You Are Here:
A Systematic Review for the U.S. Preventive Services Task Force and for an American College of Physicians Clinical Practice Guideline
Release Date: April 2012
By Howard A. Fink, MD, MPH; Areef Ishani, MD, MS; Brent C. Taylor, PhD, MPH; Nancy L. Greer, PhD; Roderick MacDonald, MS; Dominic Rossini, MD; Sameea Sadiq, MD; Srilakshmi Lankireddy, MD; Robert L. Kane, MD; and Timothy J. Wilt, MD, MPH.
The information in this article is intended to help clinicians, employers, policymakers, and others make informed decisions about the provision of health care services. This article is intended as a reference and not as a substitute for clinical judgment.
This article may be used, in whole or in part, as the basis for the development of clinical practice guidelines and other quality enhancement tools, or as a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.
This article was first published in Annals of Internal Medicine on April 17, 2012 (www.annals.org).
Background: Screening and monitoring for chronic kidney disease (CKD) could lead to earlier interventions that improve clinical outcomes.
Purpose: To summarize evidence about the benefits and harms of screening for and monitoring and treatment of CKD stages 1 to 3 in adults.
Data Sources: MEDLINE (1985 through November 2011), reference lists, and expert suggestions.
Study Selection: English-language, randomized, controlled trials that evaluated screening for or monitoring or treatment of CKD and that reported clinical outcomes.
Data Extraction: Two reviewers assessed study characteristics and rated quality and strength of evidence.
Data Synthesis: No trials evaluated screening or monitoring, and 110 evaluated treatments. Angiotensin-converting enzyme inhibitors (relative risk, 0.65 [95% CI, 0.49 to 0.88]) and angiotensin II–receptor blockers (relative risk, 0.77 [CI, 0.66 to 0.90]) reduced end-stage renal disease versus placebo, primarily in patients with diabetes who have macroalbuminuria. Angiotensin-converting enzyme inhibitors reduced mortality versus placebo (relative risk, 0.79 [CI, 0.66 to 0.96]) in patients with microalbuminuria and cardiovascular disease or high-risk diabetes. Statins and β-blockers reduced mortality and cardiovascular events versus placebo or control in patients with impaired estimated glomerular filtration rate and either hyperlipidemia or congestive heart failure, respectively. Risks for mortality, end-stage renal disease, or other clinical outcomes did not significantly differ between strict and usual blood pressure control. The strength of evidence was rated high for angiotensin II–receptor blockers and statins, moderate for angiotensin-converting enzyme inhibitors and β-blockers, and low for strict blood pressure control.
Limitations: Evidence about outcomes was sometimes scant and derived from post hoc analyses of subgroups of patients enrolled in trials. Few trials reported or systematically collected information about adverse events. Selective reporting and publication bias were possible.
Conclusion: The role of CKD screening or monitoring in improving clinical outcomes is uncertain. Evidence for CKD treatment benefit is strongest for angiotensin-converting enzyme inhibitors and angiotensin II–receptor blockers, and in patients with albuminuria combined with diabetes or cardiovascular disease.
Primary Funding Source: Agency for Healthcare Research and Quality
Chronic kidney disease (CKD) is defined as kidney dysfunction (glomerular filtration rate [GFR] <60 mL/min per 1.73 m2) or kidney damage (usually reflected by albuminuria) that persists for at least 3 months (Figure 1) (1). Eleven percent of U.S. adults aged 20 years or older have CKD, of whom 95% have early disease (stages 1 to 3) (2). Prevalence of CKD stages 1 to 3 increases markedly with older age and is strongly associated with medical conditions, such as diabetes, hypertension, and cardiovascular disease (CVD). Chronic kidney disease is usually asymptomatic until advanced, and progression varies. However, CKD stages 1 to 3, as well as reduced GFR and albuminuria independently, increase the risk for many adverse health outcomes, including CVD, end-stage renal disease (ESRD), and mortality (3, 4).
Strategies that are proposed to prevent CKD-associated complications include screening selected patients for CKD, monitoring patients with CKD stages 1 to 3 for changes in kidney function or damage, and treating patients with CKD stages 1 to 3 for their CKD, or, more often, for its associated conditions and cardiovascular risk factors.
Because the effects of these interventions are uncertain, we conducted this systematic review to evaluate the evidence about the clinical benefits and harms of screening for and monitoring and treatment of CKD stages 1 to 3. This report was intended to provide an evidence base to guide recommendations on CKD from the U.S. Preventive Services Task Force and the American College of Physicians Clinical Guidelines Committee.
We followed a protocol developed with stakeholder input. The Appendix Figure shows the analytic framework and key questions we used to guide this review. The full technical report, which incorporated peer review and public comments, is available on the Agency for Healthcare Research and Quality (AHRQ) Web site (5).
We searched MEDLINE to identify randomized, controlled trials (RCTs) published from 1985 to 25 November 2011. We manually reviewed reference lists of relevant articles and articles suggested by experts. For complete search strategies, go to Appendix 1.
We applied separate eligibility criteria for CKD screening, monitoring, and treatment (Appendix 2). Trained reviewers examined titles, abstracts, and full articles for eligibility. A second reviewer evaluated a 10% sample of abstracts. When discrepancies were identified, all abstracts initially reviewed by 1 reviewer were reviewed by a second reviewer. Randomized, controlled trials that included participants who at least approximated the definitions for CKD stages 1 to 3 were considered to be eligible for the questions about CKD monitoring and treatment. Only English-language studies were included.
Data Extraction and Quality Assessment
For each article, a first reviewer extracted details on study design, participant characteristics, outcomes, and adverse events and rated study quality. A second reviewer checked the extracted data for accuracy. A priori, we selected mortality and ESRD as our primary efficacy outcomes, followed by clinical cardiovascular events (for example, myocardial infarction [MI], stroke, and congestive heart failure [CHF]), and composite vascular and renal outcomes that included these outcomes. Biochemical outcomes, such as halving of GFR, doubling of serum creatinine, and conversion from microalbuminuria to macroalbuminuria, were considered secondary and are reported in Supplements 1, 2, and 3 (available at www.annals.org) . By using criteria developed by the Cochrane Collaboration (6), we rated individual RCT quality as good, fair, or poor on the basis of the adequacy of allocation concealment (7), blinding, reporting of reasons for attrition, and how analyses accounted for incomplete data. By using methods developed by AHRQ and the Effective Health Care Program (8), we evaluated overall strength of evidence for mortality and ESRD outcomes for each treatment comparison on the basis of the criteria of risk for bias, consistency, directness, and precision (Appendix Table 1). We resolved discrepancies in quality and strength of evidence ratings by discussion and consensus.
Data Synthesis and Analysis
We pooled results if clinical heterogeneity of patient populations, interventions, and outcomes was minimal. Data were analyzed in Review Manager 5.0 (Cochrane Collaboration, Oxford, United Kingdom). Random-effects models were used to generate pooled estimates of relative risks (RRs) and 95% CI. Statistical heterogeneity was summarized by using the I2 statistic (9). When there were few RCTs for a given treatment and no overlap of reported outcomes, we synthesized the data qualitatively.
Role of the Funding Source
This review was funded by AHRQ, and the American College of Physicians Clinical Guidelines Committee provided support for manuscript preparation. Staff at AHRQ and a technical expert panel, including members of the American College of Physicians Clinical Guidelines Committee and U.S. Preventive Services Task Force and others, helped to develop and refine the scope, and assisted with review of draft manuscripts. AHRQ granted copyright assertion before the manuscript could be submitted for publication, although the authors are solely responsible for the content and decision to submit it for publication.
Our literature search for RCTs of CKD screening yielded 335 references; 321 were excluded after review of the title and abstract, and the remainder were excluded after review of the full text. Our search for RCTs of monitoring of CKD stages 1 to 3 yielded 920 references, with 901 excluded after review of the title and abstract, and the remainder excluded after review of the full text. Our MEDLINE search for RCTs of treatment of CKD stages 1 to 3 yielded 5291 references, with 4187 excluded after review of the title and abstract and 1012 excluded after review of the full text, leaving 92 eligible trials. Eighteen additional eligible RCTs of CKD treatment were initially identified from trial or systematic review reference lists or by technical expert panel members or reviewers, for a total of 110 eligible RCTs of treatment of CKD stages 1 to 3 (Figure 2).
In asymptomatic adults, what evidence is there that systematic CKD screening improves clinical outcomes or is associated with harms?
We found no RCTs of CKD screening in adults who were asymptomatic with or without recognized risk factors for CKD incidence, progression, or complications.
In adults with CKD stages 1 to 3, what evidence is there that systematic monitoring for worsening kidney function and/or kidney damage improves clinical outcomes or is associated with harms?
We found no RCTs of monitoring adults with CKD stages 1 to 3 for worsening kidney function or damage.
Among adults with CKD stages 1 to 3, what evidence is there that treatment improves clinical outcomes?
Go to the Table for a summary of our findings.
Angiotension-Converting Enzyme Inhibitors. Nineteen eligible RCTs randomly assigned patients with CKD to treatment with angiotension-converting enzyme (ACE) inhibitors versus placebo (10–29) or no treatment (30). Nearly all trials defined CKD on the basis of albuminuria (10–25), including 1 subgroup analysis from a larger trial (16, 30). Three studies were subgroup analyses of patients with impaired estimated GFR from larger trials (20, 26, 27).
We found moderate-strength evidence that patients with CKD stages 1 to 3 who were assigned to treatment with an ACE inhibitor had no reduced risk for mortality versus placebo or no treatment (RR, 0.91 [CI, 0.79 to 1.05]; 18 trials) (10–24, 26, 27, 29). Although mortality was reduced in RCTs comprising participants with microalbuminuria (RR, 0.79 [CI, 0.66 to 0.96]; 10 trials) (10, 12–14, 16, 18, 19, 21, 24, 29), these results were driven by 1 trial comprising patients with CVD or high-risk diabetes that included 97% of the deaths in the microalbuminuria subgroup (83). However, in that trial, there was no apparent difference in treatment effect between participants with microalbuminuria (RR, 0.77 [CI, 0.64 to 0.93]) and participants overall (RR, 0.76 [CI, 0.63 to 0.92]). By comparison, risk for mortality was not reduced with an ACE inhibitor versus placebo in trials restricted to patients with impaired estimated GFR (RR, 0.94 [CI, 0.70 to 1.26]; 4 trials) (20, 24, 26, 27), including 3 subgroup analyses (20, 26, 27).
We found moderate-strength evidence that ACE inhibitors reduced risk for ESRD versus placebo in patients with CKD stages 1 to 3 (RR, 0.65 [CI, 0.49 to 0.88]; 7 trials) (12, 15–17, 22–24, 30) although this benefit seemed to be driven by 3 trials limited to participants with macroalbuminuria, most of whom also had diabetes and hypertension (RR, 0.60 [CI, 0.43 to 0.83]) (15, 22, 23). In contrast, risk for ESRD was not statistically significantly reduced in trials comprising persons with CKD defined by microalbuminuria or impaired GFR only, in whom few ESRD events occurred (P = 0.48 for interaction with trials limited to participants with macroalbuminuria). Although patients with CKD stages 1 to 3 assigned to treatment with ACE inhibitors versus placebo had no statistically significant reduction in risk for MI, stroke, or other vascular outcomes and results were mixed for composite vascular outcomes, all 3 trials reporting composite renal outcomes found a reduced risk for this outcome (15, 22, 24).
Angiotensin II–Receptor Blockers. Among 5 eligible RCTs that compared angiotensin II–receptor blockers (ARBs) with placebo (31–35), we found high-strength evidence that patients with CKD stages 1 to 3 assigned to treatment with ARBs had no reduced risk for mortality (RR, 1.04 [CI, 0.92 to 1.18]; 4 trials) (31–33, 35). Results seemed to be similar in subgroups with or without albuminuria (P = 0.26 for interaction). We also found high-strength evidence that ARBs reduced risk for ESRD in patients with CKD stages 1 to 3 (RR, 0.77 [CI, 0.66 to 0.90]; 3 trials) (31, 32, 35). However, because 99% of ESRD events occurred in patients with macroalbuminuria, most of whom also had diabetes and hypertension (31, 32, 35), we could not determine whether ARBs reduced risk for ESRD in patients with microalbuminuria or impaired GFR only and without diabetes or hypertension. In addition, risk for cardiovascular mortality, MI, CHF complications, or any other clinical vascular or renal outcome did not significantly differ between ARBs and placebo. The 1 trial that reported results stratified by CKD status found no statistically significant difference between ARBs and placebo for risk for mortality or any clinical vascular or renal outcomes in patients with CKD overall (35). However, for the 1 reported composite renal outcome, participants with albuminuria had a greater reduction in risk with ARBs versus placebo than those with no albuminuria (P = 0.01 for interaction). For all other clinical outcomes, this trial reported no statistically significant difference in treatment effect between subgroups of patients with and without reduced estimated GFR and albuminuria, although ESRD events were rare.
ACE Inhibitors Versus ARBs. Among 7 eligible RCTs that randomly assigned patients with CKD stages 1 to 3 to treatment with ACE inhibitors versus ARBs (10, 36–41), we found low-strength evidence that risk for mortality did not differ between treatment groups (RR, 1.04 [CI, 0.37 to 2.95]; 5 trials) (10, 37, 39–41). There was also no statistically significant difference between ACE inhibitors and ARBs for risk for any other reported clinical vascular or renal outcome, although few events occurred and CIs around risk estimates were wide for both mortality and all of these outcomes. No study reported ESRD outcomes.
ACE Inhibitor Plus ARB Combinations Versus ACE Inhibitor or ARB Monotherapy. Among 6 eligible RCTs that assigned patients with CKD stages 1 to 3 to treatment with ACE inhibitor plus ARB combinations versus ACE inhibitor or ARB monotherapy (35–38, 42–44), including 2 subgroup analyses (35, 36, 43), we found moderate-strength evidence that there was no statistically significant difference in risk for mortality (35–37, 42, 43) and low-strength evidence that there was no statistically significant difference in risk for ESRD (35, 36, 44). In 1 trial, combination therapy increased risk for the single reported composite renal outcome versus ACE inhibitors overall, with no significant difference in treatment effect between subgroups of patients with and without reduced estimated GFR or albuminuria (P ≥ 0.27 for interaction by CKD status) (35, 36). In a second trial, combination therapy versus ACE inhibitors reduced risk for 1 reported composite vascular outcome, although treatment benefit was similar in subgroups with and without CKD (P = 0.23 for interaction) (43).
β-blockers. Five eligible RCTs randomly assigned patients with CHF to treatment with β-blockers versus placebo and reported subgroup results in participants with impaired estimated GFR (Castagno D, McMurray J. Personal communication) (45–48). Nearly all patients were receiving an ACE inhibitor or ARB at baseline. We found moderate-strength evidence that patients with CKD stages 1 to 3 assigned to treatment with β-blockers had a reduced risk for all-cause mortality (RR, 0.73 [CI, 0.65 to 0.82]; 5 trials). Risk was also reduced for CVD mortality (RR, 0.76 [CI, 0.64 to 0.90]; 3 trials) and CHF complications. The RR between β-blocker and placebo groups did not differ by estimated GFR category for any clinical outcome in 4 trials (P > 0.2 for interaction or reported as not significant) (Castagno D, McMurray J. Personal communication) (46–48), but suggested greater risk reduction in participants with lower estimated GFR (P < 0.05 for interaction) for 4 of 9 reported clinical outcomes in 1 trial (45). No study reported renal outcomes.
Calcium-Channel Blockers. Two eligible trials randomly assigned mostly hypertensive patients with albuminuria to treatment with calcium-channel blockers versus placebo (14, 32), with virtually all clinical outcomes reported in 1 trial (32). We found low-strength evidence that calcium-channel blockers did not reduce risk for mortality (RR, 0.90 [CI, 0.69 to 1.19]; 2 trials) or ESRD (RR, 1.03 [CI, 0.81 to 1.32]; 1 trial). Although calcium-channel blockers reduced risk for MI (RR, 0.58 [CI, 0.37 to 0.92]), risk was not reduced for stroke, CHF, or composite vascular outcomes.
Thiazide Diuretics. One eligible trial randomly assigned patients with systolic hypertension to treatment with thiazide diuretics versus placebo and reported subgroup results in participants with serum creatinine levels of 119.34 µmol/L or greater (≥1.35 mg/dL) (49). We found low-strength evidence that patients with increased creatinine levels assigned to treatment with thiazide diuretics had no reduction in mortality (RR, 1.17 [CI, 0.74 to 1.85]). However, the thiazide diuretic group had a reduced risk for stroke (RR, 0.49 [CI, 0.24 to 0.99]) and for 1 of 2 reported composite vascular outcomes. In results reported only for 1 composite vascular outcome, the RR between thiazide diuretic and placebo groups did not differ between subgroups with and without increased creatinine (P = 0.96 for trend). No renal outcomes were reported.
Strict Versus Standard Blood Pressure Control. In 7 eligible trials (50–57), 6 comprised entirely (50–52, 56) or mostly (54, 55, 57) of patients with hypertension, study participants with CKD stages 1 to 3 were randomly assigned to different targets for treatment of blood pressure. Targets and medications that were used varied among trials, but the strict control target was usually approximately 10 to 15 mm Hg less than the standard control target. In trials reporting follow-up systolic and diastolic blood pressure results, mean achieved blood pressure ranged from 128 to 133 mm Hg for systolic blood pressure and 75 to 81 mm Hg for diastolic blood pressure in the strict control group versus 134 to 141 mm Hg for systolic blood pressure and 81 to 87 mm Hg for diastolic blood pressure in the standard control group (50, 51, 53, 54). The difference in achieved mean arterial pressure between treatment groups ranged from 4 to 9 mm Hg (50, 51, 53–55, 57). However, we found low-strength evidence that strict control did not reduce risk for mortality (RR, 0.86 [CI, 0.68 to 1.09]; 4 trials) (50–53) or ESRD (RR, 1.03 [CI, 0.77 to 1.38]; 3 trials) (50, 51, 53). In addition, risk for MI, stroke, or any reported composite vascular or renal outcome did not significantly differ between treatment groups. In 1 trial comprising patients with low estimated GFR in which there was no statistically significant between-group difference in risk for any of the 3 composite renal outcomes overall, a post hoc analysis reported that the strict control group had a reduced risk for 1 composite renal outcome in the subgroup with baseline protein–creatinine ratios greater than 0.22 (adjusted hazard ratio, 0.74 [CI, 0.56 to 0.99]; P = 0.09 for unadjusted interaction versus subgroup with protein–creatinine ratio ≤0.22) (56).
Statins. Among 14 eligible RCTs that compared statins with placebo (21, 58–65, 69), diet (66), or usual care (67, 68), all but 2 (21, 65) were subgroup analyses in participants with impaired estimated GFR or creatinine clearance from a larger trial. We found moderate-strength evidence that patients with CKD stages 1 to 3 assigned to treatment with statins had reduced risk for mortality compared with control (RR, 0.81 [CI, 0.71 to 0.94]; 10 trials) (21, 58–61, 64–67, 69), and low-strength evidence of no reduced risk for ESRD versus control (RR, 0.98 [CI, 0.62 to 1.56]; 2 trials) (65, 68). In addition, patients with CKD stages 1 to 3 assigned to treatment with statins had reduced risk for MI, stroke, and most reported composite vascular outcomes. However, trials consistently found no statistically significant interaction of CKD on treatment group effect for any of these clinical outcomes (59, 60, 62, 64, 67, 69).
Low-Protein Diet. Six eligible trials randomly assigned patients with CKD stages 1 to 3 to variably defined low-protein diets versus usual diets (57, 70–74). All but 1 study (70) reported results for patients with CKD stages 1 to 3 in combination with those for participants with CKD stages 4 or 5. We found low-strength evidence that low-protein diets did not reduce risk for mortality (RR, 0.58 [CI, 0.29 to 1.16]; 4 trials) or ESRD (RR, 1.62 [CI, 0.62 to 4.21]; 3 trials), although few events occurred and CIs were wide for both outcomes. Risk for a composite renal outcome was reduced in the low-protein diet group in 1 trial reporting this outcome (73).
Among adults with CKD stages 1 to 3, what evidence is there that treatment is associated with harms?
Few RCTs reported information on study withdrawals. When withdrawals were reported, they were often high and infrequently were reported separately by treatment group. Few trials reported adverse events, and these often seemed to be neither predefined nor systematically collected or reported. Adverse events reported were consistent with those reported in RCTs not limited to patients with CKD, with risk relative to placebo significantly increased for cough with ACE inhibitors, hyperkalemia with ARBs, and hypotension with β-blockers. In 1 large RCT that compared an ACE inhibitor plus ARB combination with an ACE inhibitor alone, combination treatment was associated with a significant increase in risk for cough, hyperkalemia, hypotension, and acute kidney failure requiring dialysis (RR, 1.95 [CI, 1.09 to 3.49]) (35).
We found no RCTs of CKD screening or monitoring and, thus, no direct evidence about their benefits or harms. In contrast, we found direct RCT evidence about the benefits of several treatments for patients with CKD stages 1 to 3, including ACE inhibitors, ARBs, β-blockers, and statins. Although CKD increased the absolute risk for adverse clinical vascular and renal events, other than the significantly reduced risk for ESRD with ACE inhibitors or ARBs in patients with macroalbuminuria (most of whom also had diabetes and hypertension), we found little evidence that any relative improvement in clinical outcomes with these treatments versus placebo differed between patients with CKD and those without.
The strongest evidence about the benefits and harms of systematic CKD screening versus usual care or no screening would come from RCTs that report clinical outcomes. We found no such trials. However, other studies have provided indirect evidence about these questions. Clinical and administrative data, primarily from large representative U.S. cohorts, suggest that targeted screening could identify many patients with undiagnosed CKD. First, CKD stages 1 to 3 are common in older patients (2) and in adults with specific illnesses (for example, diabetes, hypertension, and CVD) (84). Second, most persons with CKD stages 1 to 3, even those with diabetes and hypertension, are not clinically recognized (85) and do not have CKD testing in usual care (86, 87), with albuminuria measured less often than serum creatinine. Albuminuria and serum creatinine–derived estimated GFR are widely available in primary care settings, with high sensitivity and specificity for 1-time measures of renal damage or dysfunction (2). However, the risk for false-positive screening is substantial (88, 89), and these measures have unknown sensitivity and specificity for CKD as defined by persistently decreased GFR or albuminuria (90). Further, evidence from CKD treatment trials seems to differ on the basis of whether study participants have macroalbuminuria, microalbuminuria, or impaired estimated GFR, and different CKD screening tests may detect only modestly overlapping groups of patients. Thus, considerations about the potential benefit of CKD screening must be specific to the screening regimen. In that context, modeling studies have incorporated data (including CKD epidemiology, screening test characteristics, and benefits and harms of treatment) to estimate the cost-effectiveness of screening for microalbuminuria (91) or macroalbuminuria (92). These studies have concluded that, compared with usual care, targeting screening for albuminuria in older patients with diabetes or hypertension and treating patients who screen positive with ACE inhibitors or ARBs may be cost-effective. However, these modeling studies may overgeneralize CKD screening benefits. They assume that reductions in mortality risk with ACE inhibitors versus placebo reported in 1 subgroup analysis comprising patients with CKD who have albuminuria and either CVD or high-risk diabetes (16) apply to all patients with albuminuria (for example, including persons with diabetes with no other cardiovascular risk factors and patients with isolated hypertension) (91, 92). Our review did not find evidence to support this assumption.
The strongest evidence about the benefits and harms of systematic monitoring of patients with CKD stages 1 to 3 for worsening renal function or damage versus usual care or no monitoring would come from RCTs that reported clinical outcomes. We found no such trials. However, data from observational studies suggest that targeted CKD monitoring could identify many patients with unrecognized progression who are at increased risk for adverse clinical outcomes. First, several studies have reported that patients with diabetes, hypertension, hyperlipidemia, obesity, smoking, or proteinuria are more likely to have faster progression of kidney damage or dysfunction (93–95). Second, although nearly all patients with diagnosed CKD stages 1 to 3 have serum creatinine levels measured regularly in usual practice, only 30% to 40% are tested annually for albuminuria (86); as a result, albuminuria progression may be unrecognized in many patients. Third, although we are unaware of studies that report the sensitivity and specificity of estimated GFR or albuminuria for identifying persistent progression of CKD stages 1 to 3, and the risk for false-positive identification of CKD progression is unknown, categorically worsening albuminuria in patients with CKD significantly increases risk for mortality and adverse clinical vascular and renal outcomes independent of baseline albuminuria severity (96). Even accounting for RCT evidence that selected treatments improve important clinical outcomes in patients with CKD stages 1 to 3, it is uncertain from all this fragmentary evidence whether modifying treatment of worsened CKD detected by monitoring improves clinical outcomes compared with modifying treatment of worsened CKD detected by usual care. Further, we found no modeling studies that quantitatively estimated the effectiveness of any strategy for monitoring progression of CKD stages 1 to 3 followed by treatment of patients with progression versus a control strategy.
We found no RCTs or prospective observational studies of CKD screening or monitoring that reported harms. However, this does not exclude the possibility of harms associated with these interventions. Potential harms of CKD screening are adverse effects from screening and follow-up tests, including follow-up of false-positive results, psychological effects from labeling asymptomatic individuals as having the disease, medication adverse effects, increased medical visits, and increased health care costs. Potential harms of systematic monitoring of patients with CKD stages 1 to 3 for worsening kidney function or damage are adverse effects from monitoring and follow-up tests, including potentially unnecessary testing, medication adverse effects, and increased medical visits and health care costs.
The strongest RCT evidence of the benefit of treating CKD stages 1 to 3 was reduction in risk for ESRD with ACE inhibitors or ARBs. However, this benefit seemed to be limited to the subgroup of patients with CKD who have macroalbuminuria, most of whom had concomitant diabetes and hypertension. Although we found no evidence that ACE inhibitors or ARBs reduced risk for ESRD versus placebo in patients with microalbuminuria or impaired estimated GFR only, ESRD events were rare in these subgroups, and analyses of these studies had low statistical power to detect a treatment-related difference in risk for progression to ESRD. Whether our finding that ACE inhibitors reduced risk for mortality versus placebo when ARBs did not indicates a true advantage of ACE inhibitors over ARBs in patients with CKD stages 1 to 3 is uncertain. The higher prevalence of CVD in trials that compared ACE inhibitors with placebo than in those that compared ARBs with placebo may contribute to this finding. Unfortunately, the 5 RCTs in patients with CKD stages 1 to 3 that compared ACE inhibitors with ARBs and reported clinical outcomes had little power to identify a difference in risk for mortality or any vascular or renal outcome. Among patients with CKD stages 1 to 3, the relative reduction in risk for mortality and other clinical vascular and renal outcomes associated with treatment with ACE inhibitors, ARBs, β-blockers, thiazide diuretics, and statins seemed to be limited to patients with specific comorbid conditions and did not differ substantially from that found in patients without CKD. This finding suggests that populations evaluated in these trials may have a clinical indication for such treatments (for example, ACE inhibitors in patients with CVD or high-risk diabetes, β-blockers with CHF, and statins with hyperlipidemia), regardless of having CKD or CKD progression. Additional trials that randomly assigned participants with CKD stages 1 to 3 to more versus less intensive treatment showed no consistent difference in clinical outcomes between treatment groups. Interpretation of trials that compared strict versus standard blood pressure control is complicated by variability in baseline, target, and achieved blood pressures between trials. Similarly, interpretation of trials that compared low-protein with usual diets is complicated by variability in the level of protein prescribed and inclusion of participants with CKD stages 4 to 5 in addition to those with CKD stages 1 to 3. Although neither intensive intervention seemed to reduce the risk for any clinical outcome versus control therapy, given limitations in individual study quality and the few clinical events reported in these trials, future studies are likely to refine these estimates of effect. By comparison, trials that compared ACE inhibitors combined with ARBs versus ACE inhibitors or ARBs alone showed a possibly unfavorable tradeoff between improvement in 1 composite vascular outcome at the cost of increased risk for renal adverse effects, including acute kidney failure requiring dialysis.
This review is limited in part by the available literature, including our inability to identify RCTs that directly evaluated the benefits or harms of CKD screening or monitoring. Inconsistent definitions of CKD and clinical outcomes among treatment trials may limit generalizability of findings across studies. Many RCTs reported few clinical outcomes and even fewer adverse events, limiting our confidence around risk estimates for these outcomes. Because nearly all eligible trials that reported baseline GFR had a mean estimated GFR of 45 mL/min per 1.73 m2 or greater, results of this review may not apply equally to patients with lower estimated GFRs. Further, many studies were post hoc analyses of subgroups with CKD drawn from RCTs that enrolled more general populations, and many other trials involving the same populations and interventions have not reported results for their subgroups with CKD stages 1 to 3; thus, results of this review may be affected by publication bias. Although the scant attention we paid to biochemical CKD treatment outcomes, such as change in estimated GFR and albuminuria, may also be considered a limitation, our decision to focus the review on clinical outcomes was made a priori. Although these biochemistries are adverse prognostic markers, some trials have reported increases in fatal cardiovascular events (97) and in renal failure requiring dialysis (35), despite improved albuminuria.
Overall, we found no direct evidence about the benefits or harms of screening patients for CKD or for monitoring patients with CKD stages 1 to 3 for CKD progression. Indirect evidence suggested that targeting CKD screening or monitoring may be possible but that the potential benefit of these interventions was uncertain. Evidence for CKD treatment benefit was strongest for ACE inhibitors and ARBs, particularly for reduction in risk for ESRD in patients with macroalbuminuria who also have diabetes and hypertension. Future studies should compare CKD screening and monitoring with usual care on important clinical outcomes. Refined modeling studies of CKD screening and monitoring are warranted. Large-scale treatment RCTs should define CKD according to current criteria. Trials should also be designed a priori to do long-term collection of clinical vascular and renal outcomes and to report outcomes by CKD stage, albuminuria, estimated GFR categories and subcategories (that is, dividing patients with CKD stage 3 into those with estimated GFR <45 and ≥45 mL/min per 1.73 m2), and important patient characteristics. Judicious use of administrative data sets may also be informative.