Picture Naming in the Context of Image Type and Response Variety in People With and Without Aphasia

Qulity assurance:

To verify the presence and reliability of all subject data, the following items were checked during these visits using a checklist: Presence of informed consent, presence of all data requested in the questionnaire, correctness of the data copied from the source files (e.g. test protocols) and traceability of the data flow. At the end of the experiment, the correct storage of the data (informed consent of the study participants, identification list, raw data) were checked. All involved staff who had access to the clinical data were informed and trained about anonymization, ethical aspects and correct data manipulation and storage. Direct access to the source data and all relevant documents and records is guaranteed to the ethics committee upon request and presentation.

Data dictionary:

• Questionnaire on personal and health-related information of the experiment participants: Frequencies, descriptive statistics
• The Pyramids and Palm Trees test (PPTT) (Howard & Patterson 1992): passed/not passed: scores above 90% indicate correct cognitive semantics, scores below indicate impaired; 90% of the images must be correctly matched, i.e., at least 47 of 52 items.
• Colouring of Pictures test (CoPT) (De Renzi et al. 1972): passed/not passed: In order to exclude the presence of an acquired color diagnosis due to the brain injury suffered by participants with aphasia, the ability to assign colors to objects is tested with a coloration test.
• Picture Naming Test: Number of correct responses and naming latencies when picture stimuli are presented in two language varieties (Swiss dialect and Swiss standard language)

Data recording and source data:

Case Report Forms (CRF) were printed out on paper and given to experiment participants and practice partners to collect personal and health-related data. Personal and health-related source data was collected in hard copy in the CRF and hard copies were given to study participants and practice partners. Data from pretesting (PPTT, CoPT) for experiment participants with aphasia were recorded manually on the CRF, as were any comments about abnormalities during the course of the study. Data needed for analysis of naming correctness and naming latency (audio and, if consent is given, video data) were collected electronically. Data on naming correctness were also collected on the CRF. The qualitative question-guided interview at the conclusion of data collection was recorded electronically, with accompanying key words noted on the CRF.

AAT scores (from Aachener aphasia test scale) as well as other routinely collected data from patient records of participants with aphasia on the type of aphasia present was manually transferred to the CRF by the speech therapists of the collaborators. The CRFs in paper form are digitized by study staff by scanning and manually transferred to electronic data (filling in Excel lists).

Data protection and data security: confidentiality and encryption The collection, transmission, storage and analysis of health-related personal data of this experiment was carried out in accordance with Art. 5 of the HFV (Humanforschungsgesetz) and in strict compliance with Swiss legal data protection regulations. The project data was kept strictly confidential and was only accessible to authorized personnel who needed the data to fulfill their tasks within the research project. All staff were informed about data protection and confidentiality and were subject to the legal requirements. In the CRF and other project-specific documents, participants were only referred to by a unique identification code, which could only be reconstructed by the participants themselves, but which did not allow any conclusions to be drawn about the participants. The participant identification list as well as the signed informed consents were stored in a secure location at the collaborators and was only transferred to the FHNW after the end of the study, where they will be kept in a safe.

Participant data was collected using an application that was programmed for the experiment under the unique participant identification number only. Other relevant personal and health data was collected in paper form in the CRF. Audio recordings were made of study participants during the naming test and, with the consent of the study participants, they were also filmed; the image and audio sequence were stored in each case. The video recording data was only used for further research, education and training purposes with the prior consent of the experiment participants. All subject-related data was encrypted and stored on the tablet's SD card. The SD card was personally transported to the FHNW by the person conducting the study together with the CRF, where it was handed over to the study director. The information was then decrypted and analyzed on an FHNW computer. The decrypted data was stored securely on three external storage media. One was kept as a backup in the FHNW safe, and two others were used for data analysis at the Academy of Art and Design FHNW and at the School of Education FHNW. These external hard drives were kept securely in safes at the respective institutes when not in use. No original data was stored directly on the FHNW computers. Only the results of the data analysis were stored on other personal computers or secured internal servers. The encryption was managed by the project management. Only a few authorized persons know the key. A list is kept of these persons.

Storage of data:

Data obtained from the research project (raw data sets, analysis results, and anonymized documents) will be archived for at least ten years, protected from access, to ensure scientific sustainability. One hard drive is kept in the FHNW safe, one in appropriately secure locations at the IVK (HGK) and ISP (PH) institutes of the FHNW. Only authorized persons have access. Participant identification lists are destroyed. Should the data be destroyed after a minimum of ten years, all electronic data is deleted and the storage area is securely overwritten. Paper sheets such as the CRF are rendered unusable with the file shredder.

Sample size assessment:

Determining the sample size: The power analysis is based on a simulation. A baseline accuracy (e.g., the percentage correct on stimulus type 1, variety 1) for each of the n virtual participants was first defined.A certain effect size for each of the two factors, stimulus and variety was defined. Both factors are assumed to have an effect as large as an odds ratio of 1.3. This size of odds ratio was used to define the odds for each participant and thus the precision in the other cells of the design. Both effects are analyzed for each simulated data set and then analyzed using a generalized linear mixed-effects model (GLMM) with a binomial distribution of outcomes and a random intercept per participant. Power is calculated as the proportion of simulations for which this statistical model detects an effect of both stimulus type and response variety.

Plan for missing data:

Only those data sets in which participants have completed at least two cells of the design (stimulus type x response variety) are included in the analysis. To counteract dropout of experiment participants - caused by withdrawal of experiment participants, by technical problems in data collection, or by insufficiently collected data - more participants are preemptively recruited than statistically necessary (40 instead of 36 subjects per group) to ensure that there will be enough data to represent an effect, should there be one.

Statistical analysis:

The proportion of correct responses in each cell of the design (stimulus type x response variety) is analyzed using a generalized linear mixed regression model with a binomial distribution of outcomes, including a random intercept estimated for each participant.