Main findings
The study fulfilled the primary outcomes for feasibility and acceptability to continue to a full-scale trial. The expected recruitment rate to indicate feasibility stipulated in the protocol of potentially eligible participants was 50%, but the actual percentage of eligible referrals recruited exceeded this at 65%. Those who were recruited were, in the main, demographically similar to those who were not. However, family carers who cared for men with dementia were more likely to refuse to participate than those who cared for women with dementia. The research team are unsure why this should be. There was no difference in the sex of the carer who consented to the study or refused to participate, or whether carers were the care recipient’s spouse or child. The proportion of men recruited also reflects the proportion of men with dementia in the older population.
In terms of acceptability, the percentage of participants randomised to the intervention group attending four or more of six sessions was 88%, exceeding the expected value of 75%. The median number of sessions attended was six, with two people dropping out before they began the sessions (one because of lack of time and one because the person with dementia refused), but only three of those who started did not adhere to the intervention. Thus, most of those in the intervention group appeared to find it acceptable to attend all sessions once they had started.
Recruitment and follow-up
The referral rates were four potential participants per week from two memory clinics, which informs the number of trusts required for a full trial. We used JDR and recruited two people over the recruitment period. There was clearly more potential, but 19 out of 25 (76%) of those registered on the list did not respond to contact and only two of those we contacted consented. Thus, it seems that JDR is most useful as a supplementary method to recruit people rather than as the sole or main method. Generally, ≥ 80% follow-up is regarded as satisfactory, and this trial achieved 92%.
Completion of outcome measures
Very high completion rates of the validated questionnaire measures were achieved, using carers as informants. At baseline, all but one person with dementia wore the watch and 50 (81%) carers completed the sleep diary or event markers to record the person with dementia’s bedtimes and rise times. However, at follow-up 49 (79%) of the 62 people with dementia in the randomised group had ≥ 7 days of actigraphy data and 42 (82%) of their carers provided the bedtimes and rise times on the sleep diary or by using event markers to aid interpretation.
Primary and secondary outcomes for a full trial
Actigraphy as outcome and tool
It was originally envisaged that actigraphy data would be the primary outcome in a full trial. However, in the feasibility study, it was available for only 79% of randomised participants at follow-up (for ≥ 7 days; the usual ‘gold’ standard). In addition, the sleep diary or event markers, which give times of going to bed or getting up, are required to interpret the data. Only 68% of people randomised provided this, so data interpretation was difficult for some of those with records, as it relied solely on verbal reports or the researchers’ supposition. Therefore, our feasibility study indicates that we are unlikely to have the level of reliable data required to use actigraphy as a primary outcome. The research team suggest that it would be helpful to validate the use of Actiwatches in this population before considering actigraphy measures as a primary outcome in any efficacy trial.
There is, in addition, a paucity of validation data about actigraphy in this population. Although sleep efficiency is often taken as a good summary sleep measure from actigraphy in younger people, it appears not to be as well measured in older people possibly because sleep in actigraphy is inferred from time in bed and movement. Movement is, in contrast, directly measured. According to the actigraphy data, both groups were still spending a long time in bed after the intervention and it did not look as if the intervention changed this sleep behaviour in any substantial way. There was an indication from our actigraphy results that the intervention participants were more active during the day and less active during the night. These findings accord with the qualitative feedback from the carers in the study as well as the results of the validated instruments. The carers in the study and the PPI group judged the important outcomes to be that the person with dementia was less restless during the night, more awake during the day, disturbed them less during the night-time and seemed happier (which are measured in the SDI). In these circumstances, they were unsure that these measurements of sleep added additional outcome information, or were accurate.
In contrast to their disappointment with actigraphy as feedback, the carers and therapists found the information from actigraphy valuable to use at baseline to consider the rest–activity pattern and help make a plan and this would continue to be incorporated in the manual as part of the intervention in a full trial.
Overall, the place of actigraphy in a full study would, therefore, be to measure changes in daytime and night-time activity as a secondary outcome and as part of the intervention.
Validated interview measures
In the qualitative assessment participants commented on the combined length of the questionnaires, and felt that this was acceptable if participants need to be reassured that this information is being collected for the purposes of the research and will be useful. The completion rate for all validated instruments at baseline was very high, ranging from 98% to 100%.
Instruments for the person with dementia
The validated instruments to measure sleep disorder had a high rate of completion at follow-up, and the SDI was completed at follow-up by 90% of those initially recruited to the trial. This appeared to be the most practical way to measure sleep. The qualitative feedback from carers participating in the trial and PPI indicated that they felt that it reflected their experience of sleep as well and had relevance. Therefore, it is proposed as the outcome in a main trial.
Summary data for the carer-reported instruments indicated generally better scores for the intervention group and, after adjusting for site and baseline problems, there was significant improvement in daytime sleepiness and in quality of life of people with dementia, despite the small numbers in the study. There was also no increase and a possible reduction in the numbers of people who were prescribed at least one medication for sleep. These may be related to each other. Although power for these results had not been considered, the consistency in the direction of all the results suggests that they may be real. It is important that the intervention may reduce daytime sleepiness, whereas sedative medication sometimes increases it. Similarly, it is always important to consider quality of life for someone with dementia, as an intervention may improve a specific domain while reducing overall quality of life. We would therefore conclude that the DEMQoL-Proxy and ESS should be measured in a full trial. We also administered the CSRI, to assess the feasibility of its use, and suggest that it too should be included in a full trial, as this would enable the cost-effectiveness of an intervention to be calculated.
Harms
There was no clear increase or difference between groups in the use of psychotropic medication and melatonin, with the intervention group possibly being prescribed slightly fewer medications afterwards, taking into account baseline measures. This suggests that any effects were not due to psychotropics being used as rescue medication in the intervention group. There was no indication of important harms in terms of side effects in either group.
Carer outcomes
The carers in the intervention group reported significant improvements in ZBI-measured burden, and the direction of results suggested that there may be an improvement in depressive symptoms (HADS). It is important to consider the effect of sleeplessness on carers’ stress levels and mental health, and we would conclude that these measurements should be retained for a full trial. A few of the carers disliked the questionnaires used to measure carer sleep as they found it difficult to report averages. It is possible that they might be willing to record their own sleep using a sleep diary, attributing any sleep disturbance to disturbance of their relative’s sleep or any other cause. However, this was not tested and the carers were not very positive about sleep diaries for their relatives.
As carer sleep is often disturbed by the person with dementia, it seems worthwhile to measure it. In order to select one measure for future trials, two instruments were used: the PSQI and the SCI. The rate of carers who filled in these questionnaires was satisfactory, with 61 (98%) of the randomised carers completing both at baseline. Fifty-six (90%) completed the PSQI at follow-up and 57 (92%) completed the SCI. The SCI gives the information to consider whether or not criteria for insomnia have been fulfilled, whereas the PSQI does not. Overall, more than half of the carers fulfilled the SCI criteria for poor sleep at baseline. In the qualitative interviews, two carers were of the opinion that the PSQI did not measure relevant features of sleep and that it was too long. They preferred the SCI to the PSQI, and we will go with their preferences.
Additional data
One individual with dementia present during the interview also suggested that it may be important to record whether or not carers and the person they care for share a bed or bedroom as part of the assessments, as this can affect how the questionnaire is answered. This would be added to an assessment for a full trial.
Power for a full trial
To calculate the sample size required for the main study, feasibility estimates were used by calculating the SD of baseline SDI scores (2.24) and the correlation between baseline and 3-month measurements (0.57). As there is no estimate of clinically significant difference, a SD of 0.4 was used, different sizes were calculated considering the CI for ICC from the earlier START study. A full study with 2 : 1 randomisation (which is feasible from the results above) will require 230–296 participants in the intervention arm and 115–148 participants in the TAU arm (assuming an average of 15 participants per therapist, 2 : 1 randomisation and a drop-out rate of ≤ 15%, and including an inflation for the case of non-normality).
Where to recruit participants for a full trial
It was found that memory clinics (four referrals per week from two trusts) were a better source of referrals than JDR (two consented in total). JDR would be used as an adjunct rather than as the main source of referrals for a trial.
Changes to those recruited for a full trial
Participants from a range of dementia diagnoses and living situations were recruited. It had been expected that the intervention would be delivered mainly to family carers, but consenting participants chose a complex variety of methods of delivery. Slightly fewer than half were to family members by themselves. The intervention was thus delivered to family carers, paid carers and sometimes to both. Frequently the person with dementia was also included and, in one case, half of the sessions were delivered to a person with dementia alone. Delivering the intervention with the person with dementia present, although not always problematic, did present challenges for the therapists that were addressed in clinical supervision. In a few cases, the person with dementia and the family member presented conflicting views of their sleep difficulties and the potential solutions and strategies. This was especially the case when the person with dementia denied that they had any difficulties with sleep and, therefore, did not see the need for making any changes. It became clear during and after the first session for the one person with dementia who had some sessions alone that they could not retain or recall the information being discussed and, therefore, were unable to make use of the sessions. In a future trial, people with dementia would not be excluded from jointly participating in the intervention sessions; however, additional training would be built in for the therapists on delivering sessions with people with dementia present and how to manage any conflict and interpersonal challenges that arose.
It was envisaged that someone with sleep disturbance and dementia would have a paid carer or family member with them at night to ensure safety. This was not always the case. When people lived alone, the carers (whether family or paid) were unable to implement strategies, for example a scheduled bedtime or wind-down routine. It was also difficult to gain reliable information about the sleep patterns of people with dementia living alone. Therefore, people without a night-time carer would be excluded in a full trial. In addition, the intervention would be delivered only to people with dementia who have a carer also participating.
Families also found that making changes during the day was difficult if the person with dementia was alone during the day. This led to them being offered time switches for the light boxes. They found increasing activity if there was no one at home depended on the availability of services, such as going to a centre or having someone to take the person with dementia out. It is important to discuss this explicitly and consider the options available.
When paid carers attended the intervention, they were also able to implement strategies. Working with them may be important to allow people to remain living at home, as care agencies insist on full-day rates and two carers if the carer is disturbed frequently during the night. This may become financially non-viable and make it harder for someone with dementia to stay at home. Since it appeared feasible and is potentially useful, paid carers would be included in a full trial if people with dementia and their families wished.
People with a diagnosis of alcohol-related dementia or who were currently drinking were not excluded from the study. Two participants drank alcohol during the intervention sessions. These people were unable to work with a plan to change their sleep, which involved reducing alcohol, and there was also concern for the safety of our therapists visiting the homes by themselves, if the family were often absent. The participants then dropped out. Therefore, in future, anyone with current heavy drinking habits would be excluded.
It was not considered that people who were leaving the country for months or forever would be referred to the study. However, one person was leaving the UK and, although they did not fufil inclusion criteria in other ways, being in the UK for the trial and follow-up would be specified as a criterion. This would not exclude those going on holiday, only those who would be away for a period that prevented them from receiving the intervention or being assessed after the intervention.
The intervention content and delivery
Manual design and delivery
Generally carers felt that the number of sessions was appropriate and liked the balance of pictures, vignettes and text. They appreciated the direct quotations from carers. Some thought that they would continue to use the manual itself as well as strategies within it. To make it easier for participants to use the manuals in between the sessions and after the intervention, one participant’s suggestion of tabs will be used.
Topics
Carers liked the information about caring for someone with dementia, the biological processes of sleep and how dementia can affect an individual’s sleep, and found the diagrams helped them to understand why they should make changes, such as increasing natural light and activity, including exercise. They tried different components depending on individual needs. Many of them asked for specific help from relatives, which would continue after the study. They often made adaptations to the environment, for example putting signs on the bathroom door. Some realised that their relative got up during the night for a reason that they were able to address, which helped improve their relative’s comfort and reduce their own sleep disturbance.
At session 6, all carers made a plan for the future and clearly appreciated a multimodal intervention as they wanted to continue using a range of strategies. All intended to persevere with the light box (although not necessarily in the summer) and continue to increase activity or physical exercise. Many intended to carry on with a later bedtime routine, usually going to bed around half-an-hour later than before, and continued to be aware that a comfortable bedroom is important. Most carers became more aware of looking after themselves by giving themselves time, challenging negative thoughts or using relaxation. They felt that this had improved their own quality of life and made them a better carer, and so they would continue with these self-care strategies.
The carers interviewed wanted to ensure that, in future, therapists did not go through all the content of the manual on topics that were not relevant to them, for example sections about increasing activity if the person with dementia was very active. For a full trial, it will be stated at the beginning of the sessions that not all of the topics will be relevant, and, if they are irrelevant, they will be omitted, but participants will still have the information within their manual should it later become relevant (e.g. if their relative’s sleep disturbance or physical health changes, or as their dementia progresses).
The carers felt that using telecare and other assistive technology (which was often part of plans) could be further emphasised in a future trial, as this would be useful for those people with dementia alone during the day and could help manage difficult or risky behaviours.
Some carers mentioned in the qualitative interviews that they would have preferred electronic versions of the sleep diaries that were used for the actigraphy and in the intervention, and found that it was difficult to complete the next day when they had to remember when someone had been awake in the night. They suggested that the timings of the sleep diary be changed to start recording from 06:00 instead of 12:00, and day and night symbols be used to make it a lot clearer; this will be done. The research team believe that it is worth investigating the possibility of, for example, sending an automated message to prompt a carer to report information, such as bedtimes or rise times. This could be done by simply replying to the text or by writing the information in the diary.
Delivery by therapists
The therapists were clinically supervised psychology graduates who were delivering the intervention to varying combinations of carers and people with dementia. Both the participants and the PPI group felt that the therapists were taking a facilitative and supportive approach. This was essential to the therapeutic process, as it was complex for participants to overcome the barriers to helping relatives who were often unable to formulate and remember plans themselves. They valued face-to-face contact and that the strategies were delivered to them and tailored to their individual needs.
Complexity
The intervention was complex and could be individually tailored, but took only six sessions. As sleep problems are complex and diverse, this allowed it to be appropriately individualised.
Fidelity
Fidelity ratings were high, suggesting that different therapists were able to deliver the intervention consistently.
Flexibility
The intervention was offered weekly; the median time taken to deliver the intervention was 7 weeks. Most participants thought it was ideal to have 1–2 weeks between sessions, rather than necessarily aiming for weekly sessions. Therefore, a degree of flexibility in timing would be offered for an intervention in future, allowing for the intervention to take up to 3 months.
Strengths and limitations
The trial recruited people from urban and suburban environments in London only, which is a limitation to the external validity. Although there were both male and female care recipients, men were more likely to refuse. However, people from a range of age groups and with varying types and severity of dementia, relationships to the care recipient, marital status and educational backgrounds were successfully recruited. In particular, it is often the case that ethnic minorities are under-represented in studies; this study recruited ≈35% of people of ethnic minority status. In general, the findings should have good external validity.
Although one carer commented that sleep patterns in the last 2 weeks may be atypical, the analysis of such data at a group level should eliminate any systematic bias.
A very high proportion of carers remained in the study. Although the outcome assessors were blinded to outcome, the participants were not. It is possible that there was some bias in the ratings, as some carers might have felt that they had to report a positive result to please the researcher interviewing them, who was not the therapist, but in most cases had met them for screening and baseline assessment. However, the instruments are validated, and in other studies carers have frequently reported no beneficial effect in quantitative interviews.80 In addition, those in the intervention group were prescribed slightly less medication for sleep, suggesting that independent doctors, using family reports, may have felt that sleep had improved.
All participants who were interviewed and had completed the intervention liked it. Of the three carers who had started the intervention but not completed at least four sessions, only one could be recruited for a post-intervention qualitative interview. The researchers were trained in qualitative interview techniques, including how to make the interviewee feel comfortable and avoid leading questions. During the interviews, they emphasised to the carer that they wanted to hear suggestions for changes and what had not worked, as that would be very useful to inform further development of the intervention and the research programme. The slight reduction in the prescription of psychotropic medication also suggests that the participants were reporting some improvement to their doctors.
If the person with dementia lived alone, the sleep pattern was reported by the person with dementia, who may have had difficulties remembering it, although their ability to live alone may indicate that they were less impaired. The carer would, therefore, have been likely to report their relative’s assessment and this assessment may be less reliable.
Data were gathered to enable the design of a full trial, in terms of both a power calculation and primary and secondary outcomes. The primary outcome is SDI rather than actigraphy, as had been envisaged. The results were used to calculate the numbers needed for a full trial. The clinically relevant difference in SDI score is not known, but an effect size > 0.2 is usually accepted as such.81
Participants were asked to wear the Actiwatch on the same wrist (most often on the non-dominant side) at baseline and follow-up, to help consistency of interpretation of results. Actigraphy interprets sleep from movement rather than measures sleep. Detection of ‘sleep’ is made by inference from lack of activity. The accuracy of this may vary with the population, device and circumstances. The validated sleep estimation algorithms all assume the intention to sleep, so are for people in bed around their habitual bedtime.82 There is little validation in people with dementia, who may frequently be quite still while awake, move around during sleep or may sleep during the day. The families thought that the person whom they looked after was sleeping better, but the actigraphy results did not support this. They also felt that the person with dementia was less sleepy during the day, but the Actiwatch may not pick this up, because the algorithms are not expected to do so.
Interpretation
A manual was coproduced in this study that shows acceptability and feasibility. Minor changes to the manual content and delivery would be made in a full trial. These comprise adding more flexibility in timing; stating at the beginning of the sessions that not all of the topics will be relevant and omitting those that are not; discussing how to get help to increase activity early on for those who are alone during the day; emphasising interventions for safety, including telecare; and adding tabs to the manual to make it easier to use. The diary design would also be improved and the possibility of sending an automated message would also be investigated, to prompt a carer to report information, such as bedtimes or rise times, by replying to the text or by writing the information in the diary.
Changes can be made to the inclusion and exclusion criteria, such as excluding those without a family or paid carer at night, those who are not staying in the UK for the period of the study or those who are currently drinking. The primary outcome (SDI) can be stated, a full trial can be powered and appropriate secondary outcomes can be chosen.
Conclusions
This acceptability and feasibility study fulfilled its primary outcomes and indicated that the SDI, which is a questionnaire measuring sleep and completed using carer information, would be a satisfactory primary outcome.
Implications for health care
There is now a feasible and acceptable manual that can be used in a future trial, and the information to design such a trial.
Recommendations for future research
The study was not powered for efficacy, but the evidence from the validated questionnaires suggested that there was potential for efficacy. However, as it cannot be certain from this indication, this is something that could be tested in a full trial. It may be appropriate to evaluate a comprehensive intervention first, with options to see how it might be delivered more cheaply after the efficacy trial. This encompasses both improving sleep disturbance and increasing the quality of life of the person with dementia, and reducing the stress on their family carer. The recruitment and retention reflect the salience and potential benefits of the intervention, which augurs well for the next step, that is, a full trial.
