Three-year follow-up to assess the maintenance of physical activity levels

Introduction

The PACE-UP analyses showed positive effects on 12-month PA levels (see Chapter 3). We wanted to see if this effect was maintained at 3 years, as this has important implications for the NHS; specifically, would any future pedometer programme require a ‘top-up’ after 12 months? Our interventions led to an extra 33–35 minutes weekly of MVPA in bouts (an increase of about one-third from baseline) in a predominantly inactive cohort. Delivery via three nurse PA consultations had the same effect on 12-month outcomes as the simpler, cheaper postal delivery. These are exciting findings, as they show that a low-cost postal pedometer intervention increases PA levels in sedentary adults and older adults. However, it is vital to know whether 12-month effects persist at 3 years or if a further intervention boost is needed. We therefore successfully applied for additional funding from NIHR’s Health Technology Assessment programme to follow-up the trial cohort at 3 years (2 years after the previous 12-month follow-up) with both quantitative and qualitative evaluations. Both the qualitative187 and the quantitative findings188 have been published and are reproduced here under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0).

To date, little is known about the long-term sustainability of PA interventions. A meta-analysis of interventions (including pedometers) to increase PA levels in 55- to 70-year-olds included only four trials with data beyond 12 months (all self-reported). They found a limited evidence base beyond 12 months, and called for more trials with a longer follow-up period and objective PA measures.26 These findings were supported by a Cochrane systematic review23 and recent NICE guidance on PA interventions.29 The ProAct 65 + trial189 of a PA intervention found that between-group differences persisted at 2 years post intervention, but only for self-reported PA.

As well as a lack of long-term objective PA data after interventions, there is also a lack of qualitative evidence on maintenance. A literature review suggested that PA disengagement usually occurs 6 months after an intervention has ended, but called for more research to distinguish the factors that lead to successful and unsuccessful PA maintenance.190 A very small primary care study followed up participants 6 months after a pedometer-based intervention, and found some useful insights to explain how this pedometer intervention worked and how it may be developed,44 but further qualitative studies on longer-term effects are lacking.

The PACE-UP trial 3-year follow-up provides evidence on objective PA levels, 2 years after the 12-month follow-up. After the 12-month follow-up, 212 out of 322 controls (66%) received a pedometer, handbook and diary by post. They had no further input (unlike the original postal delivery group, who were telephoned by a research assistant 1 week after being sent the pedometer, to check that it had arrived, and who were asked to return their completed PA diaries for review at 3 months). The control participants being sent the pedometer by post mimics what would happen if this simple, pragmatic intervention were to be rolled out by post through routine primary care, without any further input. As described in our protocol,70 a further 64 out of 322 control participants (20%) attended a single nurse appointment after the main trial, in which they were given a pedometer, diary and handbook, but, again, received no further contact. The other intervention groups received no further intervention after the 12-month follow-up [apart from a small number of the postal delivery group, 50/312 (16%), who had a single nurse PA appointment].

Anthropometric measures did not show differences at 12 months; therefore, data on these were not collected at the 3-year follow-up. Therefore, our follow-up study focused on establishing evidence of effectiveness at 3 years in objective PA measures and had the following objectives:

• to investigate if the original nurse-support and postal delivery groups showed any persistent intervention effect (change in step count and time in MVPA in bouts) at 3 years, compared with the baseline levels
• to investigate if there were any differences between the original nurse-support and postal delivery groups in their change in objective PA levels (step count and time in MVPA in bouts) between baseline and 3 years
• to investigate if the simple postal pedometer intervention at 12 months increased objective PA levels (step count and time in MVPA in bouts) in the control group compared with baseline.

We also felt that it was important to explore how participants in the nurse-support and postal delivery groups felt about the interventions in terms of maintenance of any increase in PA levels, and what factors might help to encourage this further or to overcome barriers they had to increasing or maintaining their PA levels. Furthermore, we were interested in how the initial control group felt about the minimalist intervention that they received. Our qualitative evaluation therefore had the following objectives:

1. qualitative evaluation of both the nurse-support and postal delivery intervention groups to look at factors affecting PA levels and maintenance of any increase in PA levels at 3 years
2. qualitative evaluation of the control group to see the effect of the minimal intervention on PA levels.

Methods for quantitative physical activity evaluation

The 3-year follow-up focused on collecting the objective PA accelerometry data and other questionnaire-based self-reported outcomes by post to minimise data collection costs. To allow for seasonal variation in PA levels, the baseline, 12-month and 3-year outcomes needed to be assessed in the same calendar month; therefore, follow-up ran from October 2015 to November 2016.

Ethics approval and research governance

Ethics approval for the 3-year follow-up was granted to the trial from London, Hampstead Research Ethics Committee (reference number 12/LO/0219). NHS local research and development approval was granted to cover all of the practice sites.

Participants eligible for the 3-year follow-up

All trial participants who had not withdrawn from the trial were eligible to be followed up, even if they had not provided 3- or 12-month follow-up data. Lists of eligible participants were organised by practice and practices were asked to check whether any participants had died, moved away or developed a terminal illness or dementia since trial participation. These patients were then excluded.

Contacting participants

Eligible participants were contacted with a letter explaining the trial 3-year follow-up. A participant information sheet, consent form and a Freepost return envelope were also included, as was information on the main 12-month trial results. The letter explained that a research assistant would contact the participants by telephone in around 1 week’s time to discuss the 3-year follow-up. If they were happy to take part without further discussion, they were invited to post back the signed consent form.

Informed consent

The research assistant contacted participants by telephone approximately 1 week after sending out the letter about the 3-year follow-up to discuss any questions they might have after reading the participant information sheet. Part of the consent form included consent to contact participants for an interview to discuss their current PA levels in more detail. If the participants were happy to proceed with the 3-year follow-up, they signed and dated the informed consent sheet and returned the top copy to us.

Data collection

Once the informed consent for follow-up was agreed, the research assistant arranged a time to post out the accelerometer (GT3X+) to participants to measure their current usual PA levels for 1 week. Instructions about how to wear the accelerometer were included (on a belt, over one hip) and participants were asked to wear it for 7 consecutive days, from getting up until going to bed, as they had done previously. A diary was also provided to record what activities were done and for how long. They were also sent a health and lifestyle questionnaire to complete (see Appendix 7; this was similar to that completed previously) and a short questionnaire about self-reported PA levels to complete (the 7-day PA questionnaire used previously) after they had finished wearing the accelerometer. They were provided with a Freepost return envelope to send the accelerometer and both questionnaires back. If the accelerometer did not record at least 5 days with at least 540 minutes per day, participants were asked to rewear it for a further week. The set of recordings with the greatest number of days with at least 540 minutes per day was included in the analysis. Once accelerometers with adequate data were received, participants were posted out a £10 gift voucher.

Outcome measures

The main outcome measures (all accelerometry) used to evaluate the 3-year follow-up were as follows:

1. change in average daily step count, measured over 7 days between baseline and 3 years
2. change in time spent weekly in MVPA in ≥ 10-minute bouts between baseline and 3 years
3. change in time spent sedentary weekly between baseline and 3 years.

Although patient-reported outcome measures were collected from the health and lifestyle questionnaire (e.g. depression,75 anxiety,75 quality of life,76 self-efficacy,74 pain,77 disability80) and from the 7-day PA questionnaire (IPAQ72 and GPPAQ73), these have not been assessed further as part of this report.

Accelerometer data reduction

ActiGraph data were reduced, as described previously in Chapter 2, for the main trial. Analysis summary variables were also identical to those used in the main trial, described fully in Chapter 2.

Procedure for accounting for missing data

Only days with at least 540 minutes of registered time on the accelerometer on a given day were used. The main analysis of effect included all subjects with at least 1 satisfactory day of recording at 3 years.

Statistical methods

Statistical methods for the analysis of the 3-year follow-up are largely as described in Chapter 2 for the primary analysis at 12 months. Average daily step count at 3 years was computed from a random-effects model, allowing for day of the week and day order of wearing the accelerometer as fixed effects and participant as a random effect. The average daily step count at 3 years was then regressed on average daily step count at baseline with treatment group, age, sex, practice and month of baseline accelerometry as fixed effects and household as a random effect, in a multilevel model. The same analyses were carried out for MVPA in ≥ 10-minute bouts and daily minutes of sedentary time.

The primary analyses used the 681 participants who provided accelerometry data at 3 years. Sensitivity analyses were carried out to assess the effect of missingness:

1. Multiple imputation methods were used to impute outcome data for those missing at 3 years, assuming that outcomes were missing at random, conditional on variables in the model. We used the Stata procedure mi impute.
2. Missing-not-at-random analyses were used when it was assumed that changes in the control group from baseline to 3 years were missing at random, but the change in each of the intervention groups was ± 500 and ± 1000 steps from their missing-at-random estimate.

For this analysis, we used a mean score method,191 which has been implemented in a Stata module, rctmiss, available from Statistical Software Components at https://ideas.repec.org/s/boc/bocode.html (accessed 25 May 2018).

Methods for qualitative evaluation

Sampling

Participants consented to be contacted for a telephone interview at the same time as they consented to take part in the 3-year follow-up. In February 2016, the research assistant produced three lists of participants who had already provided 3-year accelerometry data and who had given consent to be interviewed (one list for each arm of the trial – nurse support, postal delivery, control). The trial statistician randomly sorted these three lists ready for the qualitative researchers to start approaching participants for interviews. The interviewers were blinded to participants’ previous and current PA levels.

One of the aims of the qualitative evaluation was to explore the success, or otherwise, of the minimalist intervention provided to the control group after the main trial; therefore, the following participants were excluded from this qualitative evaluation:

• control group participants who had attended a nurse consultation after the main trial
• control group participants who opted not to receive the postal pedometer after the main trial.

At 12 months, participants in the postal delivery group were also offered a nurse consultation. Again, those who attended the nurse consultation were excluded from the sampling procedure, so that all those sampled from the postal delivery group had received just the postal intervention.

The aim was to interview approximately 15–20 people from each arm of the trial (45–60 people in total), but continuing further if required, until saturation was reached. As two researchers (CB and CW) were interviewing participants, they met regularly to discuss the sampling, interview schedule and any emerging themes. Interviews were conducted until saturation of new information was reached. By looking at the participants’ demographic information it was possible to ensure that the study group included both males and females and also represented a range of different ages and ethnicities, to ensure that a wide range of views were explored.

Recruitment and informed consent

Participants were initially contacted via e-mail to arrange an appropriate time to contact them for a telephone interview. Participants without an e-mail address were called and the interview was either conducted then or arranged for later. To assess response, a detailed record was kept of when each participant was contacted, including information on who agreed, who refused and who could not be contacted. Charlotte Wahlich and Carole Beighton conducted the guided interviews with participants using a topic guide (see Appendix 7). Before the interview commenced, the participants were reminded of their initial consent to be approached for an interview; if the participants were happy to go ahead, their consent was then sought for the interview to be audio-recorded. Once the recording had started, the qualitative researchers stated the participant’s ID number to ensure confidentiality and anonymity in the subsequent transcript. On interview completion, the participants were offered a £10 high-street gift voucher to thank them for their time.

Transcribing

Interviews were promptly transcribed verbatim by an external source. Once the transcripts were received back, they were double-checked against each audio-recording by the qualitative researchers. Transcripts were also circulated to the research team to ensure consistency between the interviewers and to help assess when theme saturation had been reached.

Interview schedule

The interview schedules were developed through discussions with Tess Harris, Charlotte Wahlich, Carole Beighton, Christina Victor and Rebecca Normansell. Slightly different questions were used for participants in the intervention groups (postal delivery and nurse support) and participants in the control group (see Appendix 7). For those in the intervention groups, the aim was to explore participants’ views about PA maintenance and whether or not a ‘top-up’ intervention was required, whereas for those in the control group, the aim was to explore their views about the minimalist intervention. The interview schedules were revised slightly during data collection to ensure that the questions were clear and to include additional questions to gain a better understanding of the participants’ experiences.

Analysis

All verbatim transcripts were read repeatedly by Charlotte Wahlich and Carole Beighton. Initial line-by-line coding was conducted independently to assign conceptual ideas to important episodes within the data. Through discussion with Rebecca Normansell, Tess Harris and Christina Victor, any discrepancies were resolved; this helped to ensure that the interpretation and categorisation of the data were valid. After further discussions between Charlotte Wahlich and Carole Beighton, these codes were then refined and grouped into emergent and anticipated themes.

Results for the quantitative physical activity evaluation

Follow-up rate

Of the 1023 original trial participants, 32 had withdrawn by the end of the 12-month follow-up, a further two had died between the 12-month and the 3-year follow-up and one was excluded by their practice for health reasons. Therefore, we approached 988 participants and 681 provided adequate accelerometry data (≥ 1 day with ≥ 540 minutes wear time) for analysis, giving a 3-year follow-up rate of 69% (681/988). However, in relation to the initial trial participants providing 3-year outcome data, the 3-year follow-up rate was 681 out of 1023 (67%). The CONSORT flow diagram with 3-year follow-up data is shown in Figure 15, by randomised group.

FIGURE 15 The PACE-UP CONSORT flow diagram with 3-year follow-up data. (Figure not reproduced here; see the original article.)

Data completeness

Table 20 shows that 92% of participants (625/681) overall provided ≥ 5 days of accelerometry data at 3 years (88% of control participants, 94% of postal delivery participants and 93% of nurse-support group participants).

Objective physical activity findings

Table 20 shows the summary measures for all three groups at each time point and Table 21 shows the estimates of effect for the different groups. For the main trial outcome of steps per day, both intervention groups were still doing more than the original trial control group: 627 steps (95% CI 198 to 1056 steps) in the postal delivery group and 670 steps (95% CI 237 to 1102 steps) in the nurse-support group. The nurse-support and postal delivery groups combined did 648 steps per day (95% CI 272 to 1024 steps). The pattern was similar for the total weekly MVPA in bouts (minutes per week): 28 minutes (95% CI 7 to 49 minutes) in the postal delivery group and 24 minutes (95% CI 3 to 45 minutes) in the nurse-support group. The nurse-support and postal delivery groups combined did 26 minutes (95% CI 8 to 44 minutes). There was no difference between the groups at 3 years for sedentary time or daily wear time.

Missing data analyses

Imputation analyses (see Table 22) presents the results for missing at random, using imputations based on the different assumptions detailed in the methods section. The imputation analyses show that making adjustments for missing values has only a small effect on the primary outcome, that is, the step count estimate, and does not change the interpretation.

The missing-not-at-random analyses make a bigger impact, but only when we assume that there is a strong differential departure between the non-random effects in the control and treatment groups (solid lines in Figure 16). Even then, it is only when we assume that the missing data in the treatment groups are 1000 steps below their missing-not-at-random values, while the values in the control group are at their missing-not-at-random values, that the treatment effects become non-significant; even then, the CI is still largely positive.

FIGURE 16 Sensitivity analyses for different values of missing step counts. (a) Postal delivery group vs. control group; (b) nurse-support group vs. control group; and (c) postal delivery and nurse-support groups combined vs. control group. The figures show how different values for the missing step counts changes the treatment effects. The starting point where all missing step counts are replaced by ‘missing-at-random’ estimates in each group is represented by the zero difference estimate. Estimates in the different groups are then altered differentially over a range of scenarios. Control group – missing step counts are the same as the 3-year missing-at-random estimates. Treatment groups – missing step counts are 500 or 1000 steps lower than the 3-year missing-at-random estimates. Control group – missing step counts are 500 or 1000 steps higher than the 3-year missing-at-random estimates. Treatment group – missing step counts are the same as the 3-year missing-at-random estimates. Missing step counts are 500 or 1000 steps lower or higher than the 3-year missing-at-random estimates in all control and treatment groups (see White191 for methods). The vertical lines are the 95% CIs for the estimated treatment effects. The treatment effect becomes statistically not significant when the CI crosses the horizontal line at zero. (Figure not reproduced here; see the original article.)

Results for the qualitative evaluation

Between March and April 2016, 105 participants were randomly selected, 96 were contacted and all agreed to participate. Telephone interviews were arranged and undertaken with 60 participants (20 from each trial arm). Fifty-two participants were white and eight were non-white. Interviews lasted between 4 and 22 minutes (median 10 minutes). One participant who had difficulty hearing was e-mailed the questions to complete. In the quotations that follow, ID3Y_ refers to the participant’s ID number, F/M refers to the participant’s sex, the number following this refers to their age and N, P or C refers to whether the participant is in the nurse-support group, the postal delivery group or the control group, respectively.

Factors affecting physical activity levels and maintenance at 3 years

A key theme that emerged from our interviews was the impact that the PACE-UP trial had on participants. Most participants, regardless of what group they were in, reported an increased awareness of PA. Participants described an increased understanding about the importance of PA for health, as well as an awareness about the amount of PA required to meet their daily step count target:

It’s made me more aware of the need to actually commit to doing some exercise a day, just strolling around the house, and going to the shops occasionally doesn’t really make much difference. It doesn’t meet the sort of threshold that you need to reach to ensure that you lead a healthy lifestyle.

Participants felt that taking part in the PACE-UP trial and using a pedometer had ‘kick-started’ regular activity:

It was the PACE-UP trial that helped get me started and I think that did make a huge difference to me.

Participants highlighted different barriers to, and facilitators of, being able to stay physically active in the longer term. These barriers and facilitators were often the opposite of each other; for example, some participants saw good weather as a motivator to engage in PA, whereas others saw bad weather as a barrier to being physically active. Other important facilitators and barriers included health, self-motivation, ageing and social support. These factors were considered to be important by participants, regardless of which group they were in. For some participants, they felt that engaging in regular PA helped them to manage their health condition:

The more active I am, the better the arthritis is.

Conversely, others felt that having a health condition was a contraindication to PA:

I’ve got an ongoing problem where I get pain, so there’s no way I’m going to be going out walking if I don’t have to.

Self-motivation was seen to be an important determinant to PA:

I think it’s got to come from inside.

Some participants attributed the PACE-UP trial to providing them with this motivation:

Before the PACE-UP trial, I had no incentive. And that really did help me. That put me/gave me the first steps as it were, got me on the right track.

On the other hand, a few participants felt that since the PACE-UP trial had ended, their PA levels had decreased as a result of a lack of self-motivation:

I’m not covering anything like what I was covering when I was on the programme, but I don’t know how . . . to put myself in that mind set . . .

Some participants chose to engage in PA as a way to stay young and slow down the ageing process:

I’ve got nieces and a nephew . . . I need to be active to keep up with them because they are young. I just need to keep up with everyone else really. I don’t want to slow down and become old. Unfortunately, I’m not really motivated by anything else.

Other participants felt that their age had become a barrier to PA and were less likely to do as much as they did when they were younger:

If anything I’m getting a bit older and I’m beginning to find it a little bit more of a strain.

Many participants spoke about the importance of having friends and family to motivate them to participate in PA. Support from, and accountability to, family and friends were therefore seen as common facilitators of PA:

Maybe my motivation is not only my health, but having somebody to do it with . . . to maybe be paired up with somebody who was like-minded . . . if I’ve promised/only promised myself, then I might find excuses not to do it.

Lack of social support meant that some participants did not engage in PA:

I haven’t got anyone to walk with.

Lack of time was the most frequently cited barrier to maintaining PA. Unlike the other factors previously mentioned, ‘having time’ to engage in PA was not mentioned as a facilitator. The reasons for a lack of time included having ‘family responsibilities’ (looking after children, caring for older relatives), ‘work commitments’ or simply being ‘too busy’:

I’ve got plenty of things that encourage me, it’s just the time I find because I work full time, I just find it difficult to come home, sort of prepare meals, go to the gym, go for a walk . . . I don’t think I need any more actual motivation, I just need a bit more time!

Some participants spoke about strategies they adopted to overcome the barrier of lack of time, either by incorporating PA into their daily routine or by building up their daily PA in short bouts of activity:

I walk up the escalators to get my little bit of exercise.

As opposed to the mind set of, oh, I’ve got to do an hour in the gym. Actually 10 minutes solid walking somewhere, several times a day, actually builds stuff up.

At the end of the interview, participants were asked for their views on what additional support could be provided to aid PA maintenance. Participants were offered examples to comment on, which included regular text messages, online resources, annual nurse appointments and walking groups. Participants had varying preferences over which additional resources they would find the most beneficial. Although some participants liked the idea of a regular text message or ‘jolly little reminders’ (ID3Y47F51N) encouraging PA, others felt that these would be ‘too intrusive’ (ID3Y27M56P). Similarly, some participants liked the idea of having PA resources online to ‘open at their own time’ (ID3Y9F62C), whereas others felt that obtaining this information online would require ‘a more proactive’ (ID3Y47F51N) approach. On the whole, walking groups and nurse appointments were considered to be favourable. Some participants felt that walking groups would provide ‘more motivation to go out’ (ID3Y17M68C) and a regular nurse or other appointment would provide ‘external accountability’ (ID3Y52M66N).

As well as providing feedback on suggested possible resources, we proposed that participants also come up with additional suggestions. These suggestions included holistic appointments with a nurse to discuss both diet and PA, and more opportunities for older people:

There is not much for people over 60, there’s no real places that are easy on the doorstep.

One participant spoke about being afraid to increase their PA levels, as they were unsure of whether or not it was safe. This participant sought more guidance around riskless ways to increase PA levels:

I’m doing quite well with what I’m doing, so what’s the point of sort of having a risk of a heart attack or something like that, suddenly break in to a stride and start running, so maybe a bit of guidance on what’s going to happen to you if you do step up your exercise plan.

The effect of the minimal intervention on physical activity levels of participants in the control group

Of the 20 control group participants interviewed, 17 received the pedometer, handbook and diary after 12 months. Thirteen of these participants reported using these resources when they first received them; however, at 3 years, only four participants were continuing to monitor their steps [two with a pedometer, one with a Fitbit (Fitbit, San Francisco, CA, USA) and one with a mobile phone].

Although most of the control group participants who were interviewed felt that the PACE-UP trial had not increased their PA levels, they still stated that the PACE-UP trial had increased their awareness:

It’s made me aware that I do not do as much as I should be doing.

Some, however, did talk about changes they had made to their daily lives as a result of their increased awareness:

I’m more likely to walk to work now, rather than going on the bus.

For those who did not utilise the pedometer, some reported difficulties in using it:

I could not work the pedometer . . . could not get it going.

A few others who used the pedometer stated that discontinued use was either because it had a negative impact on them psychologically or because they had fallen out of the habit of using it regularly:

It’s quite distressing to see how little I do.

. . . there are lots of other things [that] intrude and you tend to slip back to old patterns.

There was also a suggestion that, as well as the pedometer, you needed to have someone to report back to:

It always needs to have someone keeping you aware I think.

Discussion

Main quantitative findings from the 3-year follow-up

We followed up just over two-thirds of the original trial cohort with accelerometry outcome data at 3 years (and over 90% provided ≥ 5 days of data). Compared with baseline, those in the original nurse-support and postal delivery groups were still doing significantly more steps per day and weekly time in MVPA in bouts at 3 years than the control group. There were no significant differences in outcomes between the postal delivery and nurse-supported intervention groups (as was also the case at 12 months), and there were no significant differences between the three groups in terms of wear time or sedentary time. Our sensitivity analyses looking at the potential impact of missing outcome data at 3 years suggest that it is highly unlikely that missing data have substantially biased our results. Fairly extreme departures from the missing-at-random analyses were needed to result in non-significant effects and, even then, the 95% confidence limits were largely positive. This suggests that the trial interventions had a persistent effect on objectively measured PA levels at 3 years, with no difference between intervention groups. The fact that the minimal intervention given to the control group at 12 months was not effective at increasing the participants’ PA levels suggests that the additional support given to the original trial postal intervention group (with a follow-up telephone call after 1 week and encouragement to return the completed PA diary after 3 months) was an important component of this group’s success. Both the follow-up telephone call and the encouragement to return the completed PA diaries after the intervention were not part of the intended intervention package, but, rather, were research measures as part of the process evaluation to ensure fidelity of the intervention delivery. However, this minimal support, which was not provided face to face or by a health-care professional, seems to have been important to the success of the postal intervention. The original trial postal delivery group also received the postal pedometer intervention when they had just been recruited to the PA trial, when motivation may have been higher, and they had step count targets set for them based on their baseline blinded pedometer use, whereas those receiving the materials at 12 months needed to wear the pedometer again for 1 week to set their target step count. These factors may also have been important to the success of the trial postal intervention group.

Main qualitative findings from the 3-year follow-up interviews

A key finding was that most participants discussed their increased awareness of PA, irrespective of which group they were in and regardless of whether or not they thought that the PACE-UP trial had actually increased their PA levels. Key barriers to, and facilitators of, maintaining PA were reported that were often the inverse of one another and included health, weather, self-motivation, ageing and social support. Lack of time was the most frequently cited barrier. Some participants were able to overcome lack of time by incorporating PA into their daily routine or by breaking PA down into smaller, more manageable bouts throughout the day. Participants gave us mixed feedback on how useful they thought text messages and online resources would be to help to inform future interventions to increase and maintain PA, but walking groups and nurse or other appointments to provide external accountability were broadly welcomed. Additional suggestions provided included more holistic appointments with a nurse and more opportunities for PA for older people. Participants had differing opinions over the resources they would find most beneficial, emphasising the importance of individual tailoring of some aspects of PA interventions. Only a few of the control group participants interviewed were continuing to use the pedometer provided at 12 months. Some participants were not sure how it worked, and others felt that, as well as using the pedometer, it was important to have someone to report back to. This highlights the importance of the extra contact that participants in the postal delivery group received as part of the main trial: a follow-up telephone call to check that they knew how to work the device and encouragement to send back their completed PA diaries with step count recordings after the 12-week programme.

Further discussion of the strengths and weaknesses of both the quantitative and qualitative approaches, and the implications of the findings for health care and future research, are detailed in Chapter 9.