Performance of different body weights in the Cockcroft‐Gault equation in critically ill patients with and without augmented renal clearance: A multicenter cohort

Abstract

Study objective

The primary objective was to evaluate the performance of the Cockcroft‐Gault (CG) equation with different body weights (BWs) compared to a measured creatinine clearance (mCrCl) in an intensive care unit (ICU) population with and without augmented renal clearance (ARC).

Design

Multicenter, retrospective cohort.

Setting

Two ICUs in the United States and four ICUs from a previous international observational analysis.

Patients

Adult ICU patients admitted from January 1, 2010 to July 30, 2020 with at least one mCrCl collected within the initial 10 days of hospitalization were eligible for inclusion.

Measurements and main results

The primary outcome was the performance of the CG equation in ARC (mCrCl≥130 ml/min/1.73 m²) and non‐ARC (mCrCl<130 ml/min/1.73 m²) patients. Correlation was analyzed by Pearson's correlation coefficient, bias by mean difference, and accuracy by the percentage of patients within 30% of the mCrCl. A total of 383 patients were included, which provided 1708 mCrCl values. The majority were male (n = 239, 62%), median age of 55 years [IQR 40–65] with a surgical diagnosis (n = 239, 77%). ARC was identified in 229 (60%) patients. The ARC group had lower Scr values (0.6 [0.5–0.7] vs. 0.7 [0.6–0.9] mg/dl, p < 0.001) and higher mCrCl (172.8 (SD 39.1) vs. 89.9 mL/min/1.73 m² (SD 25.4), p < 0.001) compared with the non‐ARC group, respectively. Among non‐ARC patients there was a moderate correlation (r = 0.33–0.39), moderate accuracy (range 48–58%), and low bias (range of −12.9 to 17.1) among the different BW estimations with the adjusted BW having the better performance. Among ARC patients there was low correlation (r = 0.24–0.28), low to moderate accuracy (range 38–70%), and high bias (range of −58.5 to −21.6).

Conclusions

The CG‐adjusted BW had the best performance in the non‐ARC patients, while CG performed poorly with any BW in ARC patients. Although the CG equation remains the standard equation for estimating CrCl in the ICU setting, a new, validated equation is needed for patients with ARC.

1. INTRODUCTION

Augmented renal clearance (ARC), defined as a creatinine clearance (CrCl) greater than 130 ml/min/1.73 m², affects up to 65% of critically ill patients. ¹ , ² Risk factors for development of ARC include younger age, fewer comorbidities, and intensive care unit (ICU) admission. ¹ ARC is associated with subtherapeutic concentrations of renally eliminated drugs, ³ , ⁴ , ⁵ , ⁶ may increase the risk of therapeutic failure, ⁷ and in the case of antimicrobials, may lead to development of resistance. ⁸ Many other drugs also have renal‐dependent clearance and given the importance of effective drug therapy for facilitating recovery from critical illness, an accurate estimation of renal function for optimized drug dosing is very important.

Although several calculations for estimated CrCl exist, a measured CrCl (mCrCl) is considered the most accurate method for estimating kidney function in the ICU setting, including patients with ARC. ⁹ , ¹⁰ However, obtaining a mCrCl in clinical practice can be challenging and is associated with potential inaccuracies with the collection process or lack of urinary catheterization. However, equations to estimate CrCl have not been extensively validated in critically ill patients. The primary assumption for these equations is that the serum creatinine (Scr) value is stable. ¹¹ Serum creatinine values may be acutely affected in critically ill patients by changes in nutrition, mobility status, and body composition, including muscle mass. ¹²

Various equations to estimate CrCl, including the Cockcroft‐Gault (CG), modified CG, 4‐variable and 6‐variable Modified Diet in Renal Disease (MDRD), and Chronic Kidney Disease Epidemiology Collaboration (CKD‐EPI), have been shown to result in inaccurate, biased estimates, and correlate poorly to a mCrCl in patients with ARC. ² , ⁹ , ¹⁰ , ¹³ , ¹⁴ Despite known limitations, the CG equation remains the standard method for estimating CrCl in all patients, ¹¹ , ¹⁵ including the critically ill, principally because it is established for drug dosing, including renal dose adjustments. ¹⁶ , ¹⁷ Although prior studies attempted to identify a reliable estimate for CrCl in patients with ARC, in comparison to other equations, only the CG equation was found to most accurately predict mCrCl. ⁹ , ¹⁰ However, the CG equation may still underestimate the CrCl compared to a mCrCl in critically ill patients. ⁹ , ¹⁰

As the CG equation was originally validated in a male population of predominantly normal body weight over 50 years ago, various weight metrics have been evaluated to improve accuracy among underweight and overweight/obese patients. ¹¹ , ¹⁸ , ¹⁹ Use of actual body weight (ABW) in underweight patients, ideal body weight (IBW) for normal weight patients, and adjusted body weight (AdjBW) or lean body weight (LBW) for overweight or obese patients in the CG calculation were found to lead to better estimates of CrCl in hospitalized, non‐ICU patients. ¹⁸ , ¹⁹ However, it is not known if these findings are applicable to patients with ARC. An accurate measure of renal function in patients without evidence of kidney dysfunction in the ICU setting is critical and is necessary for precise medication dosing. Additionally, the prior studies utilized ABW within their CG estimations, which may have led to inaccuracies. ⁹ , ¹⁰ , ²⁰ , ²¹ , ²² , ²³ Therefore, our primary objective was to evaluate the performance of the CG equation using different body weights compared to a mCrCl in a critically ill population in the presence or absence of ARC.

2. MATERIALS AND METHODS

This was a multicenter, retrospective cohort study between January 1, 2010 and July 30, 2020. Primary data were provided by two tertiary care hospitals from the United States, along with secondary research data from a previously conducted multicenter prospective observational study at four international sites. ²³ The study was approved by the Institutional Review Boards from all the participating sites. The lead site was Cleveland Clinic Akron General and expedited approval was granted by the Institutional Research Review Board (IRRB 20004). A waiver of informed consent was granted due to the nature of the study being minimal risk.

Patients admitted to the ICU for at least 24 h with at least one mCrCl (8‐ or 24‐h urinary measurement) were screened for inclusion. Patients were excluded for age <18 years, evidence of acute renal dysfunction (Scr >1.37 mg/dl) for any of the mCrCl days or pre‐existing end‐stage renal disease, first mCrCl collected after the initial 10 days of hospitalization, no indwelling catheter during the mCrCl collection, and not an “at risk” population. The “at risk” population for ARC included patients admitted for surgical, trauma, or neurological‐related injuries. Patients were categorized based on the presence or absence of ARC. The ARC group included patients with at least one mCrCl of at least 130 ml/min/1.73 m² or greater. The non‐ARC group consisted of those with all mCrCl less than 130 ml/min/1.73 m². Excluding patients for a Scr >1.37 mg/dl and reporting CrCl values with the units of mL/min/1.73 m² was elected to be consistent with Udy et al. ²³

Primary data were abstracted from the electronic medical records by site investigators using a standardized electronic data collection form. ²⁴ Demographic data including biologic sex, age, baseline serum creatinine, and admission anthropometric data were collected. Admission anthropometric data were utilized to calculate the IBW, ²⁵ AdjBW using 0.3 (AdjBW_0.3) and 0.4 (AdjBW_0.4) correction factors, ²⁶ LBW, ²⁷ body mass index (BMI), and body surface area (BSA) ²⁸ (Table S1). Table S1 depicts the measured CrCl and estimated CrCl formula calculated using the CG equation. ¹¹ All calculations for weights and CrCl formulas were built into the data collection tool and validated to ensure accuracy and minimize errors.

Baseline clinical data included ICU admission diagnosis, ICU length of stay, modified Sequential Organ Failure Assessment (mSOFA) scores within initial 24 h of ICU admission, and the ARC ² and Augmented Renal Clearance in Trauma Intensive Care (ARCTIC) ²⁹ scores. On the day of the mCrCl, the following variables were collected: serum creatinine, mechanical ventilation, and medications administered, including vasopressors, inotropes, hypertonic saline, diuretics, and mannitol. Serum and urinary creatinine values were determined by enzymatic methods and modified Jaffe methods (alkaline picrate) as per local laboratory practices. The primary outcome was to evaluate the performance (accuracy, bias, and correlation) of the CG equation using different body weights (IBW, ABW, AdjBW_0.3, AdjBW_0.4, LBW) as compared to a mCrCl in the presence or absence of ARC.

2.1. Statistical analysis

Correlations between each of the estimated CG CrCl for each body weight and the mCrCl were tested using Pearson's correlation coefficient. Bias was calculated as the mean difference between the estimated CG CrCl for each body weight and the mCrCl. Bland–Altman plots were used to illustrate bias relative to the mean of the measured and estimated CrCl values. Accuracy was defined as the percentage of patients with estimated CG CrCl for each body weight within 30% of the measured CrCl (i.e. P30). ¹⁸

Comparison within patients (between estimated CG‐CrCl for each body weight and the mCrCl) was performed, along with comparisons between the ARC and non‐ARC groups as independent variables. Continuous variables are reported as mean (standard deviation, [SD]) or median (interquartile range, [IQR]) and were analyzed using an independent student's t‐test or Mann–Whitney U test based on the normality for between patient comparisons. Comparisons within patients was performed using paired analyses, including paired t‐test and Wilcoxon Signed Rank test based on normality. Normality was evaluated using the Shapiro–Wilk test. Categorical variables are reported as frequency and percentage and analyzed using Pearson's chi‐squared or Fisher's exact test as appropriate. Complete case analysis was performed. For included patients, any mCrCl above 500 ml/min/1.73 m² was considered biologically implausible, and the implausible mCrCl values were excluded from the analysis. The number of participants with missing data for each variable of interest is denoted in the Table 1. Given the lack of literature, a power analysis could not be performed a priori and a convenience sample was used. A p‐value <0.05 indicated a statistically significant difference. The mCrCl values were considered independent values and a correction for repeated measures was not performed. Correction for multiplicity was not performed. All statistical analyses were performed using SAS software (version 9.4, Cary, NC).

TABLE 1.

Baseline demographics for all study patients within initial 24 h of hospital admission

Demographic variable	All patients, N = 383	ARC group, n = 229	Non‐ARC group, n = 154	p‐value
Male sex	239 (62)	159 (69)	80 (52)	0.001
Age, years, median [IQR]	55 [40–65]	49 [33–59]	63 [54–74]	<0.001
Height, cm, median [IQR]	169 [160–175]	170 [162–176]	167 [158–175]	0.042
Weights, kg, median [IQR]
Actual	72 [60–85]	73 [63–85]	70 [60–87]	0.409
Lean	56 [45–63]	57 [46–64]	53 [41–62]	0.032
Ideal	64 [55–71]	64 [57–71]	62 [52–71]	0.028
Adjusted0.3	66 [58–74]	68 [59–74]	65 [55–73]	0.102
Adjusted0.4	67 [59–75]	68 [60–75]	66 [57–74]	0.138
BMI, kg/m2, median [IQR]	25 [22–29]	25 [23–29]	25 [22–29]	0.829
BSA, m2, median [IQR]	1.8 [1.7–2.0]	1.9 [1.7–2.0]	1.8 [1.6–2.1]	0.265
Modified SOFA, n = 358, median [IQR]	3 [2–5]	3 [1–5]	3 [2–5]	0.215
ARC score, n = 357, median [IQR]	1 [1–7]	6 [1–7]	1 [1–6]	<0.001
High‐risk	118 (31)	86 (38)	32 (21)	<0.001
ARCTIC score, median [IQR]	6 [5–7]	6 [6–8]	5 [3–6]	<0.001
High‐risk	235 (61)	170 (74)	65 (42)	<0.001
Reason for ICU admission				<0.001
Surgical	293 (77)	160 (70)	133 (86)
Trauma	81 (21)	65 (28)	16 (10)
Neurologic	9 (2)	4 (2)	5 (3)
ICU LOS, days, n = 350, median [IQR]	6 [3–12]	7 [3–12]	4 [2–13]	0.095
Mechanical ventilation, n = 381	381 (69)	172 (75)	91 (60)	0.001
Concomitant medications, n = 382
Vasopressors	119 (31)	75 (33)	44 (29)	0.371
Inotropes	2 (1)	1 (0.4)	1 (1)	1.000
Hypertonic saline	2 (1)	1 (0.4)	1 (1)	1.000
Diuretics	7 (2)	5 (2)	2 (1)	0.706
Mannitol	1 (0.3)	0 (0)	1 (1)	0.403

Note: Presented as n (%) unless otherwise stated.

Abbreviations: ARC, Augmented Renal Clearance; ARCTIC, Augmented Renal Clearance in Trauma Intensive Care; BMI, body mass index; BSA, body surface area; IQR, interquartile range; ICU, intensive care unit; LOS, length of stay; SOFA, Sequential Organ Failure Assessment score.

3. RESULTS

A total of 399 patients were identified as having a mCrCl during an ICU admission, with 383 patients meeting all inclusion/exclusion criteria (Figure S1). Primary reasons for exclusion were not being an “at risk” population (n = 5), no indwelling urinary catheter (n = 4), mCrCl obtained after 10 days (n = 3), and evidence of acute renal dysfunction (n = 3). A total of 1708 mCrCl collections were available, with 4 mCrCl values excluded for biological implausibility. The majority of mCrCl were from 8‐h urine collections (n = 1631, 95%) as opposed to 24 hours (n = 77, 5%). Among included patients, the median number of mCrCl per patient was 4 [IQR 2–9]. Two‐hundred and twenty‐nine (n = 229; 60%) patients were included in the ARC group, and 154 (40%) in the non‐ARC group. There were 105 (27%) and 278 (73%) of patients from US and non‐US sites, respectively.

Baseline demographic and clinical characteristics are summarized in Table 1. The overall population was primarily male (n = 239, 62%) with a median age of 55 years [IQR 40–65] and a surgical diagnosis (n = 239, 77%). The ARC group had a greater proportion of males, lower median age, greater use of mechanical ventilation, and higher trauma‐related reason for admission compared to the non‐ARC group. There were significant differences in anthropometric data, with higher median heights, LBW, and IBW in the ARC group. There were no differences in other BW, BMI, or BSA measurements between groups. The ARC group had significantly higher median ARC and ARCTIC scores, which corresponded with more ARC patients classified as high‐risk by both the ARC and ARCTIC scores. Patients with ARC had lower median serum creatinine concentrations (0.6 [0.5–0.7] vs. 0.7 [0.6–0.9] mg/dl, p < 0.001) compared with non‐ARC patients, respectively.

The performance data for the CrCl measurements are described in Table 2. The ARC group had a mean mCrCl of 172.8 (SD 39.1) ml/min/1.73 m². The non‐ARC group had a mean mCrCl of 89.9 (SD 25.4) mL/min/1.73 m². Of the mCrCl, 801 (47%) values were at least 130 mL/min/1.73 m² or greater. Figure 1 further depicts the differences between the measured and estimated CrCl between the study groups with the values reported in Table 3. Compared to the mCrCl as the reference value, the CG equation had correlations between 0.24 and 0.28, accuracy 38–70%, and bias −58.5 to −21.6 for the ARC group (Table 2). Compared to the mCrCl, the CG equation had correlations that ranged from 0.33 to 0.39, accuracy 48–58%, and bias −12.9 to 17.1 for the non‐ARC group (Table 2).

TABLE 2.

Correlation, accuracy, and bias of different body weights in the cockcroft‐gault equation as compared to measured creatinine clearance

	Creatinine Clearance a , mean (SD)	Correlation (r)	Accuracy	Bias, Mean (SD)
ARC
n (%)	801 (47)
Measured CrCl	172.8 (39.1)	Reference	Reference	Reference
CG‐ABW	151.2 (44.5)	0.26**	70%	−21.6**
CG‐LBW	114.3 (34.7)	0.28**	38%	−58.5**
CG‐IBW	134.2 (47.4)	0.24**	51%	−38.6**
CG‐AdjBW0.3	139.3 (45.1)	0.26**	59%	−33.5**
CG‐AdjBW0.4	141.0 (44.6)	0.26**	61%	−31.8**
Non‐ARC
n (%)	907 (53)
Measured CrCl	89.9 (25.4)	Reference	Reference	Reference
CG‐ABW	107.0 (46.6)	0.39**	56%	17.1**
CG‐LBW	77.0 (34.5)	0.37**	48%	−12.9**
CG‐IBW	90.6 (44.0)	0.33**	52%	0.7
CG‐AdjBW0.3	95.5 (43.3)	0.36**	57%	5.6**
CG‐AdjBW0.4	97.1 (43.3)	0.37**	58%	7.2**

Note: Data presented as mean (SD) unless otherwise stated.

Abbreviations: 0.3, 30% correction factor; 0.4, 40% correction factor; ABW, actual body weight; AdjBW, Adjusted body weight; ARC, augmented renal clearance; CrCl, creatinine clearance; CG, Cockcroft‐Gault; IBW, ideal body weight; LBW, lean body weight; SD, standard deviation.

^a Creatinine clearance values expressed as ml/min/1.73 m².

^** p‐value < 0.001.

FIGURE 1.

Box plot comparing measured and estimated creatinine clearance between study groups. ARC, augmented renal clearance; ABW, actual body weight; AdjBW, adjusted body weight; CG, Cockcroft‐Gault equation; IBW, ideal body weight; LBW, lean body weight; mCrCl, measured creatinine clearance. **Between group comparisons with p‐value < 0.0001.

TABLE 3.

Differences in mean creatinine clearance, bias, and accuracy between ARC and non‐ARC measurements

Creatinine clearances	CrCl, Mean (SD)			Bias, Mean (SD)			Accuracy, n (%)
	ARC (n = 801)	Non‐ARC (n = 907)	p‐value	ARC (n = 801)	Non‐ARC (n = 907)	p‐value	ARC (n = 801)	Non‐ARC (n = 907)	p‐value
Measured	172.8 (39.1)	89.9 (25.4)	<0.0001
CG‐ABW	151.2 (44.5)	107.0 (46.6)	<0.0001	−21.6 (50.9)	17.1 (43.6)	<0.0001	561 (70)	507 (56)	<0.0001
CG‐LBW	114.3 (34.7)	77.0 (34.5)	<0.0001	−58.5 (44.4)	−12.9 (34.5)	<0.0001	306 (38)	438 (48)	<0.0001
CG‐IBW	134.2 (47.4)	90.6 (44.0)	<0.0001	−38.6 (53.6)	0.7 (42.9)	<0.0001	407 (51)	474 (52)	0.5499
CG‐AdjBW0.3	139.3 (45.1)	95.5 (43.3)	<0.0001	−33.5 (51.6)	5.6 (41.6)	<0.0001	472 (59)	519 (57)	0.4762
CG‐AdjBW0.4	141.0 (44.6)	97.1 (43.3)	<0.0001	−31.8 (51.1)	7.2 (41.4)	<0.0001	491 (61)	523 (58)	0.1268

Note: Creatinine clearance expressed as ml/min/1.73 m². Bias expressed as the mean difference from the measured creatinine clearance. Accuracy expressed as percentage of patients within 30% of the measured creatinine clearance.

Abbreviations: 0.3, 30% correction factor; 0.4, 40% correction factor; ABW, actual body weight; AdjBW, Adjusted body weight; ARC, Augmented Renal Clearance; CG, Cockcroft‐Gault; CrCl, Creatinine Clearance; IBW, ideal body weight; LBW, lean body weight; SD, standard deviation.

Between the study groups, there were significant differences in bias between all the BW‐adjusted estimates. Significant differences in accuracy between the ARC and non‐ARC groups for the CG‐ABW and CG‐LBW were observed, but no difference for the other BWs as depicted in Table 3. Figure S2 illustrates the bias between the measured and estimated CrCl for the study groups. Bland–Altman plots further illustrate the equation bias for the ARC (Figure S3) and non‐ARC (Figure S4) groups.

4. DISCUSSION

In a cohort of critically ill patients “at risk” for ARC, the overall incidence of ARC was 60%. This was one of the largest evaluations of critically ill patients with a significant number of CrCl measurements. Compared to the mCrCl there was low correlation for the CG estimations with all the BWs among ARC patients. However, the CG‐ABW had the highest accuracy and least bias in the ARC group. Conversely, among non‐ARC patients there was moderate correlation and accuracy compared to mCrCl, which was the highest with the CG‐adjusted BW. However, the CG‐IBW produced the lowest bias in the non‐ARC group. As highlighted in the Bland–Altman plots, as the average of mCrCl and estimated CrCl increases, there was a higher level of bias and less agreement. Although the aim of this study was to evaluate the performance of the CG equation in an ARC population, there was poor performance overall despite inputting different body weights.

While previous studies primarily investigated trauma and neurocritical care populations, ²⁰ , ²¹ our study is novel by including a variety of patients including primarily a surgical population, which was also found to have a high percentage of patients with ARC. This study has major implications for patients with ARC, as ARC may result in subtherapeutic levels of many medications, such as anti‐seizure medications, antimicrobials, anticoagulants, and others. ¹ Given that many medications are dosed based on the CG equation, incorrect CrCl estimations may lead to less precise and potentially nontherapeutic medication dosing.

The results of our study are consistent with prior work on how the CG equation performs in both ARC and non‐ARC populations. A retrospective analysis in hospitalized adults of various body weights and SCr concentrations evaluated the performance of the CG‐CrCl compared to a 24‐h mCrCl. Overall, there was a high correlation with all the body weights (range 0.792–0.906, p < 0.001). However, the IBW resulted in least bias in normal BW patients, mean difference −1.3 (95% confidence interval [CI] −3.2–0.6, p = 0.544), along with high accuracy. The AdjBW_0.4 resulted in lower bias for overweight and obese patients and provided the highest degree of accuracy. ¹⁸ Application of these results are limited as inclusion of critically ill patients was lacking and the presence of ARC was not evaluated. However, the cohort included a significant number of patients with renal dysfunction with a mean SCr value of 1.56 mg/dl (SD 1.3) and 40% with a mCrCl below 40 ml/min.

Thus far, all equations to estimate CrCl have shown to be significantly inaccurate as compared to the mCrCl in critically ill patients with ARC. However, prior literature has previously identified that the CG equation performed better than other estimations of mCrCl, including the MDRD and CKD‐EPI in critically ill patients. ⁹ Baptista et al. found no significant difference in the CG equation compared to the mCrCl (mean 135.5 ml/min/1.73m² for CG equation vs. 135.7 ml/min/1.73m² for mCrCl) using ABW but found a significant difference with the MDRD and CKD‐EPI equations (p < 0.05). However, poor correlation and poor precision (i.e., bias) with all the equations, including the CG equation, was observed. A prior study by Baptista et al., ¹⁰ performed a similar analysis of critically ill patients with ARC utilizing multiple estimations of CrCl compared to a mCrCl. Consistent with other findings, all the estimations of CrCl underestimated the mCrCl including the CG‐CrCl (median CG‐CrCl: 135 [116–171] vs. mCrCl: 162 [145–190] ml/min/1.73m², p < 0.01). This resulted in poor correlation of the estimations (ranging from 0.18 to 0.26) and a low correlation (r _s  = 0.26, p = 0.017) for the CG‐CrCl. Overall, these results are consistent with our study, including poor correlation and high risk of bias for the patients with ARC. However, the weight utilized for the CG equation was not specified in the previous studies.

A similar analysis in a surgical ICU population evaluated the performance of the CG equation in an ARC cohort. ¹⁴ Within their CG estimation, adjustments for BW were performed using ABW for BMI < 30 kg/m² and LBW for BMI > 30 kg/m². The median CG‐CrCl was significantly higher than the mCrCl (158 [122–209] mL/min vs. 148 [132–172] mL/min/1.73m², p = 0.004, respectively). This corresponded to poor correlation of r _s  = 0.343 and mean precision (i.e., bias) of −11.2 (SD 61.5). In contrast to Baptista analyses, a surgical ICU evaluation found that the CG‐CrCl estimation significantly overestimated the mCrCl. ¹⁴ This may be explained by differences in the study population, which was older and more severely ill in the surgical ICU analysis, as well as medications utilized, including a high proportion of patients on calcineurin inhibitors, nonsteroidal anti‐inflammatory drugs, and trimethoprim/sulfamethoxazole. ¹⁴ Although the surgical ICU study included the weight used for CG‐CrCl estimation, the cohort characteristics make the results less generalizable to other ICU populations. ¹⁴

There were several strengths with this study worth noting. Although a sample size calculation could not be calculated, this was one of the largest studies based on the number of patients and CrCl measurements. This study was a large cohort consisting of ARC and non‐ARC populations, which may both coexist in the ICU setting. This study was comprised of a variety of critically ill patients, ranging from trauma, surgical, and neurologic injuries. Furthermore, data were provided across multiple international sites, which increases the generalizability. Patients were limited to those without apparent renal dysfunction or those at higher risk for ARC to ensure that included patients shared a similar baseline risk for ARC in order to minimize confounding. Additionally, there were a variety of weights utilized in this study, including actual, ideal, lean, and adjusted BWs, to ensure that an extensive list of weights commonly utilized in the ICU were evaluated.

Although our study expanded on previous literature, several limitations are worth noting. The retrospective nature and secondary analysis possibly resulted in missing or incomplete data although this was found to be minimal. Systematic errors may be possible from the measured urinary collections and the lack of standardized methods for obtaining anthropometric measurements among the multiple sites. However, the presence of an indwelling urinary catheter and collection under direct supervision of a health care professional were required. Although 8‐h mCrCl collections were performed at the majority of sites, the mCrCl collections were determined by both 8 and 24 h based on the site. Previous studies have found acceptable agreement with either collection time. ³⁰ , ³¹ There were two modalities for determining creatinine values although the Jaffe method was utilized in most patients. Inulin or exogenous markers for determining the glomerular filtration rate (GFR) were not available for this study, which would provide an exact clearance, albeit these measures are not used in clinical practice. Although we aimed to exclude patients with significant renal dysfunction, critically ill patients may have had fluctuations in serum creatinine, which falls outside of the assumptions for the CG equation. ¹¹ The body size in this patient population was primarily nonobese, which may limit the generalizability, and the population was not delineated into separate obesity categories. Application of our findings to an obese population should be done cautiously and more studies are needed in this population. Given the prior literature supporting use of the CG equation in critically ill patients, our evaluation was limited to this calculation, and the results may not be generalizable to other CrCl estimations. Based on the scope of this study, there was no evaluation of changes over time, including intra‐patient variability with dynamic renal function and performance measures.

Although the CG equation remains the standardized method for drug dosing, this study may have several implications on clinical practice. There is a significant need to have an accurate estimation of CrCl, especially when urinary measurements of CrCl are not feasible to obtain or performed consistently in practice. Ultimately, this study highlights that utilizing the CG equation for critically ill patients without ARC is acceptable, but an IBW or adjusted BW should be utilized within the CG equation. Conversely, the CG equation in patients with ARC will lead to an inaccurate and less precise estimation of CrCl. An inaccurate and imprecise CrCl estimation may have a significant impact on less optimized and potentially nontherapeutic medication dosing, which may affect clinical outcomes. Although the CG equation remains a widely utilized method for estimating CrCl for ICU patients without evidence of renal dysfunction, the mCrCl may be considered for determining CrCl for ICU patients with ARC. Furthermore, future research should be aimed at determining a new equation to estimate CrCl specific to this population for those in which a mCrCl is not feasible.

5. CONCLUSIONS

In critically ill patients at risk for ARC and without acute renal dysfunction, the CG using an IBW or adjusted BW had the best performance in non‐ARC patients, while CG performed poorly with any BW in patients with ARC and should be used with caution. In critically ill patients with ARC, a mCrCl should be considered if an accurate measure of CrCl is necessary for critical drug dosing. This highlights the need for an accurate and precise estimate for CrCl, especially in a vulnerable, critically ill population. Further research is needed to determine the best equation to estimate CrCl in patients with ARC, and a new, validated equation should be investigated in future studies.

AUTHOR CONTRIBUTIONS

Michaelia D. Cucci: Original study concept, literature search, study design, data collection, analysis and interpretation of data, drafting, critical revision, and approval of manuscript. Anthony T. Gerlach: Study design, data collection, interpretation of data, drafting, critical revision, and approval of manuscript. Caroline Mangira: Analysis and interpretation of data, critical revision and approval of manuscript. Claire V. Murphy: Study design, data collection, interpretation of data, drafting, critical revision, and approval of manuscript. Jason A. Roberts: Study design, data collection, interpretation of data, drafting, critical revision, and approval of manuscript. Andrew A. Udy: Data collection, interpretation of data, drafting, critical revision, and approval of manuscript. Thomas C. Dowling: Study design, interpretation of data, drafting, critical revision, and approval of manuscript. Chanda L. Mullen: Study design, analysis and interpretation of data, drafting, critical revision, and approval of manuscript.

FUNDING INFORMATION

J.A. Roberts acknowledges funding from the Australian National Health and Medical Research Council for a Centre of Research Excellence (APP2007007) and an Investigator Grant (APP2009736) as well as an Advancing Queensland Clinical Fellowship.

CONFLICT OF INTEREST

The remaining authors have disclosed that they do not have any conflicts of interest or sources of funding.

Supporting information

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Table S1.

Cucci MD, Gerlach AT, Mangira C, et al. Performance of different body weights in the Cockcroft‐Gault equation in critically ill patients with and without augmented renal clearance: A multicenter cohort. Pharmacotherapy. 2023;43:1131‐1138. doi: 10.1002/phar.2743