Systematic Review: Sodium Bicarbonate Treatment Regimens for the Prevention of Contrast-Induced Nephropathy

Context Previous reviews suggest that sodium bicarbonate prevents contrast-induced nephropathy. Contribution This review of 9 published and 14 unpublished trials of sodium bicarbonate suggests that the effect of this agent has been overestimated. Unpublished trials found smaller effects than published trials, and formal testing confirmed publication bias. Caution Too few patients were included in the trials to determine effects on clinically relevant outcomes, such as need for dialysis. Implication Sodium bicarbonate is probably less effective at preventing contrast-induced nephropathy than is currently thought. Routine use of sodium bicarbonate for prevention of contrast-induced nephropathy is therefore premature. The Editors Contrast-induced nephropathy (CIN), which is the development of acute renal failure after administration of radiocontrast in the absence of other identifiable causes, is a leading cause of hospital-acquired acute kidney injury (1). It is defined as an increase in baseline serum creatinine level of 25% or an absolute increase of 44 mol/L (0.5 mg/dL). In addition, CIN accounts for 10% of all cases of acute kidney injury requiring hospitalization (2). In its most severe form, CIN is associated with clinically significant morbidity and mortality, including prolonged hospitalization, requirement for dialysis, and an increased risk for death (3, 4). The implementation of strategies to prevent CIN is therefore an important area of research. However, no uniform approach has been advocated, with guidelines (2) generally recommending volume expansion but giving no firm recommendation on the role of pharmacologic agents. Contrast-induced nephropathy is rare in patients with normal kidney function; however, its incidence increases by 25% in patients with preexisting renal impairment, such as those with diabetes and congestive heart failure, and with concurrent administration of nephrotoxic agents (5). Radiocontrast agents are believed to produce nephrotoxicity through acute sustained vasoconstriction and reduced renal perfusion resulting in regional hypoxia and tubular cytotoxicity (6). To date, strategies to prevent CIN have targeted renal vasoconstriction, hypoxia-induced oxidative stress, and tubular acidification. Preprocedural intravenous hydration is routinely administered; however, the evidence to support this practice is not compelling (6). Vasodilating agents, including dopamine, fenoldopam, and theophylline, and the antioxidant N-acetylcysteine also have been studied. The results of these individual studies (79) have been heterogeneous; however, 2 meta-analyses (7, 9) exploring the efficacy of N-acetylcysteine and theophylline compared with hydration alone have reported an overall beneficial effect. Recent studies and meta-analyses suggest that intravenous sodium bicarbonate may protect against CIN (1016). This protection is thought to be conferred by alkalinization of renal tubular fluid and increased urine flow (1719). In addition, animal models suggest that sodium bicarbonate may protect against formation of reactive oxygen species in the kidney (20). However, the potential benefits of intravenous sodium bicarbonate have been challenged by other studies suggesting no benefit (2123) or harm, with an increased risk for CIN compared with N-acetylcysteine or no treatment (24). The inconsistency between these findings highlights the need for a comprehensive systematic overview of all trials using intravenous sodium bicarbonate. We aimed to assess the effectiveness and safety of sodium bicarbonatebased treatment regimens for the prevention of CIN and clinical outcomes and to provide a reliable estimate of the nature and strength of any treatment effect. Methods Data Sources and Searches We performed a systematic review of the available literature according to the QUORUM (Quality of Reporting of Meta-analyses) guidelines for the conduct of meta-analyses of intervention studies. We identified relevant studies through electronic searches of MEDLINE via Ovid, PubMed, EMBASE, and the Cochrane Central Register of Controlled Trials from 1950 through December 2008. We used relevant text words and Medical Subject Heading terms that included all spellings of contrast media combined with nephropathy, renal insufficiency, nephritis, and nephrotoxicity (Appendix). We limited the search to clinical trials but did not restrict by language. To identify other relevant studies, we manually scanned reference lists from identified trials and review articles (11, 1315), and we also searched ClinicalTrials.gov and conference proceedings. We requested original data by directly contacting authors or principal investigators. Study Selection Two authors independently conducted the literature search, data extraction, and quality assessment by using a standardized approach. All completed randomized, controlled trials assessing preventive strategies for CIN that included intravenous sodium bicarbonate in 1 of the treatment groups were eligible for inclusion. We defined CIN as a 25% increase in baseline serum creatinine level or an absolute increase of 44 mol/L (0.5 mg/dL) 2 to 5 days after radiocontrast administration. We excluded studies with participants younger than 18 years. We did not restrict eligibility according to kidney function. Data Extraction and Quality Assessment Extracted data included patient characteristics (mean age, sex distribution, diabetes or hypertension status, and mean baseline creatinine level); type of imaging; inclusion and exclusion criteria; type and dose of contrast media; periprocedural hydration protocol; specific definition of CIN; treatment dose; serum creatinine level after radiocontrast injection; and the outcomes of requirement for dialysis, heart failure, and death. Quality assessment was judged on concealment of treatment allocation; similarity of both groups at baseline regarding prognostic factors; eligibility criteria; blinding of outcome assessors, care providers, and patients; completeness of follow-up; and intention-to-treat analysis (25). We quantified study quality by using the Jadad score (26). A third reviewer adjudicated any disagreement about abstracted data. Data Synthesis and Analysis We calculated relative risks and 95% CIs for individual studies before pooling data. We obtained summary estimates of overall and subgroup relative risk ratios by using a random-effects model. When either or both treatment groups of a study had no events, we added the reciprocal of the size of the opposite treatment group to each cell of the 22 table as a continuity correction factor (27). We also conducted sensitivity analyses by using continuity correction constants of various sizes (for example, 0.0001, 0.001, and 0.01) to ensure that the findings were robust. We estimated the percentage of variability across studies attributable to heterogeneity beyond chance by using the I 2 statistic (28). Publication bias was assessed by using the Egger test and represented graphically by using Begg funnel plots of the natural log of the relative risk versus its standard error (29). We explored potential heterogeneity in estimates of treatment effect attributable to each quality criterion for published studies only by using univariate meta-regression (28). We considered a P value less than 0.05 to be statistically significant for all analyses. We performed all statistical analyses with STATA, version 9.2 (Stata, College Station, Texas). Role of the Funding Source This study did not receive funding. The corresponding author, on behalf of all authors, had full access to all data in the study and had final responsibility for the decision to submit the manuscript for publication. Results Literature Search and Study Characteristics The literature search yielded 1231 articles, of which 163 were reviewed in full text on the basis of our inclusion criteria (Appendix Figure 1). Of these, 23 studies (including information on 3563 participants and 396 CIN events) were eligible for inclusion (Figure 1): 14 studies were not yet published in peer-reviewed journals but were presented at scientific sessions and reported in abstract form in conference proceedings or obtained directly from the investigators. Appendix Table 1 summarizes the characteristics of the included studies, all of which were reported since 2004. Sample size ranged from 18 to 502 participants, and total events accrued ranged from 2 to 56. Eight studies were performed in the Americas; 5 in Asia; 4 in Europe; 3 in Iran; and 1 each in Israel, Turkey, and Tunisia. Only 2 studies included patients with normal renal function; therefore, separate evaluation of patients according to baseline renal function was not possible. Mean baseline serum creatinine level ranged from 71 to 177 mol/L (0.8 to 2.0 mg/dL). Seventeen studies evaluated patients having cardiac catheterization, 5 studies evaluated patients having either cardiac catheterization or scheduled computed tomography or other arteriography, and 1 study evaluated patients having scheduled computed tomography or other arteriography. Ten studies compared sodium bicarbonate with sodium chloride; 8 studies compared sodium bicarbonate and N-acetylcysteine with sodium chloride and N-acetylcysteine; 2 studies compared sodium bicarbonate with and without N-acetylcysteine versus sodium chloride with and without N-acetylcysteine; and 1 study each compared sodium bicarbonate versus sodium chloride with N-acetylcysteine, sodium bicarbonate with and without N-acetylcysteine versus sodium chloride with N-acetylcysteine, and sodium bicarbonate with oral acetazolamide versus sodium chloride alone. The sodium bicarbonate and hydration protocols varied among the studies and treatment groups (Appendix Table 1). The average age of the participants in all studies was older than 48 years, and the proportion of men ranged from 59% to 84%. All studies included participants with diabetes (range, 24% to 58%), and 4 stud