This study aims to understand the impact of OWise on health behaviours, HRQoL and health care utilisation in early-stage breast cancer patients compared to standard care alone.
We will evaluate the effectiveness of OWise using a multi-centre, individually randomized, parallel controlled trial recruiting 122 patients. The intervention group will receive OWise plus standard care, while the control group will receive standard care alone to assess superiority. Due to the nature of the digital tool, it is not possible to blind participants or providers. Outcomes are reported directly by participants and analysis depends on the randomly assigned group. Therefore, outcome assessors and data analysts are not blinded either. Patients in both groups will complete patient-reported outcome measures (PROMs) to assess outcomes at baseline, 3 months, 6 months and 12 months from diagnosis. See Figure 2 for the SPIRIT diagram showing the schedule of enrolment and assessment and Additional File 1 for the SPIRIT-PRO checklist.
Patients will be randomly assigned (1:1) to the intervention or control group [19,20]. Randomisation will be stratified by age group and centre. Age is grouped by (1) under 60 years old and (2) 60 years old and over as internet access drops between age groups 45–55 and 55–65  and the incidence of breast cancer in the UK is evenly distributed around age 60 . The Institute of Cancer Research Clinical Trials and Statistics Unit Randomisation Service will generate the randomisation sequence and allocate the group by phone.
Females (aged 18 years or over) newly diagnosed with early-stage breast cancer as a first primary diagnosis will be eligible to take part. Eligibility was restricted to early-stage and first primary diagnoses as metastatic patients may have confounding care and psychosocial experiences and patient activation naturally increases with time after a breast cancer diagnosis . Patients must complete the baseline measure before starting anti-cancer treatment. All participating sites are located in the UK, a list of which can be found on the registration website. Exclusion criteria include private care, difficulty reading in English, significant cognitive impairments or poor mental health and no internet access.
OWise is an mHealth tool accessible online or by mobile application . The tool offers tailored medical information, a modifiable question list with tailored recommended questions, a medical terms glossary, useful links to local resources, a tracking tool for symptoms, an appointment calendar and a consultation recording device.
In the study, only patients randomised into the intervention group will receive information about OWise. The individual enrolling the patient will provide instructions for creating an account and navigating the tool. Participants will be free to use the tool as much as they wish to mimic real-world use of the application. The tool is free to download and accessible to the participant beyond the study period.
Participants in the control group will receive all standard information including leaflets and links to resources that patients usually receive at the time of a new breast cancer diagnosis. Participants in the control group will not be given information about the tool but will also not be explicitly prohibited from using the tool.
The primary objective of this study is to test whether use of OWise increases patient activation scores at three months follow-up by at least four points more than standard care.
(1)To test whether any difference in the change in patient activation between the two groups still exists after controlling for potential covariates.
(2)To test whether the use of OWise leads to a smaller decrease in health status at three months follow-up than standard care after controlling for potential covariates.
(3)To test whether the use of OWise leads to a smaller decrease in HRQoL at three months follow-up than standard care alone after controlling for potential covariates.
(4)To test whether the use of OWise leads to a lower increase in psychological distress at three months follow-up compared to standard care alone after controlling for potential covariates.
(5)To test whether the use of OWise reduces the rate of resource utilisation in the first year following diagnosis compared to standard care among patients registered in Discover, an integrated health and social care record in North West London.
(6)To test whether the use of OWise reduces the average cost per patient in the first year following diagnosis compared to standard care among patients registered in Discover.
(7)To describe the change in patient activation in the intervention group compared to the control group in the first year following diagnosis.
(8)To describe the level of OWise uptake in the intervention group in the first year following diagnosis.
(9)To describe the change in the pattern of patient activation, HRQoL, psychological distress and health status in the intervention group compared to the control group in the first year following diagnosis.
We will continuously sample all patients meeting eligibility criteria diagnosed within the recruitment period. A member of the clinical team will identify eligible patients in multi-disciplinary team meetings or clinic lists and invite potential participants at diagnosis. The name of the digital tool will not be disclosed when inviting patients. If a patient shows interest, a researcher will provide further information either in-person or over the phone. Patients can decide to take part any time before starting anti-cancer treatment.
After meeting eligibility criteria and providing written informed consent, participants will be randomised. The researcher will inform the participant of their allocation and provide instructions for accessing the online PROM collection tool. Participants in the intervention arm will be required to complete the baseline measure before using OWise.
Primary outcome measure
Patient Activation Measure (PAM–13)
Patient activation describes the knowledge, skills and confidence a person has in managing their health and care [23,24]. The PAM–13 is a 13-item questionnaire that measures patient activation . Each item has four response options from (1) ‘strongly disagree’ to (4) ‘strongly agree,’ and ‘not applicable.’ PAM–13 scores will be calculated according to the guidelines . Scores range on a scale of 1–100 corresponding to four activation levels: 1 (≤47.0) not believing activation important, 2 (47.1–55.1) a lack of knowledge and confidence to take action, 3 (55.2–67.0) beginning to take action and 4 (≥67.1) taking action . This measure has been used widely among cancer patients and across the UK [26–28] and has robust evidence of reliability and validity [23,29,30]. Patient activation, as measured by the PAM–13, can be targeted by interventions and change over time . Previous work has shown that higher patient activation is associated with HRQoL and lower health care utilisation [16–18,31].
Secondary outcome measures
European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire - Core30 (EORTC-QLQ-C30 version 3) and Updated Quality of Life Breast Cancer Module (QLQ-BR45)
This 30-item instrument measuring HRQoL has five functional scales (physical, role, cognitive, emotional and social), a global quality of life scale, eight symptom scales or items (fatigue, pain, nausea and vomiting, dyspnoea, loss of appetite, sleep disturbance, constipation and diarrhoea) and a single item assessing perceived financial impact . The QLQ-BR45 contains five functional scales or items (body image, future perspective, sexual functioning, sexual enjoyment and breast satisfaction) and 7 symptom scales or items (systemic therapy side effects, upset by hair loss, arm symptoms, breast symptoms, endocrine therapy symptoms, skin mucosis symptoms and endocrine sexual symptoms) . Scores will be calculated according to EORTC guidelines . All scores range from 0–100. Higher scores on functional scales and global quality of life indicate better function and HRQoL, respectively. Higher scores on symptom scales and items indicate higher symptom burden . The measures have strong evidence of validity and reliability in early-breast cancer patients and have been used in a number of clinical trials allowing for comparisons [33,36].
Hospital Anxiety and Depression Scale (HADS)
This 14-item questionnaire measures psychological distress, with seven items assessing anxiety and seven items assessing depression with the summed total score reflecting the level of psychological distress . Three continuous scales will be calculated (anxiety, depression and overall psychological distress) according to HADS guidelines [37,38]. Higher scores indicate more psychological distress [37,38]. The HADS has evidence of reliability and validity in early breast cancer patients [36,39].
EuroQol 5-Dimension 5-Level (EQ–5D–5L)
This instrument assessing health status consists of five items and a visual analogue scale . The items cover five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression). Each dimension has five response levels from ‘no problems’ to ‘extreme problems.’ The visual analogue scale records the patient’s self-rated health from 0–100, with the highest score indicating ‘The best health you can imagine’ and the lowest score indicating ‘The worst health you can imagine.’ Responses to each item combine to form a five-digit number that describes the patient’s health state. A corresponding index value will be assigned according to a recent valuation study conducted in England . The EQ–5D–5L has a large base of evidence and for validity and reliability in breast cancer patients and can also be used to conduct economic analysis .
Health Care Utilisation
Health care utilisation will be assessed using data routinely collected in Discover. The Discover linked dataset includes coded information on health and social care resource utilisation of individuals registered with a GP practice in the North West London region. Information is collected about the number and type of appointments from primary, secondary and social care for each patient between diagnosis and one year follow-up.
Health Care Costs
The Discover linked dataset also provides the current costs of health and social care to Clinical Commissioners in North West London based upon Commissioner local pricing. Information on the cost of each type of appointment is calculated routinely and collated together across health care settings to provide a measure of health and social care utilisation of each patient.
With patient informed consent, we will evaluate OWise uptake by reviewing timestamps that indicate logging in or modification of a specific function. This information will allow us to evaluate whether participants use the tool, which function patients use and how long the tool is used for.
Patient responses to items at each measurement time point will indicate contamination. A set of 19 items will ask participants to identify the use of supportive care services including self-management mobile phone applications or websites. If a participant says yes, we will ask them to name the source in a free-text box to determine whether or not OWise has been used. Prior to recruitment close, the statistician will assess the level of contamination and increase the sample size commensurately .
The study steering committee determined a data monitoring committee was unnecessary for this study as it poses a minimal risk to patient safety and uses only routinely collected or patient-reported data.
Patient-Reported Outcome Measure Data
Participants will complete PROMS using PROFILES (Patient Reported Outcomes Following Initial treatment and Long term Evaluation of Survivorship), an online PROM collection and data management system developed in the Netherlands and implemented at the Royal Marsden . Follow-up time points will be managed by the Royal Marsden as the coordinating centre. PROM responses cannot be viewed by the researcher or clinical team until extracted at the end of the study when it will be linked to other study data by the study identification number.
With patient informed consent, researchers will collect relevant clinical data from the local electronic patient record and store it digitally at the Royal Marsden. Clinical data will be extracted at the end of the study and linked to other study data by the study identification number.
Health care utilisation
Study participants will be identified in Discover by NHS number and flagged with the study identification number. The Imperial College Health Partners Discover team, will access a de-identified version of the data to analyse health and social care resource utilisation on behalf of the Royal Marsden. Health and Social Care resource utilisation data will be extracted at the end of the study and linked to other study data by the study identification number.
Participants will be provided with a unique invitation code linked to their study identification number to input when creating an OWise account. Timestamp data will be identified by the invitation code and extracted at the end of the study. The data will be linked to other study data by the unique invitation code.
Sample size calculations are based on the change in PAM–13 score from baseline to three months. A difference of four points is considered clinically relevant [29,45]. In a similar study, the mean PAM–13 score at baseline for the intervention group was 61.3 (SD 16.61) and 67.9 (SD 16.85) at three months . For the control group, the study reported a mean PAM–13 score at baseline of 62.1 (SD 17.30) and 62.8 (SD 14.94) at three months. Based on these findings, this study is planned to detect a mean change difference of 5.90 assuming a common standard deviation of 10.0. Using an 80% power the study will recruit 47 patients per group. This was calculated using a 2-sided test with alpha = 0.05. We will increase the sample size taking 23% attrition at three months into account (47/(1–0.23)) to 61 patients per group  and, as mentioned above, increase the sample size accordingly if contamination is found near the end of recruitment.
The CONSORT-EHEALTH recommendations for reporting randomised trials for developing and evaluating eHealth interventions will guide trial reports . Primary and secondary outcomes will be assessed using intention-to-treat analysis where all participants are analysed according to the arm to which they were randomised. We will conduct a sensitivity analysis separately as described below. The data will be analysed after all patients have completed the one year follow-up. Data will be reported descriptively at each time point. Mean and standard deviation or median and range will be reported for continuous outcomes. Frequency and percentage will be reported for categorical outcomes.
The association between potential covariates and the primary and secondary endpoints will be explored using univariate analysis. Any variables with a p-value of <0.1 will be included in the multivariable model. Multivariable analysis controlling for potential covariates associated with the particular outcome will be conducted using logistic regression for binary outcomes and multiple linear regression for continuous outcomes. Two-sided p-values of <0.05 will be considered statistically significant.
We will compare the PAM–13 score change between the intervention arm and the control arm using independent t-test. Data will be log transfored to achieve normality as appropriate. We will also compare the mean change in PAM–13 score in the intervention and control arm in a multiple linear regression model including potential covariates.
We will compare the mean change inEQ–5D–5L index score and visual analogue score, EORTC-QLQ-C30 and BR45 scale scores in the intervention and control arm in simple and multiple linear regression models including potential covariates.
We will compare the mean change in the three HADS scale scores in the intervention and control arm in simple and multiple linear regression models including potential covariates. Based on the continuous overall psychological distress score, patients are classified as ‘distressed’ when they have a score of ≥8, and ‘not distressed’ when they have a score <8. Frequency, percentages and any appropriate 95% confidence intervals of this dichotomization at baseline and 3 months will be presented. Chi-square or Fisher’s exact test will be used to compare the level of distress between the intervention and control arms.
We will present the mean rate of resource utilisation and cost per patient in the two groups by type of resource (primary, secondary and social care) and for total NHS resources used. Simple and multiple linear regression models including potential covariates will compare the mean rate of total resource utilisation and the mean cost per patient in the two groups.
We will describe the average scale scores of the four validated measures in the two groups at the four time points and show graphically the trend in scale scores in each group. We will also compare the mean change in scale scores of the measures across the four time points between the intervention and control arm using a mixed models approach.
To describe OWise uptake, the average number of times logging in at daily intervals throughout the follow up period will be described and the trend of mean logging in over time will be graphed. We will also show the average frequency of use for each function of the tool over time.
Per protocol analysis will be performed as a sensitivity analysis to assess the impact of contamination on the primary analysis. In the sensitivity analysis, all participants in the control arm that report using OWise will be excluded. If the sensitivity analysis produces results dissimilar from the primary analysis, we will determine the primary results are not robust and further research is required.
Missing data of multi-item scales will be handled according to questionnaire guidelines. Where guidelines are unavailable, items will be mean-imputed if at least half of the items from the scale are answered. Descriptive statistics are based on complete case analysis. We will analyse available data before imputation for the groups comparison and use the complete case data as a form of sensitivity analysis.
Any protocol modifications will be submitted for approval to the research ethics committee, reflected in the online registration and disseminated by email to site principal investigators and trial coordinators. To mitigate attrition, the coordinating centre will engage participants with newsletters via email or post. These will also discuss any changes to study procedures relevant to participants and results of the study. Each party involved will continue to own the data they collected, i.e. The Royal Marsden will own the clinical and PROM data, Discover on behalf of the NWL data custodians will own the health and social care resource utilisation data, and Px Healthcare will own the OWise uptake data. The statistician and health economists will have access to the final linked trial dataset. There are no plans to provide public access to the full protocol, participant-level data or statistical code. The researchers aim to publish results in a peer-reviewed journal and share via social media and conferences. Authorship will be determined by the owner of the data included in the publication.