- Email: [email protected]
Side effects of phenobarbital in toddlers; behavioral and cognitive aspects
September 1079 The Journal o f P E D I A T R I C S
361
Side effects of phenobarbital in toddlers," behavioral and cognitive aspects Cognitive and behavioral effects of phenobarbital in toddlers were assessed in a randomized, placebo-controlled study of patients who had had a febrile seizure. There were no sign(flcant differences in IQ (Binet or Baylev Scales) between placebo and phenobarbital groups after eight to 12 months of therapy. However. detrimental effects of phenobarbital were found in memory, .for which serum level &fluenced scores, and in comprehension, in that length of treatment time affected performance. Hyperactivity was not seen. Behavioral changes, reported by parents, were increased filssiness and a characteristic disturbance of sleep. These changes varied in severi(y and were classified as transient, dose related, or unacceptable. After 12 months in the study, most parents couM not distinguish between phenobarbital and placebo. Our data s~)ggest that although most toddlers do not have major side effects :from phenobarbital therapy when treated for a year. serum levels and length of time on phenobarbital shouM be kept at a minimum to reduce negative cognitive and behavioral e~ects.
Carol S. Camfield, M.D.,* Sheila Chaplin, M.A., Anna-Beth Doyle, Ph.D., Stanley H. Shapiro, Ph.D., Carl Cummings, M.D., and Peter R. Camfieid, M.D., H a l i f a x , N. S., a n d Montreal, P. Q., C a n a d a
SINCE its introduction in 1912, phenobarbital has become the most frequently used anticonvulsant in childhood. Some authorities have recommended that it be used daily as prophylaxis for children with febrile seizures. 1-~ Nonetheless, remarkably few attempts have been made to delineate the cognitive and behavioral effects of this drug in childhood. Unfavorable behavioral changes have been estimated to occur in 20% to 75% of children receiving phenobarbital, hyperactivity being the most common symptom? -; The cognitive effects in adult volunteers given a short course of phenobarbital were striking in From the Montreal Children's Hospital, Department of Pediatrics and Department of Psychology, McGill University; L W. Killam Hospital for Children, Department of Pediatrics, Dalhousie University; Department of Psychology, Concordia University," and Depqrtment of Epidemiology and Health, MeGill University. Presented in part, Canadian Congress of Neurological Sciences, Vancouver, June, 1978. *Reprint address: Izaak Walton Killam Hospitalfor Children, P.O. Box 3070, Halifax, Nova Scotia, B3J 3G9.
0022-3476/79/090361 +05500.50/0 9 1979 The C. V. Mosby Co.
areas of vigilance, reaction time, and verbal learning, despite serum levels in the generally accepted therapeutic range, s Therefore, significant behavioral and cognitive effects of phenobarbital may occur in children.
See related article, p. 403. A double-blind placebo-controlled randomized study of the efficacy of phenobarbital to prevent recurrent "benign" febrile convulsions in otherwise healthy normal toddlers has given us the opportunity to assess both the frequency and nature ofphen0barbital-induced behavioral changes as reported by parents, and the nature of phenobarbital-induced cognitive effects as measured by standard psychologic testing.
METHODS Sample selection. The complete, detailed sample selection will be reported in a further paper describing the effectiveness of phenobarbital as compared with antipyretic therapy in the treatment of febrile seizures. Briefly,
Vol. 95. No. 3, pp. 36t-365
362
Camfield et al.
Table 1. Clinical features of study patients Age at first febrile seizure 6 mo-3 yr Temperature in emergency room >38~ rectally Generalized seizure with no focal components Seizure duration <20 min Normal lumbar puncture Normal developmental and neurologic examination 2 wk after seizure No epileptic abnormality on electroencephalogram 2 wk after seizure No neonatal intensive care No major social problems Fluent English or French
90% of all children presenting to the emergency room of the Montreal Children's Hospital with an initial febrile seizure during an 18-month period beginning in September, 1975, were eligible for this study. The other 10% failed to return for follow-up. Fifty percent of those available fulfilled the clinical criteria listed in Table I, and were defined as having had a simple or "benign" febrile seizure. Three-quarters of these latter families agreed to participate in this study. The children were stratified for age, sex, and family history of febrile seizures or epilepsy, and were randomized in double-blind fashion to receive placebo or phenobarbital. All patients were given detailed instructions about_Jantipyretic therapy. Treatment plan. Placebo and phenobarbital tablets were identical in appearance (kindly supplied by Smith, Kline and Freneh Canada Ltd.) and contained 4 mg of riboflavin. Dosage, was based on 4 to 5 m g / k g / d a y of phenobarbital, and the entire amount was given as a single bedtime dose. Follow-up visits were at 1, 4, 8, and 12 months. At each visit, urine was checked for fluorescence due to riboflavin9 to determine if the child had received the medication within the past 18 hours. Parents were given the results in an attempt to maximize compliance. In addition, on each visit blood was drawn from every patient, and phenobarbital levels were determined by a standard gas-liquid chromatography method in the laboratory of Dr. A. Sherwin at the Montreal Neurologic Institute. Drug levels were screened for toxicity but the results were withheld from parents and investigators until the study was complete. At each follow-up visit, parents were systematically asked about side effects, especially sleep disturbance, behavioral change, and hyperactivity, the latter being defined as excessive motor activity and decreased attention span. If side effects intolerable to parents were reported, the dose was lowered to 2 to 3 mg/kg for two to three weeks, and then increased to 4 to 5 mg/kg if the side
The Journal of Pediatrics September 1979
effects had disappeared. If side effects did not reappear, this type of behavioral change was called transient. If side effects recurred at 4 to 5 mg/kg, the dose was lowered and maintained at 2 to 3 mg/kg. This type of behavioral change was defined as dose related. If these changes persisted at the lower dose of 2 to 3 mg/kg, medication was stopped for four to seven days and then restarted. Any child whose behavioral effects then recurred and were judged to be intolerable by parents or physicians was removed from the study. These behavioral changes were defined as unacceptable. Children with a recurrent seizure were removed from this study at the time of the recurrence. Parents of children who had completed a year on medication were asked to guess which tablet their child had been receiving. Psychologic testing. The children who had completed either eight or 12 months on medication had a doubleblind psychologic assessment during the regularly scheduled follow-up visit. Each of these children was assessed only once, at either the 8 or the 12-month visit, due to limited time available. Because of the study design outlined above, none of these children had experienced a second febrile seizure prior to the time of the psychologic testing. Either the Mental Scale of the Bayley Scales of Infant Development, which yields a Mental Development Index equivalent to an IQ score, or the Stanford Binet Intelligence Scale was used to yield an IQ score. The Binet was given if the child was more than two years of age and obtained a basal score on the test; otherwise, the Bayley test was used. The scores of those receiving the Stanford Binet Intelligence Scale were subdivided according to Valett 1~ into five classes of abilities: (1) general comprehension, (2) visual motor ability, (3) memory concentration, (4) vocabulary, and (5) judgment. These five scores represented the child's strengths and weaknesses in particular areas of intellectual ability relative to his overall IQ. Scores for each of the five subcategories of ability were derived by computi~g the number of subtests in a category passed above the subject's basal age. Basal age was defined as the highest level at which all subtests were passed. This method was employed with the rationale that the subject's strengths and weaknesses within his own range of ability, regardless of IQ, was o f interest. In addition to tests of group differences, serum levels of phenobarbital at the time of psychologic testing were correlated with the results of these five classes of abilities and with the IQ. The examiner also rated the child's behavior during the testing, using the Bayley Infant Behavior Record. A co-rater was present for 45% of the assessments; co-rater agreement on the Bayley Infant Behavior Record was 77%.
Volume 95 Number 3 RESULTS The demographic characteristics and clinical course of these patients is outlined in Table II. Behavioral changes. The frequency of reported behavioral changes is shown in Table III. The placebo and phenobarbital groups did not differ in frequency o f transient effects or of absence of side effects. However, the dose-related and unacceptable.side effects predominated in the phenobarbital group (P = 0.05, Fisher exact test). Among patients receiving placebo, transient side effects (lasting less than three weeks and not requiring dosage change) occurred in seven of 30. These were primarily increased fussiness and, to a lesser degree, sleep disturbance including awakening early, refusing to go to sleep, or slight increase in sleepiness; two of seven had transient sleep problems similar to those described below for the phenobarbital group. After four months on medication, a single patient on placebo developed a persistent skin rash and ataxia and was withdrawn from the study. No side effects were reported in 22 children receiving placebo. Among the 35 patients receiving phenobarbital, the side effects were more stereotyped but still fell into a milder transient group (8); a dose-related group (4), and an unacceptable group (3). These effects were primarily sleep disturbance and daytime fussiness. The major sleep disturbance consistently described by parents was one in which the child would awaken for several hours in the early morning (12 to 4 AM). During this time the child seemed to be wide awake and wished to play, but usually fell asleep again. This type of unusual sleep disturbance was often coupled with daytime fussiness and irritability. These children seemed dissatisfied with toys, friends, and food; nothing seemed to please them. They were argumentative and very stubborn. The characteristic sleep disturbance was reported in five of eight children on phenobarbital with transient changes, and in four of five of these the sleep disturbance was associated with daytime fussiness and irritability. Only one of eight children had these daytime symptoms transiently without sleep disturbance. The mean serum phenobarbital level in patients with transient changes was 1.5 mg/dl, which was similar to the mean of 1.4 mg/dl in patients taking phenobarbital who did not have reported side effects. Two of 35 children receiving phenobarbital were reported by their parents to have transient increased somnolence and decreased daytime activity; their drug levels at the time of their symptoms were 1.5 and 3.2 mg/dl, respectively. The four children receiving phenobarbital who had dose-related effects had both a sleep disturbance and fussiness; however, they were asymptomatic when the
Side effects of phenobarbital
3 63
Table II. Sample description and clinical course
Placebo Total number of patients Average age on entering study Sex Completed 12 mo on medication Recurrent seizure Dl~op out due to unacceptable side effects Drop out after 6-8 mo due to nonmedical reasons Patients given Bayley Scale or" Infant Development Patients given Stanford Binet Intelligence Scale
I
Phenobarbital
30 15 months
35 17 months
15 males 15 females 17
17 males 18 females 28
10 1
2 3
2
2
6
6
12
18
Table IlL Frequency of behavioral changes
Unacceptable Dose related Transient None
Placebo (N = 3O)
Phenobarbital (N = 35)
1 0 7 22
3 4 8 20
dose was lowered and kept at 2 to 3 mg/kg/day. The drug levels from these children at 4 to 5 m g / k g / d a y of phenobarbital averaged 1.9 mg/dl (range 1.5 to 2.2); whereas, at 2 to 3 m g / k g / d a y the drug levels averaged 1.1 mg/dl (range 0.7 to 1.4). Each of the three children with unacceptable side effects had a sleep disturbance and fussiness. Unfortunately, no serum phenobarbital levels were obtained from these patients. No child became hyperactive on either placebo or phenobarbital therapy, as assessed by parents and physicians, during the entire 12-month period; nor were there any significant differences between phenobarbital and placebo groups on three summary scores of emotional state, activity, and attention, as rated on the BIBR. All of the parents of children completing a year on placebo guessed correctly that their child was receiving placebo. Surprisingly, only six of 28 on phenobarbital for the entire year guessed the drug correctly; in all of these, side effects had been reported. Four of these six patients had dose-related effects, and the remaining two had reported transient somnolence. Cognitive functions. The groups did not differ in IQ
364
Camfield et al.
score, with means of 104 for placebo and 109 for phenobarbital. Multivariate analysis of the five Binet subcategories similarly yielded no significant differences between the phenobarbital and placebo groups. However, when the phenobarbital serum level at. the time of psychologic assessment was correlated with each of the five Binet subscores, there was a significant negative correlation between serum level and memory concentration subscores (r[14] = -0.52, P < 0:05). That is, when memory scores were examined relative to basic level of cognitive functioning (basal age), children with higher phenobarbital levels performed more poorly on memory concentration tasks. Consistent with this effect was a tendency, though nonsignificant, for the phenobarbital group to score lower than the placebo group on memory concentration items (P = 0.07). Although there were no significant overall differences between the two groups on the five subcategories, there were differences due to time of testing (8 vs 12 months). For children receiving phenobarbital, those tested after 12 months (n = 9) had significantly lower general comprehension scores than those tested after eight months (n = 7), P < 0.05. Patients receiving placebo (n = 6 vs 5) failed to show a similar change.* In other words, it appeared that longer exposure to phenobarbital might have reduced general comprehension scores. DISCUSSION The behavioral effects of phenobarbital in toddlers appear to fall into three categories: unacceptable, dose related, and transient. The unacceptable changes, which occurred in 9% of the children, presumably represented an idiosyncratic response to the drug and included sleep disturbance and daytime irritability. Parents could accurately identify the 11% of children with dose-related symptoms. When the dose was reduced from 4 to 5 m g / k g / d a y to 2 to 3 mg/kg/day, the serum phenobarbital levels were still generally within the accepted therapeutic range. Inconsequential behavioral changes occurring in 17% of the children were found with equal frequency among children receiving placebo and those given phenobarbital, and presumably represent the child's normal variation in mood and response to parental concern about his seizure and medication. These children were clinically recognized by the disappearance of their symptoms within a few weeks, and their parents could not distinguish phenobarbital from placebo after one year of medication. We are unaware of previous detailed descriptions of the *Samplesizesare smallerthan thosetestedbecauseof noncompliancein taking drug (n = 2) and untestablebehavior(n = 1).
The Journal of Pediatrics September 1979
effect of phenobarbital on children's sleep. Our study patients awoke in the middle of the night, seemingly fully rested and ready for normal play. Only after several hours would they go back to sleep. It was difficult to quantify this effect objectively, but the consistency with which it was reported by parents in all three categories strongly suggests that it is a specific effect on phenobarbital. Given that physicians' and parents' concerns influence their observations, the frequency of a drug's behavioral effects are best assessed by double-blind placebocontroled studies. A recent nonblind study found that 39 of 109 children with a febrile seizure treated with phenobarbital became hyperactive, as compared with 13 of 120 children in a nontreatment group. ~ The daytime fussiness and irritability of the 15 children on phenobarbital in our study who had side effects did not resemble true hyperactivity. Although the parents reported that the children were irritable, there was no increased motor activitY as assessed by parents, psychologist, or a physician. It is possible that the often mentioned phenobarbital-induced hyperactivity syndrome is seen only in children with brain damage or some other form of convulsive disorder, rather than in children with simple febrile seizures. The Bayley Scale of Infant Development and the Stanford-Binet Intelligence Test have been shown to correlate well with developmental status and moderately well with future intelligence, and are good measures of overall performance. 13 The failure to find anydifference in the mean IQ scores between phenobarbital and placebo-treated children is reassuring, and indicates that major effects on cognitive development are unlikely. However, more subtle effects suggesting need for further study were seen. Phenobarbital seemed to have an adverse effect on memory concentration tasks; scores correlated negatively with the serum drug level, despite the fact that there were no serum levels in the toxic range2 These memory effectsdo not reflect an overall deficit in cognitive functioning but only one relatj.ve to the child's basal age. However, it is possible that over a treatment period longer than eight to 12 months, the mean IQ scores in phenobarbitaltreated patients would be lower than for placebo treated. This speculation appears to be partially substantiated by the fact that general comprehension scores were significantly lower in children receiving phenobarbital for 12 months as compared with those treated for eight months. Clearly, a longitudinal assessment with finer measures of memory and concentration should be undertaken. This study does not examine the question of whether daily phenobarbital therapy or recurrent febrile seizures are more dangerous to a child. No psychologic testing was performed on children with a second febrile seizure. The literature strongly suggests that recurrent brief febrile
Vohmw 95 Number 3
seizures have no significant long-term cognitive effects, and this information should decrease the frequency o f phenobarbital use in febrile seizures. 11 However, because it will remain a useful anticonvulsant for selected patients with febrile seizures and for other children with convulsive disorders, examination of potential side effects is imPortant. Although majo r behavioral and cognitive effects o f daily phenobarbital therapy were not seen in our study, we conclude that caution should be used in exposing children to phenobarbital over many years. Careful attention should be paid to parents' observations, and the dosage o f phenobarbital should be lowered if there are side effects persisting beyond a few weeks. Serum drug levels are not good predictors of which children will develop side effects. Other than toxicity. Since memoryconcentration tasks appear to be most affected at higher serum levels; drug dosages should be reduced to provide the m i n i m u m effective level, probably about 1.5 mg/dl. ~2 Since m a n y patients are clearly protected from febrile seizures at lower serum levels, slavish adherence to the 1.5 m g / d l should be avoided if side effects are seen. Likewise. the duration of treatment with phenobarbital should be kept as short as possible, because longer exposure may adversely affect general comprehension. W e hope that more detailed psychologic tests will be developed for this age group to refine our understanding of anticonvulsant drug effects on behavioral and cognitive development. Only with such data will it be POSsible to choose appropriately from a growing list of effective anticonvulsant drugs for febrile or other seizure disorders.
Side effects o f phenobarbital
365
REFERENCES
1. Wolf SM, Carr A, Davis DC, et al: The value of phenobarbital in the child who has had a single febrile seizure: A controlled prospective study, Pediatrics 59:378, 1977. 2. Lennox-Buchthal M: Febrile convulsions, a reappraisal, Electroericephalogr Clin Neurophysiol 32(Suppl): 1973. 3. Pollack MA: Continuous phenobarbital treatment after a 'simple febrile convulsion.' Am J Dis Child 132:87, 1978. 4. Chao D~ Carter S, Gold AP: Paroxysmal disorders, in Rudolf AM, Barnett HL, and Einhorn AH, editors: Pediatrics. ed 16. New York, 1977. Appleton-Century Crofts Inc, p 1854. 5. Wolf SM. Forsythe A: Behaviour disturbance, phenobarbital. and febrile seizures. Pediatrics 61:728, 1978. 6. Thorn I: A controlled study of prophylactic long term treatment of febrile convulsions with phenobarbital, Acta Neurol Scand Suppl 60:67. 1975. 7. Heckmatt J, Houston A, Dodds K, et al: Failure of phenobarbitone to prevent febrile convulsions, Br Med J 1:559. 1976. 8. Hun SJ. Jackson. PM. Belsham A, et al: Perceptual-motor behaviour in relation to blood phenobarbital level, Dev Med Clin Neurol 10:626. 1968. 9. Jusko WJ. Levy G. and Yaffe SJ: Effect of age on intestinal absorption of r in humans, J Pharmaceutical Sci 59:487. 1970. 10. Valetl JR: A clinical profile for the Stanford Binet, J Sch Psychology 2:49. 1964. 11 Ellenberg JH. and Nelson KB: Febrile seizures and later intellectual performance, Arch Neurol 35:17, 1978. 12. Faerr O, Kastrup KW, Lykkegaard NE, et al: Successful prophylaxis of febrile convulsions with phenobarbital, Epilepsia 13:279, 1972. 13, McCall RB, Hogarty P, and Hurlburt N: Transitions in infant sensorimot0r development and the prediction of childhood IQ, Am Psychologist 27:728, 1972.
361
Side effects of phenobarbital in toddlers," behavioral and cognitive aspects Cognitive and behavioral effects of phenobarbital in toddlers were assessed in a randomized, placebo-controlled study of patients who had had a febrile seizure. There were no sign(flcant differences in IQ (Binet or Baylev Scales) between placebo and phenobarbital groups after eight to 12 months of therapy. However. detrimental effects of phenobarbital were found in memory, .for which serum level &fluenced scores, and in comprehension, in that length of treatment time affected performance. Hyperactivity was not seen. Behavioral changes, reported by parents, were increased filssiness and a characteristic disturbance of sleep. These changes varied in severi(y and were classified as transient, dose related, or unacceptable. After 12 months in the study, most parents couM not distinguish between phenobarbital and placebo. Our data s~)ggest that although most toddlers do not have major side effects :from phenobarbital therapy when treated for a year. serum levels and length of time on phenobarbital shouM be kept at a minimum to reduce negative cognitive and behavioral e~ects.
Carol S. Camfield, M.D.,* Sheila Chaplin, M.A., Anna-Beth Doyle, Ph.D., Stanley H. Shapiro, Ph.D., Carl Cummings, M.D., and Peter R. Camfieid, M.D., H a l i f a x , N. S., a n d Montreal, P. Q., C a n a d a
SINCE its introduction in 1912, phenobarbital has become the most frequently used anticonvulsant in childhood. Some authorities have recommended that it be used daily as prophylaxis for children with febrile seizures. 1-~ Nonetheless, remarkably few attempts have been made to delineate the cognitive and behavioral effects of this drug in childhood. Unfavorable behavioral changes have been estimated to occur in 20% to 75% of children receiving phenobarbital, hyperactivity being the most common symptom? -; The cognitive effects in adult volunteers given a short course of phenobarbital were striking in From the Montreal Children's Hospital, Department of Pediatrics and Department of Psychology, McGill University; L W. Killam Hospital for Children, Department of Pediatrics, Dalhousie University; Department of Psychology, Concordia University," and Depqrtment of Epidemiology and Health, MeGill University. Presented in part, Canadian Congress of Neurological Sciences, Vancouver, June, 1978. *Reprint address: Izaak Walton Killam Hospitalfor Children, P.O. Box 3070, Halifax, Nova Scotia, B3J 3G9.
0022-3476/79/090361 +05500.50/0 9 1979 The C. V. Mosby Co.
areas of vigilance, reaction time, and verbal learning, despite serum levels in the generally accepted therapeutic range, s Therefore, significant behavioral and cognitive effects of phenobarbital may occur in children.
See related article, p. 403. A double-blind placebo-controlled randomized study of the efficacy of phenobarbital to prevent recurrent "benign" febrile convulsions in otherwise healthy normal toddlers has given us the opportunity to assess both the frequency and nature ofphen0barbital-induced behavioral changes as reported by parents, and the nature of phenobarbital-induced cognitive effects as measured by standard psychologic testing.
METHODS Sample selection. The complete, detailed sample selection will be reported in a further paper describing the effectiveness of phenobarbital as compared with antipyretic therapy in the treatment of febrile seizures. Briefly,
Vol. 95. No. 3, pp. 36t-365
362
Camfield et al.
Table 1. Clinical features of study patients Age at first febrile seizure 6 mo-3 yr Temperature in emergency room >38~ rectally Generalized seizure with no focal components Seizure duration <20 min Normal lumbar puncture Normal developmental and neurologic examination 2 wk after seizure No epileptic abnormality on electroencephalogram 2 wk after seizure No neonatal intensive care No major social problems Fluent English or French
90% of all children presenting to the emergency room of the Montreal Children's Hospital with an initial febrile seizure during an 18-month period beginning in September, 1975, were eligible for this study. The other 10% failed to return for follow-up. Fifty percent of those available fulfilled the clinical criteria listed in Table I, and were defined as having had a simple or "benign" febrile seizure. Three-quarters of these latter families agreed to participate in this study. The children were stratified for age, sex, and family history of febrile seizures or epilepsy, and were randomized in double-blind fashion to receive placebo or phenobarbital. All patients were given detailed instructions about_Jantipyretic therapy. Treatment plan. Placebo and phenobarbital tablets were identical in appearance (kindly supplied by Smith, Kline and Freneh Canada Ltd.) and contained 4 mg of riboflavin. Dosage, was based on 4 to 5 m g / k g / d a y of phenobarbital, and the entire amount was given as a single bedtime dose. Follow-up visits were at 1, 4, 8, and 12 months. At each visit, urine was checked for fluorescence due to riboflavin9 to determine if the child had received the medication within the past 18 hours. Parents were given the results in an attempt to maximize compliance. In addition, on each visit blood was drawn from every patient, and phenobarbital levels were determined by a standard gas-liquid chromatography method in the laboratory of Dr. A. Sherwin at the Montreal Neurologic Institute. Drug levels were screened for toxicity but the results were withheld from parents and investigators until the study was complete. At each follow-up visit, parents were systematically asked about side effects, especially sleep disturbance, behavioral change, and hyperactivity, the latter being defined as excessive motor activity and decreased attention span. If side effects intolerable to parents were reported, the dose was lowered to 2 to 3 mg/kg for two to three weeks, and then increased to 4 to 5 mg/kg if the side
The Journal of Pediatrics September 1979
effects had disappeared. If side effects did not reappear, this type of behavioral change was called transient. If side effects recurred at 4 to 5 mg/kg, the dose was lowered and maintained at 2 to 3 mg/kg. This type of behavioral change was defined as dose related. If these changes persisted at the lower dose of 2 to 3 mg/kg, medication was stopped for four to seven days and then restarted. Any child whose behavioral effects then recurred and were judged to be intolerable by parents or physicians was removed from the study. These behavioral changes were defined as unacceptable. Children with a recurrent seizure were removed from this study at the time of the recurrence. Parents of children who had completed a year on medication were asked to guess which tablet their child had been receiving. Psychologic testing. The children who had completed either eight or 12 months on medication had a doubleblind psychologic assessment during the regularly scheduled follow-up visit. Each of these children was assessed only once, at either the 8 or the 12-month visit, due to limited time available. Because of the study design outlined above, none of these children had experienced a second febrile seizure prior to the time of the psychologic testing. Either the Mental Scale of the Bayley Scales of Infant Development, which yields a Mental Development Index equivalent to an IQ score, or the Stanford Binet Intelligence Scale was used to yield an IQ score. The Binet was given if the child was more than two years of age and obtained a basal score on the test; otherwise, the Bayley test was used. The scores of those receiving the Stanford Binet Intelligence Scale were subdivided according to Valett 1~ into five classes of abilities: (1) general comprehension, (2) visual motor ability, (3) memory concentration, (4) vocabulary, and (5) judgment. These five scores represented the child's strengths and weaknesses in particular areas of intellectual ability relative to his overall IQ. Scores for each of the five subcategories of ability were derived by computi~g the number of subtests in a category passed above the subject's basal age. Basal age was defined as the highest level at which all subtests were passed. This method was employed with the rationale that the subject's strengths and weaknesses within his own range of ability, regardless of IQ, was o f interest. In addition to tests of group differences, serum levels of phenobarbital at the time of psychologic testing were correlated with the results of these five classes of abilities and with the IQ. The examiner also rated the child's behavior during the testing, using the Bayley Infant Behavior Record. A co-rater was present for 45% of the assessments; co-rater agreement on the Bayley Infant Behavior Record was 77%.
Volume 95 Number 3 RESULTS The demographic characteristics and clinical course of these patients is outlined in Table II. Behavioral changes. The frequency of reported behavioral changes is shown in Table III. The placebo and phenobarbital groups did not differ in frequency o f transient effects or of absence of side effects. However, the dose-related and unacceptable.side effects predominated in the phenobarbital group (P = 0.05, Fisher exact test). Among patients receiving placebo, transient side effects (lasting less than three weeks and not requiring dosage change) occurred in seven of 30. These were primarily increased fussiness and, to a lesser degree, sleep disturbance including awakening early, refusing to go to sleep, or slight increase in sleepiness; two of seven had transient sleep problems similar to those described below for the phenobarbital group. After four months on medication, a single patient on placebo developed a persistent skin rash and ataxia and was withdrawn from the study. No side effects were reported in 22 children receiving placebo. Among the 35 patients receiving phenobarbital, the side effects were more stereotyped but still fell into a milder transient group (8); a dose-related group (4), and an unacceptable group (3). These effects were primarily sleep disturbance and daytime fussiness. The major sleep disturbance consistently described by parents was one in which the child would awaken for several hours in the early morning (12 to 4 AM). During this time the child seemed to be wide awake and wished to play, but usually fell asleep again. This type of unusual sleep disturbance was often coupled with daytime fussiness and irritability. These children seemed dissatisfied with toys, friends, and food; nothing seemed to please them. They were argumentative and very stubborn. The characteristic sleep disturbance was reported in five of eight children on phenobarbital with transient changes, and in four of five of these the sleep disturbance was associated with daytime fussiness and irritability. Only one of eight children had these daytime symptoms transiently without sleep disturbance. The mean serum phenobarbital level in patients with transient changes was 1.5 mg/dl, which was similar to the mean of 1.4 mg/dl in patients taking phenobarbital who did not have reported side effects. Two of 35 children receiving phenobarbital were reported by their parents to have transient increased somnolence and decreased daytime activity; their drug levels at the time of their symptoms were 1.5 and 3.2 mg/dl, respectively. The four children receiving phenobarbital who had dose-related effects had both a sleep disturbance and fussiness; however, they were asymptomatic when the
Side effects of phenobarbital
3 63
Table II. Sample description and clinical course
Placebo Total number of patients Average age on entering study Sex Completed 12 mo on medication Recurrent seizure Dl~op out due to unacceptable side effects Drop out after 6-8 mo due to nonmedical reasons Patients given Bayley Scale or" Infant Development Patients given Stanford Binet Intelligence Scale
I
Phenobarbital
30 15 months
35 17 months
15 males 15 females 17
17 males 18 females 28
10 1
2 3
2
2
6
6
12
18
Table IlL Frequency of behavioral changes
Unacceptable Dose related Transient None
Placebo (N = 3O)
Phenobarbital (N = 35)
1 0 7 22
3 4 8 20
dose was lowered and kept at 2 to 3 mg/kg/day. The drug levels from these children at 4 to 5 m g / k g / d a y of phenobarbital averaged 1.9 mg/dl (range 1.5 to 2.2); whereas, at 2 to 3 m g / k g / d a y the drug levels averaged 1.1 mg/dl (range 0.7 to 1.4). Each of the three children with unacceptable side effects had a sleep disturbance and fussiness. Unfortunately, no serum phenobarbital levels were obtained from these patients. No child became hyperactive on either placebo or phenobarbital therapy, as assessed by parents and physicians, during the entire 12-month period; nor were there any significant differences between phenobarbital and placebo groups on three summary scores of emotional state, activity, and attention, as rated on the BIBR. All of the parents of children completing a year on placebo guessed correctly that their child was receiving placebo. Surprisingly, only six of 28 on phenobarbital for the entire year guessed the drug correctly; in all of these, side effects had been reported. Four of these six patients had dose-related effects, and the remaining two had reported transient somnolence. Cognitive functions. The groups did not differ in IQ
364
Camfield et al.
score, with means of 104 for placebo and 109 for phenobarbital. Multivariate analysis of the five Binet subcategories similarly yielded no significant differences between the phenobarbital and placebo groups. However, when the phenobarbital serum level at. the time of psychologic assessment was correlated with each of the five Binet subscores, there was a significant negative correlation between serum level and memory concentration subscores (r[14] = -0.52, P < 0:05). That is, when memory scores were examined relative to basic level of cognitive functioning (basal age), children with higher phenobarbital levels performed more poorly on memory concentration tasks. Consistent with this effect was a tendency, though nonsignificant, for the phenobarbital group to score lower than the placebo group on memory concentration items (P = 0.07). Although there were no significant overall differences between the two groups on the five subcategories, there were differences due to time of testing (8 vs 12 months). For children receiving phenobarbital, those tested after 12 months (n = 9) had significantly lower general comprehension scores than those tested after eight months (n = 7), P < 0.05. Patients receiving placebo (n = 6 vs 5) failed to show a similar change.* In other words, it appeared that longer exposure to phenobarbital might have reduced general comprehension scores. DISCUSSION The behavioral effects of phenobarbital in toddlers appear to fall into three categories: unacceptable, dose related, and transient. The unacceptable changes, which occurred in 9% of the children, presumably represented an idiosyncratic response to the drug and included sleep disturbance and daytime irritability. Parents could accurately identify the 11% of children with dose-related symptoms. When the dose was reduced from 4 to 5 m g / k g / d a y to 2 to 3 mg/kg/day, the serum phenobarbital levels were still generally within the accepted therapeutic range. Inconsequential behavioral changes occurring in 17% of the children were found with equal frequency among children receiving placebo and those given phenobarbital, and presumably represent the child's normal variation in mood and response to parental concern about his seizure and medication. These children were clinically recognized by the disappearance of their symptoms within a few weeks, and their parents could not distinguish phenobarbital from placebo after one year of medication. We are unaware of previous detailed descriptions of the *Samplesizesare smallerthan thosetestedbecauseof noncompliancein taking drug (n = 2) and untestablebehavior(n = 1).
The Journal of Pediatrics September 1979
effect of phenobarbital on children's sleep. Our study patients awoke in the middle of the night, seemingly fully rested and ready for normal play. Only after several hours would they go back to sleep. It was difficult to quantify this effect objectively, but the consistency with which it was reported by parents in all three categories strongly suggests that it is a specific effect on phenobarbital. Given that physicians' and parents' concerns influence their observations, the frequency of a drug's behavioral effects are best assessed by double-blind placebocontroled studies. A recent nonblind study found that 39 of 109 children with a febrile seizure treated with phenobarbital became hyperactive, as compared with 13 of 120 children in a nontreatment group. ~ The daytime fussiness and irritability of the 15 children on phenobarbital in our study who had side effects did not resemble true hyperactivity. Although the parents reported that the children were irritable, there was no increased motor activitY as assessed by parents, psychologist, or a physician. It is possible that the often mentioned phenobarbital-induced hyperactivity syndrome is seen only in children with brain damage or some other form of convulsive disorder, rather than in children with simple febrile seizures. The Bayley Scale of Infant Development and the Stanford-Binet Intelligence Test have been shown to correlate well with developmental status and moderately well with future intelligence, and are good measures of overall performance. 13 The failure to find anydifference in the mean IQ scores between phenobarbital and placebo-treated children is reassuring, and indicates that major effects on cognitive development are unlikely. However, more subtle effects suggesting need for further study were seen. Phenobarbital seemed to have an adverse effect on memory concentration tasks; scores correlated negatively with the serum drug level, despite the fact that there were no serum levels in the toxic range2 These memory effectsdo not reflect an overall deficit in cognitive functioning but only one relatj.ve to the child's basal age. However, it is possible that over a treatment period longer than eight to 12 months, the mean IQ scores in phenobarbitaltreated patients would be lower than for placebo treated. This speculation appears to be partially substantiated by the fact that general comprehension scores were significantly lower in children receiving phenobarbital for 12 months as compared with those treated for eight months. Clearly, a longitudinal assessment with finer measures of memory and concentration should be undertaken. This study does not examine the question of whether daily phenobarbital therapy or recurrent febrile seizures are more dangerous to a child. No psychologic testing was performed on children with a second febrile seizure. The literature strongly suggests that recurrent brief febrile
Vohmw 95 Number 3
seizures have no significant long-term cognitive effects, and this information should decrease the frequency o f phenobarbital use in febrile seizures. 11 However, because it will remain a useful anticonvulsant for selected patients with febrile seizures and for other children with convulsive disorders, examination of potential side effects is imPortant. Although majo r behavioral and cognitive effects o f daily phenobarbital therapy were not seen in our study, we conclude that caution should be used in exposing children to phenobarbital over many years. Careful attention should be paid to parents' observations, and the dosage o f phenobarbital should be lowered if there are side effects persisting beyond a few weeks. Serum drug levels are not good predictors of which children will develop side effects. Other than toxicity. Since memoryconcentration tasks appear to be most affected at higher serum levels; drug dosages should be reduced to provide the m i n i m u m effective level, probably about 1.5 mg/dl. ~2 Since m a n y patients are clearly protected from febrile seizures at lower serum levels, slavish adherence to the 1.5 m g / d l should be avoided if side effects are seen. Likewise. the duration of treatment with phenobarbital should be kept as short as possible, because longer exposure may adversely affect general comprehension. W e hope that more detailed psychologic tests will be developed for this age group to refine our understanding of anticonvulsant drug effects on behavioral and cognitive development. Only with such data will it be POSsible to choose appropriately from a growing list of effective anticonvulsant drugs for febrile or other seizure disorders.
Side effects o f phenobarbital
365
REFERENCES
1. Wolf SM, Carr A, Davis DC, et al: The value of phenobarbital in the child who has had a single febrile seizure: A controlled prospective study, Pediatrics 59:378, 1977. 2. Lennox-Buchthal M: Febrile convulsions, a reappraisal, Electroericephalogr Clin Neurophysiol 32(Suppl): 1973. 3. Pollack MA: Continuous phenobarbital treatment after a 'simple febrile convulsion.' Am J Dis Child 132:87, 1978. 4. Chao D~ Carter S, Gold AP: Paroxysmal disorders, in Rudolf AM, Barnett HL, and Einhorn AH, editors: Pediatrics. ed 16. New York, 1977. Appleton-Century Crofts Inc, p 1854. 5. Wolf SM. Forsythe A: Behaviour disturbance, phenobarbital. and febrile seizures. Pediatrics 61:728, 1978. 6. Thorn I: A controlled study of prophylactic long term treatment of febrile convulsions with phenobarbital, Acta Neurol Scand Suppl 60:67. 1975. 7. Heckmatt J, Houston A, Dodds K, et al: Failure of phenobarbitone to prevent febrile convulsions, Br Med J 1:559. 1976. 8. Hun SJ. Jackson. PM. Belsham A, et al: Perceptual-motor behaviour in relation to blood phenobarbital level, Dev Med Clin Neurol 10:626. 1968. 9. Jusko WJ. Levy G. and Yaffe SJ: Effect of age on intestinal absorption of r in humans, J Pharmaceutical Sci 59:487. 1970. 10. Valetl JR: A clinical profile for the Stanford Binet, J Sch Psychology 2:49. 1964. 11 Ellenberg JH. and Nelson KB: Febrile seizures and later intellectual performance, Arch Neurol 35:17, 1978. 12. Faerr O, Kastrup KW, Lykkegaard NE, et al: Successful prophylaxis of febrile convulsions with phenobarbital, Epilepsia 13:279, 1972. 13, McCall RB, Hogarty P, and Hurlburt N: Transitions in infant sensorimot0r development and the prediction of childhood IQ, Am Psychologist 27:728, 1972.