Single-dose activated charcoal as a pre-hospital treatment for self-poisoning

Methods

A search strategy was undertaken of medical and nursing databases including Medline, Cinahl, Embase, Psychinfo, the Cochrane database of systematic reviews, together with the British Medical Journal (BMJ), Emergency Medical Journal (EMJ) and the National Institute for Health and Care Excellence (NICE) clinical guideline website. Search terms used, combined with Boolean operators ‘and’, ‘or’ and parentheses were ‘paracetamol’, ‘acetaminophen’, ‘poisoning’, ‘overdose’, ‘charcoal’, ‘activated charcoal’, ‘SDAC’, ‘pre-hospital’ and ‘randomised controlled trial’ (RCT) limited to human only studies carried out in the last 10 years.

This produced four RCTs, twenty other studies, two systematic reviews and NICE Guideline CG16 (NICE, 2004). A systematic review evaluating various interventions for paracetamol overdose was selected (Brok et al, 2006), as it represented the highest level of evidence available (Aveyard, 2010) and the title and abstract suggested strong clinical relevance to the research need of the article (Crombie, 2004). NICE Guideline CG16 (2004) was selected as it was based on a systematic review, further evaluated to produce specific advice addressed directly to ambulance services. An RCT (Cooper et al, 2005) was selected due to its high position in the hierarchy of evidence and strong clinical relevance. The fourth article, a crossover trial (Green et al, 2001) whose title stood out as highly relevant to the enquiry, was selected despite its apparent lack of randomisation due to its otherwise sound methodology.

These articles will be reviewed in the above order using tools provided by the Critical Appraisal Skills Programme (Solutions for Public Health, 2006) for the systematic review, Centre for Evidence Based Medicine (2005) for the RCT, and the AGREE II instrument (Brouwers et al, 2010) for the NICE Guidelines. Unable to find a tool to evaluate a crossover trial, This article intends to assess it looking for bias.

Critical review

Brok et al (2006) present a systematic review comparing clinical benefits of a range of treatments for paracetamol overdose. The search strategy resulted in analysis of 59 studies, with primary outcomes considered being mortality and liver transplantation, and secondary outcomes a range of clinical symptoms including plasma-paracetamol concentration. Key findings were a reduction in plasma-paracetamol concentration negatively correlating with time to SDAC and evidence in favour of SDAC over gastric lavage or ipecacuanha when given within four hours of ingestion.

A strength of the review was the thoroughness of the search strategy with studies being selected by a review team based on four criteria—one of which essential to this article's underlying question—was data stratification by time to antidote.

Weaknesses of the study were the inability to present meta-analysis due to heterogeneity of inclusion criteria and outcome measures, and no primary outcome data in the SDAC studies. Although the tabular presentation of data, ranked by time to charcoal up to one hour all showed statistically significant mean reductions in plasma-paracetamol concentration, ranging from 20% to 85%, variation in the doses given and occasional addition of substances such as laxatives made direct comparison impossible.

The authors, as advised by Aveyard (2010), graded the studies with quality of randomisation and blinding the main criteria. The authors sensibly interpreted the contribution of the lower quality evidence conservatively in their narrative conclusion.

Most studies were volunteer trials using Area Under Curve (AUC) (Barile, 2004) measurement of plasma-paracetamol concentration as the outcome measure. While not directly translating to patient outcomes, the internal validity of this measure is in its non-subjective nature. The plasma-paracetamol AUC is commonly used to provide three nomograms of concentration versus time, classifying patients into three risk groups to plan treatment (Ferner and Dear, 2011) due to patients’ variation in vulnerability to paracetamol toxicity, which can be generalised to patients encountered in practice.

NICE (2004) reaches the conclusion that SDAC should be given to all conscious paracetamol overdose patients within one hour of ingestion. A systematic review examined the efficacy of SDAC at various time intervals after ingestion and SDAC versus other interventions or no intervention, with outcome measures being largely assessment of toxin biomarkers. Key findings, presented by meta-analysis, showed clear statistical benefit in all earlier doses of SDAC in comparison with those given at later times. Laine et al (1997) (n=32) showed a mean reduction in the plasma concentration AUC of amlodipine as the test substance of 99% (p<0.0005) when SDAC was given immediately versus six hours later, but weaker though still significant evidence in favour of SDAC at two hours versus six with a mean reduction of 49% (p<0.001). Rose et al (1991) (n=10) showed benefit in reducing availability of paracetamol recovered in urine by a mean of 48% with SDAC given at 15 minutes, 44% at 30 minutes and 33% at 121 minutes. NICE (2004) concludes a clear benefit to administering SDAC up to one hour.

Strengths of the guideline development process begin with the formulation of clear clinical questions, refined by clinical importance, by the Guideline Development Committee, leading to a particularly robust search strategy involving writing to authors for unpublished data, obtaining foreign language texts, and was continued to the end of the guideline writing process to ensure they were current.

Evidence was justifiably deemed to be strong if the confidence interval surrounding the effect size lay entirely in the clinically significant range, furthermore graded as to quality using a method similar to the GRADE guidelines (Balshem et al, 2011). This not only considers risk of all sources of bias, but as recommended by Balshem et al (2011), follows a thorough examination of alternative treatments by a review team including end-users of the guidelines, giving them added credibility.

A particular strength of recommendations made by NICE (2004) is not only in the clearly described relationship between the quality of evidence and the resulting recommendations (this was a Grade B recommendation) but also in NICE’s unique position as a guideline producer with an overview of the healthcare continuum. To ensure acceptable treatment with optimum cost/benefits for the NHS as a whole is adopted, NICE undertook a thorough consultation process involving service users and healthcare providers throughout the care process. This has ensured that not only clinical parameters have been evaluated but also patient and clinician values (Straus et al, 2005), leading to a high degree of generalisability to ambulance practice.

This guideline was reviewed in November 2011. The findings of the original guideline regarding charcoal were reiterated and the recommendations further incorporated into the NHS Pathways framework specific to ambulance services (NICE, 2011), giving added confidence that these are the most up-to-date and reliable guidelines available to inform ambulance practice.

Cooper et al (2005) concur with NICE that SDAC within one hour should be the first-line treatment. This RCT of patients presenting to an urban ED randomised them by a sequentially numbered envelope system to receive SDAC or no gastro-intestinal decontamination.

Results showed a mean length of stay (LOS), the primary outcome, of 6.65 hours (Interquartile range (IQR) 4 to 14) for patients receiving SDAC versus 5.5 hours (IQR 3 to 12) for the control group. No significant differences were found in secondary outcomes (vomiting, aspiration, ventilation and death). The only death occurred in the control group, following late presentation after multiple substance ingestion.

Mention of randomisation is brief, although an attempt to avoid selection bias has been made through the envelope method. Aftercare decisions made by clinicians blinded to treatment received was designed to avoid performance bias (Greenalgh, 2010).

The study has particular strengths in its external validity (Heneghan and Badenoch, 2006) as it included adults who had taken an oral overdose (paracetamol in a third of cases) with data stratified by time to SDAC versus outcomes.

Furthermore, a standard dose of SDAC was used in all cases on an intention-to-treat basis making it generalisable to patients’ outcomes. The sample size of 327 patients satisfied the power calculation of 300 (power 0.8 p<0.05) to detect a 33% reduction in LOS.

Both groups were prognostically similar with a good level of baseline characteristic homogeneity (e.g. age, gender, substance ingested), avoiding systematic bias (Straus et al, 2005).

Although it can be agreed with the authors that there was no significant difference in outcomes between the groups, given the short lengths of stay involved the importance of the primary outcome can be questioned. Interestingly, those in the SDAC at <2 hours group had a mean LOS of 5 hours (IQR 3.0 to 9.0) compared with those in the <4 hour group of 7.5 hours (IQR 3.3 to 14.0).

Significantly, Cooper et al (2005) has looked at incidence of adverse effects although detecting no significant differences. These are patient-important outcomes; however, it can be agreed that longer follow-up would enable assessment of long term and more important outcomes such as death or liver transplant missing from the study.

The authors used their findings to reduce use of SDAC in their hospital but citing the benefits shown by volunteer studies and the fact that they did not adequately test SDAC <1 hour, wisely suggest that SDAC within one hour should continue in rural settings until a much needed RCT shows otherwise.

Like Cooper et al (2005), Green et al (2001) aimed to evaluate the effects of time to SDAC to test the assertion made by Chyka et al (2005) that SDAC is only of value if given within one hour of overdose.

This crossover study used healthy volunteers (n=10) given SDAC at one, two and three hours post-ingestion to assess reductions in plasma-paracetamol concentration relative to a control arm of no intervention. The results showed a mean reduction in paracetamol bioavailability compared with the control of 30.5% (95% CI of the difference 14.1 to 46.4) in the one hour arm, 7.7% (95% CI of the difference 0.5 to 15.9) in the two hour group and 6.2% (95% CI of the difference −7.7 to 20.1) in the three hour group.

Although no randomisation appears to have taken place, the outcome measurement was fully blinded to avoid expectation bias (Greenalgh, 2010) as the assessor measuring all samples at the end of the trial was blinded to the arm of each sample.

Complete baseline homogeneity was achieved as each volunteer was in effect their own control, with a follow-up rate of 100% achieved.

The standard 50 g dose of SDAC used was externally valid but in interpreting the results it needs to be considered that all paracetamol doses were sub-toxic and the volunteers had fasted for eight hours before each ingestion. This may have had a detrimental effect on the two hour bioavailability reduction. As acknowledged by the authors, genuine patients will not have fasted and presence of food in the stomach delaying gastric emptying may make SDAC at two hours actually more effective than shown.

The authors’ conclusion that the data support SDAC up to one hour only is agreeable. The outcome measured here is reliable, reproducible and easily quantifiable, exhibiting a dose-response effect (Greenalgh, 2010) in terms of timing rather than quantity. It is also a reasonable predictor of disease. Ferner and Dear (2011) recommend that, due to patients’ variation in susceptibility to hepatotoxicity, plasma-paracetamol concentration should be used to determine treatments, especially as Parfitt and Henry (2002) note that most patients present asymptomatically.

Newman et al (2005) recognise that completely eliminating bias in research is impossible, suggesting that the bottom line is to ask whether this research is trustworthy enough to inform your practice.

The rigour and transparency with which NICE (2004) and Brok et al (2006) conducted their reviews suggests that their findings are currently the best available evidence. Cooper et al (2005) contains some room for performance bias in its inability to fully blind carers for the patients whilst the authors of Green et al (2001) have clearly gone to some lengths to ensure blinding of outcome assessment.

Each study, although having a different viewpoint, recommends broadly the same treatment plan. NICE (2004) is responsible as a national guideline producer to examine aspects such as user group opinion and cost/benefit to the NHS as a whole in the production of its recommendations (Dawe et al, 2005). Cooper et al (2005) evaluated the treatment in the light of short-term outcome measures and the cost/benefit in their personal setting, which could be biased towards recommending a local cost-saving. Brok et al (2006) takes a holistic view of the best prevention of liver damage, hence the longer term outcomes examined such as death and liver transplant rates. Green et al (2001) not unreasonably extrapolate from reductions in biomarker concentration data (Riegelman, 2005) a clinical benefit to the average patient to support an existing clinical algorithm.

All four studies reach the conclusion that SDAC is an effective inhibitor of paracetamol absorption with timing the main discussion point. Brok et al (2006) recommend that SDAC is given within one to two hours of ingestion. NICE (2004) and Green et al (2001) find that critical to its effectiveness is its use within a one hour time frame. Cooper et al (2005) agree that this is a sensible strategy in the absence of proof otherwise.

The patient populations in the genuine patient studies including Cooper et al (2005) are all generalisable to those encountered in an ambulance setting, and with the authors all concurring on the time critical nature of the treatment required, it can be concluded that this evidence is indeed trustworthy enough to inform current practice.

Discussion

Egger et al (2001) suggest that current policy making should be based on current best knowledge. Synthesising the recommendations from the above studies given from a variety of viewpoints provides a sound proposal for change in ambulance service practice.

This is a locally relevant topic that links in with national policy in the form of NICE guidelines, National Suicide Prevention Strategy (Department of Health, 2011) and requirements to implement best practice by the National Health Service Litigation Authority (NHSLA) (2011), which puts it on a strong footing (Craig and Smyth, 2002).

Kitson et al (2008) note that tension often exists between policy imperatives and support for local developments. A survey of ambulance Trusts by Greene et al (2005) found them reluctant to implement SDAC, the majority (58%) giving lack of a protocol as the main barrier. Twenty seven trusts said they would implement the treatment if a protocol were available. Other reasons included delaying time to ED. Considering the recommendations above it can be argued that this delay is desirable and justified by the ability to administer a timely intervention pre-hospital. Subsequently, Karim et al (2001) and Thakore and Murphy (2002) found that patients presenting to EDs rarely received SDAC within an hour of toxin ingestion, recommending the introduction of pre-hospital SDAC and fast track procedures for poisoning in EDs.

Cost implications present a major barrier to implementation of change, and cost/benefits need to be evaluated (Kulier et al, 2008). Ambulance services are likely to see minimal benefit to themselves in SDAC provision with an increase in drug expenditure and increased on-scene time of ambulances. It takes the broader view of NICE (2004) with an overview of overall cost to the NHS to drive this change, reinforced by persuasion from the NHSLA to implement NICE guidance.

An additional driver could be to include treatment of overdose amongst the Clinical Performance Indicators (CPIs) through which ambulance trusts are currently measured. Including it in the educational and monitoring system already applied to conditions such as myocardial infarction and stroke would ensure robust implementation of CPIs (Ambulance Service Network NHS Confederation, 2011).

The subject of SDAC has appeared on the SWASFT’s online clinical discussion forum as recommended by Greenalgh (2010), who suggests that evidence-based knowledge should be shared socially within an organisation as a preparatory step in the implementation process.

Graham et al (2006) advise that the final phase in implementing change is monitoring, to evaluate the effectiveness of dissemination through the ‘adopter group’ to sustain use of the knowledge.

This can be categorised as a high-quality recommendation requiring a small amount of change and as such, given proper dissemination, should be readily adopted by clinicians who tend to be aware of the knowledge but wait for guidance before implementing it (Kulier et al, 2008).

Conclusions

Returning to the original clinical question, it can be concluded that the articles above, together with other literature encountered, indicate that although gaps exist in current knowledge, a strong case exists for ambulance crews to administer SDAC on scene in cases of paracetamol overdose. With much evidence being volunteer based for ethical reasons, it would be justifiable to make reasonable assumptions beyond the data (Riegelman, 2005) as to the benefits of reducing plasma-concentration levels of the toxin.

As discussed by Greene et al (2005), a protocol for use by crews seems to be the sticking point with most trusts, alongside a widely acknowledged lack of quality trials evaluating the benefits of SDAC within one hour. Clearly these issues need to be addressed in order to move ambulance trusts towards providing this treatment.

RCTs give good estimates of immediate treatment effect but longer follow-up studies could be used to evaluate long-term benefits (Glasziou et al, 2004) examining one year survival or liver transplant rate to truly evaluate the financial and social benefits of early SDAC.

The Department of Health (2011) notes that a significant number of deaths still occur in England each year despite measures to prevent paracetamol overdose. Having gained a better understanding of the aetiology of paracetamol toxicity it is clear that this patient group is particularly vulnerable to life-threatening liver damage (Berk et al, (2008).

At £11.88 for a 50 g dose (Joint Formulary Committee, 2012), this is an inexpensive and easily administered pre-hospital treatment, with its effect potentiated by early administration, comparing favourably with the enormous financial costs of one liver transplant, in the region of £77 000 (Longworth et al, 2003) with its social consequences and ‘avoidable suffering’, which as noted by Greenalgh (2010) can be caused by failure to implement evidence.

The South Western Ambulance Service NHS Foundation Trust has recently concluded a six month feasibility trial of SDAC as a pre-hospital treatment. Paramedic crews in two zones of the Trust were equipped with SDAC and following a PGD, offered the treatment to suitable patients of overdose. Results and analysis are to be published in a future article.

Key points