Polyethylene glycol bowel preparation does not eliminate the risk of acute renal failure: a population-based case-crossover study

• Introduction
• Methods
• Data source and ethical considerations
• Case-crossover study design
• Study patients
• Hazard and control periods
• Statistical analyses
• Results
• Discussion
• References

Background and study aims: Polyethylene glycol (PEG) bowel preparations are regarded as effective and safe for colonoscopy; however, recent reports have indicated a risk of acute renal failure (ARF). This population-based case-crossover study evaluated the association between PEG and ARF in screening colonoscopy patients aged ≥ 50 years.

Patients and methods: Korean Health Insurance Review and Assessment Service (HIRA) claims data from 1 January 2005 to 31 December 2009 were used in the study. The study population consisted of patients aged ≥ 50 years who were first hospitalized for ARF following colonoscopy involving PEG bowel preparation. For each patient, PEG use in a 1-, 2-, or 4-week period prior to the first hospital admission date for ARF (hazard period) was compared with PEG use in four earlier 1-, 2-, or 4-week control periods. Conditional logistic regression analysis was used to estimate odds ratios (ORs) and 95 % confidence intervals (CIs), adjusting for concomitant medications that could induce ARF.

Results: The total number of study patients was 1064 (59 % were male). A greater proportion of patients used PEG during the hazard period than during the control periods (for 4-week time window: 8.8 % vs. 3.2 %). The adjusted ORs for ARF incidence when applying the 1-, 2-, and 4-week periods were 3.1 (95 %CI 2.06 – 4.73), 2.5 (95 %CI 1.76 – 3.53), and 2.1 (95 %CI 1.61 – 4.85), respectively.

Conclusions: The use of PEG was associated with the risk of ARF. Adequate hydration and renal function monitoring should be assured before and after colonoscopy, regardless of the bowel preparation regimen used.

Introduction

Colonoscopy is a common diagnostic and therapeutic procedure. Regular screening is recommended for individuals aged ≥ 50 years [1] because colonoscopy decreases mortality from colorectal cancer [2]. Currently, polyethylene glycol (PEG) bowel preparations are the bowel cleansing solutions most frequently used before colonoscopy because PEG bowel preparations are effective and generally well tolerated [3].

PEG is a non-digestible, non-absorbable, osmotically balanced laxative lavage solution that does not alter fluid and electrolyte balance [4] [5] [6] [7]. It is stated that PEG does not cause physiological changes and can be administered to patients who have electrolyte imbalances, advanced liver disease, poorly compensated congestive heart failure, renal failure or who are elderly [6] [8].

However, several recent studies have raised concerns about safety issues related to renal impairment following PEG bowel preparation. A case report has raised the possible association of the use of PEG and acute renal failure (ARF) [9]. The patient in that case was a 55-year-old male without pre-existing renal disease who experienced severe abdominal pain and frequent diarrhea after ingesting PEG 2 hours earlier as pre-treatment for colonoscopy. He was diagnosed with pre-renal ARF, which improved after intensive fluid administration. A cohort study [10] revealed that following colonoscopy, patients aged > 65 years who did not have pre-existing renal disease were at risk for impaired renal function. We recently performed a population-based case-crossover study in patients aged ≥ 65 years, which failed to show any significant association between PEG and ARF [11] probably due to the limited number of patients included. Here we report on a large, nationwide case-crossover study that used a national health insurance database to evaluate the risk of ARF following the use of PEG among patients aged ≥ 50 years who underwent screening colonoscopy.

Methods

Data source and ethical considerations

The Korean Health Insurance Review and Assessment Service (HIRA) database was used in this study. The National Health Insurance (NHI) program in Korea was initiated in 1977 and achieved universal coverage of the population by 1989. All Koreans are entitled to medical coverage as an employed or community member under the National Health Insurance System. Healthcare providers are required to submit reports on medical services performed under health insurance policies to the HIRA for a review of the medical costs incurred. Accordingly, the HIRA database contains information on all claims, including prescribed medications and procedures for the entire population of approximately 50 million of South Korea [12]. Claims data that had been submitted by healthcare providers between 1 January 2005 and 31 December 2009 were obtained for the current study.

The study was exempted from review by the Institutional Review Board of the Seoul National University College of Medicine and Seoul National University Hospital because the researchers only accessed data from a de-identified database, which included age, sex, diagnosis, and drug prescriptions for outpatient visits, hospital admissions, medical procedures, and emergency department visits. Drug information included the brand name, generic name, prescription date, and duration and route of administration. Diagnoses were coded according to the International Classification of Disease (10th revision; ICD-10).

Case-crossover study design

A case-crossover approach was employed using cases at previous points in time as their own controls, thereby eliminating time-invariant confounders between subjects through within-subject difference comparisons [13]. A case-crossover study is a scientific method to answer the question, “Was this event triggered by something unusual that happened just before?” [14]. The study design was developed to avoid control selection bias in the Myocardial Infarction Onset Study [15], which was conducted to find the immediate determinants of myocardial infarction such as physical, psychological, or chemical triggers. The case-crossover design has been widely used as a tool to evaluate drug safety and is particularly suitable when the exposure is intermittent, when the effect on risk is immediate and transient, and when the outcome is abrupt [16]. The key feature of the design is that each case serves as its own control. The method is analogous to a crossover experiment viewed retrospectively, except that the investigator does not control when a patient starts and stops being exposed to the potential trigger [14] [17]. The hypothesis that can be tested with this design can be clarified by the following statement: “Some cases of illness or injury would not have occurred in these exposed people at these times, if they had not been exposed immediately before to the risk factor.”

Study patients

In the current study, the patient population consisted of individuals aged ≥ 50 years who underwent colonoscopy after PEG prescription prior to their first ARF hospitalization (ICD-10 code: N17). To identify ARF, primary and secondary discharge diagnoses were used. The colonoscopy was defined as being PEG-exposed if the PEG prescription date was recorded ≤ 90 days before the individual underwent colonoscopy. The index date was defined as the first hospital admission date for ARF. Patients who were hospitalized with pre-existing renal diseases (N17 – 19) including ARF and chronic renal failure (CRF) during the preceding 6 months (1 January 2005 to 30 June 2005) were excluded, as were those who had registered colon perforation-related diagnoses (K63.1, Y60.4, T81.2, K65) in the 14-day period after the colonoscopy procedure. Also excluded were the patients who were prescribed oral sodium phosphate for any reason during the study period before the index date because several recent studies have raised concerns about the safety of oral sodium phosphate preparations following reports of an association with an increased risk of serious electrolyte disturbances and renal failure [18].

Hazard and control periods

For each patient, one “hazard period” and four “control periods” were established. Time windows (1-, 2-, and 4-week periods) were used to determine the periods for assessment of drug exposure. The hazard period was defined as a 1-, 2-, or 4-week window prior to the index date. To prevent any carryover effect, a 4-week interval was chosen between the start of the hazard period and the end of the control period. Because relatively short-term window periods were chosen, the interval period was introduced between the hazard period and the four control periods to reduce the likelihood of overlapping prescriptions between these periods. Four consecutive 1-, 2-, or 4-week windows before the interval window were defined as control periods for all individual cases ([Fig. 1]). Therefore, when adopting the 1-week window period, for each patient with an ARF, hazard period, interval period, and four control periods were defined as 1 – 7 days (1 week), 8 – 35 days (4 weeks), and 36 – 63 days (4 times of each 1-week control period) prior to the index date, respectively. In the case-crossover design, the exposure to the target drugs during the single hazard period was compared with the exposure to the same drugs over four different control periods among the same case subject. Accordingly, the number having PEG exposure during the hazard period was compared with the number having PEG exposure in the control period.

Statistical analyses

Descriptive statistics were used to illustrate the characteristics of the ARF-hospitalized patients and the study population by age and sex. For the study population, the distribution of diagnoses for co-morbidities was analyzed. Co-morbidities were identified by the appropriate ICD-10 codes in claims made 6 months prior to the index date. Co-morbidities were grouped as upper gastrointestinal bleeding (K22.6, K25, K26, K27, K28, K29.0, K92.0, K92.1, K92.2, and I85.0), heart failure (I50), CRF (N18, N19), diabetes mellitus (E10 – E14), hypertension (I10 – I15), ischemic heart disease (I20 – I25), stroke (I60 – I64), chronic liver disease (K70 – K74), and cancer (C-D48). Because there could be a difference of several days or more between the date of prescription and actual administration, the PEG exposure date was defined as the day before the date of the colonoscopy.

PEG prescriptions that were dated between the hazard period and control periods were compared. PEG exposure was compared with the use of concomitant drugs included in the hazard- or control-window periods. Exposures to PEG and to other concomitant drugs were considered to be dichotomous variables in the model (exposed at least once during each specific time window [yes or no]). PEG use was compared between the hazard and control periods, and odds ratio (ORs) and their 95 % confidence intervals (95 % CIs) were calculated using conditional logistic regression analysis. By design, in the multivariable analysis controlling for time-varying confounding variables, the adjusted OR was calculated and the use of other concomitant medications that could induce ARF was simultaneously controlled for [19] [20], as their use might not be balanced between the hazard and control periods. Concomitant drugs included diuretics, angiotensin converting enzyme inhibitors, angiotensin receptor blockers, nonsteroidal anti-inflammatory drugs, aminoglycosides, beta-lactams, sulfonamides, anti-viral agents, anti-mycotics, anti-cancer drugs, and contrast media. Co-morbid conditions were regarded as time-invariant confounders so they were not included in the logistic regression model.

Subgroup analyses were conducted according to sex, age (< 65 vs. ≥ 65 years), and CRF diagnosis 6 months prior to the index date. Statistical analysis was performed using the SAS statistical application program (release 9.1; SAS Institute, Inc., Cary, North Carolina, USA).

Results

A total of 47 782 patients aged ≥ 50 years who were hospitalized for ARF between 1 January 2005 and 31 December 2009 were identified. Among the patients, 43 917 newly hospitalized patients with ARF were identified. Among these patients with ARF, 1442 had undergone colonoscopy after receiving a PEG prescription within 90 days prior to the first ARF hospitalization (more than 99 % of PEG prescription were within 3 days of the colonoscopy date). One patient who experienced colon perforation was excluded, as were 377 patients who were prescribed oral sodium phosphate. Thus, 1064 patients were included in the analysis. Among the 1064 study patients, 58 (5 %), 443 (42 %), 563 (53 %) patients were hospitalized in 2007, 2008, and 2009, respectively. Therefore, even though patients with pre-existing renal diseases only in the preceding 6 months we excluded, more than 90 % of the study patients had no renal disease during the first 3 years of the study (2005 – 2007). The details for selection of study participants are shown in [Fig. 2]. The mean ± SD age of the participants was 68.5 ± 9.5 years, and 59 % (627 patients) were male. There was a statistical age difference according to sex (P = 0.034). The proportions of co-morbidities, such as upper gastrointestinal bleeding, heart failure, chronic renal failure, diabetes mellitus, ischemic heart disease, and stroke, did not differ significantly according to sex. Hypertension was more prevalent in females (P = 0.003), and chronic liver disease (P < 0.001) and cancer (P = 0.001) occurred more often in males ([Table 1]).

[Table 2] presents the concordant and discordant pairs of PEG exposures observed among the ARF incident patients between the hazard period and the control periods according to time-window periods. Accordingly, in each time window, the number of pairs exposed in hazard periods presented in [Table 2] is equal to four times the PEG-exposed number of patients in hazard periods shown in [Table 3]. When applying a 4-week window period for the use of PEG, among a total of 4256 hazard – control pairs, 372 hazard – control period pairs were exposed to PEG during the hazard period but not the control period and 134 pairs were exposed to PEG during the control period but not the hazard period. A greater proportion of patients used PEG (i. e. 4-week time window: 8.8 % vs. 3.2 %) during the hazard period compared with the control periods. After adjusting for the use of concomitant drugs, the adjusted (a)ORs for ARF incidence in the 1-, 2-, or 4-week window periods were 3.1 (95 %CI 2.06 – 4.73), 2.5 (95 %CI 1.76 – 3.53), and 2.1 (95 %CI 1.61 – 4.85), respectively ([Table 3]).

The results of the subgroup analyses according to sex and age groups are summarized in [Table 4]. In both age groups (< 65 an  ≥ 65 years of age), the use of PEG was associated with the risk of ARF for all of the applied time-window periods. In the 50 – 64-year-old age group, the association of ARF and PEG was statistically significant only when adopting 1- or 4-week window periods in males and 4-week window period in females. In the elderly female group, the estimated aORs applying the 1-, 2-, and 4-week window periods were 5.3 (95 %CI 2.57 – 10.92), 4.2 (95 %CI 2.22 – 7.96), and 2.4 (95 %CI 1.42 – 4.01), respectively.

In the subgroup of patients without CRF, the results of association between PEG and ARF with respect to time-window periods by age groups and sex were generally similar to the results from the whole study group. In addition, regardless of whether CRF was present, the use of PEG was associated with the incidence of ARF, except when adopting the 2-week window period in patients with CRF. However, there were no associations between the use of PEG and ARF risk in males with CRF or in patients aged under 65 years with CRF. A higher aOR was evident for females with CRF when applied to the 4-week time window (14.4; 95 %CI 1.53 – 135.15) and for elderly patients ≥ 65 years of age with CRF when applied to the 1-week time window (13.5; 95 %CI 2.46 – 74.29) ([Table 5]).

Discussion

This population-based case-crossover study showed that the use of PEG was associated with a risk of ARF in patients aged ≥ 50 years undergoing colonoscopy. When several different time-window periods (1, 2, and 4 weeks) were applied, the use of PEG was consistently associated with ARF risk. Interestingly, there was a noticeably higher risk associated with PEG use in the female elderly group than in other groups. In this study, only hospitalized patients with pre-existing renal disease were excluded. Accordingly, patients who were not hospitalized for their chronic renal disease were still included. In addition, the exact indication for the colonoscopies performed was not known and as about 20 % of included patients had a history of cancer it seems likely that not all colonoscopies were performed for screening purposes. However, regardless of co-morbid conditions, the aORs for most of the categories demonstrated statistically significant increases of risk for ARF. All of the hospitalized patients with pre-existing renal diseases were excluded from the study and for this reason, most of the patients in the study did not have pre-existing renal disease or had only mild renal disease. Therefore, the subgroup of patients with CRF was not representative of the general CRF population, and the findings cannot apply to all CRF patients, and especially not those with severe disease.

A potential mechanism to explain the association between ARF and PEG preparation is dehydration through osmotic action and stimulation from bowel distension following the ingestion of a large fluid volume [21]. Volume repletion is an important but commonly overlooked component of the bowel-preparation process [22]. In addition, severe dehydration caused by excessive fluid loss and volume depletion is an important risk factor for the development of ARF [23] [24]. In the current study, female patients in the older age group had a significantly higher risk of ARF after the administration of PEG. This phenomenon may reflect that decreased kidney function is common in elderly persons and that the body volume of women is less than that of men; female elderly patients may be more affected by fluid loss than other groups.

However, these results must be interpreted in light of some potential limitations. Although ARF is generally defined as an abrupt and sustained decline in the glomerular filtration rate [25], we defined incident cases of ARF as hospitalization with a diagnosis of ARF in the HIRA database. However, because the database did not contain laboratory test results such as the serum creatinine or patients’ volume status at the time of exposure to PEG, a validation study was used to compare the diagnosis derived from the HIRA database with the actual diagnosis in the medical records. The overall positive predictive value of the diagnoses was 81.8 % in cases of hospitalized patients [26].

PEG is known to be a relatively safe bowel preparation; thus it is easy to relax our attention using this agent. Based on the current study results, however, PEG should be used with caution in all patients undergoing colonoscopy.

Competing interests: None.

Acknowledgments

This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST; No. 2010-0012292). We thank the Korean Health Insurance Review and Assessment Service for providing the insurance claims data.

• References

• 1 Rex DK, Johnson DA, Lieberman DA et al. Colorectal cancer prevention 2000: screening recommendations of the American College of Gastroenterology. Am J Gastroenterol 2000; 95: 868-877
  PubMedSearch in Google Scholar
• 2 Rabeneck L, Paszat LF, Saskin R et al. Association between colonoscopy rates and colorectal cancer mortality. Am J Gastroenterol 2010; 105: 1627-1632
  CrossrefPubMedSearch in Google Scholar
• 3 Occhipinti KE, Di Palma JA. How to choose the best preparation for colonoscopy. Nat Rev Gastroenterol Hepatol 2009; 6: 279-286
  CrossrefPubMedSearch in Google Scholar
• 4 Davis GR, Santa Ana CA, Morawski SG et al. Development of a lavage solution associated with minimal water and electrolyte absorption or secretion. Gastroenterology 1980; 78: 991-995
  PubMedSearch in Google Scholar
• 5 Beck DE, Harford FJ, DiPalma JA et al. Bowel cleansing with polyethylene glycol electrolyte lavage solution. South Med J 1985; 78: 1414-1416
  CrossrefPubMedSearch in Google Scholar
• 6 DiPiro JT, Michael KA, Clark BA et al. Absorption of polyethylene glycol after administration of a PEG-electrolyte lavage solution. Clin Pharm 1986; 5: 153-155
  PubMedSearch in Google Scholar
• 7 Nelson DB, Barkun AN, Block KP et al. Technology Status Evaluation report. Colonoscopy preparations. May 2001. Gastrointest Endosc 2001; 54: 829-832
  CrossrefPubMedSearch in Google Scholar
• 8 Dykes C, Cash BD. Key safety issues of bowel preparations for colonoscopy and importance of adequate hydration. Gastroenterol Nurs 2008; 31: 30-35
  CrossrefPubMedSearch in Google Scholar
• 9 Chun YJ, Park MK, Kim JS et al. Acute renal failure caused by oral polyethylene glycol ingestion. Korean J Gastrointest Endosc 2007; 34: 161-163
  PubMedSearch in Google Scholar
• 10 Russmann S, Lamerato L, Marfatia A et al. Risk of impaired renal function after colonoscopy: a cohort study in patients receiving either oral sodium phosphate or polyethylene glycol. Am J Gastroenterol 2007; 102: 2655-2663
  CrossrefPubMedSearch in Google Scholar
• 11 Choi NK, Chang Y, Jung SY et al. A population-based case-crossover study of polyethylene glycol use and acute renal failure risk in the elderly. World J Gastroenterol 2011; 17: 651-656
  CrossrefPubMedSearch in Google Scholar
• 12 Choi NK, Chang Y, Choi YK et al. Signal detection of rosuvastatin compared to other statins: data-mining study using national health insurance claims database. Pharmacoepidemiol Drug Saf 2010; 19: 238-246
  CrossrefPubMedSearch in Google Scholar
• 13 Delaney JA, Suissa S. The case-crossover study design in pharmacoepidemiology. Stat Methods Med Res 2009; 18: 53-65
  CrossrefPubMedSearch in Google Scholar
• 14 Maclure M, Mittleman MA. Should we use a case-crossover design?. Annu Rev Public Health 2000; 21: 193-221
  CrossrefPubMedSearch in Google Scholar
• 15 Mittleman MA, Maclure M, Tofler GH et al. Triggering of acute myocardial infarction by heavy exertion: protection against triggering by regular exertion. N Engl J Med 1993; 329: 1677-1683
  CrossrefPubMedSearch in Google Scholar
• 16 Chang CH, Chen HC, Lin JW et al. Risk of hospitalization for upper gastrointestinal adverse events associated with nonsteroidal anti-inflammatory drugs: a nationwide case-crossover study in Taiwan. Pharmacoepidemiol Drug Saf 2011; 20: 763-771
  CrossrefPubMedSearch in Google Scholar
• 17 Mittleman MA, Maclure M, Robins JM. Control sampling strategies for casecrossover studies: an assessment of relative efficiency. Am J Epidemiol 1995; 142: 91-98
  PubMedSearch in Google Scholar
• 18 Lien YH. Is bowel preparation before colonoscopy a risky business for the kidney?. Nat Clin Pract Nephrol 2008; 4: 606-614
  CrossrefPubMedSearch in Google Scholar
• 19 Guo X, Nzerue C. How to prevent, recognize, and treat drug-induced nephrotoxicity. Cleve Clin J Med 2002; 69: 289-312
  CrossrefPubMedSearch in Google Scholar
• 20 Pannu N, Nadim MK. An overview of drug-induced acute kidney injury. Crit Care Med 2008; 36: 216-S223
  CrossrefPubMedSearch in Google Scholar
• 21 Ho JM, Cavalcanti RB. A shocking bowel preparation: severe electrolyte disturbances after polyethylene glycol-based bowel preparation. J Am Geriatr Soc 2009; 57: 1729-1730
  CrossrefPubMedSearch in Google Scholar
• 22 Lichtenstein GR, Cohen LB, Uribarri J. Review article: bowel preparation for colonoscopy – the importance of adequate hydration. Aliment Pharmacol Ther 2007; 26: 633-641
  CrossrefPubMedSearch in Google Scholar
• 23 Brady HR, Brenner BM. Acute renal failure. In: et al. DL Kasper, eds. Harrison’s Principles of Internal Medicine. 16th. edn. New York: McGraw-Hill; 2005: 1644-1653
  Search in Google Scholar
• 24 Venkataraman R, Kellum JA. Prevention of acute renal failure. Chest 2007; 131: 300-308
  CrossrefPubMedSearch in Google Scholar
• 25 Nissenson AR. Acute renal failure: definition and pathogenesis. Kidney Int Suppl 1998; 66: 7-10
  PubMedSearch in Google Scholar
• 26 Park BJ, Sung JH, Park KD et al. Report of the evaluation for validity of discharged diagnoses in Korean Health Insurance database. Seoul: Seoul National University; 2003: 19-52
  Search in Google Scholar

Corresponding author

• References

• 1 Rex DK, Johnson DA, Lieberman DA et al. Colorectal cancer prevention 2000: screening recommendations of the American College of Gastroenterology. Am J Gastroenterol 2000; 95: 868-877
  PubMedSearch in Google Scholar
• 2 Rabeneck L, Paszat LF, Saskin R et al. Association between colonoscopy rates and colorectal cancer mortality. Am J Gastroenterol 2010; 105: 1627-1632
  CrossrefPubMedSearch in Google Scholar
• 3 Occhipinti KE, Di Palma JA. How to choose the best preparation for colonoscopy. Nat Rev Gastroenterol Hepatol 2009; 6: 279-286
  CrossrefPubMedSearch in Google Scholar
• 4 Davis GR, Santa Ana CA, Morawski SG et al. Development of a lavage solution associated with minimal water and electrolyte absorption or secretion. Gastroenterology 1980; 78: 991-995
  PubMedSearch in Google Scholar
• 5 Beck DE, Harford FJ, DiPalma JA et al. Bowel cleansing with polyethylene glycol electrolyte lavage solution. South Med J 1985; 78: 1414-1416
  CrossrefPubMedSearch in Google Scholar
• 6 DiPiro JT, Michael KA, Clark BA et al. Absorption of polyethylene glycol after administration of a PEG-electrolyte lavage solution. Clin Pharm 1986; 5: 153-155
  PubMedSearch in Google Scholar
• 7 Nelson DB, Barkun AN, Block KP et al. Technology Status Evaluation report. Colonoscopy preparations. May 2001. Gastrointest Endosc 2001; 54: 829-832
  CrossrefPubMedSearch in Google Scholar
• 8 Dykes C, Cash BD. Key safety issues of bowel preparations for colonoscopy and importance of adequate hydration. Gastroenterol Nurs 2008; 31: 30-35
  CrossrefPubMedSearch in Google Scholar
• 9 Chun YJ, Park MK, Kim JS et al. Acute renal failure caused by oral polyethylene glycol ingestion. Korean J Gastrointest Endosc 2007; 34: 161-163
  PubMedSearch in Google Scholar
• 10 Russmann S, Lamerato L, Marfatia A et al. Risk of impaired renal function after colonoscopy: a cohort study in patients receiving either oral sodium phosphate or polyethylene glycol. Am J Gastroenterol 2007; 102: 2655-2663
  CrossrefPubMedSearch in Google Scholar
• 11 Choi NK, Chang Y, Jung SY et al. A population-based case-crossover study of polyethylene glycol use and acute renal failure risk in the elderly. World J Gastroenterol 2011; 17: 651-656
  CrossrefPubMedSearch in Google Scholar
• 12 Choi NK, Chang Y, Choi YK et al. Signal detection of rosuvastatin compared to other statins: data-mining study using national health insurance claims database. Pharmacoepidemiol Drug Saf 2010; 19: 238-246
  CrossrefPubMedSearch in Google Scholar
• 13 Delaney JA, Suissa S. The case-crossover study design in pharmacoepidemiology. Stat Methods Med Res 2009; 18: 53-65
  CrossrefPubMedSearch in Google Scholar
• 14 Maclure M, Mittleman MA. Should we use a case-crossover design?. Annu Rev Public Health 2000; 21: 193-221
  CrossrefPubMedSearch in Google Scholar
• 15 Mittleman MA, Maclure M, Tofler GH et al. Triggering of acute myocardial infarction by heavy exertion: protection against triggering by regular exertion. N Engl J Med 1993; 329: 1677-1683
  CrossrefPubMedSearch in Google Scholar
• 16 Chang CH, Chen HC, Lin JW et al. Risk of hospitalization for upper gastrointestinal adverse events associated with nonsteroidal anti-inflammatory drugs: a nationwide case-crossover study in Taiwan. Pharmacoepidemiol Drug Saf 2011; 20: 763-771
  CrossrefPubMedSearch in Google Scholar
• 17 Mittleman MA, Maclure M, Robins JM. Control sampling strategies for casecrossover studies: an assessment of relative efficiency. Am J Epidemiol 1995; 142: 91-98
  PubMedSearch in Google Scholar
• 18 Lien YH. Is bowel preparation before colonoscopy a risky business for the kidney?. Nat Clin Pract Nephrol 2008; 4: 606-614
  CrossrefPubMedSearch in Google Scholar
• 19 Guo X, Nzerue C. How to prevent, recognize, and treat drug-induced nephrotoxicity. Cleve Clin J Med 2002; 69: 289-312
  CrossrefPubMedSearch in Google Scholar
• 20 Pannu N, Nadim MK. An overview of drug-induced acute kidney injury. Crit Care Med 2008; 36: 216-S223
  CrossrefPubMedSearch in Google Scholar
• 21 Ho JM, Cavalcanti RB. A shocking bowel preparation: severe electrolyte disturbances after polyethylene glycol-based bowel preparation. J Am Geriatr Soc 2009; 57: 1729-1730
  CrossrefPubMedSearch in Google Scholar
• 22 Lichtenstein GR, Cohen LB, Uribarri J. Review article: bowel preparation for colonoscopy – the importance of adequate hydration. Aliment Pharmacol Ther 2007; 26: 633-641
  CrossrefPubMedSearch in Google Scholar
• 23 Brady HR, Brenner BM. Acute renal failure. In: et al. DL Kasper, eds. Harrison’s Principles of Internal Medicine. 16th. edn. New York: McGraw-Hill; 2005: 1644-1653
  Search in Google Scholar
• 24 Venkataraman R, Kellum JA. Prevention of acute renal failure. Chest 2007; 131: 300-308
  CrossrefPubMedSearch in Google Scholar
• 25 Nissenson AR. Acute renal failure: definition and pathogenesis. Kidney Int Suppl 1998; 66: 7-10
  PubMedSearch in Google Scholar
• 26 Park BJ, Sung JH, Park KD et al. Report of the evaluation for validity of discharged diagnoses in Korean Health Insurance database. Seoul: Seoul National University; 2003: 19-52
  Search in Google Scholar