THERACALC ESTIMATE SELECTION GUIDE
CrCl · eGFR · Cystatin C
Use the estimate the drug label or institutional protocol specifies. When drug labeling is written around Cockcroft–Gault creatinine clearance (CrCl), use CrCl as the primary estimate.
When estimates disagree and the difference crosses a dosing band, choose the more conservative value for:
- Nephrotoxic drugs (vancomycin, aminoglycosides, colistin, cefepime)
- Narrow therapeutic index drugs
- Renally eliminated drugs in frail, elderly, or low-muscle-mass patients
Four common clinical observations about CrCl vs. eGFR discordance, assessed against the evidence:
| Suggestion | Status | Evidence-based assessment |
|---|---|---|
| Frail / low-muscle-mass elderly: eGFRcr reads high, CG reads lower & more conservative | ✅ CORRECT | A frequent and clinically important discordance pattern, especially in geriatric and long-term-care populations. Sarcopenic patients generate less creatinine relative to their true GFR, causing Scr to appear deceptively low and eGFRcr to read falsely optimistic. CG-CrCl often stays lower because body weight (typically low in frail patients) constrains the numerator, partially correcting for the low-creatinine effect — although CG can still overestimate true GFR in this population. Clinically, this discordance commonly underlies q8h vs. q12h cefepime or vancomycin dosing conflicts in frail older adults. |
| Cystatin C lands in the middle — between CG-CrCl and eGFRcr | ⚠️ NUANCED | Cystatin C-based eGFR (eGFRcys) or combined eGFRcr–cys generally moves the estimate downward from creatinine-only eGFRcr when muscle mass is low. However, cystatin C does not reliably land “in the middle” between CG-CrCl and eGFRcr. In frail, sarcopenic patients it commonly falls at or below CrCl — sometimes considerably so — because creatinine-based formulas tend to overestimate GFR. In this population combined eGFRcr–cys is often lower than eGFRcr and can be lower than both eGFRcr and CG-CrCl, reflecting improved correction for muscle-mass effects. Revised framing: cystatin C and eGFRcr–cys reliably pull the estimate down toward (and occasionally beyond) CrCl in sarcopenia; they should not be expected to simply split the difference. |
| Obese elderly (BMI 30–35): less discordance overall; CysC adjusts mildly | ⚠️ PARTLY CORRECT | Discordance in obese elderly patients is more variable, not simply smaller. Direction of disagreement depends heavily on: (a) which body weight is used in CG (actual vs. ideal vs. adjusted), and (b) whether eGFRcr is de-indexed for body surface area when comparing values in mL/min. Using actual body weight in CG for an obese patient can push CrCl higher than de-indexed eGFRcr, creating discordance opposite to the frail pattern. Key practical point: institutional policy should specify the CG weight choice; using adjusted body weight (IBW + 0.4 × [ABW − IBW]) in CG for obesity is generally preferred to prevent spuriously high CrCl from driving inappropriate dose escalation. |
| At BMI >38: CrCl sometimes reads higher than eGFRcr | ✅ CORRECT | Correct, and more pronounced when actual body weight is used in the CG formula. The weight term in CG amplifies the clearance estimate in proportion to body size, so in marked obesity CG-CrCl can exceed even de-indexed eGFRcr. Action: never use actual body weight in CG for obese patients for the purpose of dose escalation. Recalculate with adjusted weight per institutional standard, or use eGFRcr as a cross-check before increasing dose. |
| Estimate | What it measures | Key strengths | Key limitations | Best use case |
|---|---|---|---|---|
| Cockcroft–Gault CrCl | Estimated drug clearance (mL/min) | Matches most PK studies & package inserts. Used in the majority of dose-adjustment RCTs. | Sensitive to weight choice (actual vs. ideal vs. adjusted). Low muscle mass lowers Scr and can still allow overestimation. | Default when the drug label cites CrCl. Preferred in frail/elderly when cystatin C is unavailable. |
| eGFRcr (CKD-EPI 2021) | Kidney function indexed to 1.73 m² BSA | Auto-reported in most contemporary EHRs. Standardized and recommended for CKD staging and monitoring. | Indexed units complicate direct comparison with CrCl; requires de-indexing (× BSA/1.73) for mL/min. Overestimates GFR in sarcopenia/frailty and some extremes of body size. | When the drug label cites eGFR, or when the patient has average body size and estimates agree. |
| CysC eGFR / combined eGFRcr–cys | GFR estimate less dependent on muscle mass | Best accuracy overall. Less biased in frail, sarcopenic patients. Best tie-breaker. | Not routinely available in all labs; affected by inflammation, steroids, thyroid disease, and some transplant populations; local validation is important. | Order when CrCl vs. eGFRcr disagree in frail/elderly patients. Optimal tie-breaker for high-risk drug dosing. |
- Check the drug reference first. If labeling is written around CG-CrCl, use CrCl as the primary estimate. Most renal dosing evidence and historical PK trials are CG-based; follow label and institutional policy where specified.
- Calculate CG-CrCl correctly. In obese patients, use adjusted body weight (IBW + 0.4 × [ABW − IBW]) unless your institutional policy specifies otherwise. Avoid using actual weight for dose escalation in obesity to prevent inflated CrCl.
- Compare CG-CrCl with eGFRcr. To compare directly, de-index eGFRcr by multiplying by BSA ÷ 1.73 to convert to mL/min, using CKD-EPI 2021 creatinine as the reference equation. If both estimates fall in the same dosing band → use the drug-label metric.
- If estimates cross a dosing band threshold: ask whether the drug is high-risk — narrow therapeutic index, nephrotoxic, or renally cleared. For these agents, favor the lower estimate for initial dosing in frail/elderly patients.
- If cystatin C is available: use combined eGFRcr–cys (CKD-EPI) as the tie-breaker. It generally pulls the estimate down from eGFRcr toward (or below) CrCl and is more accurate in frail patients. Account for non-GFR influences (inflammation, steroids, thyroid disease).
- In non–steady-state AKI or recovery: do not rely on any single formula. Use serial creatinine trends, urine output, and therapeutic drug monitoring. Treat any GFR estimate as a rough starting point rather than a definitive value in AKI or recovery.
| Clinical scenario | Expected pattern | Practical choice |
|---|---|---|
| Average-size adult — stable creatinine, usual muscle mass, no extremes of weight or age | CG-CrCl and de-indexed CKD-EPI 2021 eGFRcr are usually within 10–15 mL/min of each other and fall in the same dosing band for most drugs. | Use the metric specified by the drug label or institutional protocol. Clinical impact of the difference is minimal. ✓ Either estimate acceptable for most agents. |
| Older adult without marked sarcopenia or obesity — age >65, creatinine stable | Estimates often differ only modestly. After de-indexing eGFRcr for comparison, numbers are usually close enough to agree on dose category. | Either estimate is acceptable for lower-risk drugs. For high-risk agents (vancomycin, aminoglycosides, cefepime) use the lower estimate when bands differ. |
| Obese elderly with CG calculated using adjusted body weight — ABW used correctly; BMI 30–38 | When adjusted weight is used in CG, CrCl and de-indexed eGFRcr are often fairly close — much more so than when actual weight is used. | Estimates are comparable when weight input is correct. Do NOT use actual body weight in CG for obesity — the most common source of artificially inflated CrCl. |
| Scenario | CrCl vs. eGFRcr pattern | Which estimate to use | How cystatin C changes the picture |
|---|---|---|---|
| MOST COMMON Frail / low-muscle-mass elderly — sarcopenia, low Scr, low body weight (classic geriatric patient) |
eGFRcr: reads too HIGH — Scr artificially low from low muscle mass → falsely optimistic GFR. CG-CrCl: reads LOWER — body weight constrains estimate, partially correcting for low Scr. | Use CG-CrCl as the safer default when cystatin C is unavailable, with awareness that CG can still overestimate. Order cystatin C when the difference crosses a dosing band and will influence dosing (e.g., q8h vs. q12h cefepime). | CysC or combined eGFRcr–cys typically pulls the estimate DOWN from eGFRcr — often at or below CrCl, not just “in the middle.” It frequently validates the conservative CrCl-based dose or supports going lower, and rarely supports the more aggressive eGFRcr-based dose in this population. |
| VARIABLE Obese elderly (BMI 30–35) — age >60, moderate obesity |
Direction is VARIABLE — depends on weight used in CG and whether eGFRcr is de-indexed: adjusted BW → CrCl often close to de-indexed eGFRcr; actual BW → CrCl may read HIGHER than eGFRcr. | If CG used adjusted weight → estimates are often comparable. If actual weight was used → do NOT use that CrCl for dose escalation; recalculate with adjusted weight and cross-check against de-indexed eGFRcr. | CysC typically lowers the estimate modestly versus eGFRcr, but interpretation is less clear-cut than in frailty. Higher fat mass can raise CysC production, making results less predictable than in sarcopenic patients. |
| HIGH RISK Marked obesity (BMI >38), especially if actual body weight used in CG |
CG-CrCl: may read significantly HIGHER than eGFRcr when actual weight is used — opposite of the frailty pattern. eGFRcr may be the more conservative value here. | Avoid reflexively using actual-weight CrCl for dose escalation. Recalculate CG with adjusted weight; if adjusted-weight CrCl remains higher than de-indexed eGFRcr, cross-check eGFRcr before increasing dose. | CysC often reveals that actual-weight CrCl is too optimistic, pulling the estimate toward a more realistic value. Useful when high-risk dosing decisions hinge on whether CG-CrCl is reliable. |
| UNDER-DOSING Very muscular patient — bodybuilder, athlete, or high muscle mass for age |
eGFRcr: may read TOO LOW — high muscle mass → high Scr production → Scr elevated relative to true GFR. CrCl may be more appropriate when under-dosing is the concern. | CG-CrCl is often more appropriate when the concern is under-dosing (e.g., subtherapeutic antibiotic exposure). De-index eGFRcr to the patient’s actual BSA before deciding; cystatin C or eGFRcr–cys often supports more standard or aggressive dosing. | CysC typically produces a HIGHER estimate than eGFRcr in muscular patients — it is not affected by high muscle mass, so it correctly indicates better true GFR and may support more standard or aggressive dosing. |
| MOST UNRELIABLE Acute kidney injury or active recovery — rapidly rising or falling Scr |
Any single-point estimate is unreliable — creatinine has not equilibrated with true GFR. Rising Scr: true GFR is LOWER than the formula suggests. Falling Scr (recovery): true GFR may be HIGHER than the formula suggests. | No single formula is reliable. Drive dosing with serial creatinine trends, urine output, and drug levels (AUC monitoring for vancomycin; aminoglycoside levels). Use empiric dose reduction for renally eliminated drugs until the trajectory is clearer. | CysC responds faster than creatinine to GFR changes (1–2 day lag vs. 3–5 days for Scr), so it can add information in rapidly changing function. It does NOT fully solve non–steady-state problems — it still lags and is affected by inflammation and steroids common in AKI. |
Example 1 — Frail elderly: cefepime dosing FAVOR CG
Presentation: 80-year-old male, Scr 1.0, weight 55 kg (low muscle mass), height ≈168 cm. eGFRcr (CKD-EPI 2021) = 76 mL/min/1.73m² (≈70 de-indexed). CG-CrCl = 46 mL/min. Sepsis, starting cefepime.
eGFRcr-based dosing → 2 g q8h · CG-CrCl-based dosing → 2 g q12h.
Decision: Low weight and sarcopenia inflate eGFRcr; even a “normal” Scr of 1.0 understates the muscle-loss effect, so eGFRcr is the optimistic outlier. CG-CrCl (46) crosses cefepime’s renal threshold (<60) into the q12h band. Default to CG-CrCl → 2 g q12h. Order cystatin C; combined eGFRcr–cys will likely confirm ~45–55 mL/min.
Bottom line: q8h here risks accumulation-related neurotoxicity. CG-CrCl is the safer starting point.
Example 2 — Obese elderly: weight choice matters CHECK WEIGHT
Presentation: 67-year-old female, Scr 1.0, weight 102 kg, height 165 cm. IBW ≈57 kg, adjusted BW ≈75 kg. CG actual weight = 88 mL/min. CG adjusted weight = 65 mL/min. eGFRcr (CKD-EPI 2021) = 62 mL/min/1.73m² (≈77 de-indexed; BSA 2.16).
Decision: Do NOT use the actual-weight CG of 88 — it is inflated by the weight term. Adjusted-weight CG (65) and indexed eGFRcr (62) now agree closely; the actual-weight value is the lone outlier. Note that de-indexing eGFRcr for this large-BSA patient raises it to ~77, so for a nephrotoxic or narrow-TI agent favor the lower of adjusted-weight CG (65) and indexed eGFRcr (62). Order cystatin C if the dose category hinges on it.
Bottom line: in obesity, actual-weight CG is often the optimistic outlier; adjusted weight and indexed eGFRcr should anchor dosing decisions.
Example 3 — Vancomycin: no formula is sufficient MONITOR LEVELS
Principle: For vancomycin, any estimating formula is only a starting point for the initial dose. AUC-guided monitoring (target AUC/MIC 400–600 for MRSA) matters more than forcing one formula to be “correct.”
Initial load: use the CG-based estimate for the first dose; in frail patients err toward the lower interval. Maintenance: AUC-guided adjustment using Bayesian software — do not anchor on the initial formula-based estimate past the first dose.
Cystatin C can help refine the initial estimate in frail patients, but drug-level feedback is always superior to any formula for ongoing dosing.
Example 4 — Muscular patient: under-dosing risk FAVOR CrCl / CysC
Presentation: 32-year-old male, weight 100 kg, height 183 cm, high muscle mass, Scr 1.4 (elevated from muscle, not kidney injury). eGFRcr (CKD-EPI 2021) = 68 mL/min/1.73m² (≈89 de-indexed). CG-CrCl = 107 mL/min. Neutropenic fever, starting cefepime.
Decision: High muscle mass raises creatinine production, so Scr 1.4 overstates renal impairment and eGFRcr reads falsely LOW (68). Trusting it would push toward reduced dosing and subtherapeutic β-lactam exposure in a patient who needs aggressive coverage. CG (107) and de-indexed eGFRcr (89) both indicate normal GFR; cystatin C — independent of muscle — would read higher and confirm. Use the full dose (2 g q8h); do not reduce.
Bottom line: in the muscular patient, eGFRcr is the pessimistic outlier — under-dosing is the main danger, so CG and cystatin C should guide toward full dosing.
Example 5 — AKI / non–steady state: the formula is a snapshot TREND + LEVELS
Presentation: 70-year-old male, 80 kg, septic, on vancomycin + cefepime. Scr rising 0.9 (day 0) → 1.8 (24h) → 2.6 (48h); urine output falling.
The trap: plugging Scr 2.6 into CG gives ~30 mL/min, which looks like stable stage-4 CKD. It is not. Because Scr is still climbing, it lags true GFR — actual clearance right now is LOWER than 30. Dosing as if CrCl were a stable 30 will under-correct and risk accumulation.
Decision: Do not anchor on any single-point estimate. Reduce or hold renally cleared drugs, dose to the trend (serial Scr, urine output) and to drug levels (vancomycin AUC; consider holding aminoglycosides). Cystatin C responds faster (1–2 day lag vs. 3–5 for Scr) and can flag the change earlier, but does not fully resolve non–steady state.
Bottom line: in flux, the formula reflects the past. Trajectory plus levels drive dosing.
When Cockcroft–Gault CrCl and CKD-EPI eGFR (using current race-free 2021 creatinine and creatinine–cystatin C equations where available) disagree for medication dosing, apply the estimate specified by drug labeling or institutional protocol. If estimates fall into different dosing bands and the medication carries meaningful toxicity risk, favor the more conservative estimate — particularly in frail older adults, low-muscle-mass patients, and those receiving nephrotoxic or narrow-therapeutic-index agents. When cystatin C is available and validated in the local laboratory, use the combined eGFRcr–cys equation as the preferred adjudicator: it improves accuracy and often resolves falsely elevated creatinine-only estimates in sarcopenic patients. In acute kidney injury or non–steady-state renal function, no single formula is reliable; use serial creatinine trends, urine output, and therapeutic drug monitoring to guide dosing decisions, and treat any GFR estimate as a rough starting point rather than a definitive value.