Use of Telehealth During the COVID-19 Pandemic Among Practicing Maternal-Fetal Medicine Clinicians

Methods

This cross-sectional study drew data from the Society for Maternal-Fetal Medicine (SMFM) Telehealth Survey distributed to U.S.-based regular and associate physician members in December 2020. It anonymously gathered information on provider demographics, practice-setting characteristics, use of telehealth before and during the pandemic, defined as beginning on March 11, 2020, and provider perceptions of telehealth. A reminder e-mail was sent 1 week after the initial e-mail and the survey was closed 2 weeks after. Participants gave informed consent when they initiated the survey, which on average took 12 min to complete. The survey instrument is available in Supplementary Appendix SA1.

Only physicians who were providing direct patient care with nonmissing data on main outcomes were included (n = 373). The outcomes were use of any (vs. no) live video visits during the pandemic and, among users, the extent of use of live video visits. We queried on the maximum percentage of outpatient visits that were live video telehealth visits since the start of the pandemic. Those who responded 0% were categorized as “No video visits” whereas nonzero response options were categorized as “Any.” The extent of use of live video visits among users were dichotomized into 1–25% (“low users”) and 26–100% (“high users”). We chose this classification based on the low proportion of use relative to in-person visits by obstetricians in a prior national study.³

We compared characteristics of the survey participants with those of the general SMFM membership to assess representativeness of our sample. We ran bivariate analyses to assess the association between the independent variables and the two outcomes using χ² for categorical variables and t tests for continuous variables. An alpha level p ≤ 0.05 defined statistical significance and was used to select potential independent variables or covariates in the multivariate models. We used logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CIs) to quantify the association between each independent variable and each outcome. Following our conceptual model, we ran simultaneous regression models that included combinations of the significant predisposing and enabling variables.

The Akaike information criterion (AIC) helped us select the best fitting model (Model 1). AIC penalizes models with larger numbers of explanatory variables that do not have much explanatory power.¹⁷ We performed sensitivity analyses to test the potential contribution of years of obstetric practice, urbanicity, and practice location; although these covariates were not significantly associated with the outcome in bivariate analyses, they were forced into the simultaneous regressions (Model 2). Furthermore, we described the association between the extent of live video visit use among users and provider satisfaction with telehealth and desire for future use, using bivariate analyses. Analyses were performed using Stata IC 16.1.¹⁸ Our study was deemed exempt from human subjects review by the Stanford University Research Compliance Office (Protocol no. 56986).

Results

The SMFM has 2,912 associate and regular members, representing about 85% of MFM physicians nationally.¹⁹ Survey invitations were sent electronically to 2,573 members with known and active e-mail addresses and they were given 2 weeks to complete; 422 providers responded, for a response rate of 16.4%. After dropping ineligible respondents, the total sample yielded 373 participants from 46 states and Washington D.C.

Demographics of survey respondents are listed in Table 1. Compared with available SMFM membership data, our survey oversampled women, those over age 65, and clinicians working in multiple practice settings; we under sampled clinicians working in university and independent practice settings.

Telehealth use prepandemic was reported by one third to almost half of the clinicians, depending on the modality, whereas ∼88% of the clinicians reported any live video use during the pandemic (Table 1). Compared with those who reported no live video visit uptake, users were more likely to be younger (p = 0.02), to provide comprehensive prenatal care (p < 0.01) and/or inpatient care (p = 0.02), practice in university settings (p = 0.01), engage in various telehealth use modalities prepandemic (p ≤ 0.01), and to perceive challenges with software or hardware usability (p < 0.01), reimbursement (p = 0.05) and patient access to the internet or data plans (p < 0.0001).

Any use of live video visits was associated with predisposing and enabling factors (Table 2). Among the predisposing factors, we observed a 1% decrease in use of live video visits for every incremental year in physician age (95% CI = 0.94–1.0) and an almost threefold increase in use among those providing comprehensive prenatal care (crude OR = 2.76; 95% CI = 1.46–5.19) and inpatient care (crude OR = 2.91; 95% CI = 1.15–7.36). Compared with clinicians practicing solely in university settings, clinicians who practiced in multiple settings were far less likely to use live video visits (crude OR = 0.22; 95% CI = 0.10–0.52). Nonetheless, only type of practice setting remained significantly associated with uptake of live video visits when adjusting for significant predisposing and enabling covariates (adjusted OR [aOR] = 0.25; 95% CI = 0.10–0.62) (Model 1, Table 2).

Among the enabling factors, use of telehealth before the pandemic increased the odds of any live video use during the pandemic. Specifically, RPM more than doubled the odds (crude OR = 2.5; 95% CI = 1.20–5.21), as did any remote ultrasound imaging interpretation and consultation (crude OR = 2.52; 95% CI = 1.28–4.97); any communication using an online patient portal increased the likelihood fourfold (crude OR = 4.18; 95% CI = 1.72–10.14). Other co-occurring enabling factors that were associated with increased likelihood of live video use were perceived barriers, including software/hardware usability (crude OR = 3.17; 95% CI = 1.52–6.59); reimbursement/insurance coverage (crude OR 1.99; 95% CI = 1.00–3.99); and patient access to internet or data plans (crude OR = 4.14; 95% CI = 1.99–8.63).

After adjusting for type of practice setting and the significant enabling variables—only three enabling factors remained significantly associated with any live video use: any prepandemic communication using an online patient portal (aOR = 3.69; 95% CI = 1.47–9.25); perceived software or hardware usability barriers (aOR = 2.49; 95% CI = 1.14–5.42); and perceived patient access to internet or data plan barriers (aOR = 3.47; 95% CI = 1.58–7.60) (Model 1, Table 2). Model 2 (Table 2) shows that the associations with multiple practice settings and perceived patient access barriers became even stronger when forcing in years of obstetric practice, urbanicity, and practice locations. Model 2 was a better fitting model according to AIC values.

Approximately 44% (n = 143/327) of clinicians who adopted live video visits were high users, defined as conducting >25% of live video visits relative to all visits. The rest (56%) were low users. Type of practice setting was the only predisposing variable associated with high versus low use (Table 3). Compared with clinicians practicing in university settings, who were the highest users, clinicians practicing in community and government settings were less likely to be high users (crude OR = 0.41; 95% CI = 0.23–0.73).

Enabling factors associated with high versus low use were prepandemic communication using an online patient portal (crude OR = 1.77; 95% CI = 1.13–2.78) and prepandemic use of remote ultrasound imaging interpretation and consultation (crude OR = 1.8; 95% CI = 1.16–2.80). In addition, perceived barriers related to training providers and staff (crude OR = 1.64; 95% CI = 1.05–2.57); reimbursement/insurance coverage (crude OR = 1.58; 95% CI = 1.01–2.47) and patient access to internet or data plans (crude OR = 2.20; 95% CI = 1.40–3.44) about doubled the odds of high versus low use.

In adjusted models, only type of practice setting, prepandemic use of ultrasound imaging interpretation and consultation (aOR = 1.64; 95% CI = 1.04–2.59), and perceived patient access to internet or data plans (aOR = 1.83; 95% CI = 1.14–2.93) remained significantly associated with high versus low use (Model 1, Table 3). When forcing in years of obstetric practice, urbanicity, and practice location, the association with type of practice setting was no longer significant. Instead, a practice located in the Midwest (vs. the Northeast) showed lower odds of high adoption (aOR = 0.46; 95% CI = 0.21–0.98) (Model 2, Table 3). Model 2 was a better fitting model according to the AIC values.

Among live video visit users, high users were more likely to strongly agree that services through telehealth are excellent, whereas low users tended to disagree, be neutral or not strongly agree (p < 0.01) (Table 4). Similarly, high users were far more likely than low users to strongly agree that “telehealth is a positive thing for my clinic” (p < 0.001). Although most clinicians reported that they wanted to continue offering live video telehealth services after the pandemic ends or relaxes, more high users had this wish (92% vs. 99%; p < 0.01).

Discussion

This survey found that as of December 2020 there was widespread telehealth use among practicing MFM clinicians. This finding is consistent with previous surveys of telehealth use among practicing obstetric care providers.^3,4 Furthermore, we found that 44% of practicing clinicians were high users defined as using live videos for at least 25% of all visits and that 9% used live videos for >75% of all visits. A previous survey of practicing obstetricians showed that only 9% were using telehealth for >50% of their total visits in June 2020, corroborating the higher use among MFM clinicians compared with other obstetric care providers.³ Despite the higher use among MFM clinicians, the majority employed it in no more than one out of four visits. This pattern suggests a selective use with patients, perhaps to provide closer monitoring for patients with complex pregnancies, such as gestational diabetes or pre-eclampsia, or patients with geographic barriers to in-person care.

Our multivariable models showed that several predisposing and enabling factors were associated with any live video visit use during the pandemic. For example, the higher likelihood of use among clinicians in practices that communicated through patient portals may indicate a more complex electronic health record system, such as Epic, which facilitates easier integration of live video visits into their workflow.²⁰ Evidence further suggests that digital modes of communication are most effective for patients and clinicians who have already established relationships incorporating flexible use of telecommunication.²¹

When adjusting for covariates, clinicians who used live video visits (vs. no use) were more likely to perceive barriers to software and/or hardware, which could include challenges integrating this technology into existing clinical workflows such as rooming. In addition, clinicians who used live video visits were more likely to perceive patient barriers to access the internet or data plans due to lack of high-speed internet connectivity or a HIPAA-compliant videoconferencing platform. These results indicate that perceived technical barriers were important drivers of telehealth use among clinicians. Studies of patient perceptions corroborate provider perceptions of patient barriers: patients commonly report internet or mobile data access barriers; however, patients also perceive advantages to using telehealth such as reduced travel and childcare expenditures and time away from work.²²

Notably, only perceived patient barriers remained significant concerns after forcing in years of obstetric practice, urbanicity, and practice location. In contrast, perceived reimbursement barriers, found to be important barriers in other studies,²³ were not associated with live video use in our models, perhaps because telehealth insurance coverage expanded considerably during the pandemic.²⁴ The main deterrent of use was the type of practice setting. Clinicians who practiced in multiple settings were far less likely to adopt live video use perhaps because of the temporary nature of their assignments or the difficulty in learning multiple systems.

Type of practice setting also influenced the amount of use of live video visits among adopters. High (vs. low) users were far more likely to practice in university settings than in community or government settings, as did clinicians who engaged in prepandemic use of ultrasound imaging interpretation and counseling. University-based MFM providers—who are more abundant in the Northeast—and those who provide remote ultrasound interpretation usually are part of larger health care systems with more widespread technological support. Furthermore, the likelihood of high use was almost double among clinicians who perceived internet access barriers among their patients, suggesting that the perception of patient access barriers increases with more telehealth use.

Finally, high users are more likely to have positive attitudes toward telehealth use and to want to continue using it in the future. Provider satisfaction and experience are important quality indicators in health care.²⁵ Provider advantages such as improved continuity of care with patients facing challenges to attend office visits, flexible scheduling, increased clinical support for local providers caring for high-risk patients, expanding the reach to underserved areas, and reducing overhead costs associated with in-person visits are some advantages of telehealth identified by SMFM.²⁶ Whether telehealth services are maintained at the same level after the pandemic will require ongoing investigation; however, there may be utility in expanding the use of hybrid models combining in-person and telehealth visits.

STRENGTHS AND LIMITATIONS

Strengths of this study include the geographic diversity and variety of clinical practice settings of the respondents and the rich array of variables included in our multivariate models. Among limitations, although our response rate was similar to other recent physician online surveys,^19,27 the 16.4% response rate for this survey was low, potentially contributing to selection bias. In addition, although MFM providers generally care for pregnant persons considered to be “high-risk,” it is unknown whether clinicians preferentially applied telehealth to low-risk or high-risk patients, or both. Our study was limited to in-depth understanding of usage of live video visits; change across other telehealth modalities has not been thoroughly examined.

Moreover, we did not assess the patient perspective, nor provider perceptions of perceived patient safety or privacy and security concerns or cost-related factors as potential barriers to use. During a pandemic when prenatal care practices are suddenly modified, pregnant women may feel abandoned and miss face-to-face encounters with their providers. Furthermore, the cross-sectional study design only captured one given moment, and conditions have changed throughout the pandemic. There is need for follow-up research that is longitudinal, especially given the rapidly changing health care delivery environment and includes pregnancy outcomes of pregnant persons who received prenatal care through telehealth.

IMPLICATIONS

Our study demonstrates that digital telehealth infrastructure is a critical enabler of virtual patient care during the pandemic. Moreover, providing obstetric care through telehealth is associated with ample SMFM clinician satisfaction and MFMs with high utilization are interested in continuing telehealth use postpandemic, potentially improving access to care and working to eliminate disparities. Key perceived barriers to telehealth use identified by SMFM clinicians signal opportunities for advocacy efforts by obstetricians at the state and national levels.

Future studies should identify how these barriers affect start-up of new programs and expansion of existing telehealth programs. Advocacy to promote expansion of telehealth include addressing barriers to access, such as the “digital divide” and safety net provider capacity limitations, by bolstering access to the internet, smartphones, and data plans in communities. In addition, providing grants and technical assistance to safety net providers can increase their capacity to provide telehealth services.¹⁵

Ongoing research is needed to help guide providers on when and how to implement telehealth along the maternal-newborn health care continuum²⁸ in both inpatient and outpatient settings.²⁰ Since telehealth has not been equitably practiced in underserved communities, exacerbating inequalities, efforts should be made to overcome barriers, such as internet access or smartphone availability. Our cross-sectional study compared pre- and postpandemic telehealth use. Prospective studies measuring provider use and satisfaction with telehealth could overcome retrospective bias. Finally, more research is required to elucidate the patient perspective and clarify what barriers patients face in accessing telehealth resources.

Conclusions

Live virtual visits have been widely used by MFM clinicians providing care during the pandemic. Although this telehealth modality enjoys ample satisfaction and is likely to be continued postpandemic, additional studies are warranted to study optimal patient populations and effectiveness of virtual prenatal and postnatal care.

Authors’ Contributions

Conceptualization (lead), methodology (lead), writing—original draft (lead), supervision (lead), formal analysis (supporting), funding acquisition (lead), and review and editing (equal) by S.G. Formal analysis (lead), visualization (lead), data curation (supporting), project administration (lead), and review and editing (equal) by G.N.K. Formal analysis (supporting), visualization (supporting), and review and editing (equal) by G.I.C. Conceptualization (supporting), resources (supporting), data curation (lead), formal analysis (supporting), and review and editing (equal) by P.C.T. Conceptualization (supporting), resources (lead), formal analysis (supporting), and review and editing (equal) by S.H.S.

Disclosure Statement

No competing financial interests exist.

Funding Information

This study was partially funded by the Wallace Center for Maternal, Child and Adolescent Health Center at UC Berkeley. The sponsor funds projects consistent with its mission, giving academic freedom to investigators to design studies, collect, analyze, and interpret data and publish their study.

Supplementary Material

Supplementary Appendix SA1