How to Determine Baseline Creatinine with Multiple Serial Values
When multiple serial creatinine values are available, use the mean of all outpatient creatinine values from 7-365 days before the current assessment, as this method shows the highest correlation (0.91) with expert-adjudicated baseline values. 1
Recommended Approach: Hierarchical Algorithm
First Choice: Mean Outpatient Value (7-365 Days)
- Calculate the arithmetic mean of all outpatient creatinine measurements obtained between 7 days and 365 days before the current value
- This method demonstrated the highest intra-class correlation coefficient (0.91) with nephrologist-adjudicated baseline values 1
- Available in approximately 81% of hospitalized patients 1
- Key advantage: Smooths out normal biological variability (14-17% reference change value) and avoids values influenced by acute illness prodrome 1
Second Choice: Most Recent Inpatient or Outpatient Value (7-365 Days)
- Use if calculating a mean is impractical
- Correlation of 0.88 with adjudicated baseline 1
- Higher availability (93% of patients) makes this pragmatic for automated systems 1
- Allows inclusion of previous inpatient stays if >7 days away
Third Choice: Most Recent Outpatient Value Alone (7-365 Days)
- Moderately lower correlation (0.84) but still acceptable 1
- Simpler to implement in clinical practice
Methods to AVOID
Do NOT Use Admission Creatinine
- Higher than true outpatient baseline, leading to missed community-acquired AKI 1
- Reduces AKI detection sensitivity
Do NOT Use Minimum Inpatient Value (First 7 Days)
- Lower than true outpatient baseline 1
- Overestimates AKI prevalence and severity
- May represent the nadir of AKI itself, not true baseline
- Despite KDIGO's earlier suggestion of using lowest hospitalization creatinine, this was before publication of validation data showing its limitations 1
Do NOT Use Values Within 7 Days of Current Assessment
- Risk of using creatinine already elevated by illness prodrome 1
- Underestimates extent of AKI
- Only 29% of patients have values this recent 1
Use Imputation ONLY as Last Resort
- Back-calculation using MDRD equation (assuming GFR 75 ml/min/1.73m²) is prone to error 1
- Underestimates baseline creatinine and overestimates GFR, particularly in CKD patients 1
- Recent research suggests using population mean GFR or GFR of 100 ml/min may be more accurate than 75 ml/min 2
Important Caveats
Time Window Considerations
- Avoid going beyond 365 days when possible—older values have lower correlation with true baseline as kidney function may have changed 1
- Extended look-back (7-730 days) improves availability but reduces correlation with true baseline 1
Distinguish Outpatient vs. Inpatient Values
- Outpatient values are superior to inpatient values for baseline determination 1
- Inpatient creatinine values may already reflect acute illness
Account for Biological Variability
- Normal creatinine variability (reference change value) is 14-17% 1
- In CKD patients with elevated baseline creatinine, small absolute rises represent progressively smaller percentage increases, making them more likely to reflect random variation 1
Special Population: CKD Patients
- Creatinine levels drift over time in CKD patients 3
- Consider adjusting the most recent creatinine value for time elapsed and sex to reduce false-positive AKI detection 3
- Standard back-estimation equations perform poorly in CKD populations 2, 3
Practical Implementation
When reviewing a patient's chart with multiple values:
- Identify all outpatient creatinine values from 7-365 days before current assessment
- Calculate the mean of these values
- If no outpatient values exist, use the most recent value (inpatient or outpatient) from this window
- Only resort to imputation if absolutely no historical values are available
- Document your methodology clearly in the medical record, as different baseline definitions significantly affect AKI classification and mortality prediction 4, 5
The choice of baseline method has substantial clinical implications—studies show that lack of proper baseline creatinine leads to smaller AKI cohorts with higher mortality (23% vs. 17-18%) due to missed diagnoses 4.