The Tayside Committee on Medical Ethics approved the study which was carried out in accordance with the Declaration of Helsinki. Written, informed consent was obtained.
Participants were seen in their own homes on three occasions. Study visits lasted approximately an hour. At the first visit, consent was taken, randomisation took place and the participant was asked to wear an accelerometer on the hip during waking hours for 7 days (a validated device  for measuring physical activity, RT3 Tri-axial Research Tracker, Stay Healthy Inc., USA). The self regulation intervention was carried out at the second visit (7 days later) along with collection the secondary outcome data. Outcome measures at the end of the study were collected on the third and final visit, three months later.
Inclusion and exclusion criteria
Women aged over 70 years who were insufficiently active or sedentary, i.e. no participation in moderate-intensity physical activity of at least 30 minutes at least 5 days per week or at least 20 minutes of continuous vigorous-intensity activity three or more times a week  were included in the study. Women were asked about their participation in physical exercise and walking at a preliminary phone call before any visits took place. Women fulfilling physical activity recommendations, resident of institutional care, housebound (unable to increase outdoor walking), having moderate to severe cognitive impairment (MMSE score < 18) precluding informed consent, having significant visual impairment and so unable to read pedometer count screen, wheelchair bound or unwilling to participate were excluded from the study.
Recruitment and randomisation
Participants were recruited from a single GP practice (see figure 1) via the well established Scottish Primary Care Research Network (SPCRN). The GP principal provided a list of all women aged 70 years and over, excluding those who should not be approached because of terminal illness, recent bereavement, severe heart failure/COPD/dementia or nursing home dwellers. The GP wrote to the women inviting them to take part in the study, including a pre-paid reply envelope. Those accepting the invitation were telephoned and asked whether they were housebound, how much exercise they were taking, whether they were visually impaired or using a wheelchair. At the first face-to-face visit, a mini mental state examination was performed.
Randomisation was performed by an individual not otherwise involved in the study using a computer-based random number generator. The participants were allocated at random, in a 3:2 ratio, to one of two groups: pedometer plus self-regulation intervention ("pedometer group") or self-regulation intervention alone ("advice group"). The pedometer group were further randomised to one of three target groups: a 10%, 15% or 20% monthly increase in step count (steps taken per day). The average daily pedometer count over three consecutive days (at baseline before intervention) was taken and used to set a target of achieving a 10%, 15% or 20% increase in steps during the first month. If the participant met her target step count, it was increased again after the first and second months. If she had not met her target, 10%, 15% or 20% was added to the average number of steps they achieved over the three day period preceding the telephone call.
Development of the self-regulation intervention and protocol for delivery
Self-regulation theory emphasises the role of goal setting, planning and self-monitoring in behaviour change . Theory-based advice was given to each participant in the form of individualised activity action plans and coping plans. Additional file 1 shows the protocol that was developed for the delivery of this advice. The protocol was developed using the Coding Manual to Identify Behaviour Change Techniques in Published Intervention Descriptions . The intervention was delivered by a senior research nurse who had received training from two experienced health psychologists (authors DJ and FFS). First, each participant was given advice about the health benefits of increasing physical activity both verbally and in pamphlet form after collection of baseline data at the second visit. Then action plans and coping plans were discussed and written with each participant in her own home after the baseline data had been collected. The action plans were designed to increase participants' physical activity levels (mainly through walking) and the coping plans were to identify how to cope with possible barriers towards increasing their walking . A graded approach to increasing walking was given with clear advice on when and where to walk and how to schedule time for physical activity. Each participant was given a daily activity diary to complete with logs of either pedometer count or time spent walking outdoors. These diaries were returned to the researcher, in a supplied stamped addressed envelope, to assess compliance with diary keeping. Each participant was contacted by telephone once a week for the first month and then fortnightly thereafter until the end of the study to provide motivation, encouragement and to troubleshoot any problems.
Validation of pedometers
Two different types of pedometer were used in this study in order to ascertain how accurate and user-friendly they were in our study population: the Omron HJ-005 spring levered and the Omron HJ-113 piezoelectric pedometers. The spring levered device is worn on the waistband; a horizontal lever-arm moves up and down in response to vertical movement and opens and closes an electrical circuit. This device must remain vertical to be effective . In this study the piezoelectric device was worn around the neck although it can be worn on the waistband. It contains a horizontal cantilevered beam with a weight at the end that compresses a piezoelectric crystal when subjected to movement, it is less susceptible to tilt than the spring levered device . At the randomisation visit (visit 2), each participant in the pedometer group was asked to read the count screen of each device and to walk 100 steps at their usual pace wearing both pedometers in the presence of the researcher. They were also asked which device they preferred. The more accurate device was given to the participant and they were asked to wear it to monitor their step count during waking hours.
The outcome measures were assessed at baseline (before the intervention) and after 3 months. Outcome measures were collected on the same day on both occasions and baseline outcomes were assessed before implementation of the intervention. The primary outcome measure was change in daily activity levels which was measured by accelerometry (RT3 Tri-axial Research Tracker, Stay Healthy Inc., USA), a device the size of a pager worn on the waistband during waking hours for a 7-day period which has been validated for use in an elderly population . Accelerometers were worn for 7 days by the participant, tri-axial data was collected in 1 minute epochs, counts < 250 or > 3000 were discarded as spurious . Counts were totalled over each 24 hr period (midnight – midnight), the 1st set of 24 hour data was discarded (incomplete day) and missing days were excluded from analysis. Counts (per minute) per day for valid days were recorded.
Secondary outcome measures were: assessment of lower limb function using a validated performance score  consisting of three 0–4 point scales summarising performance on three tests of lower extremity function – usual walking speed over 3 meters, standing balance and repeated chair stands; health related quality of life was assessed using the EuroQuol questionnaire which provides a brief measure of health status with self ratings on mobility, self-care, anxiety and depression, usual activities and pain together with a global rating of health state which has been shown to be reliable at interview in an older population ; and depression and anxiety were assessed using the Hospital Anxiety and Depression Scale . This pilot study was not powered to detect changes in outcome measures, but to demonstrate feasibility of data collection for a larger trial.
Descriptive information was generated for readability of the pedometer screen, accuracy of step counting for the two pedometer types, adherence with daily activity diaries and drop out rate. The correlation between pedometer counts and accelerometry at baseline was examined using Pearson's correlation (2 tailed), and the difference in change in accelerometer count and secondary outcomes from visit 1 to visit 3 between the two groups was analysed by t-test assuming that variances were not the same. Comparisons between target groups were carried out using analysis of variance followed by Bonferroni corrected pair-wise tests. In addition, a test for trend across target groups was carried out. Where variables were non-Normally distributed, non-parametric tests such as Mann-Whitney were utilised. All analyses were implemented in SPSS.