Portal de Periódicos da CAPES

Risk of Kidney Injury Following Oral Phosphosoda Bowel Preparations

2007; American Society of Nephrology; Volume: 18; Issue: 12 Linguagem: Inglês

10.1681/asn.2007040440

1533-3450

Steven M. Brunelli, James D. Lewis, Meera Gupta, Sherif M. Latif, Mark G. Weiner, Harold I. Feldman,

Colorectal Cancer Screening and Detection

Case reports and case series suggest a potential link between oral sodium phosphosoda used in preparation for outpatient colonoscopy and kidney injury, but controlled studies are lacking. We performed a case-control study nested within a cohort of patients with baseline serum creatinine ≤1.5 mg/dL who underwent outpatient colonoscopy. We defined a case of kidney injury as a rise in serum creatinine ≥0.5 mg/dL and/or 25% between values obtained during the 6 months prior and during the 6 months following colonoscopy (n = 116). We found that exposure to phosphosoda was not more common among patients with incident kidney injury (adjusted odds ratio 0.70; 95% CI 0.44–1.11), and sensitivity analyses that considered other definitions of kidney injury did not suggest a different conclusion. Therefore, despite a plausible link, the current data do not support an association between oral phosphosoda and kidney injury at 6 months follow-up among patients with baseline serum creatinine ≤1.5 mg/dL. Further studies are warranted to validate and generalize our findings. Nearly 14 million colonoscopies are performed each year in the United States.1 Bowel preparation is usually achieved using either polyethylene glycol or oral sodium phosphosoda (OPS). In many cases, OPS is the preferred bowel regimen on the basis of better tolerability, cost-effectiveness, and efficacy.2–4 Several uncontrolled case reports and case series5–9 suggested a potential link between receipt of OPS and nephrocalcinosis, acute kidney injury, and/or chronic kidney disease (CKD). Although data in animal models support a potentially causal relationship between OPS and kidney injury,10–13 to date, no controlled human studies have examined this potential association. CKD's mounting prevalence (20 million cases in the United States presently14), increasingly recognized morbid consequences,15 and unclear cause, in many cases, combine to make identification of novel risk factors a major public health concern. Considering the vast number of patients who are exposed to OPS each year, even a modest increase in risk conveyed by these agents could translate into a large number of cases of CKD in absolute terms. We undertook the following nested case-control study to determine whether an association exists between incident kidney injury and use of OPS to prepare patients undergoing routine outpatient colonoscopy at three University of Pennsylvania Health System–affiliated units. RESULTS A total of 8218 colonoscopies were identified between January 1, 2004, and February 1, 2006. Among these, 2237 had adequate associated creatinine data to qualify for inclusion in the source cohort. The case definition was met in 141 instances, rendering 132 unique case patients; of these, bowel preparation data were available for 116 (Figure 1). Among the 2096 colonoscopies for which the case definition was not met, 398 unique patients were randomly selected as control subjects; of these, bowel preparation data were available for 349. Patients for whom exposure data were not available (n = 65) did not differ from those for whom they were (n = 465), except on the basis of clinical site (Table 1). Median (interquartile range) baseline creatinine was 1.0 (0.8 to 1.1) and 0.8 mg/dl (0.75 to 1.1 mg/dl) among control subjects and case patients, respectively. The mean increase in creatinine concentration after colonoscopy was 0.41 and 0.00 mg/dl among case patients and control subjects, respectively. The mean (SD) and median (interquartile range) change in serum creatinine were 0.12 (0.34) and 0.1 mg/dl (−0.1 to 0.2 mg/dl) among OPS-unexposed patients and 0.08 (0.26) and 0.0 mg/dl (−0.1 to 0.2 mg/dl) among OPS-exposed patients (P = 0.28 for difference by OPS exposure status). Case patients were significantly more likely than control subjects to be female, to have congestive heart failure, and to have been exposed to diuretics (Table 2). No other significant baseline differences between case patients and control subjects existed. Time to latest and maximum serum creatinine in the 6-mo postcolonoscopy period or latest available creatinine did not differ according to OPS exposure status (data not shown). OPS-exposed patients had significantly fewer measurements of serum creatinine in the 6 mo after colonoscopy (2.5 ± 3.5 versus 3.9 ± 8.1; P = 0.004) but a similar number of measurements overall (7.0 ± 11.6 versus 8.7 ± 13.2; P = 0.25). Association between Incident Kidney Injury and OPS Exposure Among 116 case patients for whom exposure data were available, 66 (57%) were exposed to OPS; among 349 control subjects for whom exposure data were available, 230 (66%) were exposed to OPS. The crude odds ratio (OR) for OPS exposure case was 0.68 (95% confidence interval [CI] 0.44 to 1.08). There was little confounding. In the fully adjusted model, including demographics, comorbid disease status, mediation exposure, and colonoscopy indication, exposure to OPS was not associated with incident creatinine elevation (adjusted OR 0.70; 95% CI 0.44 to 1.11). Less saturated models yielded nearly identical results (Table 3). Inclusion of baseline estimated GFR (eGFR) as a covariate did not appreciably alter our findings (data not shown). Sensitivity Analyses Several different definitions for case status were used in sensitivity analyses. Because OPS exposure may be associated with acute kidney injury without persistent elevation in serum creatinine (but with lasting clinical implications), we performed a sensitivity analysis in which case definition was defined on the basis of maximum serum creatinine in the 6-mo postcolonoscopy window (SA-1). Because a 25% rise in serum creatinine may in some cases not be the result of actual kidney injury, we performed sensitivity analyses in which case status was defined as an absolute rise of 0.5 mg/dl within 6 mo after colonoscopy (SA-2). Because kidney injury may require >6 mo to become apparent, sensitivity analyses in which case definition was based on a change in serum creatinine between baseline and last available creatinine (SA-3) were performed. A total of 119, 27, and 72 patients met the case definition (and had available exposure data) for SA-1, SA-2, and SA-3, respectively. Results of each of the sensitivity analyses were nearly identical to the primary analysis (Table 3). In addition, we examined patients with documented rises in serum creatinine of >1 mg/dl. Three case patients had an increase in serum creatinine of this magnitude; only one was exposed to OPS. Considering that our source cohort consisted of >2200 people with a 64% exposure prevalence to OPS, the point estimate for incidence of kidney injury using this more restrictive definition would be seven and 14 cases per 10,000 among OPS-exposed and -unexposed patients, respectively. Effect Modification of the Association between OPS Exposure and Incident Kidney Injury Interactions on the basis of gender, baseline CKD (eGFR 1.5 mg/dl) are at increased risk for kidney injury, we did not include these patients for two reasons. First, previous work suggested that OPS is associated with nephrotoxicity among patients with normal kidney function: Markowitz et al.8 reported that all but one of the 21 cases of kidney injury that they observed had baseline creatinine ≤1.5 mg/dl; therefore, 95% of the patients in that series would have met the inclusion criteria for our study. Second, OPS is contraindicated in patients with kidney disease and, in our institution, carefully avoided in this setting. Focusing on this group with CKD would have yielded a small group of patients with a very low exposure prevalence to OPS. Such a study not only would have been grossly underpowered as a result, but also would have missed any opportunity to detect the association of OPS and kidney injury among individuals with normal kidney function at baseline, the group of patients for whom understanding this relationship is most clinically relevant. Given that we did not investigate whether OPS contributes to worsening kidney function among patients with baseline CKD, our findings should not be extrapolated to that population. At 6 mo of follow-up, in patients with baseline serum creatinine ≤1.5 mg/dl, there was no apparent increase in the risk for incident kidney injury defined as a 0.5-mg/dl and/or 25% rise in serum creatinine. Additional studies are necessary to validate and generalize findings. CONCISE METHODS The protocol was approved by the University of Pennsylvania institutional review board. Patients and Sites We conducted a case-control study nested within a source cohort of patients who underwent outpatient colonoscopy at one of three University of Pennsylvania Health System–affiliated units (two hospital-based and one free-standing unit) between January 1, 2004, and February 1, 2006. January 1, 2004, was chosen because it was the date when it became standard clinical practice to record bowel preparation type in the medical chart. The presence of colonoscopies and relevant demographics, diagnoses, medications, and laboratories were identified through the Pennsylvania Integrated Clinical and Administrative Research Database (PICARD) system, a database of resource use and clinical findings collected through the daily operation of the University of Pennsylvania Health System. To be included in the source cohort, patients were also required to have evidence of at least one visit to the health system predating the colonoscopy (to provide opportunity for capture of covariate data) and at least one measurement of serum creatinine in the 6 mo before and 6 mo after colonoscopy (to enable longitudinal assessment of kidney function). Patients with a baseline serum creatinine >1.5 mg/dl were excluded (see the Discussion section). Identification of Case Patients and Control Subjects Baseline serum creatinine was defined as the most proximate measurement before colonoscopy; follow-up creatinine was defined as the latest measurement recorded in the 6-mo postcolonoscopy window. For patients meeting the case definition at more than one colonoscopy, only the first episode was considered. Control subjects were chosen by simple random selection from among those not meeting the case definition at a ratio of 3:1 (control subjects:case patients). Data Collection Exposure data and colonoscopy indication were abstracted from colonoscopy reports by two investigators who were blinded to case/control status. Patients were considered exposed when their bowel preparation was recorded as either Fleet's Oral Phosphosoda (C.B. Fleet Co., Lynchburg, VA) or Visicol (Salix Pharmaceuticals, Morrisville, CA) and unexposed when bowel preparation was recorded as another agent or none. Colonoscopy indication was categorized as either screening/surveillance or symptomatic. Additional covariates of interest included age; race; gender; clinical site; congestive heart failure; diabetes; and receipt of ACEI, ARB, or diuretic. Baseline eGFR was estimated by the four-variable MDRD equation. Diabetes was defined by diagnosis of diabetes (International Classification of Diseases, Ninth Revision code 250.xx); elevated glycosylated hemoglobin (>6%) predating colonoscopy; or a prescription for oral hypoglycemic medication, insulin, or diabetic testing materials spanning the date of colonoscopy. Congestive heart failure was defined by an International Classification of Diseases, Ninth Revision diagnosis (425.xx, 428.xx, 401.x1, 404.x1, or 404.x3) predating colonoscopy. Patients were considered exposed to ACEI, ARB, or diuretics when they had an active prescription spanning the date of colonoscopy. Statistical Analyses Bivariable measures of association were determined by χ2 testing for categorical variables and t test or Wilcoxon rank sum test for continuous variables, as appropriate. The primary outcome of interest was the adjusted exposure OR for OPS in case patients versus control subjects. The exposure OR based on survival sampling of control subjects provides an unbiased estimate of the risk ratio when the disease in question is rare, as in this case (132/2337 = 0.059).28,29 Adjusted OR were determined via multiple logistic regression models. Prespecified interactions on the basis of gender,8 ACEI/ARB exposure,8 and baseline CKD were examined by inclusion of cross-product terms (with exposure) and compared with parent models via likelihood ratio test. A similar series of models was created for each of the sensitivity analyses—SA-1, SA-2, and SA-3—except that outcome status was reclassified according to alternative case definitions (see the Results section for description). Power Analysis Power analyses were conducted assuming a fixed sample size of 132 case patients. Assuming a 3:1 control subject:case patient ratio, an exposure prevalence of 50% among control subjects provided 80%; the study had 80% power to detect a true OR of 1.77 (or greater), at a significance level of 0.05. All analyses were performed using Stata 9.0 (Stata Corp., College Station, TX). DISCLOSURES None.Figure 1: Flow diagram for selection of case patients and control subjects.Table 1: Comparison of baseline characteristics between patients whose bowel preparation exposure data were and were not availableaTable 2: Baseline characteristics of case patients and control subjectsaTable 3: OPS exposure and kidney injury based on logistic regression modelsaTable 4: OR of association between incident kidney injury and OPS exposure according to ACEI/ARB exposure status, gender, and baseline CKDaThis study was supported in part by National Institutes of Health institutional training grant for clinical nephrology T32-DK-07785 (S.M.B.). S.M.B. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.