686
ORIGINAL CONTRIBUTIONS
nature publishing group
ENDOSCOPY
see related editorial on page x
Hyponatremia and Sodium Picosulfate Bowel Preparations in Older Adults Matthew A. Weir, MD, MSc1,2, Jamie L. Fleet, BSc2, Chris Vinden, MD3, Salimah Z. Shariff, PhD2,4, Kuan Liu, MMath2,4, Haoyuan Song, BPharm5, Arsh K. Jain, MD, MSc1,2,4, Sonja Gandhi, BSc2, William F. Clark, MD1,2 and Amit X. Garg, MD, PhD1,2,4,6 OBJECTIVES:
Bowel preparations are commonly prescribed drugs. Case reports and our clinical experience suggest that sodium picosulfate bowel preparations can precipitate severe hyponatremia in some older adults. At present, this risk is poorly quantified. We investigated the association between sodium picosulfate use and the risk of hyponatremia in older adults.
METHODS:
We conducted a population-based retrospective cohort study using six linked administrative databases in Ontario, Canada. All Ontario residents over the age of 65 years who filled an outpatient bowel preparation prescription before colonoscopy were eligible. We enrolled new users of either sodium picosulfate (n = 99,237) or polyethylene glycol (n = 48,595). The primary outcome was hospitalization with hyponatremia within 30 days of the bowel preparation assessed by database codes. The secondary outcomes were hospitalization with urgent head computed tomography (CT) (a proxy for acute central nervous system disturbance) and all-cause mortality.
RESULTS:
The baseline characteristics of the two groups, including patient demographics, comorbid conditions, and concomitant medications, were nearly identical. Compared with polyethylene glycol, sodium picosulfate was associated with a higher risk of hospitalization with hyponatremia (absolute risk increase: 0.05%, 95% confidence interval (CI): 0.04–0.06%, relative risk (RR): 2.4, 95% CI: 1.5–3.9), but not hospitalization with urgent CT head (RR: 1.1, 95% CI: 0.7–1.4) or mortality (RR: 0.9, 95% CI: 0.7–1.3).
CONCLUSIONS: Sodium picosulfate bowel preparations lead to more hyponatremia than polyethylene glycol. There
was no evidence of increased risk of acute neurologic symptoms or mortality. The absolute increase in risk of hospitalization with hyponatremia remains low but may be avoidable through appropriate fluid intake or preferential use of polyethylene glycol in some older adults. Am J Gastroenterol 2014; 109:686–694; doi:10.1038/ajg.2014.20; published online 4 March 2014
INTRODUCTION Millions of colonoscopies are performed worldwide each year (1–3). The safety and effectiveness of this procedure depend on the adequacy with which the bowel can be cleansed (4,5). A variety of effective oral bowel preparations are available, all with specific limitations (6). Isosmotic bowel preparations contain an isosmotic fluid and a nonabsorbable polymer (polyethylene glycol) that causes fluid to be retained in the colon. These bowel preparations are considered safer for patients at risk of electrolyte disturbances or circulating volume changes, because they do not cause significant shifts of
electrolytes or water across the colon wall (5,6). However, patients are required to consume 2–4 l of polyethylene glycol-containing solutions to adequately prepare the bowel. This, combined with poor palatability, limits their effectiveness (7–9). Hyperosmotic bowel preparations address the limitations of polyethylene glycol-containing solutions by providing a large amount of osmotically active molecules in a smaller volume of liquid. The ingested osmoles draw fluid into the colon and stimulate evacuation. However, this process of shifting fluid into the colon can result in intravascular volume contraction and electrolyte disturbances (10,11).
1 Division of Nephrology, Department of Medicine, Western University, London, Ontario, Canada; 2Kidney Clinical Research Unit, Western University, London, Ontario, Canada; 3Division of General Surgery, Department of Surgery, Western University, London, Ontario, Canada; 4Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada; 5Schulich School of Medicine and Dentistry, Western University, London, Ontario, Canada; 6Department of Epidemiology and Biostatistics, Western University, London, Ontario, Canada. Correspondence: Matthew A. Weir, MD, MSc, ALL-139A University Hospital, 339 Windermere Road, London, Ontario, Canada, N6A 5A5. E-mail: [email protected] Received 26 August 2013; accepted 14 January 2014
The American Journal of GASTROENTEROLOGY
VOLUME 109 | MAY 2014 www.amjgastro.com
Sodium picosulfate with magnesium oxide is a commonly prescribed hyperosmotic bowel preparation. In our clinical practice, we have encountered a number of patients hospitalized with significant hyponatremia after sodium picosulfate use. Three studies have identified asymptomatic hyponatremia (12–14), and three case reports have described hyponatremia-induced seizures in patients using sodium picosulfate (15–17). Although the risk appears greatest in older adults, it remains poorly quantified. Thus, we conducted this study to characterize the 30-day risk of hospitalization with hyponatremia in a large population of older adults prescribed sodium picosulfate in routine care. We compared these patients with a similar group of older adults prescribed polyethylene glycol. We hypothesized a higher risk of hyponatremia in patients prescribed sodium picosulfate vs. polyethylene glycol.
Patients
METHODS
Bowel preparations
Study design and setting
We conducted a population-based retrospective cohort study using health administrative data from Ontario, Canada. Ontario is Canada’s most populous province with ~13 million residents who receive universal access to hospital and physician services (Statistics Canada). Ontario’s 1.8 million residents over the age of 65 years also receive prescription drug coverage. The Research Ethics Board at Sunnybrook Health Sciences Centre approved the prespecified protocol for this study, and we have reported it according to established guidelines for observational studies (18).
Over the course of this study, the provincial drug plan provided coverage for two sodium picosulfate-containing bowel preparations, both comprising sodium picosulfate with magnesium oxide and citric acid (10 mg per sachet of sodium picosulfate, 3.5 g per sachet magnesium oxide, and 12 g per sachet of citric acid) (20,21). Four formulations of polyethylene glycol-containing bowel preparations were available during the study period, all of which had an osmolality of 280 mOsM, contained 60 g/l of polyethylene glycol, and had virtually identical electrolyte compositions except for CoLyte, which contained 80 mEq/l of sulfate compared with 40 mEq/l in the other three brands (Table 1) (22–25).
Data sources
Outcomes
We used six linked databases housed at the Institute for Clinical Evaluative Sciences to conduct this study. We ascertained vital statistics, including mortality, from the Registered Persons Database. The Registered Persons Database records the demographic information for people issued a provincial health card. We used the Ontario Drug Benefits database to ascertain prescription drug exposure and drug-related baseline characteristics. The Ontario Drug Benefit records prescription drug use for patients over the age of 65 years and has an error rate of < 1% (19). We identified admissions to hospital and baseline characteristics using the Canadian Institute for Health Information Discharge Abstract Database. The Canadian Institute for Health Information Discharge Abstract Database records hospital admissions and related diagnostic and procedural information. For similar information relating to emergency department admissions, we used the National Ambulatory Care Reporting System database. We used the Ontario Health Insurance Plan database to ascertain information relating to physician services, such as computed tomography (CT) of the head and colonoscopy. We determined physician specialty by using the Institute for Clinical Evaluative Sciences Physician Database. The Institute for Clinical Evaluative Sciences Physician Database pools information from three data sources and contains information about physician demographics, specialty training, certification, and practice location.
All outcomes were assessed in the 30 days after the bowel preparation prescription date.
© 2014 by the American College of Gastroenterology
We restricted enrollment to patients over the age of 66 years to ensure the availability of at least 1 year of drug-use data. Using records from 1 January 2006 to 31 January 2012, we identified new prescriptions for bowel preparations containing either sodium picosulfate or polyethylene glycol. Prescriptions were considered new if there was no other bowel preparation prescription filled in the preceding 120 days. To exclude prescriptions given for constipation, we eliminated those with durations of more than 3 days. We excluded patients who had bowel cleansing for bowel surgery (28% of the cohort). We excluded any patient with recent colonoscopy and those with more than one study bowel preparation or a prescription for a nonstudy bowel preparation (in order to compare mutually exclusive groups). Finally, we excluded those receiving chronic dialysis before their study drug prescription, as their sodium is regulated by the dialysis prescription.
Primary outcome: hospitalization with hyponatremia. We defined the primary outcome of hospital admission with hyponatremia as any admission with the E87.1 International Classification of Diseases, 10th revision diagnostic code (hypo-osmolality and hyponatremia) in any of the Canadian Institute for Health Information Discharge Abstract Database’s diagnostic fields. In Ontario, trained coders assign diagnostic codes to every hospital encounter on the basis of data from the patient’s chart. The Canadian Coding Standards do not permit coders to interpret laboratory test results, but they can record a laboratory-based condition if the physician has documented the diagnosis. For example, E87.1 would not be recorded for a serum sodium concentration of 120 mmol/l unless it was accompanied by a physician’s interpretation of “hyponatremia”. As a laboratory value becomes more extreme (i.e., lower levels of serum sodium), a code is more likely to be present for a particular diagnosis. In Ontario validation studies, the presence of E87.1 identifies patients with a median serum sodium level of 123 mmol/l (interquartile range: 119–126 mmol/l) and a decrement of 8.0 mmol/l (4.0–13.0) mmol/l from the baseline prehospital encounter serum sodium measurement. Although the specificity is The American Journal of GASTROENTEROLOGY
687
ENDOSCOPY
Sodium Picosulfate and Hyponatremia
688
Weir et al.
ENDOSCOPY
Table 1. Composition of polyethylene glycol-containing bowel preparations available during the study period Component
Klean-Prep
GoLytely
PegLyte
CoLyte
Sodium (mmol/l)
125
125
126
125
Sulfate (mmol/l)
40
40
40
80
diagnostic code (hyperkalemia) found in any of the Canadian Institute for Health Information Discharge Abstract Database’s diagnostic fields.
> 99%, the sensitivity is limited, particularly for milder forms of hyponatremia (26).
Post hoc analyses. Peer-reviewer commentary prompted us to conduct two post hoc analyses. To explore the effect of the prescribers’ specialty on the risk of hyponatremia, we grouped prescribers into three categories (gastroenterology and internal medicine, general surgery, and family medicine) and added a three-level interaction term to our multivariable model. To more strictly define hyponatremia associated with bowel preparations, we first explored the temporal relationship between colonoscopy and admission with hyponatremia among patients who experienced both of these outcomes. This group of patients allowed us to use colonoscopy as a proxy for the timing of bowel preparation ingestion. Using this proxy, we excluded cases where admission with hyponatremia occurred more than 72 h after colonoscopy.
Secondary outcome: hospitalization with CT head. We were interested in the potential neurologic sequelae of hyponatremia in this population because of case reports describing seizures in patients with sodium picosulfate-associated hyponatremia. We used hospitalization with evidence of an urgent CT head as a proxy for the presence of acute neurologic conditions such as delirium, seizures, and decreased level of consciousness. The typical practice in our region is to perform a CT head when a patient presents to hospital with any acute neurologic finding, and as the coding of CT scans is linked to remuneration we expect this proxy outcome to be recorded accurately (fee for service codes have high sensitivity and specificity (27); this differs from ascertaining delirium or decreased level of consciousness on their own, which are generally poorly coded in administrative data). To limit this outcome to acute neurologic conditions at the time of hospital presentation, we only considered CT head scans performed in the first 7 days of a hospital admission, or in the emergency department in the 2 days preceding the admission. We expected urgent CT head scans performed for reasons other than the effects of bowel preparationinduced hyponatremia to occur at similar frequencies in the two groups. We have successfully used this outcome of urgent CT in other population-based drug safety studies to characterize the risk of drug-induced acute neurologic findings (28).
Statistical methods. We compared the prevalence of baseline characteristics between the sodium picosulfate and polyethylene glycol groups using standardized differences. This metric describes differences between group means as a percentage of the pooled s.d. Those greater than 10% are considered meaningful (30,31). The risk of developing an outcome was expressed in both relative and absolute terms. Absolute risk was also expressed as the “number needed to harm” (1 divided by absolute risk difference), a measure that indicates how many patients needed to receive sodium picosulfate to cause harm to one patient who would not have experienced the event if they had received polyethylene glycol (a lower number indicating greater harm). We used multivariable logistic regression analyses (PROC LOGISTIC) to estimate odds ratios and 95% confidence intervals (CIs). We adjusted for the following 12 a priori-selected covariates: age, sex, chronic kidney disease, chronic liver disease, heart failure, bowel cancer, use of anticonvulsants, antidepressants, antipsychotics, benzodiazepines, loop diuretics, or thiazide diuretics. We also adjusted for rural area of residence and year of cohort entry after we noted imbalances in these characteristics between the two groups. Odds ratios were interpreted as relative risks (RR; appropriate, given the incidence rate observed). We conducted all analysis with SAS 9.2 (SAS Institute, Cary, NC).
Chloride (mmol/l)
35
35
35
35
Bicarbonate (mmol/l)
20
20
20
20
Potassium (mmol/l)
10
10
10
10
Polyethylene glycol (g/l)
59
60
59
60
Osmolality (mOsM)
280
280
280
280
Secondary outcome: all-cause mortality. In our region, the Registered Persons Database records the vital statistics of all residents covered by the provincial health-care plan. It has been shown to have a sensitivity of 97.8% and specificity of 100% for the finding of death (29). Tracer outcome: hospitalization with hyperkalemia. We tested the specificity of our findings by determining each group’s risk of hyperkalemia, an outcome that we did not expect to be influenced by the choice of bowel preparation. We defined hyperkalemia as any admission 30 days following the prescription date with the E87.5. International Classification of Diseases, 10th revision The American Journal of GASTROENTEROLOGY
RESULTS Figure 1 shows the steps of cohort assembly. Over 6 years, we identified 147,832 patients who met enrollment criteria, of whom 99,237 (67%) filled a prescription for sodium picosulfate and 48,595 (33%) for polyethylene glycol (Table 2). Twenty-four baseline characteristics were similarly distributed across the two groups, with standardized differences < 0.1. However, patients living in rural settings were more likely to receive sodium picosulfate, and those prescribing sodium picosulfate were slightly more likely to be surgeons. Sodium picosulfate prescriptions became more common over the study period, whereas polyethylene VOLUME 109 | MAY 2014 www.amjgastro.com
Sodium Picosulfate and Hyponatremia
689
Study drug prescriptions during the accrual period n = 313,536 1. Death at or before prescription, n = 90 2. Prescription for > 1 study drug, n = 1862 3. Prescription for a non-study drug, n = 61,296
Remaining eligible prescriptions n = 250,288 Restriction to new users Prescription for any bowel preparation in the preceding 120 days, n = 36,942
Remaining eligible prescriptions n = 213,346 Restriction to screening colonoscopy (1) 1. Colonoscopy in the preceding 14 days, n = 3,848 2. Bowel surgery in the preceding 14 days, n = 37 3. Hospital admission on the prescription date, n = 133 4. Emergency department admission on the prescription date, n = 789
Remaining eligible prescriptions n = 208,539 Restriction to screening colonoscopy (2) Bowel surgery in the 30 days following the qualifying prescription, n = 46,003
Remaining eligible prescriptions n = 162,536 Exclusion of patients receiving chronic dialysis Any chronic dialysis procedural code or fee code in the preceding 120 days, n = 557
Remaining eligible prescriptions n = 161,979 Restriction to the level of the patient Exclude repeat prescriptions, n = 14,147
Number of eligible patients n = 147,832
Figure 1. Flow diagram of cohort construction.
glycol prescription rates were unchanged throughout. Interestingly, despite the recommendation that patients with chronic kidney disease receive polyethylene glycol owing to its electrolyte neutrality, a number of these patients received sodium picosulfate. The majority of patients in both groups underwent colonoscopy within 30 days of filling their prescription. Procedure or billing codes for colonoscopy were detected in 78.6% of patients prescribed sodium picosulfate and 77.3% prescribed polyethylene glycol. Primary outcome
Results of the primary and secondary outcomes are presented in Table 3. Those prescribed polyethylene glycol made up the referent group. Prescription for sodium picosulfate was associated with a higher risk of hospital admission with hyponatremia (adjusted RR: 2.4, 95% CI: 1.5–3.9). Expressed in absolute terms, © 2014 by the American College of Gastroenterology
a prescription for sodium picosulfate was associated with a 0.05% higher incidence of hospital admission with hyponatremia (95% CI: 0.04–0.06). The corresponding number needed to harm was 1,903 (95% CI: 1,645–2,257). Secondary outcomes
Table 3 shows that no difference was found in the risk of hospital admission with urgent CT head on comparing sodium picosulfate with polyethylene glycol (adjusted RR: 1.1, 95% CI: 0.9–1.4). There was also no significant difference in the risk of all-cause mortality between the two groups (adjusted RR: 0.9, 95% CI: 0.7–1.3). Tracer outcome
The risk of hospital admission for hyperkalemia was not significantly different between the two groups (adjusted RR: 0.8, 95% CI: 0.4–1.6). The American Journal of GASTROENTEROLOGY
ENDOSCOPY
Data-cleaning exclusions
690
Weir et al.
Table 2. Baseline characteristics for sodium picosulfate and polyethylene glycol users
ENDOSCOPY
Sodium picosulfate (n=99,237)
Polyethylene glycol (n=48,595)
Standardized difference (%)a
Demographics 72 (69–77)
73 (69–78)
8
Female sex, no. (%)
Age, median (IQR)
52,825 (53.2)
25,015 (51.5)
4
Lowest income quintile, no. (%)
17,455 (17.6)
9,095 (18.7)
3
Rural location, no. (%)
16,216 (16.3)
4,324 (8.9)
23
Year of cohort entry, no. (%) 2006
5,595 (5.6)
8,166 (16.8)
36
2007
12,643 (12.7)
7,655 (15.8)
9
2008
15,940 (16.1)
8,162 (16.8)
2
2009
20,313 (20.5)
8,284 (17.1)
9
2010
22,120 (22.3)
7,991 (16.4)
15
2011
20,853 (21.0)
7,785 (16.0)
13
1,773 (1.8)
552 (1.1)
5
2012 (January only) Prescriber specialtyb, no. (%) Gastroenterology
24,042 (24.2)
14,603 (30.1)
13
General surgery
38,078 (38.4)
11,021 (22.7)
35
Family physician
14,342 (14.5)
10,305 (21.2)
18
Internal medicine Missing
3,769 (3.8)
2,017 (4.2)
2
15,573 (15.7)
8,617 (17.7)
5
35,239 (35.5)
16,633 (34.2)
3
c
Comorbidities , no. (%) Malignancy Chronic kidney disease
5,322 (5.4)
3,652 (7.5)
9
Chronic liver disease
3,885 (3.9)
2,447 (5.0)
6
Coronary artery disease
32,422 (32.7)
16,962 (34.9)
5
Diabetes mellitus
17,207 (17.3)
9,598 (19.8)
6
7,383 (7.4)
4,309 (8.9)
5
ACE inhibitors
30,939 (31.2)
15,942 (32.8)
4
ARB
Heart failure Medications, no. (%)
21,333 (21.5)
10,347 (21.3)
0
Anticonvulsants
4,164 (4.2)
2,284 (4.7)
2
Antidepressants
15,257 (15.4)
7,480 (15.4)
0
Antineoplastic medications
3,812 (3.8)
1,915 (3.9)
1
Antipsychotics
1,660 (1.7)
865 (1.8)
1
Benzodiazepines
14,303 (14.4)
7,339 (15.1)
2
Beta blockers
26,446 (26.7)
13,657 (28.1)
3
Calcium channel blockers
25,690 (25.9)
13,619 (28.0)
5
6,459 (6.5)
3,655 (7.5)
4
Loop diuretics Non-ASA NSAIDs
17,512 (17.7)
8,664 (17.8)
0
Statins
48,069 (48.4)
24,158 (49.7)
3
Thiazide diuretics
17,779 (17.9)
9,003 (18.5)
2
ACE, angiotensin converting enzyme; ARB, angiotensin receptor blocker; ASA, acetylsalicyclic acid; IQR, interquartile range; no., number; NSAID, nonsteroidal anti-inflammatory drug. a The standardized difference is the difference between group means expressed as a percentage of that characteristic’s pooled standard deviation. It is less sensitive to sample size than hypothesis testing. Standardized differences greater than 10% are considered significant. b Prescriber specialties are recorded in the OHIP database. Percentages do not total 100 because specialties responsible for < 1% of prescriptions were not included in the table. c Comorbidities were identified using administrative codes in the five years preceding the qualifying bowel preparation prescription.
The American Journal of GASTROENTEROLOGY
VOLUME 109 | MAY 2014 www.amjgastro.com
Sodium Picosulfate and Hyponatremia
691
Number of events (%) Picosulfate n=99,237
Polyethylene glycol n=48,595
Unadjusted relative risk (95% CI)
Adjusteda relative risk (95% CI)
Absolute risk difference, % (95% CI)
Number needed to harm (95% CI)b
Hyponatremia
93 (0.09)
20 (0.04)
2.4 (1.4–3.7)
2.4 (1.5–3.9)
0.05 (0.04–0.06)
1,903 (1,645–2,257)
CT head
257 (0.26)
126 (0.26)
1.0 (0.8–1.2)
1.1 (0.9–1.4)
—
—
Mortality
139 (0.14)
81 (0.17)
0.8 (0.6–1.1)
0.9 (0.7–1.3)
—
—
CI, confidence interval; CT, computed tomography. (—) Indicates not relevant to calculate given no difference in relative risk. For each analysis, patients prescribed polyethylene glycol were the referent group. a Adjustment was by multivariable logistic regression modeling with the following variables: age, sex, rural status, year of cohort entry, chronic kidney disease, chronic liver disease, heart failure, bowel cancer, use of anticonvulsants, antidepressants, antipsychotics, benzodiazepines, loop diuretics, or thiazide diuretics. b The number needed to harm is presented to facilitate interpretation; it does not imply causality.
Table 4. Colonoscopy and admission for hyponatremia: proportions and timing Admissions with hyponatremia Sodium picosulfate (N=93)
Polyethylene glycol (N=20)
No colonoscopy during observation period, no. (%)
56 (60.2)
11 (55.0)
Colonoscopy during observation period, no. (%)
37 (39.8)
9 (45.0)
Colonoscopy before admission, no. (%)
15 (16.1)
Median interval, days (IQR) Colonoscopy after admission, no. (%) Median interval, days (IQR)
3 (1–11) 22 (23.7) 4 (2–12)
a
4 (–) a
8 (–)
IQR, interquartile range; no., number. a To prevent the identification of patients, cell sizes of < 6 or cell sizes that could be used to derive the value of small cells cannot be reported.
Post hoc analyses
Prescriber specialty. When we stratified the cohort according to the specialty of the prescribing physician (gastroenterology and internal medicine vs. general surgery vs. family medicine), we found that prescriber specialty did not modify the risk of hospital admission with hyponatremia (P for interaction = 0.8). Stricter definition of admission for hyponatremia. Table 4 shows the timing of hospital admission with hyponatremia and colonoscopy among the patients who experienced both within the 30day observation period. The majority of patients admitted with hyponatremia did not receive colonoscopy during the observation period (60% of sodium picosulfate and 55% of polyethylene glycol). Among those who did have colonoscopy, a small proportion had the procedure before admission with hyponatremia. To remove cases of hyponatremia that are less likely to be the result of bowel preparation, we excluded admissions with hyponatremia that occurred more than 3 days following colonoscopy. This © 2014 by the American College of Gastroenterology
resulted in the loss of 10 events from the picosulfate group and less than 6 events from the polyethylene glycol group (privacy law prevents the reporting of cell sizes less than 6). Using this stricter definition, prescriptions for sodium picosulfate again associated with a higher risk of hospital admission with hyponatremia (adjusted RR: 2.9, 95% CI: 1.7–5.1). This estimate of risk was not materially different from that found in the primary analysis (adjusted RR: 2.4, 95% CI: 1.5–3.9).
DISCUSSION In this population-based cohort, we observed a significant 2.4fold increase in the RR of hospital admission with hyponatremia in patients prescribed sodium picosulfate compared with polyethylene glycol. Using the proxy measure of urgent CT head, there appeared to be no difference in acute neurologic events between the groups. The choice of bowel preparation also had no influence on 30-day all-cause mortality. The increase in the risk of hyponatremia we observed with sodium picosulfate is a concerning finding. Although patients in our study did not require more urgent CT head scans, they did experience symptoms that were severe enough to warrant hospital admission. Therefore, our failure to detect the sequelae of hyponatremia using a surrogate outcome measure should not be taken as reassurance that serious health outcomes did not occur. Although the absolute risk of hyponatremia was low, three key factors should influence the interpretation of the observed absolute risk. First, colonoscopy is a very common procedure. In 2002, ~14 million screening colonoscopies were performed in the United States (3). If our estimate of risk was applied to this population, 12,600 admissions with hyponatremia would be expected. Second, the diagnostic codes for hyponatremia at hospital admission have a sensitivity of only 10%, and some patients with severe hyponatremia are not admitted to the hospital or their serum sodium is not tested (26); therefore, our study underestimated the true risk by at least an order of magnitude. Third, colonic perforation is considered an important complication of endoscopy, and its risk (0.03%) is similar in magnitude to the risk of hyponatremia that we observed with sodium picosulfate (0.09%) (32). The American Journal of GASTROENTEROLOGY
ENDOSCOPY
Table 3. Outcomes of hospitalization with hyponatremia, hospitalization with computed tomography of the head, and mortality 30 days following a bowel preparation prescription
ENDOSCOPY
692
Weir et al.
There are several reasons why we are confident in our study’s findings. First, the tracer outcome of hyperkalemia showed no significant signal. Although residual confounding can always affect observational data, this finding provides some reassurance that its effects are not strong. Second, we attempted to limit the influence of indication bias by restricting our cohort to patients preparing for colonoscopy, rather than including patients prescribed bowel preparations for other common reasons such as constipation or preparing for abdominal and pelvic surgery. Nearly 80% of the cohort underwent colonoscopy within 30 days, suggesting that the indication for bowel preparation was relatively homogenous throughout our cohort. Third, as mentioned, the median serum sodium concentration of coded hyponatremia is 123 mmol/l (interquartile range: 119–126 mmol/l) (26). Therefore, most patients identified in this study had what would be considered clinically significant hyponatremia, despite our inability to detect its clinical impact through the proxy measure of urgent CT head. Finally, the pathophysiology of sodium picosulfate-induced hyponatremia fits well within established knowledge of serum sodium regulation. Circulating volume contraction has been demonstrated in patients taking sodium picosulfate through reductions in body weight (10,33–35), increases in hemoglobin concentration (10), postural tachycardia (10), and saline-remediable postural hypotension (33,35). Decreased circulating volume is a powerful stimulus for the release of antidiuretic hormone (36), which limits the kidney’s ability to excrete electrolyte-free water. In this physiologic setting, the intake of moderate quantities of water can lead to hyponatremia, which in some reports has been severe enough to cause seizures (15–17). Instructions included with sodium picosulfate recommend the consumption of between 3 and 4 l of clear liquid following its ingestion. Although the importance of consuming electrolytecontaining liquids in addition to water is stressed in these documents, the quantity of electrolytes consumed is at the discretion of the patient and could be limited by the nausea they commonly experience (37,38). The increased risk of hospital admission with hyponatremia that we observed with sodium picosulfate should not simply discourage its use. Sodium picosulfate is a very effective and welltolerated agent, and adequate bowel preparation is paramount for effective colonoscopy (9,39–41). Rather, our findings should serve as a reminder to those prescribing sodium picosulfate that electrolyte-free water must be consumed in moderation during its use. Our study has some important limitations. As with all observational studies, the exposures we studied were not randomly assigned; therefore, confounding variables may have influenced our estimates of risk. Although the two groups were well balanced across 24 baseline characteristics and the results proved robust even after adjustment for important predictors of hyponatremia, unknown variables or those that could not be measured may still have affected our findings. For example, adherence to the fluid intake instructions likely had a role in the development of hyponatremia, but this information was not available to us. In addition, physicians may have prescribed sodium picosulfate or The American Journal of GASTROENTEROLOGY
polyethylene glycol based on patient characteristics that may also have influenced the risk of hyponatremia. For example, patients with more severe manifestations of comorbidities (such as kidney disease or heart failure) may have had both a lower likelihood of tolerating the large volume of polyethylene glycol and a higher risk of hyponatremia. However, these characteristics were well balanced across the two groups. Our study was also limited by the use of administrative data to define exposures and outcomes. We included patients who filled prescriptions for bowel preparations, but we were unable to include those who purchased these drugs over-the-counter. Our previous validation work showed that our definition of hyponatremia significantly underestimates the true number of events, and this probably led to an inappropriately low estimate of the absolute risk. We planned to replicate this study using actual laboratory data, but the subset for which these data were available was too small to draw any meaningful conclusions (3% of the entire cohort, 27 total events). The main cohort’s low event rate also precluded meaningful subgroup analysis. We have previously used urgent CT head as a proxy for drug-induced delirium (28), but this may have been too blunt a metric to detect subtle findings associated with moderate hyponatremia. The pathophysiology of bowel preparation-associated hyponatremia requires hyponatremia to occur soon after the bowel preparation is consumed. Although we accurately ascertained the dates bowel preparation prescriptions were filled, we could not determine when they were ingested. To explore this issue, we used the date of colonoscopy as a proxy for the date of bowel preparation ingestion among the patients who experienced both colonoscopy and admission with hyponatremia during the observation period (post hoc analysis, Table 4). When we excluded cases of hyponatremia that occurred more than 72 h after colonoscopy, our findings did not change. Although this is reassuring, the uncertainty in the exact timing of bowel preparation ingestion remains. Polyethylene glycol was chosen as the referent bowel preparation, because it is considered to be electrolyte-neutral. The only reports of hyponatremia-associated polyethylene glycol have been with a hyperosmotic combination of Gatorade and Miralax (42,43). Numerous observational studies (44,45), clinical trials (46–50), and practice guidelines (6) support the assertion that polyethylene glycol in its common formulation is electrolyteneutral. There is no ideal bowel preparation, and our study highlights one important risk associated with sodium picosulfate. As sodium picosulfate is an effective bowel preparation, the risk of inadequate bowel cleansing must be weighed against the risk of hyponatremia. In addition, because water intake has an important role in sodium picosulfate-associated hyponatremia, physicians and manufacturers should consider our findings when instructing patients on the safest possible use of this product. ACKNOWLEDGMENTS
We thank Brogan, Ottawa, for use of its Drug Product and Therapeutic Class Database, Gamma Dynacare for use of the outpatient VOLUME 109 | MAY 2014 www.amjgastro.com
laboratory database, and the team at London Health Sciences Centre, St Joseph’s Health Care, and the Thames Valley Hospitals for providing access to the Cerner laboratory database. We also thank Natasha Chandok for her expertise in endoscopy diagnostic codes and Melanie Beaton for her help in preparing the manuscript. This study was conducted at the Institute for Clinical Evaluative Sciences (which is funded by an annual grant from the Ministry of Health and Long-term Care (MOHLTC)). CONFLICT OF INTEREST
Guarantor of the article: Matthew A. Weir, MD, MSc. Specific author contributions: Study design, data acquisition, analysis and interpretation, manuscript drafting, revising, and approval: Matthew A. Weir; data acquisition, analysis and interpretation, manuscript revising, and approval: Jamie L. Fleet, Chris Vinden, and Salimah Z. Shariff; data acquisition, manuscript revising, and approval: Kuan Liu and Haoyuan Song; analysis and interpretation, manuscript revising, and approval: Arsh K. Jain and William F. Clark; design, data acquisition, analysis and interpretation, manuscript revising, and approval: Amit X. Garg. Financial support: This project was conducted at the Institute for Clinical Evaluative Sciences (ICES) Western Site. ICES is funded by an annual grant from the Ontario Ministry of Health and Longterm Care. ICES@Western is funded by an operating grant from the Academic Medical Organization of Southwestern Ontario. The opinions, results, and conclusions reported in this paper are those of the authors and are independent of the funding sources. Matthew A. Weir also received funding from the MOHLTC. Amit X. Garg was supported by a Clinician Scientist Award from CIHR. Potential competing interests: None.
Study Highlights WHAT IS CURRENT KNOWLEDGE
3Sodium picosulfate is commonly prescribed and effective. 3Sodium picosulfate draws fluid into the bowel and can cause changes in serum electrolyte concentrations. 3Symptomatic hyponatremia in patients using sodium picosulfate has been described in case reports.
WHAT IS NEW HERE
3Hyponatremia is 2.4 times more likely to occur with sodium picosulfate than with polyethylene glycol. 3The absolute risk increase is considerable, given the frequency of colonoscopy.
REFERENCES 1. Bowles CJ, Leicester R, Romaya C et al. A prospective study of colonoscopy practice in the UK today: are we adequately prepared for national colorectal cancer screening tomorrow? Gut 2004;53:277–83. 2. Schultz SE, Vinden C, Rabeneck L. Colonoscopy and flexible sigmoidoscopy practice patterns in Ontario: a population-based study. Can J Gastroenterol 2007;21:431–4. 3. Seeff LC, Richards TB, Shapiro JA et al. How many endoscopies are performed for colorectal cancer screening? Results from CDC’s survey of endoscopic capacity. Gastroenterology 2004;127:1670–7.
© 2014 by the American College of Gastroenterology
4. Froehlich F, Wietlisbach V, Gonvers JJ et al. Impact of colonic cleansing on quality and diagnostic yield of colonoscopy: the European Panel of Appropriateness of Gastrointestinal Endoscopy European multicenter study. Gastrointest Endosc 2005;61:378–84. 5. Landreneau SW, Di Palma JA. Update on preparation for colonoscopy. Curr Gastroenterol Rep 2010;12:366–73. 6. Wexner SD, Beck DE, Baron TH et al. A consensus document on bowel preparation before colonoscopy: prepared by a task force from the American Society of Colon and Rectal Surgeons (ASCRS), the American Society for Gastrointestinal Endoscopy (ASGE), and the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES). Gastrointest Endosc 2006;63:894–909. 7. Heymann TD, Chopra K, Nunn E et al. Bowel preparation at home: prospective study of adverse effects in elderly people. BMJ 1996;313: 727–8. 8. Hosoe N, Nakashita M, Imaeda H et al. Comparison of patient acceptance of sodium phosphate versus polyethylene glycol plus sodium picosulfate for colon cleansing in Japanese. J Gastroenterol Hepatol 2012;27:1617–22. 9. Tan JJ, Tjandra JJ. Which is the optimal bowel preparation for colonoscopy - a meta-analysis. Colorectal Dis 2006;8:247–58. 10. Barker P, Trotter T, Hanning C. A study of the effect of Picolax on body weight, cardiovascular variables and haemoglobin concentration. Ann R Coll Surg Engl 1992;74:318–9. 11. Lieberman DA, Ghormley J, Flora K. Effect of oral sodium phosphate colon preparation on serum electrolytes in patients with normal serum creatinine. Gastrointest Endosc 1996;43:467–9. 12. Miki P, Lemos CR, Popoutchi P et al. Comparison of colon-cleansing methods in preparation for colonoscopy--comparative efficacy of solutions of mannitol, sodium picosulfate and monobasic and dibasic sodium phosphates. Acta Cir Bras 2008;23 (Suppl 1): 108–11; discussion 111. 13. Ryan F, Anobile T, Scutt D et al. Effects of oral sodium picosulphate Picolax on urea and electrolytes. Nurs Stand 2005;19:41–5. 14. Takada H, Ambrose NS, Galbraith K et al. Quantitative appraisal of Picolax (sodium picosulfate/magnesium citrate) in the preparation of the large bowel for elective surgery. Dis Colon Rectum 1990;33:679–83. 15. Dillon CE, Laher MS. The rapid development of hyponatraemia and seizures in an elderly patient following sodium picosulfate/magnesium citrate (Picolax). Age Ageing 2009;38:487. 16. Frizelle FA, Colls BM. Hyponatremia and seizures after bowel preparation: report of three cases. Dis Colon Rectum 2005;48:393–6. 17. Lewis M, Rugg-Gunn F, Don C et al. Bowel preparation at home in elderly people. Patients should be warned not to drink too much or too little fluid. BMJ 1997;314:74. 18. von Elm E, Altman DG, Egger M et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol 2008;61:344–9. 19. Levy AR, O’Brien BJ, Sellors C et al. Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol 2003;10:67–71. 20. Repchinsky C. Pico-Salax monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 1756. 21. Repchinsky C. Purg-Odan monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 1904. 22. Anon. PegLyte monograph. Pendopharm Products. Available at: http:// www.pendopharm.com/peglyte/. Accessed 8 November 2012. 23. Anon. GoLytely Monograph. Braintree Laboratories Monograph, Available at: http://www.nulytely.com/about-golytely.htm. Accessed 8 November 2012]. 24. Repchinsky C. CoLyte Monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 595. 25. Repchinsky C. Klean-Prep Monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 1229. 26. Gandhi S, Shariff SZ, Fleet JL et al. Validity of the International Classification of Diseases 10th revision code for hospitalisation with hyponatraemia in elderly patients. BMJ Open 2012;2:e0020. 27. Williams J, Young W. A summary of studies on the quality of health care administrative databases in Canada. Institute for Clinical Evaluative Sciences: Toronto, Ontario, 1996. 28. Tawadrous D, Dixon S, Shariff S et al. Delirium in older adults from standard dose histamine2-receptor antagonists. Manuscript in preparation.
The American Journal of GASTROENTEROLOGY
693
ENDOSCOPY
Sodium Picosulfate and Hyponatremia
ENDOSCOPY
694
Weir et al.
29. Jha P, Deboer D, Sykora K et al. Characteristics and mortality outcomes of thrombolysis trial participants and nonparticipants: a population-based comparison. J Am Coll Cardiol 1996;27:1335–42. 30. Austin P. Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research. Commun Stat Simulat Comput 2009;38:1228–34. 31. Mamdani M, Sykora K, Li P et al. Reader’s guide to critical appraisal of cohort studies: 2. Assessing potential for confounding. BMJ 2005;330:960–2. 32. Chukmaitov A, Bradley CJ, Dahman B et al. Association of polypectomy techniques, endoscopist volume, and facility type with colonoscopy complications. Gastrointest Endosc 2013;77:436–46. 33. German K, Chandiramani VJ, Stephenson TP. A study of the effects of Picolax on body weight, cardiovascular variables and haemoglobin concentration. Ann R Coll Surg Engl 1993;75:70. 34. Hoy SM, Scott LJ, Wagstaff AJ. Sodium picosulfate/magnesium citrate: a review of its use as a colorectal cleanser. Drugs 2009;69:123–36. 35. Sanders G, Mercer SJ, Saeb-Parsey K et al. Randomized clinical trial of intravenous fluid replacement during bowel preparation for surgery. Br J Surg 2001;88:1363–5. 36. Koeppen, Stanton B. Physiology of body fluids. In Renal Physiology, Mosby Physiology Monograph Series, 4th edn. Elsevier Inc.; Philadelphia, PA, 2013, 206 pp. 75–7. 37. Drew PJ, Hughes M, Hodson R et al. The optimum bowel preparation for flexible sigmoidoscopy. Eur J Surg Oncol 1997;23:315–6. 38. Swarbrick MJ, Collins MC, Moore DJ et al. A comparative trial of magnesium citrate (Citramag) and Picolax for barium enema bowel preparation. Clin Radiol 1994;49:379–81. 39. Hawkins S, Bezuidenhout P, Shorvon P et al. Barium enema preparation: a study of low-residue diet, “Picolax” and “Kleen-Prep”. Australas Radiol 1996;40:235–9. 40. Love J, Bernard EJ, Cockeram A et al. A multicentre, observational study of sodium picosulfate and magnesium citrate as a precolonoscopy bowel preparation. Can J Gastroenterol 2009;23:706–10.
The American Journal of GASTROENTEROLOGY
41. Schmidt LM, Williams P, King D et al. Picoprep-3 is a superior colonoscopy preparation to Fleet: a randomized, controlled trial comparing the two bowel preparations. Dis Colon Rectum 2004;47:238–42. 42. Nagler J, Poppers D, Turetz M. Severe hyponatremia and seizure following a polyethylene glycol-based bowel preparation for colonoscopy. J Clin Gastroenterol 2006;40:558–9. 43. Gerard DP, Holden JL, Foster DB et al. Randomized trial of gatorade/polyethylene glycol with or without bisacodyl and NuLYTELY for colonoscopy preparation. Clin Transl Gastroenterol 2012;3:e16. 44. Cohen CD, Keuneke C, Schiemann U et al. Hyponatraemia as a complication of colonoscopy. Lancet 2001;357:282–3. 45. Walia R, Steffen R, Feinberg L et al. Tolerability, safety, and efficacy of PEG 3350 as a 1-day bowel preparation in children. J Pediatr Gastroenterol Nutr 2013;56:225–8. 46. Clarkston WK, Tsen TN, Dies DF et al. Oral sodium phosphate versus sulfate-free polyethylene glycol electrolyte lavage solution in outpatient preparation for colonoscopy: a prospective comparison. Gastrointest Endosc 1996;43:42–8. 47. Cohen SM, Wexner SD, Binderow SR et al. Prospective, randomized, endoscopic-blinded trial comparing precolonoscopy bowel cleansing methods. Dis Colon Rectum 1994;37:689–96. 48. da Silva MM, Briars GL, Patrick MK et al. Colonoscopy preparation in children: safety, efficacy, and tolerance of high- versus low-volume cleansing methods. J Pediatr Gastroenterol Nutr 1997;24:33–7. 49. Hwang KL, Chen WT, Hsiao KH et al. Prospective randomized comparison of oral sodium phosphate and polyethylene glycol lavage for colonoscopy preparation. World J Gastroenterol 2005;11:7486–93. 50. Kastenberg D, Chasen R, Choudhary C et al. Efficacy and safety of sodium phosphate tablets compared with PEG solution in colon cleansing: two identically designed, randomized, controlled, parallel group, multicenter phase III trials. Gastrointest Endosc 2001;54: 705–13.
VOLUME 109 | MAY 2014 www.amjgastro.com
ORIGINAL CONTRIBUTIONS
nature publishing group
ENDOSCOPY
see related editorial on page x
Hyponatremia and Sodium Picosulfate Bowel Preparations in Older Adults Matthew A. Weir, MD, MSc1,2, Jamie L. Fleet, BSc2, Chris Vinden, MD3, Salimah Z. Shariff, PhD2,4, Kuan Liu, MMath2,4, Haoyuan Song, BPharm5, Arsh K. Jain, MD, MSc1,2,4, Sonja Gandhi, BSc2, William F. Clark, MD1,2 and Amit X. Garg, MD, PhD1,2,4,6 OBJECTIVES:
Bowel preparations are commonly prescribed drugs. Case reports and our clinical experience suggest that sodium picosulfate bowel preparations can precipitate severe hyponatremia in some older adults. At present, this risk is poorly quantified. We investigated the association between sodium picosulfate use and the risk of hyponatremia in older adults.
METHODS:
We conducted a population-based retrospective cohort study using six linked administrative databases in Ontario, Canada. All Ontario residents over the age of 65 years who filled an outpatient bowel preparation prescription before colonoscopy were eligible. We enrolled new users of either sodium picosulfate (n = 99,237) or polyethylene glycol (n = 48,595). The primary outcome was hospitalization with hyponatremia within 30 days of the bowel preparation assessed by database codes. The secondary outcomes were hospitalization with urgent head computed tomography (CT) (a proxy for acute central nervous system disturbance) and all-cause mortality.
RESULTS:
The baseline characteristics of the two groups, including patient demographics, comorbid conditions, and concomitant medications, were nearly identical. Compared with polyethylene glycol, sodium picosulfate was associated with a higher risk of hospitalization with hyponatremia (absolute risk increase: 0.05%, 95% confidence interval (CI): 0.04–0.06%, relative risk (RR): 2.4, 95% CI: 1.5–3.9), but not hospitalization with urgent CT head (RR: 1.1, 95% CI: 0.7–1.4) or mortality (RR: 0.9, 95% CI: 0.7–1.3).
CONCLUSIONS: Sodium picosulfate bowel preparations lead to more hyponatremia than polyethylene glycol. There
was no evidence of increased risk of acute neurologic symptoms or mortality. The absolute increase in risk of hospitalization with hyponatremia remains low but may be avoidable through appropriate fluid intake or preferential use of polyethylene glycol in some older adults. Am J Gastroenterol 2014; 109:686–694; doi:10.1038/ajg.2014.20; published online 4 March 2014
INTRODUCTION Millions of colonoscopies are performed worldwide each year (1–3). The safety and effectiveness of this procedure depend on the adequacy with which the bowel can be cleansed (4,5). A variety of effective oral bowel preparations are available, all with specific limitations (6). Isosmotic bowel preparations contain an isosmotic fluid and a nonabsorbable polymer (polyethylene glycol) that causes fluid to be retained in the colon. These bowel preparations are considered safer for patients at risk of electrolyte disturbances or circulating volume changes, because they do not cause significant shifts of
electrolytes or water across the colon wall (5,6). However, patients are required to consume 2–4 l of polyethylene glycol-containing solutions to adequately prepare the bowel. This, combined with poor palatability, limits their effectiveness (7–9). Hyperosmotic bowel preparations address the limitations of polyethylene glycol-containing solutions by providing a large amount of osmotically active molecules in a smaller volume of liquid. The ingested osmoles draw fluid into the colon and stimulate evacuation. However, this process of shifting fluid into the colon can result in intravascular volume contraction and electrolyte disturbances (10,11).
1 Division of Nephrology, Department of Medicine, Western University, London, Ontario, Canada; 2Kidney Clinical Research Unit, Western University, London, Ontario, Canada; 3Division of General Surgery, Department of Surgery, Western University, London, Ontario, Canada; 4Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada; 5Schulich School of Medicine and Dentistry, Western University, London, Ontario, Canada; 6Department of Epidemiology and Biostatistics, Western University, London, Ontario, Canada. Correspondence: Matthew A. Weir, MD, MSc, ALL-139A University Hospital, 339 Windermere Road, London, Ontario, Canada, N6A 5A5. E-mail: [email protected] Received 26 August 2013; accepted 14 January 2014
The American Journal of GASTROENTEROLOGY
VOLUME 109 | MAY 2014 www.amjgastro.com
Sodium picosulfate with magnesium oxide is a commonly prescribed hyperosmotic bowel preparation. In our clinical practice, we have encountered a number of patients hospitalized with significant hyponatremia after sodium picosulfate use. Three studies have identified asymptomatic hyponatremia (12–14), and three case reports have described hyponatremia-induced seizures in patients using sodium picosulfate (15–17). Although the risk appears greatest in older adults, it remains poorly quantified. Thus, we conducted this study to characterize the 30-day risk of hospitalization with hyponatremia in a large population of older adults prescribed sodium picosulfate in routine care. We compared these patients with a similar group of older adults prescribed polyethylene glycol. We hypothesized a higher risk of hyponatremia in patients prescribed sodium picosulfate vs. polyethylene glycol.
Patients
METHODS
Bowel preparations
Study design and setting
We conducted a population-based retrospective cohort study using health administrative data from Ontario, Canada. Ontario is Canada’s most populous province with ~13 million residents who receive universal access to hospital and physician services (Statistics Canada). Ontario’s 1.8 million residents over the age of 65 years also receive prescription drug coverage. The Research Ethics Board at Sunnybrook Health Sciences Centre approved the prespecified protocol for this study, and we have reported it according to established guidelines for observational studies (18).
Over the course of this study, the provincial drug plan provided coverage for two sodium picosulfate-containing bowel preparations, both comprising sodium picosulfate with magnesium oxide and citric acid (10 mg per sachet of sodium picosulfate, 3.5 g per sachet magnesium oxide, and 12 g per sachet of citric acid) (20,21). Four formulations of polyethylene glycol-containing bowel preparations were available during the study period, all of which had an osmolality of 280 mOsM, contained 60 g/l of polyethylene glycol, and had virtually identical electrolyte compositions except for CoLyte, which contained 80 mEq/l of sulfate compared with 40 mEq/l in the other three brands (Table 1) (22–25).
Data sources
Outcomes
We used six linked databases housed at the Institute for Clinical Evaluative Sciences to conduct this study. We ascertained vital statistics, including mortality, from the Registered Persons Database. The Registered Persons Database records the demographic information for people issued a provincial health card. We used the Ontario Drug Benefits database to ascertain prescription drug exposure and drug-related baseline characteristics. The Ontario Drug Benefit records prescription drug use for patients over the age of 65 years and has an error rate of < 1% (19). We identified admissions to hospital and baseline characteristics using the Canadian Institute for Health Information Discharge Abstract Database. The Canadian Institute for Health Information Discharge Abstract Database records hospital admissions and related diagnostic and procedural information. For similar information relating to emergency department admissions, we used the National Ambulatory Care Reporting System database. We used the Ontario Health Insurance Plan database to ascertain information relating to physician services, such as computed tomography (CT) of the head and colonoscopy. We determined physician specialty by using the Institute for Clinical Evaluative Sciences Physician Database. The Institute for Clinical Evaluative Sciences Physician Database pools information from three data sources and contains information about physician demographics, specialty training, certification, and practice location.
All outcomes were assessed in the 30 days after the bowel preparation prescription date.
© 2014 by the American College of Gastroenterology
We restricted enrollment to patients over the age of 66 years to ensure the availability of at least 1 year of drug-use data. Using records from 1 January 2006 to 31 January 2012, we identified new prescriptions for bowel preparations containing either sodium picosulfate or polyethylene glycol. Prescriptions were considered new if there was no other bowel preparation prescription filled in the preceding 120 days. To exclude prescriptions given for constipation, we eliminated those with durations of more than 3 days. We excluded patients who had bowel cleansing for bowel surgery (28% of the cohort). We excluded any patient with recent colonoscopy and those with more than one study bowel preparation or a prescription for a nonstudy bowel preparation (in order to compare mutually exclusive groups). Finally, we excluded those receiving chronic dialysis before their study drug prescription, as their sodium is regulated by the dialysis prescription.
Primary outcome: hospitalization with hyponatremia. We defined the primary outcome of hospital admission with hyponatremia as any admission with the E87.1 International Classification of Diseases, 10th revision diagnostic code (hypo-osmolality and hyponatremia) in any of the Canadian Institute for Health Information Discharge Abstract Database’s diagnostic fields. In Ontario, trained coders assign diagnostic codes to every hospital encounter on the basis of data from the patient’s chart. The Canadian Coding Standards do not permit coders to interpret laboratory test results, but they can record a laboratory-based condition if the physician has documented the diagnosis. For example, E87.1 would not be recorded for a serum sodium concentration of 120 mmol/l unless it was accompanied by a physician’s interpretation of “hyponatremia”. As a laboratory value becomes more extreme (i.e., lower levels of serum sodium), a code is more likely to be present for a particular diagnosis. In Ontario validation studies, the presence of E87.1 identifies patients with a median serum sodium level of 123 mmol/l (interquartile range: 119–126 mmol/l) and a decrement of 8.0 mmol/l (4.0–13.0) mmol/l from the baseline prehospital encounter serum sodium measurement. Although the specificity is The American Journal of GASTROENTEROLOGY
687
ENDOSCOPY
Sodium Picosulfate and Hyponatremia
688
Weir et al.
ENDOSCOPY
Table 1. Composition of polyethylene glycol-containing bowel preparations available during the study period Component
Klean-Prep
GoLytely
PegLyte
CoLyte
Sodium (mmol/l)
125
125
126
125
Sulfate (mmol/l)
40
40
40
80
diagnostic code (hyperkalemia) found in any of the Canadian Institute for Health Information Discharge Abstract Database’s diagnostic fields.
> 99%, the sensitivity is limited, particularly for milder forms of hyponatremia (26).
Post hoc analyses. Peer-reviewer commentary prompted us to conduct two post hoc analyses. To explore the effect of the prescribers’ specialty on the risk of hyponatremia, we grouped prescribers into three categories (gastroenterology and internal medicine, general surgery, and family medicine) and added a three-level interaction term to our multivariable model. To more strictly define hyponatremia associated with bowel preparations, we first explored the temporal relationship between colonoscopy and admission with hyponatremia among patients who experienced both of these outcomes. This group of patients allowed us to use colonoscopy as a proxy for the timing of bowel preparation ingestion. Using this proxy, we excluded cases where admission with hyponatremia occurred more than 72 h after colonoscopy.
Secondary outcome: hospitalization with CT head. We were interested in the potential neurologic sequelae of hyponatremia in this population because of case reports describing seizures in patients with sodium picosulfate-associated hyponatremia. We used hospitalization with evidence of an urgent CT head as a proxy for the presence of acute neurologic conditions such as delirium, seizures, and decreased level of consciousness. The typical practice in our region is to perform a CT head when a patient presents to hospital with any acute neurologic finding, and as the coding of CT scans is linked to remuneration we expect this proxy outcome to be recorded accurately (fee for service codes have high sensitivity and specificity (27); this differs from ascertaining delirium or decreased level of consciousness on their own, which are generally poorly coded in administrative data). To limit this outcome to acute neurologic conditions at the time of hospital presentation, we only considered CT head scans performed in the first 7 days of a hospital admission, or in the emergency department in the 2 days preceding the admission. We expected urgent CT head scans performed for reasons other than the effects of bowel preparationinduced hyponatremia to occur at similar frequencies in the two groups. We have successfully used this outcome of urgent CT in other population-based drug safety studies to characterize the risk of drug-induced acute neurologic findings (28).
Statistical methods. We compared the prevalence of baseline characteristics between the sodium picosulfate and polyethylene glycol groups using standardized differences. This metric describes differences between group means as a percentage of the pooled s.d. Those greater than 10% are considered meaningful (30,31). The risk of developing an outcome was expressed in both relative and absolute terms. Absolute risk was also expressed as the “number needed to harm” (1 divided by absolute risk difference), a measure that indicates how many patients needed to receive sodium picosulfate to cause harm to one patient who would not have experienced the event if they had received polyethylene glycol (a lower number indicating greater harm). We used multivariable logistic regression analyses (PROC LOGISTIC) to estimate odds ratios and 95% confidence intervals (CIs). We adjusted for the following 12 a priori-selected covariates: age, sex, chronic kidney disease, chronic liver disease, heart failure, bowel cancer, use of anticonvulsants, antidepressants, antipsychotics, benzodiazepines, loop diuretics, or thiazide diuretics. We also adjusted for rural area of residence and year of cohort entry after we noted imbalances in these characteristics between the two groups. Odds ratios were interpreted as relative risks (RR; appropriate, given the incidence rate observed). We conducted all analysis with SAS 9.2 (SAS Institute, Cary, NC).
Chloride (mmol/l)
35
35
35
35
Bicarbonate (mmol/l)
20
20
20
20
Potassium (mmol/l)
10
10
10
10
Polyethylene glycol (g/l)
59
60
59
60
Osmolality (mOsM)
280
280
280
280
Secondary outcome: all-cause mortality. In our region, the Registered Persons Database records the vital statistics of all residents covered by the provincial health-care plan. It has been shown to have a sensitivity of 97.8% and specificity of 100% for the finding of death (29). Tracer outcome: hospitalization with hyperkalemia. We tested the specificity of our findings by determining each group’s risk of hyperkalemia, an outcome that we did not expect to be influenced by the choice of bowel preparation. We defined hyperkalemia as any admission 30 days following the prescription date with the E87.5. International Classification of Diseases, 10th revision The American Journal of GASTROENTEROLOGY
RESULTS Figure 1 shows the steps of cohort assembly. Over 6 years, we identified 147,832 patients who met enrollment criteria, of whom 99,237 (67%) filled a prescription for sodium picosulfate and 48,595 (33%) for polyethylene glycol (Table 2). Twenty-four baseline characteristics were similarly distributed across the two groups, with standardized differences < 0.1. However, patients living in rural settings were more likely to receive sodium picosulfate, and those prescribing sodium picosulfate were slightly more likely to be surgeons. Sodium picosulfate prescriptions became more common over the study period, whereas polyethylene VOLUME 109 | MAY 2014 www.amjgastro.com
Sodium Picosulfate and Hyponatremia
689
Study drug prescriptions during the accrual period n = 313,536 1. Death at or before prescription, n = 90 2. Prescription for > 1 study drug, n = 1862 3. Prescription for a non-study drug, n = 61,296
Remaining eligible prescriptions n = 250,288 Restriction to new users Prescription for any bowel preparation in the preceding 120 days, n = 36,942
Remaining eligible prescriptions n = 213,346 Restriction to screening colonoscopy (1) 1. Colonoscopy in the preceding 14 days, n = 3,848 2. Bowel surgery in the preceding 14 days, n = 37 3. Hospital admission on the prescription date, n = 133 4. Emergency department admission on the prescription date, n = 789
Remaining eligible prescriptions n = 208,539 Restriction to screening colonoscopy (2) Bowel surgery in the 30 days following the qualifying prescription, n = 46,003
Remaining eligible prescriptions n = 162,536 Exclusion of patients receiving chronic dialysis Any chronic dialysis procedural code or fee code in the preceding 120 days, n = 557
Remaining eligible prescriptions n = 161,979 Restriction to the level of the patient Exclude repeat prescriptions, n = 14,147
Number of eligible patients n = 147,832
Figure 1. Flow diagram of cohort construction.
glycol prescription rates were unchanged throughout. Interestingly, despite the recommendation that patients with chronic kidney disease receive polyethylene glycol owing to its electrolyte neutrality, a number of these patients received sodium picosulfate. The majority of patients in both groups underwent colonoscopy within 30 days of filling their prescription. Procedure or billing codes for colonoscopy were detected in 78.6% of patients prescribed sodium picosulfate and 77.3% prescribed polyethylene glycol. Primary outcome
Results of the primary and secondary outcomes are presented in Table 3. Those prescribed polyethylene glycol made up the referent group. Prescription for sodium picosulfate was associated with a higher risk of hospital admission with hyponatremia (adjusted RR: 2.4, 95% CI: 1.5–3.9). Expressed in absolute terms, © 2014 by the American College of Gastroenterology
a prescription for sodium picosulfate was associated with a 0.05% higher incidence of hospital admission with hyponatremia (95% CI: 0.04–0.06). The corresponding number needed to harm was 1,903 (95% CI: 1,645–2,257). Secondary outcomes
Table 3 shows that no difference was found in the risk of hospital admission with urgent CT head on comparing sodium picosulfate with polyethylene glycol (adjusted RR: 1.1, 95% CI: 0.9–1.4). There was also no significant difference in the risk of all-cause mortality between the two groups (adjusted RR: 0.9, 95% CI: 0.7–1.3). Tracer outcome
The risk of hospital admission for hyperkalemia was not significantly different between the two groups (adjusted RR: 0.8, 95% CI: 0.4–1.6). The American Journal of GASTROENTEROLOGY
ENDOSCOPY
Data-cleaning exclusions
690
Weir et al.
Table 2. Baseline characteristics for sodium picosulfate and polyethylene glycol users
ENDOSCOPY
Sodium picosulfate (n=99,237)
Polyethylene glycol (n=48,595)
Standardized difference (%)a
Demographics 72 (69–77)
73 (69–78)
8
Female sex, no. (%)
Age, median (IQR)
52,825 (53.2)
25,015 (51.5)
4
Lowest income quintile, no. (%)
17,455 (17.6)
9,095 (18.7)
3
Rural location, no. (%)
16,216 (16.3)
4,324 (8.9)
23
Year of cohort entry, no. (%) 2006
5,595 (5.6)
8,166 (16.8)
36
2007
12,643 (12.7)
7,655 (15.8)
9
2008
15,940 (16.1)
8,162 (16.8)
2
2009
20,313 (20.5)
8,284 (17.1)
9
2010
22,120 (22.3)
7,991 (16.4)
15
2011
20,853 (21.0)
7,785 (16.0)
13
1,773 (1.8)
552 (1.1)
5
2012 (January only) Prescriber specialtyb, no. (%) Gastroenterology
24,042 (24.2)
14,603 (30.1)
13
General surgery
38,078 (38.4)
11,021 (22.7)
35
Family physician
14,342 (14.5)
10,305 (21.2)
18
Internal medicine Missing
3,769 (3.8)
2,017 (4.2)
2
15,573 (15.7)
8,617 (17.7)
5
35,239 (35.5)
16,633 (34.2)
3
c
Comorbidities , no. (%) Malignancy Chronic kidney disease
5,322 (5.4)
3,652 (7.5)
9
Chronic liver disease
3,885 (3.9)
2,447 (5.0)
6
Coronary artery disease
32,422 (32.7)
16,962 (34.9)
5
Diabetes mellitus
17,207 (17.3)
9,598 (19.8)
6
7,383 (7.4)
4,309 (8.9)
5
ACE inhibitors
30,939 (31.2)
15,942 (32.8)
4
ARB
Heart failure Medications, no. (%)
21,333 (21.5)
10,347 (21.3)
0
Anticonvulsants
4,164 (4.2)
2,284 (4.7)
2
Antidepressants
15,257 (15.4)
7,480 (15.4)
0
Antineoplastic medications
3,812 (3.8)
1,915 (3.9)
1
Antipsychotics
1,660 (1.7)
865 (1.8)
1
Benzodiazepines
14,303 (14.4)
7,339 (15.1)
2
Beta blockers
26,446 (26.7)
13,657 (28.1)
3
Calcium channel blockers
25,690 (25.9)
13,619 (28.0)
5
6,459 (6.5)
3,655 (7.5)
4
Loop diuretics Non-ASA NSAIDs
17,512 (17.7)
8,664 (17.8)
0
Statins
48,069 (48.4)
24,158 (49.7)
3
Thiazide diuretics
17,779 (17.9)
9,003 (18.5)
2
ACE, angiotensin converting enzyme; ARB, angiotensin receptor blocker; ASA, acetylsalicyclic acid; IQR, interquartile range; no., number; NSAID, nonsteroidal anti-inflammatory drug. a The standardized difference is the difference between group means expressed as a percentage of that characteristic’s pooled standard deviation. It is less sensitive to sample size than hypothesis testing. Standardized differences greater than 10% are considered significant. b Prescriber specialties are recorded in the OHIP database. Percentages do not total 100 because specialties responsible for < 1% of prescriptions were not included in the table. c Comorbidities were identified using administrative codes in the five years preceding the qualifying bowel preparation prescription.
The American Journal of GASTROENTEROLOGY
VOLUME 109 | MAY 2014 www.amjgastro.com
Sodium Picosulfate and Hyponatremia
691
Number of events (%) Picosulfate n=99,237
Polyethylene glycol n=48,595
Unadjusted relative risk (95% CI)
Adjusteda relative risk (95% CI)
Absolute risk difference, % (95% CI)
Number needed to harm (95% CI)b
Hyponatremia
93 (0.09)
20 (0.04)
2.4 (1.4–3.7)
2.4 (1.5–3.9)
0.05 (0.04–0.06)
1,903 (1,645–2,257)
CT head
257 (0.26)
126 (0.26)
1.0 (0.8–1.2)
1.1 (0.9–1.4)
—
—
Mortality
139 (0.14)
81 (0.17)
0.8 (0.6–1.1)
0.9 (0.7–1.3)
—
—
CI, confidence interval; CT, computed tomography. (—) Indicates not relevant to calculate given no difference in relative risk. For each analysis, patients prescribed polyethylene glycol were the referent group. a Adjustment was by multivariable logistic regression modeling with the following variables: age, sex, rural status, year of cohort entry, chronic kidney disease, chronic liver disease, heart failure, bowel cancer, use of anticonvulsants, antidepressants, antipsychotics, benzodiazepines, loop diuretics, or thiazide diuretics. b The number needed to harm is presented to facilitate interpretation; it does not imply causality.
Table 4. Colonoscopy and admission for hyponatremia: proportions and timing Admissions with hyponatremia Sodium picosulfate (N=93)
Polyethylene glycol (N=20)
No colonoscopy during observation period, no. (%)
56 (60.2)
11 (55.0)
Colonoscopy during observation period, no. (%)
37 (39.8)
9 (45.0)
Colonoscopy before admission, no. (%)
15 (16.1)
Median interval, days (IQR) Colonoscopy after admission, no. (%) Median interval, days (IQR)
3 (1–11) 22 (23.7) 4 (2–12)
a
4 (–) a
8 (–)
IQR, interquartile range; no., number. a To prevent the identification of patients, cell sizes of < 6 or cell sizes that could be used to derive the value of small cells cannot be reported.
Post hoc analyses
Prescriber specialty. When we stratified the cohort according to the specialty of the prescribing physician (gastroenterology and internal medicine vs. general surgery vs. family medicine), we found that prescriber specialty did not modify the risk of hospital admission with hyponatremia (P for interaction = 0.8). Stricter definition of admission for hyponatremia. Table 4 shows the timing of hospital admission with hyponatremia and colonoscopy among the patients who experienced both within the 30day observation period. The majority of patients admitted with hyponatremia did not receive colonoscopy during the observation period (60% of sodium picosulfate and 55% of polyethylene glycol). Among those who did have colonoscopy, a small proportion had the procedure before admission with hyponatremia. To remove cases of hyponatremia that are less likely to be the result of bowel preparation, we excluded admissions with hyponatremia that occurred more than 3 days following colonoscopy. This © 2014 by the American College of Gastroenterology
resulted in the loss of 10 events from the picosulfate group and less than 6 events from the polyethylene glycol group (privacy law prevents the reporting of cell sizes less than 6). Using this stricter definition, prescriptions for sodium picosulfate again associated with a higher risk of hospital admission with hyponatremia (adjusted RR: 2.9, 95% CI: 1.7–5.1). This estimate of risk was not materially different from that found in the primary analysis (adjusted RR: 2.4, 95% CI: 1.5–3.9).
DISCUSSION In this population-based cohort, we observed a significant 2.4fold increase in the RR of hospital admission with hyponatremia in patients prescribed sodium picosulfate compared with polyethylene glycol. Using the proxy measure of urgent CT head, there appeared to be no difference in acute neurologic events between the groups. The choice of bowel preparation also had no influence on 30-day all-cause mortality. The increase in the risk of hyponatremia we observed with sodium picosulfate is a concerning finding. Although patients in our study did not require more urgent CT head scans, they did experience symptoms that were severe enough to warrant hospital admission. Therefore, our failure to detect the sequelae of hyponatremia using a surrogate outcome measure should not be taken as reassurance that serious health outcomes did not occur. Although the absolute risk of hyponatremia was low, three key factors should influence the interpretation of the observed absolute risk. First, colonoscopy is a very common procedure. In 2002, ~14 million screening colonoscopies were performed in the United States (3). If our estimate of risk was applied to this population, 12,600 admissions with hyponatremia would be expected. Second, the diagnostic codes for hyponatremia at hospital admission have a sensitivity of only 10%, and some patients with severe hyponatremia are not admitted to the hospital or their serum sodium is not tested (26); therefore, our study underestimated the true risk by at least an order of magnitude. Third, colonic perforation is considered an important complication of endoscopy, and its risk (0.03%) is similar in magnitude to the risk of hyponatremia that we observed with sodium picosulfate (0.09%) (32). The American Journal of GASTROENTEROLOGY
ENDOSCOPY
Table 3. Outcomes of hospitalization with hyponatremia, hospitalization with computed tomography of the head, and mortality 30 days following a bowel preparation prescription
ENDOSCOPY
692
Weir et al.
There are several reasons why we are confident in our study’s findings. First, the tracer outcome of hyperkalemia showed no significant signal. Although residual confounding can always affect observational data, this finding provides some reassurance that its effects are not strong. Second, we attempted to limit the influence of indication bias by restricting our cohort to patients preparing for colonoscopy, rather than including patients prescribed bowel preparations for other common reasons such as constipation or preparing for abdominal and pelvic surgery. Nearly 80% of the cohort underwent colonoscopy within 30 days, suggesting that the indication for bowel preparation was relatively homogenous throughout our cohort. Third, as mentioned, the median serum sodium concentration of coded hyponatremia is 123 mmol/l (interquartile range: 119–126 mmol/l) (26). Therefore, most patients identified in this study had what would be considered clinically significant hyponatremia, despite our inability to detect its clinical impact through the proxy measure of urgent CT head. Finally, the pathophysiology of sodium picosulfate-induced hyponatremia fits well within established knowledge of serum sodium regulation. Circulating volume contraction has been demonstrated in patients taking sodium picosulfate through reductions in body weight (10,33–35), increases in hemoglobin concentration (10), postural tachycardia (10), and saline-remediable postural hypotension (33,35). Decreased circulating volume is a powerful stimulus for the release of antidiuretic hormone (36), which limits the kidney’s ability to excrete electrolyte-free water. In this physiologic setting, the intake of moderate quantities of water can lead to hyponatremia, which in some reports has been severe enough to cause seizures (15–17). Instructions included with sodium picosulfate recommend the consumption of between 3 and 4 l of clear liquid following its ingestion. Although the importance of consuming electrolytecontaining liquids in addition to water is stressed in these documents, the quantity of electrolytes consumed is at the discretion of the patient and could be limited by the nausea they commonly experience (37,38). The increased risk of hospital admission with hyponatremia that we observed with sodium picosulfate should not simply discourage its use. Sodium picosulfate is a very effective and welltolerated agent, and adequate bowel preparation is paramount for effective colonoscopy (9,39–41). Rather, our findings should serve as a reminder to those prescribing sodium picosulfate that electrolyte-free water must be consumed in moderation during its use. Our study has some important limitations. As with all observational studies, the exposures we studied were not randomly assigned; therefore, confounding variables may have influenced our estimates of risk. Although the two groups were well balanced across 24 baseline characteristics and the results proved robust even after adjustment for important predictors of hyponatremia, unknown variables or those that could not be measured may still have affected our findings. For example, adherence to the fluid intake instructions likely had a role in the development of hyponatremia, but this information was not available to us. In addition, physicians may have prescribed sodium picosulfate or The American Journal of GASTROENTEROLOGY
polyethylene glycol based on patient characteristics that may also have influenced the risk of hyponatremia. For example, patients with more severe manifestations of comorbidities (such as kidney disease or heart failure) may have had both a lower likelihood of tolerating the large volume of polyethylene glycol and a higher risk of hyponatremia. However, these characteristics were well balanced across the two groups. Our study was also limited by the use of administrative data to define exposures and outcomes. We included patients who filled prescriptions for bowel preparations, but we were unable to include those who purchased these drugs over-the-counter. Our previous validation work showed that our definition of hyponatremia significantly underestimates the true number of events, and this probably led to an inappropriately low estimate of the absolute risk. We planned to replicate this study using actual laboratory data, but the subset for which these data were available was too small to draw any meaningful conclusions (3% of the entire cohort, 27 total events). The main cohort’s low event rate also precluded meaningful subgroup analysis. We have previously used urgent CT head as a proxy for drug-induced delirium (28), but this may have been too blunt a metric to detect subtle findings associated with moderate hyponatremia. The pathophysiology of bowel preparation-associated hyponatremia requires hyponatremia to occur soon after the bowel preparation is consumed. Although we accurately ascertained the dates bowel preparation prescriptions were filled, we could not determine when they were ingested. To explore this issue, we used the date of colonoscopy as a proxy for the date of bowel preparation ingestion among the patients who experienced both colonoscopy and admission with hyponatremia during the observation period (post hoc analysis, Table 4). When we excluded cases of hyponatremia that occurred more than 72 h after colonoscopy, our findings did not change. Although this is reassuring, the uncertainty in the exact timing of bowel preparation ingestion remains. Polyethylene glycol was chosen as the referent bowel preparation, because it is considered to be electrolyte-neutral. The only reports of hyponatremia-associated polyethylene glycol have been with a hyperosmotic combination of Gatorade and Miralax (42,43). Numerous observational studies (44,45), clinical trials (46–50), and practice guidelines (6) support the assertion that polyethylene glycol in its common formulation is electrolyteneutral. There is no ideal bowel preparation, and our study highlights one important risk associated with sodium picosulfate. As sodium picosulfate is an effective bowel preparation, the risk of inadequate bowel cleansing must be weighed against the risk of hyponatremia. In addition, because water intake has an important role in sodium picosulfate-associated hyponatremia, physicians and manufacturers should consider our findings when instructing patients on the safest possible use of this product. ACKNOWLEDGMENTS
We thank Brogan, Ottawa, for use of its Drug Product and Therapeutic Class Database, Gamma Dynacare for use of the outpatient VOLUME 109 | MAY 2014 www.amjgastro.com
laboratory database, and the team at London Health Sciences Centre, St Joseph’s Health Care, and the Thames Valley Hospitals for providing access to the Cerner laboratory database. We also thank Natasha Chandok for her expertise in endoscopy diagnostic codes and Melanie Beaton for her help in preparing the manuscript. This study was conducted at the Institute for Clinical Evaluative Sciences (which is funded by an annual grant from the Ministry of Health and Long-term Care (MOHLTC)). CONFLICT OF INTEREST
Guarantor of the article: Matthew A. Weir, MD, MSc. Specific author contributions: Study design, data acquisition, analysis and interpretation, manuscript drafting, revising, and approval: Matthew A. Weir; data acquisition, analysis and interpretation, manuscript revising, and approval: Jamie L. Fleet, Chris Vinden, and Salimah Z. Shariff; data acquisition, manuscript revising, and approval: Kuan Liu and Haoyuan Song; analysis and interpretation, manuscript revising, and approval: Arsh K. Jain and William F. Clark; design, data acquisition, analysis and interpretation, manuscript revising, and approval: Amit X. Garg. Financial support: This project was conducted at the Institute for Clinical Evaluative Sciences (ICES) Western Site. ICES is funded by an annual grant from the Ontario Ministry of Health and Longterm Care. ICES@Western is funded by an operating grant from the Academic Medical Organization of Southwestern Ontario. The opinions, results, and conclusions reported in this paper are those of the authors and are independent of the funding sources. Matthew A. Weir also received funding from the MOHLTC. Amit X. Garg was supported by a Clinician Scientist Award from CIHR. Potential competing interests: None.
Study Highlights WHAT IS CURRENT KNOWLEDGE
3Sodium picosulfate is commonly prescribed and effective. 3Sodium picosulfate draws fluid into the bowel and can cause changes in serum electrolyte concentrations. 3Symptomatic hyponatremia in patients using sodium picosulfate has been described in case reports.
WHAT IS NEW HERE
3Hyponatremia is 2.4 times more likely to occur with sodium picosulfate than with polyethylene glycol. 3The absolute risk increase is considerable, given the frequency of colonoscopy.
REFERENCES 1. Bowles CJ, Leicester R, Romaya C et al. A prospective study of colonoscopy practice in the UK today: are we adequately prepared for national colorectal cancer screening tomorrow? Gut 2004;53:277–83. 2. Schultz SE, Vinden C, Rabeneck L. Colonoscopy and flexible sigmoidoscopy practice patterns in Ontario: a population-based study. Can J Gastroenterol 2007;21:431–4. 3. Seeff LC, Richards TB, Shapiro JA et al. How many endoscopies are performed for colorectal cancer screening? Results from CDC’s survey of endoscopic capacity. Gastroenterology 2004;127:1670–7.
© 2014 by the American College of Gastroenterology
4. Froehlich F, Wietlisbach V, Gonvers JJ et al. Impact of colonic cleansing on quality and diagnostic yield of colonoscopy: the European Panel of Appropriateness of Gastrointestinal Endoscopy European multicenter study. Gastrointest Endosc 2005;61:378–84. 5. Landreneau SW, Di Palma JA. Update on preparation for colonoscopy. Curr Gastroenterol Rep 2010;12:366–73. 6. Wexner SD, Beck DE, Baron TH et al. A consensus document on bowel preparation before colonoscopy: prepared by a task force from the American Society of Colon and Rectal Surgeons (ASCRS), the American Society for Gastrointestinal Endoscopy (ASGE), and the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES). Gastrointest Endosc 2006;63:894–909. 7. Heymann TD, Chopra K, Nunn E et al. Bowel preparation at home: prospective study of adverse effects in elderly people. BMJ 1996;313: 727–8. 8. Hosoe N, Nakashita M, Imaeda H et al. Comparison of patient acceptance of sodium phosphate versus polyethylene glycol plus sodium picosulfate for colon cleansing in Japanese. J Gastroenterol Hepatol 2012;27:1617–22. 9. Tan JJ, Tjandra JJ. Which is the optimal bowel preparation for colonoscopy - a meta-analysis. Colorectal Dis 2006;8:247–58. 10. Barker P, Trotter T, Hanning C. A study of the effect of Picolax on body weight, cardiovascular variables and haemoglobin concentration. Ann R Coll Surg Engl 1992;74:318–9. 11. Lieberman DA, Ghormley J, Flora K. Effect of oral sodium phosphate colon preparation on serum electrolytes in patients with normal serum creatinine. Gastrointest Endosc 1996;43:467–9. 12. Miki P, Lemos CR, Popoutchi P et al. Comparison of colon-cleansing methods in preparation for colonoscopy--comparative efficacy of solutions of mannitol, sodium picosulfate and monobasic and dibasic sodium phosphates. Acta Cir Bras 2008;23 (Suppl 1): 108–11; discussion 111. 13. Ryan F, Anobile T, Scutt D et al. Effects of oral sodium picosulphate Picolax on urea and electrolytes. Nurs Stand 2005;19:41–5. 14. Takada H, Ambrose NS, Galbraith K et al. Quantitative appraisal of Picolax (sodium picosulfate/magnesium citrate) in the preparation of the large bowel for elective surgery. Dis Colon Rectum 1990;33:679–83. 15. Dillon CE, Laher MS. The rapid development of hyponatraemia and seizures in an elderly patient following sodium picosulfate/magnesium citrate (Picolax). Age Ageing 2009;38:487. 16. Frizelle FA, Colls BM. Hyponatremia and seizures after bowel preparation: report of three cases. Dis Colon Rectum 2005;48:393–6. 17. Lewis M, Rugg-Gunn F, Don C et al. Bowel preparation at home in elderly people. Patients should be warned not to drink too much or too little fluid. BMJ 1997;314:74. 18. von Elm E, Altman DG, Egger M et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol 2008;61:344–9. 19. Levy AR, O’Brien BJ, Sellors C et al. Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol 2003;10:67–71. 20. Repchinsky C. Pico-Salax monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 1756. 21. Repchinsky C. Purg-Odan monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 1904. 22. Anon. PegLyte monograph. Pendopharm Products. Available at: http:// www.pendopharm.com/peglyte/. Accessed 8 November 2012. 23. Anon. GoLytely Monograph. Braintree Laboratories Monograph, Available at: http://www.nulytely.com/about-golytely.htm. Accessed 8 November 2012]. 24. Repchinsky C. CoLyte Monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 595. 25. Repchinsky C. Klean-Prep Monograph. In: Repchinsky C (ed). Compendium of Pharmaceuticals and Specialties. Canadian Pharmacists Association: Ottawa, ON, Canada, 2010, 1229. 26. Gandhi S, Shariff SZ, Fleet JL et al. Validity of the International Classification of Diseases 10th revision code for hospitalisation with hyponatraemia in elderly patients. BMJ Open 2012;2:e0020. 27. Williams J, Young W. A summary of studies on the quality of health care administrative databases in Canada. Institute for Clinical Evaluative Sciences: Toronto, Ontario, 1996. 28. Tawadrous D, Dixon S, Shariff S et al. Delirium in older adults from standard dose histamine2-receptor antagonists. Manuscript in preparation.
The American Journal of GASTROENTEROLOGY
693
ENDOSCOPY
Sodium Picosulfate and Hyponatremia
ENDOSCOPY
694
Weir et al.
29. Jha P, Deboer D, Sykora K et al. Characteristics and mortality outcomes of thrombolysis trial participants and nonparticipants: a population-based comparison. J Am Coll Cardiol 1996;27:1335–42. 30. Austin P. Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research. Commun Stat Simulat Comput 2009;38:1228–34. 31. Mamdani M, Sykora K, Li P et al. Reader’s guide to critical appraisal of cohort studies: 2. Assessing potential for confounding. BMJ 2005;330:960–2. 32. Chukmaitov A, Bradley CJ, Dahman B et al. Association of polypectomy techniques, endoscopist volume, and facility type with colonoscopy complications. Gastrointest Endosc 2013;77:436–46. 33. German K, Chandiramani VJ, Stephenson TP. A study of the effects of Picolax on body weight, cardiovascular variables and haemoglobin concentration. Ann R Coll Surg Engl 1993;75:70. 34. Hoy SM, Scott LJ, Wagstaff AJ. Sodium picosulfate/magnesium citrate: a review of its use as a colorectal cleanser. Drugs 2009;69:123–36. 35. Sanders G, Mercer SJ, Saeb-Parsey K et al. Randomized clinical trial of intravenous fluid replacement during bowel preparation for surgery. Br J Surg 2001;88:1363–5. 36. Koeppen, Stanton B. Physiology of body fluids. In Renal Physiology, Mosby Physiology Monograph Series, 4th edn. Elsevier Inc.; Philadelphia, PA, 2013, 206 pp. 75–7. 37. Drew PJ, Hughes M, Hodson R et al. The optimum bowel preparation for flexible sigmoidoscopy. Eur J Surg Oncol 1997;23:315–6. 38. Swarbrick MJ, Collins MC, Moore DJ et al. A comparative trial of magnesium citrate (Citramag) and Picolax for barium enema bowel preparation. Clin Radiol 1994;49:379–81. 39. Hawkins S, Bezuidenhout P, Shorvon P et al. Barium enema preparation: a study of low-residue diet, “Picolax” and “Kleen-Prep”. Australas Radiol 1996;40:235–9. 40. Love J, Bernard EJ, Cockeram A et al. A multicentre, observational study of sodium picosulfate and magnesium citrate as a precolonoscopy bowel preparation. Can J Gastroenterol 2009;23:706–10.
The American Journal of GASTROENTEROLOGY
41. Schmidt LM, Williams P, King D et al. Picoprep-3 is a superior colonoscopy preparation to Fleet: a randomized, controlled trial comparing the two bowel preparations. Dis Colon Rectum 2004;47:238–42. 42. Nagler J, Poppers D, Turetz M. Severe hyponatremia and seizure following a polyethylene glycol-based bowel preparation for colonoscopy. J Clin Gastroenterol 2006;40:558–9. 43. Gerard DP, Holden JL, Foster DB et al. Randomized trial of gatorade/polyethylene glycol with or without bisacodyl and NuLYTELY for colonoscopy preparation. Clin Transl Gastroenterol 2012;3:e16. 44. Cohen CD, Keuneke C, Schiemann U et al. Hyponatraemia as a complication of colonoscopy. Lancet 2001;357:282–3. 45. Walia R, Steffen R, Feinberg L et al. Tolerability, safety, and efficacy of PEG 3350 as a 1-day bowel preparation in children. J Pediatr Gastroenterol Nutr 2013;56:225–8. 46. Clarkston WK, Tsen TN, Dies DF et al. Oral sodium phosphate versus sulfate-free polyethylene glycol electrolyte lavage solution in outpatient preparation for colonoscopy: a prospective comparison. Gastrointest Endosc 1996;43:42–8. 47. Cohen SM, Wexner SD, Binderow SR et al. Prospective, randomized, endoscopic-blinded trial comparing precolonoscopy bowel cleansing methods. Dis Colon Rectum 1994;37:689–96. 48. da Silva MM, Briars GL, Patrick MK et al. Colonoscopy preparation in children: safety, efficacy, and tolerance of high- versus low-volume cleansing methods. J Pediatr Gastroenterol Nutr 1997;24:33–7. 49. Hwang KL, Chen WT, Hsiao KH et al. Prospective randomized comparison of oral sodium phosphate and polyethylene glycol lavage for colonoscopy preparation. World J Gastroenterol 2005;11:7486–93. 50. Kastenberg D, Chasen R, Choudhary C et al. Efficacy and safety of sodium phosphate tablets compared with PEG solution in colon cleansing: two identically designed, randomized, controlled, parallel group, multicenter phase III trials. Gastrointest Endosc 2001;54: 705–13.
VOLUME 109 | MAY 2014 www.amjgastro.com
