A Randomised Open‐Label Comparison of Dexamphetamine and Methylphenidate in Children With Attention Deficit Hyperactivity Disorder (ADHD): Symptom Response, Adherence and Weight Impact Over 12 Months

ABSTRACT

Aim

Stimulant medications dexamphetamine and methylphenidate are well‐established for the treatment of Attention Deficit Hyperactivity Disorder (ADHD), but few studies directly compare their efficacy. We aimed to compare the therapeutic and adverse effects of immediate‐release formulations of dexamphetamine and methylphenidate during the initial treatment of children with ADHD.

Methods

This open‐label study was conducted in public and private specialist paediatric clinics in New South Wales, Australia, between 2016 and 2020. One hundred stimulant‐naïve children and adolescents with DSM‐5‐confirmed ADHD were randomised 1:1 to dexamphetamine or methylphenidate. ADHD diagnosis was based on detailed clinical assessment supported by parent and teacher DSM‐5‐based ratings. Dose titration followed a standardised 4‐week weight‐based protocol monitored with weekly school‐based IOWA Conners ratings, after which dose adjustment or switching was made as clinically indicated. Participants were followed for 12 months.

Results

Study participants (mean age 9.08 [SD 2.87] years; range 4.5–16.4 years; 73% male) showed significant ADHD symptom improvement on either medication at all time periods (p = 0.001), with no significant between‐group difference on repeated‐measures ANOVA. At 12 months, 62 participants remained on their allocated medication, with no apparent preference between methylphenidate and dexamphetamine (34/50 [68%] vs. 28/50 [56%], p = 0.15). Both medications were associated with weight loss which was greater on dexamphetamine at 3 months (−1.44 [SD 2.42] kg vs. −0.31 [SD 0.97] kg; p = 0.005).

Conclusions

Stimulant medications dexamphetamine and methylphenidate appear comparable in short‐term symptom response and 12‐month persistence after initial randomised allocation, but dexamphetamine may be associated with greater weight loss.

Trial Registration: Australian New Zealand Clinical Trials Registry number: ACTRN12616000569404

Key Points.

• What is already known on this topic

  • ○
    Dexamphetamine and methylphenidate are both highly effective first‐line medications for ADHD.
  • ○
    Appetite suppression, weight change, and sleep or behavioural adverse effects are important considerations when selecting stimulant medication.

• What this paper adds

  This study provides randomised 12‐month comparative data on adherence and anthropometric outcomes after initial allocation to dexamphetamine or methylphenidate.

  • ○When allocated at random and titrated using a standardised protocol, dexamphetamine was associated with greater weight loss.
  • ○Adverse effects leading to medication change were broadly similar between the two medications.
  • ○Most children did not require a change from their allocated medication during 12 months of follow‐up.

1. Introduction

Attention deficit hyperactivity disorder (ADHD) is a common neurodevelopmental disorder of childhood. They must fulfil diagnostic criteria to a level that is developmentally inappropriate, persistent, present across settings and associated with functional impairment. Management is multimodal and includes behavioural, educational and family‐based interventions; not all children require medication, but stimulant medication is commonly used when symptoms remain impairing despite these measures. Methylphenidate and dexamphetamine are both effective stimulant medications for children with ADHD, as documented in recent systematic reviews and network meta‐analyses [1, 2, 3]. However, direct long‐term comparisons remain limited. The Australian Evidence‐Based Clinical Practice Guideline for ADHD does not strongly elevate one stimulant over another [4], although the National Institute for Health and Care Excellence (NICE) guidelines favour methylphenidate for the initial stimulant [5]. In clinical practice, response heterogeneity means that more than one stimulant may need to be trialled [6, 7].

Appetite suppression and weight loss are amongst the most clinically important adverse effects of stimulant medication, with secondary effects on growth [4, 8, 9]. Observational studies suggest that these effects may be more pronounced with amphetamine preparations, whilst other adverse effects, such as insomnia, irritability and mood change, also influence treatment choice. Existing literature therefore suggests clinically relevant tolerability differences, but the magnitude and persistence of such differences remain uncertain in routine practice [10, 11]. More recent evidence on growth and cardiovascular safety reinforces the importance of prospective height, weight, blood pressure and pulse monitoring during stimulant treatment [7, 12, 13].

There are a few head‐to‐head randomised studies comparing methylphenidate and dexamphetamine directly and those that do exist have generally been of short duration. In 1997, Efron et al. reported a short‐term double‐blind crossover trial in which 125 children took either methylphenidate for 2 weeks followed by dexamphetamine for 2 weeks or vice versa [14, 15]. They found methylphenidate had a greater therapeutic effect and fewer non‐responders. Dexamphetamine was associated with more severe negative emotional symptoms and more insomnia. Both were associated with appetite suppression but weight change was not reported. A continuation study over the following 6–9 months showed no significant difference in efficacy; however, randomisation was not continued, and each participant remained on their preferred stimulant [16]. By contrast a double‐blind crossover study by Elia et al. that used each drug in a range of doses found no significant differences in efficacy, or in number or severity of side effects. Ongoing management was left to personal preference, with more than half of those who had been pre‐treated opting for a change of stimulant. However, continuing the randomised allocation over a longer term would be better suited to evaluating adherence, switching patterns and anthropometric change over time.

The aims of this study were (1) to compare the effects of dexamphetamine and methylphenidate on symptoms of ADHD and ODD during the 4‐week titration of immediate‐release formulations, (2) to compare 12‐month persistence on the allocated medication, and (3) to compare adverse effects and reasons for medication change. Review time points after initial titration were 3, 6 and 12 months.

2. Methods

This study was conducted in a paediatric clinic in New South Wales, Australia, run by a single academic paediatrician specialising in ADHD. Between 2016 and 2020, parents of psychotropic‐naïve children aged 4–17 years with ADHD were invited to enrol. All participants had persistent functional impairment across settings despite behavioural interventions. Diagnosis was based on detailed clinical assessment, supported by DSM‐5‐based parent and teacher ratings using the Nepean ADHD Project Symptom Scale (NAPSS). Co‐morbidities were assessed clinically rather than with a standardised research battery, and children were excluded if ADHD was not the primary condition requiring treatment, if there was intellectual disability, a medical contraindication to stimulant medication, previous psychotropic treatment or other complexity requiring a different treatment priority. ODD was not an exclusion.

Study participants (n = 100) were randomised 1:1 to dexamphetamine or methylphenidate using a predetermined schedule with a block size of four (randomised using a dice throw). Allocation was concealed in a sealed envelope until enrolment. Immediate‐release tablets of dexamphetamine 5 mg or methylphenidate 10 mg were titrated as tolerated according to a standardised weight‐based 4‐week protocol [17] (Table S1), with assumed clinical equivalence of dexamphetamine 5 mg and methylphenidate 10 mg. Figure 1 illustrates symptom change during this protocolised titration period.

FIGURE 1.

Morning symptom scores during the protocolised 4‐week weight‐based titration. Both medications were associated with significant symptom reductions at each week of titration (p = 0.008 to p < 0.001; Table 3).

Medication efficacy during titration was monitored with the IOWA Conners rating scale completed by the child's teacher at baseline and weekly with each dose change [18]. Ratings were collected for behaviour in the morning and afternoon for each dose of medication.

After the initial dose titration, adjustments to medication were made as clinically indicated or according to family preference. Reasons for change were recorded, including appetite suppression, adverse behavioural effects, sleep difficulty or other tolerability concerns. Changing to a modified‐release formulation of the allocated stimulant, or changing from dexamphetamine to lisdexamfetamine, was regarded as remaining on the allocated medication.

Participants were reviewed after 3, 6 and 12 months of treatment, or more frequently if clinically indicated. Parent and child perceptions of benefit and adverse effects were recorded at clinical review, and side effects were monitored using a standardised adverse‐effect rating scale [19]. Measurements of blood pressure, height and weight were taken at every visit to the nearest 1 mm and 100 g respectively, using a wall‐mounted stadiometer and electronic scales. Measurements were made with shoes and outdoor clothing removed and without reference to previous measurements.

Baseline comparability between groups was assessed for age, sex, height, weight and BMI. The sample size was estimated based on the IOWA Conners scales using the standard approximation n = 16/f ² for independent‐samples comparisons (80% power; two‐sided α = 0.05). Assuming a standard deviation of approximately 4 points and a clinically meaningful between‐group difference of 2.5 points, approximately 41 participants per group would be required; the recruited sample of 50 participants per group was therefore considered adequate. However, incomplete weekly data reduced the sample available for analysis to around 25 per group. Continuous variables were compared using t‐tests and categorical variables using the chi‐square test. Repeated‐measures ANOVA was used to analyse efficacy during titration and weight and height over 12 months. The study was not matched or stratified by ADHD subtype, co‐morbidity, or baseline symptom severity beyond the clinical assessment undertaken at enrolment. No formal intention‐to‐treat analysis was undertaken.

Ethics approval was granted by the Nepean Blue Mountains Human Research Ethics Committee (15/37‐HREC/15/NEPEAN). Written informed consent was obtained in all cases, and children were invited to assent. Prescribing in this study conformed to the NSW Ministry of Health criteria [20].

3. Results

From 2016 to 2020, 212 children were assessed for the study, of whom 49 were excluded, and 63 declined participation, mainly because parental drug preference precluded randomisation (23 for methylphenidate, 22 for lisdexamfetamine and 2 for dexamphetamine) (Figure 2). Of those excluded, 9 had more clinically complex presentations, 23 had previously taken psychotropic medication and 17 either did not meet full diagnostic criteria in two settings or had incomplete data (including 4 affected by telehealth consultations during COVID‐19).

FIGURE 2.

Study flow chart. DEX = dexamphetamine; MPH = methylphenidate; NS = non‐stimulant treatment (guanfacine).

Of the 100 participants randomised, the majority were male (73% vs. 27%, p = 0.001). Those eligible but not included were not significantly different by age (9.45 [SD 3.38] vs. 9.08 [SD 2.87] years, p = 0.46) or sex (71% male, p = 0.83). There was no significant difference in gender balance or baseline age, height, weight or BMI between the participants randomised to dexamphetamine or methylphenidate (Table 1).

TABLE 1.

Participant characteristics of 100 children randomised to dexamphetamine or methylphenidate.

Categorical variables	Dexamphetamine (n = 50)	Methylphenidate (n = 50)	Test statistic	p (two‐sided)
	N	N	Chi‐square
Male	36	37	0.05	0.82
Female	14	13	0.05	0.82
Private consulting rooms	36	27	3.21	0.07
Public hospital clinic	14	23	3.21	0.07

Continuous variables mean (SD)	Dexamphetamine (n = 50)		Methylphenidate (n = 50)		Test statistic	p (two‐sided)
Data categories (mean, SD, t‐statistic)	Mean	SD	Mean	SD	t‐test
Baseline age at recruitment (years)	9.11	2.62	9.07	3.15	−0.08	0.94
Baseline height (cm)	135.2	16.9	135.2	17.8	0.02	0.98
Baseline weight (kg)	34.8	12.5	34.7	15.1	−0.03	0.98
Baseline BMI (kg/m2)	18.5	3.5	18.2	3.8	−0.47	0.64

At baseline, 50 participants were allocated to each medication. By 12 months, 62 remained on their allocated medication: 34/50 (68%) in the methylphenidate group and 28/50 (56%) in the dexamphetamine group. Eighteen participants were lost to follow‐up. Of the remainder, nine changed from dexamphetamine to methylphenidate, four changed from methylphenidate to dexamphetamine, five ceased medication (four from the dexamphetamine arm) and two changed from methylphenidate to non‐stimulant treatment (guanfacine). Although more children in the methylphenidate group remained on their allocated medication at 12 months, this difference was not statistically significant (chi‐square = 2.0, p = 0.15).

Clinically significant adverse effects leading to medication change were similar across groups and were predominantly appetite suppression, behavioural change and sleep difficulty (Table 2).

TABLE 2.

Clinically significant adverse effects leading to medication change.

Change	N	Timing (months)	Reason
DEX to MPH	5	3	Appetite suppression; rebound ADHD symptoms; emotional upset
	2	6	Appetite suppression; insomnia
	1	12	Appetite suppression; aggression
MPH to DEX	1	6	Anxiety
	3	12	Weight loss; behavioural change
MPH to NS	1	3	Motor tics
	1	6	Emotional instability

Note: Multiple reasons could be recorded for a single child, and categories were simplified from parent report and clinical review. Not included in the table: one child who changed from DEX to MPH immediately after randomisation.

Abbreviations: DEX = dexamphetamine; MPH = methylphenidate; NS = non‐stimulant treatment (guanfacine).

There was a significant reduction in symptom scores for inattention/overactivity (IO) and oppositional/defiant (OD) behaviour for both medications compared with baseline at all time points (Table 3). Repeated‐measures ANOVA demonstrated no significant between‐group differences in efficacy on any IOWA Conners item (Table 4).

TABLE 3.

Symptom scores for inattention/overactivity and oppositional/defiant behaviour during the 4‐week titration, with paired comparisons to baseline for participants contributing data at each week.

	N	Inattention/overactivity		Oppositional/defiant
		AM	PM	AM	PM
		Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)
DEX baseline (range of means) a	38	10.1 (3.9) (9.7–10.8)	10.9 (4.0) (10.5–11.4)	6.1 (4.9) (5.5–6.6)	7.4 (5.3) (7.0–7.9)
DEX week 1	33–34	7.7 (3.3)	8.5 (3.8)	4.5 (4.0)	6.0 (4.9)
DEX week 2	34	5.7 (3.0)	6.5 (3.5)	3.6 (4.2)	4.7 (4.3)
DEX week 3	29–30	4.1 (3.0)	4.8 (3.3)	2.5 (3.9)	3.4 (4.2)
DEX week 4	26	3.2 (2.3)	3.7 (2.9)	2.0 (2.4)	2.8 (3.4)
MPH baseline (range of means) a	38	9.6 (4.0) (9.6–10.0)	10.7 (4.0) (10.6–11.0)	5.2 (4.8) (5.2–5.8)	5.9 (5.0) (5.9–6.8)
MPH week 1	37	6.2 (4.2)	7.5 (4.6)	3.1 (3.4)	4.1 (4.4)
MPH week 2	34–35	5.3 (3.4)	6.8 (4.1)	3.1 (4.0)	3.6 (4.4)
MPH week 3	29–31	3.6 (2.8)	4.9 (2.6)	2.2 (3.8)	2.9 (3.6)
MPH week 4	23–25	2.0 (2.4)	3.3 (3.1)	1.2 (2.7)	2.2 (4.0)

^a Statistical analysis used paired t‐tests within each medication group. Because paired analyses were used, baseline N varies across weeks and represents only participants with both baseline and follow‐up data for the relevant paired comparison.

TABLE 4.

Repeated‐measures ANOVA comparing IOWA Conners item scores between dexamphetamine and methylphenidate over the 4‐week titration period.

	df	Sum sq.	Mean sq.	F	p
Inattention/overactivity
Fidgeting AM	3.00	3.00	1.00	1.80	0.15
Fidgeting PM	3.12	2.54	0.81	1.50	0.22
Hums/odd noises AM	2.50	2.00	0.80	1.55	0.21
Hums/odd noises PM	3.07	1.44	0.47	1.08	0.36
Excitable/impulsive AM	2.62	0.91	0.95	0.58	0.61
Excitable/impulsive PM	2.82	0.96	0.34	0.49	0.68
Inattentive/distracted AM	2.99	1.69	0.57	0.95	0.42
Inattentive/distracted PM	3.17	0.93	0.29	0.61	0.62
Fails to finish AM	2.92	2.04	0.70	1.06	0.37
Fails to finish PM	3.07	2.03	0.66	1.38	0.25
Oppositional/defiant
Quarrelsome AM	2.90	2.21	0.76	1.48	0.23
Quarrelsome PM	2.43	1.13	0.47	0.70	0.53
Acts ‘smart’ AM	2.76	2.04	0.66	1.67	0.18
Acts ‘smart’ PM	2.69	2.23	0.91	1.23	0.30
Temper outbursts AM	2.47	1.35	0.55	0.74	0.51
Temper outbursts PM	2.65	1.63	0.61	0.71	0.53
Defiant AM	2.97	1.50	0.50	0.94	0.42
Defiant PM	2.76	1.51	0.55	0.80	0.49
Uncooperative AM	2.97	3.17	1.07	1.52	0.21
Uncooperative PM	3.09	3.63	1.17	1.86	0.14

Abbreviations: df = degrees of freedom; Mean sq. = mean square; Sum sq. = sum of squares.

Salient ADHD symptoms were measured in the morning and afternoon using the IOWA Conners rating scale at baseline and at the end of the week on each dose of the titration. Repeated‐measures ANOVA was used to determine whether there were significant between‐group differences on individual items. No statistically significant differences were found between dexamphetamine and methylphenidate on these item‐level measures over the five time points (Table 4).

A one‐way repeated‐measures ANOVA was conducted to examine the relationship between allocated medication and weight across four time points (baseline, 3 months, 6 months and 12 months) (Figure 3). Mauchly's test indicated violation of sphericity (X ²[5] = 39.14, p = 0.03); therefore, Greenhouse–Geisser corrected tests are reported (ε = 0.66). Weight differed significantly by allocated medication over time, F(1.98, 98.83) = 6.97, p = 0.03, with less weight loss on methylphenidate than dexamphetamine. These analyses were based on completers only. Participants who changed stimulant medication, ceased medication, had missing data at time points or were lost to follow‐up were excluded from subsequent analysis (methylphenidate n = 20; dexamphetamine n = 28). There were no significant baseline differences between those who completed 12 months on the study medication and those who did not. Weight loss was greatest at 3 months (1.44 kg [SD 2.42] on dexamphetamine vs. 0.31 kg [SD 0.97] on methylphenidate; p = 0.005). At 12 months, those allocated to methylphenidate had gained 1.26 kg (SD 2.64) and those allocated to dexamphetamine had lost 0.84 kg (SD 3.51) from baseline (p = 0.006).

FIGURE 3.

One‐way repeated‐measures ANOVA of weight over time in children treated for 12 months with methylphenidate (n = 30) or dexamphetamine (n = 22). The between‐group effect on weight was statistically significant (Greenhouse–Geisser corrected F[1.98, 98.83] = 6.97, p = 0.002).

The analysis was repeated for height across the same four time points. Mauchly's test again indicated violation of sphericity (X ²[5]=66.76, p = 0.001); therefore, Greenhouse–Geisser corrected tests are reported (ε = 0.54). Height was not significantly affected by allocated medication over time, F(1.61, 77.18) = 2.27, p = 0.083 (Figure 4). Participants who changed stimulant medication, ceased medication, had missing data during time points or were lost to follow‐up were excluded (methylphenidate n = 21; dexamphetamine n = 29). There was no significant difference in height gain at 12 months between groups (4.9 cm [SD 2.7] for methylphenidate vs. 4.4 cm [SD 1.9] for dexamphetamine; p = 0.38).

FIGURE 4.

One‐way repeated‐measures ANOVA of height over time in children treated for 12 months with methylphenidate (n = 29) or dexamphetamine (n = 21). The between‐group effect on height was not statistically significant (Greenhouse–Geisser corrected F[1.61, 77.18] = 2.27, p = 0.083).

4. Discussion

Both dexamphetamine and methylphenidate were associated with substantial improvement in IOWA Conners scores during titration, with no evidence of a between‐group difference in efficacy. This is consistent with previous direct‐comparison studies, whilst recognising that incomplete weekly data reduced sensitivity to detect small differences [14, 21, 22]. This finding is also consistent with contemporary systematic reviews showing that stimulant medications remain amongst the best‐supported pharmacological treatments for ADHD in school‐aged children, whilst also noting that direct head‐to‐head evidence between individual stimulant classes remains limited [23, 24]. That 62% of the cohort were still using their allocated treatment at 12 months, with no significant difference in attrition between dexamphetamine and methylphenidate, would appear to indicate a level of parental satisfaction with both stimulant medications.

Our finding of comparable efficacy for each stimulant replicates the findings of Elia et al. but not those of Efron et al., as they found methylphenidate significantly more effective. However, like the study of Elia et al., we used a range of doses, whilst Efron et al. assumed a 2:1 equivalence and used fixed doses of 0.3 mg/kg methylphenidate or 0.15 mg/kg dexamphetamine given twice daily. The initial advantage to methylphenidate was lost in the open‐label continuation phase, which might suggest that where dose optimisation is permitted by flexible dose adjustment, both medications are of similar efficacy.

Our finding of greater weight loss with dexamphetamine accords with earlier observational work and with prior concerns regarding stimulant tolerability [11, 25]. To our knowledge, this is the first randomised study to demonstrate this difference in weight trajectory after 12 months on the allocated medication. However, our data should be interpreted as a clinically important signal rather than a definitive dose‐adjusted causal estimate, because anthropometric analyses were not framed as multiplicity‐corrected confirmatory analyses. The finding should also be interpreted in the context of broader growth literature: long‐term methylphenidate exposure has been associated with small reductions in height and weight in meta‐analysis, whereas the 2‐year ADDUCE study did not find evidence supporting clinically important growth suppression with methylphenidate; together, these data support careful longitudinal monitoring rather than assuming uniform growth effects across stimulant types [13, 26].

The main strengths of this study are the randomised treatment allocation, the duration of follow‐up and the consistency of management across the cohort. Because all participants were treated within the practice of a single clinician, the prescribing approach, follow‐up and clinical decision‐making were relatively consistent. In addition, symptom monitoring during titration relied on independent external observers (teacher ratings obtained in the school setting). We also used rigorous inclusion criteria: all participants were naïve to any psychometric medication and had no intellectual disability or major co‐morbidities. Only two recruited children were under 5 years of age, so this subgroup represented a very small proportion of the sample. Nevertheless, their inclusion may modestly limit generalisability to settings that apply higher age thresholds.

Important limitations should be acknowledged. The open‐label, single‐clinician design introduces the possibility of clinician and observer bias in dose adjustment, switching decisions and attribution of adverse effects. The study also relied on pragmatic clinical assessment rather than a formal semi‐structured diagnostic interview, and co‐morbidities were assessed clinically rather than with a standardised research battery. ASD, anxiety and eating‐related problems were considered at study entry, but they were not systematically recorded as stratification variables.

Weekly IOWA Conners ratings were complete for only approximately 50% of the cohort. Furthermore, the IOWA Conners is a brief treatment‐monitoring measure rather than a comprehensive clinical endpoint. No formal intention‐to‐treat analysis was undertaken. The study was not powered specifically to detect modest differences in remaining on allocated medication at 12 months, and the effective sample for some efficacy comparisons was substantially smaller than the recruited sample. The findings should therefore be interpreted cautiously.

Twelve months remains a limited period for evaluating growth outcomes. Height velocity over this interval is strongly influenced by age and pubertal stage, which likely contributed to extraneous variability in the comparison. Future studies should therefore include pubertal staging, dose exposure, treatment interruptions, appetite measures and baseline BMI or weight category, as these factors may modify stimulant‐associated changes in weight and height.

Overall, the findings are most applicable to pragmatic specialist prescribing contexts in which either methylphenidate or dexamphetamine may be selected as an initial stimulant. Head‐to‐head comparison studies are important for providing the scientific basis for clinical decision‐making and for guideline recommendations with regard to recommending one stimulant over the other.

5. Conclusion

This randomised 12‐month study suggests that both dexamphetamine and methylphenidate provide similar short‐term symptom improvement and show comparable overall adherence, suggesting that either may be recommended for treatment efficacy and general tolerability. However, dexamphetamine may be associated with greater weight loss.

These findings relate to a small cohort and should be interpreted cautiously. Nonetheless, they align with current Australian guidelines that support the use of either medication for efficacy, whilst favouring methylphenidate when weight loss is a particular concern.

Funding

The authors have nothing to report.

Ethics Statement

Ethics approval was granted by the Nepean Blue Mountains Human Research Ethics Committee (15/37‐HREC/15/NEPEAN; 2019/ETH08868).

Conflicts of Interest

A.P. discloses personal fees from Medcast and Novartis, and non‐financial support from Shire/Takeda, outside the submitted work; and book royalties from Disruptive Publishing (ADHD Made Simple). The other authors declare no conflicts of interest.

Supporting information

Data S1: jpc70487‐sup‐0001‐Supinfo.docx.

Table S1: Standardised initial 4‐week weight‐based titration protocol [17].

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.