DOI: 10.1111/1471-0528.15578 General obstetrics
www.bjog.org
Incidence, causes and correlates of maternal
near-miss morbidity: a multi-centre
cross-sectional study
SA Oppong,a A Bakari,b AJ Bell,c Y Bockarie,d JA Adu,e CA Turpin,f SA Obed,a RM Adanu,g
CA Moyerc
a School of Medicine and Dentistry, University of Ghana, Accra, Ghana b Suntresu Government Hospital, Ghana Health Service, Kumasi,
Ghana c University of Michigan, Ann Arbor, MI, USA d Cape Coast Teaching Hospital, Cape Coast, Ghana e School of Medical Sciences,
University of Cape Coast, Cape Coast, Ghana f School of Medical Sciences, Kwame Nkrumah University of Science and Technology, Kumasi,
Ghana g School of Public Health, University of Ghana, Accra, Ghana
Correspondence: SA Oppong, Department of Obstetrics and Gynaecology, School of Medicine and Dentistry, University of Ghana,
PO Box 4236, Accra, Ghana. Email: wak72@yahoo.com
Accepted 25 November 2018. Published Online 24 January 2019.
Objective To explore the incidence and factors associated with miss to mortality ratio of 4.6:1. Cause of near-miss was pre-
maternal near-miss. eclampsia/eclampsia (41.0%), haemorrhage (12.2%), maternal
sepsis (11.1%) and ruptured uterus (4.2%). A major factor
Design Cross-sectional study with an embedded case–control study.
associated with maternal near-miss was maternal fever within the
Setting Three tertiary referral hospitals in southern Ghana. 7 days before birth (OR 5.95, 95%CI 3.754–9.424). Spontaneous
onset of labour was protective against near-miss (OR 0.09 95% CI
Population All women admitted to study facilities with
0.057–0.141).
pregnancy-related complications or for birth.
Conclusion For every maternal death, there were nearly five
Methods An adapted version of the WHO Maternal Near Miss
maternal near-misses. Women having a fever in the 7 days before
Screening Tool was used to identify maternal near-miss cases.
delivery were six times more likely to experience a near-miss than
These were compared with unmatched controls (uncomplicated
women not having fever.
deliveries) in a ratio of 1:2.
Keywords Maternal mortality, maternal near-miss, maternal near-
Main outcome measures Incidence of maternal near-miss,
miss indicators.
maternal near-miss to maternal mortality ratio, and cause of and
factors associated with maternal near-miss. Tweetable abstract Maternal near-miss exceeds maternal death by
5:1, with the leading cause of maternal near-miss was pre-
Results Out of 8433 live births, 288 maternal near-miss cases and
eclampsia/eclampsia.
62 maternal deaths were identified. In all, 454 healthy controls
were recruited for comparison. Maternal near-miss and maternal Linked article This article is commented on by F Okonofua,
death incidence ratios were 34.2 (95% CI 30.2–38.1) and 7.4 (95% p. 762 in this issue. To view this mini commentary visit
CI 5.5–9.2) per 1000 live births, respectively with a maternal near- https://doi.org/10.1111/1471-0528.15619.
Please cite this paper as: Oppong SA, Bakari A, Bell AJ, Bockarie Y, Adu JA, Turpin CA, Obed SA, Adanu RM, Moyer CA. Incidence, causes and correlates
of maternal near-miss morbidity: a multi-centre cross-sectional study. BJOG 2019;126:755–762.
was a target of no more than 185 maternal deaths per
Introduction
100 000 live births. Ghana ended 2015 with a maternal
Low-income countries, especially those in sub-Saharan mortality ratio of nearly twice the target: 350 per 100 000
Africa, continue to bear a disproportionate burden of live births.2 Pregnant women in Ghana continue to die
maternal morbidity and mortality.1 Ghana is one of the from largely preventable causes, including haemorrhage,
countries in sub-Saharan Africa that failed to reach the hypertensive diseases and complications related to termina-
millennium development goal for maternal health, which tion of pregnancy.3–7 These mortality figures are only a
ª 2018 Royal College of Obstetricians and Gynaecologists 755
Oppong et al.
small part of the story. For every woman who dies, there the School of Medicine and Dentistry, University of Ghana.
are many others who survive severe, life-threatening com- KATH is located in Kumasi, the second largest city in
plications that may have long-lasting sequelae.8–10 Ghana, with a population of about 2 million.13 It serves as
Morbidity and mortality can be seen as a continuum. the referral centre for most of mid-Ghana, overseeing
Maternal health can be described as ranging from normal approximately 11 000 deliveries per year, and serving as the
pregnancy with no complications at one end through mild teaching hospital affiliated with the Kwame Nkrumah
non-life-threatening complications and life-threatening University of Science and Technology School of Medical
complications, to death at the other end of the continuum. Sciences. CCTH is located to the west of Accra at the
As such, a maternal near-miss has been defined as ‘a coastal town of Cape Coast, about 100 miles from Accra. It
woman who nearly died from a life-threatening complica- serves as the main referral hospital for most of the rural
tion of pregnancy, delivery, or up to 42 days after termina- Central and parts of the Western region of Ghana, oversee-
tion of pregnancy but survived’.11 Depending on the ing approximately 2800 deliveries per year and serving as
context and the specific criteria used, the incidence of the teaching hospital of the University of Cape Coast,
maternal near-miss ranges from 0.6% to > 30% of all live School of Medical Sciences.
births.12 There has been one previous study from Ghana,
which estimated that maternal near-misses outnumbered Determining the number of live births
mortality at a ratio of about 3:1 in a single setting.7 Hence, Summary statistics were collected reflecting the total num-
the true incidence of maternal near-miss in Ghana is not ber of deliveries, live births and maternal deaths for the
known; in addition, the causes of, as well as factors associ- study period from each of the three participating hospitals’
ated with, maternal near-miss are not well understood. records. In addition, trained research assistants reviewed all
This is in stark contrast to the well-developed literature admission and delivery records as well as caesarean opera-
describing the causes of maternal mortality.3–7 tion record books to record the total number of deliveries
As mortality events become less common, maternal during the study period.
near-misses are likely to become an increasingly important
metric by which quality of care can be measured. We Identifying and classifying maternal near-misses
explored maternal near-misses at three of Ghana’s four ter- Each day, research assistants reviewed the admission and
tiary-care obstetric units to identify the burden, cause of, delivery records for all pregnant women aged 18–49 years
as well as factors associated with, maternal near-miss mor- who delivered at one of the three study sites or were
bidity. referred with pregnancy or delivery-related complications
The primary aim of this study was to estimate the inci- up to 42 days after termination of the pregnancy. Compli-
dence of maternal near-miss morbidity per 1000 live births. cations were defined as any indication in the medical
The secondary aims were (i) to determine the maternal record that a woman had experienced a pregnancy- or
near-miss to mortality ratio, and (ii) to use an embedded delivery-related complication (haemorrhage; pre-eclampsia,
case–control study to determine factors associated with eclampsia, or other hypertensive disorder; premature rup-
maternal near-miss. ture of membranes; premature delivery; oligohydramnios;
gestational diabetes; miscarriage; complication from termi-
nation of pregnancy; obstructed labour/failure to progress;
Method
malposition; shoulder dystocia; emergency caesarean sec-
Study design tion; placenta praevia; or any other noted pregnancy or
We performed a multi-centre, cross-sectional study with an delivery complication). This list of complications was used
embedded case–control study. across all three study sites.
All records were screened for eligibility using the modi-
Setting fied version of the WHO Maternal Near Miss Screening
This study was conducted at three tertiary referral hospitals Tool,14 which was simplified for use in settings without all
in southern Ghana over a 4-month period (1 April to 31 of the laboratory tests and intervention procedures available
July 2015). The study sites were the maternal and neonatal as included in the full WHO instrument (see Supplementary
units of the Korle-bu Teaching Hospital (KBTH) in Accra, material, Appendix S1). Participants were assessed for the
and the Cape Coast Teaching Hospital (CCTH) and Komfo presence of specific symptom-based criteria (such as severe
Anokye Teaching Hospital (KATH) in Kumasi. haemorrhage, severe pre-eclampsia, eclampsia, sepsis or sys-
The KBTH is located in Accra, the capital city of Ghana, temic infection or ruptured uterus, assessed by attending
with a population of about 3.2 million.13 It is the largest obstetrician/gynaecologist), intervention-based criteria (such
referral hospital in Ghana, overseeing approximately 11 000 as use of blood products, laparotomy or admission to the
deliveries annually and serving as the teaching hospital for intensive care unit) or organ dysfunction-based criteria
756 ª 2018 Royal College of Obstetricians and Gynaecologists
Maternal near-miss morbidity in southern Ghana
(such as cardiovascular, respiratory, renal, haematological, standard deviations. Bi-variate statistics and test of associa-
hepatic, neurological or uterine dysfunction). A positive tions were performed with Pearson chi-square test for cate-
response to any of the above qualified a woman as a ‘near- gorical variables and Student’s t test for continuous
miss’. This classification was double-checked by an attend- variables. Assuming an initial a of 0.05, Bonferroni’s cor-
ing physician (consultant) at each study location. Women rection was conducted to identify a level of significance of
were excluded if they died during childbirth or were unwill- P < 0.002 to account for multiple comparisons. A back-
ing to provide consent. Incidental maternal deaths would ward stepwise multiple logistic regression was performed
have been excluded; however, no incidental maternal deaths with covariates that were significant in the bivariate analysis
were identified during the study period. to determine the odds ratio (OR) and 95% CI for maternal
near-miss. The resulting model was re-run using a gener-
Identification of ‘controls’ for embedded case– alised linear mixed model with site treated as a random
control study effect to account for differences across sites. Coefficients
As described above, research assistants went through each were exponentiated to allow for reporting of odds ratios.
medical record and completed a unique screening form for
each patient, indicating which of the near-miss criteria Ethical approval
were applicable. If the screening form indicated that none This study and its components were reviewed and
of the criteria above for definition of complicated preg- approved by the institutional review boards of University
nancy or delivery was met, the patient was determined to of Ghana for KBTH site (MS-Et/M.7 – P4.5/214-2015 on
be ‘uncomplicated’ and hence qualified as a ‘control’ for 10 March 2015), University of Cape Coast for CCTH
the purpose of this study. For each near-miss ‘case’ identi- (UCCIRB/EXT/2015/02 on 8 April 2015) and at the KATH
fied, the next two uncomplicated normal vaginal deliveries (CHRPE on 25 January 2015) as well as the University of
were selected as controls. Michigan (HUM00097103 on 16 February 2015).
Identification of cause of near-miss
Results
The primary cause of each case of maternal near-miss was
determined by physician review to determine primary and During the study period, a total of 8433 live births were
contributing factors associated with maternal near-miss. recorded across the three study centres, and a total of 288
Primary cause of maternal near-miss was based on the women were identified to be maternal near-misses
WHO International Classification of Diseases, 10th revi- (Table 1), yielding an overall maternal near-miss incidence
sion,15 with contributing conditions of anaemia, HIV infec- of 3.42%, or an incidence ratio of 34.2 maternal near-
tion, previous caesarean section, prolonged/obstructed misses per 1000 live births (95% CI 30.2–38.1). (Table 2)
labour, sickle cell anaemia, and sickle cell crisis. Each case There were 79 (27.5%), 120 (41.7%) and 89 (30.8%) near-
was then reviewed by one of the study investigators to ver- miss cases at study sites I (CCTH), II (KATH) and III
ify cause of near-miss. (KBTH), respectively. During the same period, a total of 62
A Qualtrix-based (Qualtrics, Provo, UT, USA), struc- maternal deaths were recorded, yielding a maternal mortal-
tured, interviewer-administered questionnaire was used to ity incidence ratio of 7.35 (95% CI 5.5–9.2) per 1000 live
record demographic data (including reported material births. This translates to a near-miss to maternal mortality
assets, which were combined to calculate a wealth index), ratio of 4.6:1 (95% CI 3.4–6.0).
pregnancy-related data and delivery outcome information. Table 3 illustrates the primary and underlying causes of
Clinical and laboratory data were also abstracted from maternal near-misses overall, as well as by site. The pri-
patients’ medical records. mary cause of maternal near-miss was severe pre-eclamp-
sia/eclampsia (n = 110, 38.2%), severe haemorrhage (n =
Data analysis 35, 12.1%), ruptured uterus (n = 12, 4.2%) and maternal
Data were imported and analysed with STATA version 13.1 sepsis (n = 10, 3.6%). The secondary or contributing causes
(College Station, TX, USA). Study data were compared of maternal near-miss were as follows: anaemia (n = 81,
against official hospital tallies to validate the number of 28.1%), hypertensive disorder (n = 78, 27.1%), infection
deliveries and live births recorded for each month. Mater- during pregnancy (n = 49, 17.0%), obstructed labour
nal near-miss incidence was calculated by dividing the (n = 39, 13.5%) or other obstetric complications (n = 41,
number of maternal near-misses recorded by the number 14.2%). The distributions of these causes were not consis-
of live births recorded, in keeping with the predominant tent: site I (CCTH) contributed a higher proportion of
method in the existing literature.12 anaemia and pregnancy-related infections than the other
Frequencies and basic descriptive statistics were calcu- two sites; site II (KATH) had much lower percentages of
lated for all variables, including proportions, means and women with pre-eclampsia and eclampsia; and site III
ª 2018 Royal College of Obstetricians and Gynaecologists 757
Oppong et al.
(KBTH) contributed a disproportionate number of women less likely to have delivered a live infant, and they were more
with pre-eclampsia and eclampsia as well as hypertensive likely to have been referred from other facilities to the ter-
disorders (see Table 3). A total of 454 women were tiary centres for care during labour and/or delivery.
recruited as healthy controls. The distribution of controls Table 5 illustrates the results of the generalised linear
was 149 (32.8%) from site I, 172 (37.9%) from site II and mixed model with site as a random effect. Previous cae-
133 (29.3%) from site III. Table 4 illustrates the socio- sarean section, multiple pregnancy and whether the baby
demographic and health-related variables for both cases was alive at birth were not significant in multivariate analy-
and controls. There were no significant differences between ses and were removed one-by-one to yield the final model
cases and controls in terms of any of the socio-demo- shown in Table 5. This model shows that women who expe-
graphic variables measured, including age, maternal and rienced a fever within the 7 days before birth were nearly
partner’s education, marital status, wealth, or health insur- six times more likely to have a maternal near-miss than
ance status. In terms of health-related variables, maternal women who did not have a fever in the days leading up to
near-miss was associated with previous caesarean section, birth (OR 5.94, 95% CI 3.65–9.68, P < 0.001). Women who
non-spontaneous onset of labour for the index birth, multi- had been referred from another hospital were 1.5 times
ple births, lower infant birthweight, and fever < 7 days more likely to experience a near-miss than women who had
before delivery. Mothers who experienced a near-miss were not been referred (OR 1.5, 95% CI 0.99–2.34, P < 0.054);
however, this finding was not statistically significant.
Women whose labour began spontaneously were signifi-
Table 1. Identification of maternal near-miss in three tertiary
hospitals in southern Ghana cantly less likely to experience a maternal near-miss than
those women whose labour did not start spontaneously.
Symptom-based criteria 146 (19.7) Women with a maternal near-miss were more likely to have
[yes to any of the following: severe a baby with a lower birthweight than controls.
haemorrhage (n = 19), severe pre-eclampsia
(n = 30), eclampsia (n = 84), sepsis or
systemic infection (n = 12), ruptured Discussion
uterus (n = 1)]
Intervention-based criteria 189 (25.5) Main findings
[yes to any of the following: use of This study assessed cases of maternal near-miss across three
blood products (n = 98), laparotomy tertiary care centres in southern Ghana, finding an overall
(n = 67), admission to intensive incidence of 34.2 per 1000 live births. The overall maternal
care unit (n = 102)] near-miss-mortality ratio was 4.6:1 with slight variability
Organ-dysfunction-based criteria 60 (8.1)
across sites. We found that, overall, severe pre-eclampsia/
[yes to specific indicators for
cardiovascular (n = 28), respiratory eclampsia, haemorrhage, uterine rupture and sepsis were
(n = 23), renal (n = 5), haematological the leading causes of maternal near-miss, although this var-
(n = 16), hepatic (n = 4), neurological ied by site. This is in line with the leading causes of mater-
(n = 6) and uterine dysfunction (n = 5)] nal mortality in Ghana, which include termination of
Met no criteria 454 (61.2) pregnancy (19.6%), postpartum haemorrhage (19.2%),
Met one category of criteria 202 (27.22) hypertensive disorders (9.1%), obstructed labour (7.3%)
Met two categories of criteria 64 (8.6)
and sepsis (6.4%).16 The single strongest factor associated
Met three categories of criteria 22 (3.0)
with a maternal near-miss was fever within the 7 days
Table 2. Maternal near-miss incidence and incidence ratio at three tertiary hospitals
Study sites Total
I II III
Number of live births 1149 3596 3688 8433
Number of maternal deaths 11 32 19 62
Number of maternal near-miss cases identified 79 120 89 288
Maternal near-miss incidence (per 1000 live births; 95% CI) 68.7 (53.6–83.9) 33.4 (27.4–39.3) 24.1 (19.1–29.1) 34.2 (30.2–38.1)
Maternal mortality per 1000 live births (95% CI) 9.5 (3.9–15.2) 8.9 (5.8–11.9) 5.5 (2.8–7.4) 7.4 (5.5–9.2)
Maternal near-miss : mortality ratio (95% CI) 7.2:1 (3.8–13.6) 3.8:1 (2.5–5.5) 4.4:1 (2.8–7.7) 4.6:1 (3.4–6.0)
758 ª 2018 Royal College of Obstetricians and Gynaecologists
Maternal near-miss morbidity in southern Ghana
Table 3. Primary and contributing causes of maternal near-misses overall and by study site
I (n = 79) Study sites III (n = 89) Overall
n (%) II (n = 120) n (%) (n = 288)
n (%) n (%)
Primary causes
Pre-eclampsia and eclampsia 38 (48.1) 9 (7.5) 63 (70.8) 110 (38.2)
Haemorrhage 8 (10.1) 22 (18.3) 5 (5.6) 35 (12.1)
Ruptured uterus 5 (6.3) 5 (4.2) 2 (2.3) 12 (4.2)
Sepsis 6 (7.6) 3 (2.5) 1 (1.2) 10 (3.8)
Underlying or contributing causes
Anaemia 45 (56.9) 15 (12.5) 21 (23.6) 81 (28.1)
Hypertensive disorders 9 (11.4) 8 (6.7) 61 (68.5) 78 (27.1)
Infection during pregnancy 45 (57.0) 1 (0.8) 3 (3.4) 49 (17.0)
Obstructed labour 13 (16.5) 19 (15.8) 7 (7.9) 39 (13.5)
Other obstetric complications 2 (2.5) 4 (3.3) 35 (39.3) 41 (14.2)
I: Cape Coast Teaching Hospital (CCTH)
II: Komfo Anokye Teaching Hospital (KATH)
III: Korle-Bu Teaching Hospital (KBTH)
Table 4. Socio-demographic and health-related variables for maternal near-miss cases and controls at three tertiary hospitals in southern Ghana
Variable name Cases Controls t-test statistic
(Total n = 742) (n = 288) (n = 454) (P-value)
Mean (SD) Mean (SD)
Maternal age (n = 445) 29.7 (6.4) 28.8 (6.3) 
1.47 (0.15)
Maternal education (years) (n = 447) 8.9 (4.0) 8.9 (4.4) 
0.11 (0.91)
Husband’s education (years) (n = 351) 10.5 (3.9) 10.5 (3.8) 0.21 (0.83)
Assets (on 0–16 point scale)* (n = 327) 9.1 (3.4) 9.2 (3.4) 0.31 (0.75)
Mean number of prior births (n = 732) 1.7 (1.7) 1.7 (1.7) 0.11 (0.91)
Number of antenatal care visits (n = 351) 6.7 (2.4) 6.9 (2.6) 0.83 (0.40)
Infant birthweight (kg) (n = 678) 2.653 (0.803) 3.080 (0.597 7.91 (<0.001)†
Cases % (n) Controls % (n) Chi-square statistic (P value)
Married (n = 742) 38.5 (111) 43.4 (197) 1.71 (0.19)
Polygamous marriage (n = 742) 3.5 (10) 1.8 (8) 2.17 (0.14)
Has national health insurance (n = 447) 93.6 (147) 96.2 (279) 1.51 (0.22)
Delivered in a healthcare facility (n = 444) 99.4 (154) 98.9 (286) 0.17 (0.67)
Previous caesarean section (n = 740) 28.8 (83) 13.7 (62) 25.46 (<0.001)†
Mode of delivery index pregnancy (n = 742) 3.4 (0.18)
Vaginal delivery 86.5 (249) 89.7 (407)
Caesarean section 8.7 (25) 7.9 (36)
Spontaneous labour (n = 712) 47.4 (129) 89.09 (392) 149.4 (<0.001)†
Fever < 7 days before labour (n = 705) 39.1 (104) 13.4 (59) 61.3 (<0.001)†
Anaemia during pregnancy (n = 702) 35.9 (95) 33.9 (148) 0.04 (0.85)
Multiple pregnancy (n = 726) 7.1 (20) 3.8 (17) 3.72 (0.05)
Live infant at birth (n = 715) 91.1 (255) 97.2 (423) 13.2 (<0.001)†
Referred from other healthcare facility (n = 731) 49.1 (139) 35.9 (161) 12.4 (<0.001)†
*Assets include aggregate list of 16 different material goods owned or available within the household, including car/truck, motorcycle, bicycle,
electricity, solar light, refrigerator, television, DVD/VCR, radio, sewing machine, stereo system, electric/box iron, fan, mobile phone, electric/gas
cooking stove, donkey cart/push truck, kerosene stove, personal computer.
†Factors associated with maternal near-miss
ª 2018 Royal College of Obstetricians and Gynaecologists 759
Oppong et al.
Table 5. Generalised linear mixed model (with site as a random effect) for predictors of maternal near-miss at three tertiary hospitals in southern
Ghana
Variable Odds ratio Standard error Z P value 95% CI
Infant birthweight 0.99 0.0001 
5.07 <0.001* 0.998–0.999
Spontaneous onset of labour 0.09 0.245 
9.81 <0.001* 0.056–0.146
Fever in 7 days before delivery 5.94 0.248 7.16 <0.001* 3.649–9.681
Referred from another facility 1.52 0.530 1.93 0.054 0.992–2.344
*Significant predictors of maternal near-miss in a negative binomial regression.
leading up to birth. Women who reported experiencing a observed difference may be due to the inclusion of two other
fever in the days before delivery were nearly six times more tertiary care centres in Ghana that are outside the capital city.
likely to experience a near-miss than women who did not This ratio is also closer to the 5.6:1 reported by some authors
have a fever, even after controlling for site differences. from India.16 Cause of death data were not complete for the
maternal deaths that occurred in each of the participating
Implications for practice healthcare facilities, hence comparison of cause of death and
This study suggests that there is a need to screen for, and cause of near-miss within each facility or across the study sites
pay additional attention to, women who report a fever in was not possible. However, we obtained national-level cause
the 7 days leading up to delivery to avert their progression of maternal death data from the Global Burden of Disease
to potentially life-threatening complications. In view of the Study,17 allowing us to compare the leading causes of maternal
known relationship between maternal near-miss and mor- death in Ghana with the leading primary causes of maternal
tality it may be helpful to investigate how febrile morbidity near-miss. We found the leading causes of maternal near-
contributes to maternal near-miss, especially as the two misses at three of the four largest hospitals in the country to
identified leading causes of maternal near-miss were pre- be fairly consistent with the leading causes of maternal mortal-
eclampsia/eclampsia and haemorrhage. ity in Ghana, suggesting that review of cases of maternal near-
miss are indeed a viable alternative to review of maternal mor-
Strengths and limitations of the study tality cases. With decreasing maternal deaths in many places
This study has several strengths. First, data were collected around the world, maternal near-miss identification and
at three tertiary care centres in Ghana, ensuring diversity in maternal near-miss audit are becoming more useful methods
our sample. We also collected data prospectively, not to review the quality of care provided.
relying upon retrospective chart reviews. We included One interesting aspect of our findings includes the lack
socio-demographic factors not typically included in clinical of significant association between socio-demographic fac-
studies. However, despite its strengths, our data collection tors and maternal near-miss. Maternal age, maternal educa-
window was limited and so we cannot explore seasonal or tion, husband’s education, household size, wealth, parity,
year-to-year differences. We also did not collect quality- gravidity, religion and marital status were not associated
of-care indicators that may have helped contextualise with maternal near-misses. This is different from other
differences found across sites in terms of variability in the studies, which found age,18 gravidity,19 maternal educa-
number and causes of maternal near-misses. tion19 and partner’s education20 to be significantly associ-
ated with maternal near-misses. Notably, one study in
Interpretation Brazil found that social and demographic characteristics of
This study adds to the literature in several ways. In Ghana, the mother were not directly linked to maternal near-miss
most previous studies have focused on mortality with few status, although such factors were linked to differences in
studies evaluating morbidity.3–7 We found the incidence of care seeking, which was then in turn linked to near-miss
maternal near-misses across three tertiary care centres dis- status.21 In our study, care seeking – e.g. number of ante-
persed over a wide geographic area to be higher than that natal visits – was not significantly different between cases
previously reported from one hospital in Accra, which iden- of near-miss and controls. One possible explanation of the
tified 94 maternal near-misses out of 3206 live births, or a limited role of social factors in our study is that we com-
maternal near-miss incidence of 29.3 per 1000 live births.7 pared healthy controls with near-misses, rather than com-
The maternal near-miss to mortality ratio in our study paring near-misses with those women who died. It is
(4.6:1) is almost twice the 2.5:1 reported despite using the possible that women who experience a near-miss are more
same maternal near-miss criteria in both studies. This similar to healthy controls in terms of socio-demographic
760 ª 2018 Royal College of Obstetricians and Gynaecologists
Maternal near-miss morbidity in southern Ghana
characteristics than they are to women who do not survive
Supporting Information
a life-threatening complication.
Additional supporting information may be found online in
the Supporting Information section at the end of the
Conclusion article.
In conclusion, this study identified nearly 300 cases of Appendix S1. Maternal near-miss tool. &
maternal near-miss, reflecting a maternal near-miss inci-
dence of approximately 34 per 1000 live births with
approximately five cases of maternal near-miss for every References
maternal death. Near misses in our study were caused pre- 1 World Health Organization. Trends in Maternal Mortality 1990–
dominantly by hypertensive disorders, haemorrhage, uter- 2013. 2014 [http://apps.who.int/iris/bitstream/10665/112682/2/9789
ine rupture and sepsis, with the single largest correlate in a 241507226_eng.pdf?ua=1] Accessed 29 September 2017.
2 WHO, UNICEF, UNFPA, The World Bank. Trends in Maternal Mortality:
multivariate model being the presence of fever in the
1990–2008. [http://whqlibdoc.who.int/publications/2010/9789241500265_
7 days before delivery. Spontaneous onset of labour was eng.pdf]. Published 2010. Accessed 12 August 2017.
also associated with a lower risk of maternal near-miss. 3 Zakaria AY, Alexander S, van Roosmalen J, Buekens P, Kwawukume
EY, Frimpong P. Reproductive age mortality survey (RAMOS) in
Accra, Ghana. Reprod Health 2009;6:7.
Disclosure of interests 4 Adu-Bonsaffoh K, Oppong SA, Bilinla G, Obed SA. Maternal deaths
None declared. Completed disclosure of interests form attributable to hypertensive disorders in a tertiary hospital in Ghana.
available to view online as supporting information. Int J Obstet Gynecol 2013;123:110–3.
5 Gumanga SK, Kolbila DZ, Gandau BBN, Munkaila A, Malechi H.
Trends in maternal mortality in Tamale Teaching hospital. Ghana
Contribution to authorship Med J 2011;45:105–10.
SAO, AB, AJB, RMA and CAM conceived the idea and devel- 6 Asamoah BO, Moussa KM, Stafstrom M, Musinguzi G. Distribution of
oped the research question, SAO, AB, AJB, YB, JAA, CAT causes of maternal mortality among different socio-demographic
groups in Ghana; a descriptive study. BMC Public Health 2011;11:159.
and SO conducted the experiment to obtain the data; SAO,
7 Tuncalp O, Hindin MJ, Adu-Bonsaffoh K, Adanu RM. Assessment of
AB, AJB and CAM analysed the data, SAO and CAM wrote maternal near-miss and quality of care in a hospital-based study in
the first draft of the manuscript, and all authors reviewed Accra, Ghana. Int J Gynecol Obstet 2013;123:58–63.
and approved the final manuscript before submission. 8 Tuncalp O, Hindin MJ, Adu-Bonsaffoh K, Adanu RM. Listening to
women’s voices: the quality of care of women experiencing severe
Details of ethics approval maternal morbidity in Accra, Ghana. PLoS ONE 2012;7:e44536.
9 WHO. Beyond the numbers: Reviewing maternal deaths and com-
This study and its components were reviewed and plications to make pregnancy safer. [http://www.who.int/maternal_
approved by the institutional review boards of University child_adolescent/documents/9241591838/en/]. Accessed 12 August
of Ghana for KBTH site (MS-Et/M.7 – P4.5/214-2015 on 2017.
10 March 2015), University of Cape Coast for CCTH 10 Tuncalp O, Hindin M, Souza P, Cho D, Say L. The prevalence of
maternal near-miss: a systematic review. BJOG 2012;119:653–61.
(UCCIRB/EXT/2015/02 on 8 April 2015) and at the KATH
11 Say L, Souza JP, Pattinson RC. Maternal near-miss: towards a
(CHRPE/AP/021/15 on 25 January 2015) as well as the standard tool for monitoring quality of maternal health care. Best
University of Michigan (HUM00097103 on 16 February Pract Res Clin Obstet Gynecol 2009;23:287–96.
2015), which was the data coordinating centre. 12 Kaye DK, Kakaire O, Osinde MO. Systematic review of the
magnitude and case fatality ratio for severe maternal morbidity in
Funding Sub-Saharan Africa between 1995–2010. BMC Pregnancy Childbirth
2011;28:65. https://doi.org/10.1186/1471-2393-11-65
The study was funded by the U.S. National Institutes of 13 Ghana Statistical Service. 2010 Population and Housing Census.
Health, 1-D43-TW-009353-01 grant provided to the 14 Souza JP, Cecatti JG, Haddad SM, Parpinelli MA, Costa ML, Katz L,
University of Michigan. The funding agency played no role et al. The WHO maternal near-miss approach and the maternal
in study design, data analysis, writing the manuscript or severity index model (MSI): tools for assessing the management of
severe maternal morbidity. PLoS ONE 2012;7:e44129.
the decision to publish this manuscript.
15 World Health Organization, editor. The WHO Application of ICD-10
to Deaths During Pregnancy, Childbirth and the Puerperium, IDC
Acknowledgements MM. Geneva: World Health Organization; 2012, 68 p.
We wish to acknowledge the authorities of the KBTH, 16 Kassebaum NJ, Bertozzi-Villa A, Coggeshall MS, Shakelford KA, Steiner
KATH and CCTH for their immense support during the C, Heuton KR, et al. Global, regional, and national levels and causes of
maternal mortality during 1990–2013: a systematic analysis for the
study. We also wish to acknowledge Miss Elorm Kudzawu,
Global Burden of Disease Study 2013. Lancet 2014;384:980–1004.
Zelda Arku, Magdalene Torto and Dominic Nyarko for 17 Roopa PS, Verma S, Rai L, Kumar P, Pai MV, Shetty J. “Near Miss”
their roles in participant recruitment and administrative obstetric events and maternal deaths in a tertiary care hospital: an
support in conducting this study. audit. J Pregnancy 2013;2013:1–5.
ª 2018 Royal College of Obstetricians and Gynaecologists 761
Oppong et al.
18 Pafs J, Musafili A, Binder-Finnema P, Klingberg-Allvin M, Rulisa S, 20 Nansubuga E, Ayiga N, Moyer CA. Prevalence of maternal near miss
Essen B. Beyond the numbers of maternal near-miss in Rwanda – a and community-based risk factors in Central Uganda. Int J Gynecol
qualitative study on women’s perspectives on access and Obstet 2016;135:214–20.
experiences of care in early and late stage pregnancy. BMC 21 Domingues RMSM, Dias MAB, Schilithz AOC, Leal MDC. Factors
Pregnancy Childbirth 2016;16:257. associated with maternal near miss in childbirth and the postpartum
19 Nakumuli A, Nakubulwa S, Kakaire O, Osinde MO, Mbalinda SN, period: findings from the birth in Brazil National Survey, 2011–2012.
Nabirye RC, et al. Maternal near misses from two referral hospitals Reprod Health 2016;13(Suppl 3):115.
in Uganda: a prospective cohort study on incidence, determinants
and prognostic factors. BMC Pregnancy Childbirth 2016;16:24.
Maternal near-miss morbidity: is this evidence of maternal health
quality in sub-Saharan Africa?
F Okonofua
Obstetrics and Gynaecology, College of Medical Sciences, Women’s Health and Action Research Centre, University of Benin, Benin City,
Edo State, Nigeria
Linked article: This is a mini commentary on SA Oppong et al., pp. 755–762 in this issue. To view this article visit
https://doi.org/10.1111/1471-0528.15578
Published Online 20 February 2019.
The article by Samuel Oppong et al. for potential confounders to identify how To date, near misses have not been
makes interesting reading (Oppong et al. mortality was prevented among near used consistently as a measure of quality
BJOG 2019; 126:755–62). It describes a misses. of care in regions with high rates of
study carried out in three tertiary referral Nevertheless, the study is important for maternal mortality. We suggest that the
hospitals in southern Ghana that investi- two main reasons. First, using the three- incidence of near misses is a key indica-
gated the incidence of and factors associ- delay model proposed by Thaddaeus and tor of quality of care, because it accounts
ated with maternal near-miss morbidity. Maine (Social Sci Med 1994;36:1091–110), for all components of emergency obstet-
The study used the WHO Maternal Near- it suggests that although women may ric care in referral facilities. Also, any
Miss Screening tool to identify maternal experience delays in accessing referral near miss is likely to share risk factors
near misses among 8433 live births and facilities, concentrating efforts on compli- with maternal deaths, so indicating what
reported a maternal near-miss rate of 34.2 cations through emergency obstetric care can be done to avert more deaths. Fur-
per 1000 live births compared with a can be effective in preventing maternal thermore, it should motivate providers
maternal mortality ratio of 740 per deaths. This study is a salutary reminder and policy-makers to take concrete steps
100 000 live births. This implied a near to health workers and policy-makers in to aggressively manage women with sev-
miss to mortality ratio of nearly 5 : 1, sub-Saharan Africa that much can be ere complications. This is also a call for
indicating that nearly five deaths are achieved if there is high-level willingness larger studies that use appropriate con-
averted for every reported maternal death. and determination to prevent maternal trols to enable the identification of speci-
The authors compared the results with deaths. fic actions that lead to prevention of
unmatched controls with uncomplicated Second, the reported higher incidence of maternal deaths in women experiencing
deliveries and concluded that women maternal near misses in this study suggests severe pregnancy complications.
experiencing fever within 7 days of deliv- improved quality of maternal health care in
ery were six times more likely to experi- the referral hospitals. Previous studies Disclosure of interest
ence a near miss compared with women report higher case fatality rates from Friday Okonofua was a reviewer of
without fever. obstetric complications in sub-Saharan the original article. Completed disclo-
Although this study is novel and is one Africa. Although this study was not sure of interests form available to view
of a few large studies that report near-miss designed to document quality of care in the online as supporting information.&
maternal morbidity in sub-Saharan Africa, referral facilities, the reported high rates of ª 2019 Royal College of Obstetricians
it can be debated whether a better control near misses suggests lower case fatalities and Gynaecologists
would not have been women with similar associated with complications, and there-
complications who died. Alternatively, the fore improved quality of emergency obstet-
authors could have statistically controlled ric care.
762 ª 2018 Royal College of Obstetricians and Gynaecologists