| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Drug Monitoring and Toxicology |
1 Baylor University Medical Center, Dallas, Texas.
aAddress correspondence to this author at: Baylor University Medical Center, 3500 Gaston Avenue, Dallas, TX 75246. Fax 214-820-4837; e-mail m.emmett{at}baylorhealth.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
Methods: Diarrhea was induced in healthy volunteers by ingestion of bisacodyl, senna, or a control laxative (n = 11 for each laxative group). Samples of urine and diarrheal stool were sent in blinded fashion to the clinical reference laboratory for bisacodyl and senna analysis.
Results: TLC testing for bisacodyl-induced diarrhea revealed a sensitivity of 73% and specificity of 91% when urine was tested and sensitivity and specificity of 91% and 96%, respectively, when stool was analyzed. When diarrhea was induced by senna, the TLC assay for senna failed to identify even a single urine or stool specimen as positive (zero% sensitivity).
Conclusions: Considering the expected prevalence of surreptitious laxative abuse in patients with chronic idiopathic diarrhea (2.4%–25%, depending on the clinical setting), TLC of urine or stool for bisacodyl by this reference laboratory would often produce misleading results, and testing for senna would have no clinical value. The major problems are false-positive tests for bisacodyl and false-negative tests for senna.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
Because the diagnosis of surreptitious ingestion of laxatives is so often missed, it has been recommended that patients with idiopathic chronic diarrhea be routinely tested for laxative ingestion before invasive or costly diagnostic and therapeutic procedures (8)(17)(18). The 2 most frequently abused laxatives are bisacodyl (the active ingredient in Ex-Lax®, Correctol®, and Dulcolax®) and anthraquinones such as senna (the active ingredient in Senokot®, Castoria®, and Black Draught®) (8)(19)(20)(21). Ingestion of these laxatives can be detected by analysis of stool or urine, for the drugs per se, or for their metabolic products.
As far as we can determine, in the US all stool and urine specimens submitted to hospital laboratories for laxative screening are directed to a single reference laboratory. The procedure used by this laboratory is high-performance thin layer chromatography (TLC), and this technique has become the de facto standard for diagnosing surreptitious abuse of these laxatives, both in clinical practice and in legal proceedings (22). In support of this practice, research laboratories have reported 100% sensitivity and specificity when using TLC testing of urine to detect ingestion of bisacodyl or senna (19)(23)(24)(25). However, these validation studies were performed by investigators who were not blinded to the laxatives ingested by the study participants, and it is possible that prior knowledge of laxative intake contributed to the perfect results. Moreover, a 2-year pilot study from the UK concluded that some laboratories approached 100% reliability in detection of stimulant laxatives in urine, whereas others were <50% reliable (26). The tested laboratories were presumably located within different hospitals, and presumably did not include external reference laboratories.
We know of no previous study on the reliability of laxative testing by reference laboratories, where good performance is usually taken for granted. Therefore, we conducted an experiment to assess the accuracy of the only reference laboratory in the United States which offers such testing. This clinical reference laboratory (CRL) was not informed in advance of our study because prior notification is known to influence results of proficiency testing (27)(28)(29).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
The sample size requirements for this study were based on a normal approximation to the binomial distribution for comparing a single proportion to a known proportion (30). The known proportion is assumed to be 0.99, which corresponds to a sensitivity of 99% for TLC detection of bisacodyl and senna in urine (approximating the perfect results actually reported by research laboratories that developed and evaluated this method of laxative detection) (19)(23)(24)(25). A substantial difference from this known proportion is conjectured to be an absolute 20% decrease (i.e., 0.79). With a one-sided 
-level of 0.05 and statistical power of 
90%, the required sample size was calculated to be 11. Therefore, each of the 3 study arms (bisacodyl, senna, and control laxatives) involved 11 participants.
tlc assay
According to information provided by CRL, their assay for bisacodyl and senna is based on the method described by de Wolff et al. (19). The de Wolff procedure consists of high-performance TLC in 2 systems, after pretreatment of urine or stool with ß-glucuronidase and subsequent column extraction. Reference standards for bisacodyl include bisacodyl itself as well as its metabolic products (mainly dihydroxy compounds). After ingestion of bisacodyl by healthy people, the stool contains unchanged bisacodyl, whereas the urine contains only bisacodyl metabolites (19).
Recommended reference standards for anthraquinone laxatives consist of rhein and danthron. Rhein is the metabolic product of senna and cascara, and its presence in urine indicates ingestion of one of these drugs. Rhein is not found in stool after senna ingestion, and this is presumably why the de Wolff method does not detect senna ingestion by analysis of stool (19). The danthron standard is used only for detection of danthron ingestion (21). Danthron was withdrawn from the US market in 1987 because of its association with development of hepatic and intestinal tumors in laboratory animals.
CRL reports results for bisacodyl and senna as either "positive" or "none detected".
experimental protocol
The studies were approved by the Institutional Board for Human Protection of Baylor University Medical Center and informed consent was obtained. Experiments were conducted in healthy adult volunteers, 12 men and 10 women, whose mean age was 32 years (range 18–57). Most participants were employees of our hospital or students at a nearby seminary; the remaining participants were spouses or friends of these 2 groups of people. All study participants denied use of laxatives, or recent use of any other medication. They were paid a stipend for their participation. Some individuals participated in more than one arm of this experiment. The interval between such experiments was always >1 week, much longer than the time required for complete elimination of bisacodyl metabolites (24 h) or senna metabolites (32 h) from the urine (19)(31).
On experimental test days, participants were admitted to a clinical research laboratory at 7 AM, after they had fasted for 10 h. Under the direct supervision of the authors, participants ingested bisacodyl, senna, or a control laxative (Milk of Magnesia, castor oil, or polyethylene glycol). The doses of bisacodyl and senna were higher than those used by previous investigators (Table 1
). The order of laxative ingestion was random. Participants ate no food for 8 h after laxative ingestion, after which they were provided a standard meal and resumed a normal diet. Fluid intake was unrestricted, and no other medications were ingested. Each participant had a personal bathroom, which was used for no other purpose on experimental test days.
|
During the 24 h after laxative ingestion, each bowel movement was quantitatively collected in separate containers. Stool frequency and the weight of each bowel movement specimen were measured. Urine was collected quantitatively in a single container during the 24-h period. Stool and urine specimens were stored in a refrigerator (3–6 °C) or in a Styrofoam cooler with freezer packs (4–10 °C). At the end of the 24-h experimental period, aliquots of the specimens were frozen (–11 °C).
The assignment of fictitious names and labeling of samples were performed by 2 authors (JHS and CAS) to double-check their accuracy. Urine and stool samples from the same person were not identified as having a common origin. Submitted urine samples were aliquots of 24-h urine collections. Each submitted stool sample was taken from a single bowel movement after the onset of diarrhea. Frozen samples, packaged in dry ice, were mailed to CRL by overnight express.
statistical analysis
The reference standard for ingestion of bisacodyl, senna, or a control laxative was our direct observation of laxative ingestion. The sensitivity and specificity of TLC for detection of bisacodyl or senna ingestion, as performed by CRL, were calculated from laboratory reports from CRL, by use of standard methods (32). For calculations of statistical significance, one-sided testing was used and a P value of 0.05 or less was considered statistically significant. To test the hypothesis that the observed sensitivities and specificities were <99% and to determine confidence intervals, a normal approximation to the binomial distribution was used. In addition, Fisher’s exact test was used to compare sensitivity and specificity of urine vs stool.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
. Diarrhea also developed in all participants who ingested the 3 control laxatives (data not shown). There were no unexpected adverse events or complications associated with these experiments.
Results of bisacodyl and senna testing by CRL are shown in Fig. 1
and Table 2
. For purposes of calculating specificity and predictive value, ingestion of senna was considered an additional control for the bisacodyl assay, and vice versa.
|
|
Eight of 11 urine specimens from participants with bisacodyl-induced diarrhea tested positive for bisacodyl. The observed sensitivity of urine testing for bisacodyl was therefore 73%, lower than the expected sensitivity of 99% (P <0.001). Two of 22 urine samples from participants who had diarrhea due to laxatives other than bisacodyl also tested positive for bisacodyl, yielding a specificity of 91% (P <0.001). One false-positive urine test for bisacodyl was from a participant ingesting senna; this person had received bisacodyl in a prior experiment 7 weeks earlier. The other false-positive urine result came from a participant ingesting polyethylene glycol, who had never taken bisacodyl.
As shown in Fig. 1
and Table 2
, TLC testing for bisacodyl in stool had a higher sensitivity and specificity than TLC testing of urine, but the differences were not statistically significant. For stool and urine testing, there was no correlation between 24-h stool weight or urine volume and the likelihood of false-positive or false-negative tests for bisacodyl.
Of the 22 submitted urine and stool specimens from 11 participants with senna-induced diarrhea, none were reported as positive for senna, yielding a sensitivity of 0% (P <0.001 compared with the perfect results reported by research laboratories). Similarly, all the urine and stool samples from participants who ingested nonsenna laxatives were negative for senna, yielding a specificity of 100%.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
The results for bisacodyl and senna were strikingly different. Therefore they will be discussed separately.
bisacodyl
TLC analysis of urine for detection of bisacodyl ingestion was less sensitive and less specific than results reported by research laboratories. It is important to emphasize, however, that we submitted samples to CRL in a blinded fashion, whereas published studies from research laboratories made no claim that samples were analyzed in a blinded fashion. It is therefore possible that the differences between the present and previous results reflect observer bias in the historical data rather than technical inaccuracy on the part of CRL.
Stool specimen results obtained by CRL were somewhat more accurate than those achieved with urine specimens, possibly reflecting high concentrations of bisacodyl in stool and relatively low concentrations of bisacodyl metabolites in urine (19). Although most current recommendations advise testing of urine rather than stool specimens for detection of laxatives, our results suggest that stool analysis is just as good, and possibly better, for detection of bisacodyl.
Based on the observed sensitivity and specificity obtained with stool samples analyzed by CRL (Table 2
), we calculated the predictive value of a positive test for bisacodyl in stool as a function of the prevalence of bisacodyl ingestion in various populations (Fig. 2
). The actual prevalence of bisacodyl ingestion among patients with cryptogenic diarrhea under the care of general practitioners has been examined in 3 studies (8)(20)(33). Their combined findings revealed surreptitious ingestion of bisacodyl in 8 of 330 patients (2.4%). At this prevalence, only 29% of positive stool tests for bisacodyl would correctly identify that a patient had ingested bisacodyl, and 71% would represent false-positive results. On the other hand, bisacodyl testing would be more accurate later in the diagnostic evaluation of patients with idiopathic diarrhea, after extensive testing had failed to reveal a cause of diarrhea, when the prevalence of factitious diarrhea is 25% (4). Even at this higher prevalence, however, 15% of positive stool tests for bisacodyl would be false positives (Fig. 2
), and 9% of negative tests would be false negatives. If urine results were used, the inaccuracy of interpretation of positive and negative tests would be higher than depicted in Fig. 2
for stool.
|
If a positive test from either a urine or a stool sample from an individual was considered a positive result for that person, the sensitivity of TLC bisacodyl testing by CRL would be 100%, but the specificity would fall to 86%. If the prevalence of bisacodyl ingestion in a population were 10%, then more than half of positive tests (from either urine or stool) would represent false positives.
senna
Our participants ingested 69–103 mg of senna (8–12 tablets) and developed diarrhea, yet TLC testing by CRL did not detect senna in any of the 11 urine or 11 stool samples that were submitted. These findings are remarkable, given published reports from research laboratories that TLC of urine detects senna with 100% sensitivity (19)(23)(24)(25). The discrepancy cannot be explained by the amount of ingested senna, because the doses ingested by our participants were higher than the doses used in the published studies. Because CRL receives mailed samples, whereas the published data may have been derived from samples collected on site, it is possible that delays and differences related to packaging and mailing might contribute to poor performance by CRL. Laxatives and their metabolic products are stable in urine over time, however, and special collection or storage conditions are believed to be unnecessary (26). The cause(s) of the discrepancy are therefore unknown, although the failure to detect even a single instance of senna-induced diarrhea suggests a technical problem in one of the essential steps of the TLC assay, such as the failure to use an appropriate standard (rhein) or failure to pretreat samples with ß-glucuronidase (19).
Anthraquinone drugs are used in 50% of patients with factitious diarrhea (3)(4)(8). In the US, all anthraquinone-based laxatives sold in pharmacies contain senna, although products containing cascara can be purchased in health food stores. Because the TLC urine assay for senna and cascara depends on detection of the same metabolic product (rhein) (19)(20)(21)(23), the low sensitivity of the TLC assay for senna may portend a similarly low sensitivity of TLC for detection of cascara.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
It is important for clinicians to know that a positive test for bisacodyl increases the likelihood of surreptitious bisacodyl ingestion but that a positive test might also represent a false-positive result, and that a negative test for bisacodyl does not rule out bisacodyl-induced disease. In our opinion, TLC testing for anthraquinone laxatives such as senna should not be performed until a reliable test has been established. Therefore, as with many other factitious diseases, physicians must use clinical evidence from multiple sources to establish a diagnosis of surreptitious laxative ingestion, rather than rely primarily on laboratory tests.
|
Acknowledgments |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  R. Jones and G. Rubin Acute diarrhoea in adults BMJ, June 15, 2009; 338(jun15_1): b1877 - b1877. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
