Article, Toxicology

Accuracy of trans-abdominal ultrasound in a simulated massive acute overdose

a b s t r a c t

Introduction: Toxic ingestions are a common cause of morbidity and mortality in the United States. In 2013, there were nearly 2 million reported ingestions and nearly 1000 deaths from poisonings. There is no well-validated im- aging study for confirming the presence of pills in the stomach of an overdose patient. There are case reports of ultrasound used for confirmation of pills in the stomach, and we are aware of one prospective trial to evaluate sonography for this application.

Objective: To determine the sensitivity and specificity of Transabdominal ultrasound for visualization of ingested pills. Methods: Randomized, blinded, placebo-controlled study. Twenty Subjects were randomized to one of two groups. The experimental group ingested 50 enteric-coated placebo capsules with 1 liter of fluid. The control group ingested only 1 liter of fluid. At time 0, 60, and 90 minutes after ingestion, expert sonographers performed transabdominal ul- trasound in an attempt to visualize pills in the stomach.

Results: At time 0, 60, and 90 minutes post-ingestion, ultrasound had a sensitivity of 62.5% (41.1%, 81.9%), 20.8% (4.6%, 37.1%), and 25% (7.7%, 42.3%) respectively. The specificity at time 0, 60, and 90 minutes post-ingestion were 58.3%

(21.9%, 61.4%), 79.1% (62.9%, 95.5%), and 70.8% (52.6%, 89.0%) respectively.

Conclusion: There is poor sensitivity and specificity for ultrasound detection of pills in the stomach at time 0; these values fell substantially at 60 and 90 minutes post-ingestion. In this pilot study, we did not find ultrasound to be a use- ful screening tool for detecting pills in the stomach.

Introduction

Background

medication overdose is a common presenting complaint to the Emer- gency Department (ED) and is a significant cause of morbidity and mor- tality. In 2013, there were nearly 2 million reports of poisonings from ingestions, and nearly 1000 deaths attributed to the same [1]. In many cases, traditional ED decontamination therapy with activated charcoal, whole bowel irrigation with polyEthylene glycol, and gastric lavage are limited to the first hour post-ingestion. Beyond this timeframe, these in- terventions are unlikely to be helpful because most pills have theoretically dissolved or moved out of the stomach. Unfortunately, the time of inges- tion is often unknown, and patients may be unable or unwilling to pro- vide the type or quantity of the overdose.

Study objectives“>? Presentations: This research was presented in March 2015 at the Western SAEM Regional conference in Tuscon, Arizona.

* Corresponding author at: 9040A Fitzsimmons Drive, Tacoma, WA 98431. Tel.: +1 253 968 0596.

E-mail address: [email protected] (S. Sullivan).

Importance

In acute toxic ingestions, diagnostic imaging is sometimes utilized in situations when there is uncertainty about the quantity or nature of in- gestion. Plain radiography and computed tomography have been de- scribed for this purpose, but both add time to a time-sensitive condition, and both have poor reliability and involve radiation exposure [2,3]. In the medical literature, there are a few case reports and one small prospective trial of ultrasound used for confirmation of pills in the stomach. [4-7] There is little data on the accuracy of ultrasound in the setting of massive overdose. Furthermore, there have been no trials that evaluate the use of ultrasound to evaluate for pills in the stomach at multiple time periods. If effective, ultrasound would allow visualization of ingested capsules in the stomach, and could potentially be used to ex- pand the window of decontamination therapy beyond 60 minutes.

Study Objectives

The objectives of this pilot study were to determine the sensitivity and specificity of transabdominal ultrasound for visualization of pills immediately after ingestion, and at 60 and 90 minutes post-ingestion.

http://dx.doi.org/10.1016/j.ajem.2016.04.030 0735-6757/

1456 S. Sullivan et al. / American Journal of Emergency Medicine 34 (2016) 1455-1457

Materials & Methods

Study Design and Selection of Participants

This randomized, prospective, blinded placebo-controlled pilot study was approved by the institutional review board. The study sub- jects were healthy volunteers, of both genders and assorted ages from the Department of Emergency Medicine. The sonographers were ex- perts in the field who had completed a one-year fellowship in Emergen- cy ultrasound. Prior to the study, the sonographers reviewed ultrasound approaches and techniques in visualizing the stomach and practiced scanning on a true positive and true negative live model of pills in the stomach as a base reference. The approach began anteriorly in the epi- gastric and left upper quadrant areas, and the probe was moved laterally to patient’s left flank as needed for visualization. Fig. 1 demonstrates this technique, and Figs. 2 and 3 demonstrate the sonographic view of a true negative and true positive.

Recruitment of subjects was performed by the associate investigator under the supervision of an ombudsman appointed by the institution to avoid any appearance of coercion. The risks of pill ingestion were discussed, and each participant gave informed consent. Subjects were in- cluded if they were healthy, active duty persons aged 18-65. They were excluded if they were pregnant females (we accepted verbal confirmation of non-pregnancy), had a history of inflammatory bowel disease or were prone to gastroIntestinal discomfort, or had any of the following symp- toms in the 24 hours preceding the experiment: indigestion, fever, nau- sea, vomiting, diarrhea, or abdominal pain. Subjects were also excluded if they reported a history of aspiration, cerebrovascular accident, dyspha- gia, or any other difficulty swallowing. Lastly, anyone who ingested any pills or capsules or violated the nothing by mouth requirements within 6 hours of the study was excluded from participation.

Twenty subjects meeting the above criteria were randomized to one of the two study groups. Randomization was performed with a computer- ized random-number generator using the “Research Randomizer” pro- gram at http://www.randomizer.org/form.htm. Study participants assigned to odd numbers were placed into the intervention (overdose) group, and those assigned even numbers were placed into the control group. Subjects randomized to the overdose group ingested 50 enteric- coated placebo capsules with 1 liter of water. Subjects in the control group ingested 1 liter of water with no capsules. All participants were given 20 minutes to complete their ingestions, and those in the overdose group were limited to swallowing no more than three capsules at a time for safety. A physician supervisor who was not participating in the exper- iment was present at all times. All participants had nothing by mouth (NPO) 6 hours prior to the ingestion time, and everyone completed the ingestion of pills and water or water alone in the allotted timeframe.

Data Collection and Processing

Three expert sonographers were blinded to the overdose history of the participants. They performed transabodominal ultrasound at 3 time intervals to determine if they could visualize pills in the stomach

or not. The scans were performed using either the Zonare Z One(TM) (Zonare Medical Systems, Mountain View CA) or the SonoSite(TM) M- Turbo (FUJIFILM SonoSite, INC, Bothell, WA). ultrasound scanning with the Zonare Z One(TM) was performed with the C4-1 curved-phased array transducer. The ultrasound scanning with the SonoSite(TM) M- Turbo was performed with a C60 series transducer. The subjects’ faces and pelvises were also covered with a sheet to avoid bias that would come from recognizing the same patient at different time intervals.

The three time intervals studied were 0, 60, and 90 minutes post- ingestion. Time 0 minutes post-ingestion started at the completion of the 20 minute consumption period, after the subjects had ingested all 50 cap- sules with 1 liter of water, or the 1 liter of water for the control group. The sonographers indicated their qualitative impression on the presence or ab- sence of pills on a sheet of paper for each study participant. “Yes” indicated that pills were seen, “No? indicated that pills were not seen. This data was then transferred manually by the principal investigator to an electronic spreadsheet using Microsoft Excel 2010(TM) (Microsoft Corporation, Red- mond, WA). The associate investigator independently verified the data.

Outcome Measures

The primary outcome measure was the ability to correctly determine the presence or absence of pills at the three time intervals using ultrasound.

Primary Data Analysis

The data was tabulated and analyzed using Microsoft Excel software to calculate sensitivity, specificity and confidence intervals. We used an exact confidence interval on proportion to determine the number need- ed in our study. We predicted a 90% sensitivity and 90% specificity as we determined this would be clinically useful. In order to get this, 72 scans in the placebo group and 72 scans in the overdose group would be need- ed and this would give a 95% confidence interval of 83% to 97%.

Results

A total of 20 participants completed the study, 10 in the overdose group and 10 in the control group, with three sonographers and a total of 144 scans. At time 0 minutes post-ingestion, we found a sensitiv- ity (95% confidence interval) and specificity (95% confidence interval) of 62.5% (43.1%, 81.9%) and 58.3% (21.9%, 61.4%), respectively. At the later time intervals of 60 and 90 minutes, the values for sensitivity were 20.8% (4.6%, 37.1%) and 25% (7.7%, 42.3%), respectively. The values for specificity at time 60 and 90 minutes were 79.1% (62.9%, 95.4%), and 70.8% (52.6%, 89.0%), respectively.

Discussion

Our study was the first of its kind in that it evaluated the utility of ul- trasound for detecting a known massive overdose in human subjects immediately following ingestion. Our study was also unique in that it evaluated the utility of ultrasound for this purpose at multiple time

Fig. 1. The approach used for stomach visualization with bedside ultrasound. In this case, the Zonare Z One(TM) C4-1 curved-phased transducer, oriented to the subject’s right and/or ceph- alad, was placed anteriorly in the epigastric/left upper quadrant and moved laterally to obtain a view of the stomach.

S. Sullivan et al. / American Journal of Emergency Medicine 34 (2016) 1455-1457 1457

Fig. 2. Scan of the stomach (from Lateral LUQ) of a true negative at Time 0 min using the Zonare Z One(TM) (Zonare Medical Systems, Mountain View CA).

intervals, in the hopes of potentially extending traditional decontamina- tion windows. Our results, unfortunately, do not demonstrate the utility of ultrasound in either scenario. However, although it was not a primary endpoint, we did develop a subjective appreciation for how difficult it is to swallow 50 enteric-coated capsules.

The use of ultrasound for the purpose of detecting ingested pills has been studied in human models before. Nordt et al. (2011) describes a randomized blinded study with ingestions of ten tablets per participant. The results showed poor inter-observer agreement for the detection of the ingestions. Despite this, there have been smaller studies showing that ingested pills can be visualized in the gastrointestinal tract. [4,5] Amitai et al. (1992) report successfully being able to detect 30 pills in vitro by ultrasound. This paper also suggests that different types of in- gestions may result in variable success of pill detection, as pills with sustained release or enteric coating were more consistently seen com- pared to immediate release formulations.

Limitations

This was a pilot study, and our small sample size limits the transfer- ability of our results to the general population. Regarding the number of total scans completed, using 3 sonographers, 20 participants, and 3 scans per participant would have yielded a total of 180 scans for the

Fig. 3. Scan of the stomach (from Lateral LUQ) of a true Positive at Time 0 min using the Zonare Z One(TM) (Zonare Medical Systems, Mountain View CA).

study. Due to logistical constraints that prevented our sonographers from completing all the scans, we only achieved a total of 144 scans. As a result, we were not able to analyze inter-observer variability based upon our resulting data set. Although this was sufficient to calcu- late sensitivity and specificity, we did rely on redundant scans by multi- ple sonographers. The wide confidence intervals support using more subjects in further study.

We made the assumption that all sonographers were of comparable skill, which may or may not have been the case. Also noteworthy, we only utilized one type of pill; an enteric-coated placebo capsule. This study did not evaluate co-ingestions, which are commonly seen in clinical practice, nor did it evaluate actual medications which could have an effect on gastric emptying [8]. We selected 50 capsules as an arbitrary but reason- able number to simulate a massive and potentially fatal overdose of even over-the-counter medications such as aspirin, but the number of ingested pills or capsules in clinical practice could obviously be highly varied.

The inclusion of 1 L of fluid in both groups deserves explanation. We acknowledge that actual patients presenting with acute overdose do not necessarily ingest such a consistent quantity of fluid. However, we the- orized that ingestion of a generous and standardized quantity of fluid should make it easier to visualize the pills, as fluid both distends the stomach and provides a better sonographic window to visualize any pills. We believe that this design supports our findings that the diagnos- tic accuracy of ultrasound for pill detection is poor, because despite the ingestion of 1 L of water and using only expert sonographers (optimal conditions), we failed to demonstrate reliable detection.

Finally, our population was limited to healthy, active duty military personnel. Applied to a larger population, there may be greater variance of age, body mass index, or altered gastrointestinal anatomy which may pose new technical challenges for the sonographer.

Conclusion

Bedside ultrasound in this pilot study poorly identified the presence of pills in the stomach, and the sensitivity of ultrasound for this purpose de- creased as time from ingestion increased. Given the results and limitations of this study, we cannot recommend ultrasound as a screening tool for de- tecting capsules in an overdose, nor can a claim be made to use ultrasound to detect capsules more than an hour after ingestion to extend traditional gastrointestinal decontamination windows. Further studies that have larg- er numbers of participants, recruited from actual patient populations, and address the non-fasting status of emergency department patients would be of benefit to make definitive conclusions. Secondary goals could look at different pill formulations, and techniques such as 3-D imaging to fur- ther evaluate point-of-care ultrasound for this application.

The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or reflecting the views of the Department of the Army, the Department of Defense, or the Unit- ed States Government.

References

AJEM