In Reply: —We thank Dr. Chappell et al. for their interest in our article and insightful review. They first question the abstract’s conclusion that “The use of vasopressor and diuretics is also associated with acute renal failure,” and specifically highlight concern regarding the applicability of the conclusion to “healthy individuals.”1Dr. Chappell et al. seem to be using the concepts of “healthy patients” and “patients with previously normal renal function” interchangeably. We agree that the patient population we examined does not represent “healthy patients,” as demonstrated by the variety of comorbidities affecting the patients and delineated in table 1 of the original article.1However, the data are explicitly based on patients with normal preoperative renal function, given that we excluded patients with preexisting renal dysfunction or the failure to demonstrate a preoperative estimated creatinine clearance of 80 ml/min or greater. As we stated in the limitations section, although these criteria do exclude patients with renal dysfunction, they probably also exclude healthy patients who did not warrant preoperative serum creatinine testing. Our article does not attempt to make commentary regarding “healthy patients,” simply those with normal renal function.
Next, Dr. Chappell et al. suggest that it is inappropriate to claim an association between vasopressor or diuretic administration and acute renal failure (ARF) because of their analysis of the details in table 5 of the original article. They compared the 0.8% and 1.5% ARF incidences experienced by patients receiving vasopressor and diuretics, respectively, to existing literature documenting an ARF rate of 1–5% for hospitalized patients.2–4They conclude that an association cannot exist because 0.8% and 1.5% are less than the 1–5% incidence described in epidemiologic studies.2–4We disagree with this interpretation. The quoted 1–5% incidence is for an entirely different patient population: all hospitalized patients, without regard for their reason for admission. The 1–5% incidence presumably also includes patients at very high risk for ARF: urologic surgery patients, cardiac surgery patients, and patients receiving intravenous contrast postoperatively. Most importantly, that population and literature include patients with preexisting renal dysfunction, a group well known to be at high risk for postoperative ARF. A careful review of the original article’s table 5 demonstrates that among low-, medium-, and medium–high-risk patients, 1.5% of those who received a diuretic experienced ARF, whereas only 0.3% of those who did not receive a diuretic experienced ARF. Similarly, among low-, medium-, and medium–high-risk patients, 0.8% of those who received a vasopressor infusion experienced ARF, whereas only 0.4% of those who did not receive a vasopressor infusion experienced ARF. We believe these are the most relevant comparisons. Most importantly, given that diuretic and vasopressor infusion administration were identified as independent predictors in a logistic regression analysis, they are independently associated with the ARF outcome. Our abstract conclusion only states this observation, without interpreting causation or speculating on pathophysiology.
Dr. Chappell et al. also highlight an important element of our data that may have warranted additional attention in the discussion section: the observation that intraoperative urine output is not associated with postoperative ARF. We completely agree with Dr. Chappell et al. that this is an important observation that contrasts existing clinical assumption and demands increased focus. We were reticent to expound more aggressively on this observation given that our data could not discern a causal relation or extract out the effect of fluid administration, preoperative fasting, or the timing of diuretics. We are uncomfortable concluding that these data suggest “limiting the crystalloid amount to reduce perioperative complication” as Dr. Chappell et al. suggest. In addition, Dr. Chappell et al. have brought to our attention a typographical error in table 5: The label of the second row from the bottom should include a greater-than symbol rather than a less-than symbol to read: “Urine >0.5 ml · kg−1· h−1.” This typographical error does not change the interpretation of the data: Urine output was not associated with ARF. In fact, most patients who experienced ARF did not demonstrate oliguria. We are delighted that Dr. Chappell et al. support our hypothesis-generating work. We hope these data will spur “further prospective investigations” as they suggest.
Dr. D’souza et al. raise an interesting point regarding the choice of the Cockcroft–Gault formula versus the Modification of Diet in Renal Disease formula when estimating creatinine clearance. As mentioned in the discussion section, the use of a single serum marker as a measure of renal function during a nonsteady postoperative state suffers from questionable accuracy, regardless of which formula is chosen. The Modification of Diet in Renal Disease formula was derived from data in patients with existing chronic kidney disease.5Later, it was modified to incorporate race-specific variations, providing additional accuracy in African-Americans. Conversely, the Cockcroft–Gault formula was derived using patients with and without chronic kidney disease.6The Cockcroft–Gault formula suffers from the absence of any race-specific measures. The Modification of Diet in Renal Disease formula suffers from the absence of any weight-based measures. We used the Cockcroft–Gault formula for several reasons: (1) It is more accurate across a broad range of renal function,7(2) it incorporates weight and the effect of weight on anticipated normal serum creatinine, (3) it is used more widely in pharmacologic dosing practice, and (4) the Modification of Diet in Renal Disease formula is known to underestimate glomerular filtration rate in patients with normal renal function.7Identifying patients with normal preoperative renal function was the foundation of our methodology and guided us to the use of the Cockcroft–Gault formula.
*University of Michigan Medical School, Ann Arbor, Michigan. firstname.lastname@example.org