When is emergency endoscopy required for caustic ingestions?

This is the first of a series of blog posts on new research in emergency toxicology. We deal with all sorts of poisons here in Virchester, so be prepared for anything.

Caustic substances are commonly used in the household and the workplace. Some are heavily alkaline, like bleach and ammonia. Others are acidic, like toilet cleaners and battery fluid. All have the potential to cause significant damage when ingested by humans, whether as an accident or as part of a suicide attempt. Injury usually occurs through corrosion of the oral mucosa, oesophagus, and stomach, although the airway may be affected if the substance is also aspirated or if there is “overspill” from the digestive tract.

The immediate consequences of caustic lesions may include upper airway swelling and gastro-oesophageal perforation. There is no “antidote” to prevent these complications. Some guidelines – e.g. UK National Poisons Information Service (NPIS) – recommend consumption of a diluent such as water or milk as first-aid. Still, the evidence that this provides anything more than symptomatic benefit is weak. Attempts to dilute the caustic substance should not delay transfer to hospital.  

Upon arrival to the emergency department (ED), these patients require early assessment to identify critical airway, oesophagus, or stomach injuries. In some ingestions, these will be clinically obvious. Patients who have perforated, for example, will generally present shocked and with clear signs of peritonitis or mediastinitis. Severe airway burns may manifest as stridor, hoarseness, or coughing.

In most cases, it is impossible to confidently rule in or out a significant corrosive injury from examination alone. This is where endoscopy comes in. Oesophago-gastroduodenoscopy (OGD) enables direct visualisation of caustic lesions and identification of patients at imminent risk of perforation – although CT imaging is typically also required before they are listed for surgery. OGD can also be helpful in estimating the risk of long-term complications, such as stricture formation.

Unfortunately, little evidence is available to help emergency clinicians decide when urgent OGD is required after caustic ingestion. Are there specific symptoms that we can use to guide our referrals? Does the nature of the ingestion—accidental or deliberate—matter in any way?

These questions were tackled in a paper published by Medical Toxicology earlier this year^​1​. The study aimed to determine if there are patients in whom we can forego emergency OGD after caustic ingestion and potentially manage with observation alone.

Abstract

Background: Caustic ingestions are relatively uncommon, but remain a significant source of morbidity. Patients with caustic injury often undergo an urgent EGD, although it is not clear if an EGD is routinely needed in an asymptomatic patient. The study has two primary objectives; 1) to determine the utility of routine EGD in asymptomatic suicidal caustic ingestions; 2) to determine if asymptomatic unintentional acidic ingestions can be managed with observation alone, similar to basic ingestions.

Methods This retrospective study, which took place at 14 hospitals in three countries evaluated all patients who presented with a caustic ingestion between 2014–2020. The presence of symptoms and esophageal injury, demographic information, pH of ingested substance, reason for the ingestion, and outcome were recorded.

Results 409 patients were identified; 203 (46.9%) were male. The median (IQR) age was 18 (4–31) years; overall range 10 months to 78 years. Suicidal ingestions accounted for 155 (37.9%) of cases. Dysphagia or dysphonia were more likely in those with significant esophageal injury compared to those without (59.3% vs. 12.6% respectively; OR 10.1; 95% CI 4.43–23.1). Among 27 patients with significant esophageal injury, 48% were found in suicidal patients, compared with 51.9% in non-suicidal patients (p = NS). On multivariate regression, there was no difference in the rate of significant esophageal injury among suicidal vs. non suicidal patients (aOR 1.55; p = 0.45, 95% CI 0.45–5.33). Most ingestions involved basic substances (332/409; 81.2%). Unknown or mixed ingestions accounted for 25 (6.11%) of the ingestions. Significant esophageal burns were found in 6/52 (11.5%) of acid ingestions, compared with 21/332 (6.3%) of basic ingestions. Of the 42 cases of acidic ingestions without dysphagia or odynophagia, 2 (4.8%; 0.58–16.1%) had significant esophageal burns, compared with 9 (3.2%; 95% CI 1.4–5.9%) of the 284 basic ingestions; p = 0.64). On multivariate logistic regression, patients with acidic ingestions were not more likely to experience a significant burn (aOR 1.7; p = 0.11, 95% CI 0.9–3.1) compared to those with basic ingestions. No patient with significant esophageal burns was asymptomatic.

Conclusion In this study, there was no statistical differences in the rates of significant burns between acidic and basic caustic ingestions. There were no significant esophageal injuries noted among asymptomatic patients.

What kind of study are we looking at?

This was a retrospective chart review conducted over six years (2014-2020) at fourteen hospitals in the United States, Canada, and Qatar.

Tell me about the patients

The study involved 409 adult patients presenting to hospital after ingestion of a caustic substance accidentally (62.1%) or deliberately (37.9%). They were a young sample – median age 18 – and 53.1% were female. The majority (81.2%) had ingested an acidic substance and the remainder had either consumed an alkaline (12.7%) or unknown caustic (6.1%).

Approximately half of the sample (48.7%) reported persistent pain after their ingestion, and vomiting was common (41.6%). Dysphagia and dysphonia were noted in 15.7% of patients and oropharyngeal lesions were seen in 16.4%.

Emergency OGD was performed in 150 patients.

What outcome measures were used?

The researchers were interested in significant oesophageal injuries, and they used a composite endpoint to record their occurrence. This included patients who were found to have a deep ulcer, necrotic lesion, or perforation on emergency OGD.

However, the endpoint also included patients who died after caustic ingestion, were found to have a perforation on CT imaging, or required an oesophageal intervention within the subsequent month. The researchers used this broad outcome to capture potential injuries in patients who did not undergo emergency OGD.

What were the main results of this study?

Significant oesophageal injuries were found in 27 (6.6%) of patients. All participants who met this endpoint initially presented with what the authors describe as a “serious sign or symptom” after caustic ingestion. Included under this heading were dysphagia, dysphonia, vomiting, and visible oropharyngeal lesions. Pain, notably, was excluded “[b]ecause the perception of pain is quite variable, and its assessment is subject to racial bias.”

Logistic regression analysis was used to see whether there were any useful predictors of severe burns apart from the red flags above. The most surprising result – in my view, at least – was that there was no association (p=0.45) between suicidal intent and injury severity. Age, on the other hand, seemed to matter. The odds of a patient over sixty-five incurring a significant burn were nearly three times (OR 2.9; 95% CI 1.6-5.7; p=0.001) the odds of a younger person meeting this endpoint.

The authors conclude from these results that regardless of suicidal intent or reported pain, patients who present after caustic ingestion “lack[ing] serious signs or symptoms” can forgo emergency OGD “as long as they remain asymptomatic after a period of observation.”

What should we take away from this?

We have to be careful here. This is a retrospective chart review, which is a methodology with some substantial limitations. The biggest is missing data. Complete, accurate, and available medical notes are required for retrospective data to be trustworthy. In this study, a concerning number of participants (29%) did not have an OGD or documented follow-up. It is therefore possible that nearly a third of patients complained of pain, did not go for endoscopy, left the hospital, and developed a significant oesophageal injury without this being known to the researchers.

It is also important to think about the classification of patients in this study. A high proportion swallowed caustic substances with apparent suicidal intent, but it is not clear how this was determined. Some patients may have concealed their motivations from the study doctors. Given the authors chose a fairly rare event – significant oesophageal injury – as their outcome of interest (n=27) it would only take a handful of patients who met the endpoint to be mis-classified as “accidental” to generate a type II error.

Finally, although a composite outcome was used, the majority of endpoint criteria in this study related to endoscopy findings. The interpretation of these images can be more subjective than we would expect and although a single grading system was used in this study (Zargar classification) there was no estimation of inter-rater reliability.

A prospective study would overcome some of these limitations. Unfortunately, it took a period of six years – and fourteen participating hospitals! – to generate the sample described in this paper. This is because caustic ingestions are an uncommon presentation to ED. It is expensive and time-consuming to conduct prospective studies with patients that we do not see very often.

Should this study change our practice?

Yes and no.

This study has encouraged me to be more vigilant in older patients who have swallowed caustic substances. To a lesser extent, it has also encouraged me to be more careful with accidental ingestions, which appear to share equal risk with deliberate ingestions. I would have guessed before reading this study that patients who drink corrosive agents with suicidal intent are likely to swallow a higher volume, and incur more significant burns. This is, perhaps, not the case.

At present, NPIS do not recommend that ED clinicians refer asymptomatic patients for OGD after caustic ingestion. This study provides no evidence to contradict this guidance.

NPIS do recommend urgent referral for endoscopy when patients present with drooling, dysphagia, vomiting, haematemesis, stridor, oropharyngeal lesions, or – importantly – severe pain. In this study, no patient with pain as their only presenting symptom was found to have a significant oesophageal injury. But again, I do not think we should take this at face value, as there were a high number of patients unaccounted for in the results.

Here in Virchester, we will probably continue to refer symptomatic patients for OGD.

vb

Greg Yates