Generalisability of fetal ultrasound deep learning models to low-resource imaging settings in five African countries.

Most artificial intelligence (AI) research and innovations have concentrated in high-income countries, where imaging data, IT infrastructures and clinical expertise are plentiful. However, slower progress has been made in limited-resource environments where medical imaging is needed. For example, in Sub-Saharan Africa, the rate of perinatal mortality is very high due to limited access to antenatal screening. In these countries, AI models could be implemented to help clinicians acquire fetal ultrasound planes for the diagnosis of fetal abnormalities. So far, deep learning models have been proposed to identify standard fetal planes, but there is no evidence of their ability to generalise in centres with low resources, i.e. with limited access to high-end ultrasound equipment and ultrasound data. This work investigates for the first time different strategies to reduce the domain-shift effect arising from a fetal plane classification model trained on one clinical centre with high-resource settings and transferred to a new centre with low-resource settings. To that end, a classifier trained with 1792 patients from Spain is first evaluated on a new centre in Denmark in optimal conditions with 1008 patients and is later optimised to reach the same performance in five African centres (Egypt, Algeria, Uganda, Ghana and Malawi) with 25 patients each. The results show that a transfer learning approach for domain adaptation can be a solution to integrate small-size African samples with existing large-scale databases in developed countries. In particular, the model can be re-aligned and optimised to boost the performance on African populations by increasing the recall to [Formula: see text] and at the same time maintaining a high precision across centres. This framework shows promise for building new AI models generalisable across clinical centres with limited data acquired in challenging and heterogeneous conditions and calls for further research to develop new solutions for the usability of AI in countries with fewer resources and, consequently, in higher need of clinical support.

National policies and care provision in pregnancy and childbirth for twins in Eastern and Southern Africa: A mixed-methods multi-country study.

BACKGROUND: High-risk pregnancies, such as twin pregnancies, deserve particular attention as mortality is very high in this group. With a view to inform policy and national guidelines development for the Sustainable Development Goals, we reviewed national training materials, guidelines, and policies underpinning the provision of care in relation to twin pregnancies and assessed care provided to twins in 8 Eastern and Southern African countries: Kenya, Malawi, Mozambique, Rwanda, Tanzania, Uganda, Zambia, and Zimbabwe. METHODS AND FINDINGS: We located policies and guidelines by reviewing national repositories and by contacting experts to systematically map country-level maternal and newborn training materials, guidelines, and policies. We extracted recommendations for care for twins spanning ante-, intra-, and postpartum care that typically should be offered during twin pregnancies and childbirth. We compared care provided for mothers of twins to that provided for mothers of singletons during the ante-, intra-, and postpartum period and computed neonatal mortality rates using the most recent Demographic and Health Surveys (DHS) data for each country. There was a paucity of guidance on care specifically for twin or multiple pregnancies: None of the countries provided clear guidance on additional number of antenatal care visits or specific antenatal content, while 7 of the 8 countries recommended twins to be delivered in a comprehensive emergency obstetric and neonatal care facility. These results were mirrored by DHS results of 73,462 live births (of which 1,360 were twin) indicating that twin pregnancies did not receive more frequent or intensified antenatal care. The percentage of twin deliveries in hospitals varied from 25.3% in Mozambique to 63.0% in Kenya, and women with twin deliveries were between 5 and 27 percentage points more likely to deliver in hospitals compared to women with singleton live births; this difference was significant in 5 of the 8 countries (t test p < 0.05). The percentage of twin deliveries by cesarean section varied from 9% in Mozambique to 36% in Rwanda. The newborn mortality rate among twins, adjusted for maternal age and parity, was 4.6 to 7.2 times higher for twins compared to singletons in all 8 countries. CONCLUSIONS: Despite the limited sample size and the limited number of clinically relevant services evaluated, our study provided evidence that mothers of twins receive insufficient care and that mortality in twin newborns is very high in Eastern and Southern Africa. Most countries have insufficient guidelines for the care of twins. While our data do not allow us to make a causal link between insufficient guidelines and insufficient care, they call for an assessment and reconceptualisation of policies to reduce the unacceptably high mortality in twins in Eastern and Southern Africa.

Interventions to reduce unnecessary caesarean sections in healthy women and babies.

Optimising the use of caesarean section (CS) is of global concern. Underuse leads to maternal and perinatal mortality and morbidity. Conversely, overuse of CS has not shown benefits and can create harm. Worldwide, the frequency of CS continues to increase, and interventions to reduce unnecessary CSs have shown little success. Identifying the underlying factors for the continuing increase in CS use could improve the efficacy of interventions. In this Series paper, we describe the factors for CS use that are associated with women, families, health professionals, and health-care organisations and systems, and we examine behavioural, psychosocial, health system, and financial factors. We also outline the type and effects of interventions to reduce CS use that have been investigated. Clinical interventions, such as external cephalic version for breech delivery at term, vaginal breech delivery in appropriately selected women, and vaginal birth after CS, could reduce the frequency of CS use. Approaches such as labour companionship and midwife-led care have been associated with higher proportions of physiological births, safer outcomes, and lower health-care costs relative to control groups without these interventions, and with positive maternal experiences, in high-income countries. Such approaches need to be assessed in middle-income and low-income countries. Educational interventions for women should be complemented with meaningful dialogue with health professionals and effective emotional support for women and families. Investing in the training of health professionals, eliminating financial incentives for CS use, and reducing fear of litigation is fundamental. Safe, private, welcoming, and adequately resourced facilities are needed. At the country level, effective medical leadership is essential to ensure CS is used only when indicated. We conclude that interventions to reduce overuse must be multicomponent and locally tailored, addressing women's and health professionals' concerns, as well as health system and financial factors.

Monitoring fetal growth in settings with limited ultrasound access.

Abnormal fetal growth significantly increases neonatal mortality and the risk of stillbirth. This creates the need for accurately monitoring fetal growth in all pregnancies regardless of the risk status. Several methods used in clinical practice include abdominal palpation, symphysio-fundal height measurements, and obstetric ultrasound. Of these, obstetric ultrasound remains the most reliable and objective way to monitor fetal growth. However, in most low-resource areas, access to obstetric ultrasound remains poor and this leaves the two as the only options available. This not only has effect on fetal growth monitoring but more importantly on the accuracy of pregnancy dating. To improve the current situation, we propose strategies for training of health workers, educating the public on importance of obstetric ultrasound, and improving access to basic equipment. However, interim solutions have to be implemented hand in hand with other strategies to ensure universal access to ultrasound technology for fetal growth monitoring.

Opportunities and Challenges in Realizing Universal Access to Obstetric Ultrasound in Sub-Saharan Africa.

The potential benefits of obstetric ultrasound have yet to be fully realized in sub-Saharan Africa (SSA), despite the region bearing the greatest burden of poor perinatal outcomes. We reviewed the literature for challenges and opportunities of universal access to obstetric ultrasound and explored what is needed to make such access an integral component of maternity care in order to address the massive burden of perinatal morbidity and mortality in SSA. Original peer-reviewed literature was searched in various electronic databases using a 'realist' approach. While the available data were inconclusive, they identify many opportunities for potential future research on the subject within the region that can help build a strong case to justify the provision of universal access to ultrasound as an integral component of comprehensive antenatal care.

External cephalic version in East, Central, and Southern Africa.

OBJECTIVE: To evaluate the views of maternity care providers in East, Central, and Southern Africa on external cephalic version (ECV), and its determinants, with the aim of drawing lessons for practice. METHODS: In February 2009, a cross-sectional survey using self-administered semi-structured questionnaires was conducted among delegates attending a regional conference of obstetricians and gynecologists. Descriptive statistical analysis was undertaken, and comments were analyzed for themes. RESULTS: Of the 70 questionnaires issued to eligible delegates, 64 were fully completed (response rate 91%). Seventy-nine percent of respondents did not offer ECV. Approximately a third (31%) of the practitioners offered elective vaginal breech delivery. Clinicians offering ECV did so at varying gestational ages. Clinicians not offering ECV gave various reasons including concerns about the procedure's safety and lack of training and experience with it, policy restrictions, medico-legal concerns, clinician or client reluctance, and poor results with the procedure. CONCLUSION: Overall, ECV is not widely practiced in East, Central, and Southern Africa, mainly owing to concerns related to safety and policy. Efforts aimed at reviving ECV in these regions should address these concerns. A conceptual framework of such efforts is proposed herein.

Health-related quality of life changes among users of depot medroxyprogesterone acetate for contraception.

BACKGROUND: Depot medroxyprogesterone acetate (DMPA) may have other noncontraceptive effects that could impact on the quality of life. The objective of this study was to assess the health-related quality of life changes associated with the use of DMPA for contraception. STUDY DESIGN: A prospective, observational study using the Short Form-36 quality of life questionnaire. RESULTS: After 6 months of use, the participants had an improved physical summary score, mean change [5.64 (95% confidence interval [CI], 1.87-9.4), p=.054]. There was no significant change in sexual function [5.33 (95% CI, -2.15 to 12.81), p=.0858] and mental summary score [-0.51 (95% CI, -1.90 to 2.92), p=.432]. The main side effect of DMPA was menstrual irregularity (32.5%); 17.2% of the participants found amenorrhea desirable. CONCLUSION: Besides its contraceptive efficacy, DMPA is associated with an improvement in perceived physical health with no apparent adverse effect on mental health and sexual function.

Correlation between laparoscopic and histopathologic diagnosis of endometriosis.

OBJECTIVE: To review the histopathologic diagnosis of biopsies taken following visualization of endometriosis at laparoscopy and to correlate visual with microscopic diagnoses. METHODS: A retrospective review was undertaken of medical charts with a diagnosis of endometriosis at Aga Khan University Hospital, Nairobi, Kenya, between January 2001 and October 2010. Eligibility included visual diagnosis of endometriosis at laparoscopy, with a clear record of site, size, morphology, and number of lesions. The histopathologic diagnosis of the biopsies sampled was sought. Correlation was undertaken using κ statistics for diagnostic variability. RESULTS: Of the 204 relevant records, 152 (74.5%) met the eligibility criteria; from these cases, 239 specimens were submitted for histology. The most common symptom was chronic pelvic pain (108 [71.1%]). Most biopsies were obtained from the ovary and posterior cul-de-sac. Histopathologic diagnosis was confirmed in (152 [63.8%]) specimens and correlated with Asian race, multiparity, and chronic pelvic pain. Neither the site of the lesion nor the stage of disease influenced the histopathologic diagnosis. CONCLUSION: Laparoscopic visualization of endometriosis does not always correlate with histopathologic diagnosis; several other lesions may mimic endometriosis on histopathologic examination.

Factors contributing to failure of vacuum delivery and associated maternal/neonatal morbidity.

OBJECTIVE: To determine factors contributing to the failure of vacuum delivery and to compare the neonatal and maternal morbidity associated with failed and successful procedures. METHODS: A retrospective case-control study was undertaken at Aga Khan University Hospital, Nairobi, Kenya, by review of medical charts from the period of January 2007 to December 2010. In total, 31 cases of failed vacuum delivery were compared with 124 controls where extraction was successful. The primary outcome measure was fetal malposition. Secondary outcome measures included a composite score of maternal complications, a 5-minute Apgar score below 7, an umbilical arterial pH below 7.1, and a base excess below -12. Multiple logistic regression analysis was undertaken to identify factors associated with failure of vacuum delivery. RESULTS: Demographic and labor characteristics were similar in both groups. Fetal malposition significantly contributed to the failure of vacuum delivery (odds ratio 12.7, 95% confidence interval 1.5-14.8). Failure of vacuum delivery was not associated with clinically important neonatal or maternal morbidity. CONCLUSIONS: Vacuum extraction is a safe mode of delivery where indicated, with minimal maternal and neonatal morbidity even in the event of procedural failure.

The utility of clinical care pathways in determining perinatal outcomes for women with one previous caesarean section; a retrospective service evaluation.

BACKGROUND: The rising rates of primary caesarean section have resulted in a larger obstetric population with scarred uteri. Subsequent pregnancies in these women are risk-prone and may complicate. Besides ensuring standardised management, care pathways could be used to evaluate for perinatal outcomes in these high risk pregnancies. We aim to demonstrate the use of a care pathway for vaginal birth after caesarean section as a service evaluation tool to determine perinatal outcomes. METHODS: A retrospective service evaluation by review of delivery case notes and records was undertaken at the Aga Khan University Hospital, Nairobi, Kenya between January 2008 and December 2009. Women with ≥2 previous caesarean sections, previous classical caesarean section, multiple gestation, breech presentation, severe pre-eclampsia, transverse lie, placenta praevia, conditions requiring induction of labour and incomplete records were excluded. Outcome measures included the proportion of eligible women who opted for test of scar (ToS), success rate of vaginal birth after caesarean section (VBAC); proportion on women opting for elective repeat caesarean section (ERCS) and their perinatal outcomes. RESULTS: A total of 215 women with one previous caesarean section were followed up using a standard care pathway. The median parity (minimum-maximum) was 1.01234. The other demographic characteristics were comparable. Only 44.6% of eligible mothers opted to have a ToS. The success rate for VBAC was 49.4% with the commonest (31.8%) reason for failure being protracted active phase of labour. Maternal morbidity was comparable for the failed and successful VBAC group. The incidence of hemorrhage was 2.3% and 4.4% for the successful and failed VBAC groups respectively. The proportion of babies with acidotic arterial PH (< 7.10) was 3.1% and 22.2% among the successful and failed VBAC groups respectively. No perinatal mortality was reported. CONCLUSIONS: Besides ensuring standardised management, care pathways could be objective audit and service evaluation tools for determining perinatal outcomes.