Baseline Creatinine

SELECT
    hadm_id,
    gender,
    age,
    scr_min,      -- Lowest creatinine during admission
    ckd,          -- 1 if CKD diagnosis present
    mdrd_est,     -- Estimated creatinine from MDRD
    scr_baseline  -- Final baseline determination
FROM mimiciv_derived.creatinine_baseline;

MDRD Estimation Formula

For patients without normal creatinine and without CKD, baseline is estimated:

Male patients:

scr_baseline = (75 / 186 / age^(-0.203))^(-1/1.154)

Female patients:

scr_baseline = (75 / 186 / age^(-0.203) / 0.742)^(-1/1.154)

This back-calculates creatinine assuming eGFR = 75 mL/min/1.73m^2 (lower limit of normal).

CKD Identification

CKD is identified from ICD codes:

• ICD-9: 585 (Chronic kidney disease)
• ICD-10: N18 (Chronic kidney disease)

Critical Implementation Notes

1. Adults Only: Query filters to age >= 18 (pediatric creatinine norms differ).

2. MDRD Limitations:

   • Less accurate in elderly, extremes of body size, or certain ethnicities
   • Assumes GFR = 75, which may underestimate for young healthy patients

3. Admission Bias: Using admission creatinine as baseline may underestimate for patients admitted already in AKI (AKI-on-admission).

4. CKD May Be Coded Late: ICD codes are assigned at discharge, so this technically uses future information. In most research this is acceptable.

5. Missing Values: If no creatinine measured during admission, baseline will be NULL.

6. Race Coefficient: The original MDRD had a race coefficient; this implementation does not use it, consistent with recent guidelines removing race from eGFR calculations.

Example: Baseline Distribution

SELECT
    CASE
        WHEN scr_min <= 1.1 THEN 'Normal admission Cr'
        WHEN ckd = 1 THEN 'CKD (using min)'
        ELSE 'MDRD estimated'
    END AS baseline_source,
    COUNT(*) AS n_admissions,
    ROUND(AVG(scr_baseline), 2) AS avg_baseline
FROM mimiciv_derived.creatinine_baseline
GROUP BY 1;

Example: Compare Baseline vs Admission Creatinine

SELECT
    cb.hadm_id,
    cb.scr_baseline,
    cb.scr_min AS admission_min_cr,
    ROUND(cb.scr_min / NULLIF(cb.scr_baseline, 0), 2) AS cr_ratio,
    CASE
        WHEN cb.scr_min >= cb.scr_baseline * 1.5 THEN 'AKI Stage 1+'
        ELSE 'No AKI at admission'
    END AS admission_aki_status
FROM mimiciv_derived.creatinine_baseline cb
WHERE cb.scr_baseline IS NOT NULL;

Example: Age-Stratified Estimated Baselines

SELECT
    CASE
        WHEN age < 40 THEN '<40'
        WHEN age < 60 THEN '40-59'
        WHEN age < 80 THEN '60-79'
        ELSE '80+'
    END AS age_group,
    gender,
    ROUND(AVG(mdrd_est), 2) AS avg_mdrd_baseline
FROM mimiciv_derived.creatinine_baseline
WHERE ckd = 0 AND scr_min > 1.1
GROUP BY 1, 2
ORDER BY 1, 2;

Alternative Baseline Methods

Other approaches used in literature (not implemented here):

1. Outpatient creatinine: Use pre-admission ambulatory values (requires linked outpatient data)
2. Rolling minimum: Lowest value in past 7-365 days
3. First available: First creatinine of admission (problematic if AKI present)
4. Fixed MDRD: Always use MDRD with assumed GFR (consistent but may miss true normal)

References

• KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney International Supplements. 2012.
• Siew ED et al. "Estimating baseline kidney function in hospitalized patients with impaired kidney function." Clinical Journal of the American Society of Nephrology. 2012;7(5):712-719.