Menu
Menu
Menu
Menu
Objectives To characterise adoption and explore specific clinical and patient factors that might influence pulse oximetry and oxygen use in low-income and middle-income countries (LMICs) over time; to highlight useful considerations for entities working on programmes to improve access to pulse oximetry and oxygen. Design A multihospital retrospective cohort study. Settings All admissions (n=132 737) to paediatric wards of 18 purposely selected public hospitals in Kenya that joined a Clinical Information Network (CIN) between March 2014 and December 2020. Outcomes Pulse oximetry use and oxygen prescription on admission; we performed growth-curve modelling to investigate the association of patient factors with study outcomes over time while adjusting for hospital factors. Results Overall, pulse oximetry was used in 48.8% (64 722/132 737) of all admission cases. Use rose on average with each month of participation in the CIN (OR: 1.11, 95% CI 1.05 to 1.18) but patterns of adoption were highly variable across hospitals suggesting important factors at hospital level influence use of pulse oximetry. Of those with pulse oximetry measurement, 7% (4510/64 722) had hypoxaemia (SpO 2 <90%). Across the same period, 8.6% (11 428/132 737) had oxygen prescribed but in 87%, pulse oximetry was either not done or the hypoxaemia threshold (SpO 2 <90%) was not met. Lower chest-wall indrawing and other respiratory symptoms were associated with pulse oximetry use at admission and were also associated with oxygen prescription in the absence of pulse oximetry or hypoxaemia. Conclusion The adoption of pulse oximetry recommended in international guidelines for assessing children with severe illness has been slow and erratic, reflecting system and organisational weaknesses. Most oxygen orders at admission seem driven by clinical and situational factors other than the presence of hypoxaemia. Programmes aiming to implement pulse oximetry and oxygen systems will likely need a long-term vision to promote adoption, guideline development and adherence and continuously examine impact.
The reporting of this observational study follows the Strengthening the Reporting of Observational Studies in Epidemiology statement.16 The Scientific and Ethics Review Unit of the Kenya Medical Research Institute (KEMRI) approved the collection of the deidentified data that provides the basis for this study as part of the Clinical Information Network (CIN). The CIN is run in partnership with the Ministry of Health and participating hospitals with aims to improve the quality of routine paediatric hospital data for use in improvement activities, observational and interventional research.17 18 Individual consent for access to deidentified patient data was not required. No patients were involved in the design, conduct, reporting or dissemination plans of our research except through the KEMRI ethical review process where they have representatives. From a broad context perspective, in many LMICs including Kenya, hospital management and monitoring systems are weak, with major human and material resource constraints. These challenges affect hospitals’ delivery of inpatient maternal, surgical and adult medical care as well as paediatric and neonatal care. Consequently, there is very limited organisational and resource slack to mobilise for any new purpose. Interventions such as oximetry and oxygen seeking to achieve large scale change must therefore either consider how to mobilise new resources or consider what is achievable with limited resources.19 20 We have described the broader context of the Kenyan healthcare system with reference to the paediatric burden of disease in great detail elsewhere.19 20 We report a retrospective cohort study of 18 public hospitals in Kenya largely providing first-referral level care predominantly admitting patients that present directly to the facility, and purposefully selected to be of at least moderate size and representative of different malaria transmission zones. Hospitals joined the CIN at different calendar time points between 2013 and 2017. Few of their patients may be formally referred from primacy care facilities but there are few functional referral mechanisms such as ambulance systems.21 Pulse oximetry has generally not been available outside hospitals in the public sector.22 Pneumonia is the major killer of children across the country except in settings where malaria is highly endemic (9/18 hospitals in this study).21 The hospitals receive 3 monthly clinical audit and feedback reports on the quality of care they provide for common conditions.23 Paediatric team leaders (paediatricians and nurses) met face to face once or twice annually until 2019 (before the pandemic) to discuss these reports and how to improve clinical care. From 2018, participation in more specific research studies was also discussed in meetings with hospitals. This resulted in two studies being initiated in subsets of the CIN hospitals which might have influenced the adoption of pulse oximetry. Hospital participation in these studies was not mutually exclusive: a hospital could be part of one, both or none of the research studies. Not all hospitals recruited in the same study were recruited at the same time. Summary details of these studies are provided in table 1 with greater detail on which hospitals participated in the studies provided in online supplemental tables 1 and 2. The 18 hospitals included in the study had a median of 1 pulse oximeter(s) per paediatric ward (IQR: 1–3) (online supplemental table 1). Description of studies undertaken in Clinical Information Network hospitals H15 got one pulse oximeter in June 2020 even though it was not part of any study. *We were not able to track and record introduction of pulse oximeters procured directly by hospitals or other sources, or those personally owned by healthcare workers, nor ability to repair existing pulse oximeters when required. MoH, Ministry of Health. bmjopen-2021-050995supp001.pdf Recording of pulse oximetry values has been included as part of a structured paediatric admission record (PAR)24 used since 2013. Hospitals joining the CIN agree to provide the PAR themselves and promote its use—the purpose of the PAR is to prompt admitting clinicians to fully assess children and rapidly document their findings.24 Emergency Triage Assessment and Treatment plus Admission Care (ETAT+) training25 has been used in Kenya since 2008 and was adapted in 2013 so that pulse oximetry was a recommended part of the assessment in all sick children and especially those with danger signs. Many of the junior and senior medical staff in CIN hospitals would have received ETAT+training (eg, as an undergraduate or postgraduate course or as in-service training).25 Additionally, the Basic Paediatric Protocols that are widely disseminated have specifically referred to the use of pulse oximetry—if available—for all pneumonia cases since 2016.26 The promotion of PAR use by hospitals, adaption and scale up of ETAT+training, and dissemination of the basic paediatric protocols to clinicians nationally are posited to have a system-wide effect on the adoption of pulse oximetry use. All medical admissions to paediatric wards (ages 0–13 years), but not those admitted to specific newborn units, to the selected hospitals between 1 February 2014, or the date at which a hospital joined CIN, and 31 December 2020 were eligible for inclusion in this study. We excluded children whose admission or discharge dates were missing or improbable (eg, discharge date is earlier than admission date), and those whose admission fell within 2017, a period of prolonged health worker strikes that resulted in major disruption to healthcare delivery.12 The expectation was that all would have pulse oximetry performed at admission given they were sick enough to warrant inpatient care but among these we also report the proportion of children with danger signs comprising any level of altered consciousness or with recorded signs of respiratory distress especially deserving of pulse oximetry.10 26 27 Methods of collection and cleaning of data in the CIN are reported in detail elsewhere.28 Clinical data for paediatric admissions to the hospitals within the CIN are captured through PAR forms27 that are approved by the Ministry of Health. The PAR prompts the clinician with a checklist of fields including patient biodata, clinical assessment, admission and discharge diagnoses, treatments and to record outcome (survival or death). The CIN supports one data clerk in each hospital to extract data from paper medical records, nursing charts, treatment charts and available laboratory reports each day after children’s discharge into the primary data collection tool developed in Research Electronic Data Capture. Automated error checking happens at the point of entry by daily review, every week centrally and both are complemented by regular external data quality assurance reviews.28 A minimal dataset—which is unsuitable for these pulse oximetry analyses—is collected for (1) all admissions with a burn or a surgical diagnosis to the paediatric ward(s), (2) admissions during major holiday breaks, (3) admissions when the data clerk was on leave and (4) on a random selection of records in hospitals where the workload is very high. This process is explained in detail elsewhere.28 29 To characterise adoption of pulse oximetry at admission over time, we use pooled data from the period February 2014 to November 2018 (excluding 2017) from 13 network hospitals. From December 2018 data were available from 17 hospitals after 5 joined the CIN linked to new studies, and one (H7) exited. To explore patterns in more detail, we employed two approaches: (1) for the period from December 2018 to December 2020, we plot data pooled for hospital subgroups depending on which research studies they were taking part in as these might influence pulse oximetry use; (2) we plot adoption within each hospital for the number of months it was part of the network, again excluding 2017 where relevant. This latter approach was used to reveal the extent of variability in adoption at the hospital level. For the individual patient population, we tabulated and summarised categorical data as proportions, with continuous variables reported as medians (IQR) if they were not normally distributed. From the included paediatric admission patient population data, we report the prevalence of important clinical features such as respiratory distress symptoms (eg, central cyanosis, indrawing, grunting and difficulty breathing), circulatory symptoms (eg, slow capillary refill, less than three seconds), level of consciousness, and whether the patient was being readmitted or referred to the hospital. We also report the prevalence of summary clinical features such as having any sign of respiratory distress and any danger sign and the proportion for whom oxygen was ordered as a treatment at admission. To examine two binary outcomes, (1) pulse oximetry used and, (2) oxygen ordered at admission, we modelled individual patient data. Our key explanatory variable was time as part of (or exposure to) the CIN, computed as the number of months since joining the CIN for a specific hospital. This allowed associations with these outcomes to be explored in hospitals while accounting for clustering by both hospital and time in months since the hospital joined the CIN; the analysis of pulse oximetry use employs a much larger dataset to extend our earlier findings.10 Predictors included in our models were patients’ age and sex, and the record of presence or absence of respiratory signs and symptoms, circulatory symptoms, level of consciousness, whether the patient was being readmitted or referred to the hospital. At the hospital level, we included terms for their specific identity and malaria endemicity zone. Hospital participation in one of the four research study subgroups was included as a covariable at the patient level to take account of possible effects occurring only after a hospital joined specific studies. We modelled whether oxygen was prescribed at admission with an interest in whether patients with a recorded pulse oximetry value (SpO2) <90% at admission would be started on oxygen therapy. Where the pulse oximetry value at admission was missing, based on findings from a previous CIN study, the assumption was that the clinician did not have the information.10 The analysis therefore explored (1) whether pulse oximetry values (SpO2 <90%) were associated with oxygen prescription and (2) in the absence of a pulse oximetry value <90%, which clinical signs are associated with the prescription of oxygen therapy. We included the variables of interest in hierarchical multivariate logistic regression models, with a random intercept for each hospital and a random slope for the time within hospital. This model specification commonly referred to as a growth curve model, allowed the explanatory variable of time to have a different effect for each hospital. A growth curve model typically refers to statistical methods that allow, in our case, the estimation of interhospital variability in intrahospital patterns of change over time.30 Our approach to the growth-curve model fitting is within the multilevel modelling framework, with patients nested in hospitals nested in time points.30 Different ways of specifying the growth-curve model using a multilevel modelling framework are explained in detail elsewhere.31 Before embarking on the growth-curve modelling, we examined for each hospital whether the 2017 strike resulted in a significant discontinuity in pulse oximetry use to check if treating time as months in the CIN, a continuous variable, instead of as calendar time was justifiable given the strike caused a 12-month data gap. Where there were missing patient-level data, we applied fully conditional specification multivariate imputation by chained equations32 to generate imputed datasets under the missing at random (MAR) assumption to allow us to analyse all eligible patients. Previous studies indicate that MAR is a reasonable approach for CIN data.33 We contrasted findings using imputation with the findings of the complete case analysis.
