Rational development of guidelines for management of neonatal sepsis in developing countries

Abstract

Purpose of review

This review discusses the rational development of guidelines for the management of neonatal sepsis in developing countries.

Recent findings

Diagnosis of neonatal sepsis with high specificity remains challenging in developing countries. Aetiology data, particularly from rural, community based studies are very limited, but molecular tests to improve diagnostics are being tested in a community-based study in South Asia. Antibiotic susceptibility data are limited, but suggest reducing susceptibility to first and second line antibiotics in both hospital and community acquired neonatal sepsis. Results of clinical trials in South Asia and sub-Saharan Africa assessing feasibility of simplified antibiotic regimens are awaited.

Summary

Effective management of neonatal sepsis in developing countries is essential to reduce neonatal mortality and morbidity. Simplified antibiotic regimens are currently being examined in clinical trials, but reduced antimicrobial susceptibility threatens current empiric treatment strategies. Improved clinical and microbiological surveillance is essential, to inform current practice, treatment guidelines, and monitor implementation of policy changes.

Introduction

Substantial improvements in maternal and child health have been achieved since 2000, with reductions in child deaths from infectious diseases.¹ However, reducing neonatal deaths, and stillbirths, remains challenging.² Of the estimated 6.3 million deaths in children under five years in 2013,³ 2.8 million (44%) of these deaths were in the neonatal period (0-27 days), and 0.42 (95%CI 0.27-0.69) million of neonatal deaths were due to sepsis.¹

The global burden of morbidity due to neonatal infection was estimated at ~3% of all Disability Adjusted Life Years (DALYs),⁴ but this estimate excludes long-term morbidity, due to a lack of data on impairment outcomes.⁵ With an estimated 6.9 (95%CI 5.5-8.3) million cases of possible severe bacterial infection (pSBI) in neonates in South Asia, sub-Saharan Africa and Latin America in 2012, improving our understanding of this burden is essential.⁶

Reducing the burden of neonatal mortality and morbidity depends on the scale-up of essential interventions, which if universally achieved could reduce neonatal mortality by over two thirds.⁷ These interventions include prevention and management of preterm birth, basic and comprehensive obstetric care, immediate new-born care (including resuscitation and antisepsis measures), supportive care, especially for preterm neonates (including kangaroo mother care) and effective management of neonatal infections.^7-9

Improving management of neonatal infections depends on effective guidelines, and overcoming barriers to implement them.⁷ Here the challenges of the rational development of guidelines in developing countries for the management of neonatal sepsis are identified and discussed.

Diagnosis

Neonatal diseases are difficult to diagnose clinically because signs and symptoms are much more non-specific for infectious syndromes (sepsis, pneumonia or meningitis) than in older children. The same clinical signs can be present, for example, in neonates born preterm (<37 weeks’ gestation), with intra-partum complications, or viral illnesses, with or without concurrent bacterial infection.⁶

In developing countries diagnoses are usually made by health care workers without specialist training. To support diagnoses of “possible severe bacterial infection” (pSBI) and guide treatment, clinical algorithms have been developed, described in the Integrated Management of Childhood Illness.¹⁰ These algorithms have been informed by research, including the first WHO Young Infants Clinical Signs Study (YICSS),¹¹ and the second YICSS, which included a focus on neonates in the first week of life.¹² In the second YICSS study, seven clinical signs (fast breathing (respiratory rate >60 breaths per minute), severe chest in-drawing, hyperthermia >37•5°C, hypothermia <35•5°C, no movement or movement only on stimulation, convulsions, poor feeding) had 85% sensitivity and 75% specificity for severe bacterial infection in the first week of life, with an experienced paediatrician’s diagnosis as the gold standard.¹² Subsequent meta-analysis of six studies validating the use of this algorithm against physician diagnosis (with supporting laboratory data) found a similarly high sensitivity (87%) but specificity was lower (62%).¹³ Fast breathing has been identified as having low specificity as a single sign,¹⁴ and hypothermia may be influenced by the environment, for example in Nepal just under half of all neonates were reported to have hypothermia (32.0-36.0°C).¹⁵

There are particular settings in which low specificity is problematic. If the algorithm is used in the community to facilitate identification and treatment of neonates,¹⁶ and the prevalence of neonatal infection is much lower than in those seeking health care, the positive predictive value using this algorithm will be lower (less than 50%).⁶ In clinical trials, low specificity may introduce misclassification of disease, as described for treatment trials of pneumonia.¹⁷ In current clinical trials for simplified antibiotic treatment the YICSS algorithm has been amended to increase specificity; fast breathing is not included as a single sign for case definition, the temperature threshold was increased from ≥37.5 to ≥38.0°C, and poor feeding had to be confirmed by observation. In addition, neonates with convulsions were excluded from the study of simplified treatments, as they were considered to have critical illness. It is important that standardised case definitions for possible severe bacterial infections in neonates are used, but with the flexibility to evolve, as more sensitive and specific clinical signs and diagnostic strategies are identified.

In developed countries clinical diagnosis of severe bacterial infection in neonates is supported by experienced paediatricians and systematic laboratory investigation, including conventional microbiological (blood, cerebrospinal fluid and urine), haematological, and biochemical tests. However, even with the expertise and investigations available in developed countries, diagnosis remains a challenge.¹⁸ Microbiological identification of a pathogen isolated from blood cultures can confirm a clinical diagnosis of neonatal sepsis, and has high specificity, but sensitivity is lower. Although blood culture sensitivity has been improved with automated blood culture systems (such as BACTEC, BactTALERT), it is still likely less than 80%, particularly with small (<1mL) blood volumes, common from neonatal samples, and if the colony forming unit is low (<4 CFU/mL).¹⁹

Molecular microbiology, using nucleic acid testing (NAT), may improve diagnostics. There are rapid advances in the use of NAT tests to diagnose and determine the aetiology of diarrhoea and meningitis, both in terms of sensitivity and implementation in developing countries.^20-22 However, high sensitivity can present problems in interpretation, and careful case-control studies are required to establish attributable fractions of disease using these methods.²³ Experience of molecular methods to improve diagnosis in developing countries are more limited for neonatal sepsis than for pneumonia,²⁴ diarrhoea,²⁰ or meningitis,²⁵ but NAT using the Taqman Array Card is being used in a large ongoing multi-site study of the Aetiology of Neonatal Sepsis in South Asia (ANISA),^26,27 with results awaited.

In both developed and developing settings, there is a risk of overtreatment for neonatal sepsis, as sensitivity is prioritised over specificity, due to high case fatality risk if neonatal sepsis is untreated. However, unnecessary treatment is not without risk in terms of possible side-effects of treatment, especially with over-dosing, reported in a fifth of neonatal cases in a study in district hospitals in the Kilimanjaro region of Tanzania.²⁸ Unnecessary treatment and admission to hospital also increases the risk of invasive fungal infections, hospital acquired bacterial infections, and reduced antibiotic susceptibility.²⁹ A rapid diagnostic test, if it were available, would transform care,³⁰ directing treatment to those who needed it, and preventing unnecessary antibiotic use.

Aetiology

Differences in the aetiology of neonatal sepsis in developed and developing countries may reflect true differences in disease epidemiology, as well as differences in the levels of care available to support survival of preterm and/or low birth weight neonates. They may also be due to study methodologies, including the quality of microbiological investigations, and study design, particularly whether community or health facility based. There are very limited community based studies,^31,32 of neonatal infection in developing countries.

In early onset neonatal sepsis (0-6 days of life), the most commonly reported pathogens in developing countries include Gram negatives (Escherichia coli, Klebsiella pneumoniae) and Gram positives (Staphylococcus aureus).^33-35 These early-onset infections may include hospital acquired infections, if associated with hospital delivery in developing countries, where inadequate hygiene and anti-sepsis measures contribute to early-onset infections.³⁶ For late onset neonatal sepsis the most common pathogens reported in developing countries are mainly Gram positive, including Staphylococcus aureus, Streptococcus pneumoniae, Streptococcus pyogenes, Streptococcus agalactiae and some Gram negative non-typhoidal Salmonellae sp..^33-35

Streptococcus agalactiae (Group B Streptococcus, GBS) although the leading pathogen in early onset infections in developed countries,^37,38 is less consistently reported in developing countries, particularly studies of community acquired bacteraemia. However, in a recent systematic review, although in Southeast Asia the estimated burden of GBS disease in infants was low (0.02/1000 live births (95%CI 0.00-0.07)), Africa was estimated to have the highest incidence of GBS in infants worldwide (1.21/1000 live births (95%CI 0.50-1.91)), although based on only four studies,^39-42 meeting selection criteria.⁴³ In contrast, the WHO study in The Gambia, Ethiopia, The Philippines and Papua New Guinea in the 1990s investigated 2398 infants attending as outpatients,⁴⁴ and isolated GBS in blood cultures from only two infants. Most recently, the WHO study in Bangladesh, Bolivia, Ghana, India, Pakistan and South Africa investigated 782 infants, and there were no isolates of GBS.⁴⁵ These differences may reflect true differences, particularly in disease-causing serotypes,⁴⁶ however, this study excluded the most severely ill neonates, and it is possible that with delays in accessing care, neonates with GBS died prior to arrival, especially in early onset GBS sepsis,⁴³ for which the case fatality risk is three times higher in developing countries (12·6%, 95% CI 10·8–14·9) than in developed countries (4·6%, 2·1–9·1).⁴³

The role of coagulase negative Staphylococci (CoNS) in neonatal sepsis in developing countries is not defined; in developed countries they are the leading cause of late onset sepsis,³⁷ but occur mainly in very preterm neonates in intensive care, with ventilator support and indwelling devices.⁴⁷ In developing countries, the burden of preterm birth is high,^48,49 but mortality is also very high.^1,49 Supportive care is much more limited, so indwelling devices and invasive ventilation are much less frequently used,⁵⁰ and, if in use, are restricted to referral centres in urban settings. In this context, it is unclear the extent to which CoNS acts as a neonatal pathogen; it is sometimes reported,^51-56 but at other times excluded as a contaminant.^39,45,57-60

Issues of contamination are important for other common skin colonisers, such as Staphylococcus aureus. Whilst S.aureus does cause disease, the dominance of S. aureus reported in some studies in developing countries,^41,45,61 may in part be due to contamination from skin colonisation and inadequate site sterilisation when blood cultures are taken.⁴⁵

Treatment

Empirical treatment for neonatal sepsis, recommended in current WHO guidelines is intravenous ampicillin (or penicillin) plus gentamicin for 7 days.⁶² Cloxacillin is an alternative if Staphylococcal infection is suspected. Third generation cephalosporins, increasing in availability, particularly in Asia,¹³ are used as second line for those not responding after 48-72 hours of treatment. Despite the inconsistent literature on GBS infections, neonatal treatment (but not intra-partum antibiotic prophylaxis) is also recommended for 48 hours (and then review) for neonates born to mothers with PROM (≥18h), maternal fever ≥38°C before delivery or during labour, and foul-smelling amniotic fluid.⁶²

Treatment guidelines include admission to hospital, but care seeking can be low (varying from 10-100% in low and middle income countries).⁶³ Clinical trials based in Africa and south Asia are assessing whether current antibiotic regimens can be simplified, to facilitate access to care.¹⁴ The Simplified Antibiotic Therapy Trial (SATT) Bangladesh, SATT Pakistan and African Neonatal Sepsis Trial (AFRINEST)), with harmonised protocols,⁶⁴ aim to evaluate standard 7 day treatment (daily injection procaine penicillin and gentamicin) compared to combinations of injectable antibiotics and oral amoxicillin.¹⁴

The success of treatment, however, depends on pathogens being susceptible to treatment choices. Data are limited, but there is increasing evidence for reduced susceptibility in pathogens causing both hospital acquired infections,³⁶ and community acquired infections, particularly Klebsiella sp, and Staphylococcus aureus.^33,65,66 In an assessment of WHO treatment guidelines, for hospital acquired infection, around a third of neonatal sepsis isolates were estimated to be susceptible to first line antibiotics,³⁶ and a half of Gram negatives (Klebsiella sp and E.Coli) to second line treatment.³⁶ Klebsiella pneumoniae is a particularly important neonatal pathogen, both in terms of its prevalence, ability to persist in environments,^67,68 and cause persisting outbreaks of hospital acquired infections, which can be highly resistant to treatment.⁶⁸

Community-based studies with antibiotic susceptibility data are limited, but in a systematic review including 11 studies, 57% of bacterial isolates were estimated to be susceptible to benzylpenicillin/ampicillin and gentamicin, and 56% susceptible to third generation cephalosporins.³³ Recent work supports these findings; 40 Gram negative neonatal pathogens were isolated as part of the WHO study across six countries and 38/39 (97%) showed resistance to ampicillin, 17/39 (44%) to gentamicin and 19/31 (61%) to third generation cephalosporins.⁴⁵ Alternative therapeutic options, such as fluoroquinolones and carbapenems, are limited and expensive for developing countries.

Supportive care

Respiratory support, intravenous fluids, phototherapy, temperature and glucose regulation as well as prevention of hospital acquired infections, are essential components of care. Limitations in resources mean that interventions for supportive care must be pragmatic. A report of neonatal care in training centres across Kenya found that all hospitals had oxygen, and 19/22 had resuscitation and phototherapy equipment.⁵⁰ Pragmatic interventions to improve respiratory support, beyond the provision of oxygen, have been recently described, using bubble CPAP in Uganda and Malawi, with promising findings.^69,70 Similar progress needs to be made in terms of controlled fluid delivery to neonates; there were only 12/22 training centres across Kenya with paediatric burettes.⁵⁰ Uncontrolled or inadequate provision of intravenous fluids likely results in many undetected cases of electrolyte imbalance and consequent morbidity and mortality. Simple bedside tests are available for testing glucose levels, and temperature regulation can be improved by Kangaroo Mother Care, but these interventions need adequate staffing and health system support for implementation.

Improving the evidence

In contrast to the networks being set up in developed countries to report and monitor neonatal infections,³⁷ there are few health facilities in developing countries undertaking systematic clinical surveillance. Embedding clinical surveillance and using admission pro-forma facilitates clinical audit and monitoring and improves care. It will be increasingly possible to add microbiological diagnostic techniques to clinical surveillance platforms to improve understanding of the aetiology of disease and inform treatment guidelines. Reduced antibiotic susceptibility in both hospital and community acquired infections represents an increasing threat to empiric treatment strategies, and more data are urgently needed.^36,71 Methods such as whole genome sequencing will be able to shed new light on the epidemiology of infectious diseases worldwide, including transmission,⁷² and antibiotic susceptibility.⁷³ Understanding these mechanisms in developing countries, and identifying drivers of reduced antibiotic susceptibility are essential to inform prevention and control.

Conclusion

Effective management of neonatal sepsis in developing countries is essential to achieve further reductions in child mortality. Simplified antibiotic regimens are currently being examined in clinical trials, but reduced antimicrobial susceptibility threatens current empiric treatment strategies. Improved clinical and microbiological surveillance is essential, to inform current practice, treatment guidelines, and monitor implementation of policy changes.

Key Points.

• Molecular diagnostics to determine aetiology of neonatal sepsis are being tested.

• Clinical trials to assess simplified antibiotic treatments are underway.

• Reduced antibiotic susceptibility to first and second line treatments is an urgent concern.

• Systematic, standardised surveillance of neonatal sepsis in health facilities is essential.

Abbreviations

(ANISA)

Aetiology of Neonatal Sepsis in South Asia

(CoNS)

Coagulase Negative Staphylococci

(CFU)

Colony Forming Units

(CPAP)

Continuous positive air pressure

(DALYs)

Disability Adjusted Life Years

(GBS)

Group B Streptococcus

(NAT)

Nucleic Acid Testing

(pSBI)

Possible Severe Bacterial Infection

(WHO)

World Health Organisation

(YICSS)

WHO Young Infants Clinical Signs Study