To assess the effectiveness of homeopathic potassium phosphate (Kali phos 6x), an off-the-shelf preparation, for attention problems associated with mental fatigue in healthy adults.
We recruited 86 adult volunteers via online advertisement and internal circulation of emails within the University of York. To be eligible for the study participants had to self-report difficulties in sustaining attention or be experiencing mental fatigue. Additionally they had to be able to communicate in English and consent to avoiding the use of self prescribed stimulants, such as caffeine and energy drinks, on the day of each test. We excluded those who were currently using a homeopathic preparation for any condition, people who were currently using prescribed stimulant medication such as those used for ADHD and people diagnosed with chronic fatigue syndrome or ME. Those eligible were offered information leaflets and an explanation of the study by the CI (MED). All participants gave written consent to participate.
The homeopathic preparation was a single dose of 0.6 g of lactose powder medicated with Kalium phosphoricum 6x (a decimal dilution equivalent to 1 part in 1 000 000, potentised by serial agitation) in 90% ethanol/water solution. The placebo was a single dose of 0.6 g of lactose powder treated with unmedicated 90% ethanol/water solution. There was no noticeable difference in taste or appearance between the two preparations. Both preparations were supplied by the Helios Pharmacy, London, who coded the batches A and B so that nobody at the trial centre was aware which powder was placebo and which Kali phos. The identity of powders A and B was not revealed by the pharmacy until after completion of the analysis. The pharmacy also supplied the manufacturer’s data sheet for the lactose used in both preparations.
The study design was a triple-blinded, placebo-controlled cross-over trial with two arms in which participants were randomly allocated to receive either a homeopathic or placebo preparation in period 1 and vice versa in the period 2.
In both periods the participants completed the 4-question mental fatigue sub-scale of the Chalder Fatigue Questionnaire, giving an integer score between 0 and 4
. They subsequently took a single dose of one of the randomly allocated preparations. They familiarized themselves with the software in trial runs of the psychological test of attention. They then performed the test, approximately ten minutes after taking either the homeopathic or placebo preparation. Each participant repeated the procedure at the same time of day, seven days later, those who received Kali phos in period 1 receiving the placebo preparation and vice versa. After completing the period 2 test, participants were asked whether they thought they had just taken Kali phos or placebo.
A review by Swanson et al. concluded that of three domains of assessing cognitive deficits, conflict resolution tasks (CRT) were the best at distinguishing children in ADHD diagnosed groups from control groups
. Of these CRTs the Stroop Colour-Word Test has been identified as one of the most sensitive for testing ADHD-specific deficits
. Therefore our primary outcome measure was accuracy score on the Stroop Colour-Word Test
The Stroop task involved participants being shown a colour word (e.g. ‘red’) which was coloured either congruently (e.g. the word ‘red’ was coloured red) or incongruently (e.g. the word ‘red’ was coloured green). Participants were required to respond to either the word or the colour of the word by pressing corresponding keys on the computer keyboard (1 = red, 2 = blue, 3 = green). Participants completed 2 practice blocks each containing 9 trials (6 incongruent and 3 congruent). In the first block participants were instructed to respond to the word and in the second block participants were instructed to respond to the colour of the word. Feedback on performance accuracy was provided during these practice blocks. Six test blocks, each containing 27 trials (18 incongruent trials and 9 congruent trials) were then presented. For the first, third and fifth blocks participants responded to the word whereas for the second, fourth and sixth blocks participants responded to the colour of the word. A recovery period of 30 seconds in between each test block was included so that the task was suitable for a linked functional magnetic resonance imaging investigation (this follow-on project was not undertaken). The dependent variable was participants’ accuracy score on the incongruent trials (maximum = 108). The test lasted approximately 18 minutes. The measure was adapted for computer presentation using EPrime® software
 and was used in both periods. We originally aimed to measure the speed of response as a secondary outcome, but the software was programmed with a 3-second delay before the response screen appeared and therefore this could not be reliably measured.
We wanted to be able to detect a fairly small difference in the Stroop score and therefore decided to design the study to detect an effect size of 0.3 standard deviations with power 0.90 and using a significance level of 0.05. For a cross-over trial, we would need the standard deviation of differences between repeated measures at a similar time gap as we proposed for the trial, or the correlation coefficient between such pairs of observations. We could find no pre-existing data allowing us to estimate this correlation, and a pilot study to estimate this would have been almost as time-consuming to carry out as the present study. We therefore arbitrarily set the correlation between the accuracy scores in the first and second periods to be 0.80, believing this would be a reasonable value for a continuous measurement recorded with only seven days between measurements. We estimated that 86 volunteers would be required to give the desired power.
Allocation to treatment order
Allocation was random using our software Clinstat
 to allocate 86 participants into equal groups in blocks of random sizes 4, 6, 8, or 10. For each participant, the folded papers containing powders A and B were put by JMB into clear plastic envelopes and labelled with the participant number and either “first” or “second”. The completed packages were passed to MED who administered them to the participants, all of whom were fully blind to the allocation.
Statistical analysis followed the method of Hills and Armitage
 using the description of Altman
. We have followed Senn by not testing the interaction between treatment and period
. As the assumptions of the two sample t methods used were not well met by the data, Mann Whitney U tests were used to confirm the analyses.
Ethical approval was given by the University of York Health Sciences Research Governance Committee on December 12, 2006. The trial registration number is ISRCTN16521161.