Investigating the Impact of Tailored Reports on Anxiety Amongst Cancer Patients and Their Confidants

Aim:

• Based on a randomised trial the researchers aimed to model use of leaflets in the home as a function of three factors:

1. whether patients chose the information themselves or if it was chosen for them;
2. whether it included personal information from their own medical record or not;
3. whether it included anxiety management advice or not.

• The researchers aimed to model patient and confidant's anxiety at three months as a function of these three factors and patients' use of leaflets in the home.

Patient Population: Beatson Oncology Centre (BOC). Patients thought to be eligible were to be identified from outpatient appointment diaries over fifteen months. Medical records were to be checked for all those initially identified to ensure that patients have breast, gynaecological, prostate or bladder or testicular, or selected cancers of the head or neck. Patients excluded were to be those:

• receiving palliative treatment,
• with no knowledge or understanding of their diagnosis (very few),
• with visual or mental handicap,
• with severe pain or symptoms causing distress,
• medical or nursing students,
• taking part in other trials with possibly confounding interventions, and
• those not giving informed consent.

Sample size: The researchers aimed to recruit 400 patients. This was to give 80% power to find a difference between 40% and 27% who have improved anxiety between (eg) 200 in personalised group and 200 in general group.

Recruitment: Eligible patients were to be sent a letter explaining the study and advising them that a researcher will approach them in a visit to the BOC soon after, to seek participation in the study. Patients were not to be recruited at their bad news consultation but as soon as possible thereafter and at any point in their course of contact and treatment at the Beatson. At recruitment the researcher was to further explain the study and seek written consent. If the patient consented the researcher was to arrange a time for the intervention and give the patient a questionnaire to take home to complete themselves and a questionnaire for their 'confidant'.

Baseline data: Patients were to be asked to bring the completed questionnaire with them to the intervention consultation about one week later. The questionnaire included the following sections:

1. previous computer and information use, perceived knowledge, significant others in providing emotional support, family and living arrangements,
2. a Hospital Anxiety and Depression Scale (HADS),
3. a Miller Coping Style questionnaire,
4. Helgeson's measures of social support and negative interactions.

Completion of this questionnaire should have taken about half an hour at home.

Confidants: The researchers aimed to recruit 400 'confidants', one for each patient (but estimated that it were more likely to recruit (say) 60% of 400. Patients were to be asked if they were prepared to pass on the baseline questionnaire to a person selected by them as the person they are most likely to use to seek emotional support. This may be a husband, wife, partner, but may also be a daughter, son, or friend. The confidant questionnaire has sections:

1. demographics and relationship to patient, newspaper read, frequency of discussion with patient, perceived knowledge
2. HADS,
3. Miller coping style.

The researchers did not know what proportion of patients would be prepared to pass on the questionnaire or what response rate would be received from confidants. Rees in her work with daughters of women with breast cancer found that 93% of patients passed the questionnaires on and 77% of daughters responded. If we got an overall 60% response (ie 240 before and after questionnaires) this would give us 80% power to find a difference between 40% and 23% who had improved anxiety between (eg) 120 in personalised group vs 120 in general group. The confidant population would be less than the patient population. Patients will not be excluded if there is no corresponding 'confidant'.

Randomisation for the three factors: After recruitment the RA was to randomise for each of the three binary factors (2X2X2 factorial design):

1. Interactive v Automatic;
2. Medical Record v General;
3. Anxiety reducing advice v no advice.

Patients were to randomised to the eight sub-groups in blocks of 32 and separately by breast vs rest. The eight (2X2X2) subgroups, each receive different styles of leaflet. These are detailed further below.

Baseline Data: Collected from case notes was to be severity, length of onset of cancer, treatments so far (surgery, radiotherapy, chemotherapy), date of birth, gender, address, deprivation category from postcode for all patients, and a problem and treatment list for patients in the 'medical record' groups.

Intervention: The delay between recruitment and intervention was to allow patients time to think about their participation and withdraw if desired. At the intervention, patients were to produce a report about their cancer. Patients in the 'automatic group' would not need to use the computer.

• Interactive vs automatic. Those in the 'interactive' groups were to choose from menus, see information on the screen, and press an 'add to report' button on information screens for that information which they want to include in their report. Patients' choices were to be recorded by the computer.
• Those in the medical records group were to have personal information included based on their medical record. They were to either have information automatically selected and formatted by the computer, or to see it on screen and choose what they want to include.
• Lastly, half the patients were to have an additional section offering advice for the reduction of anxiety. This section was written based on previous work in cognitive behavioural therapy for anxiety.

All reports were to include colour photos (for example of treatment rooms, staff), and diagrams (for example, anatomical) of good quality printing etc. The patient were to be given the report to take away.

Three month follow-up: At three months patients were to be sent a follow-up questionnaire with the following sections:

1. questions about their satisfaction with the information received, their use of their reports, their interactions with their family and friends, and the influence of the reports on these interactions, their perceived knowledge.
2. HADS,
3. Helgeson's social support questionnaires.

Significant others will be sent a questionnaire which has sections on:

1. questions about their satisfaction with the information they have received,
2. HADS

Measures of Outcome:

• Change in anxiety (for patient or confidant) between intervention and 3 month follow-up (HADS)
• Change in social support (for patient) between intervention and 3 months (Helgeson)
• Patient satisfaction with information score at 3 month follow-up
• Confidant satisfaction with information score at 3 month follow-up
• Whether patient had read report, shown it or discussed it with the named confidant, other family members or friends and whether the discussion was aided by use of the report.
• What patients liked and disliked about the reports.
• Amongst the interactive group, choices made between presentation styles and how well these would correspond to automatic computer choices.
• Patient and staff time, and hence costs, needed for routine delivery of each intervention will be modelled on a variety of assumptions.

Analysis: The researchers were to produce a CONSORT patient flow diagram and compare losses to follow-up by the three interventions (A,B,C). Analysis of outcomes was to be for those patients completing both intervention and follow-up. The researchers will focus on the following:

• Comparison of outcomes 1-4 by interventions A, B, and C while controlling for age, gender, cancer site, time since diagnosis, severity, coping style, using analysis of variance (factorial design) and multiple logistic regression.
• Comparison of outcomes 5 and 6 by interventions A,B, and C (qualitative).
• Modelling of anxiety (outcomes 1-2) on use of report at home (outcome 4), perceived social support, and interventions A,B,C.
• Frequencies of choices (outcome 7) and comparison with automatic selection, both statistically and through qualitative description of the differences.
• Comparison of patient and NHS costs (outcome 8) will be made between the methods used in this trial and other methods of providing information.
• Comparison of anxiety change in patient and significant other.
• Analyses will be on those completing follow-up. No substitution of data will be carried out for those randomised but not completing follow-up. The researchers will thus be able to make provisional conclusions on how the different types of information provision influence interaction, anxiety and patient satisfaction, and on the correlation between the use of information with family, and anxiety; The researchers will have information on patient selection to inform the better development of automatic systems, and costs to inform practical deployment.