Same-Drug Comparisons

• MPH vs. MPH (Fitzpatrick, Klorman, Brumaghim et al., 1992; Pelham, Sturges, Hoza et al., 1987; Srinivas, Hubbard, Quinn et al., 1992; Whitehouse, Shah, and Palmer, 1980): One of the four trials compared different enantiomers (Srinivas, Hubbard, Quinn et al., 1992), whereas the other three trials compared sustained-release with regular formulations. The study that evaluated different enantiomers of MPH found that drug-induced improvements in sustained attention was attributable to the d-MPH (Srinivas, Hubbard, Quinn et al., 1992). The studies evaluating different formulations found no differences between them but had small sample sizes (Evidence Table B1).
• DEX vs. levoamphetamine (Arnold, Huestis, Smeltzer et al., 1976): The main conclusion from this study is that both isomers were better than placebo but not significantly better than each other in terms of improvement of core symptoms and academic performance (Supplemental Tables B4 and B6). Only in one parent report of global symptoms was DEX better than levoamphetamine (Supplemental Table B4). Parents' ratings, but not teachers' or psychiatrists' ratings, showed significant placebo effects. This study included only 31 patients.

Stimulant vs. Stimulant

There were 13 trials in this group. Only one studied more than 100 patients and six included fewer than 30 patients. Two trials studied the effect of the interventions for 8 weeks, whereas most studied the stimulants for 3 weeks or less (Evidence Table B1).

• MPH vs. DEX (Arnold, Christopher, Huestis et al., 1978; Borcherding, Keysor, Cooper et al., 1989; Castellanos, Giedd, Elia et al., 1997; Efron, Jarman, and Barker, 1997; Elia, Borcherding, Rapoport et al., 1991; Gross, 1976; Kauffman, Smith-Wright, Reese et al., 1981; Matochik, Liebenauer, King et al., 1994; Winsberg, Press, Bialer et al., 1974). Of the nine trials, seven were placebo-controlled and three provided data on effectiveness outcomes of interest. Most of the results of these studies suggest that there are no differences between MPH and DEX. The two placebo-controlled studies with data on outcomes of interest showed that both stimulants were more effective than placebo for the improvement of core symptoms (Supplemental Table B4) (Arnold, Christopher, Huestis et al.,1978; Elia, Borcherding, Rappoport et al., 1991). Neither provided the number of patients randomized to the study groups or reached a score compatible with high quality (Evidence Table B1).
• One trial used a triple-blind crossover design and was designed to look at compliance with treatment (Kauffman, Smith-Wright, Reese et al., 1981). This study found that the mean individual patient compliance during MPH administration was 65 percent (0 to 100 percent) and during DEX administration was 60 percent (20 to 83 percent). The authors stressed that poor compliance may be an important reason for the variability and discordance among the results of the studies in this area.
• MPH vs. pemoline (Conners and Taylor, 1980; Stephens, Pelham, and Skinner, 1984). Both studies in this group were placebo-controlled, but only one reported data on effectiveness outcomes of interest (Conners and Taylor, 1980). This study received a score >3 points, provided information on 17 of the 20 clinically relevant elements, was published in 1980, and included 60 patients diagnosed as having minimal brain dysfunction. Patients were randomized to receive pemoline at 18.75 mg/d, MPH at 5 mg two times per day, or placebo for 8 weeks. Overall, the two drugs produced similar effects and were more effective than placebo (Evidence Tables B1, Supplemental Tables B4-B7).
• DEX vs. pemoline (Conners, Taylor, Meo et al., 1972): In this study, 81 children with diagnosis of minimal brain dysfunction were randomized to receive pemoline at 25 mg, DEX at 10 mg, or placebo for 8 weeks. This study found both drugs to be better than placebo but found few differences in terms of effectiveness and adverse effects between them.
• MPH vs. pemoline vs. DEX (Pelham, Greenslade, Vodde-Hamilton et al., 1990): This study compared administration of regular MPH at 10 mg two times per day, slow-release MPH at 20 mg/d, sustained-release DEX at 10 mg in the morning, and pemoline at 56.25 mg/d and a placebo-controlled arm. The main conclusion from this study was that the four medications had "generally equivalent and beneficial effects." However, the report did not provide the number of patients randomized or the actual data on formal comparisons of the outcomes produced by all drugs simultaneously. The comparisons between individual stimulants and placebo produced inconsistent results across the groups of outcomes (Supplemental Tables B4-B7). The report included information on 11 of the 20 clinically relevant elements and achieved a score compatible with low quality (Evidence Table B1).

Stimulants vs. Tricyclic Antidepressants

Only four studies, all comparing MPH with tricyclic antidepressants, were found. One trial compared MPH with desipramine (Rapport, Carlson, Kelly et al., 1993). This was a randomized, double-blind, crossover placebo-controlled study in which 16 patients were allocated to receive desipramine in titrated doses to achieve plasma levels between 125 and 225 mg/ml; MPH at 10, 15, or 20 mg; a combination of desipramine and MPH at the above levels; or placebo. The children included in this study had diagnoses of ADHD with depressive symptoms, ADHD and major depressive disorder, or major depressive disorder with prominent attentional symptoms. The main conclusions from this study were that MPH was consistently the most potent drug in affecting children's vigilance (Continuous Performance Test [CPT]) and ability to learn paired associations (Paired Associate Learning [PAL]). Additional improvements in performance were observed when desipramine was added. Desipramine alone, however, did not affect children's vigilance but did enhance paired association learning to a lesser extent than that achieved by MPH alone or in combination. The authors caution about the interpretation of their results because of the small sample size and the degree of psychopathology experienced by the children studied.

Three trials compared MPH with imipramine (Quinn and Rapoport, 1975; Spring, Yellin, and Greenberg, 1976; Werry, Aman, and Diamond, 1980). One did not provide data on effectiveness outcomes of interest (Spring, Yellin, and Greenberg, 1976). The other two trials provided inconsistent results. One showed nonsignificant differences between MPH and imipramine (Quinn and Rapoport, 1975), whereas the other showed some results in favor of imipramine (Werry, Aman, and Diamond, 1980).

In summary, the small group of studies comparing different formulations or enantiomers of the same stimulants suggests that there little difference between them, at least in the short term, but that the active medications are more effective than placebo. Similarly, despite the overall heterogeneity of the studies comparing different stimulants, they showed few, if any, differences among MPH, DEX, and pemoline. These findings agree with those of the AMA and AACAP reports. The sample size of the studies, however, is small.

Studies comparing stimulants with tricyclic antidepressants had many limitations and presented conflicting results. More rigorous studies are clearly needed to establish the relative effectiveness of stimulants and tricyclic antidepressants. Studies are required to compare the effects of stimulants with clonidine, buproprion, or selective serotonin-reuptake inhibitors.

Stimulant vs. a Combination of Two or More Drugs (Question: Is a Combination of Drugs More Effective Than Stimulant Alone?)

• Garfinkel, Webster, and Sloman, 1981: This study compared two doses of caffeine, combination of caffeine with MPH, MPH alone, and placebo. The sample consisted of six boys between 6 and 10 years of age attending a day treatment program at a psychiatric institution who had been referred because they were not progressing academically or socially while in a regular or special education class. The researchers found that the strongest effect was noted when caffeine was combined with MPH and that caffeine at a low dose was superior to caffeine at a high dose. This study is difficult to interpret because of the small sample size, sample selection, lack of measurement of compliance or adherence to regime, and use of teacher report only as an outcome measure.
• Gittelman-Klein, Klein, Katz et al., 1976: This study evaluated the effect of thioridazine (THIOR), MPH, and combination therapy. This study had the largest sample size in this group and obtained a high-quality score (3). It also included information on 16 of the 20 clinically relevant elements. The study used large doses of MPH (average dose of 50 mg, 1.78 mg/kg). A significant improvement was seen in global assessments of core symptoms of ADHD with MPH alone or in combination with THIOR compared with THIOR alone. Improvements in individual core symptoms, academic performance, and symptoms associated with conduct or oppositional disorders were reported by teachers but not by parents. There were no differences on the symptoms of depression (Evidence Table D1).
• Levy and Hobbes, 1996: This study examined the use of haloperidol in combination with MPH to determine whether haloperidol blocks the effect of MPH if given before MPH. The sample included 10 boys between 5 and 12 years of age who met the criteria for ADHD when DSM-III criteria were used but who also had comorbid ODD. Measure of outcome relied only on the Continuous Performance Test, and the result is difficult to interpret. MPH improves reaction time on the CPT, and this effect is eliminated by use of haloperidol before administration of MPH.
• Rapport, Carlson, Kelly et al., 1993: This study compared MPH combined with desipramine, MPH alone, desipramine alone, and placebo. The sample included 16 boys who were hospitalized with ADHD and mood disorder. The report obtained a quality score of 3 with the validated scale and included information on 17 of the clinically relevant elements (Evidence Table D1). This study found that MPH was potent in affecting children's vigilance (CPT) and their ability to learn paired associations and do the Matching Familiar Figures Test (MFFT). Desipramine had no effect on vigilance but improved learning paired associations (not as much as MPH) and was equally effective on the MFFT. Combination therapy led to some blocking of the vigilance but improved MFFT and the Stimulus Equivalence Paradigm (SEP), which suggests some additive effect on behavioral inhibition and complex problemsolving abilities. Some of the limitations of this study are the study sample that consisted of hospitalized patients who had significant coexisting mood disorder, lack of individual outcome data, and the clinical significance of outcome measures.
• Schechter and Timmons, 1985: This study used a crossover design to compare DEX at 1.6 mg and 5.0 mg, DEX at 1.6 mg combined with caffeine at 300 mg, and caffeine at 300 mg alone. The sample included 10 patients between 6 and 10 years of age who were diagnosed as having hyperkinetic syndrome using criteria other than DSM. After 2 weeks, all treatments led to significant increase in reaction time, significant decrease in errors of commission, and decreased total scores on the Conners Abbreviated Parent Rating Scale (ACRS). The report did not mention whether there was significant difference between treatments. This study is limited not only by its poor rating of quality but also because of small sample size, short duration of intervention, lack of measurement of compliance or adherence to regime, fixed doses of DEX, and use of dependent measures that are of questionable clinical significance.

Stimulant vs. Stimulant Plus Nondrug Intervention (Question: How Much Do Nondrug Interventions Add to the Effect of Stimulants?)

Of the five studies in this group, four evaluated MPH and one DEX. Four studies (Brown, Wynne, and Medenis, 1985; Conrad, Dworkin, Shai et al., 1971; Firestone, Crowe, Goodman et al., 1986; Klein and Abikoff, 1997) allowed direct head-to-head comparison of stimulant vs. nondrug intervention. The fifth study is Long, Richert, and Ashcraft, 1993. The studies are sorted alphabetically by author for ease of reading. The types of nondrug interventions vary widely and are highlighted in the summary.

• Brown, Wynne, and Medenis, 1985 (stimulant vs. stimulant plus cognitive behavioral therapy): This study compared the effect of cognitive behavioral therapy (CBT), MPH alone, and combination therapy (MPH plus CBT). Thirty hyperactive boys with academic difficulties between 6 and 11 years of age were randomized. Outcome measures were very broad, focusing on core symptoms of ADHD and academic measures. The CBT program consisted of 24 1-hour individual sessions 2 times per week over a 3-month period and consultation sessions for both parents and teachers. The results showed that only those children in the two medication treatment conditions demonstrated improvement in both attentional and behavioral ratings and that the combination therapy was not any more effective than MPH alone.
• Conrad, Dworkin, Shai et al., 1971 (stimulant vs. stimulant plus child training): This study examined the effect of DEX alone or with twice weekly "prescriptive perceptual-cognitive tutoring" (child training [CT]). The sample consisted of 68 children from a low socioeconomic class; age was not specified. The diagnosis of ADHD was based on the Schenectady Hyperkinetic Scale. The children were randomly assigned to placebo alone, DEX alone, DEX plus CT, or placebo plus CT. The duration of the interventions was not clearly reported. Neither experimental condition significantly influenced academic achievement as measured by the Wide Range Achievement Test.
• Firestone, Crowe, Goodman et al., 1986 (stimulant vs. stimulant plus parent training): In this study, the use of parent training (PT) in the treatment of ADHD was evaluated in 73 children between 5 and 9 years of age. Children were assigned to PT plus MPH, PT plus placebo, or MPH alone. Parents were asked to read a book on child management and were then asked to join a parents' discussion group. The results indicate that at 3 months there were significant differences between those in the two medication groups in terms of hyperactivity index, conduct problem, and reaction time but not the verbal grade level. Parent training added very little. At the first- and second-year followup, there were no significant differences among the groups. Some of the limitations of this study include substantial attrition over time, no report of the number of patients randomized, low-quality score, lack of measurement of compliance with treatment, and little information as to the extent of parent participation in the program.
• Klein and Abikoff, 1997 (stimulant vs. stimulant plus cognitive behavioral therapy): This study compared the effects of CBT, MPH alone, and CBT plus MPH together. The sample consisted of 89 children 6 to 12 years of age whose diagnosis of ADHD was based on the Conners Teacher Rating Scale. Children with conduct disorder were systematically excluded. Behavior therapy for home and school utilized operant conditioning principles individualized for each child. Dose titration for MPH was carried out up to a maximum of 60 mg/d for optimal efficacy. The results suggested that MPH alone and in combination with CBT were significantly better than CBT with placebo on most outcomes. Some differences between MPH alone and combination therapy include less attention-seeking behavior, better impulse control, and more cooperative behavior. Teachers and psychiatrists also rated the children more improved on the Clinical Global Improvement (CGI).
• Long, Rickert, and Ashcraft, 1993 (stimulant vs. stimulant plus bibliotherapy): This study examined the use of bibliotherapy as an adjunct to stimulant medication. Bibliotherapy refers to written behavioral management protocol. The sample consisted of 32 children between the ages of 6 and 11 who were recruited concurrently from outpatient pediatric clinics. Children were assigned to MPH alone or with bibliotherapy. Results revealed that the intensity of behavior lessened for the bibliotherapy group even though the problems remained. One of the major limitations of this study is that there was no surveillance of medication compliance during the study which could have influenced the results.

Nondrug vs. Stimulant Plus Nondrug Intervention (Question: How Much Do Stimulants Add to the Effect of Nondrug Interventions?)

There were 14 studies in this category. In total, 13 studies evaluated the effect of MPH and one the effect of DEX (Conrad, Dworkin, Shai et al., 1971). Five of these studies have been discussed in the section above (Brown, Wynne, and Medenis, 1985; Conrad, Dworkin, Shai et al., 1971; Firestone, Crowe, Goodman et al., 1986; Klein and Abikoff, 1997; Long, Rickert, and Ashcraft, 1993). The remaining nine are described below.

• Borden and Brown, 1989 (cognitive behavior therapy vs. stimulant plus cognitive behavioral therapy): This study was designed to determine whether those treated with combination of medication and CBT would tend to make external attributions for problem solutions or whether more internal attributions would be observed with CBT alone. Thus, this study did not focus on the effects of interventions on the core symptoms or associated behavior/academic difficulties in children with ADHD.
• Brown, Borden, Wynne et al., 1986 (cognitive behavior therapy vs. stimulant plus cognitive behavioral therapy): This study examined the effect of combined CBT and MPH. The study achieved a high-quality score and provided information on 14 of the 20 clinically important elements. The study included 40 patients diagnosed with ADHD using DSM-III criteria between 5 and 13 years of age and of average intellectual functioning. Both CBT and attention control (AC) involved 22 1-hour sessions over 3 months, but the difference was in the inclusion of specific problemsolving strategies that were taught in the CBT group. MPH was given at 0.3 mg/kg two times per day and was discontinued during pre- or postmeasurements. The results showed that no significant main treatment effects or interactions among treatments occurred. They also noted that the effects of stimulant medication were noticeable while children were receiving it. However, these effects dissipated rapidly upon discontinuation of the medication.
• Brown, Borden, Wynne et al., 1988: This study was designed to determine the extent to which compliance with treatments occur in treatment groups rather than to determine the effects of interventions on the core symptoms or associated behavior/academic difficulties in children with ADHD.
• Carlson, Pelham, Milich et al., 1992 (behavior therapy plus stimulant vs. regular classroom plus stimulant): Twenty-four boys between 6 and 12 years of age participating in an 8-week summer treatment program for ADHD were randomized to 2 classrooms (behavior management classroom [behavioral contingencies/modification, or BC/Mod] vs. regular classroom [AC]) and to placebo or MPH at a low dose or high dose. Children showed significantly higher rates of on-task behavior, lower rates of disruptive behavior, and higher rates of following rules when receiving MPH than when receiving placebo. The medication effects on on-task and disruptive behavior interacted with classroom setting such that in regular classroom settings, increasing the dose of MPH had a linear effect of increasing dosage. The study seems to suggest that low dosages of MPH are sufficient to maximally improve children's behavior when behavioral classroom techniques are employed.
• Christensen, 1975 (behavior therapy vs. behavior therapy plus stimulants): This study examined the effect of behavior therapy alone (BC/Mod) or in combination with MPH. The subjects were 16 "hyperactive, institutionalized retarded" children between 9 and 16 years of age who were randomly assigned to 2 groups: MPH plus BC/Mod (main element of the program was a token reinforcement system) or placebo plus BC/Mod. The results showed that the BC/Mod procedures in the presence of the placebo were just as effective as the same procedures in the presence of MPH, suggesting no added effect of MPH. However, it is important to note that MPH was given one time per day and the dose was not individually titrated.
• Hinshaw, Henker, and Whalen, 1984 (cognitive self-evaluation vs. cognitive self-evaluation plus stimulant): This study evaluated the cognitive self-evaluation procedures and MPH on social behavior in a playground setting. Twenty-four hyperactive boys between the ages of 8 and 13 who were already receiving treatment with MPH were randomized to "reinforcement alone" with MPH, "reinforcement alone" with placebo, "reinforced self-evaluation" (RSE) with MPH, or RSE with placebo. No attempts were made to standardize the medication dose, each boy remaining at his customary dosage. The results show that RSE in conjunction with MPH is superior to medication alone or RSE alone as judged by observations of social behavior on playground.
• Hinshaw, Buhrmester, and Heller, 1989 (cognitive behavioral therapy vs. stimulant plus cognitive behavioral therapy): This study examined the effect of MPH on anger control in 24 boys between the ages of 6 and 12 who were participating in a 5-week summer research program. Children received CBT using "specific curriculum" three times per week to enhance anger control during verbal provocation and were also randomized to MPH. The results indicated that MPH exerts significant effects on the global rating of self-control and physical retaliation when the child was provoked, but the data were difficult to extract. Furthermore, the outcome measures were of questionable clinical significance.
• Pelham, Carlson, Sams et al., 1993 (behavioral therapy plus stimulant vs. regular classroom plus stimulant): Thirty-one ADHD boys ranging in age from 5 to 10 years participated in an 8-week day treatment program. The boys alternated, on a weekly basis, between classrooms with behavior modification and no behavior modification and received in random order placebo or MPH at a low dose or high dose. Results revealed significant effects of both interventions with the mean effect size of medication being more than twice as great as that of the behavior modification. The addition of either dose of MPH resulted in improvement beyond the effects of behavior modification alone.
• Solanto, Wender, and Bartell, 1997 (behavioral contingencies/modification vs. behavioral contingencies/modification plus stimulants): This study evaluated the effect of MPH in combination with BC/Mod on a measure of sustained attention (CPT) in 22 children between 6 and 10 years of age who met DSM-III-R criteria for ADHD. The results indicated that MPH improved sustained attention whereas behavioral intervention alone did not. Adding behavioral intervention to MPH did not lead to further improvement. Unfortunately, this study did not explore whether children with better CPT performance had improvements in other core symptoms of ADHD.

The largest and most comprehensive study conducted to date was designed to determine the effectiveness of different treatment strategies, involving careful titration and systematic testing of drugs other than MPH if needed to achieve an adequate response.

• MTA Cooperative Group, 1999: Five-hundred-seventy-nine children, between the ages of 7 and 9, were randomly assigned to one of four 14-month-long conditions: (1) medication management (MedMgt) with monthly visits and supportive care; (2) intensive behavioral treatment (Beh) (including home- and school-based behavior management, direct behavioral management therapy for the child delivered through a summer camp program); (3) both treatments combined (Comb); or (4) assessment and referral back to community service providers (CC). Children's outcomes in clinical and functional domains were assessed throughout the study, using rating scales and direct observational measures. Comb and MedMgt treatments were clinically and statistically superior to intensive Beh and routine CC in reducing core ADHD and oppositional-aggressive symptoms. Moreover, the Comb and MedMgt approaches did not differ statistically or clinically from each other for ADHD or oppostional-aggressive symptoms. In several non-ADHD-symptom areas (parent-rated internalizing symptoms, social skills, consumer satisfaction, and reading), Comb provided modest incremental benefits over single-treatment approaches. MedMgt was superior to medication management provided in the community despite the fact that two-thirds of the CC subjects were treated with medication during the study period.

In summary, there is a lack of evidence supporting the superiority of combination therapy over stimulant alone or superiority of combination therapy over nondrug intervention alone. More research is needed to determine the added value of including nondrug interventions to the treatment scheme of patients who are already receiving stimulants, as well as the added value of treatment with stimulants when nondrug interventions fail to achieve the desired outcomes.

Studies of Desipramine

Six studies provided data on effectiveness outcomes of interest. All six of these trials showed results in favor of desipramine.

• Biederman, Baldessarini, Wright et al., 1989a: This study was described in a group of three reports (Biederman, Baldessarini, Wright et al., 1989b; Biederman, Baldessarini, Wright et al., 1993). This study, which did not report the number of patients randomized, found significant improvements in parent and teacher ratings of primary symptoms and in depression in favor of desipramine.
• Donnelly, Zametkin, Rapoport et al., 1986: In this study, Conners teacher ratings of primary symptoms improved and classroom activity level declined significantly, although scores on continuous performance and memory tasks were not significantly different. Significant increases in blood pressure and heart rate were noted in the desipramine group.
• Gualtieri, Keenan, and Chandler, 1991: In this study, desipramine at three doses (low, medium, and high) was evaluated against placebo in 12 patients. Desipramine at the medium dose resulted in the best scores on "clinical significance," global rating, and teachers' reports of inpatient behavior. Of all individual school behavior, only "fidgety" behavior showed a meaningful linear trend. In addition, this study showed that patients were least "cheerful" on medication at high doses and that motor performance was adversely affected by the drug.
• Rapport, Carlson, Kelly et al., 1993: This study compared MPH combined with desipramine, MPH alone, desipramine alone, and placebo. The sample included 16 boys who were hospitalized with ADHD and mood disorder. The report obtained a quality score of 3 with the validated scale and included information on 17 of the clinically relevant elements (Evidence Table E1). This study found that MPH was potent in affecting children's vigilance (CPT) and their ability to learn paired associations and do the MFFT. Desipramine had no effect on vigilance but improved learning paired associations (not as much as MPH) and was equally effective on the MFFT. Combination therapy led to some blocking of the vigilance but improved MFFT and SEP, which suggests some additive effect on behavioral inhibition and complex problemsolving abilities. Some of the limitations of this study are the study sample that consisted of hospitalized patients who had significant coexisting mood disorder, lack of individual outcome data, and the clinical significance of outcome measures.
• Singer, Brown, Quaskey et al., 1995: In this study, the effects of clonidine, desipramine, and placebo were compared in children with Tourette's syndrome and ADHD. On a global linear analog scale, parents rated children as significantly better on desipramine than on clonidine or placebo. Parent ratings on the Child Behavior Checklist hyperactivity subscale for boys on desipramine were significantly lower than for those on clonidine. Teacher ratings on the nervous overactive subscale were lower for boys during treatment with clonidine and desipramine than placebo.
• Wilens, Biederman, Prince et al., 1996: In this study, 68 percent (13/19) of the participants showed a positive response to desipramine (very much or much improved plus at least a 30 percent reduction in rating scale symptoms). No participants showed a favorable response to placebo. Two participants receiving desipramine dropped out, and 10 had doses lowered as a result of adverse effects. Favorable response was not related to the presence of lifetime comorbidity of anxiety or depressive disorders.

Studies of Imipramine

• Gualtieri and Evans, 1988: In this study, imipramine showed a dose-response relationship with increased scores on a selective attention task but reduced motor performance (finger-tapping speed) in nine children.
• Werry, Aman, and Diamond, 1980: Compared with placebo, imipramine reduced impulsive responses and seat movement significantly, but it did not reduce omission errors or have an impact on self-esteem. Psychiatrists' evaluation of improvement using Clinical Global Improvement showed significantly better results with imipramine compared with those with MPH and placebo. Parent ratings using the Conners Parent Questionnaire showed inconsistent results. There were no differences among the three groups when teacher ratings were used.
• Winsberg, Kupietz, Yepes et al., 1980: This study included 11 children who had failed to respond to MPH. No significant benefit was obtained either on parent and teacher ratings of the child's hyperactivity or on the patient's performance in a laboratory measure of sustained attention, except for improvements in attention measured by parents using Conners.

In summary, the studies on desipramine, regardless of their heterogeneous designs, small sample sizes, and variable quality, suggest that desipramine is more effective than placebo. The studies evaluating imipramine, however, show inconsistent results.

Studies Evaluating MPH vs. Placebo

• Brown, Borden, Wynne et al., 1986: This study, which included 40 children and adolescents, obtained a high-quality score and included information on 14 of the 20 clinically relevant elements selected for extraction. It showed no advantage of cognitive training over no training in terms of improvement of core symptoms, academic performance, and conduct/oppositional disorders.
• Brown, Borden, Wynne et al., 1988: This study evaluated compliance with medication. Subjects were randomly assigned to a 3-month protocol of MPH at 0.3 mg/kg two times per day or inert placebo two times per day and then were also assigned randomly to a cognitive therapy or AC training program. There was a trend for parents of children receiving placebo to report that more doses of medication and more therapy sessions were missed.
• Firestone, Crowe, Goodman et al., 1986: This study included children from 5 to 9 years of age referred to various outpatient clinics who met criteria for ADHD both at home and at school. The diagnostic procedure is not specified. Patients were randomly assigned to one of three groups: parent training plus placebo, MPH plus PT, or MPH only. A real asset of this study is the careful attention to enrollment, attrition, and dropout rates. Patients given MPH and the combined PT plus MPH showed equivalent improvement from pre- to posttest (3 months), and both of these arms showed greater improvement than the PT plus placebo arm in behavior and attention, but not in reading. No differences were detected after 12 and 24 months, perhaps as a result of numerous crossovers and dropouts. Subjects who switched to MPH showed improvement following the switch.
• Kupietz, Winsberg, Richardson et al., 1988: All 58 subjects in this study met criteria for ADHD and for reading disability and therefore are not typical of ADHD children in most clinics. The study also obtained a high-quality score and included information on 16 of the clinically relevant elements extracted. Behavioral improvement was noted for patients given MPH compared with those given placebo. MPH at the higher doses yielded the greatest effect. Eleven patients dropped out because of the lack of benefit or because of adverse effects. There was no improvement in academic performance, apart from improvement in reading which was observed even though posttesting was conducted off medication. There was evidence of relapse between 3 and 6 months.
• Schachar, Tannock, Cunningham et al., 1997: In this trial, 91 school-age children were randomized to receive a titrated dose of MPH or placebo two times per day for 16 weeks. This study achieved a high-quality score and included information on 17 of the clinically relevant elements extracted from the reports. The children had been referred to a psychiatric clinic for disruptive behavior problems and met criteria for ADHD based on parent and teacher interviews. Subjects were randomized to receive MPH or placebo, and their families were randomized to attend parent training or parent support groups. Medication was titrated against behavior effects and adverse effects up to a typical dose of 0.5 mg/kg given two times per day. MPH administered two times per day resulted in improvement in ADHD and ODD symptoms at school compared with placebo, but not at home (Evidence Table F1). The behavioral effects were evident within several weeks of treatment initiation. Physiological adverse effects were also evident within weeks, but some adverse effects such as irritability were not detected until 4 months after treatment initiation. This study did not measure academic outcomes.

Studies Evaluating MPH vs. Other Interventions

• Brown, Wynne, and Medenis, 1985: In this study, CBT alone resulted in minimal improvement in attention and no improvement in behavior or academic performance. The MPH and the combined treatments resulted in improved attention and behavior but little improvement in academic performance. Combined treatment was not superior to MPH alone.
• Gittelman-Klein, Klein, Katz et al., 1976: This study has been described in previous sections. Initially, the combination MPH plus THIOR was superior to MPH or THIOR alone, although combined treatment was not superior to MPH alone after 12 weeks. Adverse effects were more common with the combined treatment group. The duration of the effect of the combined treatment was longer than that of MPH only, resulting in greater improvement in parent ratings.
• Quinn and Rapoport, 1975: Patients who had participated in two independent RCTs of MPH vs. placebo and imipramine vs. placebo were followed for 1 year. Children who had been assigned to placebo in the initial studies (6 weeks) were assigned to either MPH or imipramine. Treatments were provided by community physicians. Behavioral improvement was evident in both medicated groups but not in the unmedicated group according to teacher ratings; equivalent improvement in behavior was evident in all three groups according to parents. No improvement was evident on academic, intellectual, or cognitive measures. MPH appeared to be the more effective medication, since fewer subjects discontinued treatment as a result of adverse effects or lack of improvement.

Studies Evaluating Dextroamphetamine

• Conrad, Dworkin, Shai et al., 1971: Eighty-one subjects were selected from children in kindergarten and first and second grade attending 7 schools in a lower class neighborhood in 1969. To qualify for the treatment program, children had to be rated in the top 19 percent on the teacher hyperactivity ratings and exhibit abnormalities on at least one of various measures of perceptual-cognitive impairment. Dosage of medication (or placebo) was initiated at 5 mg/d and increased to 10 to 20 mg/d based on response. Subjects who had not received at least a 60-day supply of medication were excluded from the analysis on the grounds that they were noncompliant. The tutoring program was elaborate, covered academic and emotional issues, was lengthy (2 times per week for 20 weeks), and was well described. Fidelity of the tutoring program was not assessed. Neither experimental condition influenced academic achievement significantly.
• Gillberg, Melander, von Knorring et al., 1997: This study merits special mention because it is the only long-term trial conducted outside of North America and one of the four studies published since 1991. Several features of this study are unusual and weaken the ability to generalize to other samples of ADHD children. Reasons for withdrawal from the study suggest that subjects were either nonresponders or were having adverse effects of DEX during the initial unblinded assessment of DEX effects. This limits the validity of the results of the subsequent comparison of DEX and placebo. The description of comorbidity indicates that this sample was unlike those treated with psychostimulants in North America. Seven of 62 subjects had many autistic features, 10 had mild mental retardation, and 2 had Tourette's syndrome. Hallucinations occurred more often in medicated children than has been reported in other studies (3/61), but on the whole, adverse effects were few and relatively mild. Positive effects of the drug were present 15 months after treatment was started.
• Greenhill, Rieder, Wender et al., 1973: Subjects were 9 severely hyperactive children, 6 to 16 years of age, who were unresponsive to drugs and psychotherapy and had no evidence of psychosis or medical illness. Lithium treatment was initiated while the children were inpatients and then continued for 3 months while they were outpatients. The trial consisted of an initial, nonrandomized phase in which placebo, DEX, and lithium were each administered for 1 week in a single-blind fashion. The second phase was conducted on an outpatient basis and consisted of a modified double-blind trial of 3-week phases of DEX, lithium, and placebo. Children showed no improvement or worsening of symptoms when given lithium. DEX was associated with improvement in hyperactivity and concentration compared with placebo as reported by parents, teachers, and psychologists and by several cognitive-perceptual measures. Tutoring accounted for improvement on the Information subtest of the Weschler Intelligence Scale for Children (WISC). The combination of medication and tutoring resulted in improvement in copying, motor pattern, reduction in errors on a cognitive task, reduction in hyperactivity, and increase in full-scale intelligence quotient (IQ). The interventions did not improve academic achievement.

Studies Evaluating Other Interventions

• Fehlings, Roberts, Humphries et al., 1991: This study included 26 boys, 7 to 13 years of age, who had been referred to a child development clinic. Patients in the CBT group showed greater improvement in parent-rated behavior and child-rated self-concept than those in the supportive therapy group, and these improvements persisted from posttreatment assessment at 4 months to the 5-month followup.
• Linden, Habib, and Radojevic, 1996: Subjects were 18 children, age 5 to 15 years with a "primary" diagnosis of ADHD which was based on unspecified measures (e.g., "family interview"). Subjects were referrals to an outpatient psychology clinic. Inclusion and exclusion criteria were not clearly defined. Nine subjects were assigned to EEG biofeedback treatment over 6 months, and nine subjects to a wait-list control group. Adjunctive treatment was controlled, and subjects received no medication or treatment other than that provided in the trial throughout the study. Measures were taken before and after the 6-month treatment program. Active treatment resulted in significantly greater improvement in IQ and attentiveness than did control treatment.
• MTA Cooperative Group, 1999: Five-hundred-seventy-nine children, age 7 to 9, were randomly assigned to one of four 14-month-long conditions: medication management (MedMgt) with monthly visits and supportive care, intensive behavioral treatment (Beh) (including home- and school-based behavior management, direct behavioral management therapy for the child delivered through a summer camp program), both treatments combined (Comb), or assessment and referral back to community service providers (CC). Children's outcomes in clinical and functional domains were assessed throughout the study, using rating scales and direct observational measures. Comb and MedMgt treatments were clinically and statistically superior to intensive Beh and routine CC in reducing core ADHD and oppositional-aggressive symptoms. Moreover, Comb and MedMgt approaches did not differ statistically or clinically from each other for ADHD or oppositional-aggressive symptoms. In several non-ADHD-symptom areas (parent-rated internalizing symptoms, social skills, consumer satisfaction, and reading), Comb provided modest incremental benefits over single-treatment approaches. MedMgt was superior to medication management provided in the community despite the fact that two-thirds of the CC subjects were treated with medication during the study period.

Overall, the studies in this category show a trend to general improvement over time regardless of treatment. MPH appears to reduce behavioral disturbance in ADHD children for as long as it is taken. However, the studies lack information on the reasons that so many children discontinue medication and provide little evidence for improvement in academic performance with stimulants, even though MPH treatment does appear to produce consistent behavior improvement. Lithium does not appear to be an effective alternative in subjects who do not respond to stimulants. A limitation of this category is that all but one study was restricted to school-age children. Few studies followed children for a time period equivalent to the duration children typically remain on these treatments in the clinical setting. There is little information on situation-specific outcome measures (e.g., outcomes measured at home and at school), and adverse effects are reported infrequently.

MPH vs. Placebo

Five RCTs compared MPH with placebo. These studies had very heterogeneous designs and incomplete reporting and provided contradictory results. The only study that was given a quality score >3 was in this group (Spencer, Wilens, Biederman et al., 1995). It was a randomized, double-blind, crossover comparison of titrated doses of MPH with placebo. The dose of MPH was titrated to approximately 1 mg/kg. This study included 25 patients out of 85 who had been referred. On MPH, 78 percent of patients showed end-of-treatment CGI scores <2 and 30 percent reduction in individual rating scale scores. Only 4 percent of the participants met these criteria on placebo. Of the four additional studies comparing the effect of MPH with placebo in adults (Evans, Vallano, and Pelham, 1994; Mattes, Boswell, and Oliver, 1984; Reimherr, Wender, Ebert et al., 1984; Wender, Reimherr, Wood et al., 1985), one had a single-patient design (n-of-1 design) and lasted for 1 week (Evans, Vallano, and Pelham, 1994); one study evaluated the interventions for 2 weeks but did not report the number of patients randomized to each arm (Wender, Reimherr, Wood et al., 1985); and another did not include any of the outcomes of interest (Reimherr, Wender, Ebert et al., 1984). The remaining study (Mattes, Boswell, and Oliver, 1984) included an evaluation of two groups of adults, one with evidence of residual attention-deficit disorder (ADD) with hyperactivity and the other without. In this study, no overall benefit from MPH was evident.

Additional information describing these studies can be found in Supplemental Evidence Tables G5-G8.

Studies Evaluating the Effect of Dextroamphetamine

• DeVeaugh-Geiss and Joseph, 1980: This was a single-subject study (a 20-year-old man) and a double-blind investigation of his response to DEX, secobarbital, and placebo. The authors noted a reduction of anxiety and motor activity while the subject was taking DEX.
• Matochik, Liebenauer, King et al., 1994: This study was primarily interested in the effects of MPH and DEX on cerebral glucose metabolism as measured by positron emission tomograph (PET) before and after treatment. Subjects were adults who met criteria for ADHD, met Utah criteria for ADHD in adulthood, had a definite childhood history of ADHD, had no psychiatric disorder, and were unmedicated prior to the study. Most had parents and/or children with ADHD. Medication dose was individually titrated and administered for 6 to 15 weeks. No differences in glucose metabolism with medication were observed. No direct comparison of MPH and DEX was conducted.

Studies Evaluating Other Interventions

• Conners, Levin, Sparrow et al., 1996: This 1-week trial found that nicotine at 7 mg delivered via transdermal patch yielded significant improvements in CGI ratings (56 percent much improved, 31 percent no change, and 12 percent unchanged or worse) and in self-ratings of vigor and concentration and improved performance on time estimation tasks, when compared with placebo.
• Ernst, Liebenauer, Jons et al., 1996: This trial found that selegiline treatment was not more effective than placebo.
• Wender, Reimherr, and Wood, 1981: This study examined the effects of pemoline vs. placebo. More than one-half the patients in the pemoline-treated group complained of moderate to severe adverse effects. Although the overall response to pemoline in the placebo and drug groups did not differ, adults with a childhood history of ADHD on parent reports showed a more favorable response than those without a childhood history of ADHD.
• Wilens, Biederman, Prince et al., 1996: This study, also described in the tricyclic antidepressants section, found that 68 percent (13/19) of the participants showed a positive response to desipramine (very much or much improved) plus at least a 30 percent reduction in rating-scale symptoms. No participants showed a favorable response to placebo. Two participants receiving desipramine dropped out, and 10 had doses lowered as a result of adverse effects. Favorable response was not related to the presence of lifetime comorbidity of anxiety or depressive disorders.
• Wood, Reimherr, and Wender, 1985: This was a crossover trial of phenylalanine that yielded a significant improvement in mood and mood ability over placebo, but the positive effects were lost within 3 months.

In conclusion, few RCTs are designed to evaluate the long-term effects of treatments for ADHD. Most of the studies available had incomplete reports, had small sample sizes, and were of short duration. The few studies evaluating MPH vs. placebo show contradictory results, but mostly in favor of MPH. The only study available with evaluation of the effect of DEX vs. placebo included few data of clinical relevance. Antidepressants may be effective in adults. Studies (one each) comparing pemoline, nicotine, or phenylalanine with placebo did not produce evidence in favor of these medications. No studies were designed to determine the proportion of adults with ADHD who will use, and benefit from, other interventions, particularly nonpharmacological.

Published Systematic Reviews

Three of the meta-analyses included in the review of reviews evaluated the effect of stimulants compared with placebo (Appendix C). Although these reviews had major methodological limitations, they showed consistently that stimulants improve behavior in children with diagnosis of ADHD in the short-term (Kavale, 1982; Ottenbacher and Cooper, 1983; Thurber and Walker, 1983).

Systematic Review Conducted at University of British Columbia

The report of the systematic review conducted at the University of British Columbia was assessed and obtained a quality score of 6 out of 7 using Oxman and Guyatt's methodological index (Oxman and Guyatt, 1991). This review was based on an extensive search of seven databases to August 1997, contact with pharmaceutical companies, and hand search of journals.

The review focused on RCTs comparing stimulants with placebo that provided data on children's general "ADHD behaviors" obtained with 11 teacher measures: the Abbreviated Conners Teacher Rating Scale; the Iowa Conners Teacher Rating Scale; the Attentional Problems Subscale of the Achenbach Child Behavior Checklist (Teacher Report Form [TRF]); the Child Attention Profile (CAP) Questionnaire; School Situations Questionnaire (SSQ); Revised Behavior Problem Checklist (RBPC); ADD-H Comprehensive Teacher Rating Scale (ACTeRS); Attention-Deficit Disorder Evaluation Scale (ADDES); Children's Attention and Adjustment Survey (CAAS); ADHD Rating Scale; and the Swanson, Nolan, and Pelham (SNAP). Nine behavior rating scales used to measure parent opinions were: Conners Parent Rating Scale; Conners, Loney, and Milich (CLAM) Questionnaire; Achenbach Child Behavior Checklist Attentional Problems score; Home Situations Questionnaire (HSQ); Revised Behavior Problem Checklist; Attention-Deficit Disorder Evaluation Scale; Children's Attention and Adjustment Survey; ADHD Rating Scale; and SNAP.

The authors standardized data that had been obtained with different tools to a common measure across the studies. Despite the fact that the authors acknowledged the limitations of the data available, they performed meta-analysis using DerSimonian and Laird's method. The review included 16 RCTs evaluating efficacy of stimulant treatment using teacher rating scales, 2 of which did not show advantage for the stimulants. Twelve RCTs included parent rating scales, one of which did not show advantage for the stimulants. When all the studies were combined, the overall results indicated that drug therapy was more effective than placebo. The results from the analysis of seven RCTs of MPH and three of DEX was possible only with the Conners Teacher Rating Scale and showed that both drugs were better than placebo. The only trial comparing pemoline vs. placebo also showed advantage for patients receiving pemoline.

In summary, the results of studies comparing stimulants with placebo indicate that stimulants offer benefit at least in the short term. The few data on long-term effects suggest that stimulants may improve behavioral disturbances. Little evidence shows long-term benefit of stimulant therapy on academic performance, and there is a paucity of data for long-term adverse effects.