Racial disparities in telehealth use during the coronavirus disease 2019 pandemic.

Our primary sample of 1,534 patients in the prepandemic timeframe was comprised of 20.9% African American (n = 210) and 68.9% Caucasian (n = 698) patients. The primary sample of 2,088 patients during the pandemic timeframe was comprised of 18.5% African American (n = 267) and 66.6% Caucasian (n = 963) patients. Table 1 includes patient characteristics comparing the prepandemic (n = 500) and pandemic (n = 500) cohorts. There were no statistically significant differences between patient age (32.7 ± 6.2 vs. 32.7 ± 6.0 years, P = 96), gravidity (1.29 ± 1.7 vs. 1.33 ± 1.9, P= .72), or number of living children (0.61 ± 1.1 vs. 0.56 ± 1.0, P =.49) between the prepandemic and pandemic cohorts, respectively. There was no difference in the proportion of patients who self-identified as Hispanic/Latino ethnicity between the prepandemic and pandemic cohorts (4.0% vs. 3.2%, P= .79). More patients in the prepandemic cohort identified as African American in comparison with the pandemic cohort (33.0% vs. 27.0%), although the overall composition of the cohorts with regard to race did not differ statistically significantly ( P= .20). With regards to insurance status, overall patients in the prepandemic cohort were less likely to have commercial insurance (64.4% vs. 72.8%, P= .011) in comparison with the pandemic cohort. Table 1 Medical history comparing patients in the prepandemic and pandemic cohorts. Characteristic Total n = 1,000 PrePandemic n = 500 Pandemic n = 500 P value Age (y) 32.7 (6.1) 32.7 (6.2) 32.7 (6.0) .96 Race .20 Caucasian 68 (63.8) 307 (61.4) 331 (66.2) African American 300 (30.0) 165 (33.0) 135 (27.0) Native American or Alaskan Native 4 (0.4) 3 (0.6) 1 (0.2) Asian 28 (2.8) 12 (2.4) 16 (3.2) Native Hawaiian or Other Pacific Islander 0 0 0 Other 30 (3.0) 13 (2.6) 17 (3.4) Ethnicity .79 Hispanic/Latino 36 (3.6) 20 (4.0) 16 (3.2) Non-Hispanic/Latino 882 (88.2) 439 (87.8) 443 (88.6) Did not disclose 82 (8.2) 41 (8.2) 41 (8.2) Insurance .011 Commercial 686 (68.6) 322 (64.4) 364 (72.8) Public 185 (18.5) 109 (21.8) 76 (15.2) Other 129 (12.9) 69 (13.8) 60 (12.0) Gravida 1.31 (1.8) 1.29 (1.7) 1.33 (1.9) .72 Parity: living 0.59 (1.1) 0.61 (1.1) 0.56 (1.0) .49 Reported as n (%), mean (±SD), or median [interquartile range]. Bolded P values less than .05 were considered statistically significant. Medical history comparing patients in the prepandemic and pandemic cohorts. Reported as n (%), mean (±SD), or median [interquartile range]. Bolded P values less than .05 were considered statistically significant. Table 2 describes the parameters of initial infertility consultations comparing the prepandemic and pandemic cohorts. There were no statistically significant differences in rates of missed appointments between the prepandemic and pandemic cohorts (32.2% vs. 32.4%, P =.78). The reason for missed appointments, however, differed significantly as the prepandemic cohort had greater rates of no-shows (49.4% vs. 27.8%) and fewer cancellations (50.6% vs. 72.2%) in comparison with the pandemic cohort ( P <.0001). There were no statistically significant differences in the proportion of patients who rescheduled their appointments (23.6% vs. 25.5%, P= .99) or the average days to rescheduled appointments (136.1 ± 144.4 vs. 125.6 ± 128.0, P= .75) between the prepandemic and pandemic cohorts. As expected, the prepandemic cohort had lower rates of telehealth visits in comparison with the pandemic cohort (1.2% vs. 64.2%, P< .0001). The length of visits between the prepandemic and pandemic cohorts differed significantly because the prepandemic cohort had fewer appointments lasting 30–44 minutes (15.8% vs. 31.1%) and more appointments lasting >59 minutes (38.6% vs. 19.7%) in comparison with the pandemic cohort ( P< .0001). As noted in Table 2 , there were no statistically significant differences in the diagnosis associated with infertility ( P= .094) or the proportion of patients who initiated an IVF cycle (17.4% vs. 17.3%, P= .96) between the prepandemic and pandemic cohorts, respectively. Table 2 Infertility appointment data comparing patients in the prepandemic and pandemic cohorts. Characteristic Total n = 1,000 Prepandemic n = 500 Pandemic n = 500 P value Missed Appt Yes 328 (32.8) 166 (33.2) 162 (32.4) .78 Missed Appt Reason <.0001 No show 127 (38.7) 82 (49.4) 45 (27.8) Cancelled 201 (61.3) 84 (50.6) 117 (72.2) Rescheduled Appt Yes 77 (23.5) 39 (23.6) 38 (23.5) 0.99 Days to rescheduled Appt 131.2 (136.0) 136.1 (144.4) 125.6 (128.0) 0.75 Length of Appt (min) <.0001 59 218 (29.1) 144 (38.6) 74 (19.7) Unknown 6 (0.8) 3 (0.8) 3 (0.8) Visit type <.0001 In person 673 (67.3) 494 (98.8) 179 (35.8) Telemedicine 327 (32.7) 6 (1.2) 321 (64.2) Diagnosis .094 Tubal factor 88 (11.7) 44 (11.8) 44 (11.7) Male factor 89 (11.9) 40 (10.7) 49 (13.0) Anovulation 18 (2.4) 6 (1.6) 12 (3.2) Endometriosis 50 (6.7) 32 (8.6) 18 (4.8) Leiomyoma 41 (5.5) 20 (5.4) 21 (5.6) PCOS 141 (18.8) 60 (16.1) 81 (21.5) Unexplained 90 (12.0) 42 (11.3) 48 (12.8) Other 194 (25.9) 93 (24.9) 101 (26.9) Not documented 123 (16.4) 70 (18.8) 53 (14.1) IVF initiated 130 (17.4) 65 (17.4) 65 (17.3) .96 Reported as n (%), mean (±SD). Bolded P values less than .05 were considered statistically significant. Appt = appointment; IVF = in vitro fertilization; PCOS = polycystic ovary syndrome. Infertility appointment data comparing patients in the prepandemic and pandemic cohorts. Reported as n (%), mean (±SD). Bolded P values less than .05 were considered statistically significant. Appt = appointment; IVF = in vitro fertilization; PCOS = polycystic ovary syndrome. Table 3 categorizes initial infertility consultation data by race, comparing African American patients to patients of all other races within both the prepandemic and pandemic cohorts. There were fewer African American patients compared with all other patients in both the prepandemic (33.0% vs. 67.0%, P< .0001) and pandemic (27.0% vs. 73.0%, P< .0001) cohorts. African American patients in comparison to all other patients were less likely to have commercial insurance in both the prepandemic (41.2% vs. 75.8%, P< .0001) and pandemic (57.0% vs. 78.6%, P< .0001) cohorts, although the proportion of African American patients with commercial insurance increased from the prepandemic to the pandemic timeframe. African American patients were more likely to miss their appointments compared with all other patients in both the prepandemic (47.3% vs. 26.3%, P< .0001) and pandemic (51.9% vs. 25.2%, P< .0001) cohorts. Further, African American patients compared with all other patients were more likely to no-show rather than cancel their appointments in both the prepandemic (69.2% vs. 31.8%, P< .0001) and pandemic (35.7% vs. 21.7%, P= .05) cohorts. The rates of rescheduled appointments did not differ between African American and all other patients in the prepandemic (19.2% vs. 27.3%, P= .22) and pandemic (18.6% vs. 27.2%, P= .20) cohorts, or did the days to rescheduled appointment in both the prepandemic (131.57 ± 153.9 vs. 138.9 ± 141.7, P= .84) and pandemic (130.0 ± 151.9 vs. 123.1 ± 116.3, P= .89) cohorts. Table 3 Demographic and infertility appointment data comparing African American patients to patients of all other races in both the prepandemic and pandemic cohorts. Characteristic Prepandemic P value Pandemic P value Total AA All other Total AA All other Patients 500 (100) 165 (33.0) 335 (67.0) <.0001 500 (100) 135 (27.0) 365 (73.0) <.0001 Insurance <.0001 <.0001  Commercial 322 (64.4) 68 (41.2) 254 (75.8) 364 (72.8) 77 (57.0) 287 (78.6)  Public 109 (21.8) 74 (44.8) 35 (10.4) 76 (15.2) 46 (34.1) 30 (8.2)  Other 69 (13.8) 23 (13.9) 46 (13.7) 60 (12.0) 12 (8.9) 48 (13.2) Missed Appt 166 (33.2) 78 (47.3) 88 (26.3) <.0001 162 (32.4) 70 (51.9) 92 (25.2) <.0001 Missed Appt Reason <.0001 .05  No show 82 (49.4) 54 (69.2) 28 (31.8) 45 (27.8) 25 (35.7) 20 (21.7)  Cancel 84 (50.6) 24 (30.8) 60 (68.2) 117 (72.2) 45 (64.3) 72 (78.3) Rescheduled appt 39 (23.5) 15 (19.2) 24 (27.3) .22 38 (23.5) 13 (18.6) 25 (27.2) .20 Days to rescheduled Appt 136.1 (144.4) 131.57 (153.9) 138.9 (141.7) .84 125.6 (128.0) 130.0 (151.9) 123.1 (116.3) .89 Length of Appt (min) .021 .47  <15 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)  15–29 5 (1.3) 0 (0) 5 (1.8) 6 (1.6) 2 (2.6) 4 (1.3)  30–44 59 (15.8) 19 (18.6) 40 (14.8) 117 (31.1) 29 (37.2) 88 (29.5)  45–59 162 (43.4) 39 (38.2) 123 (45.4) 176 (46.8) 30 (38.5) 146 (49.0)  >59 144 (38.6) 41 (40.2) 103 (38.0) 74 (19.7) 16 (20.5) 58 (19.5)  Unknown 3 (0.8) 3 (2.9) 0 (0) 3 (0.8) 1 (1.3) 2 (0.7) Visit type .99 .042  In person 494 (98.8) 163 (98.8) 331 (98.8) 179 (35.8) 58 (43.0) 121 (33.2)  Telemedicine 6 (1.2) 2 (1.2) 4 (1.2) 321 (64.2) 77 (57.0) 244 (66.8) Diagnosis .0002 .069  Tubal factor 44 (11.8) 25 (24.5) 19 (7.0) 44 (11.7) 14 (17.9) 30 (10.1)  Male factor 40 (10.7) 10 (9.8) 30 (11.1) 49 (13.0) 5 (6.4) 44 (14.8)  Anovulation 6 (1.6) 2 (2.0) 4 (1.5) 12 (3.2) 3 (3.8) 9 (3.0) Endometriosis 32 (8.6) 7 (6.9) 25 (9.2) 18 (4.8) 2 (2.6) 16 (5.4)  Leiomyoma 20 (5.4) 9 (8.8) 11 (4.1) 21 (5.6) 7 (9.0) 14 (4.7)  PCOS 60 (16.1) 14 (13.7) 46 (17.0) 81 (21.5) 14 (17.9) 67 (22.5)  Unexplained 42 (11.3) 4 (3.9) 38 (14.0) 48 (12.8) 7 (9.0) 41 (13.8)  Other 93 (24.9) 22 (21.6) 71 (26.2) 101 (26.9) 17 (21.8) 84 (28.2)  Not documented 70 (18.8) 21 (20.6) 49 (18.1) 53 (14.1) 15 (19.2) 38 (12.8) IVF initiated 65 (17.4) 7 (6.9) 58 (21.4) .0011 65 (17.3) 5 (6.4) 60 (20.1) .0045 Reported as n (%), mean (±SD), or median [interquartile range]. Bolded P values less than .05 were considered statistically significant. AA = African American; All other = all other races; Appt = appointment; IVF = in vitro fertilization; PCOS = polycystic ovary syndrome. Demographic and infertility appointment data comparing African American patients to patients of all other races in both the prepandemic and pandemic cohorts. Reported as n (%), mean (±SD), or median [interquartile range]. Bolded P values less than .05 were considered statistically significant. AA = African American; All other = all other races; Appt = appointment; IVF = in vitro fertilization; PCOS = polycystic ovary syndrome. In the prepandemic cohort, African American patients vs. all other patients were more likely to have appointments lasting 30–44 minutes (18.6% vs. 14.8%) and less likely to have appointments lasting 45–59 minutes (38.2% vs. 45.4%) ( P= .021). No statistically significant differences in appointment length were noted between African Americans and all other patients in the pandemic cohort ( P= .47). Most visits were in person in the prepandemic cohort, with no statistically significant differences between African American and all other patients (98.8% vs. 98.8%, P= .99). However, African American patients compared with all other patients were less likely to have a telehealth visit in the pandemic cohort (57.0% vs. 66.8%, P= .042). As noted in Table 3 , African Americans and all other patients differed significantly in documented diagnoses associated with infertility in the prepandemic cohort ( P= .0002), but no statistically significant differences were noted in the pandemic cohort ( P= .069). African American patients compared with all other patients were less likely to initiate IVF in both the prepandemic (6.9% vs. 21.4%, P =.0011) and pandemic (6.4% vs. 20.1%, P= .0045) cohorts. Table 4 includes the variables used in the multivariable regression model predicting presentation to the initial infertility consultation vs. no-show or cancellation of all the patients who were scheduled for an initial visit. When controlling for other variables, African American patients were less likely to present for an initial consultation (odds ratio [OR] 0.37, 95% confidence interval [CI] 0.28–0.50), and telehealth use predicted presentation to the initial consultation (OR 1.54, 95% CI 1.04–2.27). The timing of the appointment relative to the onset of the pandemic (OR 0.74, 95% CI 0.52–1.06) and insurance status (OR 0.87, 95% CI 0.72–1.06) were not associated with missed appointment rates. Table 4 Multivariable logistic regression model predicting presentation to the initial infertility consultation vs. no-show or cancellation of the initial infertility consultation for all scheduled patients (n = 1,000). Covariate Adjusted odds ratio a 95% CI Prepandemic vs. pandemic 0.74 (0.52, 1.06) African American race vs. all other races 0.37 (0.28, 0.50) Commercial vs. public insurance 0.87 (0.72, 1.06) Telehealth vs. in person 1.54 (1.04, 2.27) CI = confidence interval. Bolded P values less than .05 were considered statistically significant. a The model was a logistic regression model predicting when a patient presented to an initial infertility consultation vs. no-showed and cancelled, controlling for timing relative to the onset of the pandemic, race, insurance status, and type of visit. Multivariable logistic regression model predicting presentation to the initial infertility consultation vs. no-show or cancellation of the initial infertility consultation for all scheduled patients (n = 1,000). CI = confidence interval. Bolded P values less than .05 were considered statistically significant. The model was a logistic regression model predicting when a patient presented to an initial infertility consultation vs. no-showed and cancelled, controlling for timing relative to the onset of the pandemic, race, insurance status, and type of visit.

With approval from the Institutional Review Board, we performed a retrospective cohort study of patients aged between 21 and 45 years who presented for an initial infertility consultation between January 1, 2019, and June 1, 2021, at the University Hospitals Fertility Center in Cleveland, Ohio. The primary outcome was the change in the proportion of patients using telehealth who identify as African American compared with all other patients. The secondary outcome was to identify factors associated with the presentation to the initial consultation rather than no-shows or cancellations. Exploratory outcomes included the length of infertility appointments as defined by Current Procedural Terminology (CPT) and IVF initiation. Patients were excluded when they were completing a gestational carrier or egg donation cycle, seeking fertility preservation in the setting of a cancer diagnosis, or did not have a listed race in the electronic health record (EHR). Missing data for ethnicity, length of appointment, and infertility diagnosis were noted. Telehealth during the COVID-19 pandemic was implemented at our institution in March 2020. Before the onset of the pandemic, we did not routinely complete telehealth visits for initial infertility consultations. After the onset of the pandemic, we conducted new patient visits virtually or in person, with flexibility per patient request. These policies were implemented for all patients presenting to the clinic, regardless of insurance status or additional patient factors. Patients presenting between January 1, 2019, and February 29, 2020, were classified as the prepandemic cohort, and patients presenting between March 1, 2020, and June 1, 2021, were classified as the pandemic cohort. Initial infertility consultation was defined as an appointment with no previous consultation with any infertility provider (attending physician, REI fellow, or nurse practitioner) within our system or no infertility consultation within the last three years. For sample size determination, we estimated that the proportion of patients seeking care who are African American prepandemic was 10%, and after the onset of the pandemic, it was 20% on the basis of available analyses at that time ( 14 , 23 , 28 ). An priori power calculation determined that 500 patients per group would allow us a 90% power to detect a 7% difference in the proportion of patients scheduling infertility visits who were African American prepandemic compared with a pandemic, with an alpha of 0.05. To ensure adequate power and the feasibility of manual retrospective data collection, 500 patients per group were chosen. We randomly selected 500 patients for each cohort from a total available study population of 1,534 patients identified in the prepandemic timeframe and 2,088 patients identified in the pandemic timeframe that met inclusion criteria. Random selection was prioritized to capture different time points of care throughout the pandemic period to minimize the impact of fluctuations related to local surges in COVID-19 infections. A random number generator was used to assign a number between 1 and 1,534 (prepandemic cohort) or 2,088 (pandemic cohort) to each patient, and the first 500 patients in each cohort were included in the final analysis. All data were collected and stored in a secure REDCap database ( 29 ). The EHR was queried for demographic data, including age, race, ethnicity, type of insurance, and obstetric history. Variables regarding initial infertility consultation were recorded, including missed appointment rates and the reason for a missed appointment: either no-show (patient did not present to a scheduled visit) or patient-initiated cancellation. Rates of rescheduled appointments, length of infertility consultation CPT code, diagnoses associated with infertility as documented in the EHR, and initiation of IVF were also noted. Demographic, medical history, and initial infertility consultation factors were evaluated by summarizing approximately normally distributed continuous measures with means and standard deviations or frequencies and percentages and comparing them using two-sample t- tests. Categorical factors were summarized using frequencies and percentages and compared using Pearson’s chi-squared tests. P values of <.05 were considered significant. For comparisons between races, African American patients were compared with a cohort comprised of patients of Caucasian, Native American/Alaskan Native, Asian, Native Hawaiian/Other Pacific Islander, and other races, grouped as “all other” patients. We performed a multivariable logistic regression analysis to evaluate factors associated with a presentation to the initial infertility consultation compared with no-shows or cancellations for all scheduled patients. Variables were determined on the basis of the statistical significance of univariate analysis and factors that may impact compliance with scheduled appointments, including insurance status, telehealth vs. in-person visits, and timing relative to the onset of the pandemic. Data analysis was performed using GraphPad Prism (GraphPad Prism Version 9.3.1 (350) for macOS, GraphPad Software, LLC. San Diego, CA).

This large, single-center, retrospective cohort study demonstrates that the COVID-19 pandemic has resulted in a broad change in health care delivery in REI as telehealth continues to be a mainstay of practice today ( 30 , 31 , 32 ). However, the potential impact of telehealth on disparities in access to care must be considered. Here, we demonstrate that African American patients, relative to patients of other races, had a reduced likelihood of seeking infertility care after the implementation of telehealth modalities in the setting of a global pandemic, consistent with previous studies demonstrating decreased uptake of telehealth in African American populations ( 15 , 33 , 34 ). Additionally, fewer patients with public health insurance sought infertility consultation in the pandemic cohort, and African American patients compared with all other patients were more likely to have public health insurance. Moreover, we identified that African American race was independently associated with missed appointments when controlling for both insurance status and telehealth use, demonstrating that flexibility in access to care through telemedicine for an initial consult did not offset the inherent stressors occurring during the COVID-19 pandemic, including but not limited to personal and familial health challenges as well as economic vulnerability. Health disparities with regard to socioeconomic status and race in the field of REI are areas of ongoing concern. Notably, both race and socioeconomic status impact the success rates of ART. Caucasian patients and patients with higher levels of income and education disproportionately achieve a successful live birth after IVF, whereas African American, Asian, and Hispanic patients have reduced odds of live birth via ART ( 22 , 23 , 24 ). Racial and socioeconomic factors also serve as independent predictors of access to ART itself, and Caucasian patients are more likely to use ART compared with minority patients ( 22 , 23 , 25 ). In a study by Feinberg et al. ( 28 ) evaluating the equal access to care model, the use of ART in the African American study population was found to be four times that of the national average for African American patients. Interestingly, this study demonstrated a clinically significant decrease in rates of live birth and a statistically significant increase in spontaneous abortions in African American vs. Caucasian patients despite the adoption of this equal access model, indicating that other underlying factors may contribute to such disparities ( 22 , 24 , 28 ). A proposed benefit of telehealth is the mitigation of racial and socioeconomic disparities, but it may, in fact, potentiate existing disparities ( 35 ). A lack of access to appropriate technology contributes to the “digital divide,” a history of systemic disfranchizement that contributes to distrust in the evolving practice patterns of medical institutions, and payer hesitation to expand telehealth coverage contribute to disparities in access ( 36 ). Some studies have shown that African American patients, compared with Caucasian patients, are less likely to use telehealth, especially video visits, during the COVID-19 pandemic ( 15 , 33 , 34 ). Our data support decreased telehealth use by African American patients compared with all other patients in the pandemic cohort in a clinic model with flexible access to telehealth services per patient preference. Further, Cleveland’s population is 38.6% Caucasian and 47.4% African American. Thus, African American patients comprise a smaller proportion of our patient population compared with what is expected on the basis of the demographic makeup of our community ( 37 ). Moreover, telehealth use is an independent predictor of presentation to an initial infertility consultation in our cohort, highlighting its importance in connecting patients to the health care system and initiating infertility care. Although not an independent predictor in our model, patients with public health insurance, more commonly noted in African American patients, were less likely to present for an initial infertility consultation in the pandemic cohort. The COVID-19 pandemic incited a rapid expansion of coverage for telehealth services for Medicaid beneficiaries. In our state of Ohio, coverage was extended to phone-call visits in the early stages of the pandemic ( 38 ). In spite of an expansion of coverage for telehealth services, racial minorities, patients with limited English proficiency, and those with lower income levels remained less likely to use telehealth services covered by Medicaid ( 38 , 39 ). Several factors may contribute to these findings, including a lack of awareness or uncertainty regarding coverage via public health insurance for infertility care. Importantly, patients with public insurance, unlike those with commercial insurance, do not have coverage for diagnostic testing and initial infertility consultations in Ohio. Further, neither commercial nor public insurance mandates coverage for fertility treatments in Ohio. Thus, the changes made to reimbursement for telehealth services during the COVID-19 pandemic have not fully addressed the disparities in access to care for patients with public insurance. We additionally demonstrate that African American patients, compared with all other patients, were more likely to miss their appointment in both the prepandemic and pandemic cohorts, independent of both telehealth use and insurance type. These findings likely demonstrate barriers to care that are incompletely captured in our analysis. Logistical challenges because of travel distance or work constraints serve as important barriers to initiating and undergoing ART therapies in the traditional, in-person model ( 40 , 41 ). Telehealth utilization, in theory, can limit the impact of these obstacles. However, both the prepandemic and pandemic cohorts in our study demonstrate decreased IVF initiation in the African American population compared with all other patients. This mirrors studies that note increased time to treatment and decreased uptake of IVF in African American patients ( 40 , 42 ). Compliance with scheduled appointments may reflect also a general lack of trust in medical institutions and negative patient experiences with fertility treatments. Studies have demonstrated that African American patients, compared with Caucasian patients, experience a lack of culturally competent care in the field of REI ( 43 ). The underlying intangible yet untenable barriers that exist in infertility care for African American and other racial minority patients remain unaddressed in spite of telehealth implementation ( 44 , 45 ). Overall, patients in the pandemic cohort were more likely to cancel rather than no-show for their missed appointments compared with the prepandemic cohort. This trend may reflect the increasing familiarity and acceptance of telehealth by patients in our geographic area. We theorize that patients may be taking better advantage of available options to cancel and/or reschedule appointments with the use of available interfaces, engaging more with appointment management than in the prepandemic era. However, this same shift was not seen in African American patients because they were more likely to no-show than cancel their appointments in both cohorts. “No-shows” impact the flow of patient care, clinic productivity, and the cost of health care delivery ( 46 , 47 , 48 , 49 ). Further, patients with high rates of no-shows may be dismissed from practices, delaying patients’ abilities to seek care and leading to further adverse outcomes ( 49 ). Of note, our overall rates of missed and cancelled appointments are high, underlying the need to understand the inefficiencies in our current academic system that cares for a large, underserved population. Telehealth has the benefit of mitigating the waste of resources with more flexible scheduling and cancelled appointments compared with missed appointments, offering time for additional patients who are seeking care ( 50 ). Several factors, including patient age, time to scheduled appointments, and insurance status, have previously been shown to predict missed appointments in the primary care setting ( 47 , 48 , 51 ). The association of race with missed appointments is less clear ( 47 ). Reasons for missed rather than cancelled appointments include barriers to transportation, responsibilities as a caregiver, a lack of childcare, compounding stresses resulting in challenges with recalling appointment details, and communication errors ( 47 , 52 ). Telehealth has the potential to address some of these barriers, but changes in health care delivery in our cohort did not alter patterns of missed appointments in the African American population. It is important to note the decreased rates of IVF initiation in African Americans compared with all other patients in both the prepandemic and pandemic cohorts. Several factors likely impact these rates, one of which may be the patient-provider relationship built during consultations or individual provider biases ( 41 , 53 , 54 , 55 , 56 ). In our study, we used appointment length per CPT billing as a proxy for such relationship building and found that the lengths of appointments decreased in the pandemic cohort. When analyzing appointment length by race, African American patients in the prepandemic cohort were more likely to have appointments lasting >1 hour, but we noted a trend toward shorter appointments in the African American population in the pandemic cohort. It is important to note that CPT code data reflects both time spent and complexity of care and thus only serves as a proxy for the length of an appointment. Although beyond the scope of this study, these findings warrant further analysis of the impact of telemedicine on the relationships built between minority populations and providers in the setting of infertility care. There are several strengths in our study. We present a large, relatively diverse cohort of patients that is adequately powered to identify differences in telehealth usage prepandemic and after the start of the pandemic between African Americans and all other patients. We also explore an understudied area in REI, highlighting the impact of changes in health care delivery on established disparities in the field. However, there are limitations in our study design that warrant mention, including biases inherent to retrospective analyses. Limitations in granular data collection prevented differentiation between phone and video telehealth appointments. Similarly, insurance status was classified as “public” or “commercial” without capturing the intricacies of infertility coverage. Further, coverage for fertility services differs between patients with commercial and public insurance in Ohio. Although we had missing data in our study, the proportions were relatively low for variables such as ethnicity (8.2%), appointment length (0.8%), and diagnosis associated with infertility (16.4%), with minimal impact on our primary objective. An important limitation of our study is that we did not control for social determinants of health, warranting further investigations on the impact of geography, socioeconomic differences, education, and other contributors to disparities in infertility care. Further, utilizing a random convenience sample prevents nuanced tracking of trends in infertility consultation uptake throughout the year; thus, variability in volume on the basis of season and additional factors cannot be compared. However, given that surges in COVID-19 infections and unemployment were likely more impactful on care uptake in our population, this sampling method mitigated inadvertent bias in our sample. Additionally, our initial sample demonstrated an absolute increase of 57 African American patients and 265 Caucasian patients between the pandemic era and the prepandemic era. However, there was a relative decrease in African American (2.4%) and Caucasian (2.0%) patients seeking care during the pandemic compared with the prepandemic. Thus, using proportions from a random convenience sample in our study did not mask an absolute or relative increase in African American patients compared with Caucasian patients seeking care in our practice during the pandemic. Our data was collected at a single, academic institution representing an urban population in Ohio that was severely impacted by COVID-19, especially with regards to overrepresentation of African American patients in both cases and deaths resulting from the virus ( 57 ). Such factors likely impact our study’s generalizability to other institutions and locations. It is likely that patients from Native American, Asian, and other minority racial backgrounds have different experiences from Caucasian patients, but we were unable to complete a subanalysis of these trends despite of an adequate sample size. Lastly, telehealth is a relatively new modality to deliver health care, especially in the field of REI. It is not possible to comment on the health and technology literacy of our two cohorts. It is possible that those with greater comfort with technology were the predominant users of telehealth before the onset of the pandemic, whereas the pandemic era population may represent a more diverse user profile that has less comfort with technology interfaces, impacting the no-show and cancellation rates. Additionally, it is possible that these initial trends and disparities will diminish over time with increasing familiarity and use of these platforms in all populations.

Our study adds to a growing body of literature investigating the impact of telehealth in specialty clinics. Telehealth is likely to have an enduring presence, and its potential detrimental impact on REI must be evaluated cautiously as it may potentiate and exacerbate existing racial disparities. The benefits of telehealth in providing efficient and accessible infertility care for many must be matched with the identification of its shortcomings in delivering care to all and the prioritization of addressing underlying barriers to infertility care in African American and other underrepresented populations. Further studies are warranted to assess how socioeconomic disparities impact telehealth utilization and how the use of telehealth in fertility clinics impacts fertility outcomes.