This prospective single-center study was performed at Heidelberg University Hospital between August 2018 and January 2020 (Trial registration: ClinicalTrials.gov NCT03168048; https://clinicaltrials.gov/ct2/show/NCT03168048 [archived at WebCite http://www.webcitation.org/6wtWGgi0X]). Patients were screened in the outpatient clinic of the Department of Radiation Oncology, University Hospital Heidelberg. Inclusion criteria were an indication for radiotherapy to the chest or pelvis in curative intent, age of 18 years or older, a good general performance score (Karnofsky performance index ≥ 70%), and a generally outpatient course of treatment. Patients who did not fulfill these criteria or who did not provide written informed consent were excluded from the study.

The goal of this study was to evaluate the feasibility and acceptance of app-based treatment surveillance of patients undergoing curative RT. To evaluate this metric, we designed a treatment-support web-application in cooperation with OPASCA GmbH Mannheim, Germany. The main component of the app was the query of items of validated EORTC (European Organisation for Research and Treatment of Cancer) and RAND (research and development) Corporation questionnaires in a predefined sequence, daily during RT. These included assessments of disease-related quality of life as well as treatment-specific symptoms and functions, patient satisfaction, and overall well-being. According to standard of care at our department, patients received a planned doctor’s consultation appointment every week. The app featured the option to request an additional consultation appointment on demand, which was generally granted on the same day. The visiting clinician had access to the patient’s app inputs to help and guide all planned and requested appointments.

The treatment-support app was run on a department-owned tablet device, which was handed to participants during the waiting period for their daily RT appointments. During this time, they were asked to complete the items presented on that day and were offered to request a consultation appointment. The app design was very intuitive and easy to use, and no special patient education was required (screenshot example displayed in supplementary Multimedia Appendix 1). If required, radiation therapist personnel provided assistance. Participants spent around 5–10 min per day with the app and handed the tablet device back to the RT technician at the initiation of each RT session. This approach with a central input device was necessary for this pilot study due to infrastructural and data security reasons, as well as local policies. Future iterations of the app, however, will run on the participants’ personal mobile devices.

Treatment duration, dose prescription, and technical aspects of RT followed the general standard of care at the radiation oncology department of Heidelberg University Hospital. RT was performed once daily, with five fractions per week (Monday through Friday). Participation in this study did not affect the planning or course of RT. Treatment was performed in an outpatient setting, with the exception of patients receiving concomitant chemotherapy. In those cases, patients could be admitted to a hospital ward for typically 1–2 days per dose of chemotherapy, if necessary.

Within the treatment-support app, patients were asked to complete the EORTC QLQ-C30 questionnaire on the first and last day of RT, as well as the RAND PSQ-18 questionnaire on the last day of RT. Five to seven items of supplementary diagnosis-specific EORTC QLQ modules were presented daily during RT, repeating after 3–6 days (Fig. 1). The supplementary modules used were BR23 for breast cancer; CX24 for vaginal, cervical, or endometrial cancer; CR29 for rectal cancer; LC13 for lung cancer; and PR25 for prostate cancer patients. In summary, patients received between 115 and 291 app-based questions over the course of RT, depending on treatment duration and the modules used. The primary exploratory endpoint of feasibility was defined by the fraction of patients who completed at least 80% of all app-based questions. Secondary endpoints were disease-related quality of life and patient satisfaction. Also, patients were able to request a doctor’s consultation via the app, resulting in more thorough medical care compared to established treatment standards. A sample screenshot of the app surface is provided in the electronic supplementary materials (Multimedia Appendix 1). In the context of patient screening, we inquired about the use of mobile devices for personal and medical purposes using a self-designed smartphone questionnaire. These screening items were paper based and did not count towards the primary endpoint.

Study visits and corresponding questionnaires; the number of visits during RT varied due to length of treatment and instruments used

All data collected within this study was pseudonymized and stored on a central department-owned on-site server to which only the study conductors had access. The only exception to this was in the case of a requested doctor’s consultation appointment, where all app-based inputs where automatically forwarded to the visiting clinician for guidance. After study completion, a database extract of the study data was archived in the clinic’s clinical study archive and deleted from the server, which is being used for further development.

Patient-reported quality of life was evaluated with the EORTC QLQ-C30 questionnaire and its supplementary modules. The QLQ-C30 is a widely used tool for the measurement of quality of life in cancer patients provided by the European Organization for Research and Treatment of Cancer (EORTC). It consists of 30 items assessing five functional (physical, role, emotional, cognitive, social) and nine symptomatic (fatigue, nausea and vomiting, pain, dyspnea, insomnia, appetite loss, constipation, diarrhea, financial difficulties) aspects of health-related quality of life in 15 subscales, including a global health status scale. Higher scores on a functional scale express better respective functioning, while higher symptom scale scores indicate a higher symptomatic burden. Overall, the QLQ-C30 has been found to be highly reliable and consistent in the measurement of quality of life and is well established in cancer research [16]. While it provides a valid general evaluation of quality of life, several supplementary modules have been established for the assessment of more diagnosis- and treatment-specific issues patients might encounter. Generally, the supplementary modules follow the same scoring procedures as the QLQ-C30, providing additional disease-specific functional and symptom scales [17, 18]. For example, the breast cancer module BR23 contains items on body image and sexual functioning, while the lung cancer module LC13 queries typical symptoms like coughing, hemoptysis, and chest pain. Use of the QLQ-C30 within a mobile app has been found to be well accepted by patients [19].

The patient satisfaction questionnaire short form (PSQ-18) was used to assess patient satisfaction at the end of RT. It is a patient-reported measurement tool for satisfaction with medical care, provided by the RAND (research and development) Corporation as an 18-item short form of the 50-item PSQ-III [20]. Items are scored on a five-point Likert scale and can be grouped into seven dimensions of treatment satisfaction (general satisfaction, technical quality, interpersonal manner, communication, financial aspects, times spent with doctor, and accessibility and convenience). Response values are converted to score values, so that a higher score represents greater satisfaction. The PSQ-18 has been proven to be internally consistent and reliable, and is substantially correlated with its corresponding long form.

All work in relation to this study followed the Declaration of Helsinki [21] and was approved by the institutional ethics committee. Patient involvement was voluntary, and no disadvantages resulted from declining participation. Informed written consent was obtained prior to enrollment in the study. Participants were provided detailed information on the collection and storage of data, as well as the option to withdraw consent at any time during the investigation. All personal information of participants was pseudonymized after data collection.

Descriptive statistical analysis of EORTC questionnaires was performed using R version 4.0.2 and the supplemental packages QoLR and PROscore (R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria; URL https://www.R-project.org/). For the QLQ-C30 and its supplementary modules, a linear transformation of raw scores was performed to achieve values between 0 and 100, in accordance with the EORTC manual. Missing items were imputed by mean of the other scale items if clinically reasonable and if at least half of the items had been answered. Patients who answered fewer items were omitted for that timepoint.

In subgroup analysis, chi-squared test and t-test were preformed to test for significant differences between the groups. A p-value of < 0.05 was deemed statistically significant. Wilcoxon’s one-sample signed-rank test for paired data was used for the evaluation of the QLQ-C30 questionnaire.