Clinical handover communication at maternity shift changes and women's safety in Banjul, the Gambia: a mixed-methods study

A mixed-methods cross-sectional study was conducted from January to March 2018 in maternity units in all three government hospitals in the urban Greater Banjul region of The Gambia; Hospital 1 (tertiary), 2 (secondary) and 3 (secondary). The mixed-methods design included both quantitative and qualitative methodology. (See Additional File 1 details of health systems and hospitals).

Informed written consent was gained to observe staffs (including nurses, midwives, medical officers and doctors) who conduct handover on the maternity ward at shift changes for the timeframe of the study.

An observation tool covering handover content and environment was completed on the wards by three trained UK medical student researchers, each observing one shift at a time (Additional File 2). Environmental factors included: distractions, maintaining confidentiality, location, use of non-technical language, environment supportive of questions, communication method and documentation, time set aside and HCP involvement. As no prior handover standards existed in The Gambia, the observation checklist was adapted from a study in the USA [24] to also include the British Royal College of Obstetricians and Gynaecologists [25] and the American College of Obstetricians and Gynaecologists [26] guidelines. The researchers classified each woman as “in active labour” or not, and either 'high-risk' or not (obstetric emergencies or risk factors for complications as defined by the National Institute for Health and Care Excellence and WHO guidelines [18, 19]). Classification was recorded from information communicated during handover and by asking the staff afterwards. The tool was piloted in The Gambia before use. All handovers were conducted in English. To reduce bias, a similar proportion of morning, afternoon and evening shift changes were observed across the three hospitals, seven days a week, including public holidays. As handovers on different wards occurred simultaneously, researchers moved between wards to attempt to increase observations. Data were recorded onto paper and subsequently digitized using Microsoft Excell.

It was calculated that a conservative sample size of 385 individual women's handovers would detect a 50% prevalence of the handover items communicated with 95% confidence intervals with a 5% margin of error. Information topics (handover contents) recommended for good handover were categorised by Situation (10 items), Background (11 items), Assessment (3 items), Recommendations (4 items) (SBAR) [25, 26, 31, 32]. Non-parametric tests, Mann–Whitney U or Kruskal–Wallis, were used where appropriate for univariate analysis. Multiple linear regression with the number of items handed over as the dependent variable and the independent variables hospital, ward location, time of shift change, day observed, handover lead, whether in active labour, high risk woman, reason for admission, whether questions were asked at handover, whether notes were used, interruptions, handover delay, handover duration and number of women handed over was conducted to identify potential confounding factors from amongst those decided a priori using evidence from the literature. [9, 25, 26] Data were analysed using SPSS version 24.0.

The qualitative and quantitative components of the study were conducted in parallel. A purposive sampling frame included all cadres of HCPs (nurses, midwives, medical officers and doctors). Methodology combined both focus group discussions (FGDs) to gain diversified perspectives and one-to-one semi-structured interviews (SSIs) to examine personal beliefs and attitudes more comprehensively. The FGDs were challenging as the HCPs were not available at the same time. No staff took part in both FGDs and SSIs. Participants were recruited until thematic saturation was achieved.

One UK medical student trained in qualitative methodology conducted all the interviews. FGDs and SSIs were conducted at the hospital sites using a pre-determined piloted topic guide (Additional File 3). All interviews were conducted in English, spoken by all HCPs in The Gambia. The interviews were recorded, transcribed verbatim and anonymised. Field notes and analytic memos were also made after every interview.

Inductive thematic analysis based on Braun and Clarke’s six-step approach [33] was undertaken. A second researcher independently double-coded four of the most data-rich transcripts for analyst triangulation [34].

Characteristics of the handovers are given in Table 1. Of the 90 (81.8%) delayed handovers, only 4.5% were due to emergencies or situations requiring the shift-lead's attention. Emergencies included complex births and post-partum haemorrhage. Handovers in all hospitals were conducted at the end of each woman's bed except in the labour rooms in Hospital 2 where handover took place in front of a completed whiteboard located in the adjoining staff station (12% of handovers) in view of HCPs, patients and visitors. This displayed women's details including name, gravidity, parity, vital signs and physical examinations. All hospital wards had rooms available where handover could potentially have taken place confidentially, however many were already used for other purposes. No specific written protocols or pro-forma were available to assist with verbal handover. A report book, located at the nursing station on all wards, contained a brief handwritten summary of certain high-risk women on the ward and was updated at the end of each shift.

During verbal handovers, out of a possible 28 topics, a median (IQR) of 6 (5–9) SBAR items were communicated for each woman (Table 2). Handovers commonly discussed more "Situation" and "Background" topics compared to "Assessment" and "Recommendation'' topics, where in around half of handovers, no information was exchanged (see Additional File 5). No SBAR items were consistently communicated for all women (except "lead identified" and "patient location" which were implied with the bedside handover style).

Women in active labour versus not in active labour had a median of 4 more information items discussed. Those deemed high-risk also had significantly more items communicated (Tables 2 and 3). The fewest items were discussed in ante and postnatal wards, where 196 (29%) women were handed over with only a global comment.

On multivariate linear regression analysis (Table 3), the independent variables that significantly influenced the number of items handed over included the following; with fewer items discussed at the ante/postnatal ward, at the early shift change, tertiary hospital and more discussed where were women were in active labour or high-risk, admission for observation, when questions were asked and when handovers lasted longer. The regression predicted 46% of the variation in information communicated (R² = 0.46, p < 0.001).

SSIs were conducted with eight doctors, eight midwives and five nurses. Two FGDs were conducted with five midwives and four nurses in Hospital 1. Additional File 7 summarises the demographic details of all 30 participants.

Three major themes identifying barriers and facilitators to optimal handover emerged from thematic analysis these included healthcare systems factors, organisational cultural factors and individual HCP factors. These themes along with their supporting quotations are displayed in Table 4. A summary of the salient points is given below.

The quantitative and qualitative results were triangulated in order to guide intervention development [35]. The health systems, organisational cultural and individual HCP factors that interact and influence effective handover as barriers and facilitators, in our setting were applied to develop a conceptual diagram (Fig. 1). For example: need for standardised protocols is recognised by motivated individuals, health systems are required to drive development, with organisational support essential to address standardisation, use of tools and promoting inclusion of all women in shift-to-shift handover.

The important obstacles to most aspects of optimal handover were health systems factors, primarily lack of standards, protocols and training which were strongly supported by our HCPs. Studies in HICs alongside WHO standards for HICs and LICs, currently advocate guidelines and pro-forma for maternity handovers; applying tools such as SBAR for efficient communication and reducing adverse events. A simple checklist would further support information retention in staff who felt overloaded with information [5, 25, 26, 38]. A continuum of culturally-sensitive training, formal and informal, running alongside behaviour change methods (such as non-monitory incentives, handover champions, team supervision and monitoring) focussed on increasing the impact of handover, is needed to embed handover processes with the aim of improving patient outcomes [39]. These highlight low-cost and sustainable solutions [5, 30, 40, 41]. Staff awareness and experience, verbalising the potential dangers of inadequate or missed handovers, is a vital individual factor facilitating these recommendations and to improve a patient safety culture prioritising handover communication.

Other health systems barriers to adequate handover existed in that doctors have no set handover routine, obtaining information through asking on-ward staff, patients' notes or calling doctors on the last shift. Furthermore, for nurses/midwives, only the shift-leads gave and received handover which was not formally passed on to the rest of the team. This limits proactive monitoring/treatment of women, amplifying the risk of adverse events with potentially critical gaps in clinical information transferred to the majority of HCPs [7]. As suggested by our participants, promoting engagement of all junior and senior staff in handover could improve teamwork, quantity of information communicated[8] and provide valuable learning opportunities for junior staff to confidently support optimal handover culture [39, 42]. However, practically implementing a multidisciplinary team handover in these busy maternity hospitals would require a change in organisational culture with support and staffing [8].

Critically over 20% of women were missed from handover, and most others had little clinical information communicated. The WHO recommends in the interest of patient safety [43], quality improvement and optimal pregnancy outcomes, a standardised handover with clear, accurate information transfer should be performed and promoted for all pregnant women [40]. Co-ordinated care for every woman is arguably a priority safety concern in sub-Saharan Africa (SSA) in order to reduce the persistently high maternity death rates [1, 5, 44]. In The Gambia and SSA, many women admitted in maternity wards are high-risk (40% in our study with many referred from local health centres). In active labour, but also ante-natal and post-partum stages, these women by definition need close monitoring for unexpected changes in condition [5, 8]; with over half of maternal and newborn deaths occurring in the first few days post-partum [20, 44]. Health systems based intervention, addressing specific communication requirements in different stages, utilising targeted tools such as the WHO Safe Childbirth Checklist, could reduce handover variability and missed women, guiding continuity of care.

Lack of protocols led to a number of downstream barriers to comprehensive handovers. With no set times for shift-to-shift handover, timing varied largely leading to absence of either coming or leaving staff or shortened handovers, with some handovers skipped completely. Exacerbated by staff shortages, this was a major barrier blurring the transition of patient responsibility [41]. At the HCP level, prompt handover was not prioritised and significant individual barriers such as finding childcare and transport to and from work were identified. Provision of transport services for staff alongside availability of childcare centres around the hospitals were suggested and could be paid for by the HCPs (as they currently bear the cost) are potential facilitators to punctuality. These organisational interventions, could improve staff motivation and encourage HCPs’ full and equal engagement in work [45, 46].

Compared to other times, 8am handover had fewer information items communicated. Fatigue from 12-h shifts alongside handover delays may have played an important role in this [46]. Interruptions during handover were frequent and acceptable from organisational culture perspective. This is a recognised cause of shortened, inaccurate or incomplete handovers, exacerbated by lack of protected time and staff shortage [47, 48]. Addressing these systems based standards, monitoring a protected handover time and prioritising attendance over other tasks (excluding emergencies) could avoid delays, improving information transfer [29].

Handovers were commonly verbal only. Alongside a lack of standardised handover documentation and variable legibility of notes instead left staff reportedly relying upon memory or judgement for patient care. This was viewed as an organisational culture issue, where the importance of complete and legible notes was not a priority, even though many staff reported the likely result of errors in care [14, 15, 25, 49]. Practices including writing notes on scraps of paper and the report book (located at all nursing stations containing a brief handwritten summary of some high-risk women, updated each shift) was not unfamiliar and could be built upon to facilitate handover.

Whiteboards were one method employed successfully in one ward to provide handover structure and clarity. For the larger wards (30 beds), creative use of boards needs to be considered [50, 51]. Adoption of electronic databases has been shown to increase the efficiency of information exchange and was supported by our HCPs [52]. However, unreliable electricity supply, funds for establishing such a system, alongside varying HCP computer literacy, are barriers to this solution currently in The Gambia’s health system.

Reflecting on handover beyond simply provision of care, handover is intricately linked with patient satisfaction and patient-centred care [5, 47]. Holistic concepts (patient values and concerns) were scarcely discussed and all but 4% of handovers were not confidential [48]. This was an organisational culture issue with little prioritisation given to patient-centred care, likely given the clinical pressure of work and lack of human resources. For example, all hospital wards had private/staff rooms available where handover could potentially have taken place confidentially, however these were often used for other purposes. In a country where only 55% of women attend health facilities for childbirth, improving women's experiences has the potential to encourage women to seek healthcare, improving national maternity outcomes [4, 28].

Challenges to adequate handover as demonstrated in this LMIC are ongoing globally across maternity services [8, 22‐24, 47]. Handover interventions proposed utilise a low-cost, culturally-sensitive, multi-faceted approach (Table 5). This involves health system managers alongside staff participation with activities that prioritise patient-safety and patient-centred culture [5, 7, 8, 41, 43]. Importantly, our qualitative data demonstrated HCPs’ willingness and drive to provide high-quality care and improve skills. However, clear evidence and consensus on specific information and environment required at maternity handovers requires further development [8, 23, 24, 38, 47, 53]. Our framework (Fig. 1) could begin to provide a conceptual guide for this process.

This study reports the first research on handover practices in maternity care in Africa. Given the high MMR in SSA, this is a significant contribution to maternity care improvement. Although handover is location, context and culture specific, important lessons and recommendations for comparable LMIC's can be learnt by in-depth study of one handover system such as this. Further research should investigate rural regions of The Gambia and other LMICs to increase generalisability and transferability.

This study minimised selection bias by providing observations across all three-government maternity hospitals in the capital city; researchers observed all maternity wards across a random selection of days and times, including weekends and public holidays. By further triangulating quantitative and qualitative data, the richness of data was maximised to create a versatile framework.

Although the majority of the SSIs were conducted in private, some were carried out on the ward (HCPs unable to leave the ward unattended) which could have compromised participants speaking freely. Quantitative data analysis was limited by the observation tool placing equal weighting on each SBAR information item. Some items may have greater impact on patient care, particularly across different wards. Additionally, researcher presence could have influenced handover, with some HCPs conducting a more detailed handover and others omitting information. However, we consider the large sample size would have diluted these effects. Furthermore, classification of high-risk relied on information given at handover and clarifying with HCPs afterwards. This may have overestimated the quality of information discussed for this group since some high-risk women may not have been identified. The checklist used was adopted from a study in USA as we could find nothing similar from a LMIC setting. The qualitative findings however supported use of the checklist as HCPs demonstrated and volunteered the importance of similar principles, facilitating the development of specific and targeted recommendations.