Creatinine Rise on Vancomycin Plus Pip-Tazo: Query Type, Present-on-Admission Status, and the Adverse-Event Question
A 69-year-old woman is admitted with urosepsis after two days of dysuria and fever. Empiric therapy is vancomycin plus piperacillin-tazobactam. Admission creatinine is 0.9 mg/dL. On day 4 the creatinine is 2.1 mg/dL, urine output has fallen to 20 mL/hr, BUN is 48 mg/dL, and urinalysis shows muddy brown casts. Admission blood pressure was 84/50 and responded to 2 L of crystalloid. The hospitalist documents 'AKI, likely prerenal from sepsis' and continues the antibiotics. No dialysis is required. It is day 6, discharge is planned tomorrow, and the case manager asks the CDI specialist whether the AKI should be captured as present on admission.
1. Do the clinical indicators fit the documented type of acute kidney injury? Should you query, and for what?
Read the worked answer
The documented etiology and the data are in tension. Indicators favoring a prerenal picture: hypotension at admission and a BUN:creatinine ratio of roughly 23. Indicators pointing the other way: no recovery of creatinine after 2 L of crystalloid over several days, persistent oliguria at 20 mL/hr, and muddy brown casts, which are a classic marker of intrinsic tubular injury. Two nephrotoxic agents are running concurrently.
That is a clarification opportunity, not a coding decision. The CDI cannot substitute acute tubular necrosis for the physician's diagnosis. A compliant query: 'Documentation supports acute kidney injury (creatinine rose from 0.9 to 2.1 mg/dL, urine output 20 mL/hr, BUN 48, muddy brown casts on urinalysis) in the setting of vancomycin and piperacillin-tazobactam. Please clarify the type of acute kidney injury: [ ] Acute kidney failure with tubular necrosis [ ] Prerenal [ ] Drug-induced/toxic nephropathy [ ] Other: ____ [ ] Unable to determine.'
Keep the payment reality straight: ATN (N17.0) and unspecified acute kidney failure (N17.9) are both CC-level conditions, so this query is not a DRG play. It matters for clinical specificity, registry and quality data, and medication-management review. Do not oversell it to the physician or to leadership.
2. What is the correct present-on-admission determination, and how should the CDI answer the case manager?
Read the worked answer
The creatinine was 0.9 on admission and rose to 2.1 by day 4. On the evidence in the record, the acute kidney injury developed after admission, so the POA indicator is N. The CDI should say so plainly and should not be moved by arguments that a Y would 'protect' the metrics.
A POA query is appropriate only where the record genuinely supports ambiguity, for example a creatinine already trending upward on the admission panel. Here the trajectory is not ambiguous. If a query is sent, it must be framed around onset, not around the indicator value: 'Please clarify whether the acute kidney injury was present at the time of admission, based on the admission creatinine of 0.9 mg/dL on [date].'
Explain the stakes accurately: POA indicators are reported to CMS and flow into quality and hospital-acquired condition programs, public reporting, and risk adjustment for Medicare Advantage and some commercial contracts. MS-DRG payment itself is not directly changed by a POA flag. That is precisely why flipping the flag is attractive to a program under pressure, and precisely why it is not permissible.
3. The physician later volunteers that the vancomycin 'probably caused' the kidney injury. Should the CDI pursue documentation of a drug-induced etiology, and what if the pharmacy or quality department would rather it not appear?
Read the worked answer
Pursue it only if the physician's documentation is ambiguous; do not suppress it under any circumstances.
If the physician documents that the injury is drug-induced, the record should reflect an adverse effect: the drug code from the T36–T50 range with the adverse-effect character, plus the acute kidney injury. Leaving it out understates a real clinical event and misrepresents the care provided. A quality or pharmacy concern about a patient-safety flag is not a legitimate reason to omit it; documentation integrity does not bend to dashboard optics.
Equally, the CDI must not manufacture the link. If the physician says 'probably,' that is an uncertain etiology in the inpatient setting; ask a clarifying question that offers a genuine out ('unable to determine' plus 'other'), and accept the answer. If the physician says the cause is unclear, code the kidney injury without an adverse-effect code.
The right internal answer to pharmacy or quality is: 'If the event happened, the record should say so and our improvement work should start from the truth. We will not distort documentation in either direction.'