Articles

Comparison of treatment of incomplete abortion with

misoprostol by physicians and midwives at district level in
Uganda: a randomised controlled equivalence trial
Marie Klingberg-Allvin*, Amanda Cleeve*, Susan Atuhairwe, Nazarius Mbona Tumwesigye, Elisabeth Faxelid, Josaphat Byamugisha†,
Kristina Gemzell-Danielsson†

Summary
Background Misoprostol is established for the treatment of incomplete abortion but has not been systematically Published Online
assessed when provided by midwives at district level in a low-resource setting. We investigated the effectiveness and March 27, 2015
http://dx.doi.org/10.1016/
safety of midwives diagnosing and treating incomplete abortion with misoprostol, compared with physicians. S0140-6736(14)61935-8
See Online/Comment
Methods We did a multicentre randomised controlled equivalence trial at district level at six facilities in Uganda. http://dx.doi.org/10.1016/
Eligibility criteria were women with signs of incomplete abortion. We randomly allocated women with first-trimester S0140-6736(14)62287-X
incomplete abortion to clinical assessment and treatment with misoprostol either by a physician or a midwife. The *Joint first authors
randomisation (1:1) was done in blocks of 12 and was stratified for study site. Primary outcome was complete †Joint last authors
abortion not needing surgical intervention within 14–28 days after initial treatment. The study was not masked. School of Education, Health
Analysis of the primary outcome was done on the per-protocol population with a generalised linear-mixed effects and Social Studies, Dalarna
model. The predefined equivalence range was –4% to 4%. The trial was registered at ClinicalTrials.gov, number University, Falun, Sweden
(M Klingberg-Allvin PhD);
NCT01844024. Department of Women’s and
Children’s Health
Findings From April 30, 2013, to July 21, 2014, 1108 women were assessed for eligibility. 1010 women were randomly (M Klingberg-Allvin,
assigned to each group (506 to midwife group and 504 to physician group). 955 women (472 in the midwife group A Cleeve MSc,
Prof K Gemzell-Danielsson PhD),
and 483 in the physician group) were included in the per-protocol analysis. 452 (95·8%) of women in the midwife and Global Health (IHCAR),
group had complete abortion and 467 (96·7%) in the physician group. The model-based risk difference for midwife Department of Public Health
versus physician group was –0·8% (95% CI –2·9 to 1·4), falling within the predefined equivalence range (–4% to Sciences (Prof E Faxelid PhD),
4%). The overall proportion of women with incomplete abortion was 3·8% (36/955), similarly distributed between Karolinska Institutet,
Stockholm, Sweden;
the two groups (4·2% [20/472] in the midwife group, 3·3% [16/483] in the physician group). No serious adverse Department of Obstetrics and
events were recorded. Gynaecology, Mulago Hospital,
Kampala, Uganda
Interpretation Diagnosis and treatment of incomplete abortion with misoprostol by midwives is equally safe and (S Atuhairwe MSc,
J Byamugisha PhD); Department
effective as when provided by physicians, in a low-resource setting. Scaling up midwives’ involvement in treatment of of Epidemiology and
incomplete abortion with misoprostol at district level would increase access to safe post-abortion care. Biostatistics, School of Public
Health (N M Tumwesigye PhD),
Makerere University College of
Funding The Swedish Research Council, Karolinska Institutet, and Dalarna University.
Health Sciences, Kampala,
Uganda (S Atuhairwe,
Introduction low-resource settings, misoprostol in post-abortion care is J Byamugisha); and Karolinska
Unsafe abortion contributes substantially to the global a simplified, cost-effective, and resource-saving alternative University Hospital,
Stockholm, Sweden
burden of maternal mortality and morbidity.1 Most unsafe to surgical interventions.10
(Prof K Gemzell-Danielsson)
abortions occur in low-income countries where induced The scarcity of physicians in many low-income
Correspondence to:
abortion is restricted and the unmet need for contraception countries restricts women’s access to post-abortion Dr Marie Klingberg-Allvin,
is high.2 Sub-Saharan Africa has the highest rates of care.13 In Africa, the shortage of trained health-care Department of Women’s and
unsafe abortion and global maternal mortality.3 providers is greatest in rural and remote areas where Children’s Health, Karolinska
Institutet, WHO-centre, C1:05
Emergency treatment of complications from unsafe maternal mortality and morbidity is highest.14 In
Karolinska University Hospital,
abortion and spontaneous abortion, post-abortion care, Uganda, where unsafe abortion is a serious public health SE-171 76 Stockholm, Sweden
is an effective intervention to lower maternal mortality.4 issue, physicians are scarce and few midwives receive mkl@du.se
The post-abortion care model consists of emergency post-abortion care training,15 restricting access to care.
treatment of abortion-related complications and post- Task shifting is a process of delegating tasks, when
abortion contraceptive counselling and provision.5 The appropriate, to less specialised health-care providers. It
prostaglandin E1 analogue misoprostol has been suggested is a way of increasing access and productivity,16 and to
as an effective drug for the treatment of incomplete contribute to the building of cost-effective and equitable
abortion.6–8 International studies have compared and health-care services.17,18 Optimisation of the use of
found no significant difference in effectiveness between midwives would be a pragmatic response to the shortage
treatment of incomplete abortion with misoprostol, and of physicians at district level in Uganda and a strategy to
surgical treatment with manual vacuum aspiration.9–12 In decrease maternal mortality.18

www.thelancet.com Published online March 27, 2015 http://dx.doi.org/10.1016/S0140-6736(14)61935-8 1

 Articles

Despite the proven effectiveness and safety of of the study. The procedure and the instruments were
misoprostol treatment for incomplete abortion and the pilot tested at Mulago Hospital, the national referral
known advantages of task shifting, the involvement of hospital in Kampala,Uganda. Only minor revision was
midwives in diagnosing and treating incomplete abortion made to the instruments before recruitment began. We
with misoprostol has not been systematically assessed in deemed the first month to be the run-in period in which
any low-resource setting. In this trial, we aimed to assess the facilities were given time to set the routine for
whether diagnosis and treatment of first-trimester screening, enrolment, and follow-up. This was supervised
incomplete abortion with misoprostol provided by a and supported closely by the study coordinator. An
midwife is as equally effective and safe as that provided interim report was also sent to WHO after the run-in
by a physician. phase. We did not make any changes to the data collection
method or the recruitment process and thus we included
Methods the run-in period in the analysis. The study was developed
Study design and participants and coordinated by researchers at the WHO research
The study was a multicentre randomised controlled centre at Karolinska Institutet, Stockholm, Sweden, and
equivalence trial done at district level in six health-care Mulago Hospital and Makerere University, Kampala,
facilities in six districts in rural, peri-urban, and urban Uganda. The scientific and ethical review group at the
settings in central Uganda. The study protocol and trial Reproductive Health and Research Department, WHO,
followed the CONSORT guidelines for non-inferiority Geneva approved the study. We obtained further ethical
and equivalence randomised trials.19 Inclusion criteria approval from the research ethics committee at Makerere
were women with signs of incomplete abortion (ie, University (diary number 2012-129), Uganda National
bleeding and contractions during pregnancy, open Council for Science and Technology (diary number HS
cervical os, and partial expulsion). Exclusion criteria were 1314), and the Swedish regional ethical review board at
complete abortion, known allergy to misoprostol, a Karolinska Institutet (diary number 2013/2;9). All
uterine size of more than 12 weeks of gestation, suspected patients gave written informed consent.
ectopic pregnancy, unstable haemodynamic status and
shock, signs of pelvic infection, or sepsis. The health-care Randomisation and masking
facilities selected for inclusion were equipped to provide Eligible women who consented to participation were
basic and emergency obstetric services and were randomly allocated to a midwife (intervention) or a
estimated to have sufficient staff and caseload to be part physician (standard care [control]) for diagnosis and
treatment. The randomisation (1:1) was done in blocks of
12 and was stratified for study site. We used a computer
1108 assessed for eligibility random number generator to generate a list of codes
from 1 to 994 and each code was linked to one of the
98 excluded
89 ineligible two study groups. Sequentially numbered, opaque,
9 declined to participate sealed envelopes, each containing a random allocation,
1010 randomised were prepared at the coordinating centre, and later
opened in consecutive order by the research assistants
after obtaining written consent. Data management was
organised locally at the coordinating centre at Mulago
506 allocated to midwife group 504 allocated to physician group Hospital. Study protocols were collected and data entered
continuously throughout data collection. The study
2 discontinued (did not coordinator checked protocols for accuracy, corrected
receive misoprostol) protocols after discussion with research assistants, and
did continuous process evaluation. The study coordinator
4 excluded 5 excluded also provided support and guidance to the providers
1 missing value for endpoint 2 missing value for endpoint throughout the study period. The study was not masked
analysis analysis
2 discontinued intervention 1 not eligible (septic and to the study participants or providers.
because gestational age gestational age unknown)
>12 weeks 2 protocol violations
1 withdrawn consent
Procedures
Eligible providers for participation were physicians and
midwives involved in post-abortion care at the different
30 lost to follow-up 14 lost to follow-up facilities. The health-care providers were trained according
to a standardised post-abortion care training module.20,21
472 included in analysis of 483 included in analysis of The 5-day training programme focused on diagnosing
primary outcome primary outcome incomplete abortion, treatment with misoprostol and
manual vacuum aspiration, and contraceptive methods
Figure: Trial profile and counselling. Some midwives at each facility were

2 www.thelancet.com Published online March 27, 2015 http://dx.doi.org/10.1016/S0140-6736(14)61935-8

 Articles

trained to be research assistants in the study and were difference in completion rate between the two providers
responsible for eligibility screening and enrolment of ranged between –4% and +4%. To establish equivalence,
participants. The midwives did clinical assessments and with a power of 80% and two-sided 95% CI the sample
were responsible for the follow-up visits including size becomes 452 per group. Compensation for 10% loss
assessed abortion status at follow-up, gave contraceptive to follow-up gave a total sample size of 994.
counselling, and provided manual vacuum aspiration
when necessary. Clinical procedures for all enrolled Midwife Physician Total
women followed medical treatment of incomplete Age (years)
abortion according to WHO and the International
n 499 496 995
Federation of Gynaecology and Obstetrics guidelines.22,23
Mean (SD) 26·1 (6·5) 26·5 (6·5) 26·3 (6·5)
Women admitted with signs of incomplete abortion
16–19 84 (17%) 61 (12%) 145 (15%)
were screened based on self-reported last menstrual period
20–24 154 (31%) 170 (34%) 324 (33%)
and symptoms. Eligible women who consented to
25–34 196 (39%) 196 (40%) 392 (39%)
participation were randomly allocated to the midwife or
35–49 65 (13%) 69 (14%) 134 (13%)
physician group. The clinical assessment included: history
Education
taking, including last menstrual period, obstetric and
n 502 497 999
gynaecological history, and contraceptive history; general
None 33 (7%) 49 (10%) 82 (8%)
physical examination; and pelvic examination including
Primary school 227 (45%) 253 (51%) 480 (48%)
any signs of genital infections, cervical status, bleeding,
and size of the uterus. Ultrasound was not systematically Secondary school 198 (39%) 156 (31%) 354 (36%)

used. Each participant was given one single dose of 600 μg Tertiary school 44 (9%) 39 (8%) 83 (8%)

misoprostol orally.22 Additionally, they were offered Marital status

analgesics (ibuprofen or paracetamol) and oral antibiotics n 502 497 999
according to national guidelines for post-abortion care. For Married or cohabiting 373 (74%) 386 (78%) 759 (76%)
monitoring purposes, participants were advised to stay at Single 114 (23%) 101 (20%) 215 (22%)
the clinic for 4 h after swallowing the misoprostol tablets. Divorced or widowed 15 (3%) 10 (2%) 25 (2%)
Before discharge all women were offered contraceptive Occupation
counselling and provided with a follow-up date. n 500 496 996
Participants received detailed information about bleeding, Unemployed 245 (49%) 253 (51%) 498 (50%)
pain expected, abnormal symptoms after treatment (fever Formal employment 64 (13%) 52 (10%) 116 (12%)
and foul smelling vaginal discharge), and were informed Self employed 175 (35%) 177 (36%) 352 (35%)
about the importance of seeking care if such symptoms Student 16 (3%) 14 (3%) 30 (3%)
occurred. A separate protocol was used to record adverse Religion
events. The women were offered reimbursement for travel n 502 496 998
as an incentive to come for follow-up within the trial. Catholic 232 (46%) 213 (43%) 445 (45%)
Protestant 146 (29%) 160 (32%) 306 (31%)
Outcomes Muslim 93 (19%) 107 (22%) 200 (20%)
The primary outcome was a complete abortion not Born again 31 (6%) 16 (3%) 47 (5%)
needing surgical intervention within 14–28 days of the Gestational age based on clinical examination (weeks)
initial treatment. Clinical assessments of the primary n 499 494 993
outcome were physical examination (pulse, blood Mean (SD) 8·7 (2·3) 8·8 (2·1) 8·8 (2·2)
pressure, and temperature); and pelvic examination Range 1–12 4–12 1–12
including cervical status and bimanual examination of Number of pregnancies
uterine size. Secondary outcomes included bleeding, n 501 497 998
pain, and unscheduled visits. Measurements of Mean (SD) 3·4 (2·4) 3·5 (2·2) 3·4 (2·3)
secondary outcomes were the intensity of bleeding in 1 121 (24%) 99 (20%) 220 (22%)
relation to normal menstruation using a symptom diary 2–5 306 (61%) 308 (62%) 614 (62%)
card, and pain experienced after treatment, using a 6–10 64 (13%) 89 (18%) 153 (15%)
visual analogue scale. Research assistants measured
11–20 10 (2%) 1 (0%) 11 (1%)
primary and secondary outcomes at a follow-up visit
Parity
within 14–28 days after the initial visit.
n 501 497 998
0 129 (26%) 112 (23%) 241 (24%)
Statistical analyses
≥1 372 (74%) 385 (77%) 757 (76%)
We calculated the sample size with the objective of
showing two-sided equivalence assuming that the rate of Data are n (%) unless otherwise stated. *Includes women lost to follow-up.
incomplete abortions could be 4% and would apply to
Table 1: Sociodemographic background and reproductive history of participants* by provider
both types of providers.9 A pre-defined acceptable

www.thelancet.com Published online March 27, 2015 http://dx.doi.org/10.1016/S0140-6736(14)61935-8 3

 Articles
For the study protocol see
http://ki.se/en/people/krigem
Midwife
Randomised and
received intervention
Per protocol
Complete abortion
*Adjusted for age (<25 vs ≥25 years), marital status (single vs married or cohabiting), education (none or primary vs
secondary or tertiary), number of pregnancies (1 vs >1), and parity (para 0 vs ≥1).
Table 3: Outcome of treatment of incomplete abortion by provider
4 www.thelancet.com Published online March 27, 2015 http://dx.doi.org/10.1016/S0140-6736(14)61935-8
Background characteristics and categorical outcomes less were deemed significant. We entered data in EpiData
were presented with descriptive statistics. We did 3.1 and analysed with Stata version 13. We did the
the intergroup comparison using a generalised linear statistical analysis of the primary outcome with the Ime4
mixed-effects model with group as a fixed effect and package in R version 3.0.1. Safety data were viewed
health-care facility as a random effect. We estimated descriptively without any formal statistical testing and
the CI for the risk difference using 1000 bootstrap there were no changes to the study protocol after the run-
simulations. Additionally, we estimated the adjusted risk in period. The trial is registered at ClinicalTrials.gov,
difference in which the model was extended with the number NCT01844024.
following fixed effects: age (<25 vs ≥25 years), marital
status (single vs married or cohabiting), education (none Role of the funding source
or primary vs secondary or tertiary), number of The authors designed this investigator-initiated trial.
pregnancies (1 vs >1), and parity (0 vs ≥1). We estimated The funders were not involved in the design, data
the adjusted risk difference as the predicted risk collection, analysis, or interpretation of the results. All
difference at the average of all included covariates. authors had full access to the data. The corresponding
Equivalence between the two study groups can be stated author had final responsibility for the decision to submit
if the 95% CI of the risk difference lies completely within the manuscript for publication.
the pre-determined limits of equivalence (–4% to 4%).
We defined the intention-to-treat population as all Results
randomised patients with data for the primary outcome, From April 30, 2013, to July 21, 2014, we assessed
excluding patients who had withdrawn consent. The 1108 women with symptoms of incomplete abortion for
per-protocol population is a subset of the intention-to- eligibility, of whom 89 were ineligible, and nine women
treat population excluding women with major protocol declined participation. We randomly assigned
violations (eg, crossovers). Analysis of the primary 1010 women to the intervention groups (506 to the
outcome was made by per-protocol analysis with a midwife group and 504 to the physician group).
generalised linear mixed-effects model to estimate the 11 women were excluded. Two women allocated to the
risk difference between the groups. p values of 0·05 or physcian group were classified as protocol violations (ie,
crossovers) because they were assessed by midwives
Midwife (n=29) Physician (n=13) Total (n=42) when there was no physician available. One of the
Age (years)
crossovers returned with complete abortion and one was
Mean (SD) 40·5 (8·2) 42·1 (10·9) 41·0 (9·0)
lost to follow-up. After exclusion of these 11 women,
Median (IQR) 41 (35–45) 44 (31–52) 42 (34–46)
502 women received the intervention in the midwife
group and 497 in the physician group.
Sex
We did not deem one crossover enough to affect the
Female 29 (100%) 1 (8%) 30 (71%)
results in a separate intention-to-treat analysis because
Male 0 12 (92%) 12 (29%)
the per-protocol population would be almost identical to
Years of professional practice
the intention-to-treat population. 44 patients were lost to
Mean (SD) 15·4 (7·8) 14·8 (9·9) 15·2 (8·4)
follow-up. So, a total of 955 women were included in the
Range 2–30 2·5–30 2–30
per-protocol analysis (figure).
Clinical experience in post-abortion care (years)*
The baseline characteristics among participants
Mean (SD) 3·7 (3·7) 8·3 (7·3) 5·1 (5·4)
were balanced with no major difference between the
Range 0–13·5 0·5–28·5 0–28·5
two groups with regard to sociodemographic background
Data are n (%), mean (SD), or median (IQR). *Before study start. or reproductive history (table 1). The mean duration of
gestational age based on clinical examination was
Table 2: Background characteristics of participating providers of
post-abortion care providers
8·8 weeks (range 1–12 weeks; table 1).
Participating providers had similar background
characteristics although physicians had longer work
Physician Risk difference Adjusted difference experience in post-abortion care than did the midwives
(95% CI) (95% CI)* (table 2). The women lost to follow-up had a lower
502 497 ·· ·· gestational age based on clinical examination compared
with women who came back for follow-up. Otherwise
472 483 ·· ·· there was no difference in socio-demographics or
452 (95·8%) 467 (96·7%) –0·8% –0·6% reproductive history in the group of women lost to
(–2·9 to 1·4) (–2·4 to 1·2) follow-up compared with those who were not
(appendix).
The overall proportion of complete abortion was
96·2% (919/955). 452 (95·8%) women in the midwife
group and 467 (96·7%) women in the physician group
 Articles

had complete abortion. The model-based risk difference

Midwife Physician Total
for midwife versus physician group was –0·8% (95% CI
–2·9 to 1·4; table 3). There were 20 (4·2%) incomplete Bleeding since treatment

abortions in the midwife group and 16 (3·3%) in the n 472 481 953
physician group. All women with incomplete abortion Less than normal menstrual bleeding 211 (45%) 217 (45%) 428 (45%)
(n=36) needed surgical treatment and were evacuated Same as normal menstrual bleeding 185 (39%) 186 (39%) 371 (39%)
with manual vacuum aspiration after completion of the Heavier than normal menstrual bleeding 76 (16%) 78 (16%) 154 (16%)
follow-up assessment. No serious adverse events Days bleeding
were recorded. n 472 478 950
Most women reported bleeding less than or the same Mean (SD) 5·3 (2·9) 5·0 (2·7) 5·2 (2·8)
as normal menstrual bleeding after treatment (table 4). Range 1–16 1–14 1–16
Mean number of days bleeding was 5·2 (SD 2·8; range Pain following treatment using visual analogue scale 0–10
1–16). Women assessed pain after treatment with a visual n 472 478 990
analogue scale and had a mean score of 3·6 (SD 1·7; Mean (SD) 3·6 (1·8) 3·5 (1·7) 3·6 (1·7)
range 0–10). 30 (6%) women reported unscheduled visits Range 0–10 0–10 0–10
in the midwife group and 18 (4%) in the physician group. Unscheduled visit
Vaginal bleeding and abdominal pain were reported as n 472 483 955
reasons for the unscheduled visits (table 4). There were Vaginal bleeding or abdominal pain, or 30 (6%) 18 (4%) 48 (5%)
no differences in secondary outcomes between the both
groups. Reported side-effects after treatment were Data are n (%) unless otherwise stated.
nausea, vomiting, abdominal pain, chills, and fever, and
were similar in both groups (appendix). Table 4: Reported secondary outcomes

Discussion
Diagnosis and treatment of first-trimester incomplete
abortion with misoprostol by midwives was equally
effective and safe as when provided by physicians.
Women with signs of incomplete abortion who sought
care at district level in Uganda, and were diagnosed and
treated by a midwife, did not have a higher complication
rate compared with women treated by a physician. Our
findings will be useful to scale-up women’s access to
safe post-abortion care in low-resource settings and
thus contribute to a decrease in maternal mortality and
morbidity (panel).
Midwifery-led interventions have been shown to be
associated with efficient use of resources and to have a
positive effect on health outcomes. Implementation of
effective and sustainable models of care, such as
optimising the midwifery role, is central to reduce
maternal mortality and morbidity.25 Other randomised
controlled trials from settings with liberal abortion laws
show that provision of pregnancy termination by use of
manual vacuum aspiration26 and medical abortion27,28 is
equally safe and effective when provided by midwives
as when provided by physicians. These results are
further confirmed in a systematic review, although
there is little evidence.29
Our study is, to our knowledge, the first randomised
controlled trial done to assess the involvement of
midwives in the diagnosis and treatment of women for
incomplete abortion with misoprostol at district level, in a
low-resource setting. The overall complete abortion rate
in our study was similar to results in previous studies
(range 94–99%).8,10,29 Decentralisation of misoprostol use
in post-abortion care has been successful in a similar
setting, where ultrasound is not routinely used and
physicians are scarce,24 which is consistent with our See Online for appendix
findings. In Uganda, most physicians work in the central
urban region serving less than one third of the population.
As a response, task shifting is already taking place but
often without clear policies, planning, monitoring, and
assessment.18 A recent qualitative study at district level
in Uganda reports midwives as the main providers,
including complicated cases because of the absence of
physicians. However, both physicians and midwives had
a lack of adequate skills and expressed a need for
in-service training. Misoprostol was also rarely used in
post-abortion care because of restricted availability and no
treatment guidelines at the health facilities.15 Results
from this study provide scientific evidence that support
task shifting in post-abortion care and could be used in
the development and implementation of policies and
standard care guidelines in Uganda, and other similar
low-resource settings.
The Ugandan health system costs for treating women
with incomplete abortion, often caused by unsafe abortion,
is substantial, and so is the cost for the individual women
and their families.30 A task shift to midwives in providing
treatment for incomplete abortion with misoprostol at
district level will increase women’s access to safe
post-abortion care where it is needed most.31,32 It has the
potential to reduce the workload on the already
overstretched health-care providers, save resources, and
reduce the cost of unsafe abortion, at an individual and
societal level. To achieve these changes, an increase in
misoprostol use in post-abortion care is essential and
supplies at district level must be ensured. We suggest that
post-abortion care is incorporated in basic midwifery
education curricula to safeguard a future workforce with
post-abortion care skills. Safe and effective task shifting

www.thelancet.com Published online March 27, 2015 http://dx.doi.org/10.1016/S0140-6736(14)61935-8 5

 Articles
Panel: Research in context
Systematic review
We searched PubMed, CINAHL, and Web of Science for
articles published in English until Aug 15, 2014, with the
keywords: “post abortion care”, “incomplete abortion”,
“misoprostol”, “midwives”, “non-physicians”, and “midlevel
providers”. We reviewed scientific publications aiming to
assess midwives’ involvement in treating incomplete
abortion with misoprostol in low-resource settings. Existing
evidence is based on one observational study24 describing
misoprostol use in post-abortion care provided by
nurse-midwives. However, no randomised controlled trial has
assessed the effectiveness and safety of midwives diagnosing
and treating incomplete abortion with misoprostol.
Interpretation
Our findings show that midwives with a standardised training
programme can independently diagnose incomplete
abortion and successfully treat women with misoprostol at
district level in a low-resource setting. Our trial builds on
previous scientific evidence that shows medical treatment
with misoprostol and surgical treatment with manual
vacuum aspiration for treatment of incomplete abortion is
equally effective. Non-physician’s involvement in medical
abortion has been assessed to be safe in low-resource
settings. Our findings further support existing evidence
about the use of misoprostol and fill an important knowledge
gap about midwives’ competencies as main providers of
post-abortion care with misoprostol at district level.
between midwives and physicians also needs in-service
training in post-abortion care for both professional cadres,
based on updated evidence-based guidelines.
The strengths of our study are its randomised design,
low rate of loss to follow-up, and adequate power for
detection of the primary outcome. One limitation is that
several facilities were involved and the standard of care
might have differed between the facilities. However, this
is also a strength of the trial because the intervention
was successful irrespective of the local clinical standard
of care. The involved providers’ clinical experience of
post-abortion care ranged widely and differed between
midwives and physicians. However, the analyses bet-
ween facilities showed no cluster effect on equivalence
result. The larger loss to follow-up in the midwife group
compared with the physician group is a limitation.
Physicians have a higher status than midwives in
Uganda. Hence, this result could be interpreted as a
consequence of women feeling more inclined to return
for follow-up after being treated by a physician. It could
also be a consequence of individual differences in
providers’ ability to explain the importance of the follow-
up visit. The loss to follow-up analysis did not show any
differences except for gestational age. Women who were
lost to follow-up had a lower gestational age compared
with women who returned for follow-up (appendix).
6 www.thelancet.com Published online March 27, 2015 http://dx.doi.org/10.1016/S0140-6736(14)61935-8
Our findings are of relevance to low-resource settings
but could also have implications for other settings
where access to post-abortion care is restricted because
of shortages in human resources. Scaling up midwives’
involvement in post-abortion care at district level would
increase women’s access to safe and effective treatment
of incomplete abortion and thereby avoid severe
complications that might lead to maternal mortality
and morbidity.
Contributors
MK-A developed the protocol and study design, directed
implementation of the study, the data analyses, and was lead author
together with AC. AC participated in the implementation of the study,
in data collection, and analysis. SA was the coordinator for the facilities
included in the study and participated in data collection, analysis, and
report writing. NMT was the statistician and coordinated the data
management, and participated in analysis and report writing. EF took
part in the study design, participated in analysis, and report writing. JB
was the local principal investigator, was responsible for the overall
supervision of the trial, and participated in the implementation of the
study, data analysis, and report writing. KG-D was the principal
investigator, contributed to the study conception, and was responsible
for the overall supervision of the trial and participated in the study
design, implementation of the study, data analysis, and report writing.
All authors had access to the data, commented on the manuscript
drafts, and approved the final version submitted.
Declaration of interests
We declare no competing interests.
Acknowledgments
UNDPA/UNFPA/WHO/World Bank Special Programme of Research,
Development and Research Training in Human Reproduction provided
technical support. The funders were the Swedish Research Council
(521-2009-2605), Karolinska Institutet, and Dalarna University. We thank
the women who participated for sharing their experiences and attending
follow-up. We thank all the providers involved in the study who
remained committed and motivated throughout.
References
1 Ganatra B, Tuncalp Ö, Johnston HB, Johnson BR Jr,
Gülmezoglu AM, Temmerman M. From concept to measurement:
operationalizing WHO’s definition of unsafe abortion.
Bull World Health Organ 2014; 92: 155.
2 Shah IH, Åhman E, Ortayli N. Access to safe abortion: progress and
challenges since the 1994 International Conference on Population
and Development (ICPD). Contraception 2014; published online
April 13. DOI:10.1016/j.contraception.2014.04.004.
3 WHO, UNICEF, UNFPA, The World Bank, UNPD. Trends in
maternal mortality: 1990 to 2013. Estimates by WHO, UNICEF,
UNFPA, The World Bank and the United Nations Population
Division. Geneva: World Health Organization, 2013.
4 Rasch V. Unsafe abortion and postabortion care—an overview.
Acta Obstet Gynecol Scand 2011: 90: 692–700.
5 Escandon I, Turner K. Indicators for the essential elements of
postabortion care. PAC Consortium, 2006.
6 Weeks A. Clarifying the role of misoprostol in obstetrics.
Afr Health Sci 2001; 1: 52–54.
7 Weeks A, Alia G, Blum J, et al. A randomized trial of misoprostol
compared with manual vacuum aspiration for incomplete abortion.
Obstet Gynecol 2005; 106: 540–47.
8 Gemzell-Danielsson K, Fiala C, Weeks A. Misoprostol: first-line
therapy for incomplete miscarriage in the developing world. BJOG
2007; 114: 1337–39.
9 Neilson JP, Gyte GM, Hickey M, Vazquez JC, Dou L. Medical
treatments for incomplete miscarriage. Cochrane Database Syst Rev
2013; 3: CD007223.
10 Shochet T, Diop A, Gaye A, et al. Sublingual misoprostol versus
standard surgical care for treatment of incomplete abortion in
five sub-Saharan African countries. BMC Pregnancy Childbirth 2012;
12: 127.

11 Dabash R, Ramadan MC, Darwish E, Hassanein N, Blum J,
Wikinoff B. A randomized controlled trial of 400-μg sublingual
misoprostol versus manual vacuum aspiration for treatment of
incomplete abortion in two Egyptian hospitals. Int J Gynaecol Obstet
2010; 111: 131–35.
12 Dao B, Blum J, Thieba B, et al. Is misoprostol safe, effective and
acceptable alternative to manual vacuum aspiration for postabortion
care? Results from a randomised trial in Burkina Faso, West Africa.
BJOG 2007; 114: 1368–75.
13 Yarnall J, Swica Y, Winikoff B. Non-physician clinicians can safely
provide first trimester medical abortion. Reprod Health Matters
2009; 17: 61–69.
14 Gessessew A, Barnabas GA, Prata N, Weidert K. Task shifting and
sharing in Tigray, Ethiopia, to achieve comprehensive emergency
obstetric care. Int J Gynaecol Obstet 2011; 113: 28–31.
15 Paul M, Gemzell-Danielsson K, Kiggundu C, Namugenyi R,
Klingberg-Allvin M. Barriers and facilitators in the provision of
post-abortion care at district level in central Uganda—a qualitative
study focusing on task sharing between physicians and midwives.
BMC Health Serv Res 2014; 14: 28.
16 Fulton BD, Scheffler RM, Sparkes SP, Auh EY, Vujicic M, Soucat A.
Health workforce skill mix and task shifting in low income countries:
a review of recent evidence. Hum Resour Health 2011; 9: 1.
17 Pereira C, Mbaruku G, Nzabuhakwa C, Bergstrom S, McCord C.
Emergency obstetric surgery by non-physician clinicians in
Tanzania. Int J Gynaecol Obstet 2011; 114: 180–83.
18 Nabudere H, Asiimwe D, Mijumbi R. Task shifting in maternal and
child health care: an evidence brief for Uganda.
Int J Technol Assess Health Care 2011; 27: 173–79.
19 Piaggio G, Elbourne DR, Pocock SJ, Evans SJ, Altman DG.
Reporting of noninferiority and equivalence randomized trials:
extension of the CONSORT 2010 statement. JAMA 2012;
308: 2594–604.
20 Turner KL, Huber A, eds. Woman-centered, comprehensive abortion
care: trainer’s manual, 1st edn. Chapel Hill, NC: Ipas, 2004.
21 Turner KL, ed. Medical abortion training guide, 2nd edn. Chapel Hill,
NC: Ipas, 2010.
www.thelancet.com Published online March 27, 2015 http://dx.doi.org/10.1016/S0140-6736(14)61935-8
Articles
22 Blum J, Winikoff B, Gemzell-Danielsson K, Ho PC, Schiavon R,
Weeks A. Treatment of incomplete abortion and miscarriage with
misoprostol. Int J Gynaecol Obstet 2007; 99: 186–89.
23 WHO. Safe abortion: technical and policy guidance for health
systems. Geneva: World Health Organization, 2012.
24 Gaye A, Diop A, Shochet T, Winikoff B. Decentralizing postabortion
care in Senegal with misoprostol for incomplete abortion.
Int J Gynaecol Obstet 2014; 126: 223–26.
25 Renfrew MJ, McFadden A, Bastos MH, et al. Midwifery and quality
care: findings from a new evidence-informed framework for
maternal and newborn care. Lancet 2014; 384: 1129–45.
26 Warriner IK, Meirik O, Hoffman M, et al. Rates of complication in
first-trimester manual vacuum aspiration abortion done by doctors
and mid-level providers in South Africa and Vietnam: a randomised
controlled equivalence trial. Lancet 2006; 368: 1965–72.
27 Warriner IK, Wang D, Huong NT, et al. Can midlevel health-care
providers administer early medical abortion as safely and effectively
as doctors? A randomised controlled equivalence trial in Nepal.
Lancet 2011; 377: 1155–61.
28 Kopp Kallner H, Gomperts R, Salomonsson E, Johansson M,
Marions L, Gemzell-Danielsson K. The efficacy, safety and
acceptability of medical termination of pregnancy provided by
standard care by doctors or by nurse-midwives: a randomised
controlled equivalence trial. BJOG 2014; published online July 18.
DOI:10.1111/1471-0528.12982.
29 Renner RM, Brahmi D, Kapp N. Who can provide effective and safe
termination of pregnancy care? A systematic review. BJOG 2013;
120: 23–31.
30 Vlassoff M, Mugisha F, Sundaram A, et al. The health system cost
of post-abortion care in Uganda. Health Policy Plan 2014; 29: 56–66.
31 Haddad LB, Nour NM. Unsafe abortion: unnecessary maternal
mortality. Rev Obstet Gynecol 2009; 2: 122–26.
32 Berer M. Provision of abortion by mid-level providers: international
policy, practice and perspectives. Bull World Health Organ 2009;
87: 58–63.
7