Abstract
Background and objectives Being a woman is a well established risk factor for the development of cardiothoracic surgery–associated AKI. In striking contrast, women are less likely to develop AKI associated with noncardiac surgical procedures than men. In an attempt to ascertain why being a woman might be protective for ischemic AKI after general surgery but deleterious in patients undergoing cardiothoracic surgery, we examined cardiothoracic surgery–associated AKI in greater detail.
Design, setting, participants, & measurements We performed a systematic review and meta-analysis of cardiothoracic surgery–associated AKI studies published between January of 1978 and December of 2015 to further explore the relationship between sex and cardiothoracic surgery–associated AKI.
Results Sixty-four studies were identified that provided sex-specific data regarding the incidence of cardiothoracic surgery–associated AKI among 1,057,412 subjects. Using univariate analysis, women were more likely than men to develop AKI postoperatively (odds ratio, 1.21; 95% confidence interval, 1.09 to 1.33; P<0.001). However, when the analysis was restricted to the 120,464 subjects reported in 29 studies that used the Acute Kidney Injury Network criteria, the RIFLE criteria, or the Kidney Disease Improving Global Outcomes criteria to define AKI, there was no significant sex–related difference in risk. Seventeen studies used multivariate analysis to assess risk factors for cardiothoracic surgery–associated AKI and provided sex–specific odd ratios. Among the 1,587,181 individuals included in these studies, the risk of developing cardiothoracic surgery–associated AKI was not significantly associated with sex (odds ratio, 1.04; 95% confidence interval, 0.92 to 1.19; P=0.51). However, when the analysis was restricted to the 5106 subjects reported in four studies that used the Acute Kidney Injury Network criteria to define AKI, the risk of developing AKI was significantly lower in women compared with in men (odds ratio, 0.75; 95% confidence interval, 0.65 to 0.87; P<0.001).
Conclusions Our systematic review and meta-analysis contradict the generally held consensus that being a woman is an independent risk factor for the development of cardiothoracic surgery–associated AKI.
- acute renal failure
- cardiac surgery
- sex
- acute kidney injury
- Cardiac Surgical Procedures
- Consensus
- female
- humans
- Incidence
- male
- Multivariate Analysis
- risk factors
- Specialties, Surgical
- Thoracic Surgery
Introduction
Sex dimorphism is a well recognized feature of chronic progressive kidney disease (1). Being a woman is thought to play a protective role, mediated by differences in the hormonal milieu (1). Although less well appreciated, sex dimorphism has also been clearly established in the development of ischemic AKI (2). There is a lower incidence of AKI after noncardiac surgery in women compared with men (3–5). In striking contrast, being a woman has repeatedly been identified as predicting the development of cardiothoracic surgery–associated AKI (CSAKI) as well as the increased morbidity and mortality associated with this complication (6–21). The Kidney Disease Improving Global Outcomes (KDIGO) Clinical Practice Guideline for Acute Kidney Injury consensus report clearly recognizes the association between women and an increased risk for CSAKI (22). The vast majority of review articles written on the subject also identify women as being at greater risk for CSAKI (6–21). Moreover, several cardiovascular surgery renal risk stratification algorithms developed to assess the preoperative risk of CSAKI include women as a risk factor (23–25). In an attempt to ascertain why being a woman might be protective in patients undergoing general surgical procedures but detrimental in patients undergoing cardiothoracic surgery, we examined CSAKI in greater detail by performing a systematic review and meta-analysis of CSAKI studies published between January of 1978 and December of 2015.
Materials and Methods
Search Strategy and Selection Criteria
We conducted a systematic review and meta-analysis of the English literature to evaluate the reported incidence of AKI after cardiac surgery in men and women (Figure 1). This systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses protocol (26).
Identification of studies for meta-analysis.
We searched PubMed and EMBASE for English language articles published between January 1, 1978 and January 1, 2015. The following medical subject heading terms were used: cardiac surgery, cardiothoracic surgery, cardiovascular surgery, heart surgery, AKI, acute renal failure, and acute renal insufficiency. We also examined the references of the articles that were recovered for additional resources and included those studies that satisfied the inclusion criteria. At least two reviewers independently reviewed titles and abstracts of the articles that were found in the database searches to identify eligible studies. Full text versions of the selected studies were analyzed in detail. Any case-control or cohort study of subjects who underwent cardiothoracic surgery (coronary artery bypass graft [CABG], thoracic aorta, and/or cardiac valve surgery) was eligible. We only included studies that reported data on the incidence of AKI in men versus women. We excluded studies in subjects <18 years of age, studies in which the sample size was <500 patients, and studies in which <15% of the patients were women. We also excluded studies of subjects that underwent cardiac transplant surgery or endovascular repair of cardiac valves.
Definition of AKI
We accepted studies that defined AKI by investigator-created criteria, the Acute Kidney injury Network (AKIN) criteria, the KDIGO criteria, the RIFLE criteria, or the requirement for RRT (22,27,28). Investigator-created criteria for AKI included a specified increase in the serum creatinine level, a specified decline in calculated GFR, or a combination of these criteria. The primary outcome was defined as AKI after cardiac surgery by any of these definitions.
Data Extraction
All studies were examined for duplication of data. Attention was given to the reporting clinical center, years covered, and overlap with larger regional or national databases. If there was an estimated overlap of 33% or more of individuals between any two studies, the smaller study was excluded. To determine study quality, the studies were independently assessed by the authors using the Newcastle Ottawa Score for cohort and case-control studies (29). Disagreements on the scores of individual studies were resolved by discussion. Studies with scores less than six were excluded from the final analysis.
Statistical Analyses
Data were analyzed using a random effects model with RevMan, version 5.3 (The Cochrane Collaboration). We analyzed separately studies that used univariate versus multivariate analysis.
Results
Sixty-four studies were identified that provided sex-specific data regarding the incidence of AKI associated with cardiothoracic surgery among 1,057,412 patients (293,689 women and 763,723 men) (11,23–25,30–89). Seventeen studies included only patients who underwent isolated coronary artery bypass grafting, and two studies included only subjects who underwent isolated valvular surgery, whereas the remaining 45 studies included a mixture of subjects who underwent isolated or combined procedures. AKI was defined by the RIFLE criteria in 12 studies, the AKIN criteria in 11 studies, the KDIGO criteria in six studies, the need for RRT in 11 studies, and investigator-defined criteria in the remaining 24 studies. Women were more likely than men to develop CSAKI (odds ratio [OR], 1.21; 95% confidence interval [95% CI], 1.09 to 1.33; P<0.001) (Figures 2 and 3). When the analysis was restricted to the 57,259 subjects reported in 11 studies that used the need for RRT to define AKI, women were at even greater risk to develop CSAKI compared with the full cohort (OR, 1.70; 95% CI, 1.19 to 2.44; P=0.004). In the RRT subset, the incidence of CSAKI among women was 4.2% (695 of 16,554) compared with 2.6% among men (1056 of 40,705). In contrast, when the analysis was restricted to the 22,062 subjects reported in the 12 studies that used the RIFLE criteria to define AKI, the risk of developing CSAKI was not significantly influenced by sex (OR, 1.11; 95% CI, 0.97 to 1.27; P=0.13). In the RIFLE subset, the incidence of CSAKI among women was 26.0% (1758 of 6763) compared with 24.3% among men (3720 of 15,299). Similarly, when the analysis was restricted to the 69,438 patients reported in the 11 studies that used the AKIN criteria to define AKI, the risk of developing CSAKI was not significantly influenced by sex (OR, 0.94; 95% CI, 0.81 to 1.09; P=0.38). In the AKIN subset, the incidence of CSAKI among women was 9.1% (1728 of 18,914) compared with 10.1% among men (5118 of 50,524). When the analysis was restricted to the 28,964 individuals reported in the six studies that used the KDIGO criteria to define AKI, the risk of developing CSAKI was again not significantly influenced by sex (OR, 0.84; 95% CI, 0.66 to 1.07; P=0.16). In the KDIGO subset, the incidence of CSAKI among women was 27.0% (2226 of 8231) compared with 28.5% among men (5903 of 20,733).
Increased AKI risk for women in 64 studies with sex-specific data. 95% CI, 95% confidence interval; M.H., Mantel Haenszel.
High variability amongst 64 studies with sex-specific data. Each circle represents a separate study.
Fifty-five studies were identified that used multivariate analysis to evaluate the factors that contributed to CSAKI, with sex as one of the independent variables (11,23,25,30,31,33–35,38–43,45–50,52,54,55,57,58,60–64,66,68,71–73,75,77,80,83,85,86,89–102); 17 of these 55 studies provided sex-specific ORs (23,25,35,45,48,63,64,75,89,95–102). Individual studies evaluated a number of covariates that might interact with sex to influence the development of CSAKI. These covariates included age (100% of studies); ejection fraction or history of congestive heart failure (100%); history of diabetes mellitus (100%); reoperation (92%); type of surgical procedure (89%); baseline renal function (83%); history of hypertension (75%); prior myocardial infarction (58%); cardiopulmonary bypass time (58%); body mass index, body surface area, or weight (58%); hemoglobin level, transfusion requirement, or estimated blood loss (50%); use of intra–aortic balloon pump (50%); crossclamp time (50%); smoking (25%); use of angiotensin–converting enzyme inhibitors or angiotensin receptor blockers (8%); statin use (8%); and diuretic use (8%). Four studies included only subjects who underwent isolated coronary artery bypass grafting, and two studies included only subjects who underwent isolated cardiac valve or aortic surgery, whereas the remaining 11 studies included a mixture of subjects who underwent isolated or combined procedures. AKI was defined by the RIFLE criteria in one study, the AKIN criteria in four studies, the KDIGO criteria in two studies, the need for RRT in five studies, and investigator-defined criteria in the remaining five studies. The total group included 1,587,181 patients (473,269 women and 1,113,912 men). In this cohort, the risk of developing CSAKI was not significantly influenced by sex (OR, 1.04; 95% CI, 0.92 to 1.19; P=0.51) (Figure 4). In contrast, when the analysis was restricted to the 5106 subjects reported in four studies that used the AKIN criteria to define AKI, the risk of developing CSAKI was significantly lower in women compared with men (OR, 0.75; 95% CI, 0.65 to 0.87; P<0.001). Only three studies provided ORs for AKI stratified by stage.
No difference in sex-specific odds ratio in multivariate studies analyzed. 95% CI, 95% confidence interval; IV, inverse variance.
We observed a high degree of statistical heterogeneity in our meta-analyses (univariate: I2=94%; multivariate: I2=95%). However, statistical heterogeneity was entirely eliminated among studies that used multivariate analysis and defined AKI by the AKIN criteria (I2=0%). Among studies that used univariate analysis and defined AKI by the KDIGO criteria, statistical heterogeneity was low (I2=21%). Absent one outlier, the degree of statistical heterogeneity was moderate among studies that used univariate analysis and defined AKI by the KDIGO criteria (I2=41%). The same was true among studies that used univariate analysis and defined AKI by the AKIN criteria (I2=59%).
Thirty-eight of the 55 studies that used multivariate analysis evaluated the effect of sex on the risk of CSAKI but failed to provide sex-specific ORs (11,30,31,33,34,38–43,46,47,49,50,52,54,55,57,58,60–62,66,68,71–73,77,80,83,85,86,90–94). AKI was defined by the RIFLE criteria in 10 studies, the AKIN criteria in six studies, the KDIGO criteria in one study, the need for RRT in four studies, and investigator-defined criteria in the remaining 17 studies. This cohort included 101,835 patients (27,115 women and 74,720 men). All 38 studies failed to find any significant effect of sex on the incidence of CSAKI.
Discussion
Sex dimorphism is a well recognized feature of chronic progressive kidney disease (1). Although less well appreciated, sex dimorphism has also been clearly established in ischemic AKI (2). Large epidemiologic studies show that women exhibit a lower incidence of AKI after noncardiac surgery (3–5,103–106). Kheterpal et al. (5) used data from 57,075 general surgery procedures performed in the United States between 2005 and 2006 to create an AKI preoperative risk stratification model. In this analysis, being a man was found to be an independent predictor of AKI (OR, 1.4; 95% CI, 1.2 to 1.7). Grams et al. (4) analyzed 161,185 veterans (96.3% men) who underwent major surgery (including cardiac surgery) between 2004 and 2011 and found being a man to be an independent risk factor for AKI defined by the KDIGO criteria, by univariate analysis as well as in a fully adjusted model.
AKI is an important complication of cardiothoracic surgery that confers substantial additional morbidity and mortality (6–21). Before the development of standardized criteria to define AKI, this complication was reported in 0.3%–40% of patients depending on the definition of AKI used by the investigator, the complexity of the surgical procedures, and the prevalence of preexisting risk factors in the study population (6–21). AKI requiring RRT was reported to occur in 0.37%–5% of patients (6–21). Numerous investigators have attempted to identify risk factors that predispose to the development CSAKI. Risk factors that have repeatedly been identified include baseline renal function, the type and complexity of the surgical procedure, age, women, emergency surgery, prior cardiac surgery, reoperation, need for intra–aortic balloon pump, cardiogenic shock, the length of cardiopulmonary bypass, and comorbidities, such as insulin–requiring diabetes mellitus, peripheral vascular disease, congestive heart failure, and chronic obstructive lung disease (6–21). More recently, investigators have identified preoperative hemoglobin, nadir hemoglobin during cardiopulmonary bypass, and the number of blood transfusions as additional risk factors (50,57,88,107–109). More controversial are the roles played by on- versus off-pump surgery and aprotinin use (41).
Being a woman has repeatedly been identified in these studies as an independent risk factor for the development of CSAKI as well as the increased morbidity and mortality associated with this complication (6–21). Most review articles dealing with CSAKI recognize being a woman as an independent risk factor for its development on the basis of retrospective and prospective studies, local databases, and national and international registry data (6–21). The Nephrology Self-Assessment Program identifies “demographic factors such as female gender…as independent risk factors of [cardiovascular surgery–associated] AKI” (110). In addition, the KDIGO consensus statement clearly recognizes being a woman as an important risk factor for CSAKI (22). In this context, several cardiovascular surgery renal risk stratification algorithms have been developed to assess the preoperative risk of CSAKI. Three prominent scoring systems include being a woman as a risk factor for AKI (23–25). In contrast, several other CSAKI risk algorithms do not include sex as a risk factor (34,47,72,75,83,90), and two scoring systems recognize being a woman as renoprotective (35,111).
Among the difficulties in comparing data from the various studies of CSAKI are the differing criteria used to define AKI. Although many studies used investigator-created criteria, others used the RIFLE, the AKIN, or the KDIGO criteria. The need for RRT is an inadequate end point to define CSAKI because of subjectivity inherent in the decision to initiate dialysis (24,112,113). In a cohort of patients with AKI who underwent propensity score matching, Wilson et al. (113) found that dialysis was more likely to be initiated in women than in men. Noyez (92) applied five different investigator–created definitions of AKI obtained from the published literature to his patient cohort and found that the incidence of CSAKI varied between 4.9% and 38.1% depending on the definition used. Sex was found to predict CSAKI by univariate analysis using one of the definitions of AKI, but it was not found to predict CSAKI when the other definitions were substituted. Similarly, we found that the incidence of CSAKI varied widely depending on the definition used. AKI complicated cardiothoracic surgery in 5.3% of patients when all studies were combined, 28.2% of patients in those studies in which the RIFLE criteria are used, 14.0% of patients in those studies in which the AKIN criteria are used, and 26.6% of patients in those studies that used the KDIGO criteria. RRT was required in 3.1% of patients.
When we combined all studies that provided sex-specific data for univariate analysis, irrespective of the definition of CSAKI used, being a woman increased the risk of CSAKI. In contrast, being a woman was not a risk factor in those studies in which the RIFLE, the AKIN, or the KDIGO criteria were used to define CSAKI. Focusing solely on those studies that used multivariate analysis to assess CSAKI risk, when we combined all studies irrespective of the definition of CSAKI used, we found no association between sex and AKI risk. However, being a woman was found to be renoprotective by multivariate analysis in those studies that used the AKIN criteria to define CSAKI.
These disparate conclusions may merely reflect the differing definitions of CSAKI, because the criteria used to define AKI clearly influence the outcome. Alternatively, this simple explanation may mask more complex interactions between sex and cardiovascular surgery outcomes. Moreover, failure of the covariates used in multivariate analyses to capture these interactions may obscure the true relationship between sex and CSAKI. Most studies have shown that women present for cardiovascular surgery with a higher burden of comorbidities and worse cardiovascular status than men (36,88,102,114–117). This disparity may arise from sex-based biases in the delivery of health care or because of a more aggressive course of cardiac disease in women (102). It has been suggested that referral bias leads women to undergo coronary artery revascularization or cardiac valve surgery at a more advanced stage of disease (102,114). Women undergoing cardiovascular surgery also show a higher prevalence of risk factors that have been associated with a poor outcome. An analysis of the Society of Thoracic Surgery National Database reported that women undergoing coronary artery bypass grafting surgery between 1994 and 1996 were older; had a higher prevalence of comorbidities, such as diabetes mellitus, hypertension, and peripheral vascular disease; and were more likely to undergo nonelective emergency procedures and more complicated valve surgeries than men (114). Examination of a subset of the same database for the years 1996 and 1997 showed that women were older and more likely to have diabetes and hypertension but less likely to have smoked or have triple-vessel disease than men (88,115). Mehta et al. (88) also reported that women undergoing cardiovascular surgery showed a higher risk profile than men because of more advanced age; lower body surface area and preoperative hematocrit; greater prevalence of congestive heart failure, cardiogenic shock, recent myocardial infarction, unstable angina, and shock; and increased likelihood of undergoing reoperation or high-risk procedures requiring longer cardiopulmonary bypass time and increased numbers of transfusions.
Nadir hematocrit during cardiopulmonary bypass has recently been identified as an additional independent risk factor for CSAKI (50,88,108,109). Women exhibit lower preoperative hematocrit levels and because of their smaller body surface area, experience greater hemodilution during cardiopulmonary bypass, resulting in a lower nadir hematocrit (88,108). Paradoxically, at the lowest nadir hematocrit levels, women have a lower risk of CSAKI than men, suggesting that women may tolerate severe hemodilution better than men (88). Because of lower nadir hematocrit levels, women are also more likely to undergo a blood transfusion, which itself is an established independent risk factor for CSAKI (118). Adding further complexity to these interactions, the greater hemodilution that women experience in the perioperative setting may, in turn, reduce the rate of detection of AKI in women compared with men (119).
Being a woman was associated with a reduced risk of CSAKI among 503,478 subjects who underwent isolated CABG between the years 1997 and 1999 as reported in a multivariate analysis of The Society of Thoracic Surgeons National Adult Cardiac Surgery Database (OR, 0.94; 95% CI, 0.89 to 1.00), whereas being a woman was associated with an increased risk of CSAKI among 774,881 subjects entered into the same registry who underwent isolated CABG between 2002 and 2006 (OR, 1.25; 95% CI, 1.21 to 1.31) (89,100) These discrepant data from the same registry are consistent with our conclusion that sex is not an independent risk factor for the development of CSAKI. Although differing criteria to define CSAKI may explain these inconsistent data, clearly, the anatomic and physiologic differences between the sexes did not change between the years 1997–1999 and 2002–2006. We believe that factors related to but not inherent in one’s sex are responsible for these differing outcomes. We point to difficulties in accurately quantifying differences in the severity of comorbidities and other preoperative risk factors that may not be adequately captured in the covariates used in the multivariate analyses.
In conclusion, our meta-analysis contradicts the generally held consensus that being a woman is an independent risk factor for the development of CSAKI. These disparate results may merely reflect different definitions of CSAKI. However, this simple explanation may mask more complex interactions between sex and cardiovascular surgery. We believe that factors related to but not inherent in one’s sex may be responsible for and reflect a greater burden of comorbidities in women undergoing cardiovascular surgery because of disparities in health care delivery or a more aggressive course of cardiovascular disease in women.
Disclosures
None.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received March 23, 2016.
- Accepted August 16, 2016.
- Copyright © 2016 by the American Society of Nephrology