Three Routine Blood Tests May Predict Survival After Major Amputation in Older Diabetic Patients

For older adults with diabetes, losing a leg to a diabetic foot complication is one of the most devastating turning points in the disease. Major lower extremity amputation, whether below the knee or above it, marks the end stage of a cascade of nerve damage, blocked arteries, and non-healing wounds. What has long troubled surgeons and geriatricians alike is that the operation itself is often not the deadliest part of the story. Many of these patients die within the following two years, and clinicians have had few reliable tools to tell them, before the first incision is made, who is most at risk. A new retrospective cohort study from Istanbul now suggests that the answer may already be sitting in the patient’s chart, encoded in three of the most routine blood tests in medicine: hemoglobin, platelet count, and albumin.

The research, published in BMC Geriatrics by Mete Özer and colleagues at Başakşehir Çam and Sakura City Hospital, followed 72 patients aged 65 and older who underwent transtibial or transfemoral amputation for diabetic foot complications between January 2021 and June 2023 at a tertiary referral center. Vital status was tracked through July 31, 2025, giving the team a long observation window. The numbers they found are stark. Thirty-nine of the 72 patients, or 54.2 percent, died within 24 months of surgery. Kaplan–Meier analysis estimated 24-month survival at just 45.8 percent, with a 95 percent confidence interval of 34.1 to 56.8 percent. In other words, a patient entering the operating room for a major amputation in this cohort had roughly a coin flip’s chance of being alive two years later.

To find out what separated the survivors from those who died, the researchers turned to Cox proportional hazards regression, the standard statistical machinery for modeling time-to-event data. They tested a panel of preoperative laboratory parameters, all of which are drawn routinely before surgery. Three emerged as independent predictors of mortality. Higher hemoglobin was associated with increased risk, with an adjusted hazard ratio of 1.302 per unit increase (95 percent CI 1.041 to 1.628; p = 0.021). Lower platelet counts carried a slightly elevated hazard, with an adjusted hazard ratio of 0.9966 per platelet (95 percent CI 0.9939 to 0.9994; p = 0.017). Most striking was albumin, the liver-synthesized protein that serves as a classic marker of nutritional status and systemic inflammation: each unit decrease in albumin carried an adjusted hazard ratio of 0.868 (95 percent CI 0.813 to 0.927; p < 0.001), making it the strongest single laboratory signal in the model.

Some of these directions may seem counterintuitive at first glance. Higher hemoglobin, usually a sign of good oxygen-carrying capacity, predicted worse survival here. The authors’ framing suggests this likely reflects the physiology of chronic disease rather than a paradox: in patients with severe peripheral artery disease and chronic inflammation, elevated hemoglobin can accompany hemoconcentration, chronic hypoxia-driven erythropoiesis, and underlying cardiopulmonary strain. Low platelet counts, meanwhile, may reflect platelet consumption in the microvascular thrombosis that characterizes advanced diabetic vascular disease, or suppressed production in the context of systemic illness. Albumin’s role is the least surprising. Hypoalbuminemia is a well-established composite marker, integrating malnutrition, chronic inflammation, liver synthetic function, and capillary leak, all of which are common in geriatric patients whose diabetic foot disease has reached the point of amputation.

From these three biomarkers, the team built a simple 3-point score using receiver operating characteristic analysis and Youden-index-based thresholds, a method that picks the cutoff value maximizing the sum of sensitivity and specificity. One point was assigned for hemoglobin above 10.3 g/dL, one for a platelet count below 342.5 × 10⁹ per liter, and one for albumin below 28.5 g/L. The stratification that resulted was dramatic. Patients with a score of 2 or higher had an estimated 24-month survival of only 8.0 percent, compared with 66.0 percent among those scoring 0 or 1, a difference that was highly significant on log-rank testing (p < 0.001). The score achieved an area under the curve of 0.798 for predicting 24-month mortality, and the underlying Cox model earned an optimism-corrected concordance index of 0.712 after bootstrap correction, a technique that estimates how much the model’s apparent performance is inflated by being fit to the same data it is evaluated on.

Crucially, the researchers did not stop at building their own score; they benchmarked it against three established prognostic indices that clinicians already use in related contexts. The Prognostic Nutritional Index, which combines albumin and lymphocyte count, achieved an AUC of 0.735. The Geriatric Nutritional Risk Index, built from albumin and weight metrics, reached 0.739. The Hemoglobin–Albumin–Lymphocyte–Platelet score, a composite designed in oncology, managed only 0.590, barely better than chance. The new three-biomarker score showed numerically higher discrimination than all of them, suggesting that the specific combination and thresholds derived from this amputation population capture something the generic indices miss.

The clinical implications, if the findings hold up, are considerable. Major amputation decisions in frail older adults are among the most wrenching in surgery, balancing limb loss against the burden of a procedure that may offer limited life expectancy in return. A preoperative score computed from blood drawn for routine pre-anesthetic workup could help surgeons, anesthesiologists, and families have more honest conversations about goals of care, the intensity of perioperative support, and whether palliative pathways deserve equal consideration. It could also help stratify patients in future trials of post-amputation rehabilitation, wound care, and vascular interventions, where heterogeneous baseline risk often obscures treatment effects.

The authors are careful, appropriately so, about the limits of what they have built. The score is derived and evaluated in the same cohort of 72 patients, a single-center sample that cannot establish how the tool performs on unseen patients. Bootstrap optimism correction was applied only to the fixed three-biomarker Cox model, not to the full score derivation pipeline, so the reported AUC of 0.798 likely overstates true performance to some degree. Quantitative measures of ischemia severity were unavailable, and standardized assessments of neuropathy and frailty, both powerful predictors in geriatric surgical populations, were not incorporated. The researchers explicitly state that the score should be regarded as exploratory, that it should not replace comprehensive clinical and vascular assessment, and that external validation in independent cohorts is required before any clinical use.

Those caveats are worth emphasizing because the temptation to over-read a clean, memorable score is real. A 3-point tool built from three numbers is easy to remember and easy to compute at the bedside, and that simplicity is both its appeal and its danger. The thresholds, hemoglobin above 10.3 g/dL, platelets below 342.5 × 10⁹ per liter, albumin below 28.5 g/L, were tuned to this specific cohort, and cutoffs derived this way often shift when tested elsewhere. Still, the biological coherence of the three signals, spanning oxygen transport, thrombotic consumption, and nutritional reserve, gives the finding a plausibility that purely data-driven models often lack.

What the study ultimately offers is a hypothesis worth testing at scale: that the most common blood panel in the hospital already contains a meaningful forecast of survival after major amputation in older diabetic patients. With more than half of such patients dying within two years, and with existing nutritional and inflammatory indices performing only modestly, even a modest improvement in preoperative risk discrimination could change how these operations are planned and discussed. Until external validation arrives, the score belongs in the research column rather than the clinic, but it points toward a future where the decision to amputate is informed not only by what surgeons see in the wound, but by what the blood has been saying all along.

Subject of Research: Preoperative laboratory-based prediction of 24-month mortality after major lower extremity amputation in geriatric diabetic foot patients

Article Title: Development of an exploratory preoperative laboratory-based score for 24-month mortality after major lower extremity amputation in geriatric diabetic foot patients

Article References: Özer, M., İnce, Y., Korkmaz, T., Talmaç, M. A., & Özer, M. (2026). Development of an exploratory preoperative laboratory-based score for 24-month mortality after major lower extremity amputation in geriatric diabetic foot patients. BMC Geriatrics. https://doi.org/10.1186/s12877-026-08468-6

Image Credits: AI Generated

DOI: 10.1186/s12877-026-08468-6

Keywords: diabetic foot, major amputation, geriatrics, mortality prediction, prognostic score, albumin, hemoglobin, platelet count, Cox regression, nutritional index, external validation, peripheral artery disease