http://www.diva-portal.org
Postprint
This is the accepted version of a paper published in The Lancet. This paper has been peer-reviewed but does not include the final publisher proof-corrections or journal pagination.
Citation for the original published paper (version of record):
Klingberg-Allvin, M., Cleeve, A., Atuhairwe, S., Tumwesigye, N M., Faxelid, E. et al. (2015) Comparison of treatment of incomplete abortion with misoprostol by physicians and midwives at district level in Uganda: a randomised controlled equivalence trial.
The Lancet, 385(9985): 2392-2398
http://dx.doi.org/10.1016/S0140-6736(14)61935-8
Access to the published version may require subscription. N.B. When citing this work, cite the original published paper.
Permanent link to this version:
1
Title: Comparison of treatment of incomplete abortion with misoprostol by physicians 1
and midwives at district level in Uganda: a randomised controlled equivalence trial 2
3 4
*Marie Klingberg-Allvin, associate professor, School of Education, Health and Social 5
Studies, Dalarna University, Department of Women´s and Children´s Health, Karolinska 6
Institutet, 7
8
*Amanda Cleeve, PhD-student, Department of Women´s and Children´s Health, Karolinska 9
Institutet 10
11
Susan Atuhairwe, MD, Department of Obstetrics and Gynaecology at Mulago Hospital and 12
Makerere University College of Health Sciences. 13
14
Nazarius Mbona Tumwesigye, PhD, Biostatistics and Demography, School of public health,
15
Makerere University College of Health
16 17
Elisabeth Faxelid, professor, Global Health (IHCAR), Department of Public Health Sciences, 18
Karolinska Institutet 19
20
**Dr Josaphat Byamugisha, associate professor, Department of Obstetrics and Gynaecology 21
at Mulago Hospital and Makerere University College of Health Sciences. 22
23
**Kristina Gemzell, Danielsson, professor, Department of Women´s and Children´s Health, 24
Karolinska Institutet and Karolinska University Hospital, Stockholm, Sweden 25
26 27
*Shared first authorship 28
** Shared last authorship 29
30
Corresponding author: 31
Associate professor Marie Klingberg-Allvin, 32
Department of Women´s and Children´s Health, Karolinska Institutet 33
Postal address: 34
WHO-centre, C1:05 35
Karolinska University Hospital 36
SE-171 76 Stockholm, Sweden 37 Telephone: +46707174414 38 e-mail: mkl@du.se 39 40 41 42 43
2 Introduction
44
Unsafe abortion contributes substantially to the global burden of maternal mortality and 45
morbidity.1 Majority of unsafe abortions occur in low-income countries where induced 46
abortion is restricted and the unmet need for contraception is high.2 Sub-Saharan Africa caters 47
for the highest rates of unsafe abortion and global maternal mortality.3 48
Emergency treatment of complications from unsafe abortion and spontaneous abortion, post-49
abortion care (PAC), is identified as an effective intervention to decrease maternal mortality.4 50
The PAC model consists of emergency treatment of abortion related complications and post 51
abortion contraceptive counselling and provision.5 The prostaglandin E1 analogue 52
misoprostol has been suggested as an effective tool in treatment of incomplete abortion.6-8
53
International studies have compared and found no significant difference in effectiveness 54
between treatment of incomplete abortion with misoprostol, and surgical treatment with 55
Manual Vacuum Aspiration (MVA).9-12 In low resource settings misoprostol in PAC is a 56
simplified, cost effective and resource saving alternative to surgical interventions.10 57
The lack of physicians in many low-income countries limits women’s access to PAC.13 In 58
Africa the shortage of trained health care providers is greatest in rural and remote areas where 59
maternal mortality and morbidity is highest.14 In Uganda, where unsafe abortion constitutes a 60
serious public health issue, physicians are scarce and few midwives receive PAC training,15 61
restricting access to care. Task shifting is a process of delegating tasks, when appropriate, to 62
less specialized health care providers. It has been identified as a way of increasing access and 63
productivity,16 and to contribute to the building of cost effective and equitable health care 64
services.17,18 Optimizing the use of midwives would be a pragmatic response to the shortage 65
of physicians at district level in Uganda and a strategy to decrease maternal mortality.18 66
Despite the proven effectiveness and safety of misoprostol treatment of incomplete abortion 67
and the known advantages of task shifting, the involvement of midwives in diagnosing and 68
treating incomplete abortion with misoprostol has not been systematically evaluated in any 69
low resource setting. In this trial, we aimed to assess whether diagnoses and treatment of first 70
trimester incomplete abortion with misoprostol provided by a midwife is equally effective and 71
safe as that provided by a physician. 72
73
Methods 74
Trial design and participants
3
The study was designed as a multicentre randomised controlled equivalence trial conducted at 76
district level in six health care facilities in six districts in rural, peri-urban and urban settings 77
in central Uganda. The study protocol and trial followed the CONSORT guidelines for non-78
inferiority and equivalence randomised trials.197 The study was conducted between April 2013 79
and July 2014. The study was not masked. Inclusion criteria were women with signs of 80
incomplete abortion (i.e. bleeding and contractions during pregnancy, opening cervix and no 81
expulsion). Exclusion criteria were complete abortion, known allergy to misoprostol, a uterine 82
size of more than 12 weeks of gestation, suspected ectopic pregnancy, unstable hemodynamic 83
status and shock, signs of pelvic infection and/or sepsis. The health care facilities selected for 84
inclusion were equipped to provide basic and emergency obstetric services and were 85
estimated to have sufficient staff and caseload in order to be part of the study. The procedure 86
and the instruments were pilot tested at Mulago Hospital, the national referral hospital in 87
Kampala. Only minor revision was made to the instruments before recruitment started in the 88
current study. The first month was considered the run in period where the facilities were given 89
time to set the routine for screening, enrolment and follow up. This was supervised and 90
supported closely by the study coordinator. An interim report was also sent to WHO 91
following the run in phase. No changes were made to the data collection tool or the 92
recruitment process and thus the run in period was included in the analysis. The study was 93
developed and coordinated by researchers at the WHO research centre at Karolinska 94
Institutet, Stockholm, and Mulago Hospital/Makerere University, Kampala. The study was 95
approved by the Scientific and ethical review group at the Reproductive Health and Research 96
Department, WHO, Geneva. Ethical approval was further obtained from the Research Ethics 97
Committee, Makerere University, Dnr: 2012-129, Uganda National Council for Science and 98
Technology Dnr: HS 1314 and the Swedish regional ethical review board at Karolinska 99
Institutet Dnr: 2013/2;9. 100
101
Intervention and procedure
102
Eligible providers for participation were physicians and midwives involved in PAC at the 103
different facilities. The health care providers were trained according to a standardized PAC 104
training module.20,21 The five-day training programme focused on diagnosing incomplete 105
abortion, treatment with misoprostol and MVA, and contraceptive methods and counselling. 106
A number of midwives at each facility were trained to be research assistants in the study and 107
were responsible for eligibility screening and enrolment of participants. The midwives 108
conducted clinical assessments and were responsible for the follow up visits including 109
4
assessed abortion status at follow up, gave contraceptive counselling and provided MVA 110
when necessary. Clinical procedures for all enrolled women followed medical treatment of 111
incomplete abortion according to WHO and the International Federation of Gynaecology and 112
Obstetrics (FIGO) guidelines.22,23 113
114
Women admitted with signs of incomplete abortion was screened based on self reported Last 115
Menstrual Period (LMP) and symptoms. Clinical signs of incomplete abortion included lower 116
abdominal pain or cramping, vaginal bleeding, an open cervical os, a history of amenorrhoea 117
and sometimes partial expulsion of products of conception. Eligible women who consented to 118
participation were randomly allocated to a midwife (intervention) or a physician (standard 119
care/control) for diagnosis and treatment. The clinical assessment included: (i) history taking, 120
including LMP, obstetric and gynaecological history, and contraceptive history; (ii) general 121
physical examination; (iii) pelvic examination including any signs of genital infections, 122
cervical status, bleeding, and size of the uterus. Ultrasound was not systematically used. Each 123
participant was given one single dose of 600mcg misoprostol orally.22 In addition they were 124
offered analgesics in the form of Ibuprofen or Paracetamol, and oral antibiotics according to 125
national guidelines for PAC. For monitoring purposes, participants were advised to stay at the 126
clinic for four hours after swallowing the misoprostol tablets. Before discharge all women 127
were offered contraceptive counselling and provided with a follow up date. Participants were 128
given detailed information regarding bleeding, pain expected, and abnormal symptoms 129
following treatment (fever and foul smelling vaginal discharge) and were informed about the 130
importance of seeking care if such symptoms occurred. A separate protocol was used to 131
record adverse events. The women were offered reimbursement for travel as an incentive to 132
come for FU within the trial. 133
134
Outcomes
135
Primary and secondary outcomes were measured by research assistants at a follow up visit,
136
14 days + 2 weeks (within 14 to 28 days), after the initial visit. The primary outcome was 137
complete abortion not requiring surgical intervention within 14 to 28 days of the initial 138
treatment. The clinical assessments of the primary outcome were; (i) physical examination 139
(pulse, blood pressure and temperature); (ii) pelvic examination including cervical status and 140
bimanual examination of uterine size. Secondary outcomes included: (i) bleeding; (ii) pain; 141
(iii) and un-scheduled visits. Measurements of secondary outcomes were: (i) the intensity of 142
bleeding in relation to normal menstruation using a symptom diary card; (ii), and pain 143
5
experienced following treatment, using a visual analogue scale (VAS). 144
145
Sample size
146
The sample size was calculated with the objective of showing two-sided equivalence 147
assuming that the rate of incomplete abortions could be four per cent and would apply to both 148
types of providers. A pre defined acceptable difference in completion rate between the two 149
providers ranged between -4% to +4%. To establish equivalence, with a power of 80% and 150
two-sided 95% CI, the sample size becomes 452 per arm. Compensating for 10% loss to 151
follow up gave a total sample size of 994. 152
153
Randomisation
154
The randomisation was 1:1 conducted in blocks of 12 and was stratified for study site. A 155
computer random number generator was used to generate a list of codes from 1 to 994 and 156
each code was linked to one of the two study groups. Sequentially numbered, opaque, sealed 157
envelopes, each containing a random allocation, were prepared at the coordinating centre, and 158
later opened in consecutive order by the research assistants after obtaining written consent. 159
Data management was organised locally at the coordinating centre at Mulago Hospital. Study 160
protocols were collected and data entered continuously throughout data collection. The study 161
coordinator checked protocols for accuracy, corrected protocols after discussion with research 162
assistants, and conducted continuous process evaluation. The study coordinator also provided 163
support and guidance to the providers throughout the study period. 164
165
Statistical analyses
166
Background characteristics and categorical outcomes were presented using descriptive 167
statistics. The intergroup comparison was done using a generalized linear mixed-effects 168
model with group as a fixed effect and health care facility as a random effect. The confidence 169
interval for the risk difference was estimated using 1000 bootstrap simulations. In addition, 170
the adjusted risk difference was estimated where the model was extended with the following 171
fixed effects: age (<25 vs. 25+), marital status (single vs. married/cohabiting), education 172
(none/primary vs. secondary/tertiary), number of pregnancies (1 vs. >1), and parity (0-para vs. 173
multipara). The adjusted risk difference was estimated as the predicted risk difference at the 174
average of all included covariates. Equivalence between the two study groups can be stated if 175
the 95% CI of the risk difference lies completely within the pre-determined limits of 176
equivalence (-4% to 4%). The intention-to-treat (ITT) population was defined as all 177
6
randomised patients with data concerning the primary outcome, excluding patients who had 178
withdrawn consent. The per-protocol (PP) population is a subset of the ITT population 179
excluding women with major protocol violations (e.g. cross-overs). Analysis of the primary 180
outcome was made by PP analysis using generalized linear mixed-effects model to estimate 181
the risk difference between the groups. P-values ≤0·05 were considered statistically 182
significant. Data was entered in EpiData 3.1 and analysed using Stata version 13. Statistical 183
analysis of the primary outcome was analysed using the Ime4 package in R version 3·0·1. 184
Safety data were viewed descriptively without any formal statistical testing and there were no 185
changes to the study design following the pilot phase. We have uploaded the study protocol to 186
the Karolinska Institutet website: http://ki.se/en/people/krigem. The trial is registered at 187
ClinicalTrials.org NCT 01844024. 188
189
Role of the funding source 190
Funder of the study was UNDPA/UNFPA/WHO/World bank special programme of research, 191
development and research training in human reproduction, WHO, Geneva, the Swedish 192
research council (521-2009-2605), Karolinska Institutet and Dalarna University. The authors 193
designed this investigator-initiated trial. Funders of the study were not involved in the design, 194
data collection, analysis or interpretation of the results. All authors have had full access to the 195
data collected in the study. The corresponding author had final responsibility for the decision 196
to submit the manuscript for publication. 197
198
Results 199
A total of 1108 women with symptoms of incomplete abortion were assessed for eligibility, of 200
whom 89 were ineligible and nine women declined participation. A total of 1010 women were 201
randomly assigned to the intervention (506 to midwife and 504 to physician). Eleven women 202
(four in the midwife group and seven in the physician group) were excluded. In the midwife 203
group, one woman had missing values for endpoint analysis, two women had a gestational age 204
of >12 weeks and one was excluded because of withdrawn consent. In the physician group, 205
two women did not receive misoprostol after clinical examination thus the intervention was 206
discontinued, one woman was not eligible (septic and gestational age unknown), and two had 207
missing values for the endpoint analysis. In addition, two women allocated to physicians were 208
classified as protocol violations (e.g. cross-overs) as they were assessed by midwives when 209
there was no physician available. One of the crossovers returned with complete abortion and 210
one was lost to follow up. After exclusion of these 11 cases the total number of women that 211
7
received the intervention was 502 in the midwife group and 497 in the physician group. Table 212
1 shows the socio-demographic background and reproductive history of participants, 213
including women lost to follow up. The baseline characteristics among participants were 214
balanced with no major difference between the two groups with regard to socio-demographic 215
background or reproductive history. The mean duration of gestational age based on clinical 216
examination was 8·8 weeks (range 1-12 weeks) (Table 1). 217
218
One crossover was not considered enough to affect the results in a separate ITT analysis, as 219
the PP population would be almost identical as the ITT population. Total loss to follow up 220
was 44 (4·5%) of whom 30 (5·9%) were in the midwife group and 14 (2·8%) in the physician 221
group. Thereby, a total of 955 women, 472 women in the midwife group and 483 women in 222
the physician group, were included in the PP analysis (Figure 1). 223
224
Participating providers had similar background characteristics although physicians had longer 225
work experience in PAC (Table 2). The women lost to follow up had a lower gestational age 226
based on clinical exam compared with women who came back for follow up. Otherwise there 227
was no difference in socio-demographics or reproductive history in the group of women lost 228
to follow up compared with those who were not (Suppl Table 1). 229
230
Primary outcome
231
The overall proportion of complete abortion was 96.2%. The proportion of women with 232
complete abortion among midwives and physicians was 95·8% (n=452) and 96·7% (n=467) 233
respectively. The model based risk difference for midwife versus physician group was -0·79% 234
(95% CI- -2·90 to 1·35) (Table 3). The total number of incomplete abortions was 20 (4·2%) 235
for midwives and 16 (3·3%) for physicians. All women with incomplete abortion (n=36) 236
required surgical treatment and were evacuated with MVA after completing the follow up 237
assessment. No serious adverse events were recorded. 238
239
Secondary outcomes
240
The majority (83·8%) reported bleeding less than or similar to normal menstrual bleeding 241
following treatment. Mean number of days bleeding was 5 (range 1-16). Women evaluated 242
pain following treatment, using VAS, with a mean score of 3·6 (range 0-10). The number of 243
women reporting unscheduled visits was 30 (6·4%) in the midwife group and 18 (3.7%) in the 244
physician group. Vaginal bleeding and abdominal pain were reported as reasons for the 245
8
unscheduled visits (Table 4). There were no significant differences in secondary outcomes 246
between the groups. Reported side effects following treatment were nausea, vomiting, 247
abdominal pain, chills and fever and were similar in both groups (Suppl Table 2). 248
249
Discussion 250
Diagnosis and treatment of first trimester incomplete abortion with misoprostol by midwives 251
was equally effective and safe as when provided by physicians. Women with signs of 252
incomplete abortion, seeking care at district level in Uganda who were diagnosed and treated 253
by a midwife did not experience a higher complication rate compared with women treated by 254
a physician. Our findings will be useful to scale up women’s access to safe PAC in low 255
resource settings and thus contribute to a decrease in maternal mortality and morbidity. 256
257
Midwifery led interventions have been shown to be associated with efficient use of resources 258
and to have a positive impact on health outcomes. Implementation of effective and sustainable 259
models of care, such as optimizing the midwifery role, is central in order to reduce maternal 260
mortality and morbidity.24 Other RCT studies from settings with liberal abortion laws show 261
that provision of pregnancy termination using MVA25 and medical abortion26,27 is equally safe 262
and effective when provided by midwives as when provided by physicians. These results are 263
further confirmed in a systematic review, although the evidence is limited.28 264
265
Our study is, to our knowledge, the first RCT conducted to evaluate the involvement of 266
midwives diagnosing and treating women for incomplete abortion with misoprostol at district 267
level, in a low resource setting. The overall complete abortion rate in our study was 96.2% 268
similar to results in previous studies (range 94.4-99).8,10,28 Decentralization of misoprostol use 269
in PAC has been found successful in a similar setting, where ultrasound is not routinely used 270
and physicians are scarce,29 consistent with our findings. In Uganda, the majority of 271
physicians work in the central urban region serving less than one third of the population. As a 272
response, task shifting is already taking place but often without clear policies, planning, 273
monitoring and evaluation.18 A recent qualitative study at district level in Uganda reports 274
midwives as the main providers, including complicated cases due to the absence of 275
physicians. However, both physicians and midwives revealed a lack of adequate skills and 276
expressed a need for in-service training. Misoprostol was also reported to be rarely used in 277
PAC due to limited availability and treatment guidelines at the health facilities.15 Results from 278
this study provide scientific evidence that support task shifting in PAC and may be used in the 279
9
development and implementation of policies and standard care guidelines in Uganda, and 280
other similar low resource settings. 281
282
The Ugandan health system costs in treating women with incomplete abortion, often caused 283
by unsafe abortion, is substantial, and so is the cost for the individual women and their 284
families.30 A task shift to midwives in providing treatment of incomplete abortion with 285
misoprostol at district level will increase women’s access to safe PAC where it is needed 286
most.31,32 It has the potential to reduce the workload on the already overstretched health care 287
providers, save resources, and reduce the cost of unsafe abortion, at an individual and societal 288
level. In order to achieve these changes, an increase in misoprostol use in PAC is essential 289
and supplies at district level must be ensured. The authors suggest that PAC is incorporated in 290
basic midwifery education curricula in order to safeguard a future workforce with PAC skills. 291
Safe and effective task shifting between midwives and physicians also requires in-service 292
training in PAC for both professional cadres, based on updated evidence based guidelines. 293
The strengths of our study are its randomised design, low rate of loss to follow up, and 294
adequate power for detection of the primary outcome. One limitation is that several facilities 295
were involved and the standard of care might have differed between the facilities. However, 296
this is also a strength of the trial since the intervention was successful irrespective of the local 297
clinical standard of care. The involved providers’ clinical experiences of PAC ranged between 298
0-28,5 years (median 3·7) and differed between midwives and physicians. However, the 299
analyses between facilities showed no cluster effect on equivalence result. The larger loss to 300
follow up in the midwife group compared with the physician group is a limitation. Physicians 301
have a higher status than midwives in Uganda. Hence, this result could be interpreted as a 302
consequence of women feeling more inclined to return for follow up after being treated by a 303
physician. It could also be a consequence of individual differences in providers’ ability to 304
explain the importance of the follow up visit. The lost to follow up analysis did not show any 305
statistically significant differences except for gestational age. Women who were lost to follow 306
up had a significantly lower gestational age compared with women who returned for follow 307 up (Suppl Table 1). 308 309 310 311 312 313 314
10 Panel: Research in context
315 316
Systematic Review 317
Our literature review was conducted primarily in PubMed and included keywords: ”post 318
abortion care”, ”incomplete abortion”, ”misoprostol”, ”midwives”, “non-physicians” and 319
”midlevel providers”. We have reviewed scientific publications aiming to assess midwives’ 320
involvement in treating incomplete abortion with misoprostol in low resource settings. 321
Existing evidence in relation to the topic is based on one observational study describing 322
misoprostol use in post abortion care provided by nurse-midwives. However, no randomised 323
controlled trial has evaluated the effectiveness and safety of midwives diagnosing and treating 324
incomplete abortion with misoprostol. 325
326
Interpretation 327
Findings from our study show that midwives with a standardized training programme can 328
independently diagnose incomplete abortion and successfully treat women with misoprostol at 329
district level in a low resource setting. Our trial builds on previous scientific evidence 330
showing that medical treatment with misoprostol and surgical treatment with Manual Vacuum 331
Aspiration (MVA) for treatment of incomplete abortion is equally effective. Non-physician’s 332
involvement in medical abortion has been evaluated to be safe in low resource settings. Our 333
findings further support existing evidence concerning the use of misoprostol and fill an 334
important knowledge gap regarding midwives’ competencies as main providers of post 335
abortion care using misoprostol at district level. 336
337
Conclusion 338
Midwives are able to safely diagnose and treat first trimester incomplete abortion with 339
misoprostol at district level in Uganda. The findings are of relevance to low resource settings 340
but could also have implications for other settings where access to PAC is limited due to 341
shortages in human resources. Scaling up midwives’ involvement in PAC at district level 342
would increase women’s access to safe and effective treatment of incomplete abortion and 343
thereby avoid severe complications that may lead to maternal mortality and morbidity. 344
345
Contributors 346
MKA developed the protocol and study design, directed implementation of the study, the data 347
analyses and was lead author together with AC. AC participated in the implementation of the 348
11
study, in data collection and analysis of the study. SA was the coordinator for the facilities 349
included in the study and participated in data collection, analysis and report writing. NMT 350
was the statistician of the study and coordinated the data management, participated in analysis 351
and report writing. EF took part in the study design, participated in analysis and report 352
writing. JB was the local principal investigator and responsible for the overall supervision of 353
the trial and participated in the implementation of the study, data analysis and report writing. 354
KGD was the principal investigator, contributed to the study conception and was responsible 355
for the overall supervision of the trial and participated in the study design, implementation of 356
the study, data analysis and report writing. All involved authors have had access to the data, 357
commented on the manuscript drafts and approved the final version submitted. 358
359
Conflict of interests 360
We declare that we have no personal or financial conflicts of interests. 361
362
Acknowledgements 363
The women who participated are acknowledged for sharing their experiences and for coming 364
back for follow up. We thank all the providers involved in the study who stayed committed 365
and motivated all the way. 366
367
References 368
1 Ganatra B, Tuncalp Ö, Johnston HB, Johnson BR Jr, Gülmezoglu AM, 369
Temmerman M. From concept to measurement: operationalizing WHO's 370
definition of unsafe abortion. Bull World Health Org 2014; 92(3): 155. 371
2 Shah IH, Åhman E, Ortayli N. Access to safe abortion: progress and challenges 372
since the 1994 International Conference on Population and Development 373
(ICPD). Contraception 2014. doi: 10.1016/j.contraception.2014.04.004. 374
3 WHO, UNICEF, UNFPA, The World Bank, UNPD. Trends in maternal 375
mortality: 1990 to 2013. Estimates by WHO, UNICEF, UNFPA, The World 376
Bank and the United Nations Population Division. Geneva: World Health 377
Organization, 2013. 378
4 Rasch V. Unsafe abortion and postabortion care - an overview. Acta Obstet 379
Gynecol Scand 2011: 90(7): 692-700.
380
5 Escandon I, Turner K. Indicators for the Essential Elements of Postabortion 381
care. PAC Consortium, 2006. 382
12
6 Weeks A. Clarifying the role of misoprostol in obstetrics. Afr Health Sci 2001; 383
1(2): 52-4. 384
7 Weeks A, Alia G, Blum J, Winikoff B, Ekwaru P, Durocher J, Mirembe F. A 385
randomized trial of misoprostol compared with manual vacuum aspiration for 386
incomplete abortion. Obstet Gynecol 2005; 106(3): 540-7. 387
8 Gemzell-Danielsson K, Fiala C, Weeks A. Misoprostol: first-line therapy for 388
incomplete miscarriage in the developing world. BJOG 2007; 114(11): 1337-39. 389
9 Neilson JP, Gyte GM, Hickey M, Vazquez JC, Dou L. Medical treatments for 390
incomplete miscarriage. The Cochrane Database Syst Rev 2013; 3: CD007223. 391
10 Shochet T, Diop A, Gaye A et al. Sublingual misoprostol versus standard 392
surgical care for treatment of incomplete abortion in five sub-Saharan African 393
countries. BMC Pregnancy Childbirth 2012; 12: 127. 394
11 Dabash R, Ramadan MC, Darwish E, Hassanein N, Blum J, Wikinoff B. A 395
randomized controlled trial of 400-µg sublingual misoprostol versus manual 396
vacuum aspiration for treatment of incomplete abortion in two Egyptian 397
hospitals. Int J Gynaecol Obstet 2010; 111(2): 131-5. 398
12 Dao B, Blum J, Thieba B et al. Is misoprostol safe, effective and acceptable 399
alternative to manual vacuum aspiration for postabortion care? Results from a 400
randomised trial in Burkina Faso, West Africa. BJOG 2007; 114(11): 1368-75. 401
13 Yarnall J, Swica Y, Winikoff B. Non-physician clinicians can safely provide 402
first trimester medical abortion. Reprod Health Matters 2009; 17(33): 61-9. 403
14 Gessessew A, Barnabas GA, Prata N, Weidert K. Task shifting and sharing in 404
Tigray, Ethiopia, to achieve comprehensive emergency obstetric care. Int J 405
Gynaecol Obstet 2011; 113(1): 28-31.
406
15 Paul M, Gemzell-Danielsson K, Kiggundu C, Namugenyi R, Klingberg-Allvin 407
M. Barriers and facilitators in the provision of post-abortion care at district level 408
in central Uganda - a qualitative study focusing on task sharing between 409
physicians and midwives. BMC Health Serv Res 2014; 14: 28. 410
16 Fulton BD, Scheffler RM, Sparkes SP, Auh EY, Vujicic M, Soucat A. Health 411
workforce skill mix and task shifting in low income countries: a review of recent 412
evidence. Hum Resour Health 2011; 9(1): 1. 413
13
17 Pereira C, Mbaruku G, Nzabuhakwa C, Bergstrom S, McCord C. Emergency 414
obstetric surgery by non-physician clinicians in Tanzania. Int Jour Gynaecol 415
Obstet 2011; 114(2): 180-3.
416
18 Nabudere H, Asiimwe D, Mijumbi R. Task shifting in maternal and child health 417
care: an evidence brief for Uganda. Int J Technol Assess Health Care 2011; 418
27(2): 173-9. 419
19 Piaggio G, Elbourne DR, Pocock SJ, Evans SJ, Altman DG. Reporting of 420
noninferiority and equivalence randomized trials: extension of the CONSORT 421
2010 statement. JAMA 2012; 308(24): 2594-2604. 422
20 “Woman-centered, comprehensive abortion care: Trainer’s manual (first ed.)” 423
K. L. Turner & A. Huber (Eds.), Chapel Hill, NC: Ipas. (2004) 424
21 Medical abortion training guide (second ed.) K. L. Turner (Ed.), Chapel Hill, 425
NC: Ipas. (2010). 426
22 Blum J, Winikoff B, Gemzell-Danielsson K, Ho PC, Schiavon R, Weeks A. 427
Treatment of incomplete abortion and miscarriage with misoprostol. Int J 428
Gynaecol Obstet 2007; 99: 186–189.
429
23 WHO. Safe Abortion: Technical and Policy Guidance for Health Systems. 430
Geneva: World Health Organization, 2012. 431
24 Renfrew MJ, McFadden A, Bastos MH et al. Midwifery and quality care: 432
findings from a new evidence-informed framework for maternal and newborn 433
care. Lancet 2014. doi: 10.1016/S0140-6736(14)60789-3 434
25 Warriner IK, Meirik O, Hoffman M et al. Rates of complication in first-trimester 435
manual vacuum aspiration abortion done by doctors and mid-level providers in 436
South Africa and Vietnam: a randomised controlled equivalence trial. Lancet 437
2006; 368(9551): 1965-72. 438
26 Warriner IK, Wang D, Huong NT et al. Can midlevel health-care providers 439
administer early medical abortion as safely and effectively as doctors? A 440
randomised controlled equivalence trial in Nepal. Lancet 2011; 377(9772): 441
1155-61. 442
27 Kopp Kallner H, Gomperts R, Salomonsson E, Johansson M, Marions L, 443
Gemzell-Danielsson K. The efficacy, safety and acceptability of medical 444
termination of pregnancy provided by standard care by doctors or by nurse-445
midwives: a randomised controlled equivalence trial. BJOG 2014. doi: 446
10.1111/1471-0528.12982. 447
14
28 Renner RM, Brahmi D, Kapp N. Who can provide effective and safe termination 448
of pregnancy care? A systematic review*. BJOG 2013; 120(1): 23-31. 449
29 Gaye A, Diop A, Shochet T, Winikoff B. Decentralizing postabortion care in 450
Senegal with misoprostol for incomplete abortion. Int J Gynaecol Obstet 2014; 451
126(3): 223-6. 452
30 Vlassoff M, Mugisha F, Sundaram A et al. The health system cost of post-453
abortion care in Uganda. Health Policy Plan 2014; 29(1): 56-66. 454
31 Haddad LB, Nour NM. Unsafe abortion: unnecessary maternal mortality. Rev 455
Obstet Gynecol 2009; 2(2): 122-6.
456
32 Berer M. Provision of abortion by mid-level providers: international policy, 457
practice and perspectives. Bull World Health Organ 2009; 87(1): 58-63. 458