METHODS

The Adverse Drug Reactions On-line Information Tracking (ADROIT) database is the MCA’s national database for ADRs. It contains details of reports of suspected ADRs that have been reported to the MCA since 1964 via the YCS.

The ADROIT database was searched for reports received by the MCA for children aged 16 years or less where the outcome of the suspected reaction to a drug was fatal. The database was searched from 1964 until December 2000. Reports associated with vaccines or overdose were excluded from the analysis. For each report the data retrieved included, where available: date of report and reaction; report number; sex, age, and weight of patient; suspected drug name and dose; duration of therapy; indication for suspected drug; suspected reaction; outcome of the reaction; certified cause of death; other drugs taken by the patient; and other medical history.

Each report was then assigned a report number in chronological order based on the lists prepared by the MCA. The group of drugs to which the suspected drug belonged was added to each report. The reports were then grouped according to the type of drug reported as associated with the suspected reaction. Groups that were suspected of causing a number of deaths were looked at more closely to identify any trends within the reports.

Where the suspected ADR listed as causing death was the same as the disease being treated, we listed death as being either a result of the underlying disease or as a sudden unexplained death. The possible causal association between drug exposure and the suspected ADR was not formally assessed in this study. However, the authors have reviewed the case details and have recorded the ADR they considered most likely to be associated with death. The age distribution of the patients for whom reports had been received was determined. The distribution of the number of reports received over time was also investigated.

RESULTS

There were 390 deaths with 389 suspected drugs. The median age of children for whom reports were received was 5 years, with the greatest number of reports for infants in the first year of life.

A wide range of drugs were reported as being associated with children’s deaths; table 1⇓ outlines the classes of drugs. The number of reported deaths has increased, with at least 10 deaths each year in the past decade (table 2⇓). There has, however, been an increase in the total number of yellow cards received for children, with little change in the proportion of yellow cards associated with fatalities over the past 25 years (0.37–1.00%) Table 3⇓ shows the wide variety of suspected ADRs. Table 4⇓ shows the 10 drugs most frequently associated with fatalities.

DISCUSSION

There is increasing interest in the safety of medicines used in children.⁸ There have been numerous studies of ADR surveillance of children in hospital, and a systematic review found that approximately 9% of children experience ADRs while in hospital.⁹ No previous studies have looked at fatalities in association with ADRs in children.

This study of fatalities associated with suspected ADRs in children has utilised the YCS which, while it has a proven track record in detecting signals of drug safety,¹⁰ cannot reliably be used to assess causality. Details that may explain the deaths, such as the nature of the underlying disease or co-administered drugs, may be lacking from the report. Thus the YCS acts largely as an early warning system generating hypotheses of previously unrecognised adverse reactions, as well as identifying increases in the frequencies or severity of previously recognised reactions. It is important to realise that the scheme invites reports of suspected ADRs. Limitations of the YCS include under reporting of reactions, lack of a denominator (that is, exposure to medicines), reporting rates being affected by factors such as time that the drug has been on the market, any media attention, and the variable quality of the data received.

The broad range of suspected ADRs associated with fatalities described in this paper shows that paediatric patients of all ages are prone to ADRs that are traditionally thought of as being restricted to adults. A wide range of medicines may be responsible for the ADRs. We believe that the 331 deaths reported as associated with suspected ADRs is likely to be an underestimate, as it is well recognised that ADRs are significantly under reported. Our study has not looked at evaluating the benefit of medicines, but for all the medicines studied the overall benefit to children is likely to be far greater than the risk. The evaluation of drug toxicity in paediatric patients is essential in order to improve the clinical use of medicines in a section of the population, where scientific information is limited.¹¹

The group of drugs most likely to be associated with a suspected ADR associated with death was anticonvulsants. Even if we exclude the seven cases where the fatality was thought to be associated with a seizure that failed to respond to treatment, there were still a significant number of deaths associated with anticonvulsants. The association between sodium valproate and hepatotoxicity is well recognised.⁴ However, the possibility of an underlying metabolic disease being responsible for some of the deaths needs to be considered, as this has been previously reported.¹² Guidelines for reducing the risk of sodium valproate hepatotoxicity were first described in 1987.⁴ These guidelines suggested either completely avoiding the use of sodium valproate or only using it with caution in children aged 2 and under, those with developmental delay, and those taking other anticonvulsants. It is not known how many of the children had developmental delay, but we note that since 1987 there have been four children aged 2 years or under who died following a suspected ADR of hepatotoxicity. Almost one third of the fatalities reported in children with epilepsy were associated with the newer anticonvulsants. Prospective studies are required to evaluate the risk to children of both the new anticonvulsants and older anticonvulsants such as sodium valproate.

The association between propofol and metabolic acidosis was first described in the UK in 1992.⁵ The propofol infusion syndrome was described in detail in 1998.¹³ Despite warnings from the Committee on Safety of Medicines⁷ and the manufacturer, doctors involved in the care of critically ill children are still continuing to use propofol as a sedative. As recently as 1999 our study found two reported fatalities. We are concerned that doctors may not have taken the advice given.

There is increasing recognition that pain is inadequately treated in children. NSAIDs are useful in the management of pain. Increasing use of NSAIDs may, however, be associated with an increased risk of toxicity. Gastrointestinal perforation in association with NSAIDs is well recognised in adults, but often not considered in children. There were, however, four deaths in children aged 9 days to 15 years following gastrointestinal perforation. The association between salicylates and Reye’s syndrome is well established. In the UK, aspirin is now contraindicated in children under the age of 12 years. A review of Reye’s syndrome in the USA shows a significant number of cases involving children between the ages of 12 and 15 years.¹⁴ The four deaths included in this report in children aged 12–14 years following the development of Reye’s syndrome in association with aspirin, in combination with two cases recently described,¹⁵ support the recent recommendation that salicylates should be avoided in children under the age of 16 years with a febrile illness.¹⁶

Inhaled corticosteroids have played a significant role in the management of asthma, and it is important to recognise that the vast majority of children do not experience ADRs. A recently published report described symptomatic adrenal insufficiency in association with the use of inhaled corticosteroids.¹⁷ The two fatalities reported in our study in association with the use of inhaled corticosteroids involved recognised complications of systemic corticosteroids—that is, adrenal insufficiency and sepsis.

Surfactant replacement therapy for the treatment of hyaline membrane disease in the neonatal period has been a major advance in the care of newborn infants and has saved many lives.¹⁸ Pulmonary haemorrhage is a recognised ADR associated with surfactant therapy.¹⁹ The relative risk of pulmonary haemorrhage, however, is small in comparison with the documented benefits of surfactant therapy.

Health professionals need to be aware of the risk of ADRs in children. Initiatives such as the establishment of a Paediatric Regional Monitoring Centre in Trent have shown that increased awareness of the risk of ADRs can be achieved.²⁰ Prospective studies are the only way to determine the risk benefits of medicines and the true incidence of ADRs.^21,22 A prospective study of 122 instances where diazepam was used for children with acute seizures in a children’s emergency department identified 11 cases of respiratory depression.²² Studies are required to assess the risk–benefit of medicines, and the pharmaceutical industry, drug regulatory authorities, and the Department of Health need to work with paediatricians and paediatric clinical pharmacists to ensure that such data are collected. The American Paediatric Pharmacology Research Unit Network has shown that such studies can be carried out,²³ and we believe there is a clear need for a similar network in the UK. Prospective studies of the risk–benefit of new anticonvulsants in children with epilepsy, and NSAIDs for pain and pyrexia, are needed to ensure that children receive optimal therapy.

Doctors need to be more aware of guidelines which recommend avoiding medicines in certain high risk groups. The use of propofol in the critically ill child and sodium valproate in young children under the age of 3 years, with developmental delay or polypharmacy, are examples where we hope further deaths might be avoided. An evidence based approach to drug therapy in children²⁴ is needed to minimise drug toxicity, which at worst could lead to death. Following guidelines and a greater awareness of the toxicity of certain medicines in specific age groups or situations is likely to minimise fatalities in children.