Impairments of attention and speed of information processing are among the most common persisting cognitive changes evident following traumatic brain injury (TBI). Neuropathological changes following TBI involving subcortical white matter and predominantly frontotemporal cortical distribution, potentially disrupt attentional neural networks, including ascending catecholaminergic and serotonergic pathways.1 Altered catecholamine levels have been associated with poorer outcomes following TBI.2

While there is strong evidence of reduced information processing speed, evidence for impairments of sustained attention, working memory, divided attention and strategic control over attention has been mixed.3 Willmott and colleagues4 found evidence of slowed information processing, reduced working memory and more errorful performances on tasks with increasing requirements for strategic control over attention in TBI participants relative to controls.

The use of methylphenidate to enhance attention following TBI has been proposed as “rational pharmacotherapy”.5 Methylphenidate, a central nervous system stimulant, increases release and blocks reuptake of dopamine and noradrenaline, resulting in increased synaptic and extracellular concentrations.6 Actions at D1 and α₂-adrenergic receptors have been linked to the facilitative effects of methylphenidate on prefrontal cortical modulation of attention and working memory.7

Systematic reviews indicate sufficient evidence to support guidelines for use of methylphenidate in treating deficits of attention and speed of processing following TBI but methodological limitations of studies have prevented development of treatment standards.8 Findings from previous class II or III clinical studies with methylphenidate in those with chronic injuries have been mixed.9^–12 Whyte et al have conducted two well designed randomised controlled trials in patients with chronic TBI demonstrating improvement in response speed, caregiver ratings and on-task behaviour with methylphenidate.13 14

By reducing the impact of symptomatic attentional deficits, methylphenidate could enable individuals with TBI to benefit more from concurrent therapy while still being involved in intensive rehabilitation. The few previous studies conducted in the early rehabilitation phase have, however, been confounded by methodological limitations, including lack of blinding,15 16 difficulty discriminating drug effects from those attributable to spontaneous recovery,16 17 failure to evaluate performance concurrently with methylphenidate administration18 and small sample sizes.

Using a randomised, double blind, placebo controlled, crossover design in inpatient rehabilitation TBI participants, the present study aimed to investigate the impact of methylphenidate administration versus placebo on:

• performance on measures of different aspects of attention (speed, working memory, strategic control and sustained attention to response),

• therapists’ ratings of attentional behaviour during therapy.

It was hypothesised that methylphenidate would enhance performance on tests of information processing speed, working memory, strategic control and sustained attention, and therapists’ ratings of attentional behaviour, compared with placebo. We also aimed to identify predictors of response to methylphenidate, hypothesising that injury severity and baseline speed would predict response to methylphenidate.

METHODS

Participants

Consecutive inpatient admissions to Epworth Rehabilitation Centre, Melbourne, Australia, from March 2004 to September 2006 with a recent history of moderate to severe TBI were assessed for eligibility (fig 1). Participants were aged between 16 and 60 years, had sufficient understanding of English, adequate physical and cognitive abilities to undertake the tasks and demonstrated attentional impairment at the baseline assessment. Exclusion criteria included any report by participants of previous significant neurological history, treatment with methylphenidate for attentional disturbance, treatment for drug or alcohol dependence or current treatment with psychoactive medications. Power calculations, based on previous data,13 indicated that a sample size of 35 would give 80% power at the 0.05 level to detect a drug intervention effect size of 0.6. A target sample size of 40 was chosen.

There were no statistically significant differences between those who agreed and those who declined to participate in terms of gender (χ² = 1.657, df = 1, p = 0.198), age (t = −1.942, df = 68, p = 0.056), socioeconomic status (t = 1.699, df = 66, p = 0.094), duration of post-traumatic amnesia (PTA) (t = 0.251, df = 66, p = 0.802) or worst Glasgow Coma Scale (GCS) score (t = −1.992, df = 56, p = 0.051), although there was a trend for younger participants with more severe injuries to agree to participate. TBI sample demographics are outlined in table 1. While the mean time since injury was 68 days, the median was 47 days, and more than 80% of the sample was in the first 3 months post-injury.

Measures

Ruff 2 and 7 Selective Attention Test (2&7)20

In the automatic condition, the numbers 2 and 7 to be cancelled are embedded among letters, so processing is based on single step retrieval of information from long term memory whereas in the controlled condition participants must search for these numbers embedded among other numbers, thereby requiring strategic selection and working memory.20 A speed raw score was calculated for the automatic (2&7 ASRS) and controlled (2&7 CSRS) conditions, respectively.

Selective Attention Task (SAT)21

This reaction time task has two conditions: Simple (SSAT) and Complex (CSAT). On the CSAT, working memory demands are increased as participants are required to retain and rehearse an additional verbal rule. Measures included mean reaction time (RT) for correct responses for both conditions, and errors and misses for CSAT.

Four Choice Reaction Time Task (4CRT)22

A single arrow appears on the screen and two variations in the task resulted in four conditions (DC, DI, SC, SI). Identification: the task variable for this processing stage was signal similarity. Response selection: this stage was manipulated through stimulus–response compatibility. Mean RT for correct responses for each condition and errors for the DI condition were recorded.

Sustained Attention to Response Task (SART)23

Participants pressed the space bar each time a number appeared on the screen, except when the number 3 (target) appeared. This task was employed to evaluate ability to sustain attentional control across task duration. Errors to non-targets (non-3 errors) and errors to targets (3 errors) were recorded.

Symbol Digit Modalities Test (SDMT)24

Number of items completed in 90 s was recorded.

Letter Number Sequencing Task (LNS)25

Scaled score (LNS SS) was a measure of working memory.

Wechsler Test of Adult Reading (WTAR)26

Used to estimate premorbid IQ (FSIQ).

Rating Scale of Attentional Behaviour (RSAB)27

Side Effects Questionnaire

Adapted with permission from the Moss Rehabilitation Research Institute Side Effects Monitoring Form.

Pulse and blood pressure

Diastolic and systolic, mean arterial pressure and pulse were recorded.

Procedure

Approval was obtained from Monash University and Epworth HealthCare ethics committees, and participants provided informed consent. The design was a randomised, crossover, double blind, placebo controlled trial. Following baseline assessment 3 days previously, participants were seen for six sessions (Monday, Wednesday, Friday), over a 2 week period. The sessions were in three blocks: days 1 and 2; days 3 and 4; days 5 and 6. One session of each block was assigned as methylphenidate, the other as placebo. Participants were allocated by simple randomisation to an administration sequence according to a random number table by independent pharmacists who dispensed the medication. At no time during recruitment or data collection was the neuropsychologist, treating team or participant aware of allocation. Methylphenidate (Ritalin) was administered at a dose of 0.3 mg/kg twice daily, rounded to the nearest 2.5 mg, at 08:00 and 12:00, concealed in opaque gelatin capsules. This resulted in four possible dosages assigned as follows: 15 mg twice daily (n = 11); 20 mg twice daily (n = 21); 25 mg twice daily (n = 7); and 30 mg twice daily (n = 1). Lactose in identical capsules served as placebo. Participants refrained from caffeine and smoking for 2 h prior to testing.

Participants completed neuropsychological measures, a Side Effects Questionnaire, drug allocation guess and had their pulse and blood pressure measured 90–120 min after capsule administration. Test administration order was constant within, but randomised across, participants. Participants were rated on the RSAB in daily physiotherapy sessions.

Data analysis

Primary outcome measures were performances on attentional tests. Analysis was done by intention to treat. Results from the 3 methylphenidate days and the 3 placebo days were averaged, resulting in one score for each variable for each drug condition. Missed sessions (randomised methylphenidate = 3, randomised placebo = 3) were not made up, and results averaged across the two remaining sessions. Only those test variables that differentiated TBI and control subjects in the baseline assessment study4 were evaluated in the present drug study. Planned comparisons were performed on the primary outcome measures. Parametric analyses were applied to the RT data. RT analyses were performed using mean and median reaction times with the same results. For brevity only the means are reported. The error data distributions did not meet parametric assumptions, and were analysed using non-parametric statistics. Bonferroni corrections were made for multiple comparisons. Cohen’s d effect sizes are documented. Two way within subjects ANOVAs were utilised to investigate whether methylphenidate exerted differential effects on speed and strategic control of attention on three tasks (2&7, SAT and 4CRT). Secondary outcome measures were scores on the RSAB. To determine predictors of response to methylphenidate, mean differences scores between the methylphenidate and placebo conditions for the four neuropsychological test measures with greatest drug effect sizes were transformed into z scores. An average z score (drug response), representing response to methylphenidate, was then calculated. Pearson correlations investigated associations between this variable and injury severity and baseline performance. Multiple regression using the enter method was conducted to predict drug response from baseline SDMT and worst GCS.

RESULTS

No participant was withdrawn because of adverse events. A full report of the safety data from this trial will be published separately.

Blinding

Separate Pearson’s χ² tests for each session revealed no relationship between actual randomisation (methylphenidate/placebo) and randomisation guess for sessions 1, 2, 3, 5 or 6 (p>0.2 for all), with guesses approximating chance. For session 4 this association was significant (χ² = 10.00, df = 1, p<0.01), with 75% of respondents in each randomisation category correctly guessing allocation.

Attentional task performance

A corrected α level of 0.008 was applied for multiple drug versus placebo comparisons on paired t tests for the computerised RT tasks. A paired t test revealed that methylphenidate significantly facilitated speed on the automatic (2&7ASRS) but not the controlled (2&7 CSRS) condition of the 2&7 Test (table 2). This effect was evident in a significant repeated measures ANOVA randomisation by condition interaction F(1,39) = 8.38 (p<0.01). On the SAT, methylphenidate resulted in significantly faster speed of information processing on the simple (SSAT) but not the complex (CSAT) condition. Repeated measures ANOVA revealed a marginally significant main effect of randomisation F(1,39) = 4.24 (p = 0.046), a significant effect of condition F(1,39) = 110.16 (p<0.001) and a non-significant interaction F(1,39) = 1.09 (p = 0.302). A similar pattern was observed on the 4CRT task, with significantly faster RTs with methylphenidate compared with placebo on two of the four conditions (DC and SC); however, on the more complex conditions (DI and SI) there was no significant difference. All three ANOVA main effects were significant: randomisation F(1,38) = 10.07 (p<0.01), similarity F(1,38) = 17.86 (p<0.001) and compatibility F(1,38) = 41.52 (p<0.001). The randomisation by compatibility interaction approached significance F(1,38) = 3.75 (p = 0.060), indicating that methylphenidate was having a greater effect on processing speed in the compatible than in the incompatible condition. All other 4CRT interactions were not significant (p>0.05). It should be noted that all comparisons indicated improvement with methylphenidate but for the complex conditions the effect was not significant. For the CSAT RT comparison, the effect size was 0.14. For an effect of that size, the statistical power is 0.134 at α = 0.05, indicating that the sample size was not sufficient to detect possible existing small differences between drug conditions on complex tasks. Methylphenidate also resulted in significantly faster speed of information processing on the SDMT (table 2).

Wilcoxon signed ranks tests for accuracy data are provided in table 3. There was no significant reduction in errors or misses on the CSAT or DI condition of the 4CRT with methylphenidate compared with placebo. Similarly, there was no reduction in errors to targets (3 errors) or errors to non-targets (non-3 errors) on the SART with methylphenidate. It is noteworthy, however, that there was also no significant increase in errors with methylphenidate. Methylphenidate had no significant effect on working memory performance on the LNS task.

Prediction of response to methylphenidate

Pearson correlations revealed that slower baseline information processing speed (SDMT) was significantly correlated with lower GCS scores (r = 0.377, n = 33, p = 0.015) and longer PTA duration (r = −0.545, n = 33, p = 0.001). SDMT correlated significantly with drug response (r = −0.509, n = 33, p = 0.001), as did worst GCS (r = −0.421, n = 33, p = 0.007). There was, however, no significant correlation between PTA duration and drug response (r = 0.272, n = 33, p = 0.063). Regression analyses were used to predict drug response. Using the enter method, a significant model emerged, adjusted R square = 0.320, p = 0.003. Worst GCS (β = −0.267, t = −1.642, p = 0.111) alone did not significantly contribute to the prediction but baseline SDMT was a significant predictor (β = −0.408, t = −2.510, p = 0.018).

Rating scales

The most commonly reported attentional difficulties rated by physiotherapists were: been unable to pay attention to more than one thing at a time; tired easily; seemed lethargic; been easily distracted; needs prompting to get on with things. Although all items on the RSAB indicated improvement with methylphenidate, Wilcoxon signed rank tests revealed that no differences between ratings made on methylphenidate versus placebo reached statistical significance (table 4).

DISCUSSION

Results from this study demonstrated that methylphenidate had a robust effect in improving speed of information processing. Importantly, this faster information processing speed did not come at the cost of reduced accuracy. Regression analyses indicated a relationship between injury severity and response to methylphenidate, mediated by baseline processing speed. Those with greater injury severity, as measured by worst GCS, and slower baseline information processing speed, demonstrated a greater drug response. When greater strategic control over attention was required, however, due to increased working memory or response selection demands, methylphenidate was less effective. Participants’ ability to guess randomisation did not reliably improve across the six sessions, indicating successful blinding.

The present study is the first well controlled study, and the largest ever, to demonstrate the efficacy of methylphenidate in improving processing speed in the inpatient rehabilitation phase. The fact that slowness in processing speed is one of the most common and debilitating early TBI sequelae across the spectrum of injury severity, and a strong predictor of functional outcome,28 highlights the importance of facilitating cognitive speed in the early stages post-injury.

These results are consistent with those reported in individuals with chronic TBI by Whyte and colleagues14 who found that methylphenidate reduced “initial speed” but did not influence performance on factors encompassing dual task decrement in RT, working memory demands, errors on the SART or staff ratings on the RSAB.

The finding that methylphenidate did not significantly facilitate speed of information processing or accuracy on more complex task conditions requiring greater use of working memory and strategic control over attention was somewhat surprising. Working memory demands are increased with the CSAT as participants are required to retain and rehearse an additional verbal rule. The role of catecholaminergic augmentation with agents such as methylphenidate in the activation and modulation of working memory function in the prefrontal cortex is well established.29 However, in the present study, performance on tasks with greater working memory demands was not facilitated with methylphenidate. Potential confounds included the fact that methylphenidate acts on both dopaminergic and noradrenergic systems. Even within a single neurotransmitter system, variable response to catecholaminergic agents has been identified. Thus agents such as methylphenidate which increase the endogenous release of norepinephrine may have opposing effects on working memory.30 A power calculation on the CSAT RT also suggested that the sample size of 40 may not have been sufficient to detect possible existing small effects on these more complex tasks. Future studies should therefore aim for larger samples.

Errors on the SART are thought to represent a drift from controlled processing into more automatic responding in TBI.23 The lack of significant drug effect on the SART suggests that methylphenidate does not facilitate the strategic control of attention in assisting TBI participants to maintain their response set.

Although attentional ratings on the RSAB suggested that attentional behaviour did improve on days when methylphenidate was administered, the comparison with ratings on placebo did not reach statistical significance. This lack of significance may reflect the fact that therapists were not carrying out observations in a context in which it was possible to focus systematically on patients’ speed of information processing. Future studies should focus everyday attentional observations on behaviours on which methylphenidate is expected to have an effect, in a controlled environment.

As there have been very few studies of methylphenidate in the inpatient rehabilitation phase post-injury, we aimed to minimise the incidence of adverse events by excluding those in whom methylphenidate might pose a risk. The resultant number of ineligible participants is a possible limitation of the present study. Given the relative safety of the drug in this cohort, future studies should aim to recruit from a wider TBI group.

In summary, this study has clearly demonstrated the efficacy of methylphenidate in facilitating speed of information processing in TBI rehabilitation inpatients. This resulted in a non-significant trend towards improved attentional behaviour in therapy. There was, however, no apparent effect on working memory or on strategic control of attention. Further trials of longer duration, in larger samples in the initial rehabilitation phase, appear to be warranted based on these findings.