence rate ratio (IRR, 95% CI), of peripheral neuropathy (PN) with hospitalized fall or fracture counts, ARIC participants 2016 to 2019.

Supplementary Figure S1. Hazard ratios (95%CI) for peripheral neuropathy (PN) with incident fall or fracture in subgroups defined by age, sex, and diabetes status.

Key points.

• Among older adults in the Atherosclerosis Risk in Communities (ARIC) Study, peripheral neuropathy (PN) as assessed by monofilament testing was associated with risk of falls and fractures.

• The age-, sex-, and race-adjusted 3-year risk of incident fall or fracture was 26% among older adults with PN, and the association was independent of traditional risk factors for falls and fractures.

Why does this paper matter?

Decreased sensation in the foot may be an under-recognized risk factor for falls and fractures in the older general population.