Study Characteristics

Seventy eight studies (67 RCTs and 11 observational studies) were identified that compared N-acetylcysteine with IV saline. Of these, 74 reported on CIN directly, and three reported on serum creatinine or glomerular filtration rate without reporting the incidence of CIN. Of the studies reporting on CIN directly, we found 54 RCTs that compared N-acetylcysteine plus IV saline with IV saline with or without placebo, published between 2002 and 2014, which we included in a meta-analysis. The number of patients in each trial ranged from 40 to 3382, and the study populations were very heterogeneous across the studies. Study patients had renal dysfunction at baseline (defined as baseline serum creatinine greater than 1.2 mg/dl) in 35 studies.^28-62 The mean age of patients included in the studies was 55 to 79 years, the mean percentage of patients with diabetes was 39 percent (range 0% to 100%), and the mean percentage of females was 32 percent (range 12% to 59%).

Across all of the studies included in the meta-analysis, 4749 patients received IV saline with or without placebo, and 4775 received N-acetylcysteine. The route and dose of N-acetylcysteine varied between studies. Forty studies administered N-acetylcysteine orally,^{28-33,36-43,45-47,49,50,52-56,59-74} 13 administered it intravenouly,^{34,35,44,48,51,57,58,75-80} and one used a combination of IV and oral N-acetylcysteine.⁸¹ Thirty-four studies,^{28-36,39,41-47,49-52,56,59-63,65,67,68,70,71,74,78} used a low-dose of N-acetylcysteine (1200 mg/day or less), and 18 studies used a higher dose (greater than 1200 mg/day)^{37,38,40,48,53-55,57,58,64,66,69,75-77,79-81} One study had one arm with low-dose N-acetylcysteine, a second arm with high-dose N-acetylcysteine, and a control arm that received a placebo in IV saline.⁸¹

Contrast media was administered intravenously in seven studies,^{36,44,49,57,62,68,79} not described in one study,⁴⁶ and intra-arterially in the remaining studies. Seven studies used IOCM,^{32,36,39,69,70,76} six used either IOCM or LOCM;^{28,29,60,67,69,79} one used IOCM, LOCM, or HOCM;⁶⁹ one did not report the contrast media type,⁷³ and the remainder used LOCM.

Variation existed in the protocols for giving fluids, with studies using 0.45 percent saline; normal saline; 5 percent dextrose in normal saline, or alone; or Ringer's lactate solutions. The studies administered varying volumes and used three definitions of CIN: 0.5 mg/dl absolute increase, 25 percent increase in serum creatinine, and a combination of both. All of the studies except three measured the change in serum creatinine between 48 and72 hours. One measured the change in serum creatinine at 24 hours,⁴⁸ one measured it between 48 and 96 hours,⁶⁹ and one study measured the change five days after contrast media administration⁷¹ (Appendix E, Evidence Table E-4).

Contrast-Induced Nephropathy

The 54 RCTs comparing N-acetylcysteine plus IV saline to IV saline with or without placebo in the reduction of CIN showed a range of results included in the meta-analyses: seven reported a clinically important reduction in the risk of CIN that was statistically significant, 20 reported a clinically important reduction in the risk of CIN that was not statistically significant, 10 did not show a clinically important reduction in the risk of CIN, 12 did not show a clinically important increased risk of CIN, two showed a clinically important increased risk of CIN that was not statistically significant, and three showed a clinically and statistically significant increased risk of CIN.

The pooled risk ratio of CIN, using the DerSimonian and Laird random effects model, was 0.78 (95% CI: 0.59 to 1.03) for high-dose N-acetylcysteine (greater than 1200 mg/day), indicating that, on average, the effect is at a level consistent with a clinically unimportant reduction in CIN (Figure 3). There was moderate statistical heterogeneity across studies with an I-squared of 38%. The pooled risk ratio for CIN from the studies using intra-arterially administered contrast media and high-dose N-acetylcysteine was 0.78 (95% CI: 0.55 to 1.12) (high-dose N-acetylcysteine with intra-arterial contrast media administration pooled risk ratio was run with Knapp-Hartung method). Two studies used IV contrast media and high-dose N-acetylcysteine, and their results were too imprecise to draw conclusions (pooled risk ratio 0.55; 95% CI: 0.12 to 2.62). Using Harbord's modified test for small study effects, we did not find evidence of asymmetry in results by study precision (bias coefficient of -0.61, standard error of 0.66, p=0.37). The strength of evidence was low that high-dose N-acetylcysteine with IV saline had a small clinically unimportant effect in preventing CIN compared with IV saline with or without placebo. (Table 3; see Appendixes F and G for study limitations).

Figure 3

Meta-analysis of high-dose* N-acetylcysteine plus IV saline versus IV saline with or without placebo for the prevention of contrast-induced nephropathy. * High-dose N-acetylcysteine refers to studies that administered more than 1200mg N-acetylcysteine (more...)

Table 3

Summary of the strength of evidence: N-acetylcysteine plus IV saline versus IV saline with or without placebo.

The pooled risk ratio for CIN using a random effects model for low-dose N-acetylcysteine (1200 mg/day or less) was 0.75 (95% CI: 0.63 to 0.89), indicating that, on average, the small effect is consistent with a borderline clinically important reduction in CIN (Figure 4). The statistical heterogeneity of the studies was low, with an I-squared of 0%. The pooled risk ratio using the Knapp-Hartung method for the studies using IV contrast media and low-dose N-acetylcysteine was 0.62, but in this small subset of five studies, the confidence interval was so wide that we cannot rule out a clinically important increased risk (95% CI: 0.18 to 2.10). For studies using intra-arterially administered contrast media and low-dose N-acetylcysteine, the pooled risk ratio was 0.77 (95% CI: 0.66 to 0.91) indicating that, on average, the benefit is at a level consistent with a clinically unimportant reduction in CIN. Using Harbord's modified test for small study effects, we did not find evidence of asymmetry in results by study precision (bias coefficient of -0.70, standard error of 0.44, p=0.123). Overall, the strength of evidence was low that low-dose N-acetylcysteine with IV saline had a small clinically unimportant effect in preventing CIN compared with IV saline with or without a placebo (Table 3; see Appendixes F and G for study limitations).

Figure 4

Meta-analysis of low-dose* N-acetylcysteine plus IV saline versus IV saline with or without placebo for the prevention of contrast-induced nephropathy. * Low-dose N-acetylcysteine refers to studies that administered 1200mg or less of N-acetylcysteine (more...)

We performed stratification analyses to investigate the influence of contrast media osmolality on the effect of N-acetylcysteine. The pooled risk ratio of CIN, using a random effects model, for studies using LOCM was 0.69 (95% CI: 0.58 to 0.84), indicating that, on average, the difference is consistent with a clinically important reduction in CIN with N-acetylcysteine in patients receiving LOCM, but the confidence interval does not rule out a clinically unimportant difference (Figure 5).The statistical heterogeneity across studies was low, with an I-squared of 19 percent. The strength of the evidence was moderate that in patients receiving LOCM, N-acetylcysteine with IV saline had a clinically important reduction in CIN. The pooled risk ratio for CIN from studies of N-acetylcysteine using IOCM was 1.12 (95% CI: 0.74 to 1.69). The confidence interval was wide enough for N-acetylcysteine when IOCM was used to suggest possible harm without any indication of a clinically important benefit (Figure 6). The strength of the evidence was low that in patients receiving IOCM, N-acetylcysteine with IV saline did not have a clinically important decrease in CIN. The estimates of effect are remarkably stable across different types of studies with a 20 to 30 percent reduction, which is near the edge of what we defined to be a minimally important difference. The variation is mainly in the CIs, which is likely due to variation in the number of people in the different studies.

Figure 5

Meta-analysis of N-acetylcysteine plus IV saline versus IV saline with or without placebo for the prevention of contrast-induced nephropathy when low-osmolar contrast is used. %=percent; CAD=coronary artery disease; CI=confidence interval; CIN=contrast (more...)

Figure 6

Meta-analysis of N-acetylcysteine plus IV saline versus IV saline with or without placebo for the prevention of contrast-induced nephropathy when iso-osmolar contrast is used. %=percent; CAD=coronary artery disease; CI=confidence interval; CIN=contrast (more...)

We also performed stratification analyses to investigate the influence of the route of N-acetylcysteine administration. The pooled risk ratio for CIN, using a random effects model, for patients who received oral N-acetylcysteine was 0.77 (95% CI: 0.65 to 0.92), indicating that, on average, the difference is not clinically important. The pooled risk ratio for CIN for patients who received IV N-acetylcysteine (run with the Knapp-Hartung method) was 0.90 (95% CI: 0.72 to 1.12), indicating that the difference is not clinically important (Figure 7).

Figure 7

Meta-analysis of oral and IV route of N-acetylcysteine plus IV saline versus IV saline with or without placebo for the prevention of contrast-induced nephropathy. %=percent; ACS=acute coronary syndrome; CAD=coronary artery disease; CI=confidence interval; (more...)

Our sensitivity analysis, which removed one study at a time, did not show any significant impact on the estimated effect of N-acetylcysteine. When we examined the variation of risk ratio estimates according to baseline characteristics of the study population, we did not observe any meaningful difference by age, sex, baseline renal function, or the presence or absence of diabetes mellitus. There was no trend in the effect size by year of the study publication (Figure 7). When we examined how the results of studies of N-acetylcysteine varied in forest plots organized by the number of study limitations, we did not see any pattern indicative of a trend by study quality.

Thirteen of the 67 RCTs reporting on CIN were not included in the meta-analyses for a variety of reasons, including missing data, dosage differences, and inclusion criteria differences (see Appendix E, Evidence Table E-5).^67,82-90 In addition to the studies that reported on the incidence of CIN, three studies reported on changes in serum creatinine (Appendix E, Evidence Table E-6) and/or glomerular filtration rate (Appendix E, Evidence Table E-7) without reporting the incidence of CIN.^91-93 In those nine studies, the mean change in serum creatinine or glomerular filtration rate did not differ enough between groups to meet the definition of CIN.

Eleven observational studies were included in the studies we reviewed.^94-104The results of the observational studies were similar to those reported in the RCTs.

Other Outcomes

Of the 77 studies investigating development of CIN when comparing N-acetylcysteine plus IV saline with a placebo with or without IV saline, 35 also included data on secondary outcomes. Twenty eight reported patients' needs for renal replacement therapy,^{28,30,33,35,37-39,41,44-46,51,53,55,56, 59,61,69-71,80-85,87,89} seven reported cardiac events,^{31,38,40,53,70,71,82} 14 reported mortality,^{30,35,38,39,41, 44,53,59,69,76-78,81,83} and nine reported length of hospitalization (Appendix E, Evidence Table E-8).^{35,47,56,64, 71,76-78,83}

Of the 20 studies that examined the need for renal replacement therapy, only seven reported p-values and one reported a statistically insignificant, and clinically non-significant difference between groups (risk ratio: 0.87; 95% CI: 0.17-4.35).⁶⁹ The remaining studies reporting on the need for renal replacement therapy did not report statistics. One study, Marenzi et al.,2006,⁸¹ reported a statistically significant and clinically important difference in mortality between the placebo arm and the N-acetylcysteine arms, with more in-hospital deaths in the placebo arm (placebo: 13/119 (11%); standard dose N-acetylcysteine: 5/115 (4%); high-dose N-acetylcysteine: 3/118 (3%), p=0.007).⁸¹ Two studies reported significant findings for length of hospitalization. Hsu et al., 2007⁷¹ showed a statistically significant and clinically important reduction in length of hospitalization in the N-acetylcysteine arm (placebo: mean 8.1 days, standard deviation (SD) 4.1); low-dose N-acetylcysteine arm (mean 5.2 days, SD 1.5); p=0.04)).⁷¹ Kay et al., 2003⁴⁷ also showed a statistically significant reduction in length of hospitalization in the N-acetylcysteine arm, but the difference was not clinically important (placebo: mean 3.9 days, SD 2.0); low-dose N-acetylcysteine: mean 3.4 days, SD 0.9: p=0.02).⁴⁷ No clinically important or statistically significant differences were reported for cardiac events.

Overall, the strength of evidence was low that N-acetylcysteine plus IV saline did not differ from IV saline without N-acetylcysteine in the need for renal replacement therapy, cardiac events, or the length of hospitalization. (Table 3; Appendix E, Evidence Table E-8; see Appendix G for study limitations). Most of the studies addressing these outcomes had at least one important study limitation (frequently lacking documentation of allocation concealment or blinding of participants and personnel). The results generally were consistent in the direction of impact of N-acetylcysteine. However, the effect estimates were imprecise. The studies addressing mortality had insufficient strength of evidence to support a conclusion because they had important study limitations, with inconsistent and imprecise effect estimates.

Study Characteristics

Thirty articles were identified that compared IV sodium bicarbonate with IV saline (28 RCTs and 2 observational studies). Nineteen RCTs^{36,46,56,58,70,74,109-121} published between 2004 and 2014 were included in the meta-analysis; the two observational studies were not included in the meta-analysis.^122,123

In these studies, CIN was defined three ways (Appendix E, Evidence Tables E-1, E-3, E-10): five defined it as a 25 percent or greater increase in serum creatinine, one defined it as a 0.5 mg/dl or greater increase in serum creatinine, and seven defined it as either a 25 percent or greater increase or a 0.5 mg/dl or greater increase in serum creatinine.

A total of 1748 patients were included in the control arms, and 1750 patients were included in the sodium bicarbonate arms. The mean age of patients was 65.8 years (range 59 to 77 years). The mean percentage of diabetes patients was 44 percent (range 6–100%) and the mean percentage of female patients was 29.4 percent (range 5–48%). Contrast media administration was intra-arterial in fourteen studies,^{36,56,58,70,74,109,111-113,115-117,119-121} IV in two studies, ^110,114 both IV and intra-arterial in three studies.^46,110,118 Two studies used IOCM,^36,115 and the other studies used LOCM (Appendix E, Evidence Tables E-2, E-10).

Contrast-Induced Nephropathy

Six studies concluded that IV sodium bicarbonate administration reduced the incidence of CIN when compared with IV saline, while thirteen reported no difference in the incidence of CIN between the IV sodium bicarbonate and IV saline intervention arms. The meta-analysis indicated that administration of IV sodium bicarbonate did not differ from IV saline in the risk of CIN (pooled risk ratio 0.93; 95% CI: 0.68 to 1.27), with a point estimate indicating a difference that was not clinically important, and a wide confidence interval that did not rule out the possibility of an important reduction or important increase in CIN (see Figure 8). However, as shown in Figure 8, IV sodium bicarbonate with IV saline was more effective than IV saline in preventing CIN, with a clinically important benefit, in a subset of 11 studies using LOCM (pooled risk ratio 0.65; 95% CI: 0.33 to 1.25), but not in the subset of 7 studies using IOCM (pooled risk ratio 1.02; 95% CI: 0.70 to 1.48). The strength of evidence was low for these conclusions (Table 4; see Appendixes F and G for study limitations) because many of the studies reporting on CIN had important study limitations (frequently lacking allocation concealment or blinding of participants and personnel), and the results were inconsistent. Overall, the studies had moderate heterogeneity, with an I-squared of 33 percent (p=0.07) (Figure 8). Using Harbord's modified test for small study effects, we found no evidence of asymmetry in the distribution of results by study precision (bias coefficient of -0.55, standard error of 0.96, p = 0.57).

Figure 8

Meta-analysis of IV sodium bicarbonate versus IV saline for the prevention of contrast-induced nephropathy. %=percent; 1/2NS=0.45% saline; CHF=congestive heart failure; CI=confidence interval; CIN=contrast induced nephropathy; CKD=chronic kidney disease; (more...)

Table 4

Summary of the strength of evidence: IV sodium bicarbonate versus IV saline.

For a variety of reasons, 8 of the RCTs reporting on CIN were not included in the meta-analysis (Appendix E, Evidence Table E-11).^124-131 One study did not report on CIN as an outcome, but did report on serum creatinine. The mean change in serum creatinine from baseline in this study did not meet any definition of CIN (Appendix E, Evidence Table E-12).

There were two observational studies, and they both reported the benefits of sodium bicarbonate administration to reduce CIN. A study by Tamai et al.¹²² reported a significant difference in CIN for patients who received a high dose of sodium bicarbonate (833mEq/L) versus those who received a low dose (160 mEq/L). The study by Buhiraja et al.¹²³ showed a significant difference in CIN in patients who received sodium bicarbonate versus those who received normal saline. We did not factor the observational studies into the strength of evidence since the outcomes were in the same direction as the RCTs.

Other Outcomes

Of the studies that compared the risk of CIN using IV sodium bicarbonate with the risk of CIN using IV saline, 13 included data on secondary outcomes. Of these, 11 reported participants' needs for renal replacement therapy,^{46,56,70,110-112,115-117,119,130} four reported on cardiac events,^{56,70,114,115} three reported on hospitalization or length of stay,^110,112,120 and six reported on mortality.^{110-112,115,117,120} (Appendix E; Evidence Table E-13). The overall strength of evidence was low that the mortality rates and the need for renal replacement therapy did not differ between IV sodium bicarbonate and IV saline (Table 4; see Appendixes F and G for study limitations). The studies addressing the need for renal replacement therapy and mortality had medium study limitations, were consistent in the direction of effect, and were imprecise, due to wide confidence intervals and small study populations. Only one study reporting on cardiac outcomes¹¹⁴ reported a statistically significant difference between groups in favor of IV sodium bicarbonate (p=0.03). The remainder of the studies either reported statistically insignificant differences between groups or did not report statistics. The evidence was insufficient to determine whether or not cardiac events or length of hospitalizations differed between IV sodium bicarbonate and IV saline (Table 4; Appendix E, Evidence Table E-13).

Adverse events were reported in 11 studies. Data were only recorded if specific adverse events were reported or if the study reported no adverse events (Appendix E, Evidence Table E-14). Adverse events were not reported in a standardized manner and were rarely analyzed in these studies. As a result, we were unable to draw any firm conclusions as to whether or not the incidence of adverse events differed between IV sodium bicarbonate and IV saline.

Study Characteristics

Our search identified seven RCTs^{36,46,56,58,70,74,132} with a total study population of 1619 that compared N-acetylcysteine plus IV saline with IV sodium bicarbonate (number analyzed=930) and two observational studies.^97,133 Contrast media included iodixanol,^36,58,70 ioversol,¹³² iohexol,^46,74 and ioxaglate.⁵⁶ Contrast media were administered intravenously in one study⁴⁶ and intra-arterially in the other six studies. The seven studies were completed between 2007 and 2014 and were conducted in the United States,⁵⁸ Italy,³⁶ Denmark,⁷⁰ Argentina,¹³² Iran,⁷⁴ and Turkey.^46,56 The mean age of patients in these studies ranged from 59 to73. The study population for three of the RCTs included only individuals with kidney dysfunction.^36,56,132 The patients in one study⁵⁸ had kidney dysfunction alone (17%), diabetes mellitus alone (59%), or both (24%). Patients in the study by Kama, et al.⁴⁶ were considered to be at moderate or high risk of developing CIN (73% had an estimated glomerular filtration rate of 60 mL/min/1.73 m² or less). Only 8 percent of the patients in the study by Thayssen et al.⁷⁰ had an estimated glomerular filtration rate less than 60 mL/min/1.73 m². The percentage of patients with diabetes mellitus ranged from 8.5 percent to 68 percent. The studies had a total follow up period of 48 hours to 30 days; the outcomes of CIN were reported at 48 hours;^56,74 at 48 to 72 hours;^46,70,132 at 24, 48, and 120 hours (5 days)³⁶ (personal communication with Diego Castini, April 28, 2014); and at 24, 48, and 168 hours (7 days).⁵⁸ (Appendix E, Evidence Tables E-1, E-3, E-15)

All studies compared N-acetylcysteine plus IV saline (sometimes in 5% dextrose in water) with IV sodium bicarbonate. However, in the studies by Thayssen⁷⁰ and Kama,⁴⁶ all arms also received IV normal saline.

Our search identified two observational studies^97,133 comparing N-acetylcysteine plus IV saline with IV sodium bicarbonate. There were 977 study participants. The first study was published in 2009 and was conducted in Israel,¹³³ and the other⁹⁷ was published in 2008 and conducted in the United States. The mean age of patients ranged from 60 to 71. All of the patients had comorbid disease at baseline in both studies.

Contrast-Induced Nephropathy

The incidence of CIN in the IV sodium bicarbonate groups ranged from 4.5 to 40.0 percent and from 4.7 to19.4 percent in the N-acetylcysteine plus IV saline groups. Three of the RCTs favored IV sodium bicarbonate, three favored N-acetylcysteine plus IV saline, and one was equivocal because it had very few CIN events⁵⁸ (Appendix E, Evidence Table E-16).

The overall pooled risk ratio for CIN in the RCTs comparing IV sodium bicarbonate with the combination of N-acetylcysteine and IV saline, using the Knapp-Hartung method, was 1.11 (95% CI: 0.51 to 2.41). The point estimate of the risk ratio indicates a very small increase in risk with sodium bicarbonate that was less than clinically important. The CI was too wide to rule out the possibility of either an important decrease or important increase in risk. The studies were inconsistent and had moderate heterogeneity, with an I-squared of 24 percent (Figure 9). The Harbord's modified test for small study effects did not show evidence of asymmetry in results by study precision (bias coefficient of -0.65, standard error of 1.80, p=0.735). The strength of evidence was insufficient to support a conclusion about the comparative effectiveness of these two interventions in the ability to prevent CIN (Table 5; Appendix E, Evidence Table E-16; see Appendixes F and G for study limitations).

Figure 9

Meta-analysis of N-acetylcysteine plus IV saline versus sodium bicarbonate for the prevention of contrast-induced nephropathy. %=percent; CI=confidence interval; N=sample size; NAC=N-acetylcysteine; NaHCO3=sodium bicarbonate; p=p-value; RR=risk ratio (more...)

Table 5

Summary of the strength of evidence: N-acetylcysteine plus IV saline versus sodium bicarbonate.

Limitations of this comparison included the small number of studies, the varying regimens of fluid administration and N-acetylcysteine dosing, and the variations in follow up time. Four of the studies were exclusively in individuals with kidney disease (a population at higher risk for CIN), although the inclusion criteria were not exactly the same across all studies. One of the RCTs was conducted in individuals with either kidney dysfunction or diabetes mellitus. Another potential concern with the Ratcliffe, et al. study⁵⁸ was that only 66 percent of the participants completed the study.⁵⁸

In the observational studies, the rate of CIN was similar in both groups' comparison groups. The results of the observational studies were similar to those reported in the RCTs regarding the comparison of the risk of CIN with N-acetylcysteine plus IV saline against IV sodium bicarbonate (Appendix E, Evidence Table E-16).

Other Outcomes

Of the seven RCTs that compared N-acetylcysteine plus IV saline with IV sodium bicarbonate for the development of CIN, five reported on secondary outcomes, including the need for renal replacement therapy, cardiac events, and mortality.^{36,46,56,70,132} However, insufficient evidence existed to support firm conclusions about the comparative effects of N-acetylcysteine versus sodium bicarbonate for the outcomes of need for renal replacement therapy, cardiac events, or mortality (Table 5, see Appendixes F and G for study limitations). In those studies, no statistically significant difference was reported, no cases were reported, or statistics were not reported.

Although all of these studies reported on specific adverse events or reported that there were no adverse events, adverse events were not reported in a standardized manner, and were rarely analyzed. Thus, we were not able to draw any firm conclusions about whether or not the incidence of adverse events differed between N-acetylcysteine with IV saline and IV sodium bicarbonate (Appendix E, Evidence Table E-18).

Study Characteristics

Our search identified 19 RCTs^137-150 and one observational study on statins (Appendix E, Evidence Tables E-1, E-3, E-19).¹⁵¹ The 19 RCTs included 10,574 participants. Eight studies compared statins with placebo,^{138,139,144,145,152-155} one compared statin plus N-acetylcysteine plus sodium bicarbonate with N-acetylcysteine plus sodium bicarbonate,¹³⁷ and four compared statin plus N-acetylcysteine plus saline with N-acetylcysteine plus saline.^{141,142,146,156} The remainder of the studies compared statin with statin,^143,148,149 statin plus saline with saline and chronic statin plus saline,¹⁴⁰ low-dose statin plus probucol with high-dose statin plus probucol,¹⁵⁰ and statin to statin plus probucol¹⁴⁷. Contrast media used included iodixanol,^137,142-146 iopromide,^138,148 iobitridol,¹³⁹ iohexol,^140,143 and iopamidol.^141,147,150 Contrast media were administered intra-arterially in all studies.

These studies were completed between 1997 and 2015 and were conducted in Italy,^{137,139,142,146} China,^{138,143,145,147,150,153,157,158} Turkey,^{140,141,148,154} Korea,^144,149,152 Iran,¹⁵⁵ and Egypt.¹⁵⁶ In all of the RCTs, the mean age of patients ranged from 54 to 76 years. The percentage of patients with chronic kidney disease at baseline ranged from 4 percent to 100 percent and the percent of patients with diabetes mellitus ranged from 15 percent to 100 percent (Appendix E, Evidence Tables E-1, E-3, E-19).

The observational study,¹⁵¹ with a study population of 28,871, compared statin therapy prior to the procedure with the absence of statin therapy. The contrast media used were not specified but all were administered intra-arterially. This study was completed between 1997 and 2003 and was conducted in the United States. In this study, the mean age of patients was 64. The percentage of patients with chronic kidney disease was not specified, while the percentage of patients with diabetes mellitus was 30 percent (Appendix E, Evidence Tables E-1, E-3, E-19).

Contrast-Induced Nephropathy

We conducted two separate meta-analyses on the studies of statins to reduce the incidence of CIN in patients receiving intra-arterial contrast. One included eight studies on statin-naïve patients that compared statin plus IV saline with IV saline alone.^{138,139,144,145,152-155} The other included five studies: four compared statins plus N-acetylcysteine plus IV saline with N-acetylcysteine plus IV saline,^{141,142,146,156} and one compared statins plus N-acetylcysteine plus IV sodium bicarbonate with N-acetylcysteine plus IV sodium bicarbonate.¹³⁷ The remaining six studies were not included in the meta-analyses; they either included comparisons that were not similar enough to analyze^143,147-150 or did not include a CIN outcome.¹⁴⁰(Appendix E, Evidence Table E-20).

When evaluating the efficacy of prophylactic statin administration compared with IV fluids alone in the prevention of CIN, four studies^{138,139,145,154} found both a statistically significant and clinically important reduction in CIN (above our 25% threshold for a minimally important difference) in the intervention arm. One study found a borderline clinically important difference.¹⁴⁴ Three studies did not show either a clinically or a statistically significant reduction.^152,153,155 The largest study of the group with positive findings (n=2998) found a significant reduction with statin administration in the general study population but not in the post-hoc subgroup analyses of statin naïve versus statin non-naïve participants.¹⁴⁵ This study had a high risk of bias based on the five criteria described in the methods for assessing risk of bias for individual studies (Appendix F), but its effect estimate was in the same direction as the other three studies in the meta-analysis (which had fewer study limitations). An additional study¹⁴² evaluated the occurrence of CIN in the nonstandard time frame of 5 days and therefore was not included in the meta-analysis; this study did not demonstrate a clinically or statistically significant difference between the intervention and control arms (Figure 10).

Figure 10

Meta-analysis of statins plus IV fluids versus IV fluids with or without placebo for the prevention of contrast-induced nephropathy in patients receiving intra-arterial contrast. %=percent; CI=confidence interval; CIN=contrast induced nephropathy; CKD=chronic (more...)

In a meta-analysis of the eight studies with a CIN endpoint ranging from 48 to 72 hours after contrast media administration,^{138,139,144,145,152-155} the pooled estimate of the effect of statin plus IV fluids compared with IV fluids alone demonstrated a clinically important but statistically insignificant reduced risk of CIN with statin use (pooled risk ratio 0.68; 95% CI: 0.39 to 1.20). A sensitivity analysis demonstrated that no study unduly influenced the overall statistical significance of the pooled estimate, and a stratified analysis showed no substantial difference in estimation of effect by statin type, as the point estimates of effect were all clinically important. No statin type had a 95% CI that was fully in the range consistent with a clinically important effect The estimate for rosuvastatin, from four studies (risk ratio 0.69; 95% CI: 0.47 to 1.02) was clinically important, but the CI was wide enough to not rule out the possibility of an unimportant effect.^{145,152,153,155} The estimate for atorvastatin, three studies (risk ratio 0.41; 95% CI: 0.02 to 2.71) was clinically important, but the CI was wide enough to not rule out the possibility of an unimportant effect. While the point estimate of the effect of simvastatin (risk ratio 0.75; 95% CI: 0.17 to 3.28) was not clinically important, the confidence interval was so wide that we cannot rule out the possibility of a clinically important benefit or harm. Note that atorvastatin was the only drug for which there was more than one study. A meta-regression was not conducted, due to the small number of studies. We saw no trends in the data that pointed to differences in groups by age, kidney function, diabetes status, or sex. The studies on statins had a medium risk of bias, and consistently showed a benefit in reducing CIN in favor of the statin drug with a relatively precise resulting estimate of the effect. Harbord's modified test for small study effects did not demonstrate evidence of asymmetry in results by study precision (bias coefficient of -1.49, standard error of 1.11, p=0.227). We concluded that the strength of evidence was low for demonstrating that a statin plus IV fluids was more effective than IV fluids alone at preventing CIN (Table 6; see Appendixes F and G for study limitations).

Table 6

Summary of the strength of evidence: statins plus IV fluids versus placebo with or without fluids and statins plus N-acetylcysteine versus N-acetylcysteine alone in patients receiving intra-arterial contrast.

When evaluating the efficacy of statin administration plus N-acetylcysteine plus IV saline (or IV sodium bicarbonate) compared with N-acetylcysteine plus IV fluids (or IV sodium bicarbonate) in the prevention of CIN, four studies^{137,141,146,156} found both a statistically significant and clinically important reduction in CIN (above our 25% threshold for a minimally important difference) in the statin arm. One study showed a statistically non-significant (p=0.86) reduction that was clinically insignificant.¹⁴²

In a meta-analysis of studies with a CIN endpoint,^{137,141,142,146} the pooled estimate of the effect of statin plus N-acetylcysteine plus IV fluids (saline or sodium bicarbonate) compared with N-acetylcysteine plus IV fluids (saline or sodium bicarbonate) demonstrated a clinically important and statistically significant reduced risk of CIN with statin use (pooled risk ratio 0.52; 95% CI: 0.29 to 0.93) with a number needed to treat of 18 (95% CI: 13.44 to 34.72) (see Figure 11). However, the CI for the risk ratio was wide enough that we cannot rule out the possibility of a clinically unimportant difference. A meta-regression was not conducted due to the small number of studies. We saw no trends in the data that pointed to differences in groups by age, kidney function, diabetes status, or sex. Harbord's modified test for small study effects did not demonstrate evidence of asymmetry in results by study precision (bias coefficient of -0.63, standard error of 1.68, p=0.735). We concluded that the strength of evidence was low for demonstrating that a statin plus N-acetylcysteine plus IV fluids was more effective than N-acetylcysteine plus IV fluids at preventing CIN, when considering study limitations, directness, consistency, and precision (Table 6; see Appendixes F and G for study limitations).

Figure 11

Meta-analysis of statins plus N-acetylcysteine plus IV fluids versus N-acetylcysteine plus IV fluids with or without placebo for the prevention of contrast-induced nephropathy in patients receiving intra-arterial contrast. %=percent; CI=confidence interval; (more...)

One study comparing atorvastatin to IV saline¹⁴⁰ did not report on CIN outcomes. This study reported on the change in serum creatinine and estimated glomerular filtration rate. No difference was reported in serum creatinine levels 48 hours after the procedure, and estimated glomerular filtration rate was significantly lower in the atorvastatin group 48 hours after the procedure (Appendix E, Evidence Table E-20).

Two studies reported on the incidence of CIN in participants receiving a statin versus a statin plus probucol.^147,150 Han, 2013¹⁵⁰ compared low-dose atorvastatin plus probucol with high-dose atorvastatin plus probucol as well as with high-dose atorvastatin. No significant difference in CIN incidence was found between the groups 48 hours after the procedure. Li, 2014¹⁴⁷ compared atorvastatin with atorvastatin plus probucol. No significant difference in CIN was reported between groups (Appendix E, Evidence Table E-20).

Three studies compared either different dosages of the same statin^143,149 or different statins.¹⁴⁸ Jo, 2014¹⁴⁹ found no significant difference between high-dose and low-dose atorvastatin in preventing CIN. Kaya, 2013¹⁴⁸ found no significant difference between atorvastatin and rosuvastatin in preventing CIN. Xinwei, 2009¹⁴³ found a significantly lower incidence of CIN in patients receiving high-dose simvastatin when compared with low-dose (Appendix E, Evidence Table E-20).

One observational study reported on statins versus IV saline and found a significant decrease in CIN in the group receiving statins.¹⁵¹ The results were similar to those reported in the RCTs comparing statins with IV saline.

Four articles published in Chinese and one in Arabic were reviewed to determine if findings published in non-English language journals were different than those published in English-language journals. Three studies compared statins with IV saline and found significantly significant reductions in CIN in the statin intervention group^159,160 or higher estimated glomerular filtration rate in the statin group (statistical significance not reported).¹⁶¹ These results were generally consistent with the English-language RCTs comparing statins with IV saline. One study compared low-dose statins with high-dose statins and found no significantly significant difference between groups.¹⁶² Another compared rosuvastatin plus furosemide with furosemide and found no significant difference in CIN incidence between groups.¹⁶³

Other Outcomes

Secondary outcome reporting was not consistent across studies. Need for renal replacement therapy was reported in three comparing statins to IV saline,^144,145,156 and three comparing statins plus N-acetylcysteine to N-acetylcysteine,^137,142,146 two comparing statins by dose of administration,^145,149 one comparing different statins.¹⁵⁷ One study comparing statins¹⁵⁷ and one comparing statin to IV saline reported on mortality.¹⁴⁵ Three comparing statins plus N-acetylcysteine to N-acetylcysteine, and one comparing statins by dose of administration¹⁴⁹ also reported on mortality. Only p-values were reported for need for renal replacement therapy and mortality and none reached a significance of p less than 0.05. Two studies reported on length of stay or hospitalization, both of which compared statins to IV saline.^139,144 One study showed no difference between groups while the other, Patti et al., 2011¹³⁹ showed a statistically significant difference (p=0.007) favoring the use of statins. Cardiac events were reported in five studies, two for statins versus IV saline,^145,157 two for statins plus N-acetylcysteine versus N-acetylcysteine,^146,156 and one compared statins by dose.¹⁴⁹ Statistical significance was reported only in the set of three studies comparing statins to IV saline. Two of these studies reported no statistically significant difference between groups,^146,164 and the other reported a statistically significant difference (p=0.02) in favor of statins.¹⁴⁵ Two studies comparing statins to IV saline reported on hospital length of stay reporting no comparisons between groups.^139,144 The strength of evidence was insufficient regarding whether or not statins had an impact on any of these secondary outcomes (Table 6; Appendix E, Evidence Table E-21; see Appendixes F and G for study limitations). No clinically important or statistically significant differences were seen in the need for dialysis; very few events were reported.^{137,142,144-146,149,150,156,157} Five studies reported cardiac outcomes^{145,146,149,156,157} and did not report consistently across outcomes. Of the six studies that reported mortality by intervention group, none showed a statistically significant or clinically important difference; the strength of evidence was insufficient, however, because very few deaths were reported, with results that were too imprecise and inconsistent.^{137,142,145,146,149,157} The strength of evidence was insufficient to determine if statins were effective at reducing length of hospitalization (Table 6; Appendix E, Evidence Table E-21; see Appendix G for study limitations).^139,144

Adverse events were reported in five studies. We were not able to draw any conclusions as to whether or not the incidence of adverse events differed between statins and IV fluids (Appendix E, Evidence Table E-22).¹⁴³

Study Characteristics

We found a total of five studies that reviewed the role of adenosine antagonists in the prevention of CIN: four examined theophylline,^{31,68,167,168} and one examined aminophylline.⁶⁶ All five were RCTs. One⁶⁸ used IV contrast media and the others used contrast media that were administered intra-arterially.^{31,66,167,168} Four studies used LOCM agents, ^{66,68,168 31} and one used IOCM.¹⁶⁷ All studies used IV saline prior to and after the procedure, and administered intervention drugs prior to and after the procedure. Two studies used elevated serum creatinine as an inclusion criterion,^31,167,168 one included only those with at least one risk factor for CIN,¹⁶⁸ one used coronary artery disease as an inclusion criterion,⁶⁶ and one included a population without kidney disease or diabetes mellitus.⁶⁸ The followup for all of the studies was between 48^31,66,167 and 72 hours^68,168 for CIN outcomes (Appendix E, Evidence Tables E-1, E-3, E-23).³¹ The studies were published from 2008⁶⁸ through 2012.¹⁶⁸ (Appendix E, Evidence Tables E-1, E-3, E-23). Four of the studies had more than one important study limitation,^31,68 and one had low risk of bias based on the five criteria described in the methods for assessing risk of bias for individual studies (Appendix F).¹⁶⁸ Some of the studies had low scores for allocation generation,^31,68 allocation concealment,^31,66,68 masking of intervention,^31,66,68 and incomplete outcome reporting.^68,167

We identified one observational study that compared an adenosine antagonist with IV saline in 52 patients.¹⁶⁹ The country of origin was not identified in this study. The average age ranged from 71 to 72, 44 percent of patients had diabetes mellitus, and all patients had been diagnosed with renal insufficiency.

Contrast-Induced Nephropathy

Regarding the intra-arterial administration of contrast media: the results of our primary analysis were mixed with regard to the incidence of CIN with adenosine antagonists plus IV saline compared with IV saline. Of the three studies that only examined theophylline against IV saline, two showed a clinically important increase in CIN in the theophylline group that was not statistically significant,^68,167 and one demonstrated a clinically important reduction in CIN in the theophylline group that was statistically significant.¹⁶⁸ Other studies compared intra-arterial administration of contrast media containing multiple comparison arms.^31,66 In the two studies with multiple comparisons, the arms involving the adenosine antagonists had less CIN than the IV saline arms; however, one study³¹ examined theophylline in combination with N-acetylcysteine and not on its own (Figure 12).

Figure 12

Meta-analysis of adenosine antagonists plus IV saline versus IV saline for the prevention of contrast-induced nephropathy. %=percent; CI=confidence interval; CIN=contrast induced nephropathy; N=sample size; NS=normal saline (0.9%); p=p-value; RR=risk (more...)

In the meta-analysis exploring all studies involving a comparison between adenosine antagonists plus IV saline and IV saline alone, the confidence interval was so wide that we could not rule out a clinically important decrease or increase (pooled risk ratio with Knapp-Hartung method, 0.80; 95% CI: 0.01 to 44.48) (Figure 12). The strength of evidence was insufficient to support a conclusion about the effect of adenosine agonists on the risk of CIN because the study results were imprecise and inconsistent, and the study limitations were medium (Table 7; see Appendix G for study limitations).

Table 7

Summary of the strength of evidence: adenosine antagonists plus IV saline versus IV saline.

Only one study⁶⁸ examined the effect of theophylline in a population for which contrast media was administered IV. It demonstrated a clinically important increased risk of CIN with theophylline that was not statistically significant (Figure 12).

One of the studies was not included in our meta-analysis.³¹ It included N-acetylcysteine in one of the interventions and the p-value was calculated across the three arms (Appendix E, Evidence Table E-24).

The results of the observational studies were similar to those reported in the RCTs regarding the comparison of the risk of CIN with aminophylline versus IV saline.¹⁶⁹

Other Outcomes

Four of the five studies reporting on adenosine antagonists reported on other outcomes. Two studies reported no events for the need for renal replacement therapy, cardiac events, mortality, and length of stay.^31,167 Two additional studies reported no cardiac events.^68,168 The strength of evidence was insufficient to determine the effect of adenosine antagonists on the need for renal replacement therapy, cardiac events, length of hospital stay or mortality (Table 7; Appendix E, Evidence Table E-25; see Appendix G for study limitations).

Adverse events were not reported in a standardized manner and were rarely analyzed, so we were unable to draw any conclusions around whether or not the incidence of adverse events differed between adenosine antagonists versus fluids (Appendix E, Evidence Table E-26).

Study Characteristics

Our search identified six RCTs on use of hemodialysis or hemofiltration with a total study population of 790 patients. These trials compared renal replacement therapy with IV fluids; four assessed the use of hemodialysis^59,172-174 and two assessed the use of hemofiltration.^175,176 All of the studies included patients with chronic kidney disease who were undergoing cardiovascular interventions. Only one study included patients undergoing additional procedures.¹⁷³ In all of the studies, contrast media included LOCM and was administered intra-arterially (two studies also administered it intravenously).^172,173 These studies were completed between 1998 and 2007 and were conducted in Germany,^59,172,174 Italy,^175,176 and Switzerland.¹⁷³ The mean age of patients ranged from 57 to 70. All studies included patients with different stages of chronic kidney disease at baseline; the percentage of patients with diabetes mellitus ranged from 23 to 64 percent.

Our search identified three observational studies with a total study population of 503 patients; these studies compared renal replacement therapy with IV fluids; one study assessed the use of hemodialysis¹⁷⁷ and two assessed the use of hemofiltration.^178,179 All studies included patients with chronic kidney disease who were undergoing cardiovascular interventions. Contrast media included LOCM in all studies and was administered intra-arterially in all studies. These studies were completed between 1991 and 2013 and were conducted in Japan^177,179 and Italy.¹⁷⁸ The mean age of patients ranged from 69 to 83. All studies included patients with different stages of chronic kidney disease at baseline, and the percentage of patients with diabetes mellitus ranged from 41 to 68 percent. Hemodialysis was started in all of the studies after the contrast media was administered, while hemofiltration was started before contrast media administration; some of the hemofiltration studies started hemofiltration both before and after contrast media administration, to evaluate the effects of timing^176,178 (Appendix E, Evidence Tables E-1, E-3, E-27). All studies had important study limitations based on the five criteria described in the methods for assessing risk of bias for individual studies (Appendix F).¹⁷⁶ All studies had an increased risk of bias because of the absence of blinding of the allocated intervention. Some studies were limited by problems with allocation generation,^59,172-174 allocation concealment,^{59,172-174,175} and incomplete outcome reporting.^172,173,175

Contrast-Induced Nephropathy

None of the studies on hemodialysis reported a statistically significant difference between the use of IV fluids and hemodialysis in preventing CIN.^172-174 The incidence of CIN was similar in both groups for all of the studies comparing hemodialysis and IV saline. The only study assessing hemodialysis plus IV glucose and saline⁵⁹ found that patients on hemodialysis had higher rates of CIN at 72 hours than those on IV saline only and those receiving N-acetylcysteine (15.9% vs. 6.1% and 5.3%; p = 0.008), but this study also found that when the rate of CIN was reassessed thirty to sixty days later, this effect had disappeared. Because this study measured creatinine at time points that were different from the other studies, the studies were not comparable (Appendix E, Evidence Table E-27).⁵⁹ The pooled analysis using the Knapp-Hartung method for the three studies comparing hemodialysis with IV saline yielded a pooled risk ratio of 1.50, which is consistent with a clinically important increased risk (95% CI: 0.56 to 4.04, Figure 13).

Figure 13

Meta-analysis of hemodialysis versus IV fluids for the prevention of contrast-induced nephropathy. %=percent; CI=confidence interval; CIN=contrast-induced nephropathy; CKD=chronic kidney disease; Cr=creatinine; IV=intravenous; LOCM=low-osmolar contrast (more...)

The studies indicated that prophylactic hemodialysis does not prevent the incidence of CIN in patients with chronic kidney disease, regardless of the stage, the duration of the dialysis (from 2 to 4 hours), or the time between contrast media administration and initiation of dialysis. No benefit was found when hemodialysis was started before the contrast media was given.¹⁷⁴ The two studies that included results on contrast media clearance^172,174 demonstrated that peak levels of contrast media were lower in the hemodialysis group than in the control group during the initial hours after contrast media administration, but also showed that the effect of dialysis was no longer significant after 72 hours; after 72 hours, elimination half-life was comparable in both arms. This finding correlated with the lack of a clinical effect (Appendix E, Evidence Table E-29). The strength of evidence was low that hemodialysis does not reduce the risk of CIN and may even be harmful, because the effects of hemodialysis were consistent and direct but imprecise, the magnitude of effect was weak, and the study limitations were high (Table 8; see Appendixes F and G for study limitations).

Table 8

Summary of the strength of evidence: renal replacement therapy versus fluids.

The study by Frank et al.¹⁷⁴ was not included in the pooled analysis because it did not provide data for the incidence of CIN. It only reported an insignificant difference between arms (Appendix E, Evidence Table E-28).

The only observational study addressing this comparison showed that patients on hemodialysis had higher rates of CIN than those on IV saline, with a more harmful effect shown in those with more deteriorated renal function.¹⁷⁷

The studies comparing hemofiltration with IV fluids reported that patients with severe chronic kidney disease may have a lower incidence of CIN. In these studies, this benefit was evident only when hemofiltration was started before contrast media administration. As Marenzi et al.¹⁷⁶ showed, when hemofiltration was started after the contrast media administration, its benefit was lost and the risk for developing CIN was comparable to patients receiving IV saline only. This effect was confirmed by the observational studies. While one RCT of hemofiltration included more than 50 patients with stage 3 to 4 chronic kidney disease per arm and the other RCT included about 30 patients per arm with severe chronic kidney disease, the conclusions were similar (Appendix E, Evidence Table E-29). The Harbord's modified test for small study effects did not show evidence of asymmetrical effects by study size (bias coefficient of 4.36, standard error of 5.90, p=0.595).

The evidence was insufficient to determine whether or not hemofiltration reduced the risk of CIN in patients with pre-existing severe chronic kidney disease, because of high study limitations, small study size, and the concern that both studies were from the same authors (i.e., they were not independently replicated). The hemofiltration studies were not combined with the hemodialysis studies in the pooled analysis due to their different designs.

Other Outcomes

Five of the studies on renal replacement therapy reported on other outcomes.^173-176 Four reported on the need for renal replacement therapy; two hemodialysis studies,^59,173 and two hemofiltration studies ^175,176 Three reported on cardiac outcomes; two hemodialysis studies^173,174 and one hemofiltration studies.¹⁷⁶ Four reported on mortality; Two hemodialysis studies,^59,173 and two hemofiltration studies.^175,176

The studies comparing hemofiltration with IV saline demonstrated that patients may benefit from hemofiltration because they have a lower risk of emergency renal replacement therapy (18% vs. 0%, p <0.001),¹⁷⁵ or further renal replacement therapy (25% vs. 3%, p< 0.001¹⁷⁵ and 30% vs. 10%, p=0.02¹⁷⁶), and lower risk for mortality (14% vs. 2%, p=0.02).¹⁷⁵ This benefit was evident only when hemofiltration was started before contrast media was administered. As Marenzi et al.¹⁷⁶ showed, when hemofiltration was started after the administration of contrast, its benefit was lost and the risk for developing CIN was comparable to those patients receiving hydration only. This finding was supported by Spini et al.,¹⁷⁸ who found a higher overall mortality for the patients who had continuous renal replacement therapy only after contrast media administration (57% vs. 16%, p=0.009; Appendix E, Evidence Table E-29). There was, however, a limitation to this group of studies; the studies that compared hemofiltration versus IV fluids were confounded by the use of IV bicarbonate with the hemofiltration. Insufficient evidence was available to support a conclusion about whether hemofiltration reduces the need for renal replacement therapy (Table 8).

The strength of evidence also was insufficient to determine whether renal replacement therapy (either hemofiltration or hemodialysis) reduces the risk of other outcomes due to the heterogeneity of the studies, comparators, and outcomes measured (Table 8; see Appendix G for study limitations).

Adverse events were reported in five studies (Appendix E, Evidence Table E-30).^59,173-176 The main adverse events reported were hematomas, blood loss, urinary retention, and/or anuria. Adverse events were not reported in a standardized manner and they were rarely analyzed in these studies, so we were unable to draw any conclusions regarding whether or not the incidence of adverse events differed between patients receiving renal replacement therapy and those who did not.

Study Characteristics

Our search identified eight RCTs with a total study population of 1930 patients that compared the use of ascorbic acid with various hydration regimens and other interventions used to prevent CIN.^34,182-188 All of these studies included patients undergoing cardiovascular interventions using intra-arterial LOCM. These studies were completed between 2004 and 2013 and were conducted in Germany,^34,182 Canada,¹⁸⁴ China,¹⁸⁵ Italy,¹⁸⁷ Korea,¹⁸⁸ Saudi Arabia¹⁸⁶ and Slovenia.¹⁸³ The mean age of patients ranged from 61 to 74. The percentage of patients with diabetes mellitus ranged from 26 to 83 percent, and all studies included patients with mild or moderate chronic kidney disease but excluded patients with end-stage renal disease or those requiring hemodialysis.

Six studies compared the combination of ascorbic acid and IV fluids with IV fluids alone.^34,182-186 two of these studies added an N-acetylcysteine arm to the comparison,^34,186 and two studies only compared ascorbic acid with N-acetylcysteine added to hydration.^187,188

In all eight studies, ascorbic acid was started prior to contrast media administration, with the total doses ranging from 1 gram as a unique dose¹⁸² or split between two doses³⁴ to 7 grams split between three doses within 24 hours of contrast.^183-188 (Appendix E, Evidence Tables E-1, E-3, E-31).

Two studies had medium risk of bias,^183,185 and six had low risk of bias based on the five criteria described in the methods for assessing risk of bias for individual studies (Appendix F).^{34,182,184,186-188} The limitations were due to problems with allocation generation,^182,183,185 allocation concealment,^183,185,188 and lack of blinding regarding the allocated intervention.^183,185,186

Contrast-Induced Nephropathy

Six studies were included in our meta-analysis comparing ascorbic acid to IV saline.^34,182-186 The studies excluded from the meta-analysis included those using N-acetylcysteine in the intervention and in the control arm. (Appendix E, Evidence Table E-31) When evaluating the efficacy of prophylactic ascorbic acid administration against IV fluids alone in the prevention of CIN. Four studies^{34,183,184,186} found a reduction of CIN in the intervention arm; three found this reduction to be clinically important (beyond our 25% threshold for a minimally important difference).^183,184,186 The remaining two studies found a slight but statistically insignificant increase of CIN in the intervention arm (6.7% vs. 4.3%¹⁸² and 6.3% vs. 5.4%¹⁸⁵).

Three studies compared ascorbic acid directly with N-acetylcysteine.^34,186,188 A fourth study incorporated N-acetylcysteine into the treatment regimen of all arms.¹⁸⁷ While one of the three studies found a statistically insignificant increase in CIN with the use of ascorbic acid (4.4% vs. 1.2%)¹⁸⁸ the other two showed a slight decrease in CIN incidence in the ascorbic acid arm (24.5% vs. 27.6% ³⁴ and 3.6% vs. 8.5%¹⁸⁶). When ascorbic acid was added to N-acetylcysteine, ascorbic acid slightly increased the risk of CIN when compared with N-acetylcysteine alone (10.3% vs. 9.9%¹⁸⁷ and 9.1% vs. 8.5%¹⁸⁶).

In the meta-analysis using the Knapp-Hartung method, the pooled estimate of the effect of ascorbic acid plus IV fluids compared with IV fluids alone^34,182-186 demonstrated a statistically insignificant but clinically important reduced risk of CIN with ascorbic acid use (pooled risk ratio 0.72; 95% CI: 0.48 to 1.01) (Figure 14) A meta-analysis using the Knapp-Hartung method showed a clinically unimportant decrease in CIN in the ascorbic acid group (RR: 0.89; 95% CI: 0.34 to 2.30).(Figure 15). Our review showed no substantial difference in stratified analyses by study inclusion criteria for baseline kidney function. Harbord's modified test for small study effects did not demonstrate evidence of asymmetry in results by study precision for ascorbic acid plus IV fluid versus compared with IV fluid alone (bias coefficient of 0.39, standard error of 0.76, p =0.63). The Harbord's modified test for ascorbic acid compared with N-acetylcysteine had similar results (bias coefficient of 0.41, standard error of 1.62, p=0.843). The dose or timing of the intervention did not affect the results.

Figure 14

Meta-analysis of ascorbic acid versus IV fluids for the prevention of contrast-induced nephropathy. %=percent; AA=ascorbic acid; CI=confidence interval; CIN=contrast-induced nephropathy; CKD=chronic kidney disease; Cr=creatinine; LOCM=low-osmolar contrast (more...)

Figure 15

Meta-analysis of ascorbic acid versus N-acetylcysteine for the prevention of contrast-induced nephropathy. %=percent; AA=ascorbic acid; CI=confidence interval; CIN=contrast induced nephropathy; CKD=chronic kidney disease; H=hydration; NAC=N-acetylcysteine; (more...)

The strength of evidence was low for demonstrating that ascorbic acid plus IV fluids did not have a clinically important effect in preventing CIN compared with IV fluids alone, when considering study limitations, directness, consistency, and precision (Table 9; see Appendixes F and G for study limitations).

Table 9

Summary of the strength of evidence: ascorbic acid versus IV saline.

Other Outcomes

Other outcomes were reported in four of the studies on ascorbic acid: three on renal replacement therapy,^183,187,188 three on cardiac outcomes,^183,185,188 one on mortality,¹⁸⁸ and one on length of stay.¹⁸⁵ No clinically important or statistically significant differences were seen in the need for dialysis, but very few events were reported.^183,187,188 Findings were similar in the studies reporting on cardiac outcomes.^183,185,188 The study reporting on mortality very few deaths were reported.^187,188 There was insufficient evidence to determine if ascorbic acid was more effective than N-acetylcysteine at reducing the need for renal replacement therapy, reducing mortality, or cardiac events. The strength of the evidence was low that ascorbic acid was more effective than IV saline at reducing the need for renal replacement therapy or cardiac events, and insufficient to determine if there was an impact on length of hospitalization (Table 9; Appendix E, Evidence Table E-31; see Appendixes F and G for study limitations).

The absence of adverse events was reported only in two studies. We were not able to draw any conclusions about the incidence of adverse events based on those two reports. (Appendix E, Evidence Table E-34).

N-Acetylcysteine Versus Other Interventions

We found 24 studies comparing N-acetylcysteine with other interventions including ascorbic acid,^34,187 nebivolol,⁷² atorvastatin,¹⁴¹ aminophylline,⁶⁶ theophylline,^31,68,189 fenoldopam,^28,190,191 misoprostol,⁶⁸ IV fluids,^{58,59,126,132} allopurinol,⁹⁰ and dialysis. ⁴³ There was substantial heterogeneity across these studies in terms of: dose of N-acetylcysteine; dose, type and duration of IV fluids; sample size; and follow-up period. The definition of CIN varied across studies as well. Because of the large heterogeneity of studies, a meta-analysis was not performed. A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.

Sodium Bicarbonate Versus Other Interventions

We found four studies comparing sodium bicarbonate with other interventions not involving N-acetylcysteine.^{124,127,129,192} The comparison interventions included acetazolamide,¹²⁹ long-term versus short-term sodium bicarbonate,¹²⁹ IV sodium bicarbonate versus oral sodium bicarbonate,¹²⁴ and saline versus saline plus sodium bicarbonate. Two studies used IOCM, two used LOCM, and one used both LOCM and IOCM. There was considerable heterogeneity across studies in terms of dose of sodium bicarbonate, dose and duration of other comparators, sample size, and follow-up period. All studies with the exception of one defined CIN as an increase of serum creatinine of 25% or at least 0.5 mg from baseline. Because of the large heterogeneity of studies, a meta-analysis was not performed. A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.

N-Acetylcysteine Plus Sodium Bicarbonate Versus Other Interventions

We found eight studies comparing N-acetylcysteine plus sodium bicarbonate versus other interventions, six RCTs,^{58,128,132,187,193,194} and 2 observational.^{58,128,132,187,193-196} In all studies, sodium bicarbonate was given IV at 3 ml/kg/hour or at 1 ml/kg/hour, before and after contrast media administration. A total of two doses of N-acetylcysteine was given prior to and after contrast media administration. All studies used IOCM. However, two studies also included administration of LOCM. N-acetylcysteine plus sodium bicarbonate was compared to N-acetylcysteine plus normal saline,^128,187 Renal Guard,¹⁹³ sodium bicarbonate plus dextrose,¹³² or sodium bicarbonate alone.¹⁹⁴ The study population for all trials was comprised of patients with renal dysfunction who were undergoing coronary interventions or another major arteriographic procedure, and three of the studies only included patients with Stage 3 or Stage 4 chronic kidney disease.^132,193,194 Due to the substantial heterogeneity of the comparators, and follow-up periods, a meta-analysis was not performed. A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.

Diuretics Versus Other Interventions

We found three studies comparing the use of different diuretics (furosemide, mannitol, and acetazolamide) in combination with IV saline to prevent CIN.^17,129,197 All studies included patients undergoing cardiovascular interventions and all studies included patients with diabetes mellitus. Two studies used LOCM and one used IOCM. Two studies evaluated furosemide as the diuretic of interest.^17,197 These two studies used it as a single comparator^17,197 Diuretic administration was given IV in all of the studies, but the protocols and doses varied. One study evaluated the effects of mannitol,¹⁷ and another included acetazolamide. Due to the substantial heterogeneity of the comparators, and follow-up periods, a meta-analysis was not performed. A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.

Vasoactive Agents Versus Other Interventions

We found 13 studies comparing vasoactive agents to other interventions: 12 RCTs,^{28,68,72,190, 191,198-204} and 1 observational;²⁰⁵ four studies on fenoldopam;^{28,190,191,198} two on calcium antagonists (one with nifedipine),⁶⁸ one with the combination of amlodipine and valsartan, an angiotensin receptor blocker)²⁰²; one on benazepril (an angiotensin converting enzyme inhibitor),²⁰¹ and one on nevibolol (a beta blocker).⁷² We also include in this section two studies that investigated the need for suspending the use of an angiotensin converting enzyme inhibitor or an angiotensin receptor blocker before receiving contrast media.^203,204 One study included only patients undergoing CT imaging,⁶⁸ and the remainder of the studies included patients undergoing cardiovascular interventions. All studies included patients with diabetes mellitus, but only one performed subgroup analysis for this population.¹⁹¹ Four studies use LOCM, three used IOCM, and one used both IOCM and LOCM. The studies were very heterogeneous, from the medications included to the doses used. A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.

Antioxidants Versus Hydration

We found seven studies evaluating different antioxidant strategies for preventing CIN. The antioxidant probucol was evaluated in two of these studies,^206,207 while two investigated pentoxifylline, an antioxidant and anti-inflammatory agent,^208,209 and the other two investigated sodium-2 mercaptoethanesulfonate (MESNA), a scavenger of reactive oxygen species,²¹⁰ zinc, which has the potential to act as an “endogenous antioxidant” via increasing metallothionein,⁵⁰ and trimetazidine, an antianginal agent which decreases free radicals, decreases oxygen consumption and may also decrease renal ischemia.²¹¹ All were conducted in patients with impaired renal function (serum creatinine greater than 1.2 and less than 3.0 mg/dl) undergoing coronary interventions and receiving LOCM. A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.

Fluid Interventions

We found 13 studies comparing different fluid regimens.^{86,87,116,124,212-220} Notably, two studies compared fluids to no fluids, with one comparing 0.45% saline²¹⁴ and the other investigating normal saline.²¹⁷ Four compared oral fluids to IV normal saline,^{87,124,215,220} and three compared isotonic saline to hypotonic saline.^216,218,219 Two studies compared standard dose IV normal saline to high-dose IV normal saline.^86,116 The timing of hydration, whether prior to or after the procedure, was compared in two studies.^212,217 Saline was separately compared with dextrose or sodium bicarbonate in three studies.^87,216,217 One study compared standard IV hydration to a left ventricular end diastolic pressure guided hydration protocol.²¹³ All of these studies defined CIN as an increase in serum creatinine by 25 percent or a change in serum creatinine of 0.5mg from baseline at 48 or 72 hours. However, one study also used an increase of glomerular filtration rate from a baseline of 50 percent,²¹² while another study recorded any CIN event between one to four days. ²¹³ A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.

Dopamine Versus Other Interventions

We found three studies assessing the effectiveness of dopamine in reducing CIN in patients with impaired renal function; two RCTs,^221,222 and one observational study²²³ One of the studies compared dopamine and a placebo,²²² and another compared a combination of dopamine and furosemide to a combination of dopamine, furosemide, mannitol, and saline.²²⁴ The remaining study had three arms that compared dopamine, saline, and aminophylline.²²¹ In all of the studies, dopamine was administered prior to and after contrast media administration. In two of the studies, the dose of dopamine was 2.5 micrograms/kg/min,^221,222 and the other study used a dose of 3 micrograms/kg/ml.²²⁴ One study had no definition set for CIN,²²⁴ while the other studies defined CIN as a change in serum creatinine greater than or equal to 25 percent or greater than 0.5 mg from baseline. A more detailed description of studies in this group and a summary of outcomes can be found in Appendixes H and I.