key: cord-0334828-qn63zf47
authors: Aiken, A. R. A.; Lohr, P. A.; Lord, J.; Ghosh, N.; Starling, J.
title: Effectiveness, safety and acceptability of no-test medical abortion provided via telemedicine: a national cohort study
date: 2020-12-07
journal: nan
DOI: 10.1101/2020.12.06.20244921
sha: 7d8ab3f390b75bf0553996f778080239c63b2c22
doc_id: 334828
cord_uid: qn63zf47

Objectives To compare the effectiveness, safety and acceptability of medical abortion before and after the introduction of no-test telemedicine abortion care. Design Cohort study to assess whether a no-test telemedicine-hybrid care model (telemedicine with in-person provision only when indicated) was non-inferior to the traditional service model (blanket in-person provision including ultrasound scan). Setting The three main abortion providers in England and Wales. Participants All patients having an early medical abortion in the two months before and after the service model change. Patient demographic and clinical characteristics were compared between the cohorts to adjust for any systematic differences in the two groups. Main outcome measures Access: waiting time, gestation at abortion Effectiveness: the proportion of successful medical abortions Safety: significant adverse events defined as: haemorrhage requiring transfusion, significant infection requiring hospital admission, major surgery, death. We also examined the incidence of ectopic pregnancy and late gestation. Acceptability: Patient-reported outcomes of satisfaction, future preference, and privacy of consultation Results The study sample included 52,142 medical abortions; 22,158 in the traditional cohort and 29,984 in the telemedicine-hybrid cohort, of which 61% were provided using no-test telemedicine. The cohorts accounted for 85% of all medical abortions provided in England and Wales during the study period. Mean waiting times were 4.2 days shorter in the telemedicine-hybrid cohort, and 40% were provided at <=6 weeks' gestation compared to 25% in the traditional cohort (p<0.001). There was no difference in success rates between the two groups (98.2% vs. 98.8%, p=1.0), nor in the prevalence of serious adverse events (0.04% vs. 0.02%, p=0.557). The incidence of ectopic pregnancy was equivalent in both cohorts (0.2%, p=0.796), with no significant difference in the proportions being treated after abortion (0.01% vs 0.03%, p=0.123). In 0.04% of cases the abortion appeared to have been provided at over 10 weeks' gestation; these abortions were all completed at home without additional medical complications. In the telemedicine-hybrid group, the effectiveness for abortions conducted using telemedicine (n=18,435) was higher than for those conducted in-person (n=11,549), 99.2% vs. 98.1%, p<0.001. Acceptability was high (96% satisfied), 80% reported a future preference for telemedicine and none reported that they were unable to consult in private using teleconsultation. Conclusions Medical abortion provided through a hybrid model that includes no-test telemedicine without routine ultrasound is effective, safe, acceptable, and improves access to care.

≤ 6 weeks' gestation compared to 25% in the tradi-106 tional cohort (p<0.001). There was no difference in success rates between the two groups 107 (98.2% vs. 98.8%, p=1.0), nor in the prevalence of serious adverse events (0.04% vs. 0.02%, 108

p=0.557). The incidence of ectopic pregnancy was equivalent in both cohorts (0.2%, 109

p=0.796), with no significant difference in the proportions being treated after abortion (0.01% 110 vs 0.03%, p=0.123). In 0.04% of cases the abortion appeared to have been provided at over 111 10 weeks' gestation; these abortions were all completed at home without additional medical 112

complications. In the telemedicine-hybrid group, the effectiveness for abortions conducted 113 using telemedicine (n=18,435) was higher than for those conducted in-person (n=11,549), 114

99.2% vs. 98.1%, p<0.001. Acceptability was high (96% satisfied), 80% reported a future 115 preference for telemedicine and none reported that they were unable to consult in private us-116

ing teleconsultation. 117 118 119

Conclusions 120

Medical abortion provided through a hybrid model that includes no-test telemedicine without 121 routine ultrasound is effective, safe, acceptable, and improves access to care. 122 123 124 125 Introduction 126 Improved access to abortion care would deliver significant advantages for both healthcare 127 systems and the women who use them. Abortion is a common reason for needing healthcare -128 the global abortion rate is estimated at 39 abortions per 1000 women aged 15-49 years, 1 with 129 28% of all pregnancies in developed countries ending in abortion. 2 There is clear evidence 130 that restricting access to abortion does not reduce abortion rates, it simply makes abortion 131 less safe. 1 3 Improving access is likely to benefit those who are most vulnerable, 4 especially in 132 resource-poor settings or where care has to be self-funded. In its 2019 guideline on abortion 133 care, the National Institute for Health and Care Excellence (NICE) stated that improving ac-134 cess to abortion was a key priority. 5 135 136 Telemedicine, the use of information and communication technologies to improve patient 137 outcomes by increasing access to care and medical information, 6 has been noted to decrease 138 costs and increase convenience and safety. 7 It is an established service delivery model for 139 abortion care in many settings, 8 and it is recommended to improve access. 4 The COVID-19 140 pandemic required urgent action by clinicians and policymakers to ensure delivery of essen-141 tial health services, including abortion. In March 2020, the Royal College of Obstetricians 142 and Gynaecologists (RCOG) published guidelines to safeguard abortion care in the UK. 9 The 143 guidelines profoundly changed the way medical abortion care is delivered in Great Britain. 144

Prior to the emergence of COVID-19, all patients seeking medical abortion were required to 145 attend in-person to receive an ultrasound scan and have mifepristone administered within the 146 clinic. Under the new guidelines, consultations were encouraged to take place by telephone or 147 video call; an ultrasound scan was required only if indicated. All the British governments had 148 issued emergency legal orders by March 30 th 2020 to allow mifepristone to be used at home 149 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint along with misoprostol up to 10 weeks' gestation. [10] [11] [12] These approvals permitted abortion 150 providers to implement a telemedical service delivery model for medical abortion. 151 152 Great Britain's new pathway for no-test telemedicine abortion is unusual among existing 153 models because it is fully remote: no clinic visit, tests or ultrasound scan are required and 154 both mifepristone and misoprostol are delivered by mail or collected from a clinic for use at 155

home. This new service model thus presents an important opportunity to evaluate a poten-156 tially better way to provide medical abortion that could improve access and reduce the barri-157 ers posed by in-person care. 13 We examined and compared the effectiveness, safety and ac-158 ceptability of medical abortion provided up to 10 weeks' gestation before and after the ser-159 vice model change. 

The study population comprised all patients who accessed an early medical abortion (EMA) 165 at the three largest abortion providers in England and Wales -British Pregnancy Advisory 166 Service (BPAS), MSI Reproductive Choices (MSUK), and the National Unplanned Preg-167 nancy Advisory Service (NUPAS) -two months before and after the service model change. 168

Medical abortion is defined as the use of medications to cause termination of a pregnancy 169 without primary surgical intervention and 'EMA' applies to these procedures in the first tri-170 mester. 14 The recommended EMA regimen uses the anti-progestogen mifepristone in an oral 171 dose of 200mg and the prostaglandin analogue misoprostol by the sublingual, vaginal or buc-172 cal route. A single dose of 800mcg misoprostol used 24-48 hours after mifepristone has been 173 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint demonstrated to be highly effective up to 10 weeks' gestation. 15 Beyond 10 weeks additional 174 doses of misoprostol are needed to maintain the same degree of effectiveness. 16 175 176 Our dataset consisted of information on EMAs extracted directly from electronic service re-177 cords and included fully de-identified patient clinical and demographic characteristics, as 178 well as clinical outcomes. All data were extracted six weeks after the end of the study period 179 to ensure the reporting of complications was as complete as possible. 180 181 Two cohorts were defined. The "traditional" cohort comprises all patients having an EMA 182 between January 1 st and March 1st 2020, prior to service model change. All patients in this 183 cohort received care using the traditional pathway that included an in-person assessment and 184 an ultrasound scan. The "telemedicine-hybrid" cohort comprises all patients accessing an 185 EMA between April 6 th and June 30 th 2020, in a two month period after the service model 186 change at each provider. Patients in this cohort received care using no-test telemedicine if 187 they had low risk of ectopic pregnancy and their self-reported last menstrual period (LMP) 188 indicated a gestation of less than 10 weeks. Those who were not eligible for telemedicine had 189 an in-person assessment with ultrasound. Providers followed organisation-specific evidence-190 based policies informed by the RCOG 9 and associated decision aid 17 (figures 1 and 2) and 191 earlier RCOG and NICE guidelines for general abortion care including in-person medical 192 abortion. 18 19 193 194 In the traditional cohort, whilst some aspects of the pre-abortion consultation may have been 195 carried out by phone, an in-person assessment with ultrasound scan and in-clinic administra-196 tion of mifepristone were always required. In the telemedicine-hybrid cohort, we defined pa-197 tients served by telemedicine as those who received medications for home use (either by mail 198 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint or collection with minimal contact from a clinic) following a phone or video consultation and 199 who did not receive an ultrasound scan or other tests. Waiting time and gestation at treatment were used to assess how the two models impacted 212 access. Waiting time was defined as the interval from first contact with the abortion provider 213 to when medication was dispensed (either in-person in clinic or posted). Gestation was re-214 corded as that applicable on the date mifepristone was dispensed or provided. Analysis was 215 of mean gestation and the proportion of abortions performed at ≤ 6 weeks' gestation. 216 217

Effectiveness was defined as the proportion of medical abortions that were successful. 219

Success was defined according to the MARE Guidelines as successful expulsion of an 220 intrauterine pregnancy without need for surgical intervention. 14 Sub-categories of 221 unsuccessful abortion included: surgical management of continuing pregnancy, surgical 222 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint evacuation of retained products of conception, and continuing pregnancy where the patient 223 opted for continuation or the outcome is unknown. 224 225 Safety 226

Safety was defined according to the proportions of medical abortions that involved one or 227 more significant adverse events. We defined significant adverse events as haemorrhage re-228 quiring transfusion, significant infection requiring hospital admission, major surgery, and 229 death. We also examined the incidence of ectopic pregnancy and when it was diagnosed in 230 the care pathway. All patients referred for further diagnostics (e.g. serial β hCG monitoring) 231 but who had no further treatment are included in the ectopic pregnancy group although a pro-232 portion of these will be pregnancy of unknown locations (PULs) that includes failed early 233 intrauterine pregnancies. We analysed the proportion of cases where treatment was reported 234

to have occurred at Given the constraints of delivering healthcare during COVID-19 it was not possible to follow 240 up with all patients to capture patient-reported outcomes. Two of the providers (BPAS and 241 MSUK) collected data on patient feedback during the study period. Patients were invited to 242 provide feedback one to three weeks after treatment, either by telephone using a structured 243 interview tool (MSUK) 20 or using an online form (BPAS) 21 . We analysed data on questions 244 reporting on satisfaction, future preference and privacy. Although the questions were similar, 245 only the MSUK survey included questions on the privacy of teleconsultation. All contact was 246 by a non-clinician who had not been involved in the patient's care. 247 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint 248 Analysis 249

The primary analysis was to assess whether the telemedicine-hybrid care model (telemedicine 250 with in-person provision when indicated) was non-inferior to the traditional service model 251 (blanket in-person provision). Non-inferiority is established by comparing effectiveness and 252 safety in the two cohorts under the null hypothesis that the rates of successful abortion and 253 serious adverse events are not different. 254

We first compared patient demographic and clinical characteristics between the cohorts to 256 assess the need to covariate-adjust our hypothesis tests for systematic differences in the two 257 groups that might affect abortion outcomes. All hypothesis tests were covariate adjusted for 258 patient age, race/ethnicity, gestational age, parity, and prior abortions using logistic regres-259 sion and weighted risk differences. 22 260 261 We evaluated effectiveness by testing the alternative hypothesis that the rate of successful 262 medical abortion in the telemedicine-hybrid cohort is lower than in the traditional cohort us-263 ing a covariate-adjusted test of difference in proportions. We also performed a chi-squared 264 test to evaluate whether the distribution of unsuccessful abortion sub-categories differed be-265 tween the cohorts. We evaluated safety by testing the alternative hypothesis that significant 266 adverse events occurred at higher rates in the telemedicine-hybrid cohort compared to the 267 traditional cohort using a covariate-adjusted hypothesis test for difference of proportions. We 268 also evaluated whether the prevalence of ectopic pregnancies managed before EMA and after 269 EMA were different between the traditional and telemedicine-hybrid cohorts using chi-270 squared difference of proportions tests. 271 272 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint

The secondary analysis was to compare effectiveness and safety of medical abortion for pa-273 tients who received fully remote no-test telemedicine vs. in-person care in the telemedicine-274 hybrid cohort, primarily to assess whether any differences between the cohort service models 275 were driven by one particular group. We evaluated patient demographic and clinical charac-276 teristic differences between the two groups to assess the need to covariate-adjust our hy-277 pothesis tests for systematic differences and all hypothesis tests were covariate adjusted for 278 patient age, race/ethnicity, gestational age, parity, and prior abortions using the framework 279 described above. We note, however, that these cohorts are fundamentally different despite 280 covariate adjustment, as patients in the telemedicine-hybrid cohort were selected into in-281 person treatment based on characteristics that would affect the outcome of their abortion. We 282 performed covariate-adjusted hypothesis tests under the null hypothesis of equal effectiveness 283 and equal rates of significant adverse events in the telemedicine vs. in-person groups. 284 285 All analyses were performed using R, version 3.6.2. Statistical significance was defined using 286 an alpha level of 0.05. 287

288

The study was reviewed by the Institutional Review Board (IRB) at the University of Texas 290 at Austin and a determination was made that the research did not meet the criteria for human 291 subjects research as defined in the Common Rule (45 CFR 46) or FDA Regulations (21 CFR 292 56). Each provider ensured compliance with their own internal ethics and governance sys-293 tems. The principles of the STROBE statement, and the MARE supplement, were followed. 14 294 295

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint

In total, 52,142 medical abortions were provided during the study period; 22,158 in the tradi-297 tional pathway and 29,984 in the telemedicine-hybrid cohort. Among those that took place in 298 the telemedicine-hybrid cohort, 18,435 (61%) were provided entirely via telemedicine and 299 11,549 (39%) were provided in-person. Our sample represents 85% of the total number of 300 medical abortions performed in England and Wales during the study period. 23 The clinical 301 and demographic characteristics of patients in the two cohorts are described in Table 1 . 302 303 Access 304

Mean waiting time to treatment declined from 10.7 days (SD 19.9) in the traditional pathway 305 to 6.5 days (SD 13.5) in the telemedicine-hybrid cohort (p<0.001). Mean gestational age at 306 treatment also declined in the telemedicine-hybrid cohort resulting in 40% of abortions per-307 formed at 6 weeks' gestation or less vs. 25% in the traditional pathway (p<0.001). 308 309 Effectiveness 310

Rates of successful medical abortion were high under both service delivery models ( Table 2) : 311 98.2% in the traditional cohort vs. 98.8% in the telemedicine-hybrid cohort. We found no 312 evidence of a lower success rate with the telemedicine-hybrid pathway (p=1.0). The distribu-313 tion of the different sub-categories of unsuccessful medical abortion did not differ between 314 the cohorts (p=0.268). 315 316 Safety 317

Significant adverse events in both cohorts were rare (Table 3) . Haemorrhage requiring trans-318 fusion was reported in 8 (0.04%) cases in the traditional cohort and 7 (0.02%) cases in the 319 telemedicine-hybrid cohort. No cases of significant infection requiring hospital admission, 320 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint major surgery or death were reported. We found no evidence that significant adverse events 321 were higher in the telemedicine-hybrid cohort (p=0.557). 322

The overall incidence of ectopic pregnancy was equivalent in both cohorts, 39 (0.2%) in the 324 traditional pathway and 49 (0.2%) in the telemedicine pathway, p=0.796 (Table 4 ). The pro-325 portions managed after EMA were not significantly different between the cohorts (0.01% in 326 the traditional pathway and 0.03% in the telemedicine-hybrid pathway, p=0.123). There were 327 11 cases (0.04%) in the telemedicine-hybrid cohort where the gestational age after abortion 328 was reported as being greater than the expected 10 weeks. In all these cases, the medical 329 abortion was completed at home without additional medical complications. 330 331 Acceptability 332

Patient-reported outcome data was available from 2453 respondents. 96% were "satisfied" or 333 "very satisfied" with their care, or rated their experience as "good" or "very good". 80% re-334 ported that they would choose telemedicine in the future or that it was their preferred option, 335 with 13% choosing in-person care and the remainder being unsure. No patient reported that 336 they were unable to consult in private using teleconsultation (n=1243). 337 338

Comparison of clinical outcomes for the telemedicine vs. in-person groups in the telemedi-340 cine-hybrid cohort are shown in Tables S1-S3. Rates of successful medical abortion were 341 higher in the telemedicine group (99.2% vs 98.1%, p<0.001), and rates of significant adverse 342 outcomes were not significantly different between the two groups, 3 (0.02%) for telemedicine 343 vs. 4 (0.03%) for in-person, p=0.532. 344 345 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. Our study confirms previous literature that medical abortion is safe and effective with low 362 rates of significant complications. 25 The slight increase in effectiveness we observed in the 363 group that received telemedicine -even after controlling for lower average gestational age 364 compared to the in-person group -may be due to the ability of patients to better control the 365 timing at which they took the medication. 366

The telemedicine-hybrid model resulted in very low rates of undiagnosed ectopic pregnancy 368 and later than expected gestations. Although the rate of ectopic pregnancy in the general 369 population in the UK and USA is reported as 1-2%, 26 27 the rate reported among patients hav-370 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint ing an abortion is 10 times lower, 28 which is consistent with our findings. Ultrasound is not 371 used to screen for ectopic pregnancy in the general population -it is only used where signs 372 and symptoms suggest a need. 27 Routine screening of symptom-free women is associated 373 with a high false positive rate when the prevalence of ectopic pregnancy is low, as is the case 374 in women seeking abortion, and therefore it is unlikely there would be significant benefits. 29 375

There is no clinical justification for maintaining this inconsistency in care between women 376 wishing to continue their pregnancies and those choosing EMA. 30 31 377 378 However, given that over 200,000 people access abortion care each year in the UK alone, 379 some will inevitably have an asymptomatic ectopic pregnancy and so will proceed with hav-380 ing mifepristone and misoprostol either through telemedicine or after a false negative scan. 381

The essential issue for safety is that these are detected prior to causing harm rather than prior 382 to beginning medical abortion treatment; treatment with mifepristone and misoprostol in it-383 self will have no effect on an underlying ectopic pregnancy. Indeed, the reduction in waiting 384 times afforded by the telemedicine model may facilitate earlier detection than traditional 385 pathways where women present later, or are sent away to give additional time to visualise an 386 intrauterine pregnancy on scan. Proceeding with early medical abortion without a scan may 387 permit earlier diagnosis of a developing ectopic pregnancy owing to increased surveillance 388 and index of suspicion, for example where there is minimal bleeding after misoprostol. 9 18 389 390 The proportion of cases where gestational age was later than expected based on LMP was 391 low, as might have been expected given the evidence that women can determine the gesta-392 tional age of their pregnancy with reasonable accuracy by LMP alone. 31 Nevertheless inad-393 vertent treatment of gestations over 10 weeks is inevitable and, consistent with our findings, 394 the consequences for most are unlikely to be medically significant. 32 The 10 weeks' gestation 395 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint limit in the English government's approval order is arbitrary, and not based on evidence of 396 safety or effectiveness. The Scottish government did not stipulate a limit, leaving the decision 397 to the discretion of the clinician in consultation with their patient. Moreover, the reported 398 success of self-managed abortions at >12-24 weeks' gestation is 93%, with safety similar to 399 that expected in earlier gestations. 33 400 401

While the study is not a clinical trial and we were not able to control the assignment of pa-403 tients to treatment groups to conduct a direct comparison of fully remote telemedicine with 404 in-person care, the policy changes precipitated by COVID-19 created a rare opportunity to 405 conduct an evaluation of a major change in service delivery model with a very large sample 406 size. We were able to evaluate the outcomes of both the telemedicine-hybrid and traditional 407 in-person services as they operate in the real world, and we were able to adjust for key co-408 variates. A key strength of the study is the generalisability of our findings given that our 409 sample included 85% of all medical abortions provided in England and Wales during the 410 study period. 411

The main limitation of this study is that we were unable to actively follow-up patients post-413 abortion and therefore only significant adverse events can be reported with confidence. There 414 is a potential gap in the consistency of reporting incidents, due to some complications not 415 meeting the threshold of serious incidents, multiple routes of entry into the NHS and informal 416 communication between the NHS and abortion providers. Although it is possible that some 417 patients presented to other providers and a significant adverse event was not reported in our 418 dataset, the risk management and reporting systems within the NHS are well defined, with 419 serious incidents being routinely shared with regulators who would expect providers to en-420 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint sure actions had been taken to mitigate risks to patients in the future. More importantly, there 421 is no reason to suspect that any under-reporting that did occur would be more likely in either 422 cohort so as to introduce bias. The governing body of the NHS in England alerted all com-423 missioners of the need to report incidents relating to telemedicine and there were review 424 meetings of key stakeholders to ensure compliance. We consulted with regulators and na-425 tional agencies to ensure that we accounted for reports made directly to them. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint 446 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint where the null hypothesis is that the traditional cohort has the same effectiveness rate as the 466 telemedicine-hybrid cohort and the alternative hypothesis is that the traditional cohort has a 467 higher effectiveness rate than the telemedicine-hybrid cohort. The p-value for unsuccessful 468 medication abortion is for the chi-squared test of whether the distribution of types of failure 469 differ between the cohorts. 470 471 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint 472 Note: As explained in the methods section, the p-value was calculated using a hypothesis test 477

where the null hypothesis is that the traditional cohort the same rate of adverse events than 478 the telemedicine-hybrid cohort and the alternative hypothesis is that the traditional cohort has 479 a lower rate of adverse events than the telemedicine-hybrid cohort. 480 481 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint Supplementary Tables to Accompany Text in the Results Section  493  494  495  496  Table S1 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint 504 505 where the null hypothesis is that the in-person has the same effectiveness rate as the telemed 514 group and the alternative hypothesis is that the in-person group has a lower effectiveness rate 515 than the telemedicine group. The p-value for unsuccessful medical abortion is for the chi-516 squared test of whether the distribution of types of failure differ between in-person and tele-517 med groups. 518 519 520 521 522

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint 

Major surgery 0 (0.0) 0 (0.0) Death 0 (0.0) 0 (0.0) 526

Note: As explained in the methods section, the p-value was calculated using a hypothesis test 527

where the null hypothesis is that the in-person group has the same rate of adverse events than 528 the telemedicine group and the alternative hypothesis is that the in-person group has a lower 529 rate of adverse events than the telemedicine group. 530 531 532 533 534

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint [Note for editors -permission for reproduction obtained from RCOG] 556 557

Image from: 2020-06-04-decision-aid-for-early-medical-abortion-without-ultrasound.pdf 558 (rcog.org.uk) 559 560

If preferred, can provide original files in original Powerpoint format 561 562

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint 563 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint 564 565

. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint Acknowledgements 566 We would like to thank the Care Quality Commission (CQC), NHS England and NHS Im-567 provement (NHSE & I), and the National Child Mortality Database (NCMD) for their advice 568

and assistance in ensuring all complications reported within NHS systems have been in-569 cluded. 570 571

We are also enormously grateful to the many people who helped to ensure access to essential 572 healthcare for women needing an abortion not only continued, but was improved during the 573 COVID-19 pandemic. There are too many to list individually, but we would like to especially 574 thank the following who not only helped to implement a complete service change within a 575 few weeks, but also made this paper possible by ensuring the data was collected and accessi- . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted December 7, 2020. The Corresponding Author has the right to grant on behalf of all authors and does grant on 634 behalf of all authors, a worldwide licence to the Publishers and its licensees in perpetuity, in 635 all forms, formats and media (whether known now or created in the future), to i) publish, re-636 produce, distribute, display and store the Contribution, ii) translate the Contribution into other 637 languages, create adaptations, reprints, include within collections and create summaries, ex-638 tracts and/or, abstracts of the Contribution, iii) create any other derivative work(s) based on 639 the Contribution, iv) to exploit all subsidiary rights in the Contribution, v) the inclusion of 640 electronic links from the Contribution to third party material where-ever it may be located; 641 and, vi) licence any third party to do any or all of the above 642 643 644

The collated datasets, which include participant data with anonymised identifiers, are held by 646 AA at the University of Texas. Consideration will be given to sharing this with bone fide re-647 searchers on application. The original data resides with the co-authors' own institutions. Al-648 though the data is de-identified, some relate to very rare events and could therefore result in 649

identification. Therefore data on complications, and data arising from clinical incident re-650 ports, will be subject to the same access restrictions as those of the organisation supplying it. 651 652 653

The authors affirm that the manuscript is an honest, accurate, and transparent account of the 656 study being reported; that no important aspects of the study have been omitted; and that any 657 discrepancies from the study as originally planned (and, if relevant, registered) have been ex-658 plained. 659 660 661

None 664 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted December 7, 2020. ; https://doi.org/10.1101/2020.12.06.20244921 doi: medRxiv preprint

Unintended pregnancy and abortion by income, 668 region, and the legal status of abortion: estimates from a comprehensive model for 669 1990-2019. The Lancet Global Health

Abortion incidence between 1990 and 2014: global, 672 regional, and subregional levels and trends

The British 1967 Abortion Act-still fit for purpose?

National Institute for Health and Care Excellence (NICE). Abortion care

Review A -Accessibility and sustainability of abortion services. In: National Institute 679 for Health and Care Excellence (NICE), ed. NICE guideline NG140

Access to and sustainability of abortion services: a 681 systematic review and meta-analysis for the National Institute of Health and Care 682 Excellence-new clinical guidelines for England

World Health 686 Organization

Telemedicine 2020 and the next decade

Telemedicine for medical abortion: a systematic 690 review

The Abortion Act 1967 -Approval of a 696 Class of Places. London: English Government, 2020. 697 11. Scottish Government. Abortion -Covid-19 -Approval For Mifepristone to be Taken At 698 Home and Other Contingency Measures. Edinburgh: Scottish Government, 2020. 699 12. Welsh Government. The Abortion Act 1967 -Approval of a Class of Place for Treatment 700 for the Termination of Pregnancy (Wales) 2020

Barriers to accessing abortion services 703 and perspectives on using mifepristone and misoprostol at home in Great Britain

Medical abortion reporting of efficacy: the MARE guidelines

Outpatient medical abortion is safe and effective 710 through 70 days gestation

A non-inferiority study of outpatient mifepristone-713 misoprostol medical abortion at 64-70 days and 71-77 days of gestation

Clinical Guidelines for Early Medical Abortion at Home -England

Early medical abortion using telemedicine -acceptability 726 to patients

Client satisfaction and experience of home 729 use of mifepristone and misoprostol for medical abortion up to 10 weeks

British Pregnancy Advisory Service: a cross-sectional evaluation

Covariate-Adjusted Difference in Proportions from 733 Clinical Trials Using Logistic Regression and Weighted Risk Differences. Drug 734 information journal : DIJ / Drug Information Association

Abortion statistics during COVID-19: data 737 tables

Demand for Self-Managed Online 740

Telemedicine Abortion in Eight European Countries During the COVID-19 741 Pandemic: A Regression Discontinuity Analysis

Incidence of emergency department visits and 744 complications after abortion

Ectopic pregnancy and medical abortion

Symptom-free women at increased risk of ectopic 753 pregnancy: should we screen?

Can mifepristone medical abortion be simplified? 756 A review of the evidence and questions for future research

Alternatives to routine ultrasound for 760 eligibility assessment prior to early termination of pregnancy with mifepristone-761 misoprostol

Commentary: No-test medication abortion: A 763 sample protocol for increasing access during a pandemic and beyond

Complications of Unsafe and Self-Managed Abortion