My Blood Test Says I Have High Creatinine, Do I Have Kidney Disease - Dr. Taylor Martin

Dr. Taylor Martin explains that many clinicians see a creatinine outside the reference range, calculate an eGFR below 60, and immediately diagnose chronic kidney disease. However, reference ranges are built on bell curves of the general population, and active individuals who supplement with creatine or carry substantial muscle mass frequently fall above those curves. He notes that even when a patient is not on creatine, a high-meat meal or dehydration can transiently bump creatinine and BUN. The eGFR equation itself uses creatinine as a proxy; when creatinine is artificially elevated, eGFR is falsely lowered. Martin stresses the importance of telling your doctor about all supplements and of repeating labs while well‑hydrated before jumping to nephrology consultation.

Thomas DeLauer shares that he once had CK levels in the thousands after a long run without much thought, and it was alarming. Dr. Martin confirms that CK commonly rises above 1,000 after high‑intensity or long‑duration exercise, and even sauna use can drive it up. He explains that the clinical concern is rhabdomyolysis, where massive muscle breakdown floods the kidneys with myoglobin, leading to kidney damage. The red‑flag symptoms are 'Coca‑Cola‑colored urine' and severe, diffuse muscle pain beyond ordinary soreness. If those are absent, the CK elevation is a normal response to training. Thomas DeLauer adds his own rhabdomyolysis story — he ended up at urgent care but not inpatient — and notes his sister, an Ironman trainee, has been hospitalized multiple times, suggesting a possible genetic predisposition.

Thomas DeLauer describes his shock when his usual near‑zero CRP jumped to 12 after a viral illness and a 7‑mile morning run. His doctor, Kyle, reassured him it was expected. Dr. Martin explains that CRP is associated with endothelial inflammation and heart disease when chronically elevated, but any acute‑phase inflammation — from rolling an ankle to a bout of flu — can spike it. He also notes that excess body fat can drive CRP up chronically. The key is repeat testing when the acute stressor has resolved; if the number falls back to normal, it was a transient response. This discussion extends to ferritin and homocysteine, which can also be acute‑phase reactants and cause false alarms.

Dr. Martin explains that a patient who eats a iron‑rich meal the night before a blood draw can show high serum iron, while someone with true iron deficiency may have a perfectly normal serum iron. He emphasizes that an astute clinician examines the constellation: ferritin (iron stores), transferrin and saturation (transport capacity), and hemoglobin (functional status). He also mentions that altitude, viral infections, and hemochromatosis can affect iron levels. Martin warns that missing hemochromatosis is dangerous because excess iron slowly damages the liver, reflected in rising ALT and AST over time. Therefore, he advises trending iron and liver markers together.

Dr. Martin notes that doctors may tolerate a doubling of the upper reference limit if it’s linked to a known cause like statin use, as long as there are no symptoms. He stresses the importance of consistent labs over time to see the trajectory. Many alarming spikes are due to a recent viral illness or a night of drinking. Thomas DeLauer adds that seeing an out‑of‑range ALT can terrify patients, but often it’s just a transient bump. The discussion connects back to iron overload: if ALT/AST rise in tandem with high ferritin and transferrin saturation, hemochromatosis must be ruled out.

Dr. Martin admits he cannot recall the exact biochemical pathway off the top of his head but explains that fructosamine is an enzyme‑based marker used to confirm whether an elevated A1c truly indicates insulin resistance or is a false positive. He prefers to pair fructosamine with fasting insulin and glucose to calculate HOMA-IR, because insulin resistance is the underlying process that predates A1c elevation by years. This multi‑marker approach catches metabolic dysfunction early.

Martin describes a spectrum of AI use: on the simplest end, patients paste labs into ChatGPT or Claude and get summaries based on public information. More advanced are fine-tuned models trained on medical literature and refined by expert clinicians, which can flag patterns a busy doctor might miss. He mentions clustering models that can group patients by phenotype — for instance, spotting that a person’s lab pattern resembles a diabetic profile and prompting an A1C check that might not have been ordered. Beyond diagnosis, AI is being deployed for prior authorizations, formulary checking, and drafting patient portal responses, which currently burden doctors with hours of phone calls and message backlogs. However, Martin notes that physician adoption is mixed; a UCLA preprint showed doctors saved about 15 minutes daily but discontinued use after three months, suggesting more work is needed on workflow integration. Liability also anchors the human doctor: even when using AI, the clinician is ultimately responsible for the prescription or decision, so fully autonomous AI prescribing has little support. Thomas DeLauer adds his perspective from healthcare administration, noting that AI could reduce the top-of-funnel load, letting doctors spend more time on complex cases.

Martin explains that when red blood cell turnover is high, the A1c reflects a shorter time window and can appear elevated even if average glucose is normal. To confirm, he uses fructosamine, a marker that indicates average blood glucose over the prior 2-3 weeks and is not influenced by red cell lifespan. He also advocates pairing insulin with glucose to calculate HOMA-IR, which more directly captures insulin resistance. This constellation — fasting glucose, insulin, A1c, and fructosamine if needed — gives a far clearer picture than A1c alone. Thomas DeLauer asks if insulin testing muddies the water; Martin argues it’s essential to see the underlying physiology, because insulin resistance can start long before A1c moves.

Martin explains that the HOMA-IR score uses fasting insulin and glucose to estimate insulin resistance; a value above about 2 suggests trouble. He acknowledges that many physicians don’t routinely order fasting insulin, so the problem is often missed. AI tools that can look at the entire panel — lipids, liver enzymes, CRP, ferritin — may identify phenotypic patterns that map to insulin resistance, prompting the clinician to test accordingly. This could catch the patient 5–10 years before an A1c diagnosis.