Article Text
PDF
Abstract
Purpose To identify approaches, enablers, barriers and outcomes of facility stillbirth and neonatal death audit in low-income and middle-income countries (LMICs).
Data sources We searched MEDLINE, CINAHL Complete, Academic Search Index, Science Citation Index, Complementary index and Global health electronic databases.
Study selection Studies were considered eligible when reporting the approaches, enablers, barriers and outcomes of facility-based stillbirth and neonatal death audit in LMICs.
Data extraction Two authors independently performed the data extraction using predefined templates made before data extraction.
Results of data synthesis A total of 10 articles from 7 countries were included in the final analysis. Facility or external multidisciplinary teams performed death audits on a weekly or monthly basis. A total of 1018 stillbirths and neonatal deaths were audited. Of 18 audit enablers identified, nine were at the health provider level while 18 of 23 barriers to audit that were identified occurred at the facility level. The facility-level barriers cited by more than one study included: failure to implement change; inadequate training; limited time; increased workload; too many cases and poor documentation. Six studies reported that death audits resulted in structural improvements in physical structure, training, service organisation, supplies and equipment in the wards. Five studies reported that death audits improved the standard of care, with one study showing a significant improvement in measured standards. One study reported a significant reduction in newborn mortality rate of 29.4% (95% CI 0.6% to 2.4%; p=0.0015) and one study a reduction in perinatal mortality of 4.9% (52.8% in 2007 to 47.9% in 2008) before and after perinatal audit implementation.
Conclusion Stillbirth and neonatal death audit improves facility structures, processes of care and health outcomes in neonatal care. There is a need to enhance enablers and address barriers identified at both health provider and facility levels to improve the audit process.
- audit and feedback
- clinical audit
- infant mortality
- hospital medicine
- hospital mortality
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Introduction
Improving access to healthcare alone is not enough to improve patient outcomes.1 Recently, a focus on the quality of care (QoC) was advocated to achieve the Sustainable Development Goal (SDG) 2030 of ensuring healthy lives and promoting well-being for all at all ages. Poor QoC is not only harmful but also wastes resources that could have been used in other sectors to improve the lives of citizens.1
Despite increased facility-based births, women and babies are still dying or developing lifelong disabilities due to poor QoC.2 WHO estimates 295 000 women and 2.5 million newborns die every year during childbirth from preventable causes. Furthermore, 2.6 million stillbirths occur each year. About 98% of these deaths occur in low-income and middle-income countries (LMICs).3–5
Providing high QoC in LMICs remains a challenge and performance varies across providers.1 Implementing quality improvement is possible in these countries through identifying problems in care and adopting best practice. Table 1 summarises definitions for stillbirth and neonatal deaths. WHO has recommended auditing stillbirths and neonatal deaths to identify and implement ways to improve the quality of maternal and newborn care.6 However, progress in LMICs has been limited compared with high-income countries.7
Definitions of stillbirth and neonatal deaths
Stillbirth and neonatal death audit is the process of capturing information on the causes of deaths and analysing the QoC received, in a no-blame, interdisciplinary setting to improve the care provided to all mothers and babies.6 Through the process, the hospital staff have an opportunity to learn from the cases audited and improve care.
Many factors hinder or facilitate the successful implementation of auditing stillbirths and neonatal deaths.7 Critically, the effectiveness of audit depends on the ability to complete the audit process. Without effectively implementing the planned actions to respond to the problems identified, the audit alone cannot improve QoC.8 Also, effective audit requires a system-wide effort to support the recommended initiatives. However, challenges related to system support, formulating appropriate recommendations based on preventable factors and implementing changes have been reported.7 9
This systematic review will contribute to the existing evidence base by synthesising data on facility stillbirth and neonatal death audits and provide guidance on how to undertake a successful stillbirth and neonatal death audit initiative in LMICs. We address the following objectives:
To evaluate and synthesise the evidence on the approaches and outcomes of facility based stillbirth and neonatal death audit on QoC and perinatal and neonatal health outcomes in LMICs.
To identify enablers and barriers at health provider, facility and regional or national levels of care, to the implementation of successful stillbirth and neonatal death audits in LMICs.
Our work will serve as a guide to facility stillbirth and neonatal death audit implementation by evaluating the evidence on approaches used, outcome measures, opportunities and challenges to guide future healthcare workers undertaking similar initiatives to ensure that it is evidence based.
Methods
We registered the review on the International Register of Systematic Prospective Reviews (registration number: CRD42019148515) and used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.10
Search strategy
In September 2019, we searched MEDLINE, CINAHL Complete, Academic Search Index, Science Citation Index, Complementary index and Global health for eligible studies from January 2009 to August 2019. Included search terms were “stillb*” OR “neonat*” OR “perinatal death” OR “neonatal death” AND audit OR review (online supplemental material box 1S).
Supplemental material
Inclusion and exclusion criteria
We included studies if they met all of the following criteria: (1) studies describing approaches, enablers, barriers or reporting outcomes of stillbirth and neonatal death audits at the facility level; (2) original research article reporting either quantitative, qualitative data or both (3) study done in LMIC(s) defined and identified according to World Bank list;11 (4) studies which implemented a full audit process; (5) published in English and (6) published between 1 January 2009 and 1 September 2019 (search date). We selected studies published since January 2009 as many LMICs become proactive in addressing quality problems from this date12 and we aimed to focus the review on current practice. We excluded studies that only reported descriptive findings of audits as such reviews have been well covered elsewhere.13–15 We excluded systematic reviews as we were only interested in original research articles.
Quality appraisal
We used the checklist for reviewing disparate data developed by Hawker et al16 to appraise the studies (online supplemental material box 2S). The checklist comprises nine questions, each of which has four sub-categories, permitting summation of a methodological quality score. Each paper was rated on a scale from 9 (very poor) to 36 (good).
Data extraction and analysis
Two authors (MJG and JMM) independently screened titles and abstracts for eligibility (figure 1). A third author (MA) resolved any discrepancies. Articles approved for full-text screening were reviewed by the two authors (MJG and JMM) independently by applying the pre-established inclusion and exclusion criteria listed above; if there was disagreement, we reached consensus through discussion.
Study flow diagram. LMICs, low-income and middle-income countries.
One author (MJG) performed data extraction and quality appraisal using pre-defined templates, made by the authors before the literature search. Another author (JMM) was consulted in case of uncertainty.
Since the included studies were heterogeneous regarding design and outcomes, we used the narrative approach to synthesise the evidence. We reported characteristics related to (1) publication, (2) study, (3) audit type, (4) approaches, (5) structure outcomes, (6) process outcomes, (7) health outcomes and (8) enablers and barriers. We classified the approaches, enablers and barriers according to the Kruk and Gage schema ‘synthesising improvement approaches’.17 This guide classifies approaches at micro, meso and macrosystem levels, meaning health provider level; health facility, district or clinic level; and across health system or national level, respectively.
Results
Study selection
The searches resulted in 9759 articles across all of the databases. After excluding 4565 duplicates, we screened 5194 titles and abstracts. Of these, we excluded 5094 articles for not meeting the inclusion criteria (figure 1). The remaining 100 articles underwent full-text review. Common reasons for exclusion at this stage were not reporting outcomes of interest (n=64) and not done in LMICs (n=13) according to World Bank Classification.11 All studies were rated for quality as fair with a score ranging from 23 to 32 (online supplemental material table 1S). We finally identified ten studies as appropriate for inclusion in the synthesis.
Characteristics of studies
Ten studies from seven countries met the inclusion criteria: Tanzania (three studies), Uganda (two) and one study each from Bangladesh, Moldova, Solomon Islands, Ethiopia and Zambia. Online supplemental table 2S summarises the study designs. Six quantitative,9 18–22 two qualitative23 24 and two mixed-method studies25 26 were identified. All studies were uncontrolled and described the outcome of stillbirth and neonatal death audit, with a before or after analysis, except two studies with a qualitative design.23 24 Audits were conducted either weekly9 21 22 or monthly.18 19 Study duration ranged from 1 month24 to 48 months (online supplemental table 1S).19 The deaths audited in all studies were perinatal deaths (stillbirths and early neonatal deaths) except one study which audited deaths in neonates (0–28 days) and older children.9 The total number of cases audited ranged from a minimum of 5 to a maximum of 146 deaths per month (online supplemental table 1S). A total of 1018 stillbirths and neonatal deaths were audited.
Four studies with a qualitative or mixed-methods design23–26 interviewed facility staff and key informants to understand the process of stillbirth and neonatal death audits in hospitals. The following data collection methods were used: document review,23 24 focus group discussion (FGD)23 and in-depth interviews (IDIs).23–26 The staff and key informants interviewed for IDIs and FGD varied between studies. However, IDIs and FGD included facility staff involved in the process of auditing at the private hospital, health centre, district hospital and central hospital level. Staff included were doctors, nurses, administration staff, surgeons, family planning officers, health managers and other members of the audit committee. The number of participants in each study for IDIs ranged from 29 to 66 participants.25 26 Five doctors and six nurses participated in FGD conducted by one study.23
Stillbirth and neonatal death audit approaches
Two approaches were reported by six studies in this analysis.9 10 20–23 The first three studies used multidisciplinary facility teams to audit deaths and develop and implement recommendations.9 18 22 In these studies, the lead member of the audit team was either a senior obstetrician or paediatrician or nurse. Although the composition of multidisciplinary teams differed between studies, in all three studies it consisted of obstetricians, paediatricians, medical officers, midwives, administrators, nurses, neonatologists, neonatal fellows and neonatal and delivery unit in charges. One study had only senior staff in the audit team.18 All studies used standard mortality audit forms adapted from WHO for gathering clinical information. One study also included verbal autopsy questions for guardians and staff.18
The second three studies used a confidential inquiry approach19 or external researchers20 or external and internal auditors.21 These external multidisciplinary teams were not involved in the care of patients; they audited the cases and gave feedback to facility staff to implement recommendations.
Of six studies that reported audit approaches,9 10 20–23 five studies were prospective death audits, while one was retrospective and then prospective during reaudit.23 One study used a criteria-based audit approach, where initial death audit of fresh stillbirths was undertaken and developed standards as a benchmark to improve care.21 This was followed up by reaudit of stillbirths to assess adherence to standards. Most death audit approach activities were implemented at the facility level (eg, conducting audits,recommending solutions, implementing recommendations and training). Four studies reported engagement at the national level and two studies19 22 reported national stakeholder engagement through developing guidelines, coordinating audit and disseminating the findings. In contrast, two studies20 21 reported the use of external panel members or researchers at the national level. Online supplemental table 3S summarises the approaches.
Outcomes of stillbirth and neonatal death audit
Structure
Six studies9 10 20–23 reported structural improvements in one or more areas that improved the care of women in labour and neonates in the wards (table 2). Most changes were related to the physical structure of the ward; purchasing of essential supplies and equipment; training, staffing and organisation of services in the ward. Table templates for tables 2 and 3 were adapted from Lazzerini et al.27
Structural outcomes*
Mortality, morbidity and process of care outcomes*
Process
Five studies reported changes in the process of care. One study21 cited quantitative process outcomes against eight predefined standards, and all eight standards showed some significant improvement (table 3). Another study,19 reported improvements in standard care and case management of complications during labour and delivery. Other process outcomes reported by three studies19 20 25 were improved fetal heart rate monitoring, doppler device use, data documentation, partograph use, clinical decision making for complicated cases, referral system, the involvement of other professionals (social workers and psychologist) in the audit.
Outcomes (health)
Newborn outcomes were reported in only a few studies (table 3). Only two studies reported newborn/perinatal mortality.19 22 One study (21) reported a statistically significant reduction in mortality rates among fetuses/newborns (≥37 weeks and birth weight ≥2500 g) whih decreased significantly from 5.1 per 1000 in 2006 to 3.6 per 1000 in 2013 (with 1.5 per 1000 or 29.4% reduction, 95% CI 0.6% to 2.4%; p=0.0015).19 The second study22 reported overall perinatal mortality reduction (52.8% in 2007 to 47.9% in 2008) before and after perinatal audit implementation but this difference was not statistically significant (table 3). Demonstrated changes were attributed to improved standards of care following implementation of stillbirth and neonatal death audits. No study reported newborn morbidity outcomes. One study21 reported a reduction in the incidence of maternal obstetric complications such as obstructed labour and antepartum haemorrhage, which contributes to stillbirths, following the implementation of a fresh stillbirth audit using standards as a benchmark (table 3).
Enablers and barriers of implementing stillbirth and neonatal death audits
Four studies9 23–25 reported enablers (table 4) and five studies9 23–26 reported barriers (table 5). In total, 18 enablers were identified with nine at the health provider level, seven at the facility level, and two at the national or regional system levels (table 4). Only one enabler at the health provider level was cited by more than one study (table 4). Twenty-three barriers were identified with one at the health provider level, 18 at the facility level and four at the national system levels (table 5). Eight barriers at a facility level and one barrier at a national level were cited by more than one study (table 5).
Enablers of implementing stillbirth and neonatal death audit
Barriers of implementing stillbirth and neonatal death audit
Enablers
Most enablers (n=9) were identified at the health provider level. Audit meetings provided opportunities for teaching; learning was the only enabler mentioned by more than one study (table 4).9 24 The remaining enablers at both levels were mentioned by a single study (table 4). One study25 assessed the statistical significance of changes in enablers at both the health provider and facility levels. The enablers associated with statistical significant changes at the health provider level included attendance records of review meetings (p<0.001), knowledge of objectives of maternal and perinatal death review (MPDR) (p<0.001) and an observed improvement in care (p<0.001).25 Enablers identified at facility level included feedback (p<0.001), implementation of action (p<0.001) and the existence of an MPDR committee (p<0.001).25
Barriers
Out of 23 barriers identified at three levels by five studies,9 23–26 18 were identified at the facility level only (table 5). Barriers cited by more than one study at the facility level were inadequate formation and implementation of action plans9 24–26; audit facility team members not trained9 24 25; limited time led to the postponement of meetings9 23 24; increased workload in the ward9 24 25; health workers not aware of the death audit process25 26; too many cases to review9 25; poor documentation and inadequate information management systems23 26 and inadequate human resources (table 5).9 23 At the national level, a lack of broader engagement was the only barrier mentioned by more than one study (table 5).9 26
Discussion
This review has highlighted facility level barriers to audit that were common across studies including failure to implement action plans, inadequate training, limited time, increased workload, too many cases to review, poor documentation and inadequate information management systems. These findings will assist facility audit implementers and staff in identifying practical steps to improve the impact that audit has on reducing stillbirths and neonatal deaths.
Overall, the 10 studies included in this review support the key role that death audit plays in improving care and outcomes in perinatal and newborn care. Despite the identified barriers in this review, stillbirth and neonatal death audits resulted in improvements in facility structures, processes of care and neonatal health outcomes. We identified the following variables use of senior staff (obstetrician, paediatrician or senior nurse) to champion the audits at the facility, regular audit meetings with no blame approach, involvement of key decision-makers (doctors, nurses, health managers and administrators), implementing audit recommendations such as changes in physical structure of the ward, purchasing of essential supplies and equipment, training, staffing, and changing ward protocols and policies, suggesting that improvements need to be made in audit and service organisation to improve outcomes. More enablers have been identified at the health provider level while more barriers have been identified at the facility level. As the majority of barriers are related to the availability of staff to perform a death audit, our review has also shown that even auditing one death per week can be essential in identifying gaps in care.
This review adds to the latest evidence on how audits are performed, their outcomes on QoC and perinatal and neonatal health in LMICs. In addition, the present review has identified enablers and barriers and categorised them according to system levels to guide future implementers.
All included studies audited stillbirth and early neonatal deaths (0–7 days) except one study that included neonates from 0 to 28 days. Despite included studies resulted in significant improvement in care, it is essential to note that all studies were uncontrolled before and after studies. A review by Schouten et al28 found that uncontrolled before and after studies tend to exaggerate positive impacts as opposed to studies using a controlled design. With regard to the impact of audits on neonatal outcomes, only two studies reported newborn and perinatal mortality and no newborn morbidity outcomes were reported, suggesting this area could be explored further.
Stillbirth and neonatal death audit approaches
This review has shown the usefulness of both facility and external multidisciplinary teams in performing stillbirth and neonatal death audits. Further, criteria-based audits facilitate implementation of action plans and enable healthcare workers to evaluate standards at subsequent audits. Studies on maternal death audits reported that involving facility staff in the audit process promoted successful implementation, ownership and sustainability of the process.29 30 Most audits were conducted prospectively. The number of audited cases varied among the studies in this review, from a minimum of five cases to a maximum of 146 cases per month. Even though large numbers of cases reviewed may result in an in-depth analysis of gaps in care, such audit might pose a challenge in developing and implementing recommendations due to inadequate human and material resources in the LMIC context. In such contexts, it might be unrealistic to audit all stillbirths and neonatal deaths per month as these are often numerous. Depending on level of staffing and workload at the facility, it may be more practical for the mortality audit team to either review a selection of stillbirths and neonatal deaths or increase the frequency of meetings.6 Performing stillbirth and neonatal death audits at the departmental level is also essential in identifying gaps in care and interventions.
The majority of the approach activities were implemented at the facility level. Nambiar et al31 reported three system levels that are approached when implementing quality improvement initiatives. These are the microlevel (health providers), mesolevel (health facility team) and macrolevel (regional or national level). Despite all system levels being of value, facility level activities are central to the successful implementation of stillbirth and neonatal death audits. As described by Kaplan et al32 in their model of understanding success in quality, the facility level is responsible for quality improvement leadership, initiative support, senior leadership commitment, guidance and direction that shapes behaviour of staff pursuing quality improvement projects. Proper training of staff involved in death audits ensures quality implementation of audits and its recommendations. However, structural adjustments are required to facilitate the death audits. These adjustments include audit team characteristics, workforce focus, resource availability and data infrastructure that exist across all system levels to trigger and influence the success of death audit processes.32
While at the health provider level, the participation of individual staff is essential in the audit process, a single motivated care provider who has the capability and desire to improve performance will be of great value to the system.32 National or regional level activities like regulation, tool development, governance and dissemination need continuous coordination, as they act as external motivators that stimulate the organisation to improve the performance in death audit or any quality improvement projects.32
Outcomes of stillbirth and neonatal death audits
A previous systematic review on effects of perinatal mortality audits in LMICs reported a reduction in perinatal mortality of 30% (95% CI 21% to 38%) after the introduction of facility-based perinatal audits.8 However, this previous review focused on perinatal mortality audits (stillbirths and early neonatal deaths, 0–7 days old), which may miss late neonatal death audits or early neonatal deaths occurring in neonatal wards after discharge from the labour ward, which might give a false impression about overall neonatal mortality audits (age 0–28 days). The current review retrieved the latest evidence on outcomes of stillbirth and neonatal death audits. Overall findings varied both within and between studies. Most of the articles reported a mixture of outcomes that fell into the category of structure, process and health outcomes. Only one study reported a significant decrease in newborn mortality.
Enablers and barriers to implement stillbirth and neonatal death audits
Identification of enablers and barriers are essential for hospital management and programme planners to implement successful stillbirth and neonatal death audits that improve the QoC. In this review, 18 enablers and 23 barriers were identified at three levels, with more enablers (9) cited at the health provider level and more barriers (18) cited at facility level. Similar barriers have been reported in the previous review on maternal and perinatal death audits.7 Hospital management should prioritise both enhancing enablers identified at the health provider level to maintain staff morale and resolving barriers at facility level as they demotivate staff involved in audits to effect change. In their review, Nyamtema et al26 found that other facilities had discontinued audit meetings due to the failure of hospital management to implement audit recommendations.
Limitations
The current systematic review has some limitations, mainly relating to scope. Although the present review assembled evidence from seven different countries, located in four different regions (sub-Saharan Africa, South Asia, Europe/Central Asia and East Asia and Pacific), the published research included in this review was limited since only 10 studies were identified, thus, potentially limiting generalisability. Our search was limited to literature describing approaches, enablers, barriers or reporting outcomes of stillbirth and neonatal death audits at the facility level and to English language papers only. This limited search might have missed information regarding other elements of death audits and also studies reported in other languages. Five articles could not be retrieved, which may have included important additional information to the review. Our search only included original research articles; more information may be available in the grey literature, organisation reports, reviews, dissertations and theses, and conference proceedings. Although two authors conducted screening and eligibility assessment, data extraction and quality appraisal were primarily conducted by one author, which might have led to selection bias. However, where there was uncertainty, the second author was approached.
Conclusion
Implementation of stillbirth and neonatal death audits improves structure, process and health outcomes in maternal and neonatal care. Using a multidisciplinary facility team to conduct audits contributes to the success of the process. Despite all system levels being of value, facility-level activities are central to the successful implementation of stillbirth and neonatal death audits. Even auditing a single death is useful in the process of improving care at the facility level. Completing the audit cycle by implementing recommendations is crucial to improving perinatal outcomes. The indemnification of both audit enablers and barriers can help hospital management to improve audit processes. Researchers should aim at generating more evidence on how to implement stillbirth and neonatal death audits effectively, including sustaining the practice in oder to further improve its impact on newborn outcomes in LMICs.
Acknowledgments
The authors would like to thank Alison Derbyshire from the Liverpool School of Tropical Medicine library department for support during this review. We would also like to thank the Commonwealth Scholarship Commission for financial support and Lyle Taylor for editing the manuscript. We would like to thank Dr Lazzerini and other coauthors for allowing us to adapt table templates from their systematic review published in BMJ Open Journal, 2018.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors MJG conceived the idea of the study. MJG and JMM carried out data extraction. All authors (MJG, JMM, MA, ND and SA) were instrumental in the study’s development, in reviewing drafts of the paper and in approval of the final manuscript.
Funding This work was supported by the ‘Commonwealth Scholarship Commission’ (MWCSC, 2018, 802). This paper is part of first authors PhD studies.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.