Comparing Two Ways of Supporting Mental Health among Clinicians, Staff, and Patients during the COVID-19 Pandemic—The MHAPPS Study

BACKGROUND: The COVID-19 pandemic has dramatically affected individuals and communities, with implications for both mental and physical health. The mental health impact of COVID-19 on health care workers was expected to be particularly pronounced but was not yet well understood. As health systems struggled with historically high volumes of patients because of COVID-19, identifying innovative ways to provide evidence-based mental health support without burdening health systems was a priority. Caring text messages, such as “Hello [name], just wanted to say I hope you're doing OK and remind you that I'm here for you,” have been shown to significantly reduce suicidal ideation and behavior. Caring texts sent in clinical trials published to date have been sent by specialists that know the recipient, yet caring text programs implemented at scale rarely include the step of introducing the recipient to the sender. It is unknown whether receiving caring texts from an unknown sender is as effective as receiving caring texts from someone who knows the recipient. OBJECTIVES: Aim 1: Measure the prevalence of mental distress, including loneliness, anxiety, depression, suicidal ideation and behavior, and psychological stress, in physicians, health care clinicians, and health care staff (ie, health care professionals [HCPs]), employees, and patients served by St Luke's Health System. Describe differences associated with age, sex at birth, race, ethnicity, urban or rural residence, living situation and number of people at home, occupation, local COVID-19 prevalence, perceived risk related to COVID-19, and masking and social distancing practices. Aim 2: Compare the effectiveness of 2 versions of the Caring Contacts intervention (introductory phone call followed by text messages vs text messages alone) to reduce loneliness and improve mental health outcomes among St Luke's Health System patients, HCPs, and employees experiencing mental distress during the COVID-19 era. METHODS: Aim 1: Conduct a cross-sectional survey of mental health outcomes and their association with COVID-19 attitudes, beliefs, and practices among HCPs, staff, and primary care patients from a large rural health system in Idaho. Linear regression models with robust SEs were used to assess the association between each primary and secondary outcome and COVID-19 attitudes, beliefs, and practices, adjusting for baseline demographic characteristics, using α = .05. Aim 2: Conduct a pragmatic randomized clinical trial comparing the effectiveness of 2 versions of a Caring Contacts intervention: 1 version with an introductory phone call (Caring Contacts+) and 1 without an introductory phone call (Caring Contacts). Clinical trial participants were recruited from a subset of aim 1 participants who reported moderate or high levels of at least 1 measure of mental distress. Caring texts were sent by trained nonclinician follow-up specialists from the state suicide prevention hotline. All participants were randomly assigned to receive caring texts either with an introductory phone call (Caring Contacts+) or without an introductory phone call (Caring Contacts). Each participant received 11 standardized outgoing caring text messages over 6 months; when participants replied to text messages, follow-up specialists sent unscripted, personalized responses. Results were analyzed using linear regression with robust SEs and reported by strata—HCPs/staff and patients—using α = .05. The primary outcome was loneliness, measured using the National Institutes of Health Toolbox Social Relationship Scale for Loneliness. Secondary outcomes included suicidal ideation and suicidal behavior, as measured by the Columbia-Suicide Severity Rating Scale; perceived burdensomeness and thwarted belongingness, as measured by the Interpersonal Needs Questionnaire; and depression, as measured by the Patient Health Questionnaire. All measures were self-reported on online surveys at baseline and 6 months. The trial was single blinded. All outcome measures and study hypotheses were specified before data collection. Data were analyzed according to an intention-to-treat protocol. RESULTS: Aim 1: A total of 1687 patients (11.9% of those invited) and 1959 HCPs and staff (26.7% of those invited) completed the survey. Of the patients surveyed, 67.2% were women and 93.7% were White; of the HCPs/staff surveyed, 79% were women and 91.4% were White. In total, 32.2% of patients and 29% of HCPs/staff reported moderate to high levels of at least 1 form of mental distress. Similarly, 24.7% of patients and 20.5% of HCPs/staff reported moderate to high levels of loneliness. Patients of retirement age reported better mental health than employed patients or unemployed patients younger than 65 years of age. Among health care employees, office workers (32.3%), nurses (31.8%), and social support service employees (28.8%) reported the highest levels of mental distress. Demographic factors (older age, female sex, increased household size, care for dependents, and being employed) were significantly associated with increased adverse mental health outcomes. People who reported stronger perceived risk associated with being infected with COVID-19, infecting another person, and having someone in their direct environment die from COVID-19 had significantly worse mental health outcomes than those with lower levels of perceived risk. Aim 2: A total of 331 patients (8 adolescents and 323 adults) had a mean (SD) age of 45.5 (16.4) years, and 79% were women; 335 HCPs and staff had a mean (SD) age of 40.9 (11.8) years, and 87% were women. In total, 655 (98.3%) trial participants were retained through the end of the study. There were no significant differences in loneliness at 6 months by intervention group in either the patient group (adjusted difference of means, −1.0 [95% CI, −3.0 to 1.0]) or staff group (adjusted difference of means, 0.2 [95% CI, −1.8 to 2.2]). No secondary outcomes, including participant satisfaction or adverse effects, differed significantly across treatment groups. CONCLUSIONS: The prevalence of mental distress varied by employment status (patients) and employment type (HCPs and staff). Age; sex at birth; religious beliefs; and COVID-19 attitudes, beliefs, and practices were significantly associated with mental health outcomes. Including an initial phone call added significant operational complexity and cost without improving participant satisfaction or the effectiveness of a caring text program. Partnering with a suicide prevention hotline is a feasible way for health systems, including those in states with a limited supply of mental health care professionals, to scale up 2-way caring text programs that may help support health care workers and patients experiencing mental distress. LIMITATIONS: The study population was predominantly female (83%), White (93%), and non-Hispanic (91%), and results may not be generalizable to populations with different demographic profiles. Individuals aged 12 to 17 years were not a target population but were eligible for both aim 1 and aim 2; so few adolescents enrolled that the study cannot draw meaningful conclusions about this population. Selection bias may have been present in the patient population (only 12% of invited participants enrolled in aim 1, and nearly half of participants in both aims were unemployed) and in physicians/advanced practice HCPs (who were less likely to participate in both aims than other cadres of health care workers). As this was a comparative effectiveness trial with 2 active treatments, there was no inactive control group; therefore, it is not possible to know to what degree mental health outcomes observed in both intervention groups were because of a group effect related to stressors from the pandemic vs the effectiveness of the interventions.