The Ketogenic Diet: Adolescents Can Do It, Too


  • Accepted January 26, 2003.

Address correspondence and reprint requests to Dr. E.H. Kossoff at Jefferson 128, The Johns Hopkins Hospital, 600 North Wolfe Street, Baltimore, MD 21287-1000, U.S.A. E-mail:


Summary: Purpose: To determine both the efficacy of and compliance with the ketogenic diet in the adolescent population.

Methods: A retrospective study of 45 patients, aged 12–19 years, consecutively enrolled in a ketogenic diet program from 1994 to 2002, was performed. Thirty-seven patients were from The Johns Hopkins Medical Institutions; eight were from The University of Texas at Houston. Charts were reviewed, and patients were contacted by telephone.

Results: Six months after diet initiation, 28 (62%) of 45 remained on the ketogenic diet, with six (21%) of 28 having 50–90% seizure reduction, and eight (29%) of 28 having >90%. At 12 months, 20 (44%) of 45 remained on the diet, with seven (35%) of 20 having 50–90% seizure reduction and six (30%) of 20 achieving >90% efficacy. Only 22% discontinued the diet for perceived restrictiveness. The mean diet duration was 1.2 years. Patients with multiple seizure types did best, whereas gender, prior seizure frequency, diet ratio, and age did not influence outcome. Patients dependent on parents for daily care were more likely to remain on the diet at 6 months, but had less efficacy. Weight loss (60%) and menstrual dysfunction (45% of female subjects) were the most commonly reported side effects.

Conclusions: The ketogenic diet is as well tolerated and efficacious for adolescents with epilepsy as for the general childhood population.

The ketogenic diet is a high-fat, low-protein, nearly absent carbohydrate diet useful in the treatment of epilepsy (1). Studies have shown that 50% of children with difficult-to-control epilepsy achieve a >50% reduction in the number of seizures by using the diet (2–5).

Although it is used primarily in toddlers and children (2,6), other age groups have been investigated. Recent studies have shown both benefit and safety for infants (7,8). A preliminary study on the diet's efficacy on adults revealed that more than half achieved >50% reduction in seizure frequency, with constipation and menstrual irregularities the most common side effects (9). However, little research has been done on the efficacy of and compliance with the diet in the adolescent population.

A healthy adolescent is confronted with major psychosocial issues as part of development, such as the formation of the emerging self (10). Teenagers with epilepsy also must deal with a societal-induced stigma as a result of their disease. Their seizures often force an increased dependence on their parents, with the teenager experiencing a loss of control and personal freedom (11). As a result, behavior problems are common and are 4 times more likely to be seen in the adolescent with epilepsy than in a healthy peer (12). Adolescents may be more sensitive to weight issues because of the possible presence of both social and societal pressures, and weight gain may be associated with antiepileptic drugs (AEDs) such as valproate (VPA) (13). Weight gain also may contribute to noncompliance (14). Studies in which adolescents are a subset of the population seemed to indicate poor compliance and less efficacy in this population compared with younger children (6).

We hypothesized that adolescents were less likely to comply with the diet and achieve seizure control by using the ketogenic diet. With a collective database totaling 53 adolescents consecutively enrolled, we decided to investigate their tolerance of the diet and its efficacy in this age group.


We conducted a retrospective study of adolescents enrolled on the ketogenic diet. Since 1994, 53 patients between the ages of 12 and 19 years diagnosed with intractable epilepsy initiated the classic Johns Hopkins protocol ketogenic diet at either the Johns Hopkins Medical Institution or at the Texas Comprehensive Epilepsy Program (1). Paper and electronic medical records were reviewed in all patients, and follow-up telephone interviews focused on compliance, efficacy, and psychosocial concerns of the patient while on the diet. Side effects specifically questioned included weight loss, constipation, kidney stones, menstrual irregularities, and intercurrent illness.

Despite multiple attempts by the examiners, six of the 53 patients were unavailable for follow-up contact. Two patients had since died for reasons unrelated to the diet, and no information beyond 3 months was obtained, so these patients were excluded from the analysis. Of the 45 remaining for analysis, 37 were from the Johns Hopkins Medical Institutions, and eight were from the Texas Comprehensive Epilepsy Program. Age, seizure frequency, seizure types, and diet administration were similar between the two groups of patients, with the exception of gender (all eight patients from the Texas group were boys). All patients were seen in follow-up every 3 months, during which weights were obtained and routine laboratory studies included lipid profiles and urinalyses.

Patients were asked to clarify the reasons for discontinuing the diet—specifically whether it was believed to be ineffective, restrictive (based on personal opinion regarding the dietary impositions), secondary to an illness, or because of freedom from seizures. Efficacy was based on the decrease in the number of seizures from baseline according to patient/parental reports. The Johns Hopkins Committee for Clinical Investigation approved this study as part of an ongoing evaluation of the effectiveness of the ketogenic diet. Categoric data were analyzed by using Pearson's χ2 for independence of rows and columns. The significance level for all tests was p = 0.05.



Forty-five adolescents (25 boys, 20 girls) were identified. The mean age at diet initiation was 14.4 years, with an average age of seizure onset of 4.0 years (range, birth–11.5 years). Patients had a mean of 419 seizures per month (range, two to 3,000 seizures/month). Twenty-three adolescents had multiple seizure types (five of which had clinical Lennox–Gastaut syndrome). The remainder had complex partial (eight), generalized tonic–clonic (four), absence seizures (four), myoclonic (three), simple partial (two), and atonic (one). Patients were taking a mean of 1.9 medications at the time of diet initiation (range, 0–4). Eleven (24%) were classified as dependent on their families for care of daily life, including bathing, dressing, and toileting. Four had documented magnetic resonance imaging (MRI) abnormalities: bilateral schizencephaly, dysplasia, hemispheric atrophy, and hippocampal asymmetry. Eight adolescents had surgery before initiating the diet (including four corpus callosum sections, two right temporal lobectomies, one right frontal lobe resection, and one left parietooccipital glioma removal). Three had prior vagal nerve stimulator implants, at 1 and 3 years before diet initiation. All adolescent patients followed the standard Johns Hopkins protocol for the diet, with 21 (47%) beginning on a 3:1 ratio (1). Eight patients began on a 3.5:1 ratio, 15 on a 4:1, and one on a 4.5:1.


The average diet duration was 1.2 years (range, 0.1–6.4 years). At 3 months, 39 (87%) patients remained on the diet, 28 (62%) at 6 months, and 20 (44%) at 12 months. Five adolescents are still on the ketogenic diet, for a mean of 3.1 years (range, 0.3–6.4 years).

Efficacy is described in Table 1. At 6 months, six (21%) of the 28 adolescents who remained on the diet achieved an ≥50–90% decrease in seizure frequency, and eight (29%) achieved >90% decrease. By 12 months, seven (35%) of 20 on the diet saw a 50–90% reduction in their seizures, and six (30%) of 20 had a >90% reduction. The average number of medications decreased from 1.9 to 1.4 by 6 months of diet use. Fifteen (33%) of the 45 adolescents also reported an increase in alertness and energy.

Table 1. Efficacy and diet duration at 6 and 12 months
 Number still on diet50–90% efficacy>90% efficacy
  1. N = 45.

 6 mo28/45 (62%)6/28 (21%)8/28 (29%)
12 mo20/45 (44%)7/20 (35%)6/20 (30%)

At diet discontinuation, 25 (63%) of 40 reported inefficacy; 10 (25%) of 40 adolescents reported diet restrictiveness; and five (12%) of 40 reported an illness interfering with the diet. Of the 10 discontinuing the diet because it was too restrictive, five (50%) had been on the diet for ≥12 months, compared with 26% of patients discontinuing for that reason in a study of long-term outcomes at our institution (18). Six (60%) of these adolescents had a >50% decrease in seizures at the time of discontinuation. The reasons for discontinuation at 1 year were similar, with 16 (64%) reporting ineffectiveness, five (20%) reporting restrictiveness, and four (16%) an illness.

Influence of gender on outcome

The mean age at diet initiation was similar between boys (14.6 years) and grils (14.2 years). Mean seizure frequency did not differ, with boys experiencing an average of 328 seizures per month, and girls experiencing 532 seizures per month (p = 0.42). Eight (62%) of the 13 boys had a >50% reduction in their seizure frequency at 6 months, compared with six (40%) of the 15 girls (p = 0.26). Significant differences were not seen for weight loss, with 14 (56%) of the 25 boys exhibiting signs of weight loss of >10% of initial body weight, and 13 (60%) of the 20 girls losing weight (p = 0.43). Four (24%) boys reported discontinuing the diet because of restrictiveness; six (35%) girls also reported restrictiveness as their primary reason for diet discontinuation. Sixteen (64%) boys reported discontinuing the diet because it was ineffective, whereas nine (45%) girls reported the same. Of the five patients who remain on the diet, three are girls, and two are boys.

Influence of age on outcome

Twenty-one adolescents were younger than 14.5 years, and 24 were ≥14.5 years at diet onset. This age was both the mean and the median age of the population. The groups were similar in age at onset of seizures, number of medications, and diet duration. At 6 months, seven (47%) of 13 of the younger group reported a >50% efficacy in seizure reduction, compared with eight (62%) of 15 older adolescents (p = 0.30). At diet discontinuation, five patients in each age group reported the diet as too restrictive.

Influence of initial severity on outcome

Efficacy did not vary substantially between those who averaged more or fewer than 120 seizures per month at onset. At 6 months, 50% of those with <120 seizures/month at diet onset remaining on the diet were >50% improved, identical to 50% of those with >120 seizures/month. At 12 months, 66% of those with <120 seizures per month at diet onset were >50% improved compared with 65% of those with >120 seizures per month. Reasons for diet discontinuation were similar in the two groups, with ineffectiveness the main factor in 14 (61%) of 23 of those with <120 seizures per month compared with 11 (50%) of 22 of those with >120 seizures per month (p = 0.40).

Influence of cognitive status on outcome

Although more patients with significant cognitive impairment remained on the diet to 6 months, fewer were improved compared with more self-sufficient adolescents. In the cognitive-impairment group, eight (72%) of 11 were still on the diet at 6 months, compared with 13 (38%) of 34 nonimpaired adolescents (p = 0.05). Reasons for discontinuation were primarily ineffectiveness in both groups. However, only three (38%) of the eight dependent adolescents had >50% at 6 months, compared with nine (69%) of the 13 independent patients (p = 0.15).

Influence of diet ratio on outcome

The initial diet ratio did not affect outcome. Although seen as traditionally more restrictive because of lower carbohydrates allowed, the 4:1 and 3.5:1 ratios were found to be restrictive by four (19%) of 21 patients started on this ratio, compared with six (25%) of 24 of those on a 3:1 ratio (p = 0.63). Efficacy also was similar, with eight (62%) of 13 with >50% improvement at 6 months in those with a 3.5 or 4:1 ratio compared with six (40%) of 15 for a 3:1 ratio (p = 0.26).

Influence of seizure type on outcome

A difference in efficacy was found among the underlying seizure types. Adolescents with multiple seizure types did best, with 12 (53%) of the 23 still on the diet at 6 months >50% improved. Eight of these 12 patients were >90% improved. Other groups did less well. In the seven patients with complex partial seizures, six were <50% improved at 3 months, and one patient had stopped the diet before then. Similarly, all three of the patients with generalized tonic–clonic convulsions on the diet at 6 months were <50% improved. Both adolescents with simple partial seizures were off the diet in a week. Mixed results were seen for absence seizures (two of three still on the diet at 6 months with >50% improvement), myoclonic (one of two patients still on the diet at 3 months with 50–90% improvement), and atonic seizures (one patient with 50–90% improvement at 6 months).

Side effects

To mimic a starvation state effectively, each patient must be as near to his or her ideal body weight as possible; therefore, we frequently plan for weight loss in our patients. Twenty-seven (60%) of the adolescents reported a reduction in their weight during the time on the ketogenic diet; seven specifically exhibited an unintended weight loss of ≥25% of their initial body weight. Of the 18 patients concurrently treated with VPA (with which weight gain is a known side effect), 13 (72%) were able to lose weight. Four adolescents noted a decrease in energy in association with weight loss, all of whom were below their ideal body weight at diet initiation.

Nine (45%) girls reported menstrual problems while on the diet; specifically, amenorrhea (six) and delayed puberty (three). Only four of nine girls with menstrual irregularities had weight loss. One adolescent began hormonal therapy (combination estrogen/progesterone oral contraceptive) to induce menses; eight experienced a return of normal menses after diet discontinuation. One girl reported menorrhagia after diet discontinuation; this was subsequently treated with an oral contraceptive.

Two individuals reported an increase in seizure activity; two, thinning hair and/or hair loss, both of which were corrected with vitamin supplementation; two, increased bruising or bleeding; and two, subjective reports of stunted growth.

Psychosocial outcomes

Regarding issues of support, three of the 15 adolescents who responded to this issue noted a lack of assistance in their local medical community as being a major factor in discontinuing the diet. Although parents were not specifically questioned, 16 families independently expressed their support for their adolescent without prompting, giving specific examples of a shared positive attitude. Only three adolescents reported difficulties with the ketogenic diet in relation to peers; however, none felt it influenced compliance. Many of our adolescents reported a “strong desire” or “strong motivation” to stick to the diet. Reasons for this motivation stemmed from a desire to drive, possess seizure control, and reduce medications.

Seven patients (five girls, two boys) either had a history of depression or current depressive tendencies during ketogenic diet therapy significant enough to be mentioned by the patient or family during clinic visits. These adolescents, when asked about “depressed mood,” confirmed histories of depression during follow-up telephone contact. These patients did not have a higher seizure frequency than the other adolescents, with a mean of 128 seizures/month (range, 6 to 5,000/month). Their diet duration, at 1.1 years, was only slightly less than the overall mean of 1.2 years. Two patients were noted to have both epileptic and nonepileptic seizures.

Long-term outcomes

At this time, with the follow-up period as long as 7 years, 24 (53%) have fewer seizures than when the diet was initiated; five have the same number; and four have increased seizures. Twelve patients have unclear seizure frequency because of lack of recent follow-up. Seizure freedom was achieved in five patients, four after surgical resection and one after vagus nerve stimulation. Medications are reduced overall in nine adolescents; eight are taking the same number; and 14 are taking more medications. The current medication numbers of eight adolescents are unknown. Twenty-six (58%) reported being happy that they tried the diet.

Nine (20%) adolescents went on to have surgery after the diet, and two additional patients are considering surgery at this time. Two patients were strong candidates for surgery (one with unihemispheric atrophy and the other with focal dysplasia) but chose not to pursue surgery because of religious beliefs and to avoid the possibility of vision impairment, respectively. Vagus nerve stimulators were implanted in 11 (24%) adolescents after diet discontinuation.


No longer regarded with suspicion, the ketogenic diet has gained increasing respect in the medical profession as a useful treatment for children with intractable epilepsy (1,15). However, few adolescents are referred for the diet, as it is commonly believed to be too restrictive and not efficacious. We found no evidence to support this belief, disproving our initial hypothesis that the diet would be less efficacious and tolerated. In this study, 46% of adolescents remained on the diet for ≥12 months, nearly identical to a study of a younger population in which 47% remained on the diet at this time (2). Granted, this was not a prospective study, and even with two epilepsy centers involved, our combined population was small.

Adolescents, while dealing with issues such as peer pressure, depression, self-esteem, and other psychosocial issues, did not seem to have increased difficulty in following the diet's restrictions. Our results suggest that both family support and personal motivation are the primary factors in achieving success on the diet. Family support has been proven to correlate positively with high success in compliance in adolescents with a chronic disease (16), and our subjective reports follow that trend. Personal motivation, conversely, although also positively correlating to increased compliance, is much more likely to be present in an adolescent than in a younger child (17). Half of the patients who felt the diet was too restrictive still remained on the diet for 12 months. They appear to have been highly motivated, but the diet either interfered too much with their lives or was too difficult to maintain for the degree of seizure control they achieved.

Equally important in terms of compliance is the issue of efficacy. At 6 months, our results show that 58% had a >50% reduction in their seizures. Freeman et al. (2) reported efficacy to be slightly higher at the 6-month period, with 68% having a >50% reduction. Similarly, at 12 months, 65% of our patients remaining on the diet had a >50% reduction, in comparison to 77% of the children in Freeman's study. Our results also found that the average number of medications decreased by 26% after 6 months of being on the diet.

Regarding side effects, weight loss was often planned, with very few having excessive loss. In adolescents who are often becoming aware of their self-images, the attainment of ideal body weight can be motivation itself. Sheth (13) noted that the weight gain typically experienced by adolescents with epilepsy is resistant to diet intervention; our results appear to indicate otherwise when the tightly calorie-controlled ketogenic diet is used. The incidence of depression in our adolescents (16%) appeared to be no higher than that noted by a previous study in which 23% of adolescents with epilepsy were depressed (18). Likewise, the higher rate of girls with depression is in accordance with the literature on the incidence of depression in adolescents with a chronic disease, with twice as many girls as boys (19). The diet's effect on menses was not surprising, with 45% of our girls reporting amenorrhea or irregular cycles, although significant weight loss was seen by only half of those with menstrual irregularities. By minimizing body fat, the diet may mimic the menstrual side effects seen in starvation and certain female athletes.

The effectiveness of the diet in this age group was surprising to us. It also is surprising in view of the population enrolled in this study—the high incidence of surgical candidates, structural abnormalities, and interventions such as the vagus nerve stimulator suggests that if used in a less intractable population, it might be even more effective.


In summary, the ketogenic diet is a reasonable alternative for adolescents with difficult-to-control epilepsy. The diet is no more restrictive for this age range than for younger children and has similar efficacy. Although it is apparent that adolescents experience a much more diverse set of life issues that could negatively affect the diet, equally strong motivation factors are present. In an era of new AEDs and vagal nerve stimulation, the ketogenic diet should be considered an option for the adolescent population.


Acknowledgment: Dr. Kossoff and Ms. Mady were partially supported by the Roxanne Fund and the Vaswani Family Fund. Dr. Freeman is the Lederer Professor of Pediatric Epilepsy.