Bollettino epidemiologico nazionaleA Study of Adverse Drug Reactions in Children
Knowledge regarding the use and risk/benefit profile of drugs in pediatric practice in Italy is extremely limited. The systematic collection of information regarding the frequency, severity, and types of drugs most frequently involved in adverse reactions in the pediatric age group is therefore of particular interest, especially since information on the risks and benefits of drugs usually comes from pre-marketing clinical trials on adults and usually do not involve children.
To study adverse drug reactions in the pediatric age group, a case-control study was initiated in November, 1999 at Santobono-Pausilipon Hospital, a pediatric facility in Naples that admits 20,000 children annually. In November 2000, data collection was begun at 3 other clinical centers in Italy. Results from the first year of the study have been published elsewhere (1); this report includes an additional six months of and which further confirms initial findings.
The study involved the enrollment of all children admitted to the hospital through the emergency room, independent of their exposure to drugs, who had one of the following four diagnoses: gastro-esophageal or duodenal lesions that were endoscopically confirmed, central and peripheral nervous system problems, non-infectious mucocutaneous conditions and vasculitis, and thrombocytopenia with a platelet count <100,000. These conditions were chosen because most adverse drug events fall into one of these categories. In addition, children with adverse drug reactions not included among the above listed conditions (for example anaphylactic shock, accidental overdose, etc) were included in the surveillance system but not in the case-control study.
For each patient, parents were interviewed to construct a history of drugs and vaccines received within 3 and 6 weeks, respectively, of the initiation of symptoms leading to the present hospitalization. A case-control approach was used to analyze which, if any, drugs were associated with each adverse outcome. For purposes of the study, the cases were considered to be those children with one of the four adverse conditions, while the control group consisted of the children with the other three adverse conditions, and odds ratios (OR) were calculated for each drug. This design was used to ensure that the cases and controls were as comparable as possible, especially with respect to selection and recall bias. It should nonetheless be kept in mind that if certain drugs had caused more than one type of adverse event, the odds ratios presented may represent an underestimate of the “true” OR.
Between November 1999 and May, 2001, 325 children were admitted through the emergency room with the four conditions of interest (110 with muco-cutaneous illnesses, 39 with gastrointestinal lesions, 116 with neurologic problems, and 60 with thrombocytopenia). Among the 110 children hospitalised with muco-cutaneous illnesses, 17 (19%) had been given niflumic acid or monoflumate. Specifically, one child was diagnosed with Stevens-Johnson syndrome, 4 with Henoch-Schönlein-purpura, 8 with vasculitis, and 5 with urticaria and an exanthemous rash.
In the control group, only 10 of 215 children (5%) had been given niflumic acid. The odds ratio of developing a muco-cutaneous illness among users of niflumic acid, adjusted for concomitant use of all other drugs was 4.1 (95% confidence intervals (CI) 1.8-9.3). In addition, OR were calculated for drugs that had been taken by at least 10 children (amoxicillin + clavulinic acid and paracetamol), as well as for antibiotics as a group, which constituted overall the most frequently prescribed drug category. The OR for amoxacillin-clavulinic acid was 3.6 (95% CI 1.6-9.3) and for paracetamol was 1.4 (95% CI 0.8-2.6). Overall, the risk associated with antibiotics as 1.5 (95%CI 0.9-2.5). (Table).
Given the magnitude of the OR observed among the users of niflumic acid and antibiotics (particularly amoxacillin + clavulinic acid), it is unlikely that the results can be explained by selection or information bias. It is well-known that antibiotics may cause mucocutaneous reactions Nonetheless, given the efficacy of antibiotics, the only question in the evaluation of the relationship between benefit and risk is the appropriateness of their prescription. The case of niflumic acid is somewhat different. Given the existence of other analgesics and antipyretics that are safer in children (for example, paracetamol and ibuprofen), there are no indications that specifically require the use of drugs associated with higher risks of serious adverse events.
Following the diffusion of the results of this study, action was taken by the unit within the Ministry of Health dealing with the evaluation of drugs and pharmacosurveillance. The company producing niflumic acid sent a “Dear Doctor” letter to all the physicians in Italy containing information on the results of the study, and the surveillance of niflumic acid was extended prospectively to Santobono and the other three recently-enrolled hospitals.
1) Menniti-Ippolito F, Sagliocca L, Da Cas R,Saggiomo G, Di Nardo R, Traversa G, and the Santobono Study Group for Adverse Drug Reactions in Children. Niflumic acid and cutaneous reactions in children. Archives of Disease in Childhood, 2001, 84:430-431.
Francesca Menniti-Ippolito1, Giuseppe Traversa1, and Luciano Sagliocca 2
1Laboratory
of Epidemiology and Biostatistics, Istituto Superiore di Sanità, Rome
2Santobono-Pausilipon
Hospital Authority, Naples
Comment
Studies on drug safety conducted prior to their general use do not have the power to demonstrate an association with rare events, and often data are not available on specific population subgroups (in this case, niflumic acid in the pediatric age group). Post-marketing surveillance is specifically designed to identify unexpectedly elevated levels of adverse events associated with drugs and to spur analytical studies that permit further evaluation of a possible link between drugs and adverse events. In such systems, it is obviously important to analyze the data on a continuous basis and, if an association is discovered, to plan an appropriate intervention.
The criteria for choosing a given therapy include the efficacy and the safety profile of the drug as well as its cost. The task of the physician and pediatrician is always to prioritize the safety of the treatment while at the same time guaranteeing efficacy, while at least until recently in Italy, cost was often a secondary consideration. Often anti-inflammatory drugs are prescribed for mild illnesses (for example in upper respiratory tract infections), even if they are not expected to have a major therapeutic benefit. In the patient information package inserts as well as in the technical forms for this category of drugs, their indications include a long list of problems involving a wide variety of organ systems. The prescription of anti-inflammatory drugs, at least in pediatrics, could be drastically reduced if the marginal efficacy of these drugs for various conditions were taken into a count. When it is nonetheless necessary to use this category of drug, the aspect of safety becomes central, and in this case, there are several alternatives to niflumic acid that have a better safety profile (and in some cases are cheaper) and therefore should be preferentially prescribed.
Prescribing practices for antibiotics can also be improved. At least some of the prescriptions could be avoided. When used, the spectre of antibiotic resistance plays a role in the choice of antibiotic. For this reason, many physicians and pediatricians in Italy prefer to use the combination of amoxacillin + clavulinic acid as an altenative to cephalosporins since they believe that amoxacillin is well-tolerated and the addition of clavulinic acid increases the spectrum of coverage. This combination is frequently prescribed instead of ampicillin alone because of fear of infection with beta-lactamase producting bacteria in the absence of bacteriologic to guide antibiotic choice.
It is important to remember that no therapeutic intervention is completely free of side effects. Once again, this study reminds us that it is always necessary to balance the benefits and risks of the treatments that we prescribe.
Alberto Tozzi, Laboratorio di Epidemiologia e Biostatistica
