Observations and in-depth interviews.

In total, 23 observations and in-depth interviews (IDIs) were conducted at Zomba Central Hospital’s paediatric HDU, with healthcare providers (n = 14) and caregivers (n = 9) of nine critically ill children (aged 3 months to 5 years old) admitted to the HDU. After introducing the IMPALA monitors, study participants (healthcare providers and caregivers) were conveniently recruited for structured observations and in-depth interviews between November 9, 2022, and October 30, 2023. Healthcare providers of different cadres (nurses, clinical officers and medical doctors) working in HDU were recruited (face-to-face) during their day shifts, representing a mix of professional roles and levels of experience. Selection was based on their involvement in paediatric care, treatment and management. To enhance diversity and representativeness, caregivers of critically ill children with a range of diagnoses and socioeconomic status were also recruited during their admission to HDU. The diversity sample of critically ill children was selected based on the following criteria: age (28 days to 60 months) and categorised as (a) febrile respiratory (e.g., pneumonia, bronchiolitis and asthma), (b) febrile non-respiratory (e.g., malaria and anaemia) or (c) neurological (e.g., head injuries) conditions. These conditions were selected because they are prevalent in paediatric populations in Malawi and have varied typical progressions during admission. Caregivers included anyone accompanying the admitted children during their hospital stay. Caregivers were selected based on being above eighteen years of age and fulfilling a caregiving role. This approach ensured capturing the lived experience of providing care for critically ill children. The recruitment of participants continued until saturation, i.e., when no new or additional information was forthcoming.

Questionnaire-based survey.

The sampling frame for the questionnaire-based caregiver survey consisted of all caregivers of critically ill children aged three months to five years who were admitted to the HDU between January 1 and June 30, 2023, and consented to participate in the IMPALA clinical observational study. Children under the age of five had the highest likelihood of presenting with critical illness, and this was the population with the highest mortality. Clinical deterioration was less evident in older age groups, and disease progression develops more rapidly in younger children compared to older ones [16]. Study nurses identified eligible caregivers in collaboration with clinical staff, and additional informed consent was sought to participate in the questionnaire-based survey upon their child’s discharge from HDU. Caregivers whose child died in HDU did not participate in the surveys. A maximum of 110 caregivers were expected to be interviewed, dependent on the number of admissions and caregiver consent during the data collection period. The sampling frame for the questionnaire-based healthcare provider survey comprised all healthcare workers (i.e., the complete census of nurses and clinicians) working in the HDU from January to May 2023. We estimated the minimum detectable effect size of the change in hours spent on monitoring children before and after IMPALA for the sample of n = 24 healthcare workers, based on a one-sample mean t-test power calculation with alpha = 0.05, beta = 0.80 and a standard deviation of 1.1025, to be 0.597. Though the sample size was small, it was appropriate for the exploratory, early-stage nature of this study. It was chosen to balance statistical power with feasibility in a resource-constrained setting.

Observations.

Two structured observation guides (for children admitted to the HDU and healthcare providers) were used to collect data on various areas of critical care in the HDU, including diagnosis, communication, frequency of monitoring patient vital signs, use of monitors, response to perturbations, treatment rendered, and other tasks. Observations were conducted throughout the day shift, focusing on critically ill children and their caregivers from the point of admission to the HDU until discharge or death. Attending healthcare providers were also observed throughout the entire day shift to capture their interactions, workflows, and use of the monitoring system. A social scientist (DM) conducted the observations with support from the research nurse (AL). The researchers observed each child participant for eight hours and systematically recorded their observations using the observation guides. Structured observations were conducted with ten healthcare providers and nine children and caregivers. Over the course of 40 days, a total of 320 hours of observations were completed.

In-depth interviews.

The study team developed interview guides for healthcare providers and caregivers based on the CFIR. These guides covered topics such as experiences during admission, vital signs monitoring processes, and various constructs from the CFIR. The semi-structured guides included open-ended questions that were used flexibly and adapted during data collection. To ensure accessibility, the interview guides and consent forms were translated into the local language by a native Chichewa speaker. Initially, study nurses recruited caregivers of critically ill children admitted to the HDUs who consented to participate in the IMPALA study. Subsequently, DM obtained separate consent for the interviews and conducted them a day after the observation was completed. All interviews with caregivers were conducted in Chichewa, the local language, to facilitate understanding and expression. Some interviews with healthcare providers were conducted in English or Chichewa, depending on the provider’s preference. They took place in a private room (such as a nurse’s restroom) or a location selected by participants and lasted thirty to forty-five minutes. Healthcare workers were interviewed during lunch breaks or at the end of their shifts to minimise disruption to their daily tasks.

The questionnaires.

The self-administered EMPATHIC-65 questionnaire served as the basis for the quantitative survey tool for caregivers [17]. The EMPATHIC-65 questionnaire was initially developed and validated in the Netherlands in a cohort of 3,354 parents of children admitted to eight Dutch PICUs and subsequently adapted and validated in South Africa [18]. The EMPATHIC-65 questionnaire includes sections on general information about PICU patients, parents’ satisfaction with the quality of HDU care, and their experiences in the PICU, which were adapted to the study context and translated from English to Chichewa. The questionnaire was expanded to encompass caregiver experiences during admission and vital signs monitoring. The revised questionnaire was extensively piloted among caregivers and further adjusted before data collection. The survey instrument for healthcare providers was specifically designed for this study and included sections on provider background characteristics, training, and current role in the HDU; daily tasks and time management; and stock outs of medications and medical supplies. The study team administered the tablet-based questionnaire to the healthcare providers in their preferred language, either English or Chichewa. The caregiver survey was always administered in Chichewa.

Qualitative data.

The first author, DM – a native Chichewa speaker, conducted the in-depth interviews, which were recorded using a digital audio recorder. Subsequently, the audio-recorded data was translated directly from Chichewa to English by an experienced, independent and bilingual transcriber and native Chichewa speaker. Following transcription, DM meticulously reviewed the transcripts to ensure accuracy by cross-referencing them with the original audio recordings, confirming that the participants’ intended meanings were accurately captured. Any information that could identify the study participants was removed from the transcripts. In addition to transcription, DM maintained a field diary, documenting notes on data collection challenges and emerging issues. The audio recordings and complete translated data were securely stored on a password-protected computer. DM listened to the audio recordings and reviewed the transcripts multiple times to gain a comprehensive understanding of the issues raised. Familiarising himself with the entire dataset, DM ensured the data was clean and coherent.

The processes began with importing transcripts from Word documents into NVivo QSR version 14 software (QSR International, Victoria, Australia) for organisation, management, and thematic analysis. Subsequently, DM and LMT collaborated to define terms and concepts extracted from three transcripts, and CP validated them. DM undertook the coding process for all datasets using the agreed-upon coding framework. This collaborative effort guided the development of the codebook utilised for thematic analysis. DM and LMT utilised coding definitions from the Consolidated Framework for Implementation Research (CFIR) during the thematic process, as depicted in S3 Fig. DM systematically coded the textual data, capturing the participants’ viewpoints and narratives within the provided framework. This framework offered systematic guidance for assessing the facilitators, barriers, challenges, and opportunities for implementing IMPALA monitors in Malawi.

Quantitative data.

Frequencies and percentages were calculated for categorical variables; means and associated standard deviations (SDs) were estimated for continuous variables. The analyses include inferential statistics (95% confidence intervals) for all variables, as well as comparative statistics for the before-after difference in hours worked on child monitoring. Analyses were conducted using STATA version 17.

In-depth interviews.

In total, 23 in-depth interviews (IDIs) were conducted at Zomba Central Hospital’s paediatric HDU, with healthcare providers (n = 14) and caregivers (n = 9) of nine critically ill children (aged 3 months to 5 years old) admitted to the HDU. Of the 14 healthcare providers interviewed, 7 were nurse midwife technicians, and 5 were nursing officers. Most healthcare providers were female (n = 10). The nine caregivers interviewed were all female and mostly aged between 20 and 30 years (n = 6). S1 Table gives a detailed overview of IDI respondent characteristics.

Continuous measurement of multiple vital signs.

Healthcare providers identified key aspects of the IMPALA monitor’s design quality that support its use. A major strength was the continuous measurement of multiple vital signs, including heart rate, respiratory rate, oxygen saturation, and blood pressure.. Healthcare providers appreciated that these parameters were readily accessible during clinical assessments and ward rounds. The monitors’ wall-mounted setup and the ability to remain continuously connected to critically ill children further enhanced ease of use and supported ongoing monitoring throughout the day and night.

“We can assess the condition of the patients very easily by looking at the parameters like the heart rate, respiratory rate, oxygen saturation, and blood pressure. Even at a glance at the monitors, you can tell this patient is saturating or desaturating. In addition, it allows you to make decisions in good time”. (Nurse In charge IDI 06)

However, participants expressed concerns that the need for temperature sensors on the IMPALA monitors was a limiting factor in measuring vital signs.

“I have noticed that the IMPALA monitors do not have temperature sensors, which would be very important. We monitor patients’ temperature, and it will be good to have it on the monitor, boosting the IMPALA monitors’ usage”. (Nurse IDI 01)

Easy to read.

The healthcare providers described the IMPALA monitors as very easy to read due to the colours (blue, yellow, and red) and the minimal amount of distracting information. They could quickly and easily collect data at any time without wasting time searching for the monitors or attaching children to them.

“The IMPALA monitor is easier to use because it is stationary, uncomplicated & has colours such as red, yellow and green and alarms to alert you when the condition of a child is deteriorating. We can use it anytime we want, unlike the other mobile ones. In addition, we can monitor the children well without any interruptions”. (Nurse IDI 07)

Long-lasting backup power (up to 4 hours).

Healthcare providers appreciated IMPALA’s built-in backup battery, which provides up to four hours of power during outages. In the HDU ward, occasional voltage fluctuations did not disrupt monitoring, as the device continued to function seamlessly. In contrast, conventional monitors, which rely solely on external power, ceased operating during power cuts, making IMPALA more reliable in this setting.

“With the IMPALA monitor, there is continuous monitoring even when there is a power cut, and they keep memories, unlike the mobile monitors where there is intermittent monitoring” (Nurse IDI 01)

Alarms.

Healthcare providers and caregivers viewed the alarm feature as crucial in the clinical workflow. Alarms enabled early detection of patient deterioration, prompting timely intervention. This feature also reduced the time and effort required to locate pulse oximeters or other mobile devices during emergencies, easing the workload of healthcare providers.

“When the alarm sounded, the measurement for air (oxygen) showed that the oxygen saturation was dropping to 89. So when it happened like this, I called for the nurse, and they could increase the air (oxygen levels). I was calling the nurse to check my child”. (Caregiver IDI 02)

However, false alarms prompted nurses to ignore the monitors when they beeped, as they had challenges distinguishing between false alarms and real alarms. Indeed, approximately half (49%) of the caregivers in the survey reported that they did not hear the alarms sounding during their child’s stay on the HDU (Table 1). During observations, it was also noted that the nurses changed the alarm thresholds to high or low numbers or turned off the alarm to minimise beeping.

“I feel like the false alarms should not be there because they make us busy by making us check the monitors occasionally. We could spend the whole day in the HDU trying to check the monitors, which was tiresome on our job when they made the sound, especially when more patients were connected. Therefore, the alarms should beep only when the patient is deteriorating or the probes have disconnected (Study nurse IDI 14).

Reliability of measurements.

The IMPALA monitors increased confidence among healthcare providers by providing accurate, reliable vital sign data to support clinical decision-making, such as adjusting oxygen delivery and determining the child’s condition. Nevertheless, the healthcare providers emphasised the continued need to apply clinical expertise and to interpret and monitor outputs in conjunction with the patient’s present clinical presentation.

“I think it can work anywhere because it is a monitoring machine, and it’s effective. It shows all parameters like BP pulse rate, respiration, and oxygen saturation; these are the parameters we look for in a patient. These parameters serve as baseline data to help us determine the appropriate action to take. So, I believe it can work in all the HDUs.” (Clinician IDI 09).

However, technical issues such as sensors falling off, the absence of temperature sensors, false readings due to child movement, monitor freezing, and incorrect high blood pressure values undermined the monitors’ reliability. This led to decreased engagement with the device among clinical staff.

“The disadvantage of the IMPALA monitors is that the sensors or probes easily get disconnected when the child is too mobile, leading to false results. This affects our work, and sometimes we lose trust in the monitors. Lack of temperature sensor is also a challenge that IMPALA monitors have, affecting our work”. (Nurse IDI 07)

Human resource.

Healthcare providers reported that the availability of IMPALA monitors simplified the task of assessing and collecting vital signs for multiple critically ill children (eight to twelve critically ill children), often managed by a single nurse. The monitors enabled quick identification and prioritisation of patients in need of urgent care, particularly during assessments and ward rounds. The healthcare provider questionnaire-based surveys echo this finding. Staff reported spending less time on monitoring children, 1.8 hours daily during the IMPALA implementation pilot, compared to 3.3 hours pre-IMPALA.

“Firstly, it eases our work, especially nurses, since we are few. So, assessing the patients individually becomes a challenge. It happens that when you are busy attending to one patient, you might not be aware of how critical the other patient is and in need of haste treatment; hence, with the IMPALA monitor, it helps to prioritise which patient is to be attended to first depending on how critical the patient is”. (Nurse IDI 08)

Fit in clinical workflow (compatibility).

Healthcare providers and caregivers reported that IMPALA monitors integrated well into clinical workflows by providing real-time, reliable, and non-invasive vital signs data. This continuous monitoring supported timely decision-making, such as discharging stable patients. It enabled more efficient management of critically ill children, as nurses could quickly identify changes in the child’s condition.

“What I like about the IMPALA monitors the most is that they ease nurses’ work and are not painful when connected to our children. Instead of the nurse moving the monitors to get the vital signs, they record the data on the monitors. There was a child who was not connected to the monitor, so every time the nurse wanted to check the vital signs, they had to move the monitor from another patient to this one to record the data. But with my child, it was easy to collect the data because my child was connected to the monitor throughout”. (Caregiver IDI 07)

Training on technology.

Healthcare providers reported receiving training on using the IMPALA monitors, including how to interpret vital signs data and connect sensors to patients. This training enhanced their ability to efficiently collect and document vital signs on critical care pathway charts. However, some healthcare providers indicated that the training was limited, particularly in hands-on practice and technical features like navigating the scroll menu. Most relied on informal, on-the-job learning, which was considered insufficient, especially in the context of frequent staff rotations in the HDU. Nurses who joined after the system’s introduction often had only brief, once-off orientation sessions and lacked confidence in operating the monitors (connecting or disconnecting children to the monitors). The surveys revealed that 66.7% of healthcare providers had no critical or intensive care training, 95.8% lacked paediatric critical or intensive care training, and 70.8% were not trained in paediatric HDU monitoring before the IMPALA.

“OK, I think if there are opportunities for briefing or refresher training on IMPALA, they need to include people who were left out during the previous briefing and staff who have recently just joined. If that happens, it will be better because most staff will be knowledgeable” (Clinician IDI 10)

This lack of training hindered some healthcare providers from effectively setting up and adjusting the settings on the IMPALA monitor, leading to a lack of confidence in its use and the inability to connect the children to the monitors.

Trust and mistrust.

Some caregivers perceived the monitors as helpful in detecting a child’s prognosis and beneficial for escalating deterioration to healthcare providers. They considered the availability of different colours (green, yellow and red) helpful in alerting and informing about the child’s progress. Four out of nine female of caregivers interviewed believed that connecting critically ill children to these devices could worsen their condition and lead to death. They associated the continuous monitoring with superstitions and fears, holding the misconception that children connected to these devices were more likely to die. They reported that the most severely ill children on the monitors had a higher likelihood of dying. This belief extended to the misconception that oxygen therapy, when combined with the monitors, was a way of hastening a child’s death, which intensified their anxiety and reluctance to participate in the study.

“Other people said that IMPALA is destroying children. They claim that this causes children not to recover quickly or, even when they do, they do not wait long before returning to the HDU for admission and die”. (Caregiver IDI 09).

However, when such caregivers were asked about these beliefs, most acknowledged their fears about the potential adverse effects of using the monitors. Still, they found it difficult to align their concerns with the evidence that indicates that monitors may harm children. Over time, as they observed the monitors’ positive impact on the quality of patient care during admissions, their perceptions began to shift. This helped to dispel their initial fears and facilitated their acceptance of the devices.

“At first, I thought that the monitors were dangerous and could kill my child, but I noted that the monitors did not bring any harm to the child and helped my child to receive good care and support during hospitalisation”. (Caregiver IDI 02)