Design

This study was a retrospective, observational, descriptive review of congenital anomalies diagnosed at birth within the neonatal service at Edendale Hospital (EDH) in KZN Province, SA. The study period was from 1st January 2018 to 31st December 2018.

Study setting

EDH is a regional (secondary level) healthcare facility located in the uMgungundlovu District in the city of Pietermaritzburg. It serves a predominantly urban population of around 1.4 million mainly indigenous, Zulu-speaking African people. It has well-organized obstetric services and a 62-bed Neonatal Care Unit (NCU) catering for an average of 600 in-facility deliveries a month and referrals from the surrounding state-run healthcare clinics. This includes normal and complicated deliveries, with newborn care for well-babies and those requiring intensive care. This arrangement of a single, referral facility providing healthcare services to a relatively unchanging catchment population provided a relevant setting to meet the objectives of this study.

Study participants

As part of routine care, all in-house live births at EDH underwent a comprehensive neonatal examination within 24 hours of birth. Those identified with congenital anomalies were offered appropriate care and the BDNT was completed. These clinical records were included in the scope of this study, regardless of the gestational age of the baby. Stillbirths and other pregnancy losses (spontaneous/induced abortions, including termination of pregnancy due to foetal abnormality) were excluded as little routine clinical data were collected for these cases. Neonates with identified congenital anomalies born elsewhere and referred to EDH after birth were also excluded from the study to avoid inflating the birth prevalence rate as the number of deliveries at referral sites was not accessible within the study.

Case definition

For this study, congenital anomalies only (a sub-set of CDs) were recorded. Congenital anomalies are defined as structural or anatomical abnormalities detected at birth (by discharge in this study) and classified according to Chapter XVII of ICD-10 [19]. Based on this case definition, functional anomalies and other CDs listed elsewhere in the ICD-10 system that are not easily recognisable (i.e., without obvious dysmorphology) at birth remained unidentified. This limitation is important when interpreting the observed birth prevalence and comparing this data to other sources.

Congenital anomalies identified in the study were categorised into major and minor conditions. Major or serious anomalies are those that may result in death, limited life expectancy or lifelong disability, particularly in the absence of care [9]; whereas minor anomalies may have little impact on health status or quality of life [9,20].

Case ascertainment and data collection

Relevant cases for inclusion in this study were actively ascertained by the study team. Details of all deliveries at EDH were recorded in the birth register maintained by the Obstetric team. All live-born neonates were examined within 24 hours of delivery before discharge by a midwife or a doctor in the NCU team as a part of routine clinical care. For the study duration, the medical team comprised full-time neonatologists and general paediatricians, with the daily care team also including rotating paediatric registrars, medical officers (3-monthly) and interns (monthly). Following this assessment, healthy neonates and those diagnosed with minor congenital anomalies not requiring further care remained with their mothers in the post-natal units. Neonates and mothers with no complications were discharged as early as six hours post-delivery. Others may have been observed and routinely discharged within 24–48 hours of delivery (longer for weekend deliveries) when they were deemed fit. Sick neonates and those with serious congenital anomalies were admitted to the NCU for holistic management. This included a clinical assessment to ascertain the extent of the abnormality, to undertake appropriate diagnostic testing, clinical care, genetic counselling for parents and notification using the BDNT. These admissions were recorded in the neonatal admission register. All routine activities were managed by the NCU team. For complex cases, the team had access to a general paediatrician with a special interest and some training in clinical genetics for assistance with diagnosis, care and/or genetic counselling. A gatekeeping system in place for appropriate genetic testing ensured referral to and/or consultation with relevant genetic specialists at the tertiary hospital.

Data for this retrospective study were extracted from the birth register, neonatal admission register and the individual BDNT forms (all paper-based). Incomplete BDNT forms or more complex cases requiring a dysmorphology evaluation and syndrome identification (e.g., multiple anomalies) required the researcher to review individual clinical records and update the BDNT forms. All original BDNT forms were routinely kept on file within the NCU. As reported by Lebese et al [15], and during the feasibility assessment for this study it was important to ensure that the BDNT was accurately and fully completed for all identified cases with congenital anomalies to enhance data quality. To achieve this, the researcher/first author clinically rotated through the NCU for three months during the study period and served as a local champion for the span of the project. This was to strengthen active surveillance by raising awareness around congenital anomalies, the BDNT surveillance process, reminding staff to complete the tool and to review the completeness of the forms monthly. Completed BDNT forms were routinely submitted to the hospital coordinator as required by the NDoH. The functionality of this system outside EDH was beyond the scope of the study.

Data analysis

The in-facility birth prevalence rate was calculated in 2018 using the number of live births with congenital anomalies as the numerator and the total number of live births at EDH as the denominator, multiplied by 1 000 to report the rate per 1 000 live births as per convention. The number, percentage and birth prevalence rates of major, minor (i.e., polydactyly) and isolated congenital anomalies in diagnostic sub-categories were reported separately. To prevent double-counting, neonates with syndromes and multiple anomalies were counted only once in the overall live birth prevalence in the relevant sub-category and not for each specific anomaly. For example, a neonate with Trisomy 21 with a ventricular septal defect and a trachea-oesophageal fistula was counted under Down syndrome only and not separately (again) under both circulatory and digestive systems.

Statistical analysis was undertaken to compare birth prevalence rates for selected congenital anomalies observed with modelled birth prevalence rates for SA published previously [3,21]. Differences between these birth prevalence rates (p-values) were calculated using Chi^-squared statistical test and confidence intervals were calculated with the “test-based method” [22] using the SciStat online calculator (www.scistat.com).

Ethics

Ethical clearance was granted for this study by the University of KwaZulu-Natal Biomedical Research Ethics Committee (Ref No. BE409/18). Gatekeeper permission to conduct the study was obtained from the Chief Executive Officer at EDH and the study was registered on the National Health Research Database. All data were collected retrospectively in March 2019 from routine care records or registers during the study period and were anonymized at the point of collection with a sequential study number allocated to protect all patients’ identities and personal records. The data collection tool used was an anonymized copy of the BDNT. There was no direct patient contact in this study therefore individual patient consent was not required for ethical clearance. Collected data was stored electronically on password-protected drives and computers.

Birth prevalence

Birth prevalence studies for congenital anomalies may be useful to establish baseline rates and monitor trends, explore possible causes, develop prevention strategies, and review changes and response to interventions for specific anomalies over time.

The main aim of this study was achieved with the measurement of the birth prevalence of congenital anomalies that were easily recognisable by discharge after birth as a subset of total CDs possible during all life stages. When compared with previously published birth prevalence rates for congenital anomalies globally; averaging at 20–30 per 1 000 live births [23–26], both the rate observed (15.57 per 1 000 live births) and the rate excluding polydactyly (13.44 per 1 000 live births) in this study are approximately half the published average birth prevalence of congenital anomalies. This may be due to differences in the scope and case definitions used in these studies. Poorer case identification at the study site for reasons highlighted as study challenges and limitations may also be a factor.

The observed birth prevalence rate in this study is very similar to another SA study by Venter et al (14.97 per 1 000 live births in 1995), published over 25 years earlier [26]. This offers the opportunity for potential comparison of congenital anomalies data changes in SA before HIV/AIDS and after its successful prevention (PMTCT) in neonates. It is also comparable to previously published data from two other LMIC sites in Africa at different times using similar methodologies and case definitions, specifically in Uganda by Tann et al (17 per 1 000 live births in 2006) [27] and Abbey et al in Nigeria (20.73 per 1 000 live births in 2017) [28].

If a congenital anomalies birth prevalence of 15.57 per 1 000 live births represents the 26% that are easily recognisable at birth [26], then in a relatively constant catchment population at EDH the total CD-related birth prevalence for the study site may be approximated to around 59.88 per 1 000 live births. This is again within the range quoted for total CD-related birth prevalence globally of between 40 to 80 per 1 000 live births [9]. The modelled value for SA using 2012 as the reference year published by Malherbe et al was specified as 68 per 1 000 live births [21]. This suggests the observed birth prevalence of congenital anomalies at EDH is likely to be close to the true value. An extension of this study measuring the birth prevalence over the different age periods (neonatal, infant and under-five) would offer a clearer picture of the total CD burden.

The pattern of congenital anomalies

Previous published SA studies.

The pattern of congenital anomalies observed in this study aligned with those reported globally. The most frequently reported conditions are all defined as priority conditions in SA by the NDoH [13,14]. Adding further credibility was the comparison of birth prevalence rates of key congenital anomalies with observed with rates obtained through similar facility-based South African studies, presented graphically in Fig 1. The highest birth prevalence rate was observed by Venter and Christianson et al in the only rural study for three of the congenital anomalies profiled (DS, anencephaly and spina bifida) [26]. The highest overall birth prevalence was recorded for Congenital Talipes Equinovarus by Pompe van Meerdervoort et al in the 1970s [23]. While the birth prevalence rates were comparable for most anomalies, it was lowest for spina bifida and highest for Orofacial Clefts (OFC).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Comparison of birth prevalence rates for key congenital anomalies from the current study with rates observed by other facility-based studies in South Africa.

^a Pompe van Meedervoort 1976: Prospective,75% urban hospital-based study (Pelonomi Hospital, Bloemfontein, Free State), 10 000 live births over 3 years [23]. ^b Kromberg & Jenkins 1982: Retrospective, urban hospital-based study (Chris Hani Baragwanath Hospital, Johannesburg, Gauteng), 28 689 live births over 2 years (1976 to 1977) [24]. ^c Delport et al 1995: Prospective, urban hospital-based study (Kalafong Hospital, Pretoria, Gauteng), 17 351 live births over 3 years (1986 to 1989) [25]. ^d Venter & Christianson 1995: Prospective, rural, hospital-based (Mankweng Hospital, Limpopo), 7 617 live births over 3.5 years (1989 to 1992) [26]. ^e Saib et al 2021: Retrospective, predominantly urban, hospital-based (Edendale Hospital, Pietermaritzburg, KwaZulu Natal), 7 516 live births over one year (2018, current study).

https://doi.org/10.1371/journal.pone.0255456.g001

The high variation in birth prevalence of NTDs between the studies may be attributed to the introduction of mandatory folate fortification of staple crops in 2003 [29,30]. Similarly, higher NTD rates reported by Kromberg and Jenkins in 1982 may be indicative of improved reporting via a retrospective approach, compared to the other prospective studies [24]. This alludes to paradoxically poorer data collection in prospective versus retrospective methods due to the study challenges highlighted later in this article. Reasons for the high rates of spina bifida and anencephaly reported by Venter and Christianson et al in comparison to other studies pre-dating folate fortification are still not clear and may be attributed to the complex interaction of genetic and environmental factors [26,31,32]. The extremely high birth prevalence reported by Pompe van Meedervoort et al for Congenital Talipes Equinovarus may be due to the inclusion of both isolated and syndromic club foot whereas the current study differentiates between these aetiologies (case definitions) [23].

Based on previously reported birth prevalence rates by similar studies for oculocutaneous albinism; one of the commonest single-gene disorders in SA, at least 2 affected births would have been expected in this study [24–26]. This absence cannot be explained. Differences in birth prevalence rates of DS and other Trisomies may reflect variations in the proportions of mothers of AMA in the studies, as well as challenges in identifying neonates with these conditions at birth [33,34]. Due to incomplete data on AMA in the current and previous studies, further analysis was not possible.

Modelled SA data

In Table 4, the live birth prevalence rates observed in the current study were compared to modelled national estimates for SA in 2017 generated via the Modell Global Database of Congenital Disorders (MGDb) [3,6,35–37]. The MGDb method combines robust, observed data from well-established surveillance systems with demographic data to produce baseline (no interventions) and actual (current care) estimates, using the infant mortality rate (IMR) as a proxy to quantify available services [3,6,35].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Comparison of observed (EDH) with modelled (SA) birth prevalence rates of selected congenital anomalies.

https://doi.org/10.1371/journal.pone.0255456.t004

While there was apparent variation between birth prevalence rates for the conditions compared via the observed and modelled data (Table 4), statistical analysis revealed no significant difference between any of the compared rates. This comparison has limitations due to differences in demographics, e. g. year of study (2017 versus 2018) and the population (SA versus KZN facility) and specific indicators vary locally versus nationally. Nevertheless, this comparison may serve to validate the MGDb modelling approach, including results via the SA application [21] for the key congenital anomalies at sites that lack the ability or resources to conduct routine surveillance or research.

Study strengths

This study, one of only a few undertaken following the improved management of the HIV/AIDS crisis, confirms congenital anomalies as a significant healthcare burden for SA to address. This was the first study conducted in KZN to determine the birth prevalence of congenital anomalies, with all previous published studies undertaken in other provinces. This empiric evidence highlights the re-emergence of congenital anomalies (and CDs) in KZN and the need for similar studies in other parts of the country.

This facility-based study, incorporating active surveillance, is more feasible for large, diverse populations, particularly in LMICs such as SA where healthcare resources (both human and financial) are constrained. While this type of study may be subject to referral bias due to births by non-residents and referrals from outlying clinics before birth, it offers the advantage of obtaining high-quality data on key conditions at sentinel sites. The alternative, population-based approach is unaffected by this type of bias as it includes all (home and facility-based) deliveries. However, these studies are more suited for HIC, and smaller populations or regions or by sub-national systems due to the higher cost and infrastructure required, as has been demonstrated by poor data resulting from the population-based BDNT implemented in SA [15].

The collection of accurate data is essential for precise surveillance. Before the study, a passive surveillance approach yielded approximately 3–6 congenital anomalies per 600 deliveries on average per month at the study facility (Personal communication Dr Bhoola, Head of Clinical Unit—NCU 2017). In this study, an active surveillance approach supported by a clinical champion resulted in a higher rate of congenital anomalies being detected, averaging at 10 per 600 deliveries per month. This simple and effective quality improvement measure enhanced reporting accuracy and offers a feasible methodology for sites in other LMIC contexts. Good quality data is more useful when comparing with other studies.

Study challenges

Before a congenital anomaly can be reported it must first be accurately and timeously diagnosed. This requires appropriately trained healthcare professionals (HCPs) able to diagnose congenital anomalies and comply with reporting requirements of a surveillance system as a part of routine clinical care. Genetics content included in both medical school and nursing college curricula is insufficient [38–40], and varies greatly between institutions and countries, resulting in many HCPs lacking relevant genetics knowledge, skills, and expertise.

In SA, specialised medical genetic services are inadequate. There were only 12 practising clinical medical geneticists and a similar number of genetic counsellors countrywide in 2015, equating to 1 per 5 million and 1 per 8.4 million of the population respectively [41]. This is five times below the recommended number per population of 1 per million for clinical medical geneticists and 1 per 580 000 for genetic counsellors [41,42]. There have been minimal changes in these capacity ratios in recent years due to retirement, emigration and lack of state-posts for qualifying HCP in this sector. Despite being without a medical geneticist for decades, and the loss of an effective provincial coordinator in 2012, KZN was reported as the province with the highest reporting compliance, contributing over 50% of national surveillance data [15]. In 2018, following intense advocacy efforts, one medical geneticist was hired for the 11 million population of the KZN province and a centralised provincial genetic service is now being developed out of the tertiary hospital in Durban. Recruitment of a genetic counsellor is underway with ongoing support of two genetic nurses.

In this study (prior to the appointment of a medical geneticist in province), complex cases were referred to a paediatrician with a genetics interest in place of a medical geneticist. This individual has received training in the Medical Genetics Education Program (MGEP) and five years of experience running a specialist level referral clinical genetics clinic at the nearby tertiary hospital in Pietermaritzburg. The team of neonatologists have received no formal genetics training other than their sub-speciality training, which included access to a foeto-maternal anomaly clinic at the training site. The remainder of the medical and nursing teams lack specialised genetics training. While the gatekeeping system involving the referral of patients to genetic specialists at tertiary hospitals in the region ensures appropriate tests are requested, it places additional stress on an already poorly resourced system.

The use of paper-based systems, including the BDNT, continues to impact surveillance compliance and data quality even when integrated as part of routine clinical care. Within this study, the active, champion-based surveillance effectively improved data quality. However, a sustainable, long-term solution is needed, through the investment of relevant resources into integrating the notification of congenital anomalies as part of an effective health information system and electronic health record. By default, this would fill data gaps experienced in this study, such as missing maternal age data to establish the true extent of AMA as a risk factor for Trisomies. Such data systems are already in place in most private facilities and billing through medical aid (insurance) schemes (including ICD-10 coded diagnosis) making this smaller sector (approx. 15% of health services in SA) appealing as a feasible and additional source of reliable CD data, despite not currently contributing to the BDNT.

Some blood samples of the study population were lost by the testing laboratory. Anecdotal evidence suggests this is a common challenge in the province and largely systemic, with loss and leakage of samples occurring during transport from satellite to main laboratory sites. The additional cost implications, delayed diagnosis and patient trauma due to repeated testing highlight the need for these challenges to be addressed.

These challenges may have contributed to the lower birth prevalence observed and highlight the urgent need for capacity building at all levels, together with appropriate resource investment to improve the surveillance, care and prevention for those affected by all CDs.