What are clinical scoring systems and why so many of them?
This page is a bundle of nine different clinical scoring tools used in everyday medicine — kidney function (eGFR), atrial-fibrillation stroke and bleeding risk (CHA₂DS₂-VASc and HAS-BLED), consciousness level (GCS), newborn health (APGAR), deep-vein thrombosis probability (Wells), pneumonia severity (CURB-65), cirrhosis severity (Child-Pugh), and liver transplant priority (MELD-Na). Each was developed by regression on real patients to predict an outcome — death, a complication, a diagnosis. They are decision support: they standardize a quick, defensible estimate so that two different clinicians looking at the same patient reach similar conclusions.
Two key vocab distinctions: Risk score (CHA₂DS₂-VASc, HAS-BLED, Wells, CURB-65) = probability of an event given the patient characteristics — used to choose whether to test or treat. Severity / prognostic score (GCS, APGAR, Child-Pugh, MELD-Na) = how sick the patient is right now — used to allocate resources and predict survival. eGFR is somewhere in between: a kidney function estimate that doubles as a chronic-kidney-disease staging tool.
eGFR — how well do the kidneys filter?
GFR (glomerular filtration rate) is the volume of plasma filtered by the kidneys per minute; it is the gold standard for kidney function. Direct measurement is impractical, so the page estimates it from serum creatinine. Three formulas: CKD-EPI 2021 is the modern default and most accurate; it uses a 2-slope curve (steep for low creatinine, flat for high) so it handles both healthy kidneys and kidney failure gracefully. With the sample (male, 60 yr, Cr 1.2 mg/dL), CKD-EPI 2021 returns 69 mL/min/1.73m², placing the patient in KDIGO Stage G2 (Mildly decreased). The full ladder: G1 ≥90, G2 60–89, G3a 45–59, G3b 30–44, G4 15–29, G5 <15 mL/min/1.73m². Cockcroft-Gault (1976) returns a non-normalized creatinine clearance in mL/min (no 1.73m² correction); it is still required for some drug-dosing protocols (vancomycin, aminoglycosides) because the original dosing studies used it. The 2009 equation's race coefficient (Black patients ×1.159) was removed in 2021 as scientifically flawed.
CHA₂DS₂-VASc — should this AF patient be on a blood thinner?
AF pools blood in the left atrium, which can throw clots to the brain and cause stroke. Anticoagulants (warfarin, DOACs like apixaban) prevent most of these strokes but carry their own bleeding risk. The CHA₂DS₂-VASc score helps the decision: age ≥75 (+2), age 65–74 (+1), CHF (+1), hypertension (+1), stroke/TIA (+2), vascular disease (+1), diabetes (+1), female sex (+1). With the sample (70-year-old male, no comorbidities), score = 1 = "Consider oral anticoagulant" (modest annual stroke risk). Score 0 = no antithrombotic; 1 = consider; ≥2 (males) or ≥3 (females) = oral anticoagulation recommended. Note this score is for non-valvular AF; mechanical valves or moderate mitral stenosis need warfarin regardless.
HAS-BLED — and how likely is the blood thinner to cause harm?
HAS-BLED estimates major bleeding risk on anticoagulation — HTN (SBP ≥160), abnormal renal or liver function, stroke, bleeding history, labile INR, age ≥65, antiplatelet/NSAID drugs, alcohol ≥8 drinks/wk — +1 each. With the sample (all No), score = 0 = "Low risk (0.9 bleeds/100 patient-years)". Scores 1–2 are moderate; ≥3 signals a need for closer monitoring or modifying what is modifiable (e.g. swapping an NSAID for acetaminophen).
Glasgow Coma Scale — conscious or not?
The GCS quantifies consciousness using three independent responses: Eye opening (E, 1–4), Verbal (V, 1–5), Motor (M, 1–6) — the panel lists every anchor point. With the sample (E3, V4, M5), total = 12 = Moderate (GCS 9–12) — patient is confused but localizes pain. Total 13–15 is mild (intubation usually not needed), 9–12 is moderate (consider intubation if deteriorating), ≤8 is severe (definitive airway). The Revised Trauma Score and the AVPU scale (Alert/Voice/Pain/Unresponsive) are simpler alternatives.
APGAR — how is the newborn doing?
Virginia Apgar's 1952 score for newborn assessment is reported at 1 and 5 minutes after birth: five signs — Appearance, Pulse, Grimace, Activity, Respiration — each scored 0, 1, or 2 (the panel lists every anchor). With the sample (all 2), total = 10 = "Generally normal (reassuring)". Scores 7–10 are normal, 4–6 moderately depressed (stimulation, oxygen, suction may be needed), ≤3 severely depressed (immediate resuscitation). The 5-minute score predicts mortality better than the 1-minute; <7 at 5 minutes prompts a 10-minute evaluation.
Wells DVT — likely clot in the leg?
The Wells score for deep-vein thrombosis (DVT) sums clinical features — active cancer, immobilization or recent surgery, tenderness, whole-leg or calf swelling (>3 cm vs the other leg), pitting edema, collateral superficial veins, prior DVT, +1 each — minus 2 if an alternative diagnosis is at least as likely. With the sample (all No, alt 0), score = 0 = "Low probability (3% DVT)"; recommendation = "Consider D-dimer; if negative, DVT unlikely." Scores 1–2 are moderate (17% DVT) and ≥3 is high (75% DVT) — above 1, proceed directly to compression ultrasound.
CURB-65 — how sick is this pneumonia patient?
CURB-65 (Lim 2003) gives a quick 30-day mortality estimate for community-acquired pneumonia: Confusion (AMTS ≤8), Urea >7 mmol/L (BUN >19 mg/dL), Respiratory rate ≥30, Blood pressure SBP <90 or DBP ≤60, age ≥65. With the sample (all No), score = 0 = "Low risk (~1.5% 30-day mortality)"; management = "Outpatient management." Score 1 = also low risk (~3%); score 2 = moderate (~9%); score ≥3 = high (~22%, hospitalization recommended; ICU if 4–5). CRB-65 is the same without urea, for outpatient triage when labs are not available. All are decision support, not decisions — a CURB-65 of 0 with hypoxemia, sepsis, or social concerns should still be admitted.
Child-Pugh — how damaged is the liver?
Child-Pugh (1973) predicts surgical and 1-2-year mortality in cirrhosis using 5 inputs — bilirubin, albumin, INR, ascites, encephalopathy — each scored 1 (best) to 3 (worst) by published cutoffs. With the sample (Bili 2.0/2, Alb 3.0/2, INR 1.2/1, Ascites 1, Enceph 1), total = 7 = "Class B (Significant dysfunction)", survival ~80% 1-yr / 60% 2-yr. Class A (≤6) is compensated; Class B (7–9) is significant dysfunction; Class C (≥10) is decompensated, with ~45% 1-yr survival. It is still used to judge whether a patient can tolerate surgery.
MELD-Na — who needs a liver transplant first?
The Model for End-Stage Liver Disease (MELD) is the current U.S. liver-allocation score: it predicts 3-month waitlist mortality from bilirubin, INR, and creatinine. MELD-Na adds sodium because hyponatremia independently predicts waitlist mortality. With the sample (Bili 2.0, INR 1.2, Cr 1.0, Na 138), the calculation: 3.78×ln(2) + 11.2×ln(1.2) + 9.57×ln(1) + 6.43 = MELD = 11. Then for MELD-Na the page first constrains sodium to the [125, 137] window, so an entered Na of 138 becomes 137 in the formula (3-month mortality ~6%); at Na = 125 the correction term adds ~18 points, showing how aggressively hyponatremia drives transplant urgency.
The formula caps and floors: bilirubin, INR, and creatinine are clipped at ≥1 (none of them can be 0); creatinine is capped at 4 (or set to 4 if on dialysis); sodium is clipped to [125, 137]; the final score is clipped to [6, 40]. UNOS policy in the U.S. has moved toward MELD 3.0 (MELD-Na + albumin) since 2023.
Common misconceptions
- eGFR is not a direct measurement of GFR. It is an estimate from creatinine, which is itself a non-specific muscle marker that varies with diet (cooked meat raises creatinine transiently), exercise, and muscle mass (bodybuilders and amputees give misleading values). Cystatin C, when added, improves accuracy in elderly and CKD patients.
- CHA₂DS₂-VASc and HAS-BLED do not tell you to anticoagulate or not. They are pieces of a shared decision. A high CHA₂DS₂-VASc with a high HAS-BLED still favors anticoagulation, because the absolute benefit of preventing stroke usually outweighs the bleeding risk.
- GCS does not measure higher brain function. A patient with GCS 15 can still have severe aphasia, dense hemiplegia, or marked confusion. The FOUR score (Full Outline of UnResponsiveness) is more comprehensive.
- Wells and CURB-65 are not diagnostic. They are pre-test probability and severity scores. A low Wells does not rule out DVT (about 3% of low-probability patients still have one), and a low CURB-65 does not rule out severe pneumonia (a patient with normal BP and age can still be in respiratory failure from hypoxemia alone).
Related tools: Blood & Genetics for the electrolyte and acid-base context behind eGFR and anion-gap logic, Body Metrics for the BSA-based dosing weight used in Cockcroft-Gault, and Reproductive Health for the APGAR context and the gestational-age variables it doesn't capture.
