We offered D-IPC to 2016 consecutive patients scheduled for a contrast-enhanced CT during a study period of 20 months (June 2018–February 2020). All examinations were carried out in a large university hospital performing around 34,000 CT examinations per year. Pediatric patients, emergency patients, patients with severe sight disorders, and unconscious patients were generally not considered study participants (Fig. 1). In addition, patients who refused to perform mobile tablet workflow were not assigned to D-IPC.

Study flowchart for tablet-based, digitized informed patient consent (D-IPC). Before the tablet-based workflow started, the indication for CT was evaluated by the attending physician. The patients were then asked to answer a dedicated questionnaire in the waiting room with 22 main items. Structured patient data based on D-IPC was available on all workstations for informed consent discussion and reporting process

Before assigning patients to D-IPC, the attending radiologist evaluated the justifying CT indication based on the information provided by the referring physician and the already-archived information in the radiology information system (RIS) and the hospital information system. Patient master data were automatically transferred from the RIS worklist. The front desk in the radiology department allocated the contrast-enhanced CT questionnaires to the present patients using a desktop application of the D-IPC software (MEDePORT, Thieme Compliance GmbH, Erlangen, Germany). Patients were registered by manually typing the patient name and birthdate or in most cases by scanning a barcode from the web application on the desktop computer. The touch screen interaction was made available on a fleet of 10 mobile tablet devices (Surface Go, first-generation model, Microsoft Co., Redmond, WA, USA) using a dedicated patient-client application of the D-IPC software operating in the kiosk mode. We briefly introduced each patient, who stated sufficient German language skills, to the tablet functionality before starting the questionnaire with 22 main items in the waiting room without further assistance (e.g., by family members). The questionnaire consisted of items considering potential contraindication for contrast-enhanced CT (n = 7) and items about general patient history (n = 15). Altogether, the 22 main items are contributing to justify a safe CT indication (Tables 1, 2). The patients voluntarily answered each question. There was no constraint in answering every question. The attending radiologist discussed unanswered, critical questions during the personal informed patient discussion.

The first four general questionnaire items consider patient gender, age, body weight, and height (Table 1) but do not consider specific patient history or contraindications for CT. Clinically relevant issues for justifying CT indication (red flags; Table 2) are only covered with five or more answered questionnaire items. Therefore, patients with four or fewer answered questions were excluded from the study after being assigned to D-IPC. All excluded patients underwent assisted C-IPC. We compared the age between the study patients and patients, who were unable to finalize D-IPC after already being assigned.

In addition to general patient history (Table 1), critical questions concerning relative and absolute contraindications were defined as red flags (Table 2). The patient-client functionality also allows the patients to mark unclear questions for personalized discussion with the physician.

After completion, the tablet devices were returned to the radiographer or the front desk, and the tablet cover and screen were immediately disinfected with dedicated wipes (Cleanisept® wipes forte, Dr.Schumacher, Malsfeld, Germany). The patients were then sent to the preparation room for the patient–physician discussion.

The on-premises software in the German language was installed on a local virtual server, and all data were stored in a local database. The attending radiologist used a dedicated physician client on two mobile tablet computers with a digital pen (Surface Pro 5, Microsoft Co., Redmont, WA, USA; Fig. 2) to review the questionnaires. Updates of the patient worklist were received automatically every 30 s from the RIS. Predefined filters are available at the physician's discretion: Red flags, individually marked questions, questions answered with yes, questions answered with no, and all questions. Red flags are always displayed first. Therefore, the informed consent discussion starts with a structured overview of potential contraindications. An illustration of the user interface is available in the literature [16]. Based on D-IPC recordings, a PDF document is generated and signed on the physician's tablet by the physician and the patient. Changes in content are impossible after the PDF is signed. The finalized PDF is automatically transferred to the enterprise picture archiving and communicating system (ePACS, synedra View Diagnostic, synedra information technologies GmbH, Innsbruck, Austria). Full access to the physician client, including all recorded data and filters, was available on all reporting stations. As the D-IPC software did not provide time stamps, we cannot provide precise measurements for IPC duration. However, for the cost–benefit analysis, we estimate the mean D-IPC and C-IPC duration to be approximately five minutes.

Digital signature of structured informed patient consent. Each patient and the attending physician digitally signed and confirmed the completeness and validity of the patient data in the preparation room

The carbon dioxide (CO2) footprint for office paper sheets is reported in the literature as CO2 equivalents per sheet of paper with values of 4.29–4.74 g CO2eq [17]. Each avoided conventional IPC consisted of two sheets of office paper. Based on these assumptions, we calculated reduced CO2eq emissions achieved by paperless D-IPC. On the other hand, we also calculated the energy consumption and energy costs of the tablet devices (Table 5). Also, the tablet vendor provides estimated CO2eq emissions for the production process of Surface Pro (121 kg CO2eq) and the Surface Go devices (107 kg CO2eq) [18].

We calculated a dedicated cost–benefit analysis (CBA) for D-IPC versus C-IPC for a 5 years project duration. For this project, we assumed a constant amount of 2016 informed patient consent forms for each year (Table 5). Our approach differentiates (1) personnel, (2) IT service, (3) hardware and (4) other operational costs. In addition, we compare the performance features of D-IPC and C-IPC for quality of care and patient safety.

We estimated an average of 5 min for our assistant staff to prepare D-IPC or C-IPC and to archive C-IPC. Moreover, we estimated the average time effort for the patient–physician discussion to be 5 min. The prevalence of extensive search procedures for IPC retrieval and consecutive CT vacancy is assumed 1%. We interviewed the technicians and the support staff about the average time effort of the search process. An average time effort of 10 min was specified. All IT service costs are specified by the vendor and the local IT service department. We used current wholesale pricing in Europe to estimate hardware costs for the tablets (Surface Pro; n = 2; Surface Go; n = 10). Energy consumption measurements of the physician`s tablet are provided by the vendor [18]. For each device, the estimated service duration per day was seven working hours, which results in an average energy cost of 5.3 Euro (31 Euro cents/kilowatt-hour). Paper costs were 1 cent for each sheet of paper based on current wholesale pricing in Europe. Printing costs were 7 cents per colored page. Each conventional informed consent form consists of two sheets of paper (4032 sheets). Professional disposal of patient data is crucial. However, disposable in Germany is allowed after 30 years of data storage. Therefore, this matter of expense does not apply to this study. We analyzed D-IPC and C-IPC workflow to identify important performance features.

The vendor of the D-IPC software provided a customized script for structured database export as an extensible markup language file (.XML). We processed this data with the MS Excel 2016 software package (Microsoft Cooperation, Redmond, WA, USA) and analyzed the prevalence of red flags, marked questions, and the rate of answered questionnaire items. Also, we compared for each patient the prevalence of documented red flags in C-PR and D-IPC. The information documented with C-PR included data from previously collected analog informed consents, the written requests from the referring physicians, and the documented risk factors and complications from previous examinations, each directly documented in the RIS. The information archived during D-IPC for the same patient was retrieved from the.XML file. We also evaluated the completeness of clinical context information and the examination-related clinical questions for each CT.

If applicable, we provide mean and standard deviation in the case of normal distribution. Median and interquartile range (IQR) with 75th and 25th percentiles are given when normal distribution was not assumed. Total patient counts for several questionnaire items do not sum to the same number for every category because multiple answers for pre-existing conditions or medications were allowed for most items. Moreover, there was no constraint in answering every questionnaire item. Normally distributed data were compared with paired t-tests. The Wilcoxon signed-rank test was used for further analysis if the normal distribution was not assumed. The significance level was defined as p < 0.05. We performed statistical analysis with the software package SPSS Statistics Version 21 (International Business Machines Corporation [IBM], Somers, NY, USA).

The study complied with the Declaration of Helsinki and was approved by the local institutional review board.