GLP-1 Records in Diabetic Neuropathy

A Romanian hospital study found that patients with documented diabetic peripheral neuropathy and a GLP-1-based medication record also had more documented comorbidity than patients without such a record. Heart failure appeared in 35.7% of the GLP-1-record group and 22.2% of the comparison group^[1]. Those percentages do not show that a medicine caused heart failure. The analysis was a retrospective cross-sectional snapshot with no patient-level dates for treatment, diagnoses or outcomes.

The exposure category pooled exenatide, dulaglutide, semaglutide, fixed-ratio insulin and GLP-1 combinations, and tirzepatide. Only one patient had a tirzepatide record. The paper cannot support a tirzepatide-specific claim, a class safety verdict or a claim that these medicines improve neuropathy.

What the researchers analyzed

Researchers aggregated medication-level exports by patient identifier for 4,120 people classified as active patients at the Clinical County Emergency Hospital Bihor. The hospital-defined reporting window ran from February 2024 through February 2026, but the files did not include patient-level encounter, diagnosis or treatment dates. The authors therefore treated the dataset as an undated reporting snapshot^[2].

Every included patient had documented diabetic peripheral neuropathy, yet that classification and diabetes type came from export fields and were not independently validated. Diabetic peripheral neuropathy can involve pain, numbness, weakness and loss of protective sensation, especially in the feet and legs^[3]. This study did not grade those symptoms, nerve function or neuropathy severity.

The GLP-1-record group contained 852 patients. Exenatide accounted for 483 records, dulaglutide 142, semaglutide 127, fixed-ratio combinations 134 and tirzepatide one. Categories could overlap, so those entries should not be read as mutually exclusive treatment arms^[2]. The other 3,268 patients had no documented medication from the study's GLP-1-based list.

What differed between the groups

Mean documented diabetes duration was 14.85 years in the GLP-1-record group and 11.72 years in the comparison group. Mean documented comorbidity counts were 2.64 and 1.89, respectively. Hypertension, ischemic heart disease, stroke, nephropathy and dyslipidemia were also recorded more often in the GLP-1 group^[2]. These differences describe two unlike patient groups. They are plausible markers of treatment selection and clinical complexity rather than evidence of drug effects.

Heart failure was documented in 304 of 852 patients with a GLP-1-based record and 724 of 3,268 without one. The primary logistic model produced an adjusted odds ratio of 1.68, with a 95% confidence interval from 1.41 to 1.99. An expanded exploratory model returned 1.37, while a Poisson sensitivity model estimated a prevalence ratio of 1.45^[1]. The change across models shows how the estimate depends on adjustment choices. None of the models can recover which came first when dates are missing. A numerical association can remain precise while the underlying sequence of treatment and disease is completely unresolved.

Any lower-limb amputation was documented in 28 patients in the GLP-1-record group and 38 in the comparison group, equal to 3.3% and 1.2%. There were only 66 events in total. The authors therefore kept amputation descriptive and unadjusted rather than fitting a multivariable model^[2]. Without treatment or amputation dates, the difference cannot estimate incident limb risk or treatment harm.

Why this is an association, not a treatment result

Timing is the central problem. Cross-sectional data place characteristics in the same reporting snapshot, so they cannot show whether a medication record preceded heart failure, amputation or another condition. Clinicians may also choose GLP-1-based medicines more often for people with obesity, cardiovascular disease or a need to change glucose-lowering therapy. That selection process can create associations even when the medicine did not produce the outcome. Statistical adjustment cannot reconstruct a timeline that the source files never recorded.

The record fields introduced more uncertainty. A listed medicine did not prove that a prescription was current, dispensed or taken. Dose and adherence were absent. The analysis also lacked HbA1c, estimated glomerular filtration rate and albuminuria. Body mass index was available for only 271 patients after plausibility exclusions, or 6.6% of the cohort^[2]. That missingness prevents a reliable whole-cohort comparison of body size.

What we don't yet know

This study did not test whether GLP-1-based treatment changes neuropathy symptoms, nerve function, ulcers, amputation incidence or heart-failure risk. That question remains open. Longitudinal data with verified start dates and outcome dates would be needed to establish temporal order, while any credible treatment-effect analysis would also need better measurement of diabetes control, kidney function, vascular disease, neuropathy severity and the clinical reasons each medicine was selected.

The grouped exposure also leaves molecule-specific questions unanswered. One tirzepatide record cannot inform tirzepatide safety or neuropathy outcomes. The 127 semaglutide records were not analyzed as a separate outcome group. Readers should not use the pooled numbers to start, stop or switch either medicine. A clinician can interpret neuropathy symptoms and treatment choices in the context of the person's diabetes, kidney and cardiovascular history.

Medical disclaimer: This article is for educational and journalistic purposes only and does not constitute medical advice. Peptides discussed may be classified as prescription medicines or research chemicals depending on your jurisdiction. Always consult a qualified healthcare professional before using any peptide product. PeptideMethods.com does not sell, distribute, or facilitate the sale of any peptide product.