A retrospective descriptive study of the characteristics of deliberate self-poisoning patients with single or repeat presentations to an Australian emergency medicine network in a one year period

Deliberate self-poisoning (DSP) is a common cause for frequent presentations to the emergency department (ED), placing an increased burden on the ED [1]. Self-poisoning is the most common method of deliberate self-harm [2]. Presentation with DSP is often perceived negatively by staff involved in the treatment of these patients [3]. Staff attitudes and the patient’s ED experience have been shown to influence the patient’s decision to stay for treatment [4]. A poor ED experience may also exacerbate repeated self-harm behaviour [4]. In addition, an adverse experience may influence whether the patient seeks medical assistance for future self-harm [4]. These findings are significant considering that repeat self-poisoning behaviour results in an increased risk of subsequent successful suicide [5, 6].

In an era of increasing demand on the ED [7] and introduction of efficiency initiatives such as the 4-hour rule [8], understanding and describing frequent presenters is an important first step towards development and implementation of strategies to reduce re-attendances. Therefore, the aim of this study was to examine presentations to three Australian ED’s, in a one-year period by repeat DSP status. Characteristics examined were 1) patient characteristics, including age, gender, employment and psychiatric history; 2) presentation characteristics, including length of stay (LOS) data and disposition from the ED; 3) ED interventions with occurrence of psychiatric and toxicological consultations; and 4) type and amount of medications taken in the DSP.

This study was a retrospective chart review of DSP presentations to Monash Health’s three emergency departments from January 1^st 2011 to December 31^st 2011. Ethical approval for the study was obtained from Monash Health Human Research and Ethics Committee and Australian Catholic University. The definition of DSP was based on National Institute for Health and Clinical Excellence (NICE) definition of self-harm where self-harm is defined as ‘any act of self-poisoning or self-injury carried out by an individual irrespective of motivation’ [4]. Repeat poisoning presentations (RPP) were defined as greater than one presentation with DSP in 2011 to any of the three ED’s. A single presentation in 2011, following an episode of DSP, was termed single poisoning presentation (SPP).

A search of the Monash Health’s emergency department electronic database (SYMPHONY Version 2.29, Ascribe plc, Bolton, UK) was conducted to identify DSP presentations. Patients were included if aged ≥ 18 years presenting to Monash Health’s three EDs in 2011 with a presenting complaint of 1) Mental Health involuntary assessment required, 2) Mental Health voluntary requesting treatment or self-harm, 3) overdose/ingestion/poison/toxic exposure were included. The triage notes of each presentation were then screened and those deemed as accidental, recreational and non-DSP presentations were removed. This included presentations requesting psychiatric help or presentations for irrational/disturbed behaviour. Each DSP patient identified from the initial screening then had all their ED presentations in 2011 examined in Scanned Medical Records (SMR) for any missed DSP presentations. Emergency and hospital information about each DSP presentation was obtained from Monash Health SMR. This was then entered into a clinical database specifically designed for toxicology data acquisition. The database included information about the patient’s demographics, previous psychiatric history, toxins taken and amounts, admission and discharge dates, toxicology and psychiatric consults, critical care required, ventilation support required, discharge outcome and follow-up.

Age, employment, and marital status were determined at the index visit. The index visit was defined as the first presentation with DSP to the ED in 2011. The unemployed group included those patients who had identified themselves as unemployed or were on a disability support pension. These items were categorised together as a disability support pension may represent “hidden unemployment” [9]. Employement status distribution was analysed by repeat status. A secondary analysis was also undertaken where patients were categorised into unemployed and ‘other’. ‘Other’ was defined as either being either a student, employed, retired or undertaking home duties. Marital status was categorised into married/de facto relationship or single (single/widowed/separated/divorced).

Previous psychiatric history was obtained from emergency Enhanced Crisis Assessment Treatment Team (ECATT) or psychiatric notes from SMR. A person was deemed to have a previous psychiatric illness if they had documented personality disorder, bipolar affective disorder or schizophrenia.

Drugs ingested were categorised into major drug classes. The most common specific agents in each drug class were also examined. Drugs were described by presentation, that is, what percentage of presentations involved ingestion of a particular class of medication. Daily defined dose (DDD) was determined from the Anatomical Therapeutic Chemical/Defined Daily Dose (ATC/DDD) index 2014 World Health Organisation Collaborating Centre for Drug Statistics Methodology website [10]. Comparison of DDD between RPP and SPP was undertaken. A separate analysis of the DDD was also undertaken for patients with multiple frequent DSP presentations to the ED compared with SPP. Multiple frequent DSP attendance was defined as five or more presentations with DSP in 2011 [11].

Summary categorical statistics are expressed as number (percentage). Continuous data were examined for normality using the Shapiro-Wilk and Shapiro-Francia tests. Summary continuous statistics are expressed as mean ± standard deviation if normally distributed or median [inter-quartile range (IQR)] if not. Univariate logistic regression was used to analyse the relationship between all potential predictor variables and the binary outcome variable, RPP versus SPP. The effect size was described as odds ratio (OR) and 95% confidence interval (CI). Statistical significance was considered if the 95% CI included the value one. Hosmer-Lemeshow goodness of fit test was used to assess fit of logistic regression model. Variables were also examined for frequency of missing data. Except for LOS missing data, where single imputation was performed as explained above, complete case analysis was used to handle missing data [12]. Data analysis was performed using Stata version 12 (Stata Corporation, TX, USA).

This study confirms the results of a number of previous studies examining DSP repeat presenters observing that, this patient group was more likely to be unemployed and have a mental health illness compared to those presenting only once [13, 14]. This study found that RPP patients were more likely to be disadvantaged in other ways such as, they were more likely to abscond before care was completed and less likely to have a toxicology assessment even though similar doses of drug were ingested for most classes of medication. A similar severity of poisoning in both patient groups is suggested by the fact that ingested drug dose for most toxins was similar and the frequency of ICU admission was also not significantly different.

The proportion of DSP patients with recurrent presentations in the same year in this study was similar to that reported in other studies [14‐17]. However, two studies reported higher re-presentation proportions [13, 18]. One of these reported that 21% of DSP patients had recurrent presentations; however this was a retrospective study with a study period of ten years, which may have influenced the result [13]. The other study reported that 18% of DSP patients re-presented. The difference in representation proportion may be due to the prospective nature of the study as DSP patients, who met the study inclusion criteria, were consented to be followed over time [18].

Similar to our study, previous studies have also found that RPP patients were more likely to be unemployed [13, 18, 19], single [14], live alone [20], or have a psychiatric history [13, 18, 19]; particularly personality disorder or schizophrenia [19] than SPP patients.

Gender and age differences in RPP patients have been less consistent in previous studies, with two studies finding no gender influence on RPP behaviour [15, 16]. However, two other studies found the opposite, with female gender being a significant factor in RPP behaviour [13, 14]. Additionally, we found no association with age and RPP status. This is consistent with another Australian study [14]. Other studies have noted that RPP patients tended to be older than SPP patients [15, 19]. It is possible that there are geographical and cultural differences between studies. For example, the catchment area of the present study included a large proportion of people born overseas [21]. However, it was beyond the scope of this study to examine this further.

This study also found no difference between SPP and RPP patients in their disposition from the ED. However, an older study of self-poisoning found that RPP patients were less likely to be admitted to a hospital ward than SPP patients [15]. As RPP patients had lower triage category scores than SPP patients the authors hypothesised that RPP patients were less sick and hence less likely to be admitted.

Differences in length of stay between SPP and RPP groups have not been well characterised. A study undertaken in 1999 examined hospital LOS of the index visit for the RPP group versus the SPP group. It was found that the RPP group had statistically significant longer hospital LOS than the SPP group, but the authors felt the difference was not clinically significant [14]. Our study found that RPPs had a significantly longer LOS if critical care admission was required. However, the number of critical care admissions in the RPP group was not large and one patient contributed to three critical care admissions. In older studies of DSP patients, the proportion of patients admitted to hospital was considerably greater than in this study and other more contemporary studies. A study from 20 years ago reported that 70% of all DSP patients were admitted to an in-patient bed [16], which was considerably higher than in the present study. The lower in-patient admission rate may reflect the current practice of utilising ED short stay or observation units for patients with projected short lengths of stay (<48 hours) as is often the case with DSP patients.

Very few studies have examined whether the characteristics of toxicants ingested by SPP or RPP patients vary. In a retrospective study that examined the characteristics of patients presenting with DSP to a South Korean ED over a nine year period; the authors found that RPP patients were more likely to ingest sedatives or antidepressants whereas SPP patients tended to use sedatives or analgesics [13]. In an Australian study, the RPP characteristics for patients presenting to a Victorian hospital over a two-year period between 1993 and 1994 were examined. Those with greater than one presentation with DSP tended to take a single drug in the overdose compared to those that presented once [15]; this was not the case in this present study. However, similar to the current study, paracetamol overdose was higher in the RPP group. In another Australian study the class of drug ingested at the index visit in SPP and RPP patients was examined within one year period [14]. Benzodiazepines were most commonly ingested followed by paracetamol in both groups. In the RPP group, the third most common drug class was antipsychotics, whereas for the SPP group it was antidepressants [14]. This observation parallels the results of our study. Our study shows that RPP had higher use of quetiapine associated DSP but the ingested dose was lower in this group. Repeat DSP is often associated with pre-existing mental health conditions, such as schizophrenia and bipolar affective disorder [19], which allows easier access to antipsychotic prescription medication such as quetiapine. Concerns about the trend of increased rates of quetiapine prescribing and ambulance call outs associated with quetiapine use in recent years have been raised in Australia [22]. This study also reveals that although there was no significant difference in overall benzodiazepine ingestion between the two groups, the RPP group had statistically higher use of diazepam. This may reflect greater access to this drug in recurrent presenters.

Although there are methodological differences and over a decade has lapsed between this current study and the two previous Australian studies, they all show that the top four classes of drugs used in DSP, excluding alcohol, are benzodiazepine, paracetamol, antidepressants and antipsychotics [14, 15]. The impulsivity associated with DSP [23] may explain why paracetamol was the second most common ingested medication for patients who had repeated presentations following DSP in this current study.

This study confirms the results of other studies that alcohol co-ingestion is common in patients with DSP [1, 23]. This current study found no difference in frequency of alcohol co-ingestion between the two groups, however, we did not examine chronic alcohol use as a predictor for RPP.

We recently reported that ED nursing and medical staff felt empathy towards DSP patients who presented once but frustration towards DSP patients who presented repeatedly [24, 25]. The association of psychosocial history, unemployment and psychiatric illness, and more frequent absconding behaviour from the ED may result in staff feeling that they are dealing with patients who are difficult to manage, do not take responsibility for their actions, and do not want help. This finding is especially significant for those patients who re-present on numerous occasions. Many of the RPP patients were considered to be attention seeking, taking up resources and some staff queried whether the ED was the right place for these patients [24, 25]. The sub-analysis of the most frequent DSP attenders suggested that the severity of poisoning was similar to the single presenters. For those that ingested paracetamol the ingested dose was greater in the frequent repeat presenters, with a median DDD of 3.8. This is within the range of expected toxic ingested dose requiring antidote treatment [26]. This suggests risk assessment in the ED is important in this group.

The ED has an important role in the recognition of patients at risk of recurrent self-poisoning and re-presentation to hospital. Detection of patients at risk of re-harming may allow the implementation of evidence based interventions designed to reduce this phenomenon. This may include measures such as frequent postcards sent to DSP patients, telephone follow up, or follow up in self-harm clinics once discharged [27]. An increase in the severity of repeat DSP is associated with increased risk of future suicide completion [28]. As a result, a database of DSP patient presentations could identify those in the RPP group. Coupled with regular multidisciplinary clinical meetings, this would facilitate the development and implementation of specific care plans for patients.

The retrospective nature of this study meant that we were reliant on the integrity of the medical record data. The variables marital and employment status, documentation of toxicology and psychiatric consult, all had high levels of missing data, which may have biased the results. Variables regarding the type and amount of medications used in DSP episodes were not collected for some patients, such as those with altered conscious state, which may also have influenced the results. We used the ED presenting diagnosis to determine self-poisoning presentations and may have missed some patients presenting with DSP. Also, we only examined repeat presentations in a one year period and some patients in the single presentation group may have re-presented beyond the boundaries of the study time-frame. Patients may have presented to other hospital networks during the study period, which means that patients identified as single presentations may actually be recurrent presenters. While there were no in-hospital deaths in this cohort, we did not analyse coroner’s data to assess for completed suicide following DSP in the community. Variables that may have helped to explain repeater behaviour were not collected for this study. These included previous DSP episodes, drug and alcohol history, and associated self-injury with the DSP. Therefore, further empirical studies investigating those patients who present to the ED following repeated DSP presentations are required.

Our study provides more contemporary information regarding the characteristics of patients who present with repeated self-poisoning in an Australian setting. This patient group has pre-existing disadvantages, with increased likelihood of being unemployed and having a pre-existing mental health problem. They are also more likely to have health service inequities given the greater likelihood to abscond from the ED and lower likelihood of receiving toxicology consultation for their poisoning. However, repeat presenters had a similar severity of DSP as determined by frequency of critical care admission and similar doses and classes of drug ingested compared to single presenters. ED staff are well placed to facilitate the development and implementation of evidence based intervention and specific care plans for these patients.