Adverse drug reactions—the nature and scope of the problem

For most of recorded history, medical treatment was conducted using therapies that would today be considered complementary or alternative medicine. Medications were typically prepared—frequently by compounding—largely from botanic or other natural sources, and there was little standardization or regulation. Occasionally this approach resulted in the development of a specific therapy, of which the most notable example would be the development of digoxin for the treatment of cardiac disease by William Withering in the 18th century. In this case, a series of unique circumstances led to this discovery. Withering, who in addition to being a physician was a trained botanist, was interested in understanding how an herbal mixture used by a folk healer in Shorpshire, in the West Midlands region of England, improved dropsy (edema usually associated with heart failure) among patients who did not respond to the best medical therapy of the day. In this case, Withering’s botanic knowledge enabled him to determine that, among the 20 herbs in the mixture, only one, Digitalis purpurea or Foxglove, was likely to have biologic activity. Withering then conducted a series of clinical studies to determine what would be the most effective and safe dose of Foxglove to treat dropsy. His landmark publication in 1785, “An Account of the Foxglove and Some of Its Medical Uses; With Practical Remarks on Dropsy and Other Diseases,” is remarkable not only in the detail of the clinical investigations undertaken, but also in Withering’s candor in acknowledging that his studies had included treating patients with doses that produced toxicity. This early description of the potential for hazards associated with therapy was, unfortunately, largely unheeded.

The contemporary era of specific therapy has been most frequently attributed to the discovery of the antimicrobial activity of sulfanilamide. In 1935 Gerhard Domagk, a physician working for the pharmaceutical division of the German conglomerate IG Farbenindustrie (which previously and later was the Bayer Company) published his key paper describing the results of his work over the previous 3 years, which demonstrated that Prontosil, a compound derived from azo dyes, was able to treat streptococcal infection in mice. Subsequently it was discovered that the pharmacologically active molecule was sulfanilamide. While Sir Alexander Fleming had demonstrated the antibacterial effects of penicillin in 1929, his work had been conducted entirely in vivo and was framed in the context of using penicillin as a tool for isolation of certain bacteria. In contrast, by 1937 sulfanilamide was being routinely used clinically, with penicillin not being commonly available until the pressures of World War II led the British and American governments and industrial sectors to push the evaluation and manufacture of penicillin in industrial quantities.

The impact of introducing specific drug therapy cannot be overestimated. Infection was among the most common causes of death for patients of all ages, and the impact of effective antibiotic therapy was, in the words of Louis Thomas, “as if we could cure cancer”. However, while the ability to cure infections produced a paradigm change in medical care, this was not without cost. In fact, soon after sulfanilamide entered into routine clinical practice, rash and fever were described as associated with sulfanilamide therapy. It is likely that this was an early example of drug hypersensitivity. Crystalluria, the most notorious adverse event associated with sulfanilamide therapy, however, was toxicity produced by a solvent used in drug preparation. The early sulphonamides were very water-insoluble, and indeed crystalluria was a problem associated with sulphonamide therapy. The problem for children using sulphonamide treatment was compounded by the need to develop an alternative to the large sulphonamide tablets used at that time. An elixir of sulfanilamide was prepared using diethylene glycol as a solvent. Diethylene glycol is an excellent solvent, and a potent nephrotoxin. Elixir of sulfanilamide was responsible for more than 100 deaths, and the ensuing public outrage resulted in, among other things, the creation of the US Food and Drug Administration (FDA) and led to developments that produced today’s drug regulatory process.

These 2 adverse events—rash and fever associated with drug therapy and unexpected nephrotoxicity of a component in a drug formulation—provide examples of the diversity of effects manifested as ADRs. An ADR has been defined as “a response to a drug which is noxious and unintended and which occurs at doses normally used in man for prophylaxis, diagnosis, or therapy of disease, or for the modification of physiologic function”. This broad definition includes a wide range of possible complications of therapy, but does not include errors seen related to incorrect drug administration such as giving a drug by the wrong route or in a toxic dose. The latter events are referred to as adverse drug events.

It is now appreciated that ADRs are both common and important complications of therapy. While the rate of ADRs varies depending on a number of variables including the drug, disease being treated, and the circumstances of the individual patient, it appears that for most drugs the minimal rate of ADRs that can be expected during therapy is approximately 5%, with many drugs being associated with substantially higher rates of ADRs. Consequently, ADRs are associated with substantial morbidity and mortality. In fact, ADRs have been estimated to be between the 4th and 6th most common causes of death in the United States and Canada and have been associated with health care costs in the billions of dollars.

Adverse drug reactions—implications for child health

There are several myths that enjoy widespread circulation as to the degree and range to which drug therapies, and as a consequence ADRs, are used in children, as well as to relative risks of ADRs in children versus adults. These myths have led to complacency, which has not fostered a culture of patient safety in child health care.

The first of these myths has to do with drug utilization. It is commonly believed that children rarely receive drug therapy and when they do it is primarily in the form of antibiotics. This issue has actually been studied. All prescriptions dispensed to a cohort of 1 million Canadian children, roughly 15% of all children in Canada, in 1999 were retrieved. Over this 1-year period, the 1,031,731 Canadian children (<18 years) received 4,028,502 prescriptions, roughly 4 prescriptions per child per year. Of the drugs prescribed, antibiotics were indeed the most commonly prescribed drugs. However, they accounted for less than half of all drugs prescribed. There was a wide range— indeed more than 1300 distinct pharmacologic entities—that accounted for the remainder of the drugs prescribed ( Table 1 ). As well, the overall numbers require some context to understand actual drug use among children. Despite the large number of prescriptions, most children are well and need few if any prescription drugs. As shown in Fig. 1 , the largest amount of drug utilization—in fact, more than 70—occurs among roughly 25% of children. Thus, there are 2 distinct populations of children,children who are otherwise well and need perhaps one prescription a year, often an antibiotic or a respiratory drug, and a group of children with chronic health issues who require multiple prescriptions from a wide range of therapeutic classes.

Drug utilization by Canadian children, 1999. Most children have zero to 2 prescriptions per year, but approximately a quarter of children account for 70% of drug utilization.

( Data from Khaled LA, Ahmad F, Brogan T, et al. Prescription drug use by one million Canadian children. Paediatr Child Health 2003;8(Suppl A):6A–56A.)

The second myth is that children are at a relatively reduced risk for ADRs compared with adults. While it has been accepted for some time that the preterm neonate is a distinct exception to this, a combination of faith in the general good health of children and unique cultural issues germane to certain specialty practices has fueled this myth. In fact, children are typically at either the same risk for ADRs as adults or even more; for example, it could be argued that essentially every child treated for leukemia experiences a series of ADRs, which are a regrettably but currently necessary part of life-saving therapy.

The fact that children are at an appreciable risk for ADRs, in some cases at higher risk than adults, is not surprising when one considers the 7 generally accepted risks for ADRs. These include: (1) previous adverse reaction to another drug, (2) polypharmacy, (3) female gender, (4) extremes of age, (5) reduction in capacity of the organs of excretion (liver, kidney), (6) larger drug dose, and (7) certain genetic polymorphisms.

These risk factors often operate in synergy (eg, patients at the extremes of age —the very premature infant or the very elderly senior) often have (for different reasons) reduced liver and kidney capacity, and often (again for different reasons) are prescribed multiple drugs. As well, although the developmentally related reductions in drug clearance are well known in the case of the premature infant, the fact that the liver is the largest relative percentage of body mass at any time during life during the toddler years (with consequent implications for activation-induced adverse drug events such as valporic acid-induced hepatotoxicity, lamotrigine-induced serious skin rash, ifosfamide-induced nephrotoxicity and cefaclor-induced serum sickness-like reactions) is much less well appreciated. Children are also at special risk for adverse drug events such as 10-fold errors in medication administration. This phenonemon—almost unknown in adults—occurs in children due to a combination of drugs with child-unfriendly dosing formulations and mathematical problems on the part of some health care personnel.

Thus, there are clear and compelling reasons that suggest that children are likely to be at a substantial risk for ADRs, and in certain circumstances may be at an even higher risk than adults. Indeed, as noted previously, there are certain populations of children (eg, the premature neonate and children with cancer) for whom ADRs are essentially universal and are accepted as an undesired but currently necessary cost associated with therapy. Given the increasing variety and complexity of therapies for children, the ability of clinicians to be able to appreciate, evaluate, treat, and ultimately prevent ADRs constitutes a core clinical skill for child health care practitioners in the upcoming decades. Regrettably, despite the importance of ADRs in child health care, this is an area that is poorly taught in most pediatric residency programs.

Adverse drug reactions—implications for child health

There are several myths that enjoy widespread circulation as to the degree and range to which drug therapies, and as a consequence ADRs, are used in children, as well as to relative risks of ADRs in children versus adults. These myths have led to complacency, which has not fostered a culture of patient safety in child health care.

The first of these myths has to do with drug utilization. It is commonly believed that children rarely receive drug therapy and when they do it is primarily in the form of antibiotics. This issue has actually been studied. All prescriptions dispensed to a cohort of 1 million Canadian children, roughly 15% of all children in Canada, in 1999 were retrieved. Over this 1-year period, the 1,031,731 Canadian children (<18 years) received 4,028,502 prescriptions, roughly 4 prescriptions per child per year. Of the drugs prescribed, antibiotics were indeed the most commonly prescribed drugs. However, they accounted for less than half of all drugs prescribed. There was a wide range— indeed more than 1300 distinct pharmacologic entities—that accounted for the remainder of the drugs prescribed ( Table 1 ). As well, the overall numbers require some context to understand actual drug use among children. Despite the large number of prescriptions, most children are well and need few if any prescription drugs. As shown in Fig. 1 , the largest amount of drug utilization—in fact, more than 70—occurs among roughly 25% of children. Thus, there are 2 distinct populations of children,children who are otherwise well and need perhaps one prescription a year, often an antibiotic or a respiratory drug, and a group of children with chronic health issues who require multiple prescriptions from a wide range of therapeutic classes.

Drug utilization by Canadian children, 1999. Most children have zero to 2 prescriptions per year, but approximately a quarter of children account for 70% of drug utilization.

( Data from Khaled LA, Ahmad F, Brogan T, et al. Prescription drug use by one million Canadian children. Paediatr Child Health 2003;8(Suppl A):6A–56A.)

The second myth is that children are at a relatively reduced risk for ADRs compared with adults. While it has been accepted for some time that the preterm neonate is a distinct exception to this, a combination of faith in the general good health of children and unique cultural issues germane to certain specialty practices has fueled this myth. In fact, children are typically at either the same risk for ADRs as adults or even more; for example, it could be argued that essentially every child treated for leukemia experiences a series of ADRs, which are a regrettably but currently necessary part of life-saving therapy.

The fact that children are at an appreciable risk for ADRs, in some cases at higher risk than adults, is not surprising when one considers the 7 generally accepted risks for ADRs. These include: (1) previous adverse reaction to another drug, (2) polypharmacy, (3) female gender, (4) extremes of age, (5) reduction in capacity of the organs of excretion (liver, kidney), (6) larger drug dose, and (7) certain genetic polymorphisms.

These risk factors often operate in synergy (eg, patients at the extremes of age —the very premature infant or the very elderly senior) often have (for different reasons) reduced liver and kidney capacity, and often (again for different reasons) are prescribed multiple drugs. As well, although the developmentally related reductions in drug clearance are well known in the case of the premature infant, the fact that the liver is the largest relative percentage of body mass at any time during life during the toddler years (with consequent implications for activation-induced adverse drug events such as valporic acid-induced hepatotoxicity, lamotrigine-induced serious skin rash, ifosfamide-induced nephrotoxicity and cefaclor-induced serum sickness-like reactions) is much less well appreciated. Children are also at special risk for adverse drug events such as 10-fold errors in medication administration. This phenonemon—almost unknown in adults—occurs in children due to a combination of drugs with child-unfriendly dosing formulations and mathematical problems on the part of some health care personnel.

Thus, there are clear and compelling reasons that suggest that children are likely to be at a substantial risk for ADRs, and in certain circumstances may be at an even higher risk than adults. Indeed, as noted previously, there are certain populations of children (eg, the premature neonate and children with cancer) for whom ADRs are essentially universal and are accepted as an undesired but currently necessary cost associated with therapy. Given the increasing variety and complexity of therapies for children, the ability of clinicians to be able to appreciate, evaluate, treat, and ultimately prevent ADRs constitutes a core clinical skill for child health care practitioners in the upcoming decades. Regrettably, despite the importance of ADRs in child health care, this is an area that is poorly taught in most pediatric residency programs.

The contemporary approach to ADRs

The current dealing with ADRs is primarily clinical, and can best be remembered as 5 As: appreciation, assessment, analysis, assistance, and aftermath.

This approach starts with clinicians appreciating the possible risk of an ADR. Succinctly put, the diagnosis of an ADR necessitates the consideration that it is a possibility. While this seems intuitive, assessment of ADR risk and the communication of this risk to patients and families are often less than optimal, even in settings in which ADR risk is clear and virtually certain. Some pediatric specialities (eg, pediatric oncologists) have developed widely adopted standard operating procedures prior to starting therapy that go to considerable lengths to ensure that patients and families understand the risks associated with chemotherapy. However, such an approach has not been as well developed for many other child health disciplines. This is compounded by the complexity of therapies administered to chronically ill children as noted previously. Children who require drug therapy often require complex combinations. Given that there are over 3500 distinct therapeutic agents on the American market, it is a real challenge for health care providers to be aware of what are common and important ADRs, notably for new drugs or for drugs with which they are unfamiliar. Another layer of challenge is caused by the fact that the care of sick children is often multidisciplinary, requiring clear communication between members of the team, most notably in the area of therapeutics. In many hospitals, clinical pharmacists have indeed become essential to optimize the ability of providing safe and effective therapy.

When an adverse event occurs in the context of therapy, a distinction needs to be made as to whether the event is disease- or medication-related. While somewhat of a generalization, it is common for patients and families to associate the adverse event with therapy, while health care providers, especially those who prescribe therapy, quite commonly tend to associate the adverse event with the disorder being treated. This dichotomy serves as a useful starting point for the key diagnostic question: is the adverse event described drug-related or not.

This brings one to assessment, the second step after appreciation. To make an adequate assessment it is crucial to have a clear and complete understanding of the events that led to the suspicion of an ADR. This includes both a comprehensive history as well as a relevant physical examination. Key elements in history include the rationale for therapy, the timing and dose of the drug(s) in question, the timing and nature of the symptoms and signs of concern, any interventions used with respect to the symptoms and signs of concern, and any potential confounders. As an example, it would be distinctly inappropriate to attribute a rash to a drug if the rash had been present before the drug was given. While somewhat an extreme example, much information can be obtained from a careful temporal analysis of the course of the disorder being treated, the course of therapy, and the course of the symptoms or signs of interest. Similarly, the physical examination may be extremely helpful. As an illustration, while an urticarial rash may be strongly suggestive of drug allergy, a nonpruritic, nonurticarial erythematous rash—of the type often called maculopapular—may be more suggestive of nondrug etiologies. While a thorough history and physical examination can be of great help in assessing ADRs, laboratory and other tests are usually less helpful. Laboratory tests can indeed identify worrisome findings, but often cannot tell whether the abnormality is drug-induced or not. For example, the presence of elevated serum transaminases can signal drug-induced hepatoxicity, but the results cannot differentiate drug-induced causes from infectious (viral hepatitis) ones. The main role of laboratory tests at the present time is to identify potential nondrug-induced causes (eg, viral hepatitis) and thus exclude drug-related causes, rather than to confirm the role of drugs in the sign or symptom of interest. Testing can also confirm a clinical impression (eg, the demonstration of leukoclastic vasculitis can support the diagnosis of drug eruption). However, as will be noted, this may well change over the next decade.

There are several drugs for which laboratory testing in fact may be helpful. These include ADRs resulting from the use of penicillins as well as local anesthetics. Pencillin allergy is perhaps the most commonly stated drug allergy. While penicillin can cause ADRs mediated by a variety of mechanisms, including Gell and Coombs class 1 to 4 hypersensitivity, type 1 hypersensitivity (ie, immunoglobulin E [IgE]-mediated allergy) is the most common and most serious ADR associated with penicillin. This is 1 case of an ADR where the pathophysiological mechanism is known in reasonably complete detail and in fact has been used to develop testing to predict ADR risk. Assessment can be done in vivo using penicillin skin testing, which is predictive of the risk for penicillin allergy and can provide a reliable risk for the likelihood of a serious allergic reaction with future penicillin therapy. As well, in vitro assessment of risk can be done by determining if the patient has pencillin-specific IgE using a radioallergosorbent test (RAST). The use of in vitro assessments for penicillin allergy is less specific than skin testing but is reasonably sensitive, notably when combined with a rechallenge following a negative in vitro test. Positive results are highly significant (ie, they are associated with a significant risk for a serious ADR with future therapy), but negative in vitro results are not a guarantee of safety; thus rechallenge is important if future therapy is planned on the basis of a negative in vitro test. Assessment for penicillin allergy is of importance given both the frequency with which penicillin can induce allergy and the regrettable fact that, although penicillin allergy is common, it is not nearly as common as it is reported. When patients are asked as to which drugs they are allergic, to up to 10% will mention a penicillin allergy; however, it has been demonstrated that most patients who present with a history of penicillin allergy are in fact not allergic to penicillin. Interestingly, many parents whose children have a history of a possible ADR to penicillin remain reluctant to treat their children with penicillin despite negative testing.

In addition to assessment of the risk of penicillin allergy, in vivo testing can be useful in the evaluation of allergy to local anesthetics. In this case, testing is not to confirm or refute the association of a local anesthetic with an adverse event but rather to determine safety for future therapy. As most ADRs to local anesthetics are class-specific, typically the approach is to undertake skin testing with an agent of another class (eg, if the patient is believed to have an allergic reaction to an amide local anesthetic, then it would be appropriate to conduct skin testing using an ester local anesthetic). Patch testing has been used for the assessment of a number of compounds, but patch testing requires special expertise and is not widely available. Finally, in vitro testing can be very useful for the evaluation of possible pseudocholinesterase deficiency in patients who have had prolonged duration of effect from paralyzing agents such as succinylcholine.

Analysis follows appreciation and assessment. Analysis essentially is putting together a differential diagnosis guided by the information gathered during assessment. It is often difficult for clinicians to reach a conclusion when assessing the potential role of a drug in an adverse event, and thus several algorithms have been developed to help guide the process of determining causation. However, consistency between these algorithms is often poor; moreover, many algorithms do not take key confounders into account. The algorithm that is used most commonly is the Naranjo score, a score derived from an algorithm developed by Naranjo and colleagues for the assessment of possible ADRs in adult patients treated with neurotropic drugs. Decisions as to whether an adverse event is an unlikely, possible, probable, or definite ADR is made by adding up the score assigned to a number of variables, including timing of drug exposure, likelihood of other causes, and whether the response changed in response to changes in drug dose or use of a placebo ( Table 2 ). This algorithm was extremely useful when it was developed, but changes in medical practice and differences in approach between adults and children have made this algorithm less useful in pediatrics.

The next step is assistance, which for most ADRs is relief of symptoms and to decide whether to continue therapy. In the case of serious ADRs—notably those mediated, even in part, by the immune system—it is usually necessary to stop the offending drug. If the ADR is acute or life-threatening, such as anaphylaxis, then additional timely and acute intervention is often needed to prevent serious injury or death. As an example, there is emerging evidence that immune modulation may alter the course of serious drug hypersensitvities such as Stevens Johnson syndrome or toxic epidermal necrolysis. However, for most ADRs, the best therapy is symptom management. Part of assistance is also the evaluation as to whether the drug used needs to be replaced, if the disorder that led to the drug prescription persists, or whether the drug can be permanently discontinued as the disorder has resolved. Occasionally, treating through ADRs is indicated. This approach has merit when the symptoms are troublesome but expected to diminish in severity or even disappear over time. An example is the fine hand tremor associated with aerosol albuterol (salbutamol) therapy. This ADR relates to stimulation of the beta receptors on skeletal muscle. It is common when therapy for reactive airways disease or asthma is started but typically goes away with continued therapy.

The final step in managing an ADR is aftermath. There are several consequences of having had an ADR, some predictable and some not. While patients will often return to their usual state of health after resolution of an ADR, some ADRs have significant long-term consequences in terms of organ impairment. As an example, some patients with ifosfamide-induced nephrotoxicity will have long-term kidney dysfunction, which in rare cases can even progress to the point of requiring a renal transplant. Similar considerations apply to cisplatin-induced ototoxicity and anthracycline-induced cardiotoxicity. Thus, additional interventions may be needed.

Communication is an important part of the aftermath of an ADR that is often underappreciated. First, it is crucial that the patient and family understand what has happened and what the implications are. It is also important to ensure clear communication with other health care practitioners who are involved in managing the patient and his or her family. Finally, and especially for new drugs, it is important that the respective drug regulatory authorities (eg, the FDA) are made aware of ADRs, most notably unexpected ones. This is obviously less important for older drugs (eg, the FDA is very aware that amoxicillin therapy can result in penicillin allergy) than it is for new drugs. The cold hard reality is that drugs enter the market being tested in relatively few people, and thus serious ADRs that are occurring in a frequency of 1 case per 1000 population or less are very unikely to be detected prior to drug approval. It is worthwhile to remember that every serious drug hypersensitivity described has occurred after the drug was marketed, often as a result of a report by an astute clinician.

This, then, is the current approach to adverse events associated with drug therapy in children. This approach has been less than satisfactory in informing safe drug therapy, especially in the area of ADR prevention, and new approaches are clearly needed. What is exciting is that there are several developments underway that offer great promise in enhancing safe and effective drug therapy in children.