A literature search was conducted to review existing validated IPE questionnaires [11-14]. Although these surveys had notable and overlapping IPE domains, including communication, team functioning, and collaboration, they were not designed specifically for the evaluation of education in virtual environments. As such, the authors generated the initial item pool via question development and modification of existing validated questionnaires, including questions from the Interprofessional Socialization and Valuing Scale 9A, the Brief Attitudes Survey for Interprofessional Collaborative Learning, the Attitudes Toward Health Care Teams Scale, the Team Skills Scale, the Interprofessional Collaboration Scale, and the Simulation Effectiveness Tool–Modified (Multimedia Appendix 1) [11-14].

A total of 126 questions were compiled into the initial database, covering five key domains: (1) demographics, (2) telehealth knowledge and skills, (3) understanding interprofessional roles, (4) communication, and (5) teamwork. A core interprofessional group of 5 experts was recruited to refine the initial 126-item survey into a 50-item survey. These individuals were members of the Interprofessional Education Committee at Stony Brook University, convened to guide the development of future IPE courses. This panel included representatives from the School of Medicine, School of Nursing, School of Health Professions, and School of Social Welfare with extensive knowledge in both IPE and telehealth core competencies [9,10]. These representatives individually assigned a ranking to each item. Individual scores were summed and divided by the number of rankings assigned to each item, and the top 50 questions were chosen for widespread distribution. These were grouped into the following sets of questions: demographic-related information (questions 1-8), questions of experience in interprofessional teams (question 9), and questions of familiarity in interprofessional teams (question 10). Demographic items collected information on participants’ school; sex; ethnicity; race; age; and perceived knowledge, skill, and usability of telehealth. Originally, question 9 contained 33 items, and question 10 contained 8 items. Psychometric validation was performed exclusively for questions 9 and 10, and each section was analyzed independently.

A convenience sampling approach was used targeting graduate health professions students participating in required coursework across the School of Medicine, School of Nursing, School of Health Professions, and School of Social Welfare at Stony Brook University. Recruitment and survey distribution took place between November 2023 and June 2024 during mandatory in-class instruction associated with IPE coursework. Students were invited to complete the survey electronically during class sessions. Participation was voluntary, responses were anonymous, instructors did not have access to individual responses, no grades or evaluations were linked to participation, and no incentives were provided. A total of 520 students responded. Records were excluded if they had no response to either question 9 or question 10 (n=122), leaving 398 records. Five duplicate submissions were identified and removed, resulting in 393 records for analysis. No separate 95% completion threshold was applied.

Analyses followed standard psychometric validation procedures for scale development and exploratory factor analysis (EFA) [15-19]. Differences in the distribution of participants’ responses across the 4 schools were evaluated using chi-square tests, with exact P values generated via Monte Carlo simulation to account for small cell counts and potential sparsity in the data. The item reduction analysis focused on refinement of the 2 conceptually distinct domains: experience in interprofessional teams (question 9) and familiarity in interprofessional teams (question 10). These domains were analyzed separately a priori because experience and familiarity represent related but distinct constructs. Descriptive analyses were first conducted to examine relationships among items in questions 9 and 10 using polychoric correlation heat maps to visualize estimated correlations between ordinal items and confirm the expected separation of the 2 domains. Instrument reliability for questions 9 and 10 was evaluated using the Cronbach α, with values between 0.7 and 0.9 considered acceptable and values above 0.9 considered indicative of potential item redundancy [15-17]. Item-to-total correlations were used as an initial item performance screen to identify items with limited association with their intended domain; items with item-to-total correlations of less than 0.3 were removed before EFA. This screening step was used to reduce poorly performing items while recognizing that EFA was subsequently used to evaluate multidimensional factor structure within each domain. EFA using principal factor extraction with oblique Promax rotation was performed. Suitability for factor analysis was confirmed via the Kaiser-Meyer-Olkin (KMO) measure (threshold>0.5) [16,17]. Parallel analysis determined the number of factors to retain by comparing observed eigenvalues with those obtained from simulated data. Items were retained if they demonstrated a primary factor loading of at least 0.50, did not demonstrate substantial cross-loading, and had a difference of at least 0.20 between the primary and next-highest loading. Items that did not meet these criteria were removed to produce a clearer, more interpretable factor structure [18,19].

All statistical procedures were performed in SAS (version 9.4; SAS Institute Inc).

This study was submitted to the Stony Brook University Institutional Review Board (IRB) for preliminary review and was determined not to constitute human subject research requiring further IRB review or approval (IRB2023-00080). Participation in the survey was voluntary. Completion of the anonymous survey was considered implied consent; no written informed consent was obtained because no identifiable private information was collected and the activity was determined not to require further IRB review. Privacy and confidentiality were protected by collecting responses anonymously, storing data in deidentified aggregate form for analysis, and ensuring that instructors did not have access to individual responses. No grades, evaluations, or other academic consequences were linked to participation or nonparticipation. Participants received no compensation or incentives for completing the survey.

A total of 520 responses were received (Figure 1). After excluding records with no response to either question 9 or 10 and removing duplicate submissions, 393 records were included in the analysis, with the distribution of participant characteristics summarized in Table 1.

x`Question 9, which assessed experience in interprofessional teams, consisted of 33 items, and question 10, which assessed familiarity in interprofessional teams, consisted of 8 items. Questions 9 and 10 were analyzed separately because experience and familiarity were defined a priori as conceptually distinct constructs. A polychoric correlation heat map was used to examine correlations among items and confirmed the expected separation of questions 9 and 10 into distinct blocks (Figure 2). Item reduction was therefore conducted separately for questions 9 and 10.

The original 33-item section demonstrated strong internal consistency, with a Cronbach α of 0.94. Four items with an item-to-total correlation below 0.3 were removed. Using the remaining 29 items, the data showed high interitem correlations, as indicated by a KMO score of 0.94, confirming its suitability for the questions to move onto the next stage of evaluation for factor analyses. EFA identified 3 meaningful factors that explained 69.81%, 13.00%, and 9.58% of the variance. A total of 6, 8, and 3 items loaded clearly onto the 3 factors, resulting in a refined 17-item version.

The internal consistency of the refined scale remained strong, with an overall Cronbach α of 0.9 and factor-specific Cronbach α values of 0.88, 0.89, and 0.77. Internal consistency reliability was also assessed within each school using the reduced 17-item version, and acceptable reliability was confirmed both overall and within each factor (Table 2).

The original 8-item section demonstrated high internal consistency, with a Cronbach α of 0.94. All items showed item-to-total correlations above 0.30, making them all considered in factor analysis. A KMO score of 0.92 further confirmed strong correlations among the items (Table 3).

Parallel analysis initially indicated a single-factor solution, suggesting that all 8 items could be represented by 1 factor. However, for the purpose of item reduction, a 2-factor model was explored, and this model converged successfully. One factor contained 3 items, and the other factor contained 2 items. In the unrotated solution, factors 1 and 2 explained 93.03% and 6.98% of the variance, respectively. The simplified 5-item version maintained strong reliability, with a Cronbach α of 0.91 for the overall scale and of 0.90 and 0.88 for the 2 factors. The reduced instrument showed consistent reliability across various professional program groups, although small sample sizes from the School of Social Welfare and the School of Nursing may have slightly influenced the results.

To ensure the comprehensive evaluation of IPE in virtual environments, the final survey compilation incorporated demographic factors to enhance its applicability across diverse settings. These factors were carefully integrated to capture variations in responses attributable to participants’ backgrounds, such as age, professional discipline, and educational level and led to the final survey total questions to 32 items from the initial 51 questions (Multimedia Appendix 2). Multimedia Appendix 3 identifies each candidate item, whether it was retained in the final instrument or removed during refinement, whether it was adapted from an existing validated interprofessional or teamwork instrument vs newly developed for this study, and whether the item captures a virtual-specific competency.

This study aimed to develop and conduct preliminary psychometric validation of survey subscales assessing IPE conducted in a virtual environment. The survey development process included evaluation and thematic grouping of items adapted from previously validated IPE tools, ranking and review by an expert panel with expertise in IPE and telehealth competencies, and distribution to graduate health professions students across multiple schools. The psychometric analysis focused on 2 sections: experience in interprofessional teams (question 9) and familiarity in interprofessional teams (question 10). EFA supported refinement of question 9 from 33 items to a 17-item, 3-factor structure and refinement of question 10 from 8 items to a 5-item structure. Confirmatory factor analysis was not conducted; therefore, the findings should be interpreted as preliminary evidence supporting the internal structure and reliability of these 2 subscales.

Prior studies have described virtual and telehealth-based IPE curricula that target communication, teamwork, role clarity, learner confidence, and telehealth readiness [4-7]. The present study did not directly evaluate whether learners improved in these outcomes. Rather, it contributes to this literature by developing and preliminarily validating two survey subscales intended to measure learners’ reported experience and familiarity within virtual interprofessional settings. This distinction is important because validated measurement tools are needed before educational programs can be compared reliably across contexts or linked to learner outcomes.

The present study provides preliminary psychometric support for two subscales designed to assess learners’ experience and familiarity within virtual interprofessional environments. Several validated instruments exist to assess interprofessional attitudes, teamwork, collaboration, and simulation-based learning; however, many were developed primarily for in-person or simulation-based settings rather than for distributed, technology-mediated care [11-13]. The retained items include content related to virtual collaboration, telehealth communication, virtual care planning, and functioning within virtual interdisciplinary teams. However, because only the experience and familiarity subscales were evaluated using EFA, the findings should not be interpreted as validation of the entire survey or as definitive evidence that all virtual-specific competencies have been fully captured. Further studies are needed to confirm the factor structure, examine construct validity, and evaluate relationships with educational outcomes.

This study has several limitations. First, uneven participation across professional groups, particularly lower representation from the School of Nursing and School of Social Welfare, may have influenced reliability estimates and limited generalizability. Second, data were collected from a single academic medical center, which may further restrict broader applicability. In addition, the reliance on self-reported data introduces the potential for response and social desirability bias. The cross-sectional design precludes assessment of test-retest reliability and limits evaluation of the temporal stability of the instrument. Finally, as all data were obtained from a single survey source at a single time point, there is potential for common method variance, which may have inflated observed associations among items.

Despite these limitations, the stepwise development process and exploratory psychometric evaluation provide preliminary support for the reliability and internal structure of the two validated subscales.

The Virtual Interprofessional Education Survey includes two subscales with preliminary EFA-based psychometric support for assessing graduate health professions learners’ experience and familiarity within virtual interprofessional settings. The retained items address aspects of virtual teamwork, telehealth communication, interdisciplinary collaboration, and virtual care planning. These findings support continued refinement and evaluation of the instrument, including confirmatory factor analysis, assessment of construct validity, and testing in broader learner populations, before the tool is used to make stronger claims about educational outcomes or clinical impact.