We collected quantitative data using a 36-question survey divided into six categories: (1) demographics (multiple choice; questions 1-2); (2) patient and family background, such as location and medical history (multiple choice; questions 6-15); (3) degree of AD and general mental health (sliding scale; questions 16-23); (4) quality and quantity of health care services and communication (sliding scale; questions 24-30); (5) degree of health inequity experienced while in the hospital (sliding scale; questions 31-33); and (6) degree of interest in the use of mHealth to facilitate better communication with the clinical staff (sliding scale; questions 34-36).

The Family Inpatient Communication Survey [46,47] and senior technology acceptance model survey [48], along with other published papers addressing family-centered care of patients [20,49-51], provided guidance in the development of our research instrument. After institutional review board approval, we prepared the questionnaire in the REDCap (Research Electronic Data Capture; Vanderbilt University) survey platform. The overall reliability of the scales used was assessed using the Cronbach α, yielding a coefficient of 0.758, indicating acceptable internal consistency.

All self-reported psychological constructs within the survey, such as anxiety, depression, satisfaction, and mental health, were left to each participant’s interpretation based on their subjective experience of the terms. For example, one question asks the following: “How would you rate your anxiety after your family member was in the ICU?” Next to the question is a sliding scale with no numbers, ranging from “no anxiety” on the left to “highly anxious” on the right. Such terms were not operationalized or linked to existing standardized psychometric tools such as the Hospital Anxiety and Depression Scale or the Beck Anxiety Inventory [52].

For participant enrollment, we posted our survey aims and invitation with a URL on 16 social networking websites and Facebook community groups related to critical care and family-centered care. We also displayed posters at selected clinical sites on the University of Kentucky medical campus (Lexington, Kentucky) for further recruitment. Individuals were eligible to participate if they were (1) adults (aged ≥18 years) and (2) self-identified family members of critical care patients (in cardiac, pediatric, COVID-19, and cancer ICUs) admitted within the previous 2 years. Data collection lasted 6 months.

This research project was originally designed as a proof-of-concept study, where a limited sample would be sufficient to achieve our intended goals. Moreover, we experienced extraordinary challenges in obtaining responses. As such, we set the sample size at 100 using a convenience sampling method for participant recruitment. A post hoc sensitivity analysis was performed using G*Power (version 3.1). With a sample size of 100, an α level of .05, and statistical power set at 80%, the analysis demonstrated sufficient capacity to detect a medium effect size of r=0.28 (2 tailed). This indicates that the study was appropriately powered to identify moderate or stronger associations, aligning with the objectives of this pilot investigation. We received 97 responses in the end.

We analyzed the data using the open-source statistical software Jamovi (version 2.7.9) [53]. Jamovi is a statistical software built on R (R Foundation for Statistical Computing). Standard significance levels were used throughout the analysis, with α=.05 as the threshold for statistical significance unless otherwise specified. To analyze the data, first, we examined the extent and pattern of missingness across variables. Jamovi excludes any row (case) that has missing data for one or more variables used in the analysis. The analysis was set to Jamovi’s handling of missing data as the default. In addition to reporting descriptive statistics (eg, means, SDs, and frequencies), we conducted inferential statistical analyses to explore relationships and differences among variables.

We applied a paired-sample 2-tailed t test to identify differences in AD between ICU admission and postdischarge. Furthermore, correlation analyses were conducted to determine the strength and direction of relationships among transportation difficulties, satisfaction with the frequency of patient updates, mental health indicators (reported AD status), and family interest in mHealth solutions. However, because the data did not meet the assumptions of normality, nonparametric tests were used to assess differences (Wilcoxon W) and relationships (Spearman ρ) among variables. Further analysis details are provided in the Results section.

This study was reviewed and approved by the institutional review board (approval 83549) at the University of Kentucky, Lexington, Kentucky, United States, in accordance with the principles of the Declaration of Helsinki. The participant population was from both within and outside the University of Kentucky. Outside populations were recruited through REDCap advertising networks and through the list of websites that specialize in participant recruitment. Researchers obtained an institutional review board waiver of documentation for the informed consent process. All participation was on a voluntary basis, with no compensation offered. Research presented no more than minimal risk to the participant because it involved only survey and interview questions and obtaining consent would increase the risk related to confidentiality based on the consent document being the only link to any personal health information.

We used a nonprobability sampling technique, necessitating the collection of demographic information from family member participants. Patients spanned various age groups, with a notable concentration of those aged ≥31 years. The 97 individuals who responded to the survey represented a spectrum of ages, races, ICU patient disease types, and relationships to patients. The data showed a balance of both male and female patients. Race identification among family members was 80% (n=78) White, 18% (n=17) Black, and 2% (n=2) other races.

Regarding medical reasons for patient admission to the ICU, the percentage breakdown of the sample included cancer (n=16, 17%) and COVID-19 (n=13, 13%) as the most frequent, followed by pneumonia or infections (n=9, 9%), myocarditis (n=7, 7%), stroke (n=7, 7%), sepsis (n=7, 7%), accidents or injuries (falls, burns, cuts, or gunshot wounds; n=7, 7%), overdose or substance abuse (n=6, 6%), diabetes (n=3, 3%), meningitis (n=2, 2%), and other conditions (n=20, 21%). The largest group of patient relatives who spent time at the bedside were parents or guardians (n=34, 35%), followed by spouses (n=14, 14%), grandparents (n=12, 12%), and other relatives (n=37, 39%).

We analyzed the normality of the data by using skewness and kurtosis tests for all dependent variables. Table 1 shows how the data were normally distributed. Values were within the threshold range for both skewness (–1 to +1) and kurtosis (–3 to +3) [54]. We supplemented the skewness and kurtosis statistics with formal normality tests, specifically the Shapiro-Wilk test, which is more appropriate for small to moderate sample sizes. To further support our assessment, we generated Q-Q plots for the main continuous variables used in parametric analyses. These plots were examined to identify deviations from normality that may not be evident in statistical tests alone. The Shapiro-Wilk test results and Q-Q plots showed deviation from normality; therefore, nonparametric tests were applied for inferential statistics. Figures 1-4 provide the Q-Q plots illustrating the data.

RQ 1 addresses the degree to which family members were satisfied with available health information and communication provided by ICU clinical staff. The mean score of 2.94 (SD 1.31) on a scale from 1 to 6 shows that, on average, family members received updates once every 7 to 12 hours when away from the hospital (Table 2). Regarding the satisfaction level with the overall frequency of communication with the patients’ family members, we observed a mean score of nearly 7 on average (mean 6.62, SD 2.70) on a scale from 1 to 10. Conversely, phone communication was lower on average. Furthermore, satisfaction with the mental health services provided by the hospital for family members was at a midpoint on a scale from 1 to 10 on average (mean 4.97, SD 2.90). Mental health services as defined in the survey included mental health counseling, social services, and clinical psychologists available to family members on request (Table 2).

RQ 2 addresses the level of AD among family members while their loved ones were in the ICU, as well as after discharge. Table 3 shows that the mean scores for AD were elevated (scale from 1-10). Anxiety was consistently higher than depression in both periods (Table 3). Levels of AD were higher among family members during the ICU stay than after discharge. To determine whether this difference was statistically significant, we conducted a paired-sample t test using the Wilcoxon W. The results confirmed a significant difference between AD levels of family members during the ICU stay and after discharge (Table 4).

Prior research has established that there is difficulty in arranging transportation for family members to ICU clinical sites, which impacts their satisfaction level and frequency of receiving patient information, phone communication, and mental health services [55].

For RQ 3, we examined the correlation between difficulty arranging transportation and satisfaction with the frequency of shared information. Our findings confirmed evidence of a negative relationship between transportation difficulties and satisfaction with the frequency of information provided (r=−0.284; P=.005). This suggests that, as transportation challenges increase, families become less satisfied with the frequency of patient health information. The lack of frequent visits due to transportation issues may also suggest a heightening of AD and, subsequently, a stronger desire for more frequent updates to compensate for family members’ inability to regularly visit the ICU.

RQ 3b examined the correlation between difficulty in arranging transportation and AD. As performing multiple correlation analyses increases the risk of type I error, the Benjamini-Hochberg procedure was applied to control the false discovery rate at α=.05. As all raw P values were identical and very small (P=.001), the Benjamini-Hochberg–adjusted P values were uniformly .002, which is well below the threshold. Therefore, all tests remained significant after false discovery rate correction, indicating strong evidence against the null hypotheses (Multimedia Appendix 1).

Unlike the correlation observed with satisfaction regarding communication of patient information, we found no significant correlation between transportation difficulties and anxiety (P=.15) and depression during the patients’ ICU stay (P=.71). However, there was a positive correlation between anxiety levels and depression levels during the patients’ ICU stay (r=0.550; P=.001). Additionally, AD levels experienced during the ICU stay were significantly and positively correlated with AD levels after patient discharge (Table 5).

Regarding RQ 4, we surveyed family members about their interest in new forms of mobile technology designed to improve information flow and communication from the ICU bedside to the family’s location regarding their loved ones’ health status. Table 6 presents data indicating that family members expressed strong interest in using such a service (mean 8.34, SD 1.98), connecting with bedside nurses via SMS text message and videoconferencing (mean 8.32, SD 2.25), and having easy access to social workers to help manage their AD (mean 8.29, SD 2.03; 1=least interest; 10=high interest).

RQ 5 explored the correlation among satisfaction with communicated patient information, AD experienced by family members, and interest in using an mHealth app. As performing multiple correlation analyses increases the risk of type I error, we applied the Benjamini-Hochberg false discovery rate correction (α=.05) to adjust for multiple comparisons. Adjusted P values (q values) are reported, and correlations with q<0.05 were considered statistically significant. Thus, we found it reliable to proceed with a multiple correlation analysis (Multimedia Appendix 2).

Spearman ρ outcomes revealed significant positive correlations among these variables (P<.05; Table 7). We discovered that families experienced higher levels of anxiety during patients’ ICU stays and had significantly greater interest in the use of a mobile app that would provide direct access to social workers (r=0.326; P<.001), in using an mHealth videoconferencing app (r=0.319; P=.002), and in overall mHealth app use (r=0.322; P=.001). We also observed that family members who experienced higher levels of depression during ICU stays had significantly greater interest in SMS text message and videoconferencing communication with the bedside nurse (r=0.297; P=.003) and an app to connect with social workers (r=0.214; P=.04). Figure 5 provides a graphic review of the findings.

Similarly, we observed that satisfaction with phone communication from clinicians was positively correlated with direct access to social workers through mHealth (r=0.221; P=.03), use of an mHealth videoconferencing app (r=0.219; P=.03), and satisfaction with the frequency of medical information shared by the staff (r=0.710; P<.001).

We further confirmed that participants’ interest in an mHealth videoconferencing app was significantly and positively correlated with overall mHealth app interest (r=0.775; P<.001). Interest in mHealth to connect with social workers was significantly and positively correlated with interest in an mHealth videoconferencing app (r=0.813; P<.001) and overall mHealth app use (r=0.814; P=.001). However, there was no correlation between satisfaction with health information communicated and AD.

Finally, our results indicate a positive correlation between satisfaction with the frequency of health information received and satisfaction with phone communication from medical staff (r=0.710; P<.001). The results further indicate that family members who were more satisfied with the frequency of communication were more interested in using an mHealth videoconferencing app for frequent updates (r=0.299; P=.003), using an app to connect with social workers (r=0.280; P=.006), and using an mHealth app for updates about their family members (r=0.288; P=.005).

Our findings confirm that family members experienced high levels of AD both during their loved ones’ ICU stay and after discharge. Notably, AD levels were more pronounced while patients were in the ICU compared to after discharge. This heightened stress was especially evident during the first year of the COVID-19 pandemic, when many ICU patients died without a family member present, leading to increased stress levels among relatives [56]. Although AD can be significant during an ICU stay, the persistence of these issues after discharge, as reported in this study, raises serious concerns. This finding is further confirmed by another recent study reporting that family members exhibited symptoms of posttraumatic stress disorder for up to a year after their loved ones were discharged from the ICU [57].

We emphasize COVID-19 as a primary context for the issues raised due to the recent global impact it had on millions of families. While cancer as a reason for admission to the ICU was noted in 17% (n=16) of cases (COVID-19 constituting n=13, 13% of cases), there are over 200 types of cancer, with the National Cancer Institute identifying breast cancer, lung cancer, prostate cancer, colorectal cancer, bladder cancer, and melanoma as the most common. Due to the noncommunicable nature of these diseases, families do not experience the same types of mental health issues because they are often not isolated from their loved ones. Conversely, COVID-19, within a short period, brought to the fore the severe challenges of family-clinician communication and its impact on the mental health of families.

Our study indicated a maximum mean satisfaction level of only 6.62 (SD 2.70) on a scale from 1 to 10, reflecting a generally low level of satisfaction among family members regarding real-time communication. Difficulties in arranging transportation significantly affected their dissatisfaction with the frequency of updates. This difficulty may be attributed to the challenges of traveling to hospitals from remote areas. This finding aligns with those of previous research that suggests that the psychosocial status of family members can influence their satisfaction with communication.

Satisfaction with existing communication between clinical staff and family members of ICU patients led participants to express interest in a digital solution that could more readily provide real-time bedside health updates, as well as the ability to text and conduct videoconferences. This need was also highlighted during the COVID-19 pandemic, when visitation restrictions underscored the necessity of having alternative means of connecting family members with their loved ones. Virtual visitation via videoconferencing has been proposed as a solution, addressing the challenges faced by family members who struggled with the distance and time constraints associated with in-person visits [56,58].

Additionally, our study found that family members were not satisfied with the adequacy and availability of mental health counseling and social services provided by hospitals. This dissatisfaction further underscores an interest from families in an mHealth solution that facilitates virtual communication with social workers or mental health counseling, particularly for families of critical care patients.

In summary, our findings suggest that families were highly dissatisfied with the frequency of health updates provided by clinical staff, with lower satisfaction reported among those who faced difficulties arranging transportation or lived further from the hospital. Family members experienced high levels of AD during their loved ones’ ICU admission as well as after discharge even though their mental health challenges were reduced.

In this study, modest but statistically significant correlations were observed between family members’ reported mental health status (AD) during the patients’ ICU stay and their interest in using an mHealth app. A similar association was found between participants’ satisfaction with medical communication and their interest in a mobile app. Additionally, difficulty arranging transportation was negatively correlated with family members’ satisfaction with the frequency of medical information sharing. Overall, family members expressed interest in a digital tool that could provide access to real-time bedside information, facilitate communication with bedside nurses, and support connections with social workers.

It is important to note that these findings are based on a nonrandom sample and reflect moderate correlations; therefore, they should be interpreted with caution. Rather than indicating strong evidence of effectiveness, the results suggest potential associations that warrant further investigation. Nonetheless, when considered alongside prior research emphasizing the value of structured interprofessional frameworks in hospital and ICU settings—particularly those using standardized digital tools and empathic communication strategies [59]—the findings highlight directions for future work.

Moreover, prior studies suggest that interventions aimed at mitigating the negative psychological impacts of hospitalization may be beneficial when initiated early during ICU admission. Remote or technology-mediated approaches may improve reach and engagement, whereas individuals with greater needs may benefit from longer or in-person interventions to ensure adequate support and reduce the risk of discontinuation [60].

Building on these insights, we propose that future research explore the development and evaluation of an mHealth intervention designed to support communication between family members and ICU bedside staff. Such an app could offer timely clinical updates and secure messaging or video communication within a HIPAA (Health Insurance Portability and Accountability Act)-compliant platform. Importantly, rather than replicating the limited access often associated with existing mobile electronic medical record services, a family-centered design could acknowledge the central role of family members in patient recovery. However, rigorous studies using randomized designs are needed to assess feasibility, effectiveness, and ethical considerations before such interventions can be recommended for widespread implementation.

This study has several important limitations that should be considered when interpreting the results. First, our self-selecting sampling strategy may have introduced bias, potentially overrepresenting family members who were more comfortable with technology and, thereby, inflating interest in mHealth apps. Second, the sample size was limited due to the uniqueness of the population studied and the sensitivity of the topic (ie, family members may have found it difficult to fill out the survey due to the need to revisit a prior family event that was psychologically painful). Third, mental health status (AD) was self-reported and was not assessed using standardized scales. Considering these limitations, the findings may be interpreted as exploratory and may not be generalizable to all ICU family populations. Nevertheless, this study offers valuable insights into this important issue and identifies directions for future research.

In parallel with this study, we recently completed a quasi-experimental pilot study that investigated the feasibility, usability, and efficacy of a newly developed mHealth app to reduce the AD of family members of critical care patients with cancer, with a focus on those living in rural Kentucky. This study recruited family members of patients in the Markey Cancer Center, Blood & Marrow Transplant and Cellular Therapy Program, University of Kentucky. As such, our ongoing research includes further mHealth app design, development, and clinical testing of this intervention in both critical care and other long-term health facilities (eg, nursing homes). The aim of this translational research is to provide bedside support to family members living in remote locations throughout Kentucky and Ohio.