Demographic Data
A total of 228 cases of suspected poisoning were screened of which 152 were included (Figure 1). The median age of the included children was 39 (IQR 25-61) months, of whom 86 (56%) were male and 113 (74%) were below 5-years-of age (table 2). According to the national immunisation schedule the immunisation status was up to date in 118 (78%) children. Thirty-nine (26%) lived in informal housing (table 2).
Table 2: Baseline characteristics of the study population (N =152)
|
Variable
|
n(%)
|
Male
|
86 (56%)
|
Age
|
< 1 year
|
14 (9%)
|
1- 5 years
|
99 (65%)
|
|
>5 – 12 years
|
31 (21%)
|
> 12 years
|
8 (5%)
|
Housing
|
Formal
|
96(63%)
|
Informal
|
39 (26%)
|
Unknown
|
17 (11%)
|
Immunization Status
|
Up to date
|
118 (78%)
|
Not up to date
|
15 (10%)
|
Unknown
|
19 (12%)
|
Toxicology results
Samples for all 152 study participants were analysed by LC-MS/MS. In 72 (47%), participants had urine alone tested, six (4%) had blood alone tested and 74 (49%) had both urine and blood tested. Altogether, in 89/152 (59%) participants a substance was detected. In 16 (18%) of these the detected substances were iatrogenic secondary to administration of in-hospital care. After discounting the iatrogenic substances 73 of 152 (48%, 95% CI 40 – 56%) participants had a substance detected by LC-MS/MS.
In total, 128 (84%) of the children, 71 (55%) of whom a toxin was detected on LC-MS/MS, were classified as genuine cases of poisoning (toxin-intake-likely), while 15 (10%) of the 152 were classified as unlikely to have been poisoned(toxin-intake-unlikely). In nine (6%) of the children it was not clear whether poisoning had taken place or not(toxin-intake-unclear).
Despite being classified as genuine cases of poisoning, 57 (45%) of the 128 toxin-intake-likely children did not have a causative substance identified via LC-MS/MS as, 49 (38%) children had no substance identified and eight (6%) had iatrogenic substances identified.
In 26 (20%) of the toxin-intake-likely group in which no substance was detected, the suspected toxin was not in the LC-MS/MS reference library used. Of these, 17/26 (65%) were pesticides (11 rat ‘poison’, 4 ‘cockroach poison’, 1 ‘tick poison’ and 1 undefined pesticide).
There were eight organophosphate poisonings cases in the toxin-intake-likely group. In two of the eight organophosphate poisonings, LC-MS/MS detected other toxins (bromazepam and diphenhydramine), ingested by the same patients. Likewise, in one of the four cases of iron poisonings, trimethoprim was concomitantly identified by LC-MS/MS. Eight patients who had ingested hydrocarbons, three ethanol ingestions, two turpentine, and one each of petrol, eucalyptus oil and paraffin ingestion, had no additional substances detected by LC-MS/MS.
Five (4%) patients in the toxin-intake-likely group presented with a history of ingesting an unknown toxin, and the identity of the unknown toxin was not identified via LC-MS/MS. Cannabis was detected via LC-MS/MS in a tablet brought by one of these patients but could not be confirmed in the patient’s samples.
Nine patients in the toxin-intake-likely group presented after ingesting a substance found in the LC-MS/MS library and yet the substance was not detected by LC-MS/MS despite the patients being symptomatic. The drugs that were not detected were the following, clonazepam, diazepam, lorazepam, phenytoin, alprazolam, cannabis, antiretrovirals (Tenofovir/emtricitabine/efavirenz), chlorpromazine and tricyclic antidepressant. Of note is that two of these patients received charcoal before the LC-MS/MS was done (one case of tricyclic antidepressant toxicity and one case of chlorpromazine ingestion).
Of the 15 patients, in the toxin-intake-unlikely group, LC-MS/MS detected no substances in eight (53%) and identified iatrogenic medicines in seven (47%). Of the nine toxin-intake-unclear patients, one patient had a positive result due to iatrogenic medicines and two had positive results, but the drugs identified could not explain the clinical presentation.
Presenting history versus LC-MS/MS results in poisoning cases(toxin-intake-likely)
Further analysis was done on the 128 in the toxin-intake-likely group, comparing the clinical presentation and the LC-MS/MS result. In 88(69%) there was a history of ingesting a known toxin, 16(12%) with an unknown toxin, while 24(19%) had no history of exposure to a poison (occult poisoning). In those who had occult poisoning, the suspicion of poisoning came from the clinician’s examination findings, and/or investigations done by the attending clinician. The toxin detection rate of LC-MS/MS after removing iatrogenic medicines was then analysed in the three different groups.
In children with occult poisoning, LC-MS/MS was able to identify the toxin in 22 (92%) of the 24 compared to 42/88(48%)when a guardian reported ingestion of a known toxin (p=<0.0001), or 7/16(44%) when a guardian reported ingestion of an unknown toxin (p value = 0.003) (figure 2). In the patients who reported ingesting a seemingly ‘known’ toxin, the toxin found on LC-MS/MS was different in 15/88(17%) patients. All the 15 patients who had presented with an unknown toxin and 23(96%) of the 24 cases of occult poisoning had neurological symptoms. Overall 18/128 (14%) cases of poisoning would have been missed had LC-MS/MS not been used in this study.
Causes of poisoning
In 106(83%) of the cases, poisoning was unintentional. There were however 6/128 (5%) cases of attempted homicide and 5/128 (4%) of attempted suicide (table 3).
Table 3: Causes of poisoning (Intent), n=128
Intention
|
Frequency (N=128)
|
Unintentional
|
Self
|
99 (77.3%)
|
Caregiver medication error
|
1 (0.8%)
|
Traditional medicine
|
3 (2.3%)
|
Iatrogenic
|
3 (2.3%)
|
Intentional
|
Attempted homicide
|
6 (4.7%)
|
Caregiver/adult but not attempted homicide
|
6 (4.7%)
|
Attempted suicide
|
5 (3.9%)
|
Self but not suicide attempt
|
1 (0.8%)
|
Undetermined
|
4 (3.1%)
|
Of the six attempted homicides, two cases involved siblings from a family that had three deaths due to the same event organophosphate poisoning. In one of the patients who had been given traditional medicines, norfluoxetine, trimethoprim and diphenhydramine were detected by LC-MS/MS. Four of six children given toxins intentionally by adults received drugs of abuse- two received cannabis, one received methamphetamine and the other ethanol. The other two patients, presented with neurological symptoms, and the substances administered could not be identified.
Drugs identified by LC-MS/MS
LC-MS/MS was able to identify a total 45 different drugs after removal of iatrogenic medicines (figure 3). In the 128 toxin-intake-likely cases LC-MS/MS identified 140 toxins. The most common causative group identified by LC-MS/MS was antihistamines found in 24 (19%) patients, followed by opiates in 23 (18%) and antipsychotics in 17 (13%). The most common drugs were chlorpheniramine and haloperidol found in 9 (7%) patients each. LC-MS/MS was able to identify multiple drugs in 40(31%) of the toxin-intake likely group.
Comparison of urine and blood LC-MS/MS results
Seventy-four (74) patients had both urine and blood samples analysed on LC-MS/MS. Urine and blood LC-MS/MS yielded the exact same result in 48 (65%) patients (table 4). In 18(24%) of the participants with paired samples, toxins were detected in urine but not in blood while in 4 (5%) samples, toxins were detected in blood but not urine.
Table 4: Comparing urine and blood LC-MS/MS positivity rate (N = 74)
LC-MS/MS Result
|
Frequency (%)
|
No detected toxin in urine and blood
|
27 (37%)
|
Same toxin detected in urine and blood
|
21 (28%)
|
Different toxin detected in urine and blood
|
4 (5%)
|
Toxin detected in urine and blood, but more toxins found in urine
|
7 (10%)
|
Toxin detected in urine and blood, but more toxins found in blood
|
3 (4%)
|
Toxin detected in urine but not in blood
|
11 (15 %)
|
Toxin detected in blood but not in urine
|
1(1%)
|
Total
|
74 (100%)
|
LC-MS/MS: Liquid chromatography tandem mass spectrometry
|
|
Clinical systems involved in the poisoning cases
The most common system involved was neurological, found in 88(69%) of the toxin-intake-likely cases followed by gastrointestinal found in 49 (38%), cardiovascular in 26 (20%) and 22 (17%) were asymptomatic. Of note is that, of the 49 that had gastrointestinal symptoms 24 (49%) had the presence of the confounder of intentional induction of vomiting by the caregivers using manual induction, milk and/or saltwater. LC-MS/MS detected a substance in 58(66%) out of 88 poisoning cases with neurological symptoms compared to 13 (33%) of the 40 without neurological symptoms (p<0.0001).
Toxin-intake-likely management and outcome
According to the PSS, most cases were classified as moderate, 51 (40%), while 12(9%) were classified as none and 42 (33%) were minor and therefore required minimal supportive care. Of the 23(18%) children with a PSS severe grade, 10 (8%) required admission to the Paediatric Intensive Care Unit (PICU). Twenty-nine children (23%) were given an antidote and 6 (5%) received activated charcoal. There were no deaths.
Individualized social intervention was instituted in all the patients with removal and emergency placement occurring in 6 patients. All six attempted homicide cases were referred for forensic investigation. The mother was the perpetrator in four of the attempted homicide cases. LC-MS/MS detected a toxin in three of the attempted homicides. A total of 22 (14%) patients had an LC-MS/MS result prior to discharge. The turnaround time was 5 (IQR 3 – 7) days.