Pair eGFR with albuminuria, then frame uncertain decisions for transparent evidence appraisal.
Assess eGFR alongside urine albumin-to-creatinine ratio; persistent albuminuria for at least three months can establish chronic kidney disease despite preserved filtration.
Chronic kidney disease can be present before estimated glomerular filtration rate (eGFR) falls. Persistent albuminuria is a marker of kidney damage and can establish the diagnosis when it remains present for at least three months, even with preserved filtration. A single urine albumin concentration is vulnerable to changes in hydration, so interpret albumin using the urine albumin-to-creatinine ratio and review earlier results before applying a chronic label.
Kidney risk is two-dimensional
eGFR and albuminuria answer different but complementary questions. Lower eGFR and higher albuminuria together identify increasing risks of kidney replacement therapy, cardiovascular events, hospitalisation and death. Cardiovascular risk may dominate prognosis long before dialysis is required, so assessment should not stop at filtration or future kidney replacement.
The mechanism also matters. Loss of nephrons can remain clinically hidden while physiological reserve and compensatory hyperfiltration preserve measured filtration. Remaining nephrons enlarge and increase their workload, but this adaptation can become maladaptive. Diabetes can increase proximal glucose and sodium handling and contribute to early hyperfiltration before eGFR falls. The workload–capacity model therefore directs attention to diabetes, obesity, hypertension, heart failure, previous acute kidney injury and nephrotoxic medicines, rather than treating kidney disease as an isolated laboratory problem.
Management review should consider blood-pressure control, diabetes treatment, renin–angiotensin system blockade, sodium–glucose cotransporter 2 inhibitors when clinically suitable, and nephrotoxin exposure.
Make uncertainty explicit
The second clinical thread concerns how decisions are made when evidence is incomplete. Evidence-based practice combines the best available research with clinical expertise and patient values; it is not a requirement that every decision rests on a randomised trial or a guideline.
Start by defining the question. Treatment, diagnostic, prognostic and harm questions may require different evidence designs. Randomisation strengthens a treatment comparison by balancing known and unknown prognostic factors on average, reducing confounding, but it does not guarantee adherence, perfect measurement or applicability to every patient. Systematic reviews remain limited by the completeness and quality of their underlying data, and regulatory clinical study reports may contain important detail missing from journal publications.
Uncertainty should prompt deliberate reasoning. When a presentation does not fit the initial pattern, slow down, move from pattern recognition to analytical thinking, observe the trajectory and reassess. Guideline recommendations can distinguish established benefit, established harm and genuine uncertainty. Artificial intelligence can support analysis, but confident outputs still require verification.
Preserved eGFR does not exclude chronic kidney disease when albuminuria or another structural marker persists for at least three months. The discussion links nephron workload, hyperfiltration, cardiovascular risk and treatment aimed at protecting remaining kidney function.
Evidence-based medicine integrates research, clinical expertise and patient values rather than demanding obedience to trials or guidelines. Diagnostic uncertainty, bias, incomplete reporting, randomisation and careful use of artificial intelligence shape the practical appraisal of clinical decisions.
Before labelling chronic kidney disease, review previous results and confirm that a structural or functional abnormality has persisted for at least three months. Record eGFR and urine albumin-to-creatinine ratio together, then identify modifiable workload and capacity factors, including diabetes, hypertension, obesity, previous acute kidney injury and nephrotoxic medicines. For an uncertain decision, define whether the question concerns treatment, diagnosis, prognosis or harm before searching for evidence. Reassess when the patient’s trajectory does not match the initial pattern, and state clearly whether the available evidence indicates benefit, harm or continuing uncertainty.
An adult has an eGFR of 96 mL/min/1.73 m², but the urine albumin-to-creatinine ratio remains raised on samples taken four months apart. Does the preserved eGFR exclude chronic kidney disease?
No. Persistent albuminuria for more than three months is a marker of kidney damage and can establish chronic kidney disease despite preserved eGFR.
A randomised trial compares two treatments in patients with differing baseline prognoses. What protection does random allocation mainly provide?
Randomisation balances known and unknown prognostic factors between groups on average, reducing confounding. It does not guarantee adherence or perfect measurement.
A patient’s presentation does not fit the expected clinical pattern, and the diagnosis remains uncertain. What reasoning approach should follow?
Slow down and move from rapid pattern recognition to deliberate analytical reasoning. Observe the clinical trajectory and reassess rather than forcing an immediate precise diagnosis.
.svg)