Determinants of Maternal Near Miss: A Facility-Based Case-Control Study in Luanda, Angola

Abstract

Introduction

Understanding the determinants of maternal near miss is crucial for designing preventive strategies and health policies to reduce preventable maternal deaths. This study aimed to identify the determinants of maternal near miss among women admitted to two tertiary hospitals in Luanda, Angola.

Methods

A facility-based case-control study was conducted at two referral hospitals from June to September 2022. Case participants were recruited among women admitted who met at least one of the WHO near miss criteria. The controls were women admitted for vaginal birth, without any criteria for severe maternal morbidity. For each maternal near-miss case, three age-matched controls were selected. Bivariate analysis was used to evaluate the differences between these two groups. Multivariate unconditional logistic analysis models were constructed to identify determinants of maternal near miss. Reduced models (forward technique) were adjusted, and adjusted odds ratios and 95% confidence intervals were obtained. The level of significance was set at p < 0.05.

Results

In this study, 245 cases of maternal near miss and 735 controls were included. Maternal near miss was strongly associated with a long delay in seeking care (AOR = 27.38; 95% CI: 12.60–59.47), being referred to referral hospitals (AOR = 11.07; 95% CI: 7.25–16.91), having a preexisting medical disorder (AOR = 12.00; 95% CI: 6.25–23.04), having a previous cesarean section (AOR = 2.42; 95% CI: 1.43–4.10), and consuming alcohol (AOR = 1.59; 95% CI: 1.04–2.43).

Discussion/Conclusion

Delay in seeking care and being referred to referral hospitals were the main determinants of maternal near miss. Decision makers and health system managers must provide primary and secondary healthcare facilities with resources to manage obstetric complications before and during transfer to tertiary hospitals. Additionally, community interventions focused on overcoming barriers and constraints in seeking healthcare should be prioritized as they can positively impact the prevention of maternal near miss.

Introduction

Globally, the maternal mortality ratio (MMR) was estimated at 197 maternal deaths per 100,000 live births in 2023, after a reduction of 40.0% over 2 decades [1]. Maternal mortality continues to be an important public health problem in developing countries, particularly in sub-Saharan Africa, which accounts for approximately 70% of all maternal deaths worldwide [1].

Maternal deaths constitute a small fraction of the burden of maternal morbidity [2, 3], which is caused by a wide variety of conditions and complications that range from common non-life-threatening disorders to severe disorders with potentially life-threatening conditions for women [3, 4]. Many women who experience severe complications during pregnancy, childbirth, and puerperium survive these complications [2, 5, 6]. These surviving women were classified as near miss, defined by the WHO “as a woman who almost died, but survived a severe maternal complication, occurring during pregnancy, childbirth, or within 42 days after the end of the pregnancy” ([7], pp. 33).

There are large differences in the rate of near-miss cases across countries [5, 6, 8]. However, near-miss cases are more prevalent than maternal deaths, even in developing countries [8, 9]. A systematic review covering research from low-resource settings reported a ratio of near-miss cases to maternal deaths of 26 to 1 [9]. Maternal near-miss (MNM) cases and maternal deaths share the same pathological and risk factors related to their condition [7]. As maternal death has become less frequent, near-miss cases have been studied as surrogates of maternal deaths [10]. Therefore, near-miss cases provide valuable information for assessing the circumstantial and risk factors underlying severe maternal morbidity that proximally lead to maternal death [2, 4]. Research on MNM cases and deaths has been used to improve health systems and the quality of care [10].

MNM has multifactorial causes. Previous research conducted in sub-Saharan countries has shown that maternal age, low socioeconomic status, previous cesarean section, preexisting chronic morbidity, and barriers to access to healthcare services are predictors of near miss prevalence [11, 12]. However, the magnitude of the association between these factors and the occurrence of MNM diverges across settings [11].

Between 2000 and 2023, the MMR of Angola significantly decreased (78.7%), from 860 to 183 maternal deaths per 100,000 live births [1]. However, like most sub-Saharan African countries, Angola is still far from achieving the Sustainable Development Goal (SDG) of reducing the global MMR to less than 70 per 100,000 live births by 2030 [1]. A clear understanding of the determinants of MNM in each setting is essential to improving the quality of maternal healthcare. Therefore, this study aimed to identify the determinants of MNM among women admitted to two tertiary hospitals in Luanda, the capital of Angola.

Materials and Methods

Study Setting, Period, and Design

This case-control study was conducted in two referral public hospitals in Luanda (Lucrécia Paím Maternity and Maternity of the Augusto Ngangula General Specialized Hospital) from 8th June to 3rd September 2022. Both hospitals provide maternal and neonatal carefree of charge and are medical training centers for specialists in gynecology and obstetrics.

Study Population and Eligibility Criteria for Cases and Controls

All women admitted during pregnancy, childbirth, or within the first 42 days post-partum were considered the target population. All the case participants were admitted women with potentially life-threatening conditions and classified as having MNM according to the WHO criteria [7], adapted to low-resource settings. Controls were women admitted for a vaginal birth, without any criteria for maternal morbidity associated with pregnancy, childbirth, or puerperium, because they are considered to be healthier women. For each MNM case, we selected three age-matched controls among those admitted to the hospital within 24 h after case identification.

Sample Size Determination

The sample size was determined using the Epi Info 7 software StatCalc menu for case-control studies. The main exposure factor for near miss is the delay in seeking care, and we considered the proportion of exposed cases and controls based on a study conducted in Ethiopia, in which 34.0% of cases and 19.5% of controls took more than 36 h to seek healthcare [13]. The sample size of near-miss cases was estimated to be using a power of 90%, a confidence level of 95%, and a control-to-case ratio of three (3:1). The minimum estimated sample size was 544 women (136 cases and 408 controls) according to the Fleiss procedure. However, during the study period, more case participants were admitted. Thus, the final sample included 980 women, 245 cases, and 735 controls.

Data Collection Tool and Measurement

The data were collected through a review of medical records and face-to-face interviews. For MNM cases, the interviews were performed when the woman’s health status allowed informed consent. Two doctors and two trained midwives participated in data collection under the supervision of two obstetricians/gynecologists. The questionnaire was adapted from the WHO tool [7]. After discharge, the controls were followed up in the community during the postpartum period to ensure that they did not develop any severe maternal complications; that is, they did not become near-miss cases.

Data Quality Management

The team researcher was trained in standardized procedures when the questionnaire was completed for data collection. The pretest results served to improve the questionnaire and daily research procedures. The supervision of all the research activities was essential to verify the correct application of the methodology. All completed questionnaires were carefully checked to detect inconsistencies and omissions. The clinical data collected were compared to those recorded in the clinical process, and any errors detected were immediately corrected. The data were entered into the Epi Info 7.2.5.0 program by two independent typists (double typing), and discrepancies were resolved using the original questionnaires.

Potential Risk Factors for MNM

We considered a four-level conceptual framework previously used for assessing the determinants of MNM [14, 15]. Each level of determinants included a set of variables under a particular topic. The first level included sociodemographic characteristics such as marital status (single or with companionship), years of schooling (none, <9; 9–12 or >12 years), maternal residence (urban or periurban), and monthly family income (<USD 100; USD 100–300, or >USD 300). The second level included clinical and behavioral profiles, such as alcohol consumption, tobacco smoking, and preexisting diseases (at least one of the following diagnoses: chronic hypertension, sickle cell disease, HIV, chronic renal disease, heart disease, and/or neurologic disease). The third level included obstetric characteristics and antenatal booking status: the number of previous pregnancies with or without the occurrence of pregnancy-related adverse outcomes (abortion, fetal death, and/or neonatal death), a previous cesarean section, and an acceptable frequency of antenatal care (classified as yes if the first visit was before the 13th gestational week and at least one, two, or three antenatal visits if admission occurred up to 22 gestational weeks, up to 31 gestational weeks, and at 32 or more gestational weeks, respectively, and classified as no otherwise). The fourth level incorporated variables related to circumstances of admission to the hospital: delay in seeking care (<24, 24–36, or >36 h) and being referred from another facility.

Data Analysis

The distribution of the independent variables was obtained for both MNM cases and controls, and the differences between these two groups were assessed through chi-square tests or Fisher’s exact tests. Only variables significantly related to MNM (p < 0.05) in the bivariate analyses were included in the regression models. Multivariate nonconditional logistic analysis models were constructed to identify the determinants of MNM. Independent variables were introduced into the models based on a hierarchical approach [16], considering the four levels of determinants defined previously. Multivariate regression analyses were conducted by running sequential models from the first to the fourth level of determinants. Each sequential model included variables from the preceding model if the p value <0.05 for at least one category of the variable. Estimates of the adjusted odds ratio (AOR) and the respective 95% confidence interval (95% CI) were obtained. The statistical analysis was performed with IBM SPSS Statistics (version 25.0). The level of significance was set at p < 0.05.

Results

During the study period, a total of 980 women were included in the sample (245 nearmiss cases and 735 controls). Table 1 shows the comparison between the MNM and control groups.

Table 1.

Distribution of cases and controls according to sociodemographic and behavioral characteristics, clinical and obstetric profiles, and circumstances at admission

Variables	Case		Control		p valuea
	n = 245		n = 735
	n	%	n	%
Sociodemographic characteristics
Maternal age
<20 years	37	15.1	111	15.1	Matching variable
20–34 years	159	64.9	477	64.9
≥35 years	49	20.0	147	20.0
Marital status	​	​	​	​	0.325
Single	75	30.6	200	27.2	​
Married/cohabitation	170	69.4	535	72.8
Schooling years	​	​	​	​	<0.001
<9	110	44.9	206	28.0	​
9–12	124	50.6	469	63.8	​
>12	11	4.5	60	8.2	​
Maternal residence	​	​	​	​	0.029
Urban	69	28.2	264	35.9	​
Periurban	176	71.8	471	64.1	​
Family monthly income	​	​	​	​	0.017
<USD 100	82	33.5	195	26.5	​
USD 100–300	118	48.2	431	58.6	​
>USD 300	45	18.4	109	14.8	​
Clinical and behavioral profile
Preexisting medical conditionsb	​	​	​	​	<0.001
No	183	74.7	718	97.7	​
Yes	62	25.3	17	2.3	​
Alcohol consumption	​	​	​	​	0.004
No	170	69.4	578	78.6	​
Yes	75	30.6	157	21.4	​
Tobacco consumption	​	​	​	​	0.481
No	241	98.4	728	99.0	​
Yes	4	1.6	7	1.0	​
Reproductive history and antenatal booking status
Number of pregnancies and previous adverse pregnancy-related outcomes	​	​	​	​	0.004
Primigesta	65	26.5	247	33.6	​
2–4 with no previous adverse outcomesc	68	27.8	226	30.7	​
≥5 with no previous adverse outcomesc	22	9.0	44	6.0	​
2–4 with previous adverse outcomesc	45	18.4	142	19.3	​
≥5 with previous adverse outcomesc	45	18.4	76	10.3	​
Previous caesarian section	​	​	​	​	<0.001
No	188	76.7	667	90.7	​
Yes	57	23.3	68	9.3	​
Acceptable prenatal cared	​	​	​	​	0.501
No	133	55.0	423	57.6	​
Yes	109	45.0	312	42.4	​
Missing	3	1.2	0	0.0	​
Circumstances at admission
Delay in seeking care	​	​	​	​	<0.001
<24 h	100	40.8	510	69.4	​
24–36 h	83	33.9	215	29.3	​
>36 h	62	25.3	10	1.3	​
Referred from another facility	​	​	​	​	<0.001
No	120	49	677	92.1	​
Yes	125	51.0	58	7.9	​

^aChi-square test or Fisher’s exact test.

^bHypertension, HIV, sickle cell disease, diabetes, chronic renal, heart, and/or neurologic diseases.

^cPrevious adverse outcomes are abortion, fetal death, and/or neonatal death.

^dStart ANC before the 13th week of pregnancy; first visit prior to 22 weeks, second visit between 23 and 31 weeks, and third visit at 32 weeks or more.

In terms of sociodemographic characteristics, almost two-thirds of the cases and controls were aged 20–34 years, 15.1% were adolescents, and 20% were aged 35 years or older. The results also revealed that low education levels (44.9% vs. 28.0%, p < 0.001) and living in a peri-urban area (71.8% vs. 64.1%, p < 0.001) were more common among patients with MNM than among their control counterparts. Patients with MNM more frequently reported not only lower (33.5% vs. 26.5%; p = 0.017) but also higher family income (18.4% vs. 14.8%, p = 0.017).

In terms of clinical and behavioral profiles, we observed a greater proportion of patients with preexisting medical conditions (25.3% vs. 2.3%; p < 0.001) and alcohol consumption (30.6% vs. 21.4%; p < 0.001) among patients with MNM than among controls. Additionally, MNM cases and controls were significantly different in obstetric characteristics and antenatal booking status, such that patients with MNM were more likely to have a previous cesarean section (23.3% vs. 9.3%; p < 0.001) and have five or more previous pregnancies, either without (9.0% vs. 6.0%; p = 0.004) or with (18.8% vs. 10%; p = 0.003) a previous adverse pregnancy-related outcome. In regard to the admission to the hospital, compared with their counterparts, patients with MNM were much more likely to delay care seeking by more than 36 h (25.3% vs. 1.3%; p < 0.001) and to be referred from another facility (51.0% vs. 7.9%; p < 0.001).

Multivariate analysis was conducted by including blocks of variables according to previously defined levels of determinants, as presented in Table 2, and the final model is presented in Table 3. The factors significantly associated with increased odds of MNM were preexisting medical disorders (AOR = 12.00; 95% CI: 6.25–23.04), alcohol consumption (AOR = 1.59; 95% CI: 1.04–2.43), previous cesarean section (AOR = 2.42; 95% CI: 1.43–4.10), delay in seeking care, particularly longer delays (AOR = 27.38; 95% CI: 12.60–59.47), and being referred from another hospital (AOR = 11.07; 95% CI: 7.25–16.91).

Table 2.

Multivariate regression analyses of sociodemographic characteristics, clinical and behavioral profile, reproductive history and antenatal booking status, and circumstances of admission (hierarchical analysis)

Variables	Model 1		Model 2		Model 3		Model 4
	OR (95% CI)	p value	AOR (95% CI)	p value	AOR (95% CI)	p value	AOR (95% CI)	p value
Schooling years
<9	Reference	​	​	​	​	​	​	​
9–12	0.49 (0.36–0.68)	<0.001	0.54 (0.38–0.76)	<0.001	0.54 (0.39–0.76)	<0.001	0.78 (0.52–1.17)	0.233
>12	0.25 (0.12–0.54)	<0.001	0.27 (0.12–0.62)	0.002	0.34 (0.16–0.72)	0.005	0.90 (0.40–1.92)	0.824
Residence
Urban	Reference	​	​	​	​	​	​	​
Peri-urban	1.43 (0.97–2.12)	0.072	​	​	​	​	​	​
Family monthly income
<USD 100	Reference	​	​	​	​	​	​	​
USD 100–300	0.87 (0.61–1.24)	0.446	0.75 (0.52–1.08)	0.123	​	​	​	​
>USD 300	2.11 (1.21–3.70)	0.009	1.26 (0.74–2.17)	0.397	​	​	​	​
Preexisting medical conditionsa
No	Reference	​	​	​	​	​	​	​
Yes	​	​	14.23 (7.99–25.33)	<0.001	13.83 (7.67–24.91)	<0.001	11.86 (6.16–22.82)	<0.001
Alcohol consumption
No	Reference	​	​	​	​	​	​	​
Yes	​	​	1.72 (1.21–2.44)	0.002	1.69 (1.18–2.41)	0.004	2.41 (1.42–4.08)	0.035
Number of pregnancies and previous adverse pregnancy-related outcomes
Primigesta	Reference	​	​	​	​	​	​	​
2–4 with no previous adverse outcomes	​	​	​	​	0.87 (0.57–1.33)	0.519	​	​
≥5 with no previous adverse outcomes	​	​	​	​	0.80 (0.40–1.59)	0.527	​	​
2–4 with previous adverse outcomes	​	​	​	​	0.91 (0.56–1.47)	0.699	​	​
≥5 with previous adverse outcomes	​	​	​	​	0.95 (0.55–1.65)	0.853	​	​
Previous caesarian section
No	Reference	​	​	​	​	​	​	​
Yes	​	​	​	​	2.47 (1.56–3.92)	<0.001	1.58 (1.03–2.41)	0.001
Delay in seeking care (hours)
<24 h	Reference	​	​	​	​	​	​	​
24–36 h	​	​	​	​	​	​	1.53 (1.02–2.29)	0.040
>36 h	​	​	​	​	​	​	27.29 (12.53–59.44)	<0.001
Referred from another facility
No	Reference	​	​	​	​	​	​	​
Yes	​	​	​	​	​	​	10.68 (6.93–16.46)	<0.001

Model 1 includes sociodemographic characteristics; model 2 includes sociodemographic characteristics and clinical and behavioral profile; model 3 includes model 2 and reproductive history and antenatal booking status; model 4 includes model 3 and circumstances at admission. Previous adverse outcomes are abortion, fetal death, and/or neonatal death.

^aHypertension, HIV, sickle cell disease, diabetes, chronic renal, heart, and/or neurologic diseases.

Table 3.

Final model from multivariate regression analysis

Variables	AOR	95% CI	p value
Preexisting medical conditionsa
Yes	12.00	6.25–23.04	<0.001
Alcohol consumption
Yes	1.59	1.04–2.43	0.031
Previous caesarian section
Yes	2.42	1.43–4.10	<0.001
Delay in seeking care, h
24–36	1.53	1.03–2.30	0.037
>36	27.38	12.60–59.47	<0.001
Referred from another facility
Yes	11.07	7.25–16.91	<0.001

^aHypertension, HIV, sickle cell disease, diabetes, chronic renal, heart, and/or neurologic diseases.

Discussion

In Angola, as in other developing countries, particularly in sub-Saharan Africa, MNM remains a significant public health problem. This study assessed the determinants of MNM in two tertiary hospitals in Luanda. Our findings indicated that alcohol consumption, previous cesarean section, preexisting medical conditions, being referred from another healthcare facility, and a delay in seeking care of more than 36 h were independent predictors of MNM in our setting.

The results of our study revealed that delays longer than 36 h in seeking care are most strongly associated with MNM. Delay in seeking and accessing timely and adequate emergency obstetric care has been considered a preventable determinant of severe maternal morbidity and mortality [17–21]. Previous research conducted in Africa was summarized in a recent systematic review [11], where the delay in seeking care appeared to be a predictor of MNM based on pooled estimates from six studies. This delay has been associated with a lack of health literacy [22], including low awareness and poor recognition of danger signs [17, 18, 20, 23] and low maternal autonomy in seeking healthcare [20–22, 24]. The long distances required to access healthcare services, the lack of transportation, and the cost related to transportation are also issues that negatively impact maternal healthcare-seeking behavior [21–23]. Furthermore, the perceived poor quality of care in health facilities can discourage women from seeking healthcare [22, 23]. Community interventions have been implemented in some sub-Saharan countries to address these issues, and such interventions can improve healthcare-seeking behavior among pregnant women [18, 25]. Further research in our setting is critical to gain a clear understanding of the individual and contextual factors underlying delay in seeking early maternal care, which is important for overcoming such delay.

According to our findings, being referred from another facility strongly affects the occurrence of MNM. This finding is consistent with those from previous research conducted in Uganda [26], Eritrea [27], Ghana [28], and Ethiopia [29, 30]. A maternal referral system is critical for transferring patients from lower-level healthcare facilities, which lack resources, to tertiary hospitals with resources to provide comprehensive emergency obstetric care [31]. There is a body of research reporting several barriers and constraints that compromise the function of the referral system in low- or middle-income countries, preventing timely access to adequate maternal emergency care [31–34]. Poor road network has been considered an issue even when the distance between referring and referral facilities is short [34]. Additionally, the lack of available and affordable transportation, including ambulance shortages [31–33], as well as the absence of accompanying healthcare providers and life-support equipment during transfer [32, 33], compromises the effectiveness of the referral system. Late referral due to poor knowledge among health professionals concerning the identification of risk during pregnancy has also been reported [31]. Furthermore, an inadequate supply of drugs and medical equipment in lower-level facilities makes early and accurate diagnosis and the provision of initial treatment impossible before challenging cases are transferred to tertiary facilities [32]. Improving referral services and lower-level facilities by addressing these issues is critical for timely and adequate emergency obstetric care.

In our setting, the referral system suffers from the same barriers and restrictions identified in most sub-Saharan African countries. Both primary and secondary facilities lack sufficient human and material resources and medicine to manage obstetric emergencies before and during transfer. The road system is also precarious, even in urban areas, and support in medicalized ambulances with specialized personnel is unsatisfactory. Therefore, decision-makers and health system managers must develop strategies to mitigate the negative impact on maternal health.

Our findings indicate increased odds of MNM among women with preexisting chronic conditions, which is consistent with the findings of studies conducted in other sub-Saharan countries [11, 12]. Clinical conditions such as cardiac, lung, renal, and hepatic diseases can be aggravated by pregnancy, leading to severe maternal morbidity with a high fatality rate [35]. These non-obstetric complications account for almost 20% of all maternal deaths in sub-Saharan countries [36]. There is a need to strengthen sexual and reproductive healthcare services for timely and high-quality care, including contraception and family planning, as well as antenatal, intrapartum, and postpartum care [20]. Of particular importance is the integration of prevention, screening, treatment, and monitoring of noncommunicable diseases throughout women’s lifecycle, deserving special attention to the perinatal period [20, 37]. Many low-resource settings face shortages of health professionals, inadequate equipment for the diagnosis and monitoring of chronic conditions among pregnant women, preventing timely and adequate care [20].

In line with the findings of previous studies conducted in sub-Saharan countries [11, 12], our findings revealed that a prior history of cesarean section increases the odds of MNM. Although cesarean section is effective at saving maternal and infant lives when used for medically indicated reasons, this surgical procedure is associated with an increased risk of complications in subsequent pregnancies [38, 39] and a higher risk of maternal death, particularly in sub-Saharan countries [40]. A rate of cesarean section between 10 and 15% has been recommended by the WHO to minimize unnecessary risks for women and their infants. However, this rate has risen worldwide to levels that cause concern [39]. Healthcare providers should be aware of the risks of cesarean section and deter elective cesarean section in the absence of clinical indications.

According to our results, alcohol consumption was high among both the cases (30.6%) and the controls (21.4%) and appeared to be a predictor of MNM. Previous studies have shown a high prevalence of alcohol use among pregnant women in sub-Saharan countries. A meta-analysis conducted in sub-Saharan African countries (n = 37) revealed a pooled prevalence of alcohol consumption during pregnancy of 20.8% [41]. The study also reported that the main predictor of alcohol consumption among pregnant women is poor awareness and knowledge of the adverse effects [41]. This implies a need for health education throughout the lifespan to improve women’s knowledge of the potential health risks of alcohol consumption, particularly during pregnancy. Antenatal care is a favorable time for counseling on health issues, including the potential health risks to themselves and to their babies posed by alcohol consumption [42]. Strengthening antenatal care services with a skilled health workforce is mandatory for addressing maternal risk factors in our setting.

Maternal education has been reported to be a predictor of MNM, with women with no formal education having a higher risk of being MNM [11, 12]. According to our results, when only sociodemographic factors were considered in the regression model, we observed a gradient of risk such that women with higher educational levels presented lower odds of MNM. However, after adjusting for circumstances at admission (delay in seeking care and referral status), maternal education was no longer related to MNM. These findings suggest that maternal education plays a role in seeking care behavior and improving the decision-making capacity to seek timely and adequate maternal healthcare services, which in turn affects the incidence of MNM. This conclusion is consistent with previous research, reporting better healthcare-seeking behavior among more educated women [43].

Limitations and Strengths

To our knowledge, this is the first facility-based study to identify predictors of MNM in Angola. There are several strengths in this study. First, cases and controls were selected prospectively, which minimized recall and misclassification bias. Second, the identification of cases was performed by staff under the supervision of physicians specializing in obstetrics/gynecology and was based on the WHO near-miss approach. This approach has largely been widely used for monitoring severe maternal morbidity, particularly in low- and middle-income settings, allowing reliable and comparable data across settings. Third, in this study, we used a hierarchical approach based on a conceptual framework in which the potential determinants of MNM were considered according to different levels of determination [14, 16]. The choice of factors to be included in the multivariate regression model was based on a conceptual framework that described the hierarchical relation between a set of variables [15, 16]. This approach is the most appropriate for managing many variables, with the assumption that distal factors (first level of determination) can influence the variables included in the subsequent level of determination [16].

The findings of our study must be interpreted with caution because of some limitations. The selection of participants was limited to those who were admitted to the study hospitals. The study was carried out in two tertiary facilities receiving the most severe and complex obstetric cases. Further research should include low-level facilities to gain an in-depth understanding of MNM determinants. Although the identification of cases was based on the WHO’s MNM criteria, some laboratory parameters were not considered because of limited technical resources. In addition, due to the constraints of the blood bank in our setting, we reduced the threshold to five or more units of blood to define severe hemorrhage. Despite these limitations, the findings of this study can be used to provide more detailed information for policymakers and stakeholders in addressing MNM.

Conclusions

MNM assessment is a useful tool for understanding the predictors of maternal mortality and for evaluating maternal healthcare services. In our study, a delay of more than 36 h in seeking care and being referred from another facility strongly affected the occurrence of MNM. Improving the referral system is critical in our setting. Health authorities and policymakers should address barriers and constraints affecting the referral system to facilitate women’s access to timely and appropriate emergency obstetric care. Strategies to increase maternal education level should be implemented to improve healthcare-seeking behavior. Finally, further research is needed to understand the complexity of the factors underlying maternal morbidity and mortality, particularly those that negatively influence the demand for healthcare and the obstacles that make the referral system less efficient and effective.

Acknowledgments

The authors express their gratitude to the heads of the two hospitals for their authorization of the research, to all the study participants, and to the interviewers and others who made this work possible.

Statement of Ethics

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Independent Ethics Committee of the Faculty of Medicine of Agostinho Neto University (Deliberation No. 24/2022 of April 29). Permission for data collection was also obtained from the two hospitals. Before taking part in the study, all participants were informed of the purpose of the research. All women included in the study provided written informed consent by signing or fingerprinting the study for illiterate women and were informed that they could withdraw from the study at any time if they desired. For MNM cases, data collection was performed when the woman’s health status permitted informed consent. Data were collected confidentially and anonymously.

Conflict of Interest Statement

The authors declare that they have no conflicts of interest.

Funding Sources

This research received no external funding.

Author Contributions

M.C.: conceptualization, investigation, writing original draft, methodology, validation, visualization, writing review, and project administration. P.C. and P.M.: conceptualization, validation, visualization, writing review, and editing. C.T.: conceptualization, investigation, methodology, validation, visualization, writing review, and editing. All authors approved the final manuscript.

Funding Statement

This research received no external funding.

Data Availability Statement

All the data generated during this study are available from the corresponding author upon request.