The impact of pulse oximetry and Integrated Management of Childhood Illness (IMCI) training on antibiotic prescribing practices in rural Malawi: A mixed-methods study.

BACKGROUND: The misdiagnosis of non-malarial fever in sub-Saharan Africa has contributed to the significant burden of pediatric pneumonia and the inappropriate use of antibiotics in this region. This study aims to assess the impact of 1) portable pulse oximeters and 2) Integrated Management of Childhood Illness (IMCI) continued education training on the diagnosis and treatment of non-malarial fever amongst pediatric patients being treated by the Global AIDS Interfaith Alliance (GAIA) in rural Malawi.
METHODS: This study involved a logbook review to compare treatment patterns between five GAIA mobile clinics in Mulanje, Malawi during April-June 2019. An intervention study design was employed with four study groups: 1) 2016 control, 2) 2019 control, 3) IMCI-only, and 4) IMCI and pulse oximeter. A total of 3,504 patient logbook records were included based on these inclusion criteria: age under five years, febrile, malaria-negative, and treated during the dry season. A qualitative questionnaire was distributed to the participating GAIA providers. Fisher's Exact Testing and odds ratios were calculated to compare the prescriptive practices between each study group and reported with 95% confidence intervals.
RESULTS: The pre- and post-exam scores for the providers who participated in the IMCI training showed an increase in content knowledge and understanding (p<0.001). The antibiotic prescription rates in each study group were 75% (2016 control), 85% (2019 control), 84% (IMCI only), and 42% (IMCI + pulse oximeter) (p<0.001). An increase in pneumonia diagnoses was detected for patients who received pulse oximeter evaluation with an oxygen saturation <95% (p<0.001). No significant changes in antibiotic prescribing practices were detected in the IMCI-only group (p>0.001). However, provider responses to the qualitative questionnaires indicated alternative benefits of the training including improved illness classification and increased provider confidence.
CONCLUSION: Clinics that implemented both the IMCI course and pulse oximeters exhibited a significant decrease in antibiotic prescription rates, thus highlighting the potential of this tool in combatting antibiotic overconsumption in low-resource settings. Enhanced detection of hypoxia in pediatric patients was regarded by clinicians as helpful for identifying pneumonia cases. GAIA staff appreciated the IMCI continued education training, however it did not appear to significantly impact antibiotic prescription rates and/or pneumonia diagnosis.

Background and significance

While the diagnosis and treatment of malaria has greatly improved due to the advent of rapid diagnostic tests, non-malarial causes of fever still represent a significant burden of disease throughout sub-Saharan Africa [1]. Within this region, non-malarial fever (NMF) is critical to address in Malawi, particularly amongst patients under the age of five [2]. Of the many causes of NMF in Malawi, pneumonia accounts for 13% of deaths amongst children under the age of five, and is thus essential to accurately diagnose and treat [3]. It is estimated that one in five pediatric deaths due to pneumonia could be avoided if providers showed stronger adherence to existing diagnostic guidelines and interventions [4]. Defined as an of the alveoli, pneumonia typically presents with cough, fatigue, chest tightness, fever, sweating, and shortness of breath [5]. While the gold standard of pneumonia diagnosis usually involves the use of chest x-rays and/or sputum tests, the luxury of this diagnostic equipment is rarely afforded to those living in low-resource settings [6].

Not only does the inaccurate diagnosis of pneumonia lead to adverse patient outcomes, but it also poses a major public health concern due to the subsequent overuse of antibiotic prescriptions. Due to the vague presentation of febrile illness in pediatric patients, Malawian providers tend to over-rely on the prescription of antibiotics for all NMF diagnoses, thus furthering the development of antimicrobial resistance [7-10]. To address the issue of inaccurate diagnoses, additional testing is needed to help providers differentiate febrile patients with or without pneumonia, so that they can determine when antibiotics can be safely distributed or withheld. In the context of low-resource settings, this diagnostic methodology should ideally be cost-effective, function independently of electricity, require minimal training, and provide results that are easy to interpret [11].

Two such diagnostic resources include 1) portable pulse oximeters and 2) the Integrated Management of Childhood Illness (IMCI) continued education courses. Despite being a standard tool in most developed healthcare systems, the pulse oximeter is not widely available in Malawian health facilities, a constraint that is especially exacerbated in rural areas [12, 13]. While studies regarding the use of pulse oximeters and/or IMCI continued education courses are expanding in hospital settings, little is known about the impact of implementing either intervention on antibiotic prescriptive rates in alternative sources of health care delivery, such as mobile clinics. Characteristics inherent in mobile clinics, namely their lack of geographical permanence, may increase providers’ over-reliance on antibiotics due to concerns of patients’ febrile symptoms worsening during the time it takes for the clinic to return to a given community [10]. As 85% of the Malawian population resides in rural areas, improving the capacity of mobile clinics to diagnose and treat NMF amongst these hard-to-reach communities is crucial [14].

Addressing these issues are in the forefront of the concerns for providers working for the Global AIDS Interfaith Alliance (GAIA) in Mulanje, Malawi. GAIA is a nonprofit organization which aims to increase healthcare accessibility amongst rural communities based in southern Malawi through the use of mobile clinics that rotate between the many villages throughout this region [15]. The mobile clinics operate five days a week, often extending into the weekend, offering comprehensive primary care services and consistent treatment for conditions that heavily plague this region such as malaria, tuberculosis, and HIV, all free of charge for their patients [15]. GAIA mobile clinic providers operate under tremendously challenging conditions, given that each mobile clinic team typically includes one clinical officer, two nurses and one nurse aid, seeing an average of 160 patients per day [15]. This hardship is further compounded by underdeveloped roads, seasonal tropical storms and frequent floods that burden southern Malawi [16]. Given the transient nature of these clinics, GAIA represented an ideal platform to assess the impact of an implementation of both portable pulse oximeters and IMCI continued education courses. This study aimed to understand the impact of these interventions, both individually and together, on pediatric fever diagnosis and prescribing practices in rural Malawi. While predominantly successful in these intentions, this study did conclude with some ambiguity regarding the impact of pulse oximetry as a stand-alone intervention.

Methods

This study is a continuation of recent research regarding NMF diagnosis in Malawi which concluded that rapid point-of-care tests were needed to help providers accurately diagnose febrile patients in order to make informed decisions when prescribing antibiotics [7, 17]. Specifically, this quantitative study used a logbook review to compare provider use of the IMCI guidelines to the use of the IMCI guidelines in conjunction with pulse oximeters.

An intervention study design was employed with four comparison groups; 1) logbook review of prescribing practices from April-July 2016 prior to any interventions, referred to as control group 1; 2) logbook review of prescribing practices from April-July 2019 in clinics in which neither intervention was implemented, referred to as control group 2 to account for changes in national protocols from 2016–2019; 3) logbook review from April-July 2019 of prescribing practices at clinics receiving IMCI continued education training only; and 4) logbook review of prescribing practices from April-July 2019 at clinics receiving both IMCI continued education training and pulse oximetry training (Table 1). The data collection period took place over six weeks (May 6, 2019 –June 14, 2019). The designation of intervention versus control status to each of the five clinic sites who serviced communities in the Mulanje district was done randomly. The logbooks used for this study are filled during clinic by trained support staff, typically nurses or nurse aids, who verify the information written in each client’s health passport and then transfer the information into the logbooks. Verification of unclear information written in the health passport is done by crosschecking with the prescribing officer to ensure that correct information is recorded. A monitoring and evaluation team at GAIA conducts data quality assessment and data verification quarterly.

Table 1

Clinic sites included in each study group.

Study Group	Clinic Site 1	Clinic Site 2	Clinic Site 3	Clinic Site 4	Clinic Site 5
Control (2016)	X	X	X	X	X
Control (2019)	X	X
Intervention 1 (IMCI-only)			X
Intervention 2 (IMCI+PO)				X	X

IMCI = Integrated Management of Childhood Illness, PO = pulse oximeter

Prior to the start of the study period, GAIA providers from five mobile clinics participated in one of two scheduled IMCI continued education courses held in March and April, 2019. The structure of this five-day course involved three days of theory-based learning in a classroom setting, followed by two days of practical-based learning in a pediatric clinical setting (Mulanje District Hospital). These courses were attended by 15 providers total, including both nurses and clinical officers. The participating providers were asked to take a pre-exam prior to the course, followed by a post-exam after completion of the training course. Each mobile clinic provider was assigned to a specific mobile clinic, thus the crossover of providers between mobile clinics did not occur. Clinic Site 3 was originally intended to be included in the intervention group which employed both the IMCI continued education course and the pulse oximeters. However, due to scheduling conflicts, the clinical officer from this clinic was unable to attend the pulse oximeter training, thus excluding Clinic Site 3 from the IMCI/pulse oximetry intervention group. Clinical officers and nurses from Clinic Sites 4 and 5 did participate in the training course, thus these two clinic sites comprised the IMCI/pulse oximetry intervention group.

Participating providers then attended a brief, one-hour training course in May 2019 on the use of the pulse oximeters. The training included a demonstration in the use of pulse oximeters led by the lead investigator with instruction on how to evaluate oxygen saturation levels, followed by a practical component in which the clinical staff practiced using the pulse oximeters on one another. Additional guidance was provided in the clinics by the lead investigator when needed in the weeks following the training. The training manual used for this session was drafted from the pulse oximetry training manual published by the World Health Organization, and supplemented by professional recommendations made by clinical provider members of the research team at the University of California, San Francisco [18].

The pulse oximetry protocol was designed to triage patients’ respiratory status. Oxygen saturation levels between 95–100% were described as healthy, 90–95% as moderately hypoxic, and <90% as severely hypoxic, warranting immediate referral to the nearest hospital. Providers were instructed to use 95% oxygen saturation as a cut-off for antibiotic use. This threshold is based on previously established indicators of pneumonia: oxygen saturation <95%, lung crackles on auscultation, fever >37.8°C, and pulse rate >100 beats per minute [18, 19]. The providers were advised to use the 95% threshold as a general parameter to aid in their diagnostic decision-making, not as an absolute determination of a patient’s diagnosis and/or need for antibiotics. Ultimately, the final diagnosis and treatments prescribed were determined by the combined knowledge gained from the pulse oximeter evaluation, IMCI guidelines, and clinical observations. Thus, for patients presenting with danger signs such as fast breathing or stridor, it was recommended that the clinical staff rely on their experience-based judgment rather than the pulse oximeter measurement as a standalone indication of health status.

The clinical staff were instructed to conduct pulse oximeter evaluation for all patients who presented with a high fever and negative malaria test. For this study, patients aged 1 year and older were evaluated with the Santamedical Generation 2 SM-165 fingertip pulse oximeter, while patients under the age of 1 year were evaluated with the Hopkins Handheld neonatal pulse oximeter [20, 21]. Both standard and neonatal pulse oximeters were distributed to the two clinics in the intervention group which received both pulse oximeters and IMCI continued education training (Clinic Sites 4 and 5) (Table 1). The lead investigator was in-country throughout the duration of the data collection period, during which time they rotated throughout the five clinics, with particular focus on the two clinics which implemented pulse oximetry for quality assurance purposes. While both the clinical officers and nurses were trained in the use of pulse oximeters, it is primarily the role of the clinical officers to diagnose patients and prescribe medications, as such it was the clinical officers who were largely responsible for using the pulse oximeters when evaluating patients throughout this study period.

A total of 3,504 patient logbook records were included for analysis. Review of the GAIA patient logbooks allowed for assessment of the implementation of and adherence to IMCI guidelines and pulse oximeter parameters, and the resulting diagnosis and treatment of pediatric patients with NMF. A sample-size calculation was run to determine the minimum size for each group at a power of 80% and alpha of 0.05 to detect a 15% difference in practices between the four study groups, yielding a necessary 122 records per study group. Quantitative data extracted from logbooks included: clinic site, patient age, patient sex, date, diagnosis, and drugs prescribed. For the two clinics that implemented pulse oximeters, data were also recorded on whether a pulse oximeter was used and the resulting oxygen saturation measurement. To evaluate changes in the use of the pulse oximeter in the IMCI/pulse oximetry group over the six-week study period, data were collected regarding the percentage of U-5 NMF patients that were evaluated with a pulse oximeter during this time. All quantitative data were collected and stored using Redcap (version 9.1.0), a secure online platform designed for managing surveys and databases [22]. The medical records were stored in hand-written logbooks that were securely stored in the GAIA clinical offices in Mulanje and Limbe, Malawi.

After the data collection period, brief qualitative questionnaires were distributed to the GAIA clinical staff regarding their opinion of the IMCI training course and/or pulse oximeters. The questionnaire guide was drafted by the research team at the University of California, San Francisco in conjunction with GAIA administrators (S1 Text). The questionnaire was given verbally in English and recorded, with provider consent. The names of the providers were not recorded during the questionnaire, but their position as either clinical officer, nurse, or nurse aide was documented. The data collection, both logbook extraction and qualitative interviews, was conducted by the lead investigator (FS).

Data analysis

To assess for changes in antibiotic prescriptions, the proportion of U-5 patients presenting with NMF who were prescribed antibiotics was compared among the four study groups. These proportions reflected provider adherence to either intervention when deciding to distribute or withhold antibiotics for NMF patients. To assess for differences in the prescriptive practices among the four study groups, Fisher’s exact test and simple logistic regressions were used. Adherence to the IMCI guidelines was determined by assessing whether the diagnosis made by the provider was followed by the IMCI recommended treatment, as reflected in the GAIA patient logbooks. Adherence to the pulse oximetry protocol was evaluated based on the percentage of patients who received a normal pulse oximeter measurement of ≥95% who then received the protocol's recommended treatment of basic analgesics. P-values less than 0.05 were considered significant. Odds ratios were also reported with corresponding 95% confidence intervals to provide both the strength and direction of association. All analyses were performed using the standard statistical analysis package R (version 3.5.1) [23].

The provider responses to the open-ended questionnaires were designed to help the GAIA management staff understand the experiences of the providers throughout the study and thus were considered supplemental to this study. Therefore, strict qualitative analysis methods were not employed for this portion of the study.

Ethics approval and consent to participate

All participating GAIA providers signed informed consent. The patient information collected from the logbooks was anonymized, and no direct contact between the research team and the patients took place. As such, consent was waived for this study population. Ethical approval was obtained from both the University of California San Francisco (UCSF) Committee for Human Research (#19–27452) and the Malawi National Health Sciences Research Committee (#19/03/2262) (NHSRC).

Results

Clinician population included in study

As previously stated, the GAIA providers who participated in the IMCI continued education course were required to complete a knowledge assessment exam before and after the course. The averages of the pre-exam and post-exam scores for the participating providers are shown in Fig 1, categorized by clinical position. Both the IMCI-only and IMCI/pulse oximetry groups had nurses and clinical officers in attendance for one of the IMCI training courses, while clinics in the 2019 control group had only nurses attend the training. A significant increase in content knowledge and understanding was detected between the pre- and post-exam scores from IMCI training (p-value <0.001). In comparing the scores of the nurses to the clinical officers, no significant difference was found in the average pre-scores, post-scores, or change in scores (p >0.001). Nurse and clinical officer participation in the pulse oximeter training is outlined in Table 2.

Fig 1

Average pre- and post- exam scores for the 15 GAIA providers who participated in either the first or second IMCI continued education training course (March 4–8 or April 1–5 respectively) categorized by clinical position as either nurse or clinical officer, shown with standard deviation.

Table 2

Clinical officers and nurses present for the pulse oximeter training course from each clinic site.

	Clinical Officer	Nurse
Clinic Site 3		X X X X
Clinic Site 4	X	X X X
Clinic Site 5	X	X X X X

Patient demographics

The demographics for the patient populations included in each of the four study groups can be found in Table 3. The 2016 control data consisted of 1,960 U-5 NMF patients seen by all five clinics. The sample sizes of the three remaining study groups were significantly smaller due to division of the five clinics according to which intervention was employed. The study group with the smallest sample size was that of the IMCI-only group, which only contained data from Clinic Site 3. A significant difference in the patient age distribution was detected among the four study groups (p < 0.001).

Table 3

Demographic data for all U-5 patients presenting with NMF in each study group.

Variable	Control (2016)	Control (2019)	IMCI	IMCI + PO	p-value
Sex
Male	938 (47.9%)	260 (45.5%)	86 (48.3%)	378 (47.5%)	p = 0.80
Female	1022 (52.1%)	311 (54.5%)	92 (51.7%)	417 (52.5%)
Age
≤1 month	11 (0.6%)	7 (1.2%)	2 (1.1%)	15 (1.9%)	p <0.05
2–12 months	679 (34.6%)	198 (34.7%)	48 (27.0%)	242 (30.4%)
1	473 (24.1%)	164 (28.7%)	55 (30.9%)	211 (26.5%)
2	345 (17.6%)	67 (11.7%)	34 (19.1%)	149 (18.7%)
3	258 (13.2%)	64 (11.2%)	19 (10.7%)	83 (10.4%)
4	195 (9.9%)	71 (12.4%)	20 (11.2%)	95 (11.9%)
Clinic site
1	216 (11.0%)	306 (53.5%)	--	--
2	410 (20.9%)	265 (46.4%)	--	--
3	129 (6.6%)	--	178 (100%)	--
4	643 (32.8%)	--	--	506 (63.6%)
5	563 (28.7%)	--	--	289 (36.4%)
Total participants per study group	1,960	571	178	795

IMCI = Integrated Management of Childhood Illness; PO = pulse oximeter

Differences in antibiotic prescriptive patterns

Overall, the odds of a patient receiving antibiotics in an intervention clinic that employed both IMCI training and pulse oximeters were 7.9 times less likely compared to a patient in the 2019 control group (95% CI 6.1–10.5), 7.3 times less likely than a patient in the IMCI-only group (95% CI 4.8–11.4), and 4.0 times less likely than a patient in the 2016 control group (95% CI 3.3–4.7). The significant difference in the antibiotic prescribing rates for U-5 NMF patients across the study groups can be seen in Fig 2, with rates of 75% (1,461/1,960) in 2016, 85% (485/571) in the 2019 control group, 84% (150/178) in the 2019 IMCI-only group, and 42% (336/795) in the 2019 IMCI with pulse oximetry group (p <0.001).

Fig 2

Percentage of U-5 NMF patients prescribed antibiotics per clinic site within each study group: Control 2016, control 2019, IMCI-only, and IMCI/Pulse Oximeter (PO).

Pulse oximeter utilization

Changes in the providers’ use of the pulse oximeters for diagnosing U-5 NMF patients throughout the study period are shown in Fig 3. Utilization of the pulse oximeter was consistently higher in Clinic Site 5 during each week throughout the study period. However, Clinic Site 4 exhibited a steady increase in the use of the pulse oximeter over time.

Fig 3

Percentage of U-5 NMF patients evaluated with a pulse oximeter in clinic sites 4 and 5 over the six-week study period.

A total of 795 patients were seen during the study period in intervention clinics using the pulse oximeter. Of these, 30% (n = 239) received evaluation by pulse oximetry. Differences in patient diagnoses based on oxygen saturation cutoff can be seen in Fig 4. For patients who received pulse oximeter evaluation with a resulting oxygen saturation level greater than or equal to 95%, the most common diagnosis was common cold (31%), followed by acute respiratory infection (22%), gastroenteritis (19%), upper respiratory tract infection (18%), and sepsis (9%). Only 2% of patients with a normal oxygen saturation (≥95%) were diagnosed with pneumonia. For patients who received pulse oximeter evaluation with a resulting oxygen saturation less than 95%, the most common diagnosis was pneumonia (77%), followed by acute respiratory infection (10%), sepsis (5%), bronchitis (4%), and lower respiratory tract infection (4%). A significant increase in pneumonia diagnoses was seen for pediatric patients with an oxygen saturation level less than 95% compared to patients who had an oxygen saturation of 95% or higher (p <0.001). For the remaining patients that did not receive pulse oximeter evaluation in the IMCI/pulse oximetry group, the most common diagnoses were acute respiratory infection (38%), sepsis (19%), gastroenteritis (17%), and common cold (14%). Of these patients, pneumonia represented 0.4% of all diagnoses.

Fig 4

(Left) Most common diagnoses for patients that received pulse oximeter evaluation with a resulting oxygen saturation ≥95%. (Right) Most common diagnoses for patients that received pulse oximeter evaluation with a resulting oxygen saturation <95%. ARI = Acute Respiratory Infection; URTI = Upper Respiratory Infection; LRTI = Lower Respiratory Infection.

Qualitative provider interviews

From the brief qualitative interviews conducted with the GAIA providers, key themes were identified regarding the perceived benefits and challenges of implementing either intervention (Fig 5). For quotes regarding provider opinion of either intervention, see Tables 4 and 5.

Fig 5

Key themes from qualitative interviews held with GAIA providers regarding their opinions of the IMCI continued education course training and the pulse oximeter.

Table 4

Provider quotes related to their perceived benefits from IMCI course participation.

Benefits from IMCI course participation	Illustrative Quote
Improved illness classification	"So, if they say that the patient is coughing, we check the type of breathing the child is having. Is it fast? Is it slow? Is the child having breath weakness? Is the child having hoarse voice? Is the child sucking? If the child is not sucking, is the child eating? Is the child having diarrhea? And is the child malnourished? So, we have to check the child holistically."
	“IMCI it does help, especially with classification, because we do not just diagnose now. Now we classify the illness. So, we start with symptoms, we take each step to see what the symptoms are and how severe each symptom is. Then, we can come to the conclusion of what the child is suffering from."
Enhanced awareness of antibiotic resistance	"There is no test for antibiotic resistance. But it exists. We know that it exists because already we are being told by the Ministry of Health, and we learned in the [IMCI] training, that certain antibiotics do not work in young patients, patients under five, because resistance already exists for those antibiotics."
Increased provider confidence	"I think the IMCI course helped me to be more confident when I meet a patient. When I know the patient has these symptoms, I can be more certain when he does need medication, or when he just needs to rest at home to feel better. I do not feel that I am guessing about the child’s condition."

Table 5

Provider quotes related to their perceived benefits of pulse oximeter utilization.

Benefits from pulse oximeter use	Illustrative Quote
Improved detection of pneumonia cases	“It has helped quite a lot in terms of reaching the right diagnosis, especially pneumonia, because it is very easy to miss a child with pneumonia. They can present with normal breathing, normal temperature, normal lung sounds, and so we might think it is a minor infection. We might send them home with the wrong medications. But now, we can see that their oxygen saturation is low, below 95% concentration, and we know that they do have pneumonia and do need antibiotics immediately.”
Reduced unnecessary antibiotic prescription	“Previously, we would often prescribe antibiotics after hearing the history from the mother. We would go into the field and if we saw a child that was having difficulty breathing, he was coughing severely, and we would think maybe it would be pneumonia, we would just prescribe antibiotics. But now, when you use the pulse oximeter and see normal oxygen levels, you are indeed sure that this is not pneumonia, just an upper respiratory tract infection.”
Increased confidence of caretakers	“Yes, the pulse oximeters have increased our confidence greatly, but also the confidence of the caregivers. You explain to them that this is medical equipment to assess if the child has pneumonia. Then, when you tell them that this is just the common cold or an upper respiratory infection, they will believe you, because you have used the instrument rather than just saying to them that their child just has the common cold. They can see the results right there and it helps them to feel that their child is safe, that they are receiving proper care.”

Perceived benefits of IMCI continued education course

Provider responses to the questionnaires exploring the IMCI continued education course were predominantly positive. The stepwise function of the IMCI guidelines was regarded as especially important in helping diagnose pediatric patients. Providers felt that the enhanced classification aspect of IMCI helped them to deconstruct the symptoms presented by each patient in order to generate a holistic diagnosis.

The providers also felt they had grasped the urgency of antibiotic resistance in Malawi and the need to be more conservative with the prescription of antibiotics for febrile pediatric patients. Several providers noted the increase in antibiotic resistance that they had witnessed firsthand through working with various patient populations over the years. This experience was described as stressful, as the providers relayed feeling “frustrated” and "defeated” when children would return to clinic without any improvement from previously administered antibiotics. Lastly, many reported an augmented sense of self-confidence as a result of their participation in the IMCI training. By completing the IMCI training, many felt that they no longer had to speculate on a diagnosis.

Perceived challenges of IMCI continued education course

While participants did describe some challenges related to IMCI course participation, these challenges primarily involved issues regarding the scheduling of the course, rather than the content of the course itself. For future IMCI course arrangements, several providers felt that it would be beneficial to increase the duration of the course from five days to ten, with a larger portion of the training course devoted to the field-based practical component. One other criticism of the IMCI courses was that they had been scheduled in March and April, both overlapping with malaria season in Malawi. As this is the busiest season for the GAIA providers, several staff members recommended that future courses be arranged specifically in the dry season, so as to avoid major conflicts with the clinic schedules.

Perceived benefits of pulse oximeter utilization

Similar to the responses regarding IMCI training, the providers perceived the impact of implementing pulse oximeters in the mobile clinics to be beneficial. The improved detection of pneumonia as indicated by low oxygen saturation on the pulse oximeter was deemed especially useful. The clinical officers from the IMCI/pulse oximetry group both believed that without the use of the pulse oximeter, many pneumonia patients may have been missed due to incomplete diagnostic evaluation. Due to the lack of diagnostic capacity in the mobile clinic setting and the rapid pace of patient flow in the mobile clinics, the providers noted the value of implementing a point-of-care device that can provide results in under 30 seconds.

In addition to providing added input when considering a diagnosis of pneumonia, the clinical officers also felt that the pulse oximeters allowed them to be more conservative when prescribing antibiotics for febrile pediatric patients. This is supported by the quantitative finding which stated an approximately 50% decrease in antibiotic prescriptive rates seen in the clinics which implemented both pulse oximeters and IMCI continued education courses. The providers considered the 95% saturation threshold as a cutoff to which they could easily adhere, thus simplifying their decision to prescribe antibiotics versus analgesics. The use of the pulse oximeters was also believed to increase the confidence of the patients’ caretakers. Many of the parents and/or guardians of the pediatric patients were reported to have felt more assured when being told that antibiotics were not needed after having observed a normal oxygen saturation reading. Witnessing the use of this technology left many caretakers with a feeling of relief in knowing that their child had been evaluated thoroughly in comparison to past clinical experiences.

Perceived challenges of pulse oximeter implementation

Several challenges were addressed regarding the use of the pulse oximeters in clinic. The primary challenge identified by the providers was that of applying the pulse oximeters to infants. Often times the very young children, particularly neonates and infants, demonstrated resistance to wearing the pulse oximeter. This resistance appeared to result from pediatric patients associating the use of the pulse oximeter with the use of mRDTs, and thus anticipating a painful experience. This resistance led providers to feel as though continuing to apply the pulse oximeter to the distressed child would be unethical. Additionally, one provider mentioned that on days in which the clinical officer was not working, it was difficult to incorporate the pulse oximeter into the diagnostic routine as the providers were already struggling to manage an under-staffed clinic. Lastly, several providers noted that issues related to the implementation of any new clinical protocol were observed, namely the missed documentation of the use of the pulse oximeter in the first few weeks of the study. However, these providers believed that the process of documenting pulse oximeter evaluations had improved over the six weeks.

Discussion

We investigated the impact of implementing two interventions, pulse oximeters and IMCI continued education training, on the frequency of antibiotic prescriptions for U-5 NMF patients being treated in five mobile health clinics in Mulanje, Malawi. Data extracted from patient logbooks indicated a substantial reduction in antibiotic provision in clinics receiving both IMCI continued education training and portable pulse oximeters. However, there was no significant reduction in antibiotic prescribing practices among providers in the clinics which only implemented IMCI training. Data compiled from the provider interviews suggested additional benefits of provider participation in the IMCI training related to improved diagnostic confidence and personal empowerment.

To our knowledge, this study was one of the first to show a significant association between the use of pulse oximeters and decreased antibiotic prescriptions for febrile patients being treated in a rural mobile clinic setting. However, investigation into the impact of pulse oximeters on antibiotic prescribing practices has been expanding for urban outpatient settings in sub-Saharan Africa [24]. The rate of antibiotics prescribed in the intervention group which utilized pulse oximeters was found to be approximately half of the rate exhibited in the three remaining study groups. This outcome is consistent with previous data which found that improved identification of various causes of non-malarial fever through the use of diagnostic resources, such as point-of-care procalcitonin tests, can decrease unnecessary antibiotic consumption by 30–50% [25, 26]. This finding indicates that the relatively simple and low-cost introduction of pulse oximeters into low-resource clinics that serve rural patient communities has the potential to greatly conserve antibiotic resources. Not only can this conservation help to deter the further development of resistance, but it can also benefit clinical organizations by allowing for increased financial savings and reallocation of funds given the higher cost of antibiotics compared to basic analgesics [27, 28]. Without a dedicated follow-up component of this study, it is impossible to declare with certainty whether this finding is purely a reduction in unnecessary antibiotic prescriptions or rather a reduction in necessary antibiotics as well. However, based on previous studies done within this region, the qualitative interviews held with the GAIA providers, and recommendations made by the Malawi Ministry of Health, it is known that the issue of antibiotic over-prescribing is quite prevalent in Malawi [7–10, 29]. Thus, it can be inferred that a more conservative shift in antibiotic prescribing practices in this setting could be beneficial in the advancement of public health.

During the qualitative interviews, the GAIA clinicians unanimously expressed greater confidence in their decision to diagnose pediatric pneumonia when presented with oxygen saturation measurements within the hypoxic range. This qualitative finding, in conjunction with the significantly higher proportion of pneumonia diagnoses for patients with a low oxygen saturation measurement, represents a meaningful clinical improvement. Given the high burden that pneumonia represents amongst pediatric patients in Malawi, and elsewhere in sub-Saharan Africa, and the ability of pulse oximeters to function in limited-resource settings, this device could greatly aid in the identification and treatment of pediatric pneumonia in rural areas [3].

It is worth noting that the IMCI-only intervention group did not exhibit a significant decrease in the number and/or percentage of U-5 NMF patients prescribed antibiotics. This finding contradicts previous studies which found that IMCI courses increased providers’ adherence to guidelines related to antibiotic distribution based on patient need [30]. It should be considered, however, that this finding may be influenced by the small sample size of the IMCI-only group in comparison to the other study groups. In the majority of the provider interviews, participants discussed the benefit of the IMCI course in training them to be more conservative with antibiotic prescriptions. However, such an outcome was not reflected in the analysis. One explanation for this inconsistency could be the lack of physical confirmation provided by the IMCI guidelines compared to the objective measurements offered by the pulse oximeters. It is understandable that when a pediatric patient presents with fever, despite no other physical signs of pneumonia as outlined in the IMCI guidelines, that a provider will still prescribe antibiotics to ensure the child’s safety. Furthermore, many of the guardians who accompany U-5 children may expect to leave a clinic with antibiotics, it is also likely that they strongly insist on such a prescription [31, 32]. However, it was observed that for many guardians who were able to see that their child’s oxygen saturation results fell within the healthy range, they felt confident enough in this “new” technology to return home with basic analgesics.

The lack of significant change in antibiotic prescriptions within the IMCI-only study group should not undermine the additional benefits that this course provided to the providers. Throughout the interviews, several providers conveyed a sense of empowerment after having completed the IMCI course. The significant increase in knowledge retention shown through their pre- and post-exam scores indicates the educational benefit of this course, both in theory and practice. This rise in diagnostic confidence is helpful in maintaining morale, and thus should not be overlooked. Furthermore, the augmented confidence that many providers experienced after completing this course represents a meaningful achievement given the lack of resources allotted in this setting.

As mentioned above, the most significant finding of this study is that antibiotics were significantly reduced in the intervention group that received both IMCI continued education training and use of the portable pulse oximeter. Both clinics in this intervention group had a clinical officer present during the study period. In Malawi, as with many countries around the world, nurses are often viewed as inferior to clinical officers and/or physicians, despite both completing three years of healthcare education [33, 34]. As such, when the clinical officer is present, he or she exclusively determines patient diagnoses and medication prescriptions. During the provider interviews, nurses from the IMCI-only clinic and the 2019 control clinics who had completed IMCI training expressed the need for decreasing unnecessary antibiotic prescriptions in order to deter further development of resistance. Yet, these clinics both exhibited high rates of antibiotic prescription. This finding indicates that regardless of training, it is ultimately the clinical officer who will govern significant changes in clinical output, a practice that limits the impact of positive trainings on other provider cadres. For an intervention to significantly impact diagnostic and prescriptive trends, nurses need to be given more responsibility and authority in the process of treating patients. Indeed, other studies in similar settings are beginning to show the significant impact of further empowering nurses in their clinical duties [35].

Limitations

Due to the limited time allotted for the completion of this study, conducting patient follow-up to assess whether the diagnosis and treatment plan was appropriate was not possible. While the results from the pulse oximeter intervention group showed a significant decrease in antibiotic prescriptions, it was not possible to conclude whether this decrease in antibiotics had any impact on patient wellbeing. However, several of the collaborators on this study recently published findings which showed that 14 days after U-5 febrile patients were evaluated in the same clinical setting, outcomes were stable to improved [7, 36]. Furthermore, given the significant increase in pneumonia diagnoses seen in the IMCI/pulse oximetry group, it is more likely that the wellbeing of pediatric patients was improved given the increased detection of pneumonia cases which may otherwise have been missed. Future studies should prioritize follow-up of patient outcomes post-clinic visit.

The restriction of the IMCI-only intervention group to only one clinic site may have limited the evaluation of this intervention. Furthermore, it is possible that the significant impact of the pulse oximeter on prescriptive trends was a result of its implementation in conjunction with the IMCI training rather than as a stand-alone diagnostic intervention. Previous studies have found this combination of interventions to be optimal in improving the detection and treatment of pediatric pneumonia in resource-poor settings [37]. Future studies should also investigate the impact of these interventions by evaluating before and after prescribing practices among same clinics, rather than between clinics, to further clarify the changes that resulted directly from intervention implementation.

Finally, it is possible that clinician-level practices also affected the overall treatment patients received. However, while for the purposes of this study the clinician/clinic groupings were static based on the intervention level, in practice GAIA clinicians and/or vehicles may be on a different given route at a given time. While this helped in the establishing of the baseline from which the analysis drew, future studies could also look at the impact of clustering on changes in practice.

Conclusion

In assessing the impact of multiple interventions on the diagnosis and treatment of pediatric non-malarial fever, this study found the use of simple pulse oximetry coupled with IMCI training significantly curbed the distribution of antibiotics in a low-resource setting. This finding points to the inclusion of pulse oximeters as a tool in the fight against antimicrobial resistance globally. Furthermore, the increase in pneumonia diagnoses in clinics using pulse oximeters indicates the benefit of this device in detecting pneumonia among pediatric patients experiencing hypoxia in similar settings. While the IMCI continued education course was not found to significantly influence antibiotic prescriptive trends, it was considered to be beneficial in stimulating provider confidence and empowerment. Additional investigation is needed to determine whether the success of the pulse oximeter can be replicated independently of IMCI training or if the combination of the two interventions provides the ideal balance of educational and physical resources to aid in diagnosing NMF pediatric patients. Future studies are also needed to closely follow patient outcomes when antibiotics are withheld and pre-determined oxygen saturation levels are used to guide provider treatment decisions.

Qualitative questionnaire provided to GAIA clinical staff.

(DOCX)

Click here for additional data file.

Quantitative data extracted from logbooks.

(XLSX)

Click here for additional data file.

13 Jul 2020

PONE-D-20-13967

The impact of pulse oximetry and Integrated Management of Childhood Illness (IMCI) training on antibiotic prescribing practices in rural Malawi: a mixed-methods study.

PLOS ONE

Dear Dr. Sylvies,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please look that the reviewers have, correctly, suggested several methodological changes and clarifications to be done, and this must be observed and performed in the next version.

Please submit your revised manuscript by Aug 27 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Ricardo Q. Gurgel, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please include additional information regarding the survey or questionnaire used in the study and ensure that you have provided sufficient details that others could replicate the analyses. For instance, if you developed a questionnaire as part of this study and it is not under a copyright more restrictive than CC-BY, please include a copy, in both the original language and English, as Supporting Information.

3. Thank you for stating the following in the Funding Section of your manuscript:

"This study was funded by the Global AIDS Interfaith Alliance and the Institute of Global Health Sciences

of the University of California, San Francisco. The design of this study was led by a masters student at

UCSF Institute for Global Health Sciences in collaboration with GAIA leadership."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

"The authors received no specific funding for this work. "

4. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

5. Your ethics statement must appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please also ensure that your ethics statement is included in your manuscript, as the ethics section of your online submission will not be published alongside your manuscript.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Every day, millions of parents seek health care for their sick children, taking them to hospitals, health centres, pharmacists, doctors and traditional healers. Surveys reveal that many sick children are not properly assessed and treated by these health care providers, and that their parents are poorly advised. At first-level health facilities in low-income countries, diagnostic supports such as radiology and laboratory services are minimal or non-existent, and drugs and equipment are often scarce. Limited supplies and equipment, combined with an irregular flow of patients, leave health workers at this level with few opportunities to practice complicated clinical procedures.

These factors make providing quality care to sick children a serious challenge. WHO and UNICEF have addressed this challenge by developing a strategy called the Integrated Management of Childhood Illness (IMCI).

IMCI is an integrated approach to child health that focuses on the well-being of the whole child. IMCI aims to reduce death, illness and disability, and to promote improved growth and development among children under five years of age. IMCI includes both preventive and curative elements that are implemented by families and communities as well as by health facilities.

The strategy includes three main components: Improving case management skills of health-care staff, Improving overall health systems,Improving family and community health practices.

In health facilities, the IMCI strategy promotes the accurate identification of childhood illnesses in outpatient settings, ensures appropriate combined treatment of all major illnesses, strengthens the counselling of caretakers, and speeds up the referral of severely ill children. In the home setting, it promotes appropriate care seeking behaviours, improved nutrition and preventative care, and the correct implementation of prescribed care. So, this research is very important.

Reviewer #2: The current study which was aimed to understand the impact of two interventions, that is, IMCI continued education courses and portable pulse oximeter, both individually and together, on paediatric fever diagnosis and prescribing practices in Malawi. The authors concluded that the use of simple pulse oximetry coupled with IMCI training helped significantly curb unnecessary antibiotic prescriptions in a low-resource setting. However, this statement is not true as we don't know the outcome of these children who did not receive antibiotics compared to those who received antibiotics in IMCI + PO group. Without outcome it is not possible to say that we have reduced the “unnecessary” antibiotics. There is a likelihood that those children also needed the antibiotics.

we all know that there is no point of care diagnostic test, such as rapid diagnostic test for malaria, for pneumonia. Therefore, health care providers, using the national IMCI chart booklet which depends on clinical signs, assess, classify and treat o refer children with pneumonia. We cannot diagnose pneumonia with pulse oximeter. Pulse oximetry is used to detect hypoxaemia in children. Therefore giving it diagnostic value in pneumonia diagnosis is not correct.

Furthermore, the version of the manuscript is not suitable to be published.

Methods: It is not written well. Some specific questions are :

• Which pulse oximeters were used in the study?

• Were some or all readings of pulse oximetry?

• Please describe any supportive supervision or monitoring visits were performed for quality assurance?

• Training material for pulse oximetry?

• How many days of training on how to use pulse oximeters in children?

• Any practical training on how to use pulse oximeters in children?

• Who performed pulse oximeters?

• What was the criteria to performed pulse oximeters?

• please provide sample size calculations to see the difference between four study groups.

• Please provide more information on mobile clinics? Usual timings? Routine services? Free of charge or people need to pay for services?

• Please provide definitions of pneumonia as well as algorithm to treat or refer pneumonia cases. What about children with danger signs? Was pulse oximeter <95% is part of definition of pneumonia where IMCI + PO was the intervention?

• Line 115: the standard duration for IMCI training is of 6 days. Please give reasons why did you conduct training for 5 days?

• Line129: The rationale use for oxygen saturation as <95% for prescription of antibiotics is based on a study which studied adult population. I am not sure why did authors use this threshold level? The WHO IMCI chart booklet uses a threshold level of oxygen saturation as <90% for hypoxaemia and these children need immediate hospital referral.

• Line 135: what was the selection criteria for patient logbook records. Please explain.

• Line 138: Quantitative data extracted from logbooks included….what about sex of the child?

• Line 147: After the data collection period, brief qualitative questionnaires…..please explain more about this questionnaire, who prepare it, was it tested before use?

Results:

• Figure 1:

o caption. It seems typo error for number of providers. It should be 13, see table 2.

o Along with average scores, it seems either range or standard deviation was also presented. Please clarify this in the caption.

• Table 3:

o Regarding Age, Why were infant aged between 1 and 2 months excluded from the study?. Please give explanation in the methods section also.

o Please provide any reason why proportion of infants <1 months increased in 2019 (including control 2019, IMCI and IMCI + PO)?

o In the table denominator for calculation of percentages changes several times. Sometime it is column sum, while on the other places it is row total. Please correct this.

o Footnote: please correct Integrated management of Childhood Illness.

• Line 191 Differences in antibiotic prescription patterns. Please provide number of children received antibiotics by four study groups.

• Line 199 Pulse oximeter utilization.

o Why 30% received evaluation by pulse oximetry? Please explain.

o Please give number of children in which pulse oximetry was performed but there was no reading/missing values.

Discussion

• Line 292: no significant reduction in antibiotic prescribing….but one reason is due to small sample size in this group.

Reviewer #3: Comments

The paper tackles an important topic, one that is critical in the management of childhood illnesses. Pneumonia diagnosis in children is complex often leads to misdiagnoses and over-prescription of antibiotics. The paper evaluated the introduction of a pulse oximeter in the diagnosis of pneumonia among children under-five by trained clinicians. As appealing and relevant as the topic is, my primary concern about the paper is on the design and analysis.

1) The description of the study groups is very confusing and does not clearly separate the baseline measurement versus endline and whether the groups are comparable. Table 1 helps to understand the design better, but also shows the flaws in the analysis and comparisons that were made. The 2016 control pools data from all five sites included in the study. This is then compared to subgroups of the sites, one subgroup (sites 4 and 5) having received training on IMCI and pulse oximetry (PO), another one (site 3) received IMCI training only, and the remaining 2 (sites 1 and 2) received nothing. The pooled 2016 control group is not comparable to each individual subgroup, and the authors seem to be comparing apples and oranges. The result of 75% antibiotic prescription in the 2016 control group is not comparable to the 85% in sites 1 and 2 (2019 control), or 84% in the site 3 (IMCI only) or 42% in sites 4 & 5 (IMCI+PO). To be valid, the change must be assessed within the same group at baseline and endline. The inconsistency is further highlighted in the higher antibiotic prescription found in the 2019 control group or the IMC training only group compared to the 2016 group. The correct design and internally valid approach would be to compare changes in antibiotics prescription between baseline and endline within each site (or subgroup of sites).

2) It would be useful to also describe the profile of the clinicians and nurses that provided child care and show how they differ across sites. Figure 3 shows, for example, a progressive increase in the use of PO in one of the site that received IMCI+PO training compared to the other site that received the same training, where the use started at much higher level. It is unclear how this would be the case, but illustrates that the sites are different across.

3) It seems that nurses from the 2019 control sites have also received the IMCI only training (page 8, line 176), but this was not accounted for in the description of the results.

4) The rationale for the selection of the sites must be described.

5) The completeness of the logbooks and quality of recording must also be described and not be assumed. Any supervisory measures implemented during the study that may affect the outcomes must be acknowledged and described.

6) It is also unclear what the eligibility for the PO was. The description in the paper suggests it would fever. In the group that was trained in IMCI+PO, 30% received PO. However, among those who did not receive PO, diagnosis included symptoms that would be associated with fever as well (38% ARI, 19% sepsis). Can we assume that this group might be severely rationing the antibiotics because they were being observed?

7) Please indicate who were the data collectors, for both the extraction and the qualitative interviews. The duration of the data extraction and the number of cases extracted per logbook must also be described.

8) The positive results of the qualitative interviews in the group that received IMCI training only don’t seem to square with the high level of antibiotic prescription in that group.

9) The analysis ignored the clustering of sick children within providers. This would affect the standard errors and the inference. The logit regressions that were run were not shown. The odds ratios must be adjusted not only for this clustering effect but also for other sites and provider characteristics.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Ayele Mamo Abebe

Reviewer #2: Yes: Yasir Bin Nisar

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

21 Aug 2020

Editor Comments:

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

Based on the styling format required by PLOS ONE, we have made the following edits:

• All level 1 headings have been made into bold, 18-point font, level 2 headings into bold, 16-point font, and level 3 headings into bold, 14-point font

• We have changed all “Figure X” to “Fig X”

• We have removed all Italics from text

• We have renamed the Figure files to “FigX.pdf”

• We have reformatted the author information to match the style of PLOS ONE standards.

• We have replaced the symbols in Tables 1 and 2 with “X”

2. Please include additional information regarding the survey or questionnaire used in the study and ensure that you have provided sufficient details that others could replicate the analyses. For instance, if you developed a questionnaire as part of this study and it is not under a copyright more restrictive than CC-BY, please include a copy, in both the original language and English, as Supporting Information.

We have included the questionnaire used in this study as supporting information, uploaded under the title “S1 Text”. As the GAIA staff members are all proficient in English speaking and comprehension, having a version of the questionnaire translated into Chichewa was deemed unnecessary.

3. We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form. Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

We have removed all funding-related texted from the manuscript. Please update our Funding Statement to the following: "This study was funded by the Global AIDS Interfaith Alliance and the Institute of Global Health Sciences of the University of California, San Francisco. The authors received no specific funding for this work."

4. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

After speaking with our in-country team, we have been granted permission to publish our dataset publicly without specific names of clinics and/or clinicians involved. We have included this dataset as Supporting Information entitled “S2 Dataset”. We have updated this information in line 548 under “Availability of data and materials”.

5. Your ethics statement must appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please also ensure that your ethics statement is included in your manuscript, as the ethics section of your online submission will not be published alongside your manuscript.

We moved the section entitled “Ethics Approval and Consent to Participate” to the final paragraph of the Methods section.

Reviewer 1:

Thank you so much for your feedback, we greatly appreciate your support of our research and hope that our findings may contribute to future efforts to improve community health in low resource settings.

Reviewer 2:

Overall:

1. The authors concluded that the use of simple pulse oximetry coupled with IMCI training helped significantly curb unnecessary antibiotic prescriptions in a low-resource setting. However, this statement is not true as we don't know the outcome of these children who did not receive antibiotics compared to those who received antibiotics in IMCI + PO group. Without outcome it is not possible to say that we have reduced the “unnecessary” antibiotics. There is a likelihood that those children also needed the antibiotics.

We very much appreciate your input. Based on our literature search and conversations with the in-country clinicians, we found that the issue of inappropriate antibiotic prescription was very prevalent in this region of Malawi and needed to be addressed. However, to your point, we have made the following edits throughout the paper:

• “Without a dedicated follow-up component of this study, it is impossible to declare with certainty whether this finding is purely a reduction in unnecessary antibiotic prescriptions or rather a reduction in necessary antibiotics as well. However, based on previous studies done within this region, the qualitative interviews held with the GAIA providers, and recommendations made by the Malawi Ministry of Health, it is known that the issue of antibiotic over-prescribing is quite prevalent in Malawi. Thus, it can be inferred that a more conservative shift in antibiotic prescribing practices in this setting could be beneficial in the advancement of public health.”

• We have also addressed this issue in the limitations section of this paper: “Due to the limited time allotted for the completion of this study, conducting patient follow-up to assess whether the diagnosis and treatment plan was appropriate was not possible. While the results from the pulse oximeter intervention group showed a significant decrease in antibiotic prescriptions, it was not possible to conclude whether this decrease in antibiotics had any impact on patient wellbeing. However, several of the collaborators on this study recently published findings which showed that 14 days after U-5 febrile patients were evaluated in the same clinical setting, outcomes were stable to improved.(7,34) Furthermore, given the significant increase in pneumonia diagnoses seen in the IMCI/pulse oximetry group, it is more likely that the wellbeing of pediatric patients was improved given the increased detection of pneumonia cases which may otherwise have been missed. Future studies should prioritize follow-up of patient outcomes post-clinic visit.”

2. We all know that there is no point of care diagnostic test, such as rapid diagnostic test for malaria, for pneumonia. Therefore, health care providers, using the national IMCI chart booklet which depends on clinical signs, assess, classify and treat or refer children with pneumonia. We cannot diagnose pneumonia with pulse oximeter. Pulse oximetry is used to detect hypoxaemia in children. Therefore giving it diagnostic value in pneumonia diagnosis is not correct.

We do apologize if our wording suggested the pulse oximeter as a definitive diagnostic tool for pneumonia. Our aim was to introduce the pulse oximeters as a tool that could provide useful information for the clinician’s diagnostic decision-making, in addition to their knowledge gained from the IMCI training and clinical expertise. However, we have added the following statements throughout the paper in order to clarify this point:

• In the abstract, we removed the line, “Additionally, the pulse oximeters demonstrated the capacity to improve detection of pediatric pneumonia” and replaced it with, “Enhanced detection of hypoxaemia in pediatric patients was regarded by clinicians as helpful for identifying pneumonia cases.”

• “The providers were advised to use the 95% threshold as a general parameter to aid in their diagnostic decision-making, not as an absolute determination of a patient’s diagnosis and/or need for antibiotics. Ultimately, the final diagnosis and treatments prescribed were determined by the combined knowledge gained from the pulse oximeter evaluation, IMCI guidelines, and clinical observations. Thus, for patients presenting with danger signs such as fast breathing or stridor, it was recommended that the clinical staff rely on their experience-based judgment rather than the pulse oximeter measurement as a standalone indication of health status.” Line 156

• In the discussion, we removed the following statement, “Of the various benefits of utilizing pulse oximeters in clinic, the subsequent improvement in pneumonia detection was universally commended by providers during the interviews.” This line was replaced with: “During the qualitative interviews, the GAIA clinicians unanimously expressed greater confidence in their decision to diagnose pediatric pneumonia when presented with oxygen saturation measurements within the hypoxic range.” Line 443

Methods:

1. Which pulse oximeters were used in the study?

For patients aged approximately 1 year and older, the Santamedical Generation 2 SM-165 Fingertip Pulse Oximeter was used, while patients under the age of 1 year were evaluated with the Hopkins Handheld Pulse Oximeter. We have included this information in the methods section in line 168.

2. Please describe any supportive supervision or monitoring visits were performed for quality assurance?

The lead investigator was in-country throughout the duration of the data collection period, during which time they rotated throughout the 5 clinics, with particular focus on the 2 clinics which implemented pulse oximetry for quality assurance purposes. This information has been added to the methods section to line 173.

3. Training material for pulse oximetry?

The training schedule for pulse oximetry was drafted based on both existing training resources found online, primarily the Pulse Oximetry Training Manual published by the World Health Organization in 2011, as well as the professional recommendations made by clinical provider members of the study team. This information has been added to the methods section to line 147.

4. How many days of training on how to use pulse oximeters in children?

The training course was brief, lasting 1 hour. However, additional guidance was provided for the clinicians assigned to clinics which received this intervention in the weeks following the training, although very little guidance was needed. This information has been added to the methods section to line 142.

5. Any practical training on how to use pulse oximeters in children?

The training session began with a demonstration in the use of pulse oximeters with instructions on how to interpret the resulting oxygen saturation levels, followed by a practical component in which the clinical staff practiced using the pulse oximeters on one another. There was no specific training session in which children were present, however guidance was offered by the lead researcher in clinic when needed, particularly with the neonatal pulse oximeters. This information has been added to the methods section to line 143.

6. Who performed pulse oximeters?

While both the clinical officers and nurses were trained in the use of pulse oximeters, it is primarily the role of the clinical officers to diagnose patients and prescribe medications, as such it was the clinical officers who were largely responsible for using the pulse oximeters when evaluating patients throughout this study. This information has been added to the methods section to line 175.

7. What was the criteria to performed pulse oximeters?

When training the clinical staff in the use of pulse oximeters, they were instructed to evaluate all patients who presented with a high fever and negative malaria test. This information has been added to the methods section to line 167.

8. Please provide sample size calculations to see the difference between four study groups.

We thank the reviewer for bringing up the important issues of whether or not we had an accurate sample size and a reasonable power to detect meaningful differences between our groups. As we did detect multiple significant differences between the groups, we omitted the calculations from our paper for space, but now recognize that many readers would still wish to see it. We have therefore added the following sentence within our methods section:

"A sample-size calculate was run to determine the minimum size for each group at a power of 80% and alpha of 0.05 to detect a 15% difference in practices between the four study groups, yielding a necessary 122 records per study group." Line 182

9. Please provide more information on mobile clinics? Usual timings? Routine services? Free of charge or people need to pay for services?

The GAIA clinics operate five days a week, often extending into the weekend, offering comprehensive primary care services and consistent treatment for conditions that heavily plague this region such as malaria, tuberculosis, and HIV, all free of charge for their patients. This information has been added to the background section to line 92.

10. Please provide definitions of pneumonia as well as algorithm to treat or refer pneumonia cases. What about children with danger signs? Was pulse oximeter <95% is part of definition of pneumonia where IMCI + PO was the intervention?

We added the following statements to the background and methods sections for clarification:

• “Defined as an inflammatory infection of the alveoli in the lungs, pneumonia typically presents with cough, fatigue, chest tightness, fever, sweating, and shortness of breath.” Line 62.

• “For patients presenting with danger signs such as fast breathing or stridor, it was recommended that the clinical staff rely on their experience-based judgment rather than the pulse oximeter measurement as a standalone indication of health status.” Line 158.

• While saturations measurements <95% were not defined as diagnostic of pneumonia, the clinical staff were instructed on the association between pneumonia and moderate hypoxia and thus used the 95% threshold as a tool to help determine whether a patient’s condition warranted antibiotic use or simply analgesics. Ultimately, the decision to diagnose pneumonia and/or prescribe antibiotics was based off of the clinician’s whole evaluation of the patient, including what they learned from the IMCI training and observed of the patient’s physical presentation. This has been further clarified in the methods section in line 156.

11. Line 115: the standard duration for IMCI training is of 6 days. Please give reasons why did you conduct training for 5 days?

The IMCI training was led by an experienced team based in Malawi. When presented with the question of why they conducted the training in 5 days rather than 6, they responded: “The extra (6th) day is a provision to ensure that all participants have completed a number of practices. In this case, there were enough clinical cases and additional onsite facilitators during the training days which enabled all participants to complete within 5 days.”

12. Line129: The rationale use for oxygen saturation as <95% for prescription of antibiotics is based on a study which studied adult population. I am not sure why did authors use this threshold level? The WHO IMCI chart booklet uses a threshold level of oxygen saturation as <90% for hypoxaemia and these children need immediate hospital referral.

Our rationale for using the 95% cutoff instead of 90% was that it would be more conservative, and thus a safer approach to treating pediatric patients. By making the cutoff 95%, more children would be receiving antibiotics comparatively and would thus minimize risk. We have expanded on this point for clarification:

• “The pulse oximetry protocol was designed to triage patients’ respiratory status. Oxygen saturation levels between 95-100% were described as healthy, 90-95% as moderately hypoxic, and <90% as severely hypoxic, warranting immediate referral to the nearest hospital.” Line 151.

13. Line 135: what was the selection criteria for patient logbook records. Please explain.

Patient logbook records were included based on these inclusion criteria: age under five years, febrile, malaria-negative, and treated during the dry season, as stated in the methods section in line 33.

14. Line 138: Quantitative data extracted from logbooks included….what about sex of the child?

Thank you for catching this error. We have added patient sex to the list of data extracted from the logbooks in line 185 of the methods section.

15. Line 147: After the data collection period, brief qualitative questionnaires…..please explain more about this questionnaire, who prepare it, was it tested before use?

The questionnaire guide was drafted by the research team at the University of California, San Francisco in conjunction with GAIA administrators. This information has been added to the methods section in line 197.

Results:

1. Figure 1:

� caption. It seems typo error for number of providers. It should be 13, see table 2.

This is not a typo, as Figure 1 is referring to the 15 GAIA clinicians who participated in the IMCI study while Table 2 is referring to the 13 GAIA clinicians who participated in the pulse oximetry training.

� Along with average scores, it seems either range or standard deviation was also presented. Please clarify this in the caption.

We have clarified in the Figure 1 caption that these bars represent the standard deviation values.

2. Table 3:

� Regarding Age, Why were infant aged between 1 and 2 months excluded from the study?. Please give explanation in the methods section also.

There was no exclusion of infants aged between 1 and 2, we have changed the category in Table 3 to read as “�1 month” instead of “<1 month” for clarification.

� Please provide any reason why proportion of infants <1 months increased in 2019 (including control 2019, IMCI and IMCI + PO)?

While we are also interested in this shift in patient age between 2016-2019, we do not believe it is within the scope of this study to fully investigate this demographic trend.

� In the table denominator for calculation of percentages changes several times. Sometime it is column sum, while on the other places it is row total. Please correct this.

We have removed the row sums from Table 3 so that only column sums remain.

� Footnote: please correct Integrated management of Childhood Illness.

Thank you for catching this error, we have corrected it.

3. Line 191 Differences in antibiotic prescription patterns. Please provide number of children received antibiotics by four study groups.

We have included these numbers in our results section in line 309: “The significant difference in the antibiotic prescribing rates for U-5 NMF patients across the study groups can be seen in Figure 2, with rates of 75% (1,461/1,960) in 2016, 85% (485/571) in the 2019 control group, 84% (150/178) in the 2019 IMCI-only group, and 42% (336/795) in the 2019 IMCI with pulse oximetry group (p <0.001).”

4. Line 199 Pulse oximeter utilization.

� Why 30% received evaluation by pulse oximetry? Please explain.

The clinical staff did not use the pulse oximeters for 100% of the pediatric patients presenting with non-malarial fever. There were many instances in which the devices were not used either due to clinician confidence in the patient’s diagnosis without needing to know the oxygen saturation or due to lack of available time given the extremely fast pace of patient flow in these mobile clinics.

� Please give number of children in which pulse oximetry was performed but there was no reading/missing values.

To our knowledge, there were no instances in which pulse oximetry was performed without documentation of the resulting values.

Discussion:

1. Line 292: no significant reduction in antibiotic prescribing….but one reason is due to small sample size in this group.

We have edited the discussion to include the following statement: “It is worth noting that the IMCI-only intervention group did not exhibit a significant decrease in the number and/or percentage of U-5 NMF patients prescribed antibiotics. This finding contradicts previous studies which found that IMCI courses increased providers’ adherence to guidelines related to antibiotic distribution based on patient need. It should be noted, however, that this finding may be influenced by the small sample size of the IMCI-only group in comparison to the other study groups.”

Reviewer #3: Comments

1) The description of the study groups is very confusing and does not clearly separate the baseline measurement versus endline and whether the groups are comparable. Table 1 helps to understand the design better, but also shows the flaws in the analysis and comparisons that were made. The 2016 control pools data from all five sites included in the study. This is then compared to subgroups of the sites, one subgroup (sites 4 and 5) having received training on IMCI and pulse oximetry (PO), another one (site 3) received IMCI training only, and the remaining 2 (sites 1 and 2) received nothing. The pooled 2016 control group is not comparable to each individual subgroup, and the authors seem to be comparing apples and oranges. The result of 75% antibiotic prescription in the 2016 control group is not comparable to the 85% in sites 1 and 2 (2019 control), or 84% in the site 3 (IMCI only) or 42% in sites 4 & 5 (IMCI+PO). To be valid, the change must be assessed within the same group at baseline and endline. The inconsistency is further highlighted in the higher antibiotic prescription found in the 2019 control group or the IMC training only group compared to the 2016 group. The correct design and internally valid approach would be to compare changes in antibiotics prescription between baseline and endline within each site (or subgroup of sites).

Since the GAIA’s initiation, each clinic’s circumstances have been comparable in regard to topography, patient demographics and socioeconomic status, health conditions treated, clinical training, and prescribing practices. Therefore, we felt pooling information across all GAIA clinics was informative. Given that individual patient data from 2016 has been converted into an amalgamated dataset, we cannot know for certain whether different providers from the 2016 group showed varying prescribing practices. In the discussion we have noted that it is possible that the individual practitioner approaches from 2016 may have influenced antibiotics prescribed. As such, future studies which evaluate the before and after prescribing practices among same clinical groups would provide further clarity for this area of study.

2) It would be useful to also describe the profile of the clinicians and nurses that provided child care and show how they differ across sites. Figure 3 shows, for example, a progressive increase in the use of PO in one of the site that received IMCI+PO training compared to the other site that received the same training, where the use started at much higher level. It is unclear how this would be the case, but illustrates that the sites are different across.

Based on our observations from the clinic, we believe that the difference in pulse oximeter utilization between clinic sites was largely due to differences in each clinical officer’s priorities. For example, the clinic that exhibited higher PO utilization was led by a clinical officer who had substantially more administrative responsibility and was thus more enthusiastic about antibiotic conservation from a financial standpoint. In the clinic which showed a more hesitant use of the pulse oximeters, the lead clinical officer was more less fiscally focused and was reluctant to withhold antibiotics from potentially sick children. While we agree that this personal information is contextually useful, we believe it to be too speculative for the purposes of this study. Furthermore, it could allow for the identification of specific GAIA clinicians, which would be inappropriate. As such, we have decided to exclude these details.

3) It seems that nurses from the 2019 control sites have also received the IMCI only training (page 8, line 176), but this was not accounted for in the description of the results.

The nurse participation from the 2019 control group is accounted for at the beginning of the results section in line 232: “Both the IMCI-only and IMCI/pulse oximetry groups had nurses and clinical officers in attendance for one of the IMCI training courses, while clinics in the 2019 control group had only nurses attend the training.”

4) The rationale for the selection of the sites must be described.

The designation of intervention versus control status to each of the five clinic sites who serviced communities in the Mulanje district was done so randomly. This information has been added under the methods section in line 119.

5) The completeness of the logbooks and quality of recording must also be described and not be assumed. Any supervisory measures implemented during the study that may affect the outcomes must be acknowledged and described.

The logbooks are filled during clinic by trained support staff, typically nurses or nurse aids, who verify the information written in each client’s health passport and then transfer that information into the logbooks. Verification of unclear information written in the health passport is done by crosschecking with the prescribing officer to ensure that correct information is recorded. A monitoring and evaluation team at GAIA conduct data quality assessment and data verification quarterly. This information has been added to the methods section in line 125.

6) It is also unclear what the eligibility for the PO was. The description in the paper suggests it would fever. In the group that was trained in IMCI+PO, 30% received PO. However, among those who did not receive PO, diagnosis included symptoms that would be associated with fever as well (38% ARI, 19% sepsis). Can we assume that this group might be severely rationing the antibiotics because they were being observed?

While it is possible that the clinical officers changed their prescribing habits due to feeling observed by our research team, a similar decrease in antibiotic prescriptions was not observed for the clinics without pulse oximeter implementation during the days in which a member of the research team was present. As such, we do not believe that our presence had a significant influence over the providers’ decision to prescribe antibiotics.

7) Please indicate who were the data collectors, for both the extraction and the qualitative interviews. The duration of the data extraction and the number of cases extracted per logbook must also be described.

The data collection, both logbook extraction and qualitative interviews, was conducted by the lead investigator (FS). This information has been added to the methods section in line 202. The duration of data extraction is referenced in the methods section in line 118: “The data collection period took place over six weeks (May 6, 2019 – June 14, 2019).”

Unfortunately, it would be very difficult to calculate how many cases were extracted per logbook as the logbooks are kept as hand-written records and stored in the GAIA office in Mulanje.

8) The positive results of the qualitative interviews in the group that received IMCI training only don’t seem to square with the high level of antibiotic prescription in that group.

Yes, we discussed this contradiction in the discussion in line 455: “In the majority of the provider interviews, participants discussed the benefit of the IMCI course in training them to be more conservative with antibiotic prescriptions. However, such an outcome was not reflected in the analysis. One explanation for this inconsistency could be the lack of physical confirmation provided by the IMCI guidelines compared to the objective measurements offered by the pulse oximeters. It is understandable that when a pediatric patient presents with fever, despite no other physical signs of pneumonia as outlined in the IMCI guidelines, that a provider will still prescribe antibiotics to ensure the child’s safety. Furthermore, many of the guardians who accompany U-5 children may expect to leave a clinic with antibiotics, it is also likely that they strongly insist on such a prescription.”

9) The analysis ignored the clustering of sick children within providers. This would affect the standard errors and the inference. The logit regressions that were run were not shown. The odds ratios must be adjusted not only for this clustering effect but also for other sites and provider characteristics.

We thank the reviewer for their keen statistical considerations with regards to the analysis and issues of clustering. As we consider prescribing practices, there is certainly the possibility that the approaches used by different providers may affect the overall differences seen between groups as might characteristics of the patient population served. In these cases, it can be advantageous to cluster the analysis by facilities and/or providers themselves though a generalized estimating equation (GEE) or logistic mixed-effect model to provide a more accurate representation of the differences seen by an intervention. However, in the case of our study we ultimately decided against this for the reasons listed below:

• With regards to the patient populations: the district served by GAIA mobile clinics is a rural district of Malawi with a largely homogeneous population along the mobile clinic routes including that of demographic-, socioeconomic-, and environmental-profiles. Further, given the remote nature of the district, patients will often be as close to one mobile clinic route as to another and choose to attend one based on need and availability.

• In contrast to static facilities with dedicated staff, the nature of the mobile clinics made clustering problematic. While the pulse-oximeter intervention was given to a set group of providers for purposes of this study, in general the staffing of the mobile clinics shifts among the GAIA staff. Further, given groups of staff and vehicles are not always serving the same area. For this reason, we could not cluster the clinics when comparing pre- and post- intervention as such clustering would be artificial and not account for providers or patient characteristics that clustering seeks to address.

Recognizing the importance of how things like provider approaches could affect prescription rates, and the theoretical importance of clustering to account for this, we have added the following notes in the methods and discussion:

• It is possible that clinician-level practices also affected the overall treatment patients received. However, while for the purposes of this study the clinician/clinic groupings were static based on the intervention level, in practice GAIA clinicians and/or vehicles may be on a different given route at a given time. While this helped in the establishing of the baseline from which the analysis drew, future studies could also look at the impact of clustering on changes in practice. Line 517.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

3 Nov 2020

The impact of pulse oximetry and Integrated Management of Childhood Illness (IMCI) training on antibiotic prescribing practices in rural Malawi: a mixed-methods study.

PONE-D-20-13967R1

Dear Dr. Sylvies,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Ricardo Q. Gurgel, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank for the opportunity to review 'The impact of pulse oximetry and Integrated Management of Childhood Illness (IMCI) training on antibiotic prescribing practices in rural Malawi: a mixed-methods study' I have no further comments for the authors

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Ayele Mamo Abebe

Reviewer #2: No

9 Nov 2020

PONE-D-20-13967R1

The impact of pulse oximetry and Integrated Management of Childhood Illness (IMCI) training on antibiotic prescribing practices in rural Malawi: a mixed-methods study.

Dear Dr. Sylvies:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Professor Ricardo Q. Gurgel

Academic Editor

PLOS ONE