Program for outpatient antibiotic management of possible severe bacterial infections (PSBI) in newborns

Abbreviations

Abbreviation	Definition	Note
PSBI	Possible severe bacterial infection

Intervention Overview

Possible severe bacterial infections (PSBIs) in newborns are serious conditions that can lead to sepsis and death. According to the [WHO-PSBI-Guideline], clinical signs of PSBI include:

• Clinical severe infection:

  • Severe chest indrawing

  • High body temperature

  • Low body temperature

  • Not feeding well

  • Movement only when stimulated

• Critical illness:

  • Unconscious

  • Apnoea

  • Unable to cry

  • Cyanosis

  • Bulging fontanelle

  • Persistent vomiting

  • No movement at all

  • Not able to feed at all

  • Convulsions

  • Active bleeding requiring transfusion

Management of PSBI starts with identification of at least one sign (by a family member or health professional) followed by confirmation by a community health worker or other health professional. To facilitate the identification of PSBI, WHO recommends that community health workers counsel families on recognition of danger signs for PSBI and assess infants for signs of PSBI during postnatal care visits. [WHO-PSBI-Guideline]

The current WHO guideline on treatment for PSBI in infants less than 59 days old is referral to a hospital for inpatient management and a 7-10 day course of two injectable antibiotics - penicillin or ampicillin plus gentamicin [WHO-PSBI-Guideline]. There is a separate WHO Guideline on the management of PSBI when hospital referral is not feasible, and recommends that infants of families who do not accept or cannot access hospital care are managed in an outpatient setting with intramuscular gentamicin (once daily for 2 or 7 days) and oral amoxicillin (twice daily for 7 days) [WHO-PSBI-Guideline].

The WHO guideline for PSBI management without hospital referral relies heavily on evidence included in a recent Cochrane review of community-based antibiotic management of PSBI [PSBI-Cochrane-Review]. There has been evidence of successful implementation of outpatient management of PSBI in accordance with the WHO guideline, although not yet at scale [Nisar-et-al-2022]. Additionally, some unpublished trials suggest that outpatient management for clinical severe infection and early hospital discharge for non-critical cases is as effective as full hospital management of PSBI.

Todo

Link unpublished data when it is ready.

Comparison one of the [PSBI-Cochrane-Review] evaluated trials that compared programs to initiate and/or complete community-based antibiotic management of newborn PSBI cases in newborns when hospital referral was refused relative to the standard of care of hospital referral only. This comparison included five trials and found a summary effect of such outpatient antibiotic programs on overall mortality of RR=0.82 (95% CI: 0.68 to 0.99) with a high level of statistical heterogeneity (I^2 = 87%). Only two trials reported effects on sepsis-specific neonatal mortality, resulting in a summary relative risk of 0.78 (95% CI: 0.60 to 1.00).

Notably, direct observation of mortality rates among a population with PSBI with no intervention (treatment or referral) by researchers is not ethical. As such, trials like those included in the [PSBI-Cochrane-Review] are designed so that the control arm is the standard of care hospital referral and the intervention arm is community based antibiotic administration in addition to hospital referral. Some portion of the population in these trials is expected to refuse referral and not receive antibiotics and could serve as a control group in the derivation of the effect of antibiotics on sepsis mortality relative to no antibiotics that we desire for use in our simulation. Unfortunately, however, none of the individual studies included in [PSBI-Cochrane-Review] report sufficient information to derive such values (although the Bhandari et al. 2012 study comes the closest).

A review by [Zaidi-et-al-2011] attempted to address this challenge of deriving an effect of antibiotics relative to no antibiotics that cannot be directly observed in an experimental study for the Lives Saved Tool (LiST). Through a Delphi process to reach expert consensus, they arrived at the following effects on neonatal sepsis mortality:

• RR=0.72 for oral antibiotics in a community setting

• RR=0.35 for injectable antibiotics is a community/clinic setting

• RR=0.2 for in-patient hospital management with antibiotics

We have conducted our own assessment of alternative strategies to arrive at such values, summarized in a few slides on sharepoint here

Vivarium Modeling Strategy

This section describes how an outpatient program for newborn PSBI management intervention can be implemented and calibrated for the MNCNH Portfolio model.

Affected Outcomes

Outcome	Effect	Modeled?	Note (ex: is this relationship direct or mediated?)
Neonatal sepsis and other neonatal infections Mortality Probability \(\text{CSMRisk}_i^\text{sepsis}\)	Adjust multiplicatively using RR	Yes

Research background

Given the limitations of the approaches to estimating the effect of antibiotics relative to no antibiotics for the management of PSBI, we have decided to instead model the effect of an outpatient antibiotic program relative to the absence of such a program. This allows us to utilize the evidence from the [PSBI-Cochrane-Review] directly and avoids having to obtain specific estimates of antibiotics coverage. However, this approach has several limitations, listed in the Assumptions and Limitations section of this page.

Baseline Coverage Data

Ethiopia has integrated outpatient management of newborn PSBI cases into its national health policy starting in 2012. [Tiruneh-et-al-2024] reviews the implementation experience of this program in Ethiopia. According to this paper, the ability to manage newborn PSBI on an outpatient basis occurs at a national level in Ethiopia; however, between 60-98 percent of facilities reported stockouts of gentamicin or amoxicillin over a three month period.

According to this online resource, Pakistan does not have a national program to manage PSBI in newborns on an outpatient basis.

Based on the multi-country implementation study of outpatient antibiotic management programs for PSBI in newborns that conducted pilot programs [Nisar-et-al-2022], we assume that the baseline coverage of this intervention is zero in our modeled locations.

Outpatient neonatal antibiotic intervention baseline coverage

Location	Population	Value	Reference
Pakistan	All newborns (baseline coverage does not vary by delivery facility)	0	See Aga Khan University Medical College’s seminar here indicating no national guideline for PSBI management on an outpatient basis in addition to [Nisar-et-al-2022] pilot program implementation in modeled location and lack of scaled-up programs
Ethiopia	All newborns (baseline coverage does not vary by delivery)	0.5	Assumption of 50% functional capacity to manage PSBI on an outpatient basis given on findings from [Tiruneh-et-al-2024] of frequent stock-outs and limited health worker capacity as well as estimated coverage from this cluster randomized trial in Ethiopia [Degefie-Hailegebriel-et-al-2017]
Nigeria	All newborns (baseline coverage does not vary by delivery facility)	0	Assumption based on [Nisar-et-al-2022] pilot program implementation in modeled location and lack of scaled-up programs

Vivarium Modeling Strategy

This intervention requires adding an attribute to all simulants to specify if a neonate has access to outpatient antibiotic programs for PSBI in newborns. Since the neonatal mortality model does not explicitly represent incidence of sepsis, we will not track explicitly if a simulant receives antibiotics. Instead the model will have different cause-specific mortality rates for sepsis for individuals with and without access to outpatient antibiotic programs (implemented with a slightly confusing application of our Risk and RiskEffect components from vivarium_public_health).

The Risk component adds an attribute to each simulant indicating whether the simulant has access to antibiotics during the neonatal period, which we assume will be closely related to the facility choice during birth, i.e. home births have much lower access than in-facility births, and births in BEmONC facilities have lower access than CEmONC facilities.

To make this work naturally with the RiskEffect component, it is best to think of the risk as “lack of access to antibiotics”. With this framing, the RiskEffect component requires data on (1) the relative risk of sepsis mortality for people with lack of access to antibiotics, and (2) the population attributable fraction (PAF) of sepsis due to lack of access to antibiotics. We will use the decision tree below to find the probability of sepsis mortality with and without access to antibiotics that are logically consistent with the baseline delivery facility rates and baseline antibiotics coverage.

In Vivarium, this risk effect will modify the sepsis mortality pipeline, resulting in

\[\text{CSMRisk}_i^\text{sepsis} = \text{CSMRisk}^\text{sepsis}_{\text{BW}_i, \text{GA}_i} \cdot (1 - \text{PAF}_\text{no antibiotics}) \cdot \text{RR}_i^\text{no antibiotics}\]

where \(\text{RR}_i^\text{no antibiotics}\) is simulant i’s individual relative risk for “no antibiotics”, meaning \(\text{RR}_i^\text{no antibiotics} = \text{RR}_\text{no antibiotics}\) if simulant i accesses a facility without antibiotics, and \(\text{RR}_i^\text{no antibiotics} = 1\) if simulant i accesses a facility with antibiotics.

If there are other interventions also affecting the CSMR of sepsis, the pipeline will combine these effects, and we can write out the math for this risk explicitly as

\[\text{CSMRisk}^\text{sepsis}_{i, \text{updated}} = \text{CSMRisk}^\text{sepsis}_{i, \text{original}} \cdot (1 - \text{PAF}_\text{no antibiotics}) \cdot \text{RR}_i^\text{no antibiotics}\]

This reduces to the previous formula if there are no other interventions, and we would have

\[\text{CSMRisk}^\text{sepsis}_{i, \text{original}} = \text{CSMRisk}^\text{sepsis}_{\text{BW}_i, \text{GA}_i}\]

Where:

Risk Effect Parameters for Lack-of-Access-to-Intervention

Parameter	Value	Notes
\(\text{RR}^\text{no antibiotics}\)	\(1/\text{RR}_\text{intervention}\)	To be used in artifact in accordance with “lack of intervention” risk factor effect
\(\text{RR}_\text{intervention}\)	0.78 (95% CI: 0.60 to 1.00), lognormal distribution of uncertainty (implemented as parameter uncertainty)	[PSBI-Cochrane-Review]
mean_rr	\(\text{RR}^\text{no antibiotics} * (1 - p_\text{baseline coverage}) + p_\text{baseline coverage}\)	\(p_\text{baseline coverage}\) is baseline coverage proportion defined in the baseline coverage section above
PAF	(mean_rr - 1) / mean_rr

Scenarios

Scenario-specific coverage of the outpatient neonatal antibiotic intervention for the MNCNH simulation can be found in the neonatal component scenario table.

Generally, intervention-scenario coverage of this intervention should be 100%, indicating the presence of a fully functioning outpatient program to manage newborn PSBI with antibiotics. Note that this does not imply that 100% of newborn PSBI cases are treated with antibiotics, but rather that outpatient treatment occurs at the same rate as the trials included in the [PSBI-Cochrane-Review].

Assumptions and Limitations

• This intervention applies to the first two months of life according to the WHO guideline and we only model the first month of life, so we will not capture any averted deaths in the second month of life due to this intervention, therefore underestimating total impact.

• We assume 50% functional capacity of the outpatient newborn PSBI management program in Ethiopia in accordance with frequent stock-outs and limited health worker capacity reported in [Tiruneh-et-al-2024] and hypothesize that these issues can be resolved in an intervention scenario

• We do not model the effect of this intervention on pneumonia mortality (note that according to the WHO guideline these cases do not require inpatient treatment and outpatient management programs are expected to have a different effect on pneumonia mortality than sepsis mortality)

• Our modeling strategy does not allow for differential impact by location due to factors such as existing level of inpatient PSBI treatment rates

  • Notably, of the two trials included in the [PSBI-Cochrane-Review] that reported sepsis-specific mortality effects, [Soofi-et-al-2012] was performed in rural districts in Pakistan and [Gill-et-al-2014] was performed in a rural district in Tanzania with no hospital in the district. Therefore, it is possible we will overestimate the effect of this intervention by applying to the population at large that includes population subgroups that may have greater access to hospitals (such as those in urban areas)

• There was significant heterogeneity in the effect of the intervention in the [PSBI-Cochrane-Review] and we do not model factors that may influence the effectiveness of the intervention

• The evidence in the [PSBI-Cochrane-Review] for this intervention was graded as low quality given the significant heterogeneity and the existence of interventions besides antibiotics administered alongside the community distribution of antibiotics interventions

• Many of the trials in the [PSBI-Cochrane-Review] included additional services alongside the ability to treat PSBI on an outpatient basis, which may confound the estimate of the intervention effect (however, we lessen the degree of this bias by using the sepsis-specific mortality estimate rather than the all cause mortality estimate)

• Although the trials in the [PSBI-Cochrane-Review] were measured specifically among the population visited by community health workers participated in the study, we do not limit the intervention effect only to those who receive postnatal care visits. This is because even if a family does not receive a PNC visit, it is possible that they will seek care if/when their newborn displays signs of illness. However, by not modeling decreased coverage among the population who does not receive PNC visits, we may overestimate the effect of the intervention.

• We apply the effect of this intervention to the neonatal sepsis cause-specific mortality risk among the entire neonatal population in our model. However, babies who are hospitalized from birth through the end of the neonatal period cannot directly benefit from an outpatient intervention. While it is possible that reduced demand for inpatient antibiotic management due to the intervention will decrease capacity demand for NICUs and therefore increase quality of care among existing inpatient neonates, we do not have specific data to inform this effect. Therefore we are likely overestimating the impact of this intervention by applying its effect the entire population including hospitalized neonates (who are likely to be preterm and therefore at greatest risk of sepsis mortality in the first place).

Todo

Consider adding pneumonia as additional affected cause?

Determine if we want to make the eligible population those who recieve postnatal care visits according to the GBD covariate value. This would require us to add PNC as an additional attribute to the model and would allow us to correlate it with ANC/IFD attributes.

Consider adding a NICU variable to our model so that we can apply the intervention effect to non-hospitalized neonates only. Note that a NICU variable would also be helpful for tracking CPAP use.

Consider modeling incidence risk for neonatal sepsis to aid in costing this intervention

Validation and Verification Criteria

• Population-level mortality rate should be the same as when this intervention is not included in the model

• The ratio of sepsis deaths per birth among those without antibiotics access divided by those with antibiotics access should equal the relative risk parameter used in the model

• The baseline coverage of antibiotics in each facility type should match the values in the artifact

References

[WHO-PSBI-Guideline] (1,2,3,4)

Guideline: managing possible serious bacterial infection in young infants when referral is not feasible https://www.who.int/publications/i/item/9789241509268

[PSBI-Cochrane-Review] (1,2,3,4,5,6,7,8,9,10,11,12)

Duby J, Lassi ZS, Bhutta ZA. Community-based antibiotic delivery for possible serious bacterial infections in neonates in low- and middle-income countries. Cochrane Database Syst Rev. 2019 Apr 11;4(4):CD007646. doi: 10.1002/14651858.CD007646.pub3. PMID: 30970390; PMCID: PMC6458055.

[Degefie-Hailegebriel-et-al-2017]

Degefie Hailegebriel T, Mulligan B, Cousens S, Mathewos B, Wall S, Bekele A, Russell J, Sitrin D, Tensou B, Lawn J, de Graft Johnson J, Legesse H, Hailu S, Nigussie A, Worku B, Baqui A. Effect on Neonatal Mortality of Newborn Infection Management at Health Posts When Referral Is Not Possible: A Cluster-Randomized Trial in Rural Ethiopia. Glob Health Sci Pract. 2017 Jun 27;5(2):202-216. doi: 10.9745/GHSP-D-16-00312. PMID: 28611102; PMCID: PMC5487084.

[Gill-et-al-2014]

Gill CJ, MacLeod WB, Phiri-Mazala G, Guerina NG, Mirochnick M, Knapp AB, Hamer DH. Can traditional birth attendants be trained to accurately identify septic infants, initiate antibiotics, and refer in a rural African setting? Glob Health Sci Pract. 2014 Aug 31;2(3):318-27. doi: 10.9745/GHSP-D-14-00045. PMID: 25276591; PMCID: PMC4168634.

[Nisar-et-al-2022] (1,2,3,4)

Nisar YB, Aboubaker S, Arifeen SE, Ariff S, Arora N, Awasthi S, Ayede AI, Baqui AH, Bavdekar A, Berhane M, Chandola TR, Leul A, Sadruddin S, Tshefu A, Wammanda R, Nigussie A, Pyne-Mercier L, Pearson L, Brandes N, Wall S, Qazi SA, Bahl R. A multi-country implementation research initiative to jump-start scale-up of outpatient management of possible serious bacterial infections (PSBI) when a referral is not feasible: Summary findings and implications for programs. PLoS One. 2022 Jun 13;17(6):e0269524. doi: 10.1371/journal.pone.0269524. PMID: 35696401; PMCID: PMC9191694.

[Soofi-et-al-2012]

Soofi S, Ahmed S, Fox MP, MacLeod WB, Thea DM, Qazi SA, Bhutta ZA. Effectiveness of community case management of severe pneumonia with oral amoxicillin in children aged 2-59 months in Matiari district, rural Pakistan: a cluster-randomised controlled trial. Lancet. 2012 Feb 25;379(9817):729-37. doi: 10.1016/S0140-6736(11)61714-5. Epub 2012 Jan 27. PMID: 22285055.

[Tiruneh-et-al-2024] (1,2,3)

Tiruneh GT, Odwe G, Kamberos AH, K’Oduol K, Fesseha N, Moraa Z, Gwaro H, Emaway D, Magge H, Nisar YB, Hirschhorn LR. Optimizing integration of community-based management of possible serious bacterial infection (PSBI) in young infants into primary healthcare systems in Ethiopia and Kenya: successes and challenges. BMC Health Serv Res. 2024 Mar 5;24(1):280. doi: 10.1186/s12913-024-10679-9. PMID: 38443956; PMCID: PMC10916061.

[Zaidi-et-al-2011]

Zaidi AK, Ganatra HA, Syed S, Cousens S, Lee AC, Black R, Bhutta ZA, Lawn JE. Effect of case management on neonatal mortality due to sepsis and pneumonia. BMC Public Health. 2011 Apr 13;11 Suppl 3(Suppl 3):S13. doi: 10.1186/1471-2458-11-S3-S13. PMID: 21501430; PMCID: PMC3231886.