Key Points

• Moderate-strength evidence from studies comparing reduced versus traditional visit schedules did not indicate differences in infant outcomes between visit schedules regarding

  • gestational age at birth (2 RCTs, 2 NRCSs; mean difference [MD] 0 days),
  • being small for gestational age (3 RCTs; summary odds ratio [OR] 1.08, 95% confidence interval [CI] 0.70 to 1.66),
  • Apgar score (3 RCTs, 2 NRCSs; nonsignificant ORs, MD 0 at 1 and 5 minutes), and
  • neonatal intensive care unit (NICU) admission (3 RCTs, 2 NRCSs; summary OR 1.05, 95% CI 0.74 to 1.50).
• Low-strength evidence did not indicate differences between visit schedule groups regarding

  • maternal anxiety (3 RCTs; incomplete data reported),
  • preterm birth (1 RCT, 2 NRCSs; nonsignificant ORs), and
  • low birth weight (1 RCT, 3 NRCSs; summary OR 1.02, 95% CI 0.82 to 1.25).
• There is insufficient evidence for numerous prioritized outcomes of interest (e.g., completion of American College of Obstetricians and Gynecologists [ACOG] recommended services and patient experience measures).
• Studies did not evaluate heterogeneity of treatment effect (whether some specific groups of patients may have had better, or worse, outcomes with different visit schedules) or equity issues (whether certain classes of patients might be disadvantaged by introduction of reduced visit schedules).
• Qualitative studies suggest that

  • Providers perceive reduced visit schedules may reduce inconveniences of attending multiple appointments for individuals with low-risk pregnancies and may allow more clinic time to be directed to those with high-risk pregnancies.
  • Patients and providers expressed concerns that some patients may not have the confidence to independently manage their pregnancies under reduced care models and that fewer visits may compromise patients’ antenatal care in terms of meeting their psychosocial and physical needs. Providers expressed concerns that decades of familiarity with the traditional care model may hamper patients’ acceptance of a change in model. Patients noted emotional discomfort with reduced care and inversely positive emotion with a traditional care schedule. Providers expressed fears that reduced care may lead to staff cutbacks and emphasized their perceived need for a supportive management, without which, they expressed concerns about repercussions from management in the event of a rare adverse outcome associated with patients’ maternity care.

Summary of Findings

Overall, 10 studies (in 13 publications) compared different antenatal care schedules. Nine studies evaluated typical individual visits.^38–49 The tenth study compared (a reduced number of) group visits with traditional individual visits (and is summarized separately, below).⁵⁰ The 10 studies included five RCTs (N = 5748) and five NRCSs (N = 8987). In addition, four qualitative research studies evaluated perspectives on the number of antenatal visits from patients, providers, and clinic leadership. These are summarized following the comparative quantitative studies.

Table 1 lays out the scheduled visits during pregnancy for each group within each study. All studies compared alternative reduced schedules with a “traditional” or “standard” visit schedule (13 or 14 visits). Across studies, traditional (or standard) visit schedules were mostly consistent with the ACOG guideline of about 14 visits: every 4 weeks from week 8 to week 28, every 2 weeks from week 30 to 36, and weekly thereafter. There were some minor variations mostly due to a lack of explicit reporting of visit schedules prior to week 16, allowance for a range of timing (particularly for earlier visits), and whether a visit at week 41 was explicitly included. Overall, reported traditional schedules included 12 to 15 visits. In contrast, the number and timing of alternative visit schedules varied substantially across studies. The number of reduced-schedule visits ranged from 6 or 7 (depending on participant parity) to 10. Most studies scheduled visits at weeks 16, 24, 32, 36, 38, and 40.

The five RCTs^{39, 42, 44, 45, 48} and three NRCSs^{38, 40, 50} were funded by nonindustry sources. Two NRCSs did not report on the funding source.^{43, 49}

Tables in Appendixes C and D provide the design details, participant characteristics, detailed schedule data, and full results.

Table 1

Timing of scheduled routine antenatal visits (weeks).

Reduced Versus Traditional Schedule of Individual Visits

Nine studies compared a reduced with a traditional schedule of individual visits (one on one with the provider). Five were RCTs (in 8 articles^{39, 41, 42, 44–48}) and four were NRCSs.^{38, 40, 43, 49} The studies compared 6 or 7 to 10 scheduled visits to a traditional schedule of about 13 or 14 visits. The five RCTs included between 80 and 2692 pregnant patients (total 5748). The four NRCSs included between 214 and 3882 patients (total 8899).

Five studies were conducted in the United States (Marko 2019, McDuffie 1996, Partridge 2005, Walker 1997, and Ward 1999), three studies in England (Jewell 2000, Ross-McGill 2000, and Sikorski 1996), and one in Sweden (Berglund 1998). Most studies (implicitly or explicitly) included primarily non-minority (White) participants (about 80% to 94%) (Table 2). In contrast, Walker 1997 investigated a mostly minority population in Southern California, 74 percent of whom were Hispanic, 56 percent spoke only Spanish, and 68 percent were not born in the United States. In addition, a study in the Washington, DC area (Marko 2019) evaluated a mixed population, 30 percent of whom were African-American and 7 percent Hispanic. In most studies that reported data, about 85 to 95 percent of participants cohabitated with their partners. The average ages of participants ranged from 24 to 33 years. Only one study (Marko 2019) reported body mass index (BMI), at screening, which was similar between groups (mean 22.9 kg/m² in the reduced-schedule group and 24.9 kg/m² in the traditional-schedule group).

All but one study recruited pregnant patients who were at low antenatal risk (Table 2). In contrast, Berglund 1998 included all those receiving antenatal care (of whom 25% had undefined “obstetric [or] medical history”, 22% had complications during the current pregnancy, and an additional 10% had both); however, this study, like others, evaluated routine obstetrical care. McDuffie 1996 reported that pregestational hypertension was present in 4.5 percent and gestational hypertension (GHTN) was present in 19.0 percent of participants (similar in both groups). Sikorski 1996 reported the mean number of cigarettes smoked per day was 4.8 (not reported whether before or during pregnancy). Notably, though, each study defined risk differently.

Studies were conducted at sites that had a variety of provider patterns, but most studies did not explicitly describe who the providers were at the routine visits. In two studies (conducted in Sweden and England), midwives provided 100 percent of the routine antenatal care (Berglund 1998 and Ross-McGill 2000) (Table 3). One study (in southern California) provided routine care by certified nurse-midwives (Walker 1997). In two studies (conducted in Washington, DC and England) both obstetricians and midwives provided routine care (Marko 2019 and Sikorski 1996); although the percentage of visits by each provider was not provided. One study conducted on a U.S. military base provided care primarily by obstetrics residents and certified nurse practitioners, with oversight by supervising obstetricians (Ward 1999). One study (in Colorado) provided most care by obstetricians, but also by nurses and perinatologists (McDuffie 1996). The remaining three studies did not report the type of providers. Only three studies (McDuffie 1996, Walker 1997, Ward 1999) reported visit duration, which was 45 or 50 minutes for the intake visit and 10 to 15 minutes for return visits. Of note, the intake visit duration from these approximately 25-year-old studies may be longer than what is more commonly allotted in current practices (often 20 to 30 minutes).

The reduced schedules mostly dropped visits in the third trimester (e.g., every 4 instead of 2 weeks from weeks 28 to 36 and every other week instead of weekly thereafter) (Table 1). In two studies (Berglund 1998 and Jewell 2000), parous patients were offered one fewer visit than nulliparous patients. Also as noted above, the RCTs were mostly at low risk of bias (except for possible issues related to lack of blinding and, in one RCT, a high dropout rate), but the NRCSs suffered from lack of adjustment for any potentially confounding differences between studied groups.

Three of the NRCSs (Berglund 1998, Partridge 2005, and Ward 1999) were retrospective pre-post studies based on clinics changing their recommended numbers of visits. The fourth NRCS (Marko 2019) described itself as a “pragmatic controlled trial” that allocated participants into the experimental group versus the control group based on the operating system of the patient’s mobile phone (i.e., iOS or Android or Windows). Patients who had iPhones had fewer (9) visits but were provided with a smartphone app (Babyscripts) which delivered educational content and promoted home monitoring of blood pressure and weight with use of a Wi-Fi-connected scale and blood pressure cuff. For this study, evaluation of the difference in scheduled visits is confounded by use of the app and home monitoring. The study employed hierarchical regression models but did not describe what was adjusted for; although, notably, the study did not adjust for participant income, a potentially important confounder given that allocation was based on owning an iPhone.

Assessment of the risk of bias (RoB) of the included studies are presented in Tables C-4-1 (RCTs) and C-4-2 (NRCSs). Among the five RCTs addressing different routine schedules, four had no substantial methodological concerns and were deemed to be at low RoB (Jewell 2000, McDuffie 1997, Ross-McGill 2000, and Sikorski 1996); although the nature of the intervention precluded blinding of either the study participants or providers. One RCT (Walker 1997) had a very high dropout rate of 30 percent, although there were no differences between groups or between dropouts and retained participants; thus, this study was deemed to have some methodological concerns. All four NRCSs addressing different routine schedules were deemed to be at high RoB, primarily due to inadequate statistical analyses in four studies (i.e., they reported only crude, unadjusted analyses). As will be described, few studies evaluated any given outcome. Given the small number of studies per outcome, we could not formally assess risk of publication bias.

None of the studies evaluated or compared subgroups of patients or formally addressed heterogeneity of treatment effect (whether specific individuals or identifiable groups of patients may have better, or worse, outcomes than most patients with the implementation of reduced visit schedules). Similarly, no eligible studies evaluated equity issues (whether classes of patients, such as rural, poor, or non-English speakers, might have been further disadvantaged or have had worse outcomes with the implementation of reduced visit schedules).

Table 2

Study design and sample characteristics of studies addressing timing of scheduled routine antenatal visits.

Table 3

Summary of interventions in studies addressing timing of scheduled routine antenatal visits.

In the following text, the results are organized by outcome. We first describe the outcomes prioritized by stakeholders for which there are data (pertaining to maternal psychosocial, preference, and related outcomes; fetal, neonatal, and infant outcomes; care utilization; and harms). These are followed by other included outcomes; other included outcomes that were reported by only a single study are grouped together at the end of the list of outcomes.

Across studies, for all outcomes, there were no reported analyses of subgroups, subgroup differences, or heterogeneity of treatment effect. Except as noted, study findings were consistent with each other (generally finding no significant differences) such that no discernible differences could be found across studies that may suggest actual differences in effect by patient, provider, clinic, or other differences.

Prioritized Outcomes (Reduced Versus Traditional Visit Schedules)

We evaluated the strength of evidence (SoE) for prioritized outcomes (Table 4). Based on discussions with stakeholders, we prioritized 15 outcomes that were deemed both important to pregnant patients and their babies and also most likely to be affected by changes to routine antenatal visit schedules. These included: maternal quality of life, maternal anxiety, maternal depression, satisfaction with antenatal care, lost work time, preterm/full term birth, gestational age at birth, small for gestational age, low birth weight, Apgar score, breastfeeding, completion of ACOG recommended services, unplanned visits, NICU admissions, and delayed diagnoses.

To summarize the overall findings, moderate-strength evidence from studies comparing reduced versus traditional visit schedules did not indicate differences in infant outcomes between visit schedules regarding gestational age at birth, likelihood of being small for gestational age, likelihood of a low Apgar score, and likelihood of NICU admission. Low-strength evidence did not indicate differences between visit schedule groups regarding maternal anxiety, likelihood of preterm birth, and likelihood of low birth weight. Due to inconsistent findings across studies, there is insufficient evidence regarding satisfaction with antenatal care and number of unplanned antenatal visits. Several prioritized outcomes were not reported or were reported by only a single eligible study, precluding conclusions. These included maternal quality of life, maternal depression, lost work time, breastfeeding outcomes, completion of ACOG recommended services, and delayed diagnoses.

Table 4

Evidence profile for reduced versus traditional visit schedules.

Mental Health

Anxiety

Three RCTs (Ross-McGill 2000, Sikorski 1996, Walker 1997) reported on anxiety. Ross-McGill 2000 and Walker 1997 reported the State Trait Anxiety Inventory (STAI) anxiety scale. In Ross-McGill, scores were near-identical at multiple timepoints between groups. Walker 1997 also reported no change in anxiety scores over time or significant differences between groups (but no data were reported). The Sikorski 1999 RCT (reported in Clement 1999⁴⁶) found no difference in the anxiety component of the Hospital Anxiety and Depression Scale (HADS): mean 5.2 vs. 5.3, P = 0.78.

Overall, we found low SoE of no evidence of a difference between visit schedules in maternal anxiety (Table 4). Our confidence was reduced due to a lack of precision and incomplete outcome reporting.

Other Mental Health Outcomes

The Sikorski 1999 RCT (reported in Clement 1999⁴⁶) also found no difference in the HADS depression score (mean 5.7 vs. 6.1, P =0.97), overall parenting stress index (mean 220.5 vs. 219.9, P = 0.45), or in the percent of patients who self-reported a psychological or emotional problem (18.4% vs. 16.4%; OR 1.14, 95% CI 0.83 to 1.58) or who underwent treatment for a psychological or emotional problem (13.5% vs. 10.9%; OR 1.27, 95% CI 0.87 to 1.86).

The Sikorski 1999 RCT also found no difference in the overall Parenting Stress Index score between groups (mean 219.9 vs. 220.5, P = 0.45). According to the study, this is a validated psychometric scale that “gives an indication of the mother-child relationship” and for which high scores (not defined) are associated with “dysfunctional parenting behavior, negative interactions between parents and their child, and deviant child development.”

Based on the single study, there is insufficient evidence to yield a conclusion regarding the effect of different visit schedules on maternal depression (Table 4).

Patient Satisfaction With and Preference for Antenatal Care

Five RCTs (Jewell 2000, McDuffie 1996, Ross-McGill 2000, Sikorski 1996, and Walker 1997) and two NRCSs (Partridge 2005, Marko 2019) reported on patient satisfaction with care, but each defined the outcome differently (Appendix Table D-1-1).

The studies had inconsistent findings. There were no clear discernible reasons for different findings across the studies beyond the inconsistency in how satisfaction with care and preference for a given schedule were measured and the degree to which the two concepts were conflated.

The two RCTs from the United States found statistically significantly higher satisfaction with or preference for reduced schedules, at least in part. In a study of mostly Mexican-born, Hispanic, patients in southern California, Walker 1997 evaluated the Patient Satisfaction with Prenatal Care instrument but reported only P (and F) values between groups. Participants in the reduced-schedule group were significantly more satisfied with both provider (P = 0.02) and the antenatal care system (P = 0.04) than in the traditional-schedule group. Related to preference, McDuffie 1996 found that more patients in the reduced-schedule group stated that the number of visits was “just right” (89.2%) than the traditional-schedule group (82.8%; OR 1.40, 95% CI 1.04 to 1.87, reported P = 0.002). However, they reported that satisfaction—“quality of prenatal care” rated as excellent or good—was similar between groups (97.5% vs. 97.8%). This study was conducted in a population described as mostly White and well-educated.

In contrast, two of the RCTs from England found statistically significantly lower rates of satisfaction (in part) with reduced schedules. Jewell 2000 reported on satisfaction with care provided by midwives, family doctors, and hospitals. Patients in the reduced-schedule group were less likely to be “very satisfied” (60%) than those with traditional schedules (72%; OR 0.59, 95% CI 0.40 to 0.87). However, satisfaction with family doctors (42% vs. 51%) and with care provided by hospitals (42% vs. 54%) were statistically similar. Sikorski 1996 reported on dissatisfaction with visit frequency (defined as a preference for more visits in the reduced-schedule group, or vice versa); overall, throughout pregnancy, twice as many patients in the reduced-schedule group (32.5%) would have preferred the alternate schedule than those in the traditional-schedule group (16.2%; OR 2.50, 95% CI 2.00 to 3.11). In addition, Sikorski 1996 reported that partner dissatisfaction with visit frequency was considerably higher in the reduced-schedule group (39.8%) than the traditional-schedule group (5.8%; OR 10.75, 95% CI 7.23 to 15.99). Regarding a related outcome, Sikorski 1996 also found that patients in the reduced-schedule group were more likely to report “not feeling remembered from one visit to the next” (OR 1.52, 95% CI 1.25 to 1.89). These two English studies concluded that patients who received fewer visits were left feeling “not properly looked after”³⁹ or they had less time to talk at visits.⁴⁵ However, Ross-McGill 2000, a third RCT from England, reported that across the two groups 58% preferred a reduced schedule, but patients tended to prefer the group to which they were assigned: 82.5 percent in the reduced-schedule group would prefer a similar schedule in future pregnancies, while 65 percent in the traditional-schedule group would prefer the traditional schedule in future pregnancies.

The two United States-based NRCSs studies found no significant differences in satisfaction (or assessment of quality of care) between groups. In their pre-post study, Partridge 2005 reported that patient satisfaction measured by the Department of Defense patient satisfaction survey did not change with introduction of the reduced schedule at a military community hospital, but they did not report data (despite describing this as a primary conclusion of the study). Marko 2019 reported a modified Likert scale from a hospital consumer assessment of healthcare providers and systems survey instrument. Overall, across six timepoints between the first trimester and postpartum, the study reported no significant difference in patient satisfaction.

Overall, there were mixed findings across the five studies regarding satisfaction or preference for reduced- or traditional schedule visits. The studies were highly heterogeneous regarding the ethnicity, socioeconomic factors, and country of the pregnant individuals. Both studies noting dissatisfaction with reduced visit schedules were from England. Despite multiple studies evaluating satisfaction with antenatal care, there is insufficient evidence to yield a conclusion regarding the effect of different visit schedules on satisfaction (Table 4). The studies had inconsistent findings precluding an overall conclusion.

Delivery Timing, Gestational Age, and Infant Size

The prioritized outcomes related to delivery timing, gestational age, and infant size are highly correlated measures. Nevertheless, across studies, each measure (preterm birth, gestational age at birth, small for gestational age, and low birth weight) was reported as an independent outcome. Thus, for example, the outcome low birth weight does not account for the gestational age of the low birth weight babies.

Preterm Birth

One RCT (McDuffie 1996) and two NRCSs (Berglund 1998, Partridge 2005) reported on preterm birth at <37 weeks. Each study found no significant difference between groups. The RCT (McDuffie 1996) yielded an OR of 1.17 (95% CI 0.82 to 1.65) for births before 37 weeks and an imprecise OR of 1.25 (95% CI 0.49 to 3.18) for births before 32 weeks. The two NRCSs had similar findings. Berglund 1998 yielded ORs of 1.04 (95% CI 0.80 to 1.35) and 0.80 (95% CI 0.41 to 1.57) for births before 37 and 32 weeks, respectively. Partridge 2005 yielded an OR of 0.98 (95% CI 0.76 to 1.25) for births before 37 weeks.

Overall, we found low SoE of no evidence of a difference between visit schedules in preterm birth (Table 4). Our confidence was reduced due to methodological limitations of the studies and a lack of precision in findings.

Premature Rupture of Membranes (PROM) With Preterm Birth

Two RCTs (McDuffie 1997, Sikorski 1996) and one NRCS (Berglund 1998) reported PROM with preterm birth. All three studies found no significant difference, but mostly with some degree of imprecision in estimates of OR: McDuffie 1996 1.00 (95% CI 0.80 to 1.25), Sikorski 0.76 (95% CI 0.48 to 1.31), and Berglund 0.85 (95% CI 0.48 to 1.53).

Gestational Age at Birth

Two RCTs (Jewell 2000, McDuffie 1996) and two NRCSs (Partridge 2005, Ward 1999) reported on gestational age at birth. All studies found no difference between groups, with identical mean gestational ages in three studies (Jewell 2000, McDuffie 1996, and Ward 1999) and only an about 8-hour difference (0.05 weeks) in gestational age in Partridge 2005.

Overall, we found moderate SoE of no evidence of a difference between visit schedules in gestational age at birth (Table 4). Our confidence was reduced due to some methodological limitations.

Small for Gestational Age

Three RCTs (Jewell 2000, McDuffie 1996, Sikorski 1996) reported on the rates of infants who were small for gestational age (SGA). McDuffie 1996 and Sikorski 1996 defined small for gestational age as less than the 10^th percentile; Jewell 2000 did not define the outcome. Each study found no significant difference between groups.

Across studies, the summary OR for small for gestational age (<10^th percentile) is 1.08 (95% CI 0.70 to 1.66), also suggesting no difference in the rate of SGA related to antenatal visit schedule (Figure 2). There was moderate heterogeneity across studies, given the range of effect size estimates; although, individual studies had wide confidence intervals. Sikorski 1996 also reported small for gestational age less than the 3^rd percentile, which yielded an OR of 0.84 (95% CI 0.64 to 1.12).

Overall, we found moderate SoE of no evidence of a difference between visit schedules in small for gestational age (Table 4). Our confidence was reduced to moderate due to a lack of precision in findings.

Figure 2

Meta-analysis of reduced versus traditional visit schedules: Small for gestational age (RCTs). Abbreviations: CI = confidence interval, OR = odds ratio, RCTs = randomized controlled trials.

Low Birth Weight

One RCT (McDuffie 1996) and three NRCSs (Berglund 1998, Partridge 2005, Ward 1999) reported on low birth weight. Across studies (but not consistently), low birth weight was categorized as <2500 g, <1500 g, and <1000 g.

Across the three NRCSs, which all reported birth weight <2500 g, differences were nonsignificant and the summary OR was 1.02 (95% CI 0.82, 1.25) (Figure 3). The studies were statistically homogeneous. The RCT (McDuffie 1996) had a similar finding: OR 0.89 (0.64, 1.23). Including the RCT in the meta-analysis as a sensitivity analysis did not substantively change the findings.

For birth weight <1500 g, the two NRCSs (Berglund 1998 and Partridge 2005) and the RCT (McDuffie 1996) also found no significant differences between groups, but with less precision and a wider range of OR estimates: McDuffie 1996 1.17 (95% CI 0.39 to 3.49), Berglund 1998 0.59 (95% CI 0.28 to 1.23), and Partridge 2005 2.05 (95% CI 0.55 to 7.59).

For birthweight <1000 g, two NRCSs (Partridge 2005 and Ward 1999) found highly imprecise estimates of OR: Partridge 2005 0.64 (95% MCI 0.11 to 3.85) and Ward 1999 1.03 (95% CI 0.21 to 5.13).

Overall, we found low SoE of no evidence of a difference between visit schedules in low birth weight (Table 4). Our confidence was reduced due to methodological limitations of the studies.

Figure 3

Meta-analysis of reduced versus traditional visit schedules: Low birth weight <2500 g (NRCSs). Abbreviations: CI = confidence interval, NRCSs = nonrandomized comparative studies, OR = odds ratio.

Apgar Score

Three RCTs (Jewell 2000, McDuffie 1996, Sikorski 1996) and two NRCSs (Berglund 1998, Ward 1999) reported on Apgar score. Similar to the above gestational age and infant size measures, the analyses of Apgar score did not account for gestational age.

Three of the studies (McDuffie 1996, Berglund 1998, Ward 1999) evaluated Apgar scores <7 at 5 minutes. Each study found no significant difference between the two groups, with OR estimates of 0.62 (95% CI 0.34 to 1.11), 0.62 (95% CI 0.36 to 1.07), and 1.26 (95% CI 0.73 to 2.18), respectively.

The unadjusted NRCS Berglund 1998 found a near-significant difference, in the direction of favoring reduced schedule for Apgar score <5 at 5 minutes (OR 0.45, 95% CI 0.20 to 1.03). The other unadjusted NRCS, Ward 1999, found no significant difference for Apgar score <3 at 1 minute (OR 1.24, 95% CI 0.73 to 2.11).

Jewell 2000 and Sikorski 1996 reported Apgar scores as continuous outcomes. All mean Apgar scores were identical at 1 minute (Sikorski 1996) and 5 minutes (Sikorski 1996 and Jewell 2000).

Overall, we found moderate SoE of no evidence of a difference between visit schedules in Apgar scores (Table 4). Our confidence was reduced due to some methodological limitations.

Breastfeeding

The Sikorski 1996 RCT was the only study to report on breastfeeding outcomes. The reduced-schedule group breastfed for a nonsignificantly longer duration (29.7 weeks) than the traditional-schedule group (27.0 weeks), yielding an MD of 2.7 (95% CI −0.9, 6.3) weeks.

Based on the single study, there is insufficient evidence to yield a conclusion regarding the effect of different visit schedules on breastfeeding (Table 4).

Unplanned Visits

One RCT (Walker 1997) and two NRCSs (Berglund 1998 and Partridge 2005) reported variations of unplanned visits. Each reported a different, self-defined measure.

The RCT implied that the difference in the mean number of unscheduled “drop-in” office visits per patient between groups was not statistically significant, but there were more such visits in the reduced-schedule group: 0.23 (SD 0.57) versus 0.08 (SD 0.27) (calculated MD 0.15, 95% CI −0.04 to 0.34, P = 0.11).

The Partridge 2005 NRCS reported “outpatient visits to Labor & Delivery Unit per delivery,” but did not define this outcome further. They reported significantly more outpatient visits in the reduced-schedule group (2.9) than the traditional-schedule group (2.5), yielding a calculated difference of 0.4 (95% CI 0.1 to 0.7) visits per delivery.

The Berglund 1998 NRCS (a pre-post study) reported patient-initiated extra midwife visits, staff-initiated extra midwife visits, patient-initiated extra doctor visits, and staff-initiated extra doctor visits. Both patient- and staff-initiated extra visits to midwives were increased after institution of the reduced-schedule protocol (by 0.62 staff-initiated visits and 0.01 patient-initiated visits), but both patient- and staff-initiated doctor visits were less (by 0.15 staff-initiated visits and 0.15 patient-initiated visits). The study did not report statistical significance, but our calculations of differences between group suggested the differences were statistically significant (in this large study with 3882 participants), if not clinically significant. However, the differences in the number of staff- and patient-initiated visits found may have been driven, at least in part, by differences between groups (or between time periods) other than the number of scheduled visits.

Despite multiple studies evaluating unplanned visits, there is insufficient evidence to yield a conclusion regarding the effect of different visit schedules on the outcome (Table 4). The studies had inconsistent findings precluding an overall conclusion.

Neonatal Intensive Care Unit Admission

Three RCTs (Jewell 2000, Sikorski 1996, Walker 1997) and two NRCSs (Berglund 1998, Partridge 2005) reported on admission to NICU. The three RCTs all found no significant difference in the numbers of infants with NICU admissions. Across RCTs, the summary OR was 1.05 (95% CI 0.74 to 1.50) (Figure 4). The studies were statistically homogeneous.

The two NRCSs had similar findings. Partridge 2005 had similar rates of NICU admissions between groups (OR 0.97, 95% CI 0.77 to 1.24). The Berglund 1998 pre-post study reported no significant difference between groups (OR 0.75, 95% CI 0.31 to 1.80).

The Sikorski 1996 RCT also reported mean length of stay in the NICU, which was not significantly different between groups: MD 5.4 days (95% CI −5.9 to 16.7).

Overall, we found moderate SoE of no evidence of a difference between visit schedules in NICU admissions (Table 4). Our confidence was reduced due to imprecision in study findings.

Figure 4

Meta-analysis of reduced versus traditional visit schedules: NICU stay (RCTs). Abbreviations: CI = confidence interval, NICU = neonatal intensive care unit, OR = odds ratio, RCTs = randomized controlled trials.

Prioritized Outcomes Not Reported by Eligible Studies

Several outcomes of interest to stakeholders were not evaluated by eligible studies. Among the prioritized outcomes, these include:

• Maternal psychosocial, preference, and related outcomes

  • Quality of life
  • Lost work time
• Fetal, neonatal, and infant outcomes

  • Breastfeeding
• Care utilization

  • Completion of ACOG recommended services

Other Outcomes (Reduced Versus Traditional Visit Schedules)

Gestational Hypertension

Two RCTs (Jewell 2000 and Walker 1997) and 1 NRCS (Berglund 1998) reported on GHTN.

The two RCTS yielded imprecise (for Walker 1997, highly imprecise) estimates of effect sizes: Jewell 2000 OR 1.31 (95% CI 0.74 to 2.31), Walker 1997 OR 1.80 (95% CI 0.16 to 20.7). In contrast, the Berglund 1998 pre-post study found a substantial and significantly lower risk of diagnosed uncomplicated GHTN (excluding pre-eclampsia) with a reduced schedule (7.4% vs. 11.9%, unadjusted P <0.001; OR 0.59, 95% CI 0.48 to 0.74). Including pre-eclampsia yields similar findings (10.0% vs. 14.3%, OR 0.70, 95% CI 0.59 to 0.83). The study authors explain the difference in uncomplicated GHTN as being due to “the smaller number of observations during late pregnancy, when the majority of cases occur” (i.e., due to a lower rate of diagnosis in the reduced-schedule group, not necessarily due to a lower true risk of GHTN).

Pre-Eclampsia

Two RCTs (McDuffie 1996, Sikorski 1996) and two NRCSs (Berglund 1998, Ward 1999) reported on pre-eclampsia. All studies found no significant differences between the two groups. Across studies, the ORs ranged from 0.67 (95% CI 0.43 to 1.05) to 1.10 (95% CI 0.73 to 1.64).

Intrauterine Growth Restriction (IUGR)

One RCT (Walker 1997) and one NRCS (Berglund 1998) reported on IUGR. Both studies found no significant difference between groups. The RCT was highly imprecise, with only a single pregnancy affected by IUGR; the NRCS had an OR of 0.79 (95% CI 0.59 to 1.06).

Anemia

One RCT (Walker 1997) and one NRCS (Berglund 1998) reported on anemia. The RCT did not define anemia. Berglund 1998 defined anemia as hemoglobin <10 g/dL. The RCT was highly imprecise, with only a single episode of anemia in each study group; the NRCS had an OR of 0.90 (95% CI 0.65 to 1.26).

Preterm Labor

Two RCTs (McDuffie 1996 and Walker 1997) reported on preterm labor. The RCT was highly imprecise, with few episodes of preterm labor (OR 0.28, 95% CI 0.03 to 2.79); the NRCS risk ratio (RR) of 1.01 (95% CI 0.86 to 1.18).

Hemorrhage

Two RCTs (McDuffie 1996, Sikorski 1996) and one NRCS (Berglund 1998) reported on peripartum hemorrhage. Each study found no significant difference between groups. Across studies, the OR for hemorrhage ranged from 0.86 to 1.01.

Sikorski 1996 also reported antenatal hemorrhage, the rate of which was also similar between groups (OR 0.97, 95% CI 0.69 to 1.35).

Cesarean Delivery

Three RCTs (McDuffie 1996, Sikorski 1998, Walker 1997) reported on cesarean delivery. No NRCS adjusted for differences in patient or obstetrician preferences regarding cesarean delivery. However, we summarize one NRCS (Berglund 1998) that reported on emergency cesarean deliveries (which are unlikely to be affected by patient or provider preferences).

The three RCTs found no evidence of a difference in cesarean delivery rate between groups; however, Walker 1997 provided a highly imprecise estimate due to small numbers of cesarean deliveries. Across the three studies, the ORs ranged from 0.12 to 1.09. Meta-analysis yielded a summary OR of 0.94 (95% CI 0.72 to 1.24), with moderate heterogeneity related to the imprecise Walker 1997 study (Figure 5).

The McDuffie 1996 RCT also reported on cesarean delivery for fetal distress, with similar event rates between groups (OR 0.88, 95% CI 0.50 to 1.55). In addition, the Sikorski 1996 RCT yielded an OR of 0.81 (95% CI 1.78 to 0.36) for cesarean delivery for pregnancy-related hypertension. One NRCS (Berglund 1998) also reported on emergency cesarean, also finding similar event rates between groups (OR 0.92, 95% CI 0.67 to 1.28).

Figure 5

Meta-analysis of reduced versus traditional visit schedules: Cesarean delivery (RCTs). Abbreviations: CI = confidence interval, OR = odds ratio, RCTs = randomized controlled trials.

Induction of Labor

Two RCTs (Jewell 2000, Sikorski 1996) and two NRCSs (Berglund 1998, Partridge 2005) reported on induction of labor. Each study found no significant difference between the two groups. Across studies, the OR for induction of labor is close to null and ranged from 0.92 and 1.18. The Sikorski 1996 RCT also reported on induction of labor due to pregnancy-related hypertension, again finding no significant difference between groups (OR 0.91, 95% CI 0.57 to 1.47).

Operative Vaginal Delivery

Two RCTs (Sikorski 1996, Walker 1997) and one NRCS (Partridge 2005) reported on operative vaginal delivery, defined as either forceps or vacuum delivery (or instrumental or assisted delivery). The two RCTs found no significant difference between groups. Sikorski 1996 had similar event rates (OR 1.09, 95% CI 0.85 to 1.39), while Walker 1997 was highly imprecise (OR 0.43, 95% CI 0.04 to 4.92) related to small number of events. In contrast, the Partridge 2005 pre-post study found that significantly fewer operative vaginal deliveries occurred in the reduced-schedule group (2.9% vs. 6.6%; OR 0.42, 95% CI 0.30 to 0.59). The authors hypothesized that this may have been related to somewhat higher rates of small-for-gestational age babies and a concomitant decrease in large-for-gestational age babies after implementation of a reduced routine visit schedule, together with (otherwise unreported) use of continuous epidural analgesia.

Spontaneous Vaginal Delivery

Two RCTs (Jewell 2000 and Sikorski 1996) reported on spontaneous vaginal delivery. Both found no significant difference between groups: Jewell 2000 OR 1.16 (95% CI 0.79 to 1.71), Sikorski 1996 OR 1.04 (95% CI 0.64 to 4.98).

Infant/Fetal Mortality

None of the eligible studies explicitly defined mortality outcomes to distinguish among different perinatal deaths. Neonatal and fetal mortality are sufficiently rare that studies are generally underpowered to evaluate the outcomes.

Stillbirth

Two RCTs (McDuffie 1996, Sikorski 1996) and one NRCS (Ward 1999) reported on rates of stillbirth (or intrauterine fetal death). Stillbirth was uncommon in all studies; thus, effect estimates were highly imprecise: McDuffie 1996 RCT: OR 1.00 (95% CI 0.29 to 3.47; Sikorski 1996 RCT: OR 0.85 (95% CI 0.26 to 2.81); Ward 1999 NRCS: OR 2.07 (95% CI 0.52 to 8.32) for intrauterine fetal death.

Neonatal Mortality

Two NRCSs (Partridge 2005 and Ward 1999) reported on neonatal mortality. Both found no significant difference between groups, but the effect estimate from Ward 1999 was highly imprecise (OR 0.74, 95% CI 0.23, 2.33) and from Partridge also had a wide confidence interval (OR 0.62, 95% CI 0.27 to 1.43).

Perinatal Death

One RCT (Sikorski 1996) and two NRCSs (Berglund 1998 and Ward 1999) reported on perinatal death. All found no significant difference in event rates between groups; however, findings were imprecise. The RCT yielded an OR of 0.72 (95% CI 0.27 to 1.89). The Berglund 1998 NRCS yielded a more imprecise estimate (OR 0.86, 95% CI 0.34 to 2.18). Ward 1999 reported incidence rates of 3 per 1000 births in the reduced-schedule group versus 4 per 1000 births in the traditional-schedule group, which was statistically nonsignificant.

Perinatal Morbidity

One RCT (Walker 1997) and one NRCS (Ward 1999) reported on four different types of perinatal mortality. Due to small sample size and low event rates, the RCT provided highly imprecise estimates of effect sizes between groups for sepsis (OR 1.79, 95% CI 0.06 to 54.8), hyperbilirubinemia (OR 0.88, 95% CI 0.05 to 14.6), and respiratory distress (OR 5.63, 95% CI 0.27 to 116). The NRCS found a nonsignificantly lower rate of bronchopulmonary dysplasia in the less intensive group (OR 0.31, 95% CI 0.08 to 1.12).

Unplanned Hospital Admissions (Maternal)

Three RCTs (McDuffie 1996, Ross-McGill 2000, Sikorski 1996) and one NRCS (Berglund 1998) reported on unplanned hospital admission for the pregnant patient. Two RCTs (McDuffie 1996, Sikorski 1996) reported no differences in the numbers of patients who had unplanned inpatient antenatal admissions (OR 1.15, 95% CI 0.77 to 1.72), outpatient antenatal admissions (OR 1.13, 95% CI 0.95 to 1.34), or admission to a special care unit (OR 1.07, 95% CI 0.70 to 1.62). One RCT (Ross-McGill 2000) and the NRCS (Berglund 1998) found no significant differences in the number of hospital visits. Berglund 1998 found a less than 0.1 difference between groups in the mean numbers of total hospital visits, emergency hospital visits, and referral-based hospital visits. Ross-McGill 2000 reported only that the mean numbers of pregnancy-related hospital visits between 28 and 41 weeks were 1.2 for the reduced-schedule group and 1.4 for the traditional-schedule group.

Emergency Department Visits (Maternal)

Two RCTs (McDuffie 1996 and Walker 1997) reported on emergency department visits by the pregnant patient. McDuffie 1996 reported similar percentages of patients who made emergency department visits regardless of visit schedule (OR 1.08, 95% CI 0.89 to 1.32). Walker 1997 reported no significant difference in the mean number of emergency department visits (MD 0.06, 95% CI −0.04 to 0.16).

Other Outcomes Reported by a Single Study Each

The Sikorski 1996 RCT reported on maternal death. However, only one patient in the study died, resulting in a highly imprecise estimate of differences between groups (OR 3.08, 95% CI 0.13 to 75.7). In the same trial, among both pregnancies with malpresentation and with placenta previa, similar percentages were undiagnosed at the time of labor between the reduced-schedule and traditional-schedule groups: undiagnosed malpresentation (25.0% vs. 29.1%), undiagnosed placenta previa (33.3% vs. 32.7%).

The McDuffie 1996 RCT reported that in both groups, 1.5 percent of patients were diagnosed with diabetes, there was no significant difference between groups in the risk of placental abruption (OR 1.55, 95% CI 0.72 to 3.33), and there was a highly imprecise estimate of differences between groups of chorioamnionitis (OR 0.82, 95% CI 0.34 to 1.98).

The Partridge 2005 pre-post study reported that post-term delivery occurred less frequently after implementation of the reduced visit schedule than the prior traditional schedule (8.1% vs. 10.4%; OR 0.76, 95% CI 0.60 to 0.97). However, the study implied that this outcome may have occurred due to other changes in the clinic practice related to “management of postdates pregnancy.” The study also reported statistically significantly fewer babies large for gestational age, defined as >4000 g after implementation of the reduced schedule than the prior use of a traditional schedule (10.6% vs. 13.0%; OR 0.79, 95% CI 0.64 to 0.98). Larger infants (>4500 g) were not significantly different between groups (1.3% vs. 1.6%; OR 0.81, 95% CI 0.46 to 1.44). However, the study implied that these outcomes may have occurred due to other changes in the clinic practice related to “management of postdates pregnancy.”

Only the Walker 1997 RCT provided a clearly defined measure of adherence with assigned visit schedule. The mean number of missed (“no show”) visits were similar between groups (0.26 vs. 0.28).

The only study to report on harms, was the Sikorski 1996 RCT, which reported total and “correctly” diagnosed birthweight; although, the concept of correct diagnosis was not defined. From these measures, we calculated “incorrectly” diagnosed birthweight, which we interpreted as overdiagnoses of small for gestational age during pregnancy. About 7 percent of fetuses/infants were diagnosed as <3^rd percentile weight, only about one-third of which were categorized as “correct”; thus, about 5 percent were incorrectly diagnosed. Similarly, about 21 percent of fetuses/infants were diagnosed as <10^th percentile weight, only about one-quarter of which were categorized as “correct”; thus, 16 percent were incorrectly diagnosed. However, rates of incorrect diagnoses were similar in both groups. Overall, the OR for incorrectly diagnosed <3^rd percentile birthweight was 0.93 (95% CI 0.67 to 1.30) and for incorrectly diagnosed <10^th percentile birthweight was 1.00 (95% CI 0.82 to 1.23).

Reduced-Schedule Group Versus Traditional-Schedule Individual Visits

A single study, Tandon 2013, compared group with individual visits with a difference in the number of total scheduled routine antenatal care visits.⁵⁰ It was a retrospective parallel study comparing patients who self-selected a lower visit-frequency CenteringPregnancy program of 10 structured 2-hour group visits led by a nurse practitioner and health educator or the traditional 12 individual visits with nurse practitioners. Notably, the study did not adjust for differences between study groups.

The study was conducted at a Florida clinic that serves primarily Spanish-speaking (93%) self-identified Hispanics (76%) or Mayans (21%) mostly from Mexico and Central America (90%); 41 percent of the patients were single and 30 percent were primiparous (Table 2).

The only reported outcome pertinent to this review was satisfaction with antenatal care. The study used the Patient Participation and Satisfaction Questionnaire (PPSQ) with satisfaction with care and satisfaction with participation subscales. For both subscales, participants were statistically significantly more satisfied in the reduced-schedule group visits than with the traditional-schedule individual visits, with about 10- to 20-point differences (on an implied 100-point scale). Estimates of minimal important differences were not reported.

Similarly, participants receiving reduced-schedule group visits were substantially more likely to be very satisfied than those with traditional-schedule individual visits. Among patients in the reduced-schedule group visit study arm, 98 percent were very satisfied with time spent with the provider while only 19 percent were very satisfied in the traditional-schedule individual visit study arm (reported relative risk 5.14, 95% CI 2.85 to 9.25).

Qualitative Evidence Pertaining to Number of Visits

Four studies (Meiser 2021, Sanders 1999, Sanders 2000, and Sikorski 1995) provided qualitative data on antenatal care visit schedules,^51–54 three from practices in England (Bristol metropolitan area and London) and one from the United States (Lubbock, Texas). Two related English studies provided different perspectives—midwives (Sanders 1999) and pregnant individuals (Sanders 2000)—on their experiences with “flexible” versus “traditional” care that were evaluated in the Bristol Antenatal Care Study (Jewell 2000, included in KQ 1).³⁹ The London-based study (Sikorski 1995) sought to ascertain the view of midwives, obstetricians, and general practitioners on “possible changes in the provision of antenatal care” (in 1993) that were implemented in a later RCT, included above (Sikorski 1996).⁴⁵ The Texas-based study (Meiser 2021) sought to “better understand the patient perspective on alternative antenatal care schedules.”

Two studies collected data via questionnaires that invited open-ended responses exclusively (Sikorski 1995, Sanders 2000). One study each collected data via questionnaires and subsequent follow-up interviews (Meiser 2021) and focus group (Sanders 1999). Sample sizes ranged from 14 to 438, although the depth of qualitative analysis and data from the larger sample studies (Sikorski 1995 and Meisner 2021) was limited. All studies conducted thematic content analysis to identify themes within the qualitative data relevant to their study question.

We assessed studies to have variable risks to rigor. Sanders 2000 was assessed to have high risks to rigor (related to a lack of statement of aims and incomplete reporting of methodology) and Sikorski 1995 moderate risk to rigor (related to a lack of consideration of the relationship between researcher and participants or ethical issues). Meiser 2021 and Sanders 1999 were assessed to have low risks to rigor. Appendix C Table C-4-3 summarizes the risk to rigor assessment of the studies. Most findings had no or minor concerns regarding potential risks to rigor (Appendix D Table D-3-2 presents the GRADE-CERQual [Grading of Recommendations Assessment, Development and Evaluation - Confidence in Evidence from Reviews of Qualitative research] table). The exception related to the findings regarding knowledge and beliefs about capabilities where we had moderate concerns because the pertinent studies had higher risks to rigor. We had no concerns or minor concerns with the coherence of findings (i.e., the synthesized findings reflected the complexity and variation of the data) but minor to moderate concerns with the adequacy of several findings due to the limited number of studies, perspectives sought (often midwives and no other providers, and often providers sharing perceived perspectives of patients rather than patients themselves), and geographical/health setting representation.

We coded data provided by the four studies into seven of the Theoretical Domains Framework (TDF) domains including: knowledge; social/professional role and identity; beliefs about capabilities; beliefs about consequences; environmental context and resources; social influences; and emotion. We did not find evidence from extracted qualitative data for the remaining seven TDF domains of skills; optimism; reinforcement; intentions; goals; memory, attention, and decision processes; and behavioral regulation. Table 5 summarizes the barrier and facilitator themes related to adopting a reduced antenatal care schedule. Themes are presented within the seven TDF domains identified, organized by whether the themes related to patients, providers, or the clinic/office. Providers discussed their own, but also perceived patient perspectives. Notably, across studies, most patient-related themes were shared by providers, and few were based directly on patient-reported perspectives. No study provided either clinic leadership or partner perspectives.

Appendix D Table D-3-1 provides the extracted text from studies linked to their synthesized themes. Appendix D Table D-3-2 is the GRADE-CERQual table summarizing the findings for each TDF theme. Only one finding (related to the TDF domain social/professional role and identity) was supported by three studies; four findings (related to the TDF domains of knowledge; beliefs about capabilities, beliefs about consequences, and emotion) were supported by two studies, and two findings (related to the TDF domains of environmental context and resources; and social influences) were supported by one study.

TDF Domain: Emotion

Two studies (Meiser 2021, Sikorski 1995) provided evidence pertaining to emotion from either the patient or provider perspectives. The studies reported that patients’ distress with reduced care (and inversely emotional appreciation for appointments) and providers’ fears of staff cutbacks may pose psychological barriers to reduced visit schedules.

Table 5

Coding extracts for antenatal care visit schedules into the Theoretical Domains Framework.

Key Points

• Low-strength evidence comparing hybrid (televisit and in-person) visits and all in-person visits

  • did not indicate differences in rates of preterm births (1 RCT and 3 NRCSs; summary OR 0.93, 95% CI 0.84 to 1.03, P=0.18), and
  • did not indicate differences in rates of NICU admissions (3 NRCSs; summary OR 1.02, 95% CI 0.82 to 1.28).
• There is insufficient evidence for numerous prioritized outcomes of interest (e.g., completion of ACOG recommended services and patient experience measures).
• Studies did not evaluate heterogeneity of treatment effect (whether some specific groups of patients may have had better, or worse, outcomes with televisits) or equity issues (whether classes of patients might be disadvantaged by use of televisits).
• Qualitative studies suggest that

  • Providers’ perceive televisits may improve patients’ access to care, continuity of care, and attendance at visits and may allow individuals with low-risk pregnancies to be safely managed at home. While some providers believe that televisits may reduce training opportunities for junior clinicians, others believe televisits increase team cohesion and case discussions between junior and senior clinicians. Some providers believe there is limited evidence on the value of televisits for routine antenatal care or that existing evidence is not universally recognized.
  • Patients and providers believe televisits may improve communication and relationship-building (though some believe it may hinder it), allow tailoring to the needs of patients, and (during the COVID-19 pandemic) protect patients from COVID-19 transmission.
  • Providers believe facilitators to implementation of televisits include providers’ ease with technology and access to colleagues with experience with telemedicine, as well as guidelines on which types of visits are appropriate for televisits.
  • Some providers doubted patients’ ability to describe their symptoms virtually and expressed discomfort in their own ability to deliver care with complex televisit technologies. They also expressed concerns about potential liability issues or changes to reimbursement policies associated with the delivery of televisits.
  • Training for staff (e.g., online modules) and regular leadership meetings to learn and refine processes were noted by providers and leadership as important facilitators, respectively. The need for “onerous” training however was perceived by some providers to be a barrier to the implementation of televisits for routine antenatal care. Providers also believe that leadership support during the transition can help facilitate the implementation of televisits (and conversely, lack of buy-in from leadership, can be a barrier).
  • Patients and providers noted system supports (e.g., guidance, technology support, translation services) and home monitoring are important adjuncts to televisits. Providers noted the importance of user-friendly technology to support engagement by both providers and patients.
  • Patients and providers expressed concerns about safety and reduced quality of care with televisits.
  • Providers expressed concerns of possible worsening of health disparities related to difficulty accessing Web-based visits and internet technologies in general and noted the need for audio and visual translation to minimize inequities particularly among non-English speaking patients.
  • Providers and leadership noted that the implementation of televisits added clinic operational challenges.
  • Providers noted concerns that patients’ psychological apprehension with televisits and difficulty to initially set up televisits may pose a barrier to its uptake. Some patients noted a general desire to be able to choose to be seen in-person.
  • While some providers reported patients’ lack of privacy as a barrier others noted the attendance and participation of patient’s families during televisits as a facilitator.

Summary of Findings

Overall, seven studies (in 9 publications) compared televisits (or hybrid visits) with in-person visits for routine antenatal care.^55–63 These included two RCTs,^{55, 56} four NRCSs,^57–62 and one survey.⁶³ The studies mostly compared groups of patients who either had a hybrid schedule, including both televisits and in-person visits, with those who had in-person visits only. In contrast, the Futterman 2021 survey evaluated a single cohort of patients who had both televisits and in-person visits and asked them to compare the two visit types.⁶³ We, thus, present results from the survey separately below. Across studies, the number of televisits varied from one to six (Table 6). Three of the studies (Duryea 2021, Futterman 2021, and Palmer 2021) were conducted during the COVID-19 pandemic.^{57, 58, 63}

Three studies were funded by nonindustry sources (Butler Tobah 2019, Duryea 2021, and Pflugeisen 2017), one was unfunded (Palmer 2021), and the other three did not report information on funding source. Pflugeisen 2016 and Pflugeisen 2017 both evaluated a “Virtual Visit” program, but the patients did not appear to overlap. Pflugeisen 2016 enrolled patients from 2011 to 2013. Pflugeisen 2017 contacted patients in 2016.

Tables in Appendixes C and D provide the design details, participant characteristics, detailed schedule data, and results.

Hybrid Visits Versus In-Person Visits

Six studies compared hybrid antenatal visit schedules (both televisits and in-person visits) with in-person antenatal visits only. Two were RCTs (Barbour 2017, Butler Tobah 2019)^{55, 56} and four were NRCSs (Duryea 2021, Palmer 2021, Pflugeisen 2016, Pflugeisen 2017).^57–60 Except for Palmer 2021 (conducted in Australia), the studies were conducted in the United States, in Salt Lake City, Minnesota, Dallas, and around Seattle. The two RCTs included 200 and 267 patients each. Two NRCSs that compared pre- and post-pandemic periods (Duryea 2021 and Palmer 2021) were relatively large (N = 12,607 and 17,237), but neither compared hybrid visits during COVID-19 with in-person visits pre-pandemic or accounted for factors other than visit type that may have led to different outcomes during the pandemic. The remaining two NRCSs included 171 and 1058 participants. Palmer 2021 reported on two groups of pregnant individuals: “low” and “high” risk. These categories were not defined but high-risk individuals comprised 23 percent of the overall sample and had their care provided by obstetricians; they had more overall and more in-person visits. Low-risk individuals were seen by midwives. Consistent with our eligibility criteria, we summarize the findings only for the low-risk cohort. Findings from the high-risk cohort were similar overall to those for the low-risk cohort and can be found in the primary article.⁵⁸

Only four of the six studies reported the number of visits and timing of the televisits (Table 6), but all at least implied that ACOG (or equivalent) guidance was followed (approximately 10 to 14 scheduled visits depending on gestational age at presentation). One study (Palmer 2021) scheduled six televisits, two studies (Butler Tobah 2019 and Pflugeisen 2016) scheduled five antenatal televisits (and one additional postpartum televisit). One study (Duryea 2021 scheduled three televisits. The Barbour 2017 study (which was reported only as a conference abstract) stated only that five visits were in-clinic, but did not report the number of televisits. One study (Pflugeisen 2017) did not report visit timing and included participants who had at least one televisit. (The survey comparing experiences with both televisits and in-person visits, addressed in the next section, also did not report visit timing and included participants who had at least one televisit.) In the three studies (in two settings) that reported information, televisits were all conducted by nurse practitioners; in one of these (Butler Tobah 2019), the providers of in-person visits were not reported and in the other two (related) studies (Pflugeisen 2016, Pflugeisen 2017), in-person visits were conducted by obstetricians.

Among the five studies that reported data, the average ages of participants ranged from 27.7 to 31.4 years. Most participants included in three studies (Butler Tobah 2019, Pflugeisen 2016, Pflugeisen 2017) were White (72%-92%), one study (Duryea 2021) had a majority (76%) Hispanic patients, and two studies (Barbour 2017 and Palmer 2021) did not report information on race/ethnicity (Table 7). Among two studies that reported information on participants’ body mass index (BMI), Duryea 2021 reported an average BMI of 29 at first encounter, Butler Tobah 2019 reported an average BMI of 26 with 15 percent of the participants having a BMI over 30 at enrollment (implied), and Palmer 2021 reported a median BMI of 25 (interquartile range 22 to 29). Three studies (Butler Tobah 2019, Pflugeisen 2016, Pflugeisen 2017) reported on participants’ relationship status; the majority of participants were married, partnered, or co-habiting (71-97%). The studies varied in the percentage of patients who were nulliparous, ranging from 30 to 67 percent.

Limited information on participants’ baseline morbidities or pregnancy risk factors was reported (Table 7). Barbour 2017 included parous patients with low-risk pregnancies at <16 weeks’ gestation with a singleton. Butler Tobah 2019 included participants with low-risk pregnancies, defined as no concurrent medical or obstetric complications (e.g., hypertension, diabetes, preterm delivery history). Duryea 2021 included individuals who had a delivery of an infant weighing more than 500 g (live or stillbirth) but included participants with chronic hypertension (4.5%) and pregestational diabetes (1.9%). Notably, if the number of very-low weight infants differed between groups, this study would provide biased results related to other birth and infant outcomes. In Pflugeisen 2016, 30 percent of participants were tobacco users, 7 percent were illicit drug users, and 22 percent had a prior pregnancy loss. Pflugeisen 2017 included patients with low-risk pregnancies who were assessed and stratified as low risk by their physician (no specific criteria were reported). Duryea 2021 was conducted among patients with medical indigence (per source population) while all other studies included general population pregnant patients. In Palmer 2021, 6 percent of patients smoked during pregnancy (in high- and low-risk groups). The studies were conducted between 2004 and 2020.

The studies varied in the technologies used to enable televisits (Table 8), although most studies did not fully describe the technologies. Four studies (Barbour 2017, Palmer 2021, Pflugeisen 2016, Pflugeisen 2017) used Web-based videoconferencing, one study used only telephone visits (Duryea 2021), and one study used both Web-based and phone visits (Butler Tobah 2019). The study using only telephone visits (Duryea 2021) was conducted in a hospital system that serves a medically indigent population “without ready access to the resources and technologic literacy necessary to engage in video visits”.⁵⁷

Assessment of the RoB of the included studies are presented in Tables C-4-1 (RCTs) and C-4-2 (NRCSs). The two RCTs (Barbour 2017 and Butler Tobah 2019) did not report any issues that raise concerns of high risk of bias, except that neither study could blind patients or providers to the intervention. The Butler Tobah 2019 RCT was deemed to be at low risk of bias, but the Barbour 2017 was downgraded to medium risk of bias because reporting information was missing in this conference abstract. The four NRCSs that compared groups of patients who received hybrid or all in-person visits (Duryea 2021, Pflugeisen 2016, Pflugeisen 2017) were all deemed to be high risk of bias. Three studies did not adjust their findings for potential confounders or fundamental differences between groups of patients. Pflugeisen 2016 did adjust for baseline differences between groups with logistic regression; in addition, the providers differed between groups, confounding interpretation of the televisit intervention. As will be described, few studies evaluated any given outcome. Given the small number of studies per outcome, we could not formally assess risk of possible publication bias.

None of the studies evaluated or compared subgroups of patients or formally addressed heterogeneity of treatment effect (whether specific individuals or identifiable groups of patients may have better, or worse, outcomes than most patients with the implementation of televisits). Similarly, no eligible studies evaluated equity issues (whether classes of patients, such as non-English speakers or those without access to stable internet communication, might have been further disadvantaged or have had worse outcomes with the implementation of televisits).

Table 6

Timing of routine antenatal in-person and televisits (weeks).

Table 7

Study design and sample characteristics of studies addressing televisits.

Table 8

Description of televisits.

Study results are organized by outcome. We first describe outcomes related to prioritized outcomes for which there are data (pertaining to maternal psychosocial, preference, and related outcomes; fetal, neonatal, and infant outcomes; care utilization; and harms). These are followed by other included outcomes; other included outcomes that were reported by only a single study are grouped together at the end of the list of outcomes.

Across studies, for all outcomes, there were no reported analyses of subgroups, subgroup differences, or heterogeneity of treatment effect. Except as noted, study findings were consistent with each other (generally finding no significant differences) such that no discernable differences could be found across studies that may suggest actual differences in effect by patient, provider, clinic, or other differences.

Prioritized Outcomes (Hybrid Visits Versus In-Person Visits)

We evaluated the strength of evidence for prioritized outcomes (Table 9).

In general, outcomes were sparsely reported across studies. No outcome was evaluated by both RCTs. Low-strength evidence from studies comparing hybrid (televisits and in-person visits) and all in-person visits did not indicate differences in rates of preterm births or NICU admission of the newborn. However, there is low-strength evidence that satisfaction with antenatal care was greater in patients receiving hybrid visits. Several prioritized outcomes were not reported or were reported by only a single eligible study, precluding conclusions. These included maternal stress, lost work time, gestational age at birth, low birthweight, Apgar score (and abnormal cord blood pH), and completion of ACOG recommended services.

Table 9

Evidence profile for hybrid (televisits and in-person) versus in-person visits.

Mental Health

Pregnancy-Related Stress

One RCT (Butler Tobah 2019) reported on pregnancy-related stress as an outcome using the PreNatal Maternal Stress survey at 14, 24, and 36 weeks of gestation. Patients having hybrid visits reported having statistically significant less pregnancy-related stress than patients having in-person visits at 14 weeks (on a 2-point scale, MD −0.09, 95% CI −0.14 to −0.04) and 36 weeks (MD −0.06, 95% CI −0.11 to −0.01). Stress levels were not found to be different at 24 weeks. There was no indication of clinical importance of the differences.

Based on the single study, there is insufficient evidence to yield a conclusion regarding the effect of televisits on maternal stress (Table 9).

Patient Satisfaction With Antenatal Care

One RCT (Butler Tobah 2019) and one NRCS (Pflugeisen 2017) reported on patient satisfaction with antenatal care. The Butler Tobah 2019 RCT measured patient satisfaction by using the Modified Littlefield and Adams 16 item self-reported validated satisfaction subscale survey. Patients having hybrid visits reported a significantly higher level of satisfaction than those receiving in-person visits (on a 100-point scale, MD 15.0, 95% CI 13.4 to 16.6). The unadjusted Pflugeisen 2017 NRCS reported a higher score on overall rating of antenatal care among patients in the hybrid group compared with the all in-person group (on a 5-point Likert scale, MD 0.25, 95% CI 0.06 to 0.44) but no difference in scores for satisfaction with televisits versus in-person visits (MD 0.11, 95% CI −0.14 to 3.35). There was no indication of clinical importance of the differences.

The two studies found higher satisfaction with use of hybrid visits (although one study found no difference in satisfaction with the televisits versus the in-person visits). However, given small number of studies, the methodological limitations (particularly the lack of adjustment in the NRCS), and the lack of validation of the satisfaction measure in the NRCS, we deemed there to be low SoE suggesting increased patient satisfaction with use of televisits.

Lost Time

One RCT (Barbour 2017) reported on lost time including lost personal time and lost work time. Lost personal time was less among patients having hybrid visits compared to those having in-person visits (MD −1.28 hours, 95% CI −2.44 to −0.13). The difference in work time lost was similar to personal time, but the difference between groups was not statistically significant (MD −1.27 hours, 95% CI −3.11, 0.56).

Based on the single study, there is insufficient evidence to yield a conclusion regarding the effect of televisits on lost time (Table 9).

Delivery Timing, Gestational Age, and Infant Size

As for studies of routine visit schedules, the prioritized outcomes related to delivery timing, gestational age, and infant size are highly correlated measures. Nevertheless, across studies, each measure (preterm birth, gestational age at birth, small for gestational age, and low birth weight) was reported as an independent outcome. Thus, for example, the outcome low birth weight does not account for gestational age.

Preterm Birth

One RCT (Butler Tobah 2019) and three NRCSs (Duryea 2021, Palmer 2021, Pflugeisen 2016) reported on preterm birth (birth <37 weeks). All four studies found that preterm birth occurred at similar rates between patients having hybrid visits and in-person visits for antenatal care. The effect sizes across studies ranged from 0.96 to 1.34; the RCT provided a highly imprecise effect size estimate with few preterm birth events. Meta-analysis of the three NRCSs yielded a nonsignificant difference in risk of preterm births by visit type (OR 0.93, 95% CI 0.84 to 1.04; P=0.18); addition of the highly imprecise RCT did not affect the estimate (Figure 6). Of note, the two studies that trended toward lower risk of preterm births with hybrid visits (Duryea 2021 and Palmer 2021) both compared hybrid visits during the COVID-19 pandemic with in-person visits pre-pandemic. Neither study attempted to adjust for underlying differences between pre-pandemic and pandemic era patients such as change in work status, office work, societal stressors, and changes in diet and exercise.

The Duryea 2021 NRCS also reported births <34 weeks, with similar rates reported between the two groups (3.3% vs. 3.1%, unadjusted RR 1.08, 95% CI 0.89 to 1.31).

Overall, we found low SoE of no evidence of a difference between hybrid visits (with televisits) and all in-person visits in preterm birth (Table 9). Our confidence was reduced due to methodological limitations of the studies and a lack of precision in findings in the RCT.

Figure 6

Meta-analysis of hybrid versus in-person visits: Preterm births (RCT and NRCSs). Abbreviations: CI = confidence interval NRCSs = nonrandomized comparative studies, OR = odds ratio, RCT = randomized controlled trial..

Gestational Age at Birth

One NRCS (Pflugeisen 2016) reported on gestational age at birth. Similar gestational ages at birth (38.9 weeks vs. 39.1 weeks) were reported between patients having hybrid visits and in-person visits for antenatal care.

Based on the single study, there is insufficient evidence to yield a conclusion regarding the effect of televisits on gestational age at birth (Table 9).

Low Birth Weight

One RCT (Butler Tobah 2019) and one NRCS (Palmer 2021) reported on low birth weight. The RCT had few low-birth-weight infants (defined as <2500 g) and thus provided a highly imprecise estimate of effect size (OR 0.49, 95% CI 0.04 to 5.50) between groups. The NRCS reported on multiple definitions of low birth weight but found no differences between groups for any outcome. The OR for singletons at <3^rd percentile was 1.06 (95% CI 0.76 to 1.49), for full-term singletons at <3^rd percentile was imprecise at 0.95 (95% CI 0.46 to 1.97), and for singletons at <10^th percentile was 0.97 (95% CI 0.82 to 1.15).

We found there is insufficient evidence to yield a conclusion regarding the effect of televisits on low birth weight (Table 9). The RCT was highly imprecise; thus, only the single NRCS provided interpretable evidence. Since the conclusion is based on a single, high risk of bias NRCS, we conclude there is insufficient evidence.

Apgar Score

One RCT (Butler Tobah 2019) reported on infants with an abnormal Apgar score (<7). The study had few low Apgar events and thus provided a highly imprecise estimate of effect size (OR 1.50, 95% CI 0.25 to 9.12) between groups. The analysis did not account for gestational age.

Based on the single study, there is insufficient evidence to yield a conclusion regarding the effect of televisits on Apgar score (Table 9).

Abnormal Umbilical Cord Blood

One NRCS (Duryea 2021) reported on umbilical cord blood pH <7.0. Similar rates (0.3% vs. 0.4%, RR 0.84, 95% CI 0.46 to 1.62) were reported between patients having hybrid visits and those having in-person visits for antenatal care.

Completion of ACOG Recommended Services

One RCT (Butler Tobah 2019) reported on completion of ACOG recommended services at 36 weeks of gestation. The study found no significant difference between patients having hybrid visits and those having in-person visits in terms of having influenza vaccine (80.2% vs. 83.8, P = 0.44), Tdap booster (90.8% vs. 90.0%, P = 0.82), mid-pregnancy education (96.6% vs. 94.6%, P = 0.35), group B Streptococcus screening (97.7% vs. 99.2%, P = 0.32), and depression screening (99.2% both groups).

Based on the single study, there is insufficient evidence to yield a conclusion regarding the effect of televisits on completion of ACOG recommended services (Table 9).

Neonatal Intensive Care Unit Admission

Three NRCSs (Duryea 2021, Palmer 2021, Pflugeisen, 2016) reported on NICU admission. All found no significant differences in NICU admissions between groups with ORs ranging from 0.69 to 1.08. Meta-analysis of the three NRCSs found no difference in risk of NICU admission by visit type (OR 1.02, 95% CI 0.82 to 1.28) (Figure 7).

Overall, we found low SoE of no evidence of a difference between hybrid visits (with televisits) and all in-person visits in NICU admission (Table 9). Our confidence was reduced due to methodological limitations of the unadjusted NRCSs.

Figure 7

Meta-analysis of hybrid versus in-person visits: NICU admission (NRCSs). Abbreviations: CI = confidence interval, NICU = neonatal intensive care unit, NRCSs = nonrandomized comparative studies, OR = odds ratio.

Prioritized Outcomes Not Reported by Eligible Studies

Numerous outcomes of interest to stakeholders were not evaluated by eligible studies. Among the prioritized outcomes, these include:

• Maternal psychosocial, preference, and related outcomes

  • Quality of life
  • Anxiety or depression
• Fetal, neonatal, and infant outcomes

  • Small for gestational age
  • Breastfeeding
• Resource use

  • Unplanned visits
• Harms

  • Delayed diagnoses (e.g., gestational diabetes)

Televisits Versus In-Person Visits

One survey study (Futterman 2021) compared patient satisfaction with televisits versus in-person visits among patients who had hybrid care (i.e., they had both televisits and in-person visits).⁶³ The study was conducted in the New York City region during the COVID-19 pandemic from March 1 to May 1, 2020. The average age of the 104 included patients was 31.2 years, 46 percent were English speakers, and 74 percent were Hispanic. Race was unknown for the large majority of participants. Most participants (73%) were multiparous, and 17 percent had prior cesarean delivery. The majority (72%) received low-risk level of care while 28 percent received high-risk level of care. Antenatal care was provided by either physicians (60%) or midwives (40%).

Patient Satisfaction With Antenatal Care

The only reported outcome reported by Futterman 2021⁶³ pertinent to this review was patient satisfaction measured by Short Assessment of Patient Satisfaction (SAPS). The SAPS survey was conducted by phone. The survey includes seven questions, each of which is scored from 0 to 4, with higher scores indicating greater satisfaction. The study reported that total SAPS scores of 0 to 10 indicate “very dissatisfied”, 11 to 18 “dissatisfied”, 19 to 26 “satisfied”, and 27 to 28 “very satisfied”. The study reported median scores for all participants and compared various subgroups.

The study found a lower level of satisfaction, overall, with televisits than in-person visits (median scores 20 vs. 24, P=0.008), but both median scores were in the “satisfied” range. Among 104 participants, 10 were dissatisfied with televisits, while seven were dissatisfied with in-person visits (P=0.45). Among the participants, 67 percent were “satisfied” with televisits and 69 percent were “satisfied” with in-person visits.

The study reported median satisfaction for different subgroups of participants finding very similar median scores in all subgroups (televisit median scores 22 to 23 vs. in-person median scores 22 to 25.5). While statistically significant differences between satisfaction with televisits and in-person visits were found in some subgroups, favoring in-person visits, the findings are somewhat difficult to interpret because all median scores are within the “satisfied” range. Statistically significant differences (favoring in-person visits) were found among multiparous patients, English-speaking patients, non-Hispanic patients, those receiving low-risk care, and those receiving care from midwives (although the reported median scores for the subgroup seen by midwives were identical to each other). Furthermore, the study did not account for multiple testing across multiple subgroups. The study also compared different subgroups’ satisfaction with televisits (separate from comparisons of televisits and in-person visits), but found no statistically significant differences; although, the study was unlikely to be powered for multiple subgroup analyses.

Incorporating this study with the two studies comparing hybrid (televisits and in-person) visits versus in-person visits alone (Table 9) would add inconsistency. We would, thus, deem there to be inadequate evidence with no conclusion.

Qualitative Evidence Pertaining to Televisits

Five studies (Madden 2020, Mary 2021, Peahl 2021, Quinn 2021, and Stanhope 2022) provided qualitative data on televisits for routine antenatal care.^64–68 Four studies were based in the United States and one study (Quinn 2021) was based in England.

Madden 2020 explicitly sought to understand providers’ experiences with transitioning to use of televisits; both providers and clinic leadership were surveyed.⁶⁴ The study interviewed 11 providers (including “operations leaders”), mostly (83%) physicians, but not patients, at two affiliated New York City hospitals.

Mary 2021 sought to explore the implementation of a new perinatal telemedicine network linking Level I and II hospitals with the Level IV hospitals in Maryland.⁶⁵ The telemedicine network included both a new system for conducting televisits and telemedicine communications among the hospitals and the clinicians. The study interviewed 24 obstetric clinicians (nurse, nurse-midwives, obstetricians) and “telemedicine experts”, but not patients.

Peahl 2021 sought to understand institutional-level adoption and patient and provider experiences with an antenatal care model implemented due to the COVID-19 pandemic, which was largely delivered via televisits.⁶⁶ The study sought patient, provider, and clinic leadership perspectives. The study sent questionnaires with open-ended responses to 203 patients (26% of sample) and providers at affiliated hospitals around Ann Arbor, Michigan. The patients were mostly White (71%), with few Black or Hispanic patients (<5%). Patients mostly had private insurance (78%) and were well-educated (2/3 with bachelors or advanced degree). The providers were mostly physicians with general obstetrics and gynecology (35%) or family medicine (38%) practices; 12 percent of providers were certified nurse-midwives.

Quinn 2021 sought to assess the experiences of patients and healthcare providers with virtual antenatal clinic appointments during the COVID-19 pandemic.⁶⁷ The study sent questionnaires with open-ended responses to 240 patients and 45 healthcare professionals (both clinical staff delivering virtual care and clerical staff organizing virtual care) in a tertiary obstetric unit in Leicester, England who experienced at least one virtual consultation. The patient and healthcare professional surveys had a 62% (148/240) and 82% (37/45) response rate, respectively. Patient respondents were typically multiparous (90%), had an average age of 31 years, and were predominantly White (79%). Healthcare professional respondents reflected diverse clinical roles (consultants 38%, registrars 38%, junior doctors 16%, and midwives 4%) with diverse years of obstetric experience and experience with virtual consultations.

Stanhope 2022 sought to assess patients’ perceptions of quality of care and their satisfaction with care following changes to delivery of care during the COVID-19 pandemic in a hospital in Atlanta, Georgia.⁶⁸ The study employed a sequential mixed-methods design, first surveying patients with a quantitative survey followed by in-depth interviews with a purposive sample of those who consented to be contacted for further information. Sixteen individuals participated in the interviews, the majority of whom were non-Hispanic Black, between 20 and 35 years old, multiparous, and Medicaid-insured.

All studies except Quinn 2022 conducted thematic content analysis to identify themes within the qualitative data relevant to their study questions. Quinn 2022 reported findings from open-ended responses but did not report their qualitative analysis methods. We considered the reporting of the qualitative findings to be thin in this study due to sparse reporting or analysis of their open-ended text data.

Madden 2020, Mary 2021, and Stanhope 2022 were assessed to have low risk to rigor. Peahl 2021 and Quinn 2021 was assessed to have moderate risk to rigor due to limited conduct (and/or reporting) of qualitative methods and analysis plans, and a lack of consideration of the relationship between researcher and participants. Appendix C Table C-4-3 summarizes the risk to rigor assessment of the studies. With the exception of the domain of optimism, which was based on the finding of one thin study with moderate risk to rigor (Quinn 2021), all synthesized findings had no concerns or minor concerns with risks to rigor (Appendix Table D-3-4). We had no concerns or minor concerns with the coherence of findings (i.e., the synthesized findings reflected the complexity and variation of the data) but minor to moderate concerns with the adequacy of findings due to the limited number of studies, geographical/health setting, and population representation.

We coded data provided by the five studies into eight of the TDF domains including: knowledge; skills; beliefs about capabilities; optimism; beliefs about consequences; environmental context and resources; social influences; and emotion. We did not find evidence from extracted qualitative data for the remaining six TDF domains: social/professional role and identity; reinforcement; intentions; goals; memory, attention, and decision processes or behavioral regulation. Table 10 summarizes the barrier and facilitator themes related to adopting televisits for routine antenatal care. Themes are presented with the eight TDF domains identified, organized by whether the themes related to patients, providers, or the clinic/office. Providers, some of whom were also in clinic leadership positions, discussed their own, but also perceived patient and health system perspectives. No study provided data on partner perspectives.

Appendix D Table D-3-3 provides the extracted text from studies linked to their synthesized themes. Appendix D Table D-3-4 is the GRADE-CERQual table summarizing findings and conclusions for each TDF theme. One finding (related to the TDF domain of environmental context and resources) was supported by five studies, two findings (related to the TDF domains of beliefs about consequences and emotion) were supported by four studies, two findings (related to the TDF domains of knowledge and skills) were supported by three studies, one finding (related to the TDF domain of social influences) was supported by two studies, and two findings (related to the TDF domains beliefs about capabilities and optimism) were supported by one study.

TDF Domain: Emotion

Four studies (Madden 2020, Mary 2021, Stanhope 2022, Quinn 2021) provided evidence pertaining to emotion. The studies reported that providers perceived that patients’ psychological apprehension with televisits and general preference for in-person visits may pose a barrier to the uptake of telehealth technology. However, providers noted that patients’ appreciation for continuity of care and humanized interaction may be a facilitator.

Table 10

Coding extracts for antenatal televisits into the Theoretical Domains Framework.