HbA1c Variability Linked to Recurrent Falls in Older Adults With Diabetes

In a single-center retrospective cohort study of 286 older adults with type 2 diabetes, greater visit-to-visit variation in glycated hemoglobin A1c (HbA1c) was associated with a higher risk of recurrent falls over the following year. After adjustment for mean HbA1c and several fall-related factors, each 3.1-percentage-point increase in the between-patient standard deviation of HbA1c variability was associated with a 34% higher risk of at least two falls. The findings suggest that HbA1c variability may carry information beyond average glucose levels when assessing fall risk, although the observational study did not establish that reducing variability would prevent falls.

The study was conducted at Chongqing Jiulongpo District People’s Hospital in China and included patients aged 65 or older who received standardized outpatient diabetes care. Eligible participants had at least four routine HbA1c measurements during a 364-day exposure window, with consecutive tests at least eight weeks apart and a span of at least nine months from the first to the last measurement. The researchers also required complete ascertainment of falls during the subsequent 365 days. Records dated from 2022 through 2025 were used to construct the exposure and outcome windows. Of 341 patients screened, 55 were excluded, including 23 whose number or timing of HbA1c measurements did not meet the criteria, 12 whose fall information could not be reliably verified, eight who could not walk independently or with an assistive device, seven with conditions affecting HbA1c interpretation, and five with another type of diabetes.

The primary exposure was the HbA1c coefficient of variation, abbreviated HbA1c-CV. It was calculated by dividing each patient’s within-person HbA1c standard deviation by that patient’s mean HbA1c and multiplying by 100, expressing variability relative to the average level. The one-standard-deviation increment used in the analysis referred to the between-patient sample standard deviation of HbA1c-CV, equivalent to 3.1 percentage points, rather than the average within-patient fluctuation in HbA1c. HbA1c was measured by ion-exchange high-performance liquid chromatography and reported in United States National Glycohemoglobin Standardization Program units. Only quality-controlled results obtained during routine outpatient care were included, while measurements taken on an ad hoc basis during acute hospitalizations were excluded. The researchers also evaluated within-person HbA1c standard deviation, variability independent of the mean, and residual standard deviation after removal of a within-person linear time trend as alternative exposure measures.

Recurrent falls were defined as at least two independent events during the 365-day outcome period in which a patient unexpectedly came to rest on the ground, floor, or another lower level. Intentional acts, traffic accidents, falls from a height, and events caused by major external forces were excluded. A subsequent fall counted as a new event only if the patient had returned to the pre-event level of activity. Investigators searched structured outpatient records, telephone follow-up records, emergency department and inpatient records, nursing records, injury and fracture diagnoses, and imaging data. The outcome therefore was not limited to falls that caused injury or required medical attention, although noninjurious or unreported events could still have been missed. Two investigators blinded to HbA1c variability independently adjudicated events and dates, with disagreements reviewed by a third investigator. Across the cohort, 127 patients experienced at least one fall, 77 experienced recurrent falls, and 240 falls were recorded in total.

Patients with recurrent falls differed from those without recurrent falls in several measured characteristics. The group with recurrent falls was older on average, had a longer median duration of diabetes, and more frequently had a previous fall, severe hypoglycemia, limitations in activities of daily living, walking-aid use, cognitive or visual impairment, diabetic peripheral neuropathy, cerebrovascular disease, and orthostatic hypotension. They also had a lower estimated glomerular filtration rate, used more long-term medications, and had more medication classes associated with fall risk. Patients with recurrent falls also had higher mean HbA1c and greater variability on HbA1c-CV and all three alternative measures. The primary adjusted model included age, mean HbA1c, number of HbA1c measurements, span between the first and last measurements, previous falls during the exposure window, severe hypoglycemia, limitations in activities of daily living, and treatment considered to carry a high risk of hypoglycemia. Missing covariate data were handled with multiple imputation using 20 datasets, while the primary exposure and outcomes were complete for all participants.

Unadjusted results showed a gradient in recurrent-fall risk across HbA1c-CV tertiles. The 365-day risk was 15.8% in the lowest tertile, with a 95% confidence interval of 9.8% to 24.4%; 25.3% in the middle tertile, with a 95% confidence interval of 17.6% to 34.8%; and 39.6% in the highest tertile, with a 95% confidence interval of 30.4% to 49.6%. In the primary adjusted model, each one-standard-deviation increase in HbA1c-CV was associated with a risk ratio of 1.34 for recurrent falls (95% confidence interval, 1.16–1.55; P<0.001). Compared with the lowest tertile, the highest had an adjusted risk ratio of 2.03 (95% confidence interval, 1.19–3.47; P=0.010), while the middle tertile did not differ significantly from the reference group. A restricted cubic spline analysis detected an overall association but found no clear evidence of nonlinearity; the test for nonlinearity was not statistically significant (P=0.294).

The association was also present when researchers examined the total number of falls rather than whether recurrent falls occurred. An adjusted negative binomial model estimated an incidence rate ratio of 1.37 per one-standard-deviation increase in HbA1c-CV (95% confidence interval, 1.18–1.58; P<0.001). The highest tertile had an adjusted fall incidence rate 1.91 times that of the lowest tertile (95% confidence interval, 1.32–2.77; P=0.001), while the middle-tertile comparison was not statistically significant. The negative binomial model fit the count data better than a Poisson model. Before adjustment, the fall incidence rate rose from 52.6 falls per 100 person-years in the lowest HbA1c-CV tertile to 80.0 per 100 person-years in the middle tertile and 118.8 per 100 person-years in the highest tertile.

Sensitivity analyses produced estimates in the same direction. Replacing HbA1c-CV with within-person HbA1c standard deviation, variability independent of the mean, or residual standard deviation yielded adjusted recurrent-fall risk ratios of 1.30, 1.32, and 1.28 per one-standard-deviation increase, respectively. Among the 194 patients with exactly four HbA1c measurements, the risk ratio was 1.31 (95% confidence interval, 1.06–1.62; P=0.013). Excluding patients who had fallen during the exposure window left 198 participants and yielded a risk ratio of 1.29 (95% confidence interval, 1.01–1.65; P=0.042). Complete-case analysis produced a risk ratio of 1.35, while a ridge-penalized model with an expanded adjustment set produced 1.32 (bootstrap 95% confidence interval, 1.12–1.55). These secondary and sensitivity analyses were exploratory, and their P values were not adjusted for multiple comparisons.

The biological pathway remains uncertain. Recurrent metabolic and oxidative stress has been proposed as a possible link between long-term glycemic fluctuations and impaired nerve conduction, muscle perfusion, balance, and postural regulation, but the study did not test whether glycemic variability mediated falls. HbA1c variability may also reflect short-term glucose swings, inconsistent meals, medication adherence, treatment changes, or greater hypoglycemia exposure that were not fully captured in routine records. Peripheral neuropathy, cognitive impairment, visual impairment, orthostatic hypotension, impaired kidney function, and medicines associated with fall risk were more common among patients with recurrent falls, but the analyses did not establish their mediating roles. The persistence of the association after adjustment and across several variability measures supports its potential prognostic value without showing that changing HbA1c variability would alter fall outcomes.

The study has important limitations. Its single-center retrospective design cannot fully account for residual confounding, and information on diet, physical activity, sarcopenia, vestibular function, and the home environment was incomplete. Although the models accounted for the number of HbA1c measurements and the span between the first and last tests, measurement timing was not fully standardized, and the frequency of routine testing could still have influenced the exposure estimate. Falls were ascertained from existing medical and follow-up records, making it possible that some noninjury events went unreported. The number of recurrent-fall events was also limited, producing wide confidence intervals for some estimates. Eligibility required repeated HbA1c measurements, the ability to walk independently or with an assistive device, and verifiable fall data, which limits generalization to other healthcare settings, younger populations, and long-term care residents. The results therefore position HbA1c-CV as a possible complementary marker within multidimensional fall-risk assessment, not as a standalone predictor or a proven intervention target.

Subject of Research: Visit-to-visit HbA1c variability and recurrent falls in older adults with type 2 diabetes

Article Title: Association between visit-to-visit glycated hemoglobin A1c variability and recurrent falls over 12 months in older adults with type 2 diabetes mellitus

Article References: Hu, W., & Shao, A. (2026). Association between visit-to-visit glycated hemoglobin A1c variability and recurrent falls over 12 months in older adults with type 2 diabetes mellitus. Frontiers in Endocrinology, 17, Article 1971401. https://doi.org/10.3389/fendo.2026.1971401

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1971401

Keywords: HbA1c variability, recurrent falls, older adults, type 2 diabetes, glycemic instability, fall risk, hypoglycemia, diabetic neuropathy, retrospective cohort, Association, between, visit-to-visit