Search

ObjectivesThe objectives of this study were to (1) determine if membership of a clinical information network (CIN) was associated with an improvement in the quality of documentation of in-patient neonatal care provided over time, and (2) characterise accuracy of prescribing for basic treatments provided to neonatal in-patients if data are adequate.Design and SettingsThis was a retrospective cohort study involving all children aged ≤28 days admitted to New-Born Units (NBUs) between January 2018 and December 2021 in 20 government hospitals with an interquartile range of annual NBU inpatient admissions between 550 and 1640 in Kenya. These hospitals participated in routine audit and feedback processes on quality of documentation and care over the study period.OutcomesThe study’s outcomes were the number of patients as a proportion of all eligible patients with (1) complete domain-specific documentation scores, and (2) accurate domain-specific treatment prescription scores at admission.Findings80060 NBU admissions were eligible for inclusion. Upon joining the CIN, documentation scores in the monitoring (vital signs), other physical examination and bedside testing, discharge information, and maternal history domains demonstrated a statistically significant month-to-month relative improvement in number of patients with complete documentation of 7.6%, 2.9%, 2.4%, and 2.0% respectively. There was also statistically significant month-to-month improvement in prescribing accuracy after joining the CIN of 2.8% and 1.4% for feeds and fluids but not for Antibiotic prescriptions. Findings suggest that much of the variation observed is due to hospital-level factors.ConclusionsIt is possible to introduce tools that capture important clinical data at least 80% of the time in routine African hospital settings but analyses of such data will need to account for missingness using appropriate statistical techniques. These data allow trends in performance to be explored and could support better impact evaluation, performance benchmarking, exploration of links between health system inputs and outcomes and scrutiny of variation in quality and outcomes of hospital care.

Background Medication errors are likely common in low- and middle-income countries (LMICs). In neonatal hospital care where the population with severe illness has a high mortality rate, around 14.9% of drug prescriptions have errors in LMICs settings. However, there is scant research on interventions to improve medication safety to mitigate such errors. Our objective is to improve routine neonatal care particularly focusing on effective prescribing practices with the aim of achieving reduced gentamicin medication errors. Methods We propose to conduct an audit and feedback (A&F) study over 12 months in 20 hospitals with 12 months of baseline data. The medical and nursing leaders on their newborn units had been organised into a network that facilitates evaluating intervention approaches for improving quality of neonatal care in these hospitals and are receiving basic feedback generated from the baseline data. In this study, the network will (1) be expanded to include all hospital pharmacists, (2) include a pharmacist-only professional WhatsApp discussion group for discussing prescription practices, and (3) support all hospitals to facilitate pharmacist-led continuous medical education seminars on prescription practices at hospital level, i.e. default intervention package. A subset of these hospitals (n = 10) will additionally (1) have an additional hospital-specific WhatsApp group for the pharmacists to discuss local performance with their local clinical team, (2) receive detailed A&F prescription error reports delivered through mobile-based dashboard, and (3) receive a PDF infographic summarising prescribing performance circulated to the clinicians through the hospital-specific WhatsApp group, i.e. an extended package. Using interrupted time series analysis modelling changes in prescribing errors over time, coupled with process fidelity evaluation, and WhatsApp sentiment analysis, we will evaluate the success with which the A&F interventions are delivered, received, and acted upon to reduce prescribing error while exploring the extended package’s success/failure relative to the default intervention package. Discussion If effective, these theory-informed A&F strategies that carefully consider the challenges of LMICs settings will support the improvement of medication prescribing practices with the insights gained adapted for other clinical behavioural targets of a similar nature. Trial registration PACTR, PACTR202203869312307. Registered 17th March 2022.

The objectives of this study were to (1)explore the quality of clinical data generated from hospitals providing in-patient neonatal care participating in a clinical information network (CIN) and whether data improved over time, and if data are adequate, (2)characterise accuracy of prescribing for basic treatments provided to neonatal in-patients over time. This was a retrospective cohort study involving neonates ≤28 days admitted between January 2018 and December 2021 in 20 government hospitals with an interquartile range of annual neonatal inpatient admissions between 550 and 1640 in Kenya. These hospitals participated in routine audit and feedback processes on quality of documentation and care over the study period. The study’s outcomes were the number of patients as a proportion of all eligible patients over time with (1)complete domain-specific documentation scores, and (2)accurate domain-specific treatment prescription scores at admission, reported as incidence rate ratios. 80,060 neonatal admissions were eligible for inclusion. Upon joining CIN, documentation scores in themonitoring,other physical examination and bedside testing,discharge information, andmaternal historydomains demonstrated a statistically significant month-to-month relative improvement in number of patients with complete documentation of 7.6%, 2.9%, 2.4%, and 2.0% respectively. There was also statistically significant month-to-month improvement in prescribing accuracy after joining the CIN of 2.8% and 1.4% for feeds and fluids but not for Antibiotic prescriptions. Findings suggest that much of the variation observed is due to hospital-level factors. It is possible to introduce tools that capture important clinical data at least 80% of the time in routine African hospital settings but analyses of such data will need to account for missingness using appropriate statistical techniques. These data allow exploration of trends in performance and could support better impact evaluation, exploration of links between health system inputs and outcomes and scrutiny of variation in quality and outcomes of hospital care.

Modelling healthcare providers' knowledge while they are gaining new concepts is an important step towards supporting self-regulated personalised learning at scale. This is especially important if we are to address health workforce skills development and enhance the subsequent quality of care patients receive in the Global South, where a huge skills gap exists. Rich data about healthcare providers' learning can be captured by their responses to close-ended problems within conjunctive solution space -such as clinical training scenarios for emergency care delivery- on smartphone-based learning interventions which are being proposed as a solution for reducing the healthcare skills gap in this context. Together with sequential data detailing a learner's progress while they are solving a learning task, this provides useful insights into their learning behaviour. Predicting learning or forgetting curves from representations of healthcare providers knowledge is a difficult task, but recent promising machine learning advances have produced techniques capable of learning knowledge representations and overcoming this challenge. In this study, we train a Long Short-Term Memory neural network for predicting learners' future performance and forgetting curves by feeding it sequence embeddings of learning task attempts from healthcare providers from Global South. From this training, the model captures nuanced representations of a healthcare provider's clinical knowledge and their patterns of learning behaviours, predicting their future performance with high accuracy. More significantly, by differentiating reduced performance based on spaced learning, the model can help provide timely warning that helps support healthcare providers to reinforce their self-regulated learning while providing a basis for personalised instructional support to aid improved clinical outcomes from their professional practices.

BACKGROUND Although smartphone-based emergency care training is more affordable than traditional avenues of training, it is still in its infancy, remains poorly implemented, and its current implementation modes tend to be invariant to the evolving learning needs of the intended users. In resource-limited settings, the use of such platforms coupled with gamified approaches remains largely unexplored, despite the lack of traditional training opportunities, and high mortality rates in these settings. OBJECTIVE The primary aim of this randomized experiment is to determine the effectiveness of offering adaptive versus standard feedback, on the learning gains of clinicians, through the use of a smartphone-based game that assessed their management of a simulated medical emergency. A secondary aim is to examine the effects of learner characteristics and learning spacing with repeated use of the game on the secondary outcome of individualized normalized learning gain. METHODS The experiment is aimed at clinicians who provide bedside neonatal care in low-income settings. Data were captured through an Android app installed on the study participants’ personal phones. The intervention, which was based on successful attempts at a learning task, included adaptive feedback provided within the app to the experimental arm, whereas the control arm received standardized feedback. The primary end point was completion of the second learning session. Of the 572 participants enrolled between February 2019 and July 2019, 247 (43.2%) reached the primary end point. The primary outcome was standardized relative change in learning gains between the study arms as measured by the Morris G effect size. The secondary outcomes were the participants individualized normalized learning gains. RESULTS The effect of adaptive feedback on care providers’ learning gain was found to be g=0.09 (95% CI −0.31 to 0.46; P=.47). In exploratory analysis, using normalized learning gains, when subject-treatment interaction and differential time effect was controlled for, this effect increased significantly to 0.644 (95% CI 0.35 to 0.94; P CONCLUSIONS There is a considerable learning gain between the first two rounds of learning with both forms of feedback and a small added benefit of adaptive feedback after controlling for learner differences. We suggest that linking the adaptive feedback provided to care providers to how they space their repeat learning session(s) may yield higher learning gains. Future work might explore in more depth the feedback content, in particular whether or not explanatory feedback (why answers were wrong) enhances learning more than reflective feedback (information about what the right answers are). CLINICALTRIAL Pan African Clinical Trial Registry (PACTR) 201901783811130; https://pactr.samrc.ac.za/TrialDisplay.aspx?TrialID=5836 INTERNATIONAL REGISTERED REPORT RR2-10.2196/13034

Research is lacking on developing adaptive learning applications for training health workers in low-resource settings making student modelling approaches supporting individualised learning to remain largely unexplored. This study targeted a clinical training intervention using smartphones in a low-resource context to explore if clinicians’ performance patterns can be differentiated into distinctive groups based on an inferred proficiency level using cluster analysis. We also explored the applicability of Knowledge-Component (KC) cognitive learning models-Additive and Performance Factor Models (AFMs, PFMs) - in describing these patterns and their accuracy in predicting performance. The intervention provides simulation training on contextualised management of new-born resuscitation through a series of learning interactions that elicit responses through multiple-choice answers and interactive tasks. AFMs and PFMs were used to explore the impact of previous exposure to KCs within the learning intervention on learner performance. We demonstrate that effectiveness of low-dose-high-frequency training might be linked to successful attempts in previous learning sessions. Additionally, there exists intermediate and expert cadres of health workers who would benefit more from cascading-challenge scenarios. From these results, we propose a preliminary cognitive learning model as a basis for adaptive instructional support on smartphones for clinical training in low-resource settings.

BACKGROUND Although smartphone-based clinical training to support emergency care training is more affordable than traditional avenues of training, it is still in its infancy and remains poorly implemented. In addition, its current implementations tend to be invariant to the evolving learning needs of the intended users. In resource-limited settings, the use of such platforms coupled with serious-gaming approaches remain largely unexplored and underdeveloped, even though they offer promise in terms of addressing the health workforce skill imbalance and lack of training opportunities associated with the high neonatal mortality rates in these settings. OBJECTIVE This randomized controlled study aims to assess the effectiveness of offering adaptive versus standard feedback through a smartphone-based serious game on health care providers’ knowledge gain on the management of a neonatal medical emergency. METHODS The study is aimed at health care workers (physicians, nurses, and clinical officers) who provide bedside neonatal care in low-income settings. We will use data captured through an Android smartphone-based serious-game app that will be downloaded to personal phones belonging to the study participants. The intervention will be adaptive feedback provided within the app. The data captured will include the level of feedback provided to participants as they learn to use the mobile app, and performance data from attempts made during the assessment questions on interactive tasks participants perform as they progress through the app on emergency neonatal care delivery. The primary endpoint will be the first two complete rounds of learning within the app, from which the individuals’ “learning gains” and Morris G intervention effect size will be computed. To minimize bias, participants will be assigned to an experimental or a control group by a within-app random generator, and this process will be concealed to both the study participants and the investigators until the primary endpoint is reached. RESULTS This project was funded in November 2016. It has been approved by the Central University Research Ethics Committee of the University of Oxford and the Scientific and Ethics Review Unit of the Kenya Medical Research Institute. Recruitment and data collection began from February 2019 and will continue up to July 31, 2019. As of July 18, 2019, we enrolled 541 participants, of whom 238 reached the primary endpoint, with a further 19 qualitative interviews conducted to support evaluation. Full analysis will be conducted once we reach the end of the study recruitment period. CONCLUSIONS This study will be used to explore the effectiveness of adaptive feedback in a smartphone-based serious game on health care providers in a low-income setting. This aspect of medical education is a largely unexplored topic in this context. In this randomized experiment, the risk of performance bias across arms is moderate, given that the active ingredient of the intervention (ie, knowledge) is a latent trait that is difficult to comprehensively control for in a real-world setting. However, the influence of any resulting bias that has the ability to alter the results will be assessed using alternative methods such as qualitative interviews. CLINICALTRIAL Pan African Clinical Trials Registry PACTR201901783811130; https://pactr.samrc.ac.za/TrialDisplay. aspx?TrialID=5836 INTERNATIONAL REGISTERED REPOR PRR1-10.2196/13034

Summary Background Management of pneumonia in many low-income and middle-income countries is based on WHO guidelines that classify children according to clinical signs that define thresholds of risk. We aimed to establish whether some children categorised as eligible for outpatient treatment might have a risk of death warranting their treatment in hospital. Methods We did a retrospective cohort study of children aged 2–59 months admitted to one of 14 hospitals in Kenya with pneumonia between March 1, 2014, and Feb 29, 2016, before revised WHO pneumonia guidelines were adopted in the country. We modelled associations with inpatient mortality using logistic regression and calculated absolute risks of mortality for presenting clinical features among children who would, as part of revised WHO pneumonia guidelines, be eligible for outpatient treatment (non-severe pneumonia). Findings We assessed 16 162 children who were admitted to hospital in this period. 832 (5%) of 16 031 children died. Among groups defined according to new WHO guidelines, 321 (3%) of 11 788 patients with non-severe pneumonia died compared with 488 (14%) of 3434 patients with severe pneumonia. Three characteristics were strongly associated with death of children retrospectively classified as having non-severe pneumonia: severe pallor (adjusted risk ratio 5·9, 95% CI 5·1–6·8), mild to moderate pallor (3·4, 3·0–3·8), and weight-for-age Z score (WAZ) less than −3 SD (3·8, 3·4–4·3). Additional factors that were independently associated with death were: WAZ less than −2 to −3 SD, age younger than 12 months, lower chest wall indrawing, respiratory rate of 70 breaths per min or more, female sex, admission to hospital in a malaria endemic region, moderate dehydration, and an axillary temperature of 39°C or more. Interpretation In settings of high mortality, WAZ less than −3 SD or any degree of pallor among children with non-severe pneumonia was associated with a clinically important risk of death. Our data suggest that admission to hospital should not be denied to children with these signs and we urge clinicians to consider these risk factors in addition to WHO criteria in their decision making. Funding Wellcome Trust.

Background Childhood pneumonia is the leading infectious cause of mortality in children younger than 5 years old. Recent updates to World Health Organization pneumonia guidelines recommend outpatient care for a population of children previously classified as high risk. This revision has been challenged by policymakers in Africa, where mortality related to pneumonia is higher than in other regions and often complicated by comorbidities. This study aimed to identify factors that best discriminate inpatient mortality risk in non-severe pneumonia and explore whether these factors offer any added benefit over the current criteria used to identify children with pneumonia requiring inpatient care. Methods We undertook a retrospective cohort study of children aged 2–59 months admitted with a clinical diagnosis of pneumonia at 14 public hospitals in Kenya between February 2014 and February 2016. Using machine learning techniques, we analysed whether clinical characteristics and common comorbidities increased the risk of inpatient mortality for non-severe pneumonia. The topmost risk factors were subjected to decision curve analysis to explore if using them as admission criteria had any net benefit above the current criteria. Results Out of 16,162 children admitted with pneumonia during the study period, 10,687 were eligible for subsequent analysis. Inpatient mortality within this non-severe group was 252/10,687 (2.36%). Models demonstrated moderately good performance; the partial least squares discriminant analysis model had higher sensitivity for predicting mortality in comparison to logistic regression. Elevated respiratory rate (≥70 bpm), age 2–11 months and weight-for-age Z-score (WAZ)

In many low income countries health information systems are poorly equipped to provide detailed information on hospital care and outcomes. Information is thus rarely used to support practice improvement. We describe efforts to tackle this challenge and to foster learning concerning collection and use of information. This could improve hospital services in Kenya. We are developing a Clinical Information Network, a collaboration spanning 14 hospitals, policy makers and researchers with the goal of improving information available on the quality of inpatient paediatric care across common childhood illnesses in Kenya. Standardised data from hospitals’ paediatric wards are collected using non-commercial and open source tools. We have implemented procedures for promoting data quality which are performed prior to a process of semi-automated analysis and routine report generation for hospitals in the network. In the first phase of the Clinical Information Network, we collected data on over 65,000 admission episodes. Despite clinicians’ initial unfamiliarity with routine performance reporting, we found that, as an initial focus, both engaging with each hospital and providing them information helped improve the quality of data and therefore reports. The process has involved mutual learning and building of trust in the data and should provide the basis for collaborative efforts to improve care, to understand patient outcome, and to evaluate interventions through shared learning. We have found that hospitals are willing to support the development of a clinically focused but geographically dispersed Clinical Information Network in a low-income setting. Such networks show considerable promise as platforms for collaborative efforts to improve care, to provide better information for decision making, and to enable locally relevant research.