Background. There are little data on the immunogenicity of PCV10 and PCV13 in the same high-risk population. Methods. PCV10 and PCV13 were studied head-to-head in a randomized controlled trial in Papua New Guinea in which 262 infants received 3 doses of PCV10 or PCV13 at 1, 2, and 3 months of age. Serotype-specific immunoglobulin G (IgG) concentrations, and pneumococcal and nontypeable Haemophilus influenzae (NTHi) carriage were assessed prevaccination and at 4 and 9 months of age. Infants were followed up for safety until 9 months of age. Results. One month after the third dose of PCV10 or PCV13, >80% of infants had IgG concentrations >= 0.35 mu g/mL for vaccine serotypes, and 6 months postvaccination IgG concentrations >= 0.35 mu g/mL were maintained for 8/10 shared PCV serotypes in >75% of children vaccinated with either PCV10 or PCV13. Children carried a total of 65 different pneumococcal serotypes (plus nonserotypeable). At 4 months of age, 92% (95% confidence interval [CI] 85-96) of children vaccinated with PCV10 and 81% (95% CI 72-88) vaccinated with PCV13 were pneumococcal carriers (P = .023), whereas no differences were seen at 9 months of age, or for NTHi carriage. Both vaccines were well tolerated and not associated with serious adverse events. Conclusions. Infant vaccination with 3 doses of PCV10 or PCV13 is safe and immunogenic in a highly endemic setting; however, to significantly reduce pneumococcal disease in these settings, PCVs with broader serotype coverage and potency to reduce pneumococcal carriage are needed.

We investigated the immunogenicity, seroprotection rates and persistence of immune memory in young children at high risk of pneumococcal disease in Papua New Guinea (PNG). Children were primed with 10-valent (PCV10) or 13-valent pneumococcal conjugate vaccines (PCV13) at 1, 2 and 3 months of age and randomized at 9 months to receive PPV (PCV10/PPV-vaccinated, n = 51; PCV13/PPV-vaccinated, n = 52) or no PPV (PCV10/PPV-naive, n = 57; PCV13/PPV-naive, n = 48). All children received a micro-dose of PPV at 23 months of age to study the capacity to respond to a pneumococcal challenge. PPV vaccination resulted in significantly increased IgG responses (1.4 to 10.5-fold change) at 10 months of age for all PPV-serotypes tested. Both PPV-vaccinated and PPV-naive children responded to the 23-month challenge and post-challenge seroprotection rates (IgG ≥ 0.35 μg/mL) were similar in the two groups (80–100% for 12 of 14 tested vaccine serotypes). These findings show that PPV is immunogenic in 9-month-old children at high risk of pneumococcal infections and does not affect the capacity to produce protective immune responses. Priming with currently available PCVs followed by a PPV booster in later infancy could offer improved protection to young children at high risk of severe pneumococcal infections caused by a broad range of serotypes.

Summary In areas where Streptococcus pneumoniae is highly endemic, infants experience very early pneumococcal colonization of the upper respiratory tract, with carriage often persisting into adulthood. We aimed to explore whether newborns in high-risk areas have pre-existing pneumococcal-specific cellular immune responses that may affect early pneumococcal acquisition. Cord blood mononuclear cells (CBMC) of 84 Papua New Guinean (PNG; high endemic) and 33 Australian (AUS; low endemic) newborns were stimulated in vitro with detoxified pneumolysin (dPly) or pneumococcal surface protein A (PspA; families 1 and 2) and compared for cytokine responses. Within the PNG cohort, associations between CBMC dPly and PspA-induced responses and pneumococcal colonization within the first month of life were studied. Significantly higher PspA-specific interferon (IFN)-γ, tumour necrosis factor (TNF)-α, interleukin (IL)-5, IL-6, IL-10 and IL-13 responses, and lower dPly-IL-6 responses were produced in CBMC cultures of PNG compared to AUS newborns. Higher CBMC PspA-IL-5 and PspA-IL-13 responses correlated with a higher proportion of cord CD4 T cells, and higher dPly-IL-6 responses with a higher frequency of cord antigen-presenting cells. In the PNG cohort, higher PspA-specific IL-5 and IL-6 CBMC responses were associated independently and significantly with increased risk of earlier pneumococcal colonization, while a significant protective effect was found for higher PspA-IL-10 CBMC responses. Pneumococcus-specific cellular immune responses differ between children born in pneumococcal high versus low endemic settings, which may contribute to the higher risk of infants in high endemic settings for early pneumococcal colonization, and hence disease.

Infants in Papua New Guinea (PNG) are at a high risk of invasive pneumococcal disease, and a substantial burden of this falls on children less than six months old. PNG is planning to introduce a pneumococcal conjugate vaccine for infants in the near future, but to make the maximum impact neonatal immunization will have to be considered. To provide evidence on safety and immunogenicity for neonatal and early infant immunization, we undertook an open randomized controlled trial of 7-valent pneumococcal conjugate vaccine (7vPCV). 318 children received 7vPCV at ages 0, 1 and 2 months or at 1, 2 and 3 months or not at all. All children received 23-valent pneumococcal polysaccharide vaccine at age 9 months. This was a large and complex trial: village reporters visited participants weekly during the first year and fortnightly for a further 6 months and nurses monitored self-reported morbidity and collected many thousands of biological samples. The study team was remarkably successful in achieving the study aims, with 18-month follow-up completed on 77% of enrolled children and over 80% of scheduled samples collected. While the results of the trial will be reported elsewhere, this paper discusses the design of the study and dissects out some of the main reasons for its successful completion. Strong community engagement was an essential factor in success and the principles of equitable partnership and service provision led to a strong research partnership. A two-stage consent process, comprising primary assent followed by later informed consent, led to a high drop-out before initial enrolment, but an outstanding retention of those enrolled in the study. We conclude that factors such as strong community participation, reciprocity and a good relationship between the study team and participants are just as important as the technical elements of laboratory testing and data handling in ensuring the success of a vaccine trial in PNG.

Background Pneumonia is the most frequent cause of child mortality in less-developed countries, We aimed to establish whether the combination of benzylpenicillin and gentamicin or chloramphenicol would be better as first-line treatment in children with severe pneumonia in Papua New Guinea.Methods We did an open randomised trial in which we enrolled children aged 1 month to 5 years of age who fulfilled the WHO criteria for very severe pneumonia and who presented to hospitals in two provinces. Children were randomly assigned to receive chloramphenicol (25 mg/kg 6 hourly) or benzylpenicillin (50 mg/kg 6 hourly) plus gentamicin (7.5 mg/kg daily) by intramuscular injection. The primary outcome measure was a good or an adverse outcome.Findings 1116 children were enrolled; 559 children were treated with chloramphenicol and 557 with benzylpenicillin and gentamicin. At presentation the median haemoglobin oxygen saturation was 71% (IQR 57-77) for those allocated chloramphenicol and 69% (55-77) for those allocated penicillin and gentamicin. 147 (26%) children treated with chloramphenicol and 123 (22%) treated with penicillin and gentamicin had adverse outcomes (p=0.11). 36 children treated with chloramphenicol and 29 treated with penicillin and gentamicin died. More children treated with chloramphenicol than penicillin and gentamicin represented with severe pneumonia within 1 month of hospital discharge (p=0.03).Interpretation For children with severe pneumonia in less-developed countries the probability of a good outcome is similar if treated with chloramphenicol or with the combination of benzylpenicillin and gentamicin.

A multi-centre randomised open trial was done to determine whether moderate oral fluid restriction or intravenous fluid at full maintenance volumes would result in a better outcome for children with bacterial meningitis in Papua New Guinea, and what clinical signs could guide fluid management. Children with clinical signs and cerebrospinal fluid suggestive of bacterial meningitis received either breast milk by nasogastric tube at 60% of normal maintenance volumes (n=172) or intravenous half-normal saline and 5% dextrose at 100% of normal maintenance volumes (n=174) for the 1st 48 hrs of treatment. An adverse outcome was death or severe neurological sequelae, and a good outcome was defined as intact survival or survival with at worst mild-to-moderate neurological sequelae. The probability of an adverse outcome was 24.7% in the intravenous group and 33.1% in the oral-restricted group, but the difference was not statistically significant (RR 0.75, 0.53-1.04, p=0.08). Sunken eyes or reduced skin turgor at presentation were risk factors for an adverse outcome (OR 5.70, 95% CI 2.87-11.29) and were most strongly associated with adverse outcome in the fluid-restricted group. Eyelid oedema during treatment was also a risk factor for an adverse outcome (OR 2.54, 95% CI 1.36-4.75) and eyelid oedema was much more common in the intravenous group (26%) than in the restricted group (5%). For many children with bacterial meningitis in less developed countries, moderate fluid restriction is unnecessary and will be harmful; a normal state of hydration should be achieved but over-hydration should be avoided. Giving 100% of normal maintenance fluids, especially with intravenous hypotonic fluid, will lead to oedema in up to one quarter of children with bacterial meningitis. If additional intravenous fluids are required for children with meningitis, an isotonic solution should be used.

This supplement has presented the results of a large study of the etiology and clinical signs of serious infections in 4552 infants younger than 91 days of age in 4 developing countries, Papua New Guinea, The Philippines, Ethiopia and The Gambia.1-7 This paper summarizes the important findings of the study, discusses the implications for the management of infections in young infants in developing countries and discusses future work that could facilitate progress in this difficult area. ETIOLOGY The bacteriology results from these four very different countries produced remarkably similar results. Among infants with clinical signs indicating that infection was likely, the rate of positive blood culture ranged from 4% in The Philippines and 5% in Papua New Guinea (PNG) to 10% in Ethiopia and 11% in The Gambia. The low rate of positive culture in PNG is likely to be the result of the milder nature of illness in infants in that site compared with 16% for the other sites together. The lower culture positivity rate in The Philippines is likely to be the result of the high rate of early antibiotic administration in that community compared with the others. Among the 167 bacterial isolates from blood cultures in the study, 96 (57%) of the isolates were the three main Gram-positive pathogens, Streptococcus pneumoniae, Staphylococcus aureus and Streptococcus pyogenes. This pattern was seen across the sites, with all 3 organisms being isolated from infants in all sites. Staphylococcus aureus was seen more frequently in The Gambia than in the other sites, and S. pyogenes was seen more frequently in PNG than in the other sites. The major role of S. pneumoniae was an important finding of the study, particularly its role as a cause of meningitis in this age group. It will surprise many to find that S. pneumoniae was the cause of 43% of episodes of proven bacterial meningitis in this study and 50% of meningitis episodes in infants older than 1 week of age. Of particular concern is the serotype distribution of the pneumococcal cases. Eight of the 17 pneumococcal meningitis cases were caused by type 2, a rare cause of invasive pneumococcal disease in infants beyond the neonatal period, and a serotype not included in any of the current generation of pneumococcal conjugate vaccines. It appears that even if administered during the first few weeks of life, pneumococcal conjugate vaccines may have only a limited impact on pneumococcal meningitis in early infancy. This finding argues in favor of maternal immunization with the 23-valent pneumococcal polysaccharide vaccine, which does include type 2, as a potential strategy to prevent early infant pneumococcal disease. Other Gram-positive infections, particularly Streptococcus agalactiae (group B Streptococcus) infection, were uncommon in this study. The wide variety of enteric Gram-negative organisms causing invasive disease is as expected. Although most of these infections were found in infants younger than 2 months of age, there were infections caused by Acinetobacter spp., Klebsiella spp. and Escherichia coli in the third month of life, suggesting that susceptibility to these organisms does extend into the third month, although most cases of meningitis caused by these organisms occurred during the first month of life. E. coli was the most common Gram-negative organism found. Most of the E. coli isolates were in infants younger than 1 month of age and most were from Ethiopia. Salmonella spp. was a significant pathogen in all the sites except Papua New Guinea. This has significant implications for treatment. Before this study, recommended empiric therapy for suspected serious infections in young infants in developing countries was penicillin and gentamicin, for the reasons outlined in earlier papers.1 Although still the antibiotic of choice for S. pyogenes infections, penicillin is ineffective against staphylococcal infections in many parts of the world. Penicillin-resistant S. pneumoniae organisms are now found in most parts of the world.8 Although the clinical relevance of penicillin resistance in pneumococcal pneumonia is uncertain, few would feel comfortable treating penicillin-resistant pneumococcal pneumonia in very young infants with penicillin, and there is no doubt that penicillin is ineffective in the treatment of penicillin-resistant pneumococcal meningitis.9 Thus for penicillin-resistant pneumococcal infections in young infants, the drug of choice is a third generation cephalosporin such as cefotaxime, provided cephalosporin resistance is not a problem, This would also provide effective cover against Salmonella infection and other Gram-negative infections. Thus although this study supports the use of ampicillin and gentamicin for the initial therapy of suspected neonatal sepsis, where meningitis is present, or where infection with penicillin-resistant S. pneumoniae or Salmonella is suspected, a third generation cephalosporin should be used. Where skin sepsis or another marker of staphylococcal infection is present, an antistaphylococcal agent should be added. The virology results produced a similar picture across the different sites. Just as in developed countries respiratory syncytial virus is the most important virus affecting this age group, causing considerable respiratory morbidity, but little mortality.10 Indeed the isolation of viruses was associated with a lower risk of death in the Gambian study. Virus infections, particularly those caused by respiratory syncytial virus, are common in this age group in developing countries but appear to be rarely associated with an adverse outcome. In contrast the major role of Chlamydia trachomatis as a cause of neonatal pneumonia was well-demonstrated in Papua New Guinea.11 Particularly disturbing was the finding that C. trachomatis is associated with severe pneumonia and hypoxemia. Further information on this important finding will become available in the future after further assays of specimens from the Gambian study. OUTCOME In this study there were 247 deaths among the 4552 infants enrolled. This group provided a rare insight into early infant death in developing countries, which is a poorly understood area. Of those infants who died 63% (156 of 247) were younger than 1 month of age; of those 51% (80 of 156) were younger than 1 week of age. During the course of the analysis it was found that particular factors associated with mortality included inability to feed, low or high temperature, low white blood cell count and positive blood or CSF culture. The infant deaths in this study will be the subject of an ongoing analysis which will be reported at a later stage. CLINICAL SIGNS One of the primary objectives of this study was the use of the large amounts of clinical data collected to help derive a simple clinical algorithm that could be used to enable health workers in developing countries to identify young infants at high risk of dying for referral to the hospital for inpatient care. This analysis presented a number of challenges that were described in detail in this supplement. The huge data set was approached with a high level of statistical rigor, and the result of a careful, structured analysis was a relatively simple model involving weight, age, temperature, respiratory rate and five clinical signs.7 These were able to predict the presence of serious disease with a high degree of accuracy. Probability values attached to each score provide the clinician with real probability estimates corresponding to each score, allowing the physician to incorporate this information into a rational plan for each infant. This information should provide specialist clinicians working in developing countries with a useful guideline to predict the risk of serious illness in a very young infant. Unfortunately the use of complex probability information is not likely to be very useful for primary care situations in developing countries, where crucial decisions about the referral of infants are usually made by health workers with only minimal training. In such settings an algorithm to assist decision making must include a threshold score beyond which a child is referred. This study has shown that such a system is feasible, even for this most difficult group. Applied in settings similar to those in which this study was conducted, and using a cutoff score designed for maximum sensitivity and specificity for detecting any hypoxemia, radiographic abnormality or positive culture (12 points on the scale, 0.17 predicted probability), the algorithm described in the accompanying paper7 would result in the referral of 80% of infants with any hypoxemia, radiographic abnormality or positive culture, 88% of those with positive cultures or severe hypoxemia and 94% of those who would die of the current illness, at a cost of 49 referrals of infants without positive cultures, positive chest radiograph or hypoxemia for every 100 referrals. If this is judged to be too insensitive, the algorithm can be adjusted to produce 95% sensitivity for any hypoxemia, radiographic abnormality or positive culture (for which the cutoff in predicted risks is 0.10, corresponding to 8 points on the scale shown in Table 5 of Ref. 7). This would result in 60 referrals of infants without positive cultures, positive chest radiograph or hypoxemia for every 100 referrals. FUTURE PROSPECTS It is anticipated that during the coming decade infant mortality in developing countries will continue to decline as a result of vaccination and other specific strategies as well as general improvements in living standards and standards of health care. As the proportion of infant deaths occurring in the first 3 months increases the need to address this group will become more urgent. Although prevention must always be the first priority, the need to improve recognition and treatment of infectious diseases in this age group must also be addressed. As a result of this study we now have a clear view of the etiologic spectrum of agents causing serious infections in very young infants in developing countries. This information, combined with recent data on regional trends in antimicrobial resistance patterns, should help guide health planners and clinicians to the most appropriate antibiotic therapy for this group of infants. However, recognition and referral of sick infants remain a major challenge. The model proposed based on the analysis of this study may be usable in some settings as part of a strategy to train traditional birth attendants or other health workers in contact with neonates. It may need further simplification for use as part of a more general strategy for acute pediatric care such as the Integrated Management of Childhood Illness currently being promoted by WHO. The next step in this process should be the formal evaluation of the algorithm, using an appropriate control group, in a setting with a high neonatal mortality rate. This will demonstrate the true potential for this strategy.

Despite major advances in recent years, infant mortality rates remain high in many developing countries. In 1995, of ∼126 million infants born alive into the world, ∼8 million (6%) died during their first year, 5 million during the neonatal period and many during the second and third months of life.1 In many developing countries neonatal death is systematically underreported because of cultural reluctance to declare neonatal deaths.2 Despite this bias towards underreporting, 98% of the neonatal deaths in the world are believed to occur in developing countries, mainly in Asia and Africa, where the regional neonatal mortality rates average >40 per 1000 live births and where several countries have rates of >60.1 Because most neonatal deaths occur at home, there are few data on the causes of the deaths. It is believed that infection is a major cause, representing a progressively larger contribution in settings with higher neonatal death rates. WHO has developed a case management strategy to reduce mortality from acute respiratory infections (ARI) in developing countries.3 Health workers are trained to identify pneumonia in children >2 months of age using simple clinical signs (fast breathing and lower chest wall indrawing).4 This approach has recently been incorporated into the Integrated Management of Childhood Illness strategy, which combines ARI management with the management of the other important causes of childhood mortality, malaria, diarrheal disease, malnutrition and measles. In the first 2 months of life, when one-third of all childhood ARI deaths are believed to occur, this strategy is broadened to include sepsis and meningitis, in addition to pneumonia, because of the difficulty in distinguishing between these entities clinically. Separate guidelines have been developed for children <2 months of age and are currently being implemented in a number of developing countries as part of the Integrated Management of Childhood Illness strategy. For children age 2 months to 4 years the importance of bacteria in the etiology of pneumonia in developing countries is well-documented, with Streptococcus pneumoniae and Haemophilus influenzae causing the most severe cases.5, 6 In contrast there are inadequate data on the clinical signs and etiologic agents of serious infections in infants <2 months of age in developing countries. Bacterial infections can lead very rapidly to death in this age group, prevention of which requires early detection based on recognition of simple clinical signs by the family and health workers. Following expert clinical advice and published etiologic data, predominantly from developed countries, WHO manuals and training materials currently recommend referring young infants with suspected pneumonia, sepsis or meningitis for treatment with parenteral penicillin and gentamicin. To improve case management guidelines for health workers responsible for the care of young infants in both outpatient and inpatient settings, a multicenter study was set up under the auspices of WHO in four developing countries, Ethiopia, The Gambia, Papua New Guinea and The Philippines. This paper describes the rationale for the study and summarizes the data that existed before the study. ETIOLOGY OF SERIOUS BACTERIAL INFECTIONS IN YOUNG INFANTS IN DEVELOPING COUNTRIES The immunologically naive neonate is particularly susceptible to bacterial infections. For how long this high level of susceptibility to infection persists is unclear. During the second and third months there is a steady maturation of the immune system, although the child remains susceptible to infection with organisms like H. influenzae type b for several years. In developed countries serious neonatal bacterial infections are now usually caused by Escherichia coli or Streptococcus agalactiae (group B Streptococcus), although in the past Streptococcus pyogenes (group A Streptococcus) was a major cause and Staphylococcus aureus has been responsible for many serious nursery outbreaks. Gradual shifts in the organisms responsible for serious bacterial infections have been documented in several hospitals.7-9 Young infants in developing countries suffer a much higher mortality than in developed countries, and much of this mortality is caused by bacterial infection. There have been a number of studies on the etiology of neonatal sepsis in developing countries. The results of 15 studies describing >3000 blood culture-positive cases of neonatal sepsis in India, Africa, the Middle East and the West Indies were reviewed.10-24 In 6 studies Klebsiella spp. was the most frequently isolated organism, in 3 it was Staphylococcus aureus and in three it was E. coli. Overall the most frequently isolated organism was Klebsiella spp. followed by E. coli, Staphylococcus aureus and Pseudomonas spp. There were only 32 isolates of S. agalactiae, most from one small study in the West Indies,22 although in a more recent publication it was found to be the most important cause of neonatal meningitis in Harare, Zimbabwe.25 The other important feature of these data is the large number of infections with Klebsiella spp. and Staphylococcus aureus. Most of these studies were hospital-based and many of the infections are likely to have been hospital-acquired. However, one study from India of infections among home-delivered neonates also found many cases of Klebsiella spp. and Staphylococcus aureus infections.26 In the more recent studies the staphylococci were almost uniformly resistant to penicillin; 10 to 57% of Klebsiella were resistant to gentamicin.18 The particular problem of early neonatal sepsis associated with amnionitis was explored in detail in Addis Ababa, Ethiopia, in the 1970s. In that city the rate of fatal neonatal sepsis associated with amniotic fluid infection was estimated at 21.8/1000 live births.27 Cultures from a series of 339 autopsies yielded a wide variety of organisms. Ureaplasma urealyticum was the most commonly found organism, being found in 44% of infected placentas and in only 6% of control placentas from healthy infants. It was also recovered from 26% of infant lung specimens. In a prospective study from the same city, blood cultures were taken from 945 infants younger than 8 days of age with suspected sepsis; bacteremia was present in 344, of whom 260 were <72h old.28 These yielded a variety of organisms, particularly E. coli, Klebsiella spp. and Staphylococcus aureus. No pneumococci or H. influenzae were isolated although pneumococci were isolated from a few specimens (0.9%) in the postmortem study. There have been several studies from developing countries looking at meningitis in young infants. Soni, from the University of Garyounis, Libya, reporting on neonatal meningitis, found Klebsiella spp. and Staphylococcus aureus to be the dominant organisms.29 In two studies from Dakar, Senegal, where large numbers of patients with meningitis have been studied, enterobacteria, in particular Salmonella spp. and S. pneumoniae, were the main causes of meningitis in infants younger than the age of 3 months.30, 31 In the more recent of those studies 13 of 42 cases of meningitis in infants younger than 3 months of age were caused by S. pneumoniae and one was caused by H. influenzae. In a retrospective study of 36 cases of neonatal meningitis presenting to University College Hospital, Ibadan, Nigeria, 10 were caused by S. pneumoniae, 8 by E. coli and 2 by H. influenzae.32 Similar results were obtained in a smaller study from Yaoundé, Cameroon Republic.33 There, of 14 cases of meningitis in children younger than 2 months of age, 5 were caused by S. pneumoniae and one was caused by H. influenzae. In a recent study of H. influenzae meningitis in the Gambia, 6% of cases were <3 months old.34S. agalactiae was not found in most of these studies, but it has been shown to be an important cause of neonatal meningitis in Zimbabwe25 and South Africa (K Klugman, personal communication). Despite the large published studies of neonatal sepsis that focus on Klebsiella spp. and Staphylococcus aureus as the main causes of neonatal infections, there remained suspicion that H. influenzae and S. pneumoniae may play a previously unrecognized role in neonatal sepsis and pneumonia. Both these organisms are difficult to isolate from blood culture, and minor alterations in methodology can substantially reduce the number of isolates from a laboratory. Because vaccination of 3-month-old infants against H. influenzae type b is likely to become routine in many parts of the world and similar conjugate vaccines against S. pneumoniae are currently being evaluated, it is important to know how much disease caused by these organisms occurs before vaccination age, so that other specific measures to protect young infants, such as maternal immunization, can be explored. Several of the above studies support the suspicion that S. pneumoniae and H. influenzae are underestimated as causes of serious infections in young infants in developing countries.30-32 Both organisms are recognized as rare causes of neonatal sepsis in developed countries. Carriage rates of these organisms are much higher in neonates and older children in developing countries,35 so it is reasonable to expect that they may play a more significant role in neonatal sepsis in those countries. OBJECTIVES OF THE MULTICENTER STUDY The objectives of this study were to improve the recognition and the management of serious infections in young infants in developing countries. Recognition of serious infections must be undertaken at the community level by those health workers who have contact with young infants born into poor circumstances. In most cases these are community health workers or traditional birth attendants, usually with minimal training. Simple signs are needed that can be easily taught and applied to detect appropriate infants for referral to hospital. There are two approaches to developing an algorithm composed of such signs. One approach is to assemble a group of experienced experts in the field and develop a consensus opinion. The other is to conduct a prospective study to evaluate the possible signs. This is the course we chose to follow. Discussions held between clinicians working in developing countries before the start of the study revealed three general views of the etiology data. One view based on published studies held that because Klebsiella spp. and Staphylococcus aureus are the established causes of neonatal sepsis in developing countries, recommended therapy should be directed against these organisms and further studies are not required. The second view held that blood culture studies dominated by hospital-acquired infections from institutions with inadequate bacteriology should not be taken as a guide to the etiology of community-acquired infections that were the likely cause of most neonatal deaths and were probably caused by the same pathogens as in developed countries, S. agalactiae and E. coli. The third view, favored by most, held that existing studies were inadequate for the reasons outlined above and the causative organisms in developing countries are likely to be different to developed countries, with S. pneumoniae and H. influenzae probably being more important in developing countries. With this degree of uncertainly the need for a careful study with the use of high quality bacteriology was clear. There were so few data on the role of viruses and atypical organisms in this age group in developing countries that it was considered essential that these agents be sought in as many sites as possible. For reasons of sample size and representativeness a multicenter study was considered preferable. Sites were selected in Asia, Africa and Latin America for their ability to perform the study. The group in the Latin American site withdrew because of political instability leaving two African sites and two Asian sites, The Gambia, Ethiopia, The Philippines and Papua New Guinea.

Objective. Determine the bacterial and viral etiology of severe infection in young Papua New Guinean infants as part of a multicenter study in four developing countries aimed at improving case management guidelines.Methods. Between March, 1991, and April, 1993, children aged <3 months were recruited at the outpatient department of Goroka Base Hospital, Papua New Guinea (PNG). Children with predefined inclusion criteria were enrolled, a history was taken and clinical examination was performed. Blood and urine were collected from children with signs suggestive of severe disease together with eye, umbilical and pernasal swabs as appropriate. Nasopharyngeal aspirates (NPAs) were collected from children with and without signs of severe disease for identification of viruses and Chlamydia trachomatis by direct fluorescent antibody staining.Results. 3280 infants were triaged and 2168 enrolled, among whom 968 had signs suggestive of severe disease. Group A Streptococcus (Streptococcus pyogenes) and Staphylococcus aureus were the most important bacterial pathogens isolated from children <1 month old with severe infections, and Streptococcus pneumoniae, S. pyogenes and Staphylococcus aureus were most important in older children. Of 292 eye swabs 19 (7%) grew Neisseria gonorrhoeae. Of 116 umbilical swabs 51 (44%) grew S. pyogenes and 45 (39%) grew Staphylococcus aureus. Respiratory syncytial virus was the most important viral cause of acute lower respiratory infection.Conclusions, S, pyogenes, S. pneumoniae and Staphylococcus aureus are important causes of severe infection in young children in the PNG; highlands. It is necessary to improve access to clean water, promote hand-washing in the hospital and at home and investigate further the use of maternal immunization for the prevention of severe disease in young infants.

It is estimated that 5 million deaths occur in the neonatal period annually. Of these, ∼97% are in developing countries and >40% are believed to be caused by infection.1 In developing countries much of the effort to control neonatal mortality has concentrated on reducing the risk of perinatal infection by improving the care of pregnant mothers. For example referral of high risk pregnancies, maternal immunization for tetanus, improved care of infants after delivery with emphasis on umbilical cord care and the early introduction of breast-feeding. The treatment of infants with established infection remains unsatisfactory. Many cases of neonatal infection never reach treatment facilities, and the case-fatality rate for those that do ranges from 13 to 69%.2 One approach that has been used in dealing with other pediatric problems such as diarrhea and pneumonia has been to train clinicians, nurses and local health workers (e.g. traditional birth attendants or community health workers) to recognize clinical signs of illness and to refer children for treatment. In developing countries treatment decisions are based largely on the history and clinical examination because of the limited availability of laboratory facilities. More accurate methods of diagnosis could help improve decisions about the need for additional diagnostic tests or therapy in developing country settings. Most previous studies of the accuracy of the clinical examination in detecting serious bacterial illness in infants younger than 3 months of age have been conducted in developed countries. Studies in the US, United Kingdom and Australia3-5 have given conflicting results about the accuracy of observational scales in detecting serious bacterial illness. The small numbers of patients with serious bacterial illness who can be enrolled in developed countries have limited previous studies. This has made it difficult to evaluate the importance of large numbers of clinical variables and to use sophisticated statistical approaches to evaluate the relationship between clinical findings and diagnoses. In addition previous studies of serious bacterial illness have focused simply on the presence or absence of infection, an outcome that is less clinically relevant than one that also enables clinicians to assess the severity of illness. Finally developed country studies have placed greater reliance on the value of laboratory examinations and less emphasis on the clinical examination.3, 6 In developing countries laboratory facilities to perform tests such as the blood count and chest radiograph are often unavailable, and clinical decisions must be made without them. The purpose of this study was to develop a clinical prediction instrument that would identify infants at risk of serious bacterial illness on the basis of clinical examination findings. We hoped that the study would lead to a relatively simple method that could be used by nonphysicians and physicians to detect children at increased risk of adverse outcomes of serious bacterial illness. METHODS The design of the study has been described in detail previously.7 Briefly it was conducted at hospitals or outpatient clinics in Ethiopia, The Gambia, Papua New Guinea and The Philippines that see large numbers of sick infants. At each site infants younger than 91 days of age seen consecutively for acute care with chief complaints indicating possible infection were eligible to be included in the study. Entry criteria were intended to include infants with a wide spectrum of illness severity and to ensure that virtually all infants with serious infection would be included. Exclusion criteria are described elsewhere in this supplement.7 All infants underwent a standardized history and physical examination to assess the presence or absence and the degree of severity of signs and symptoms believed to be associated with bacterial disease (Table 1).8-10 Candidate predictors considered included demographic variables, historical variables, vital signs and physical examination findings. A pediatrician conducted the examination at three of the sites. Pediatric nurse practitioners performed the examination at the site in Papua New Guinea.TABLE 1: Clinical history and examination findings evaluated in model development Infants with prespecified symptoms associated with possible bacterial infection underwent a laboratory evaluation that included blood culture, white blood cell count and chest radiograph. Specific criteria were also used to identify infants for lumbar puncture. All infants enrolled in the study underwent pulse oximetry. Urine cultures were not systematically obtained in all sites because it was not feasible to collect suprapubic specimens or to catheterize infants in all the settings. A random sample of patients not meeting criteria for laboratory testing underwent chest radiograph, blood culture and blood count to assess the frequency of bacterial infection among infants not meeting criteria for laboratory evaluation. Decisions regarding treatment were made on a clinical basis. Bacteremia was defined as the growth of a known pathogen in cultures of blood or cerebrospinal fluid (CSF). Meningitis was defined as a positive CSF culture or >10 polymorphonuclear cells in a nonbloody CSF specimen. Patients with grossly blood-stained lumbar punctures were classified as not having meningitis unless the CSF culture was positive. The criteria used to classify organisms as contaminants are described elsewhere.7 Oxygen saturation was measured after the clinical examination with a Nellcor N-200 (Nellcor Inc., Hayward, CA) pulse oximeter as previously described. Oxygen saturation measurements from sites above sea level were adjusted to correspond to readings at sea level. Pneumonia was diagnosed on the basis of chest radiographs interpreted without clinical information by a panel of three radiologists.7, 11 Study infants were defined as having pneumonia if all the radiologists who read the radiograph considered it to show probable or definite pneumonia. Infants with uninterpretable films or with missing radiographic information were classified as having a normal radiograph. Study outcome measure. The primary study outcome measure was an ordinal scale that summarized the presence or absence of disease, as well as its severity. This measure was selected because it would allow infants to be included in the clinical prediction rule who might have negative cultures (e.g. those with low oxygen saturation and positive radiographs), but who still might benefit from treatment because they were at increased risk of bacterial infection. Data about oxygen saturation, chest radiography, blood culture, CSF culture and cell count and mortality were combined to produce a measure that grouped diagnosis in increasing severity. Diagnoses were ranked in a hierarchic fashion based on their association with the most severe outcome, death. Diagnoses not strongly related to death were examined with respect to their association with the presence of a positive blood culture or CSF result. Outcome variables with similar associations with death or bacterial infection were included in the same category. This approach produced a ranking of severity with four levels of disease severity: (1) no abnormality; (2) mild hypoxemia (90% ≧ SaO2 < 95%) or radiologic pneumonia; (3) severe hypoxemia (SaO2 < 90%) or bacteremia or meningitis; and (4) death. As will be discussed later the main study analysis considered only a three level ordinal scale that did not include death as an explicit outcome category. From a statistical standpoint an ordinal ranking assumes that the variables can be ordered in terms of clinical importance or disease severity without assuming interval spacing between the variables. Such a procedure also makes more efficient use of data for statistical analyses. Children who died >7 days after discharge from the hospital were not counted as a death for the purposes of this study. Deriving the clinical prediction rule. The goal of the statistical analysis was to select a limited set of clinical predictors that could reasonably be used in clinical settings and that could predict infants at risk of serious bacterial illness or death. The analysis was designed to reduce the total number of variables in the model by grouping related subsets of clinical findings and summarizing each group of findings with a single score. We sought to avoid selecting variables to include in the model on the basis of their individual statistical associations with specific diagnoses because the multiple tests involved in this approach would exaggerate the strength of the relationship between clinical findings and the diagnoses. The statistical methods of the analysis are described in detail elsewhere.12 The derivation of the clinical prediction rule took place in three steps. Step 1. Clustering variables. Statistical variable clustering methods were used to place the 51 individual clinical signs and symptoms into groups or "clusters" so that the correlation of clinical findings within clusters was high and the between cluster correlation was low. The clustering procedure took place before the evaluation of the association between the clusters and the diagnoses. After the clustering procedure clinicians on the study team were asked to identify and rank the relative severity of the findings within the clusters. After this exercise a check was made to identify major disagreements between physicians' scoring and the outcome patterns. The only clustering that was rescored after examining patterns of the association with the ordinal scale was the auscultation cluster. Here a decision was made to use as the only variable the presence or absence of crepitations. Clinicians' ranking and scoring decisions were carried out before any analysis relating individual clinical signs to the diagnoses. We selected a scoring method in which the cluster score was the score of the most severely abnormal finding in the cluster. Measurements of vital signs (age, temperature, respiratory rate, weight) were not included in the clusters. The weight-for-age Z score was used as the measure of infant size. We made this decision because other anthropomorphic variables of size in this age group are highly correlated with weight-for-age. Once each cluster of clinical signs was scored, we entered the cluster scores into a logistic regression model. The clustering of signs thus resulted in significant reduction in the number of variables in the model so that fewer regression coefficients needed estimation. Step 2. Estimating the overall accuracy of the clinical findings. The association of the clusters of clinical variables, vital signs measurements and age with the diagnostic outcome was assessed by fitting a logistic regression model for an ordinal response variable (continuation ratio model) using penalized maximum likelihood estimation. Penalization is a method of discounting fitted regression coefficients to avoid overfitting. All two way interactions between predictors and the category of the ordinal response being predicted were included in the model so that predictors could be included that might be associated with one level of the scale but not with another. For example the presence of crepitations might be more highly associated with chest radiographic evidence of pneumonia but less strongly associated with a positive blood culture. Continuous variables such as temperature and respiratory rate were entered in the model through the use of restricted cubic splines (piecewise cubic polynomials) to not restrict the model to the assumption of a linear or even monotonic association of these variables with the ordinal scale.12 Bootstrap methods were used to assess the reproducibility of the model after penalizing for overfitting. For the analyses reported here the main emphasis was on developing a clinical prediction instrument that assigned infants to a diagnostic category on the basis of chest radiography, blood and CSF cultures and oxygen saturation without regard to mortality. We were interested in this model because many of the deaths occurring in the context of the study took place within hours of admission. This model might identify children more likely to benefit from antibiotic therapy. Infants who died were assigned to lower level outcomes according to the presence or absence of the other diagnoses. The net effect of this was that about two-thirds of deaths were classified in another abnormal diagnostic category. The discriminating ability of the three level model was compared with that of the four level model using ordinary binary logistic regression methods. Step 3. Simplifying the model for clinical use. Simplification of the full model was accomplished by treating it as a "gold standard" against which models with fewer variables could be compared. We used ordinary least squares regression and step-down techniques to predict the predicted log odds of the full model and thereby eliminated variables that explained little of the variation in the full model.12 The objective of the selection procedure was to maintain the squared linear correlation between the full model predicted log and the reduced model log odds (R2) above 0.95 with a minimal decrease in the measure of discrimination (Somers' Dxy) against the outcome. Variables (clusters or individual variables) were deleted sequentially so that the first variable deleted was the one that resulted in the smallest drop in the squared correlation between the reduced model predictions and the "gold standard" model prediction. Actual outcomes were not used as the dependent variable in this model simplification procedure in that conducting stepwise variable selection in this way has a low chance of selecting the "correct" model because of problems of multiple comparisons and arbitrary levels for stopping.14 Assessment of model accuracy. We summarized the diagnostic accuracy of the model as its ability to discriminate between infants without any diagnostic abnormality and infants in the categories of adverse events. We report the area under the receiver operating curve (ROC) for each level in the ordinal scale. We used the overall likelihood ratio chi square to summarize the amount of information in the models. We have not presented odds ratios for individual variables because of model complexity. Bootstrap methods were used to validate the ability of the model to separate infants with no abnormalities from those in the moderate or severe category.14 The bootstrap procedure also adjusts the observed predictive accuracy for overfitting caused by evaluating the performance of the model in the same set of data used to develop it. RESULTS A total of 8418 infants less than 91 days of age were triaged in 4 sites, of whom 4552 satisfied the criteria for enrollment in the study and underwent a full history, physical examination and pulse oximetry. Table 2 lists the demographic and the clinical characteristics of the study infants at presentation. Similar proportions of infants were enrolled in Ethiopia, The Gambia and the Philippines. More infants than any of the other sites (47.6%) were enrolled in Papua New Guinea. As described previously infants in Papua New Guinea were substantially less likely to be hospitalized or to have died.7Table 2 also shows the relationship between the presence of clinical findings and diagnostic abnormalities. For example the presence of a history of fast breathing increased the probability of any abnormality (mild hypoxemia or pneumonia; severe hypoxemia bacteremia or meningitis) from 6.9% to 19.2%. It increased the probability of a more serious abnormality (severe hypoxemia, bacteremia or meningitis) from 4% to 8%.TABLE 2: Characteristics of infants included in the study and the relationship of clinical findings at presentation to diagnostic outcomes Of the 4552 infants in the study sample 2398 met the prespecified criteria for laboratory evaluation and underwent blood culture and chest radiography; 507 met the criteria for a lumbar puncture. The 67 infants with gross blood in the CSF specimen were excluded from the analysis of abnormal CSF counts. The remaining 2154 infants had no clinical findings requiring further evaluation. The 4552 infants were classified into the following diagnostic categories: 3467 (76.2%) had no abnormality; 450 (9.9%) had pneumonia or mild hypoxemia; 386 (8.5%) had bacteremia, meningitis or severe hypoxemia; and 249 (8%) died. Almost all deaths (86%) took place in the hospital. Nineteen (8%) children were taken home moribund and 17 (7%) others were known to have died as outpatients. Of 1979 children in the group not meeting criteria for laboratory evaluation, 15 (0.8%) returned for a second evaluation with pneumonia, mild hypoxemia, bacteremia, meningitis or severe hypoxemia. None of these infants presented initially with such diagnoses. To estimate the frequency of serious infection in infants not eligible for laboratory evaluation, 175 infants (8%) were selected randomly from those not eligible for complete laboratory evaluation to have blood and urine cultures obtained. The number of infants selected was less than planned because most parents in this group refused to permit blood drawing. Among the infants in whom laboratory tests were obtained, 4 (2%) had positive blood cultures. All the positive cultures occurred in infants from The Gambia who had skin infections with scabies. A fifth infant had a negative initial evaluation but returned the day after enrollment with meningitis. An additional 6 (3%) infants had mildly decreased oxygen saturation, one had a focal infiltrate on chest radiography, and the other two infants had diffuse or nonspecific chest radiographic findings. None of the infants in the sampled group died. We did not correct for the possible verification bias produced by the study design because the effect would be minimal and because the random sample appeared to have overestimated the frequency of adverse events in the group of children not meeting criteria for laboratory evaluation. Development of the clinical prediction rule. For the analyses reported here the main emphasis was on the development of a clinical prediction instrument in which infants were assigned to a diagnostic category without considering whether they died. The model had three categories rather than four, because infants who died were assigned to lower levels of severity according to the presence or absence of other diagnoses. Infants who died were reassigned as follows: 84 (33.7%) had no evidence of any other serious abnormality or were missing data about laboratory procedures and were assigned to the no abnormality category; 40 (16%) had pneumonia or mild hypoxemia; and 125 (50.2%) had bacteremia, meningitis or severe hypoxemia. For infants reclassified into the no abnormality or pneumonia/mild hypoxemia category, virtually all were suspected of having a serious bacterial illness by the clinicians caring for the patients. In approximately one-third of these cases, it was not possible to obtain a chest radiograph or a blood culture. Thus some infants who would have met the criteria to be included in a higher level may have been misidentified. The overall effect of this reclassification is therefore likely to lead to a conservative estimate of the accuracy of the model. The variable clustering procedure led to an aggregation of the 51 clinical history and examination findings into 14 clusters of variables.12 These clusters were included in a model along with the infants' age, respiratory rate, temperature and weight-for-age Z score to predict the diagnostic categories. The model for classifying infants into three levels ignoring death (no abnormality; pneumonia/mild hypoxemia; or bacteremia, meningitis, severe hypoxemia) had a bootstrap-validated ROC area of 0.836 for discriminating infants with no abnormality from those with any abnormality and an ROC area of 0.872 for discriminating infants with no abnormality from those with more serious ones. To illustrate the accuracy of the model we plotted the number of infants with any abnormality relative to the number of infants without an abnormality at each probability of disease (Fig. 1). Figure 1 also shows the likelihood ratios at a number of probability levels.Fig. 1: A, histogram comparing the number of infants with no abnormality with those with any diagnostic abnormality at each level of predicted probability; B, histogram comparing the number of infants with no severe abnormality with those with a severe diagnostic abnormality at each level of predicted probability. Likelihood ratios were estimated by the following procedure. First, the data were stratified into "normal" and "abnormal" subsets. Within each subset kernel density estimates were computed, to estimate the probability density function for the predicted probabilities. The ratio of density functions for abnormal vs. normal was then computed, and these ratios were smoothed with the use of the lowest nonparametric regression smoother. The labels "Any Abnormality" (A) and "Severe Abnormality" (B) apply to the portions of the histograms to the right of the vertical lines.The three level model was compared with one in which death was added as a fourth diagnostic category (Table 3). The accuracy of the four level model was slightly better in detecting infants at different degrees of illness severity, but the difference between the models was small. A model with only vital signs, respiratory rate and weight explained ∼60% of the total variation in outcomes. This model had an ROC area of 0.773 for predicting any diagnostic abnormality. We also compared the accuracy of the models in the four study sites and found <5% difference in ROC areas.TABLE 3: Diagnostic accuracy of three and four level logistic regression models in detecting infants with differing outcome abnormalities The reference categories for all models are: infants with no outcome abnormality; infants with any outcome abnormality; infants with severe hypoxemia, meningitis, or death; and death. All models are binary logistic models. Simplification of the model. Although the full models provided good discriminating ability, they would be difficult to use clinically because they would require clinicians to assess a large number of clinical predictors and to compute cluster scores. We simplified the three diagnostic category model to facilitate its use in clinical settings by identifying a smaller group of findings that could be used as a guideline for treatment by clinicians in small hospitals and by peripheral health workers in dispensaries and outpatient facilities.13 Ordinary least squares regression was used to eliminate progressively those groups of clinical findings that contributed little to variations in predictions given by the full (gold standard) model. This procedure was implemented for groups of findings as well as for individual clinical findings. Before the elimination procedure we excluded signs that were impractical to measure accurately in young infants (e.g. heart rate), those found to be unreliable (respiratory distress, smiling and attentiveness) and those with low prevalence (stridor and hypotonia). The procedure was conducted first for the model predicting any abnormality and then for the model predicting a more serious abnormality. This approach led to two simplified clinical prediction models with similar abilities to predict the presence of any abnormality. A model that used clusters of clinical signs included three vital signs (temperature, respiratory rate, weight-for-age), the infant's age and five clusters of clinical findings: auscultation; respiratory effort; evidence of neurologic infection; inability to feed; and lethargy. This model had an R2 of 0.973 in representing the predictions of the full model and an ROC area of 0.833 in predicting any abnormality. A second model with individual clinical findings was created by selecting a reduced model from a full model containing all the individual clinical findings. In this model binary (dummy) coding was used for each clinical finding (i.e. the findings could be used by rating them as present or absent). The approach resulted in a model with seven specific clinical findings (inability to suck, crepitations, cyanosis, history of convulsions, definite lower chest wall indrawing, failure to arouse with minimal stimulation, history of change in activity), as well as respiratory rate, age, temperature and weight for age. The model had an ROC area to discriminate children with any abnormality in the ordinal scale of 0.832 and an approximation accuracy of 0.954 against the best model. This process was repeated to create a simplified model to predict more serious abnormalities (see Appendix). In its simplest form the model predicting a serious abnormality had an ROC area of 0.866. Use of the model in clinical settings. We developed a tabular version of the simplest models for use in clinical settings (Tables 4 and 5). To create the tabular presentation individual clinical findings were scaled so that the value (or "risk points") of each finding included was proportional to its log odds ratio in the model. For continuous predictors we solved for ranges that corresponded with whole numbered points. Because respiratory rate was associated with a different probability of an abnormality depending on the infant's age, the table for respiratory rate has multiple columns. Because all of the clinical findings are scored as yes/no items, each of the seven clinical findings receives the number of points indicated in the table if the finding is present and no points if the finding is absent. For example a 4-day-old infant with a temperature of 37.9°C, respiratory rate of 76/min, weight-for-age Z score of 2 and crepitations would receive a total of 17 points and a probability of 34% of having at least mild hypoxemia or pneumonia and a probability of 12% of having severe hypoxemia, bacteremia or meningitis.TABLE 4: Tabular version of simplified clinical prediction model to predict any abnormality* TABLE 5: Probabilities of abnormalities We also compared the accuracy of the 12-item simplified 3 diagnostic category model with that of the 12 clinical signs in the WHO guidelines for the management of the sick young infant. These signs include: temperature >37.5°C or <36.5°C, baby feels cold to touch, presence of convulsions, fast breathing (>60 breaths/min), severe chest indrawing, nasal flaring, grunting, presence of a bulging fontanel, pus draining from the ear, red umbilical stump, pustules, lethargy/unconscious and less than normal movement. In detecting infants younger than 60 days of age with any abnormality, the WHO sick child criteria had an ROC area of 0.656 compared with the 3 level model ROC area of 0.838. Since the study was completed the WHO algorithm for the assessment of infants under 2 months of age has been modified to include inability to attach well to the breast, inability to suckle and inability to feed as indicators of possible serious infection. These certainly improve the predictive power of the algorithm. DISCUSSION This study indicates that clinical examination is of value in identifying infants at risk of serious bacterial illness and in estimating the severity of disease. Some of the most valuable information comes from the child's age, vital signs and size. A limited number of clinical findings add additional diagnostic information. The clinical findings can be combined into a predictive instrument that is likely to be of use clinically. Despite the simplicity of the instrument, it preserves the powerful information attained by using vital signs on a continuous basis. The World Health Organization has promoted the use of simple clinical signs in the management of sick infants. However, there have been few studies of the accuracy of the management guidelines. The clinical findings included in the current sick young infant management algorithm were quite similar to those identified in this empirical study. The greater accuracy of the findings identified here may be attributable to two characteristics of the instrument: the more detailed use of an infants' vital signs; and more accurate cutoff points for determining whether a finding is abnormal. The clinical prediction instrument uses temperature, anthropomorphic measurements and respiratory rate as continuous variables and thus avoids the loss of information that occurs when variables are categorized.15 Like other previous studies of the value of clinical findings in detecting streptococcal pharyngitis16 and dehydration,17 the results presented here suggest that empirical evaluation of the WHO clinical management algorithms can lead to more accurate information about which clinical findings to use and how to combine them. Studies from developed countries differ in their assessment of the value of the clinical examination in detecting serious bacterial illness in infants. A study by Baker3 in a US emergency room found that an existing observation scale to detect serious bacterial illness9 was of limited diagnostic value in identifying infants younger than 2 months of age with bacterial illness. In contrast a study by Hewson and Gollan4 concluded that the presence of any of five clinical findings (drowsiness, chest indrawing, generalized pallor, a history of feeding <50% of normal, decreased activity) had a sensitivity of 91% and a specificity of 72% in predicting "serious" illness or the need for hospital admission in children younger than 6 months of age. Although the present study is not directly comparable with the Hewson study, most of the same variables, except vital signs which were not considered by Hewson, were identified as important. Our findings contrast with previous studies in developed countries for several reasons. First, we enrolled a large number of infants, many of whom had severe diagnostic abnormalities. The large number of abnormalities permitted the use of more sophisticated statistical procedures than has previously been possible. This allowed us to explore more completely the relationship of the clinical findings to diagnoses. The approach to model development we used also permitted us to examine nonlinear relationships of clinical findings to such diagnoses. Forcing variables to fit a linear model form when the true relationship is nonlinear results in loss of information and discriminating ability. Indeed variables with complex relationships to the diagnoses, such as vital signs, provided considerable information and may be responsible for the greater discriminating ability of the clinical prediction rule. Our study suffered from a number of limitations. Almost one-half of the data were from Papua New Guinea, the site with the lowest proportion of severely ill children. We decided to combine data from this site with other data because there was no a priori evidence that the clinical characteristics observed there would have a different relationship to serious bacterial illness than those observed elsewhere. Another limitation was the inconsistent collection of specimens for urine culture. A failure to identify consistently urinary tract infections may have reduced the number of infants in whom bacterial illness was detected. This would have reduced the discriminating ability of the clinical findings. Finally the model has not yet been validated prospectively in a new sample of patients. Although bootstrap validation methods indicate that the rule is likely to perform well in other settings, prospective validation is needed in the future. Although we designed the study to assess the importance of "verification" bias18 by conducting blood cultures and chest radiographs on infants who did not meet the criteria for evaluation, there were fewer infants than anticipated on whom these tests were performed. The proportion of infants with positive laboratory results in this sample indicates that they may have had more significant disease than the group from which they were sampled as the mothers of sicker infants were probably more likely to accept investigation. Future projects will need to develop approaches to enroll or follow patients so that the importance of verification bias can be assessed. Taken together the approach to model development that we pursued and the decisions we made regarding variable definition are likely to have underestimated the diagnostic accuracy of the model. This study provides strong evidence of the diagnostic value of clinical findings that can be observed on examination. The approach we developed provides an empirical basis for developing management recommendations and thus extends the methods that the WHO currently uses to develop diagnostic algorithms. In addition the study suggests that an approach that permits the classification of diagnoses according to disease severity rather than into dichotomous groups is feasible and may provide a more informative approach for future studies. Finally representing the model so that the probability of a diagnosis can be computed simply from a chart will allow clinicians to account for multiple clinical findings at once and thereby estimate the probability of each of several diagnoses for any given patient. Used in a developing country setting where laboratory testing is not readily available, the use of the clinical prediction model could improve the ability of clinicians to identify sick infants. Despite the increase in accuracy beyond the WHO management criteria, the possibility exists that children with serious bacterial illness will be missed. In developing country settings the possibility of false reassurance could be reduced by careful follow-up of infants classified as low risk. In developed countries information from the clinical prediction instrument could be combined with data from the clinical laboratory to enhance clinician's diagnostic accuracy. Validation of the model in both high and low prevalence settings, as well as studies to determine how to maximize the utility and acceptability of the prediction instrument in different clinical settings and by different levels of health workers, should be conducted in the future. ACKNOWLEDGMENTS We thank the many nurses, doctors and field workers at each of the study sites who contributed so much to this study, but who are too numerous to mention individually. The support of WHO and the contribution of many WHO staffs was essential to the completion of this project. In particular J. Tulloch, J. Bryce and C. John made significant contributions to this project. Finally we acknowledge the patience and understanding of the more than 4000 families of sick infants enrolled in the 4 study sites who tolerated the intrusion of this study at a time when the families were under considerable stress.

Children aged 1-59 months admitted to Goroka Base Hospital with signs suggestive of meningitis were recruited to determine what proportion of such children have clinical or bacterial meningitis and to investigate the bacterial aetiology. A laboratory classification of definite, probable, possible, indeterminate and no meningitis was established. Thirty per cent of 697 children had a final clinical diagnosis of meningitis, 12% had culture-proven bacterial meningitis (case fatality rate 34%) and 10% had probable or possible meningitis. Inability to feed, vomiting, drowsiness, "staring eyes" and haemoglobin < 9 g/dl in addition to the classical signs of meningitis were associated with increased mortality. Isolates from cerebrospinal fluid were 62 pneumococci, 22 Haemophilus influenzae type b (Hib) and one Neisseria meningitidis. Including blood culture-proven and antigen-proven Hib disease, Hib and pneumococci accounted for 44% and 46% of bacterial meningitis, respectively, and 23% of pneumococci were intermediately resistant to penicillin. Inability to feed, bulging fontanelle, convulsions in young children, neck stiffness, fever and "staring eyes" were all independently associated with bacterial meningitis. Conjugate Hib vaccine must be given to infants as early as possible. Conjugate pneumococcal vaccines, maternal immunization with 23-valent vaccine and pneumococcal protein vaccines are under investigation for prevention of pneumococcal disease.

Despite great improvements in infant mortality rates in recent years, the mortality of young infants, particularly neonates, remains high in most developing countries. In 1995, 47 countries were believed to have neonatal mortality rates in excess of 40/1000 live births. It is estimated currently that ∼5 million neonates die in developing countries each year.1, 2 Outside the first month of life most other infant deaths occur during the second and third months of life. By contrast, in industrialized countries, neonatal mortality rates are ∼5/1000 live births and most of those deaths are premature infants in intensive care units, a group not always included in developing country data. Infection is estimated to cause 30 to 40% of neonatal deaths in developing countries.2 There are several prevailing views of the etiology of neonatal infections in developing countries. Published series suggest that Staphylococcus aureus and Klebsiella spp. are the main pathogens,2, 3 although hospital series include many hospital-acquired infections, and the bacteriologic techniques in many places would not permit the growth of fastidious organisms like Streptococcus pneumoniae and Haemophilus influenzae. Before this study many believed that S. pneumoniae and H. influenzae were important pathogens in young infants in developing countries, although there were few published reports to support this view. In industrialized countries Escherichia coli and Streptococcus agalactiae (group B Streptococcus) are the most important neonatal pathogens, and it is unclear what role these organisms have in developing countries. On the basis of available data WHO and other authorities have recommended that serious infections in very young infants in developing countries should be treated with penicillin and gentamicin initially.4 In practice many different combinations are used based on local interpretations of existing data; most regimens include a penicillin and an aminoglycoside. Some use chloramphenicol and, where they are available, third generation cephalosporins, particularly cefotaxime, are used. Although treatment of established cases is important, control of infections in young infants ultimately rests on prevention.2 Although efforts are under way in many developing countries to improve perinatal care, attention is also being given to maternal immunization as a means of protecting young infants from infection with passively acquired maternal antibody. This has been used with great success to control neonatal tetanus in developing countries5 and has recently been tried to prevent neonatal infections caused by encapsulated bacteria such as S. agalactiae, S. pneumoniae and H. influenzae.6-9 Developments in this area provide another valid reason for investigating the etiology of infections in young infants in developing countries. To address this question a multicenter project was set up to examine the etiology and clinical signs of serious infections in infants younger than 90 days of age in developing countries. The objective of the studies described in this paper was to determine the bacterial and viral causes of serious infections in young infants in developing countries. Only the bacteriology results are reported in this paper. Investigations that were not performed uniformly in all sites are reported in the site-specific papers in this supplement.10-13 These include all virology, urine culture, urine antibacterial activity and antimicrobial sensitivity results of isolates. METHODS The four sites, Ethiopia, the Gambia, Papua New Guinea and The Philippines, were chosen to represent a range of developing country settings with high neonatal mortality rates.1 The age criterion for infants to be enrolled in the study was set as 90 days or less. In industrialized countries the transitional age when traditional neonatal pathogens give way to infant pathogens is thought to be at ∼4 to 6 weeks. Because there are no data on which to base such an estimate for developing countries, a broader age range was chosen which includes the period of highest mortality. The methods are described in detail in an accompanying paper.14 Briefly infants younger than 90 days who presented ill to one of the study institutions underwent a formalized triage process. Those with any symptom or sign of infection who did not meet the exclusion criteria14 were enrolled and subjected to a standardized history and examination by a physician (all sites) or pediatric nurse (Papua New Guinea only). Those meeting predefined criteria suggestive of infection were investigated by blood culture, urine culture, hematologic examination, blood film for malaria, nasopharyngeal aspirate (where virology was performed) and lumbar puncture (where clinically indicated). Before the study laboratory procedures were optimized and standardized between sites. These are described in detail in an accompanying paper.14 Blood cultures were considered positive if a definite pathogen (e.g. S. pneumoniae, H. influenzae, Streptococcus pyogenes, S. agalactiae, Salmonella spp.) was isolated from either bottle. For organisms that could be either pathogens or contaminants (e.g. E. coli, Klebsiella pneumoniae, Staphylococcus aureus, Enterococcus faecalis (group D) a blood culture was considered positive if the organism grew from both (or one of one) bottles within 48 h of inoculation. Cultures that did not fulfil these criteria, or cultures of known contaminants such as coagulase-negative staphylococci, Micrococcus spp. and Bacillus spp. were regarded as negative. This strategy was supported by an analysis at the end of the study showing the mortality in those with contaminants to be the same as that in the culture-negative group. Nasopharyngeal swabs were cultured for Bordetella pertussis. Ethical approval for the study was obtained from the WHO Secretariat Committee for Research in Human Subjects and from the relevant local ethical committee in each of the sites involved. Because no investigations were performed that would not have been included in the good clinical care of the infants with suspected infection, only verbal consent was obtained from parents or care givers before the enrollment of an infant into the study. RESULTS In the 4 sites a total of 8418 young infants were triaged, of whom 4552 satisfied the criteria for enrollment in the study and underwent a full history, physical examination and pulse oximetry. Of these, 2398 with clinical signs suggestive of infection were investigated, including blood culture (2452), lumbar puncture (507) and chest radiograph (1868), and this comprised the study sample for the analyses presented in this paper. Included in the above figures are 175 infants without signs of serious infection who were randomly chosen for investigation including blood culture. Not all investigated infants had blood cultures performed, usually because of technical difficulties. Overall 247 infants (5% of those enrolled) died, including 19 who were judged likely to have died because of being taken from the hospital by their families, against medical advice and in a serious condition. There were some important differences between the sites with respect to the study populations. The largest number of infants was enrolled in Papua New Guinea, but in that site the infants enrolled were less ill, fewer were admitted and few died (Table 1).14TABLE 1: Characteristics of infants enrolled and investigated at each of the four study sites Figures in Columns 5 to 9 refer to children with clinical signs who were investigated. Table 2 summarizes the 167 blood culture isolates that were obtained from 2452 infants who underwent this investigation. Of the isolates 102 (61%) were Gram-positive organisms, 96 of which were either S. pneumoniae, Staphylococcus aureus or S. pyogenes. The most frequent Gram-negative isolates were E. coli and Salmonella spp. Three infants grew both Staphylococcus aureus and S. pyogenes in the blood. One infant each grew: Staphylococcus aureus in the blood and S. pyogenes in the CSF; Staphylococcus aureus and Streptococcus spp. (group G) in the blood; Staphylococcus aureus and Enterococcus faecalis in the blood. One infant grew H. influenzae in the blood and had a positive nasopharyngeal culture for B. pertussis. In addition 3 other infants (2 in Ethiopia and 1 in PNG) grew B. pertussis from NP culture.TABLE 2: Bacterial isolates from blood cultures taken from infants investigated, by age During the first month of life Staphylococcus aureus (19, 23%), S. pyogenes (17 isolates, 20%) and E. coli (15, 18%) were the most common organisms isolated from blood. S. pneumoniae was the most common organism in both the second and third month of life, accounting for 30% of blood isolates, with meningitis accounting for 47% (17 of 36) of all episodes of invasive pneumococcal disease in study infants. Four of the positive blood cultures occurred amongst the 175 infants in the systematic sample of infants without signs of systemic infection. Three of the 4 isolates (all Staphylococcus aureus) were from The Gambia where an epidemic of scabies may have contributed to the high rate of isolation of this organism. Table 3 lists the CSF isolates by age. Of the children with a positive CSF culture 28 (68%) also had a positive blood culture. These are shown in parentheses with the organisms listed in Table 3. In children younger than 7 days of age Gram-negative enteric organisms were the main causes of meningitis. During the remainder of the first month Gram-negative organisms continued to be important, but the pneumococcus was equally important. Thereafter during the second and third months of life the pneumococcus was the main cause of meningitis, accounting for 12 of 24 (50%) of cases.TABLE 3: Bacterial isolated from cerebrospinal fluid samples from infants in the study Although the general patterns of organisms isolated were similar, there were some differences between the sites. Seventeen of 34 (50%) isolates of Staphylococcus aureus came from The Gambia, 10 of the 19 (52%) isolates of E. coli came from Ethiopia, and 13 of 29 (45%) S. pyogenes isolates came from Papua New Guinea. Table 4 summarizes the clinical characteristics of infants with blood cultures that grew S. pneumoniae, S. pyogenes, Salmonella spp., Staphylococcus aureus, E. coli and H. influenzae compared with the clinical characteristics of all infants with positive blood cultures and all infants who were investigated but had negative blood cultures. The infants with S. pneumoniae infection were more likely to have fever and fast breathing. Most of the infants with S. pneumoniae infection who died had meningitis; the case fatality rate among infants with proven pneumococcal meningitis was 53%. The infants with S. pyogenes infection were also likely to be febrile with fast breathing; this group had a low case fatality rate, not significantly different to the infants who were culture-negative. Infants with Salmonella spp. infection were significantly more likely to be underweight. Infants with E. coli infection were likely to be young and underweight; the mortality in this group was high (53%).TABLE 4: Clinical characteristics of infants of children with positive blood cultures caused by the most frequently isolated pathogens compared with all children with positive and negative blood cultures Of the blood and CSF isolates of S. pneumoniae, 34 were serotyped at a reference laboratory. The serotypes were type 5 (9 cases), type 2 (9 cases), type 7F (2 cases) and 1 each of types 1, 9L, 9V, 12F, 18F, 19F, 19A, 23F, 27 and 33F. Four isolates were not factor-typed: one each of serogroups 6, 7, 10 and 12. Only 20 of 34 (59%) isolates of S. pneumoniae were of serogroups included in the current 11-valent pneumococcal conjugate vaccine. Of the 9 infants with type 2 infection 8 had meningitis. There were no clear differences between the sites with respect to serotype distribution. In particular type 2 isolates were obtained from all 4 sites. DISCUSSION At the time when this study was carried out published studies from developing countries suggested that Klebsiella spp. and Staphylococcus aureus were the most important neonatal pathogens in developing countries.2, 3 The main causes of serious infections in young infants in the four sites of this study were the classical Gram-positive primary pathogens, Staphylococcus aureus, S. pneumoniae and S. pyogenes. E. coli was the most important Gram-negative organism, followed by a wide range of enteric pathogens causing disease. Although this study confirmed the importance of Staphylococcus aureus, that organism was responsible for only 20% of blood isolates and one CSF isolate, whereas Klebsiella spp. was responsible for only 3% of blood isolates and no CSF isolates. S. pneumoniae was a major pathogen in all age groups studied, particularly after the first week of life. It was also the most important cause of meningitis; in children older than 1 week of age it accounted for 50% of all cases. Surprisingly the outcome of pneumococcal meningitis in this group was not worse than that found with pneumococcal meningitis in older children in developing countries, with about one-half of all cases dying.15 Among the serotypes of S. pneumoniae found, the importance of type 2, which represented 26% of isolates serotyped, was a surprising finding. It is not included in any of the current generation of pneumococcal conjugate vaccines. Type 2 is a serotype rarely reported in previous series; eight of the nine type 2 isolates in our study were from meningitis cases. An earlier study from PNG found type 2 to be responsible for 11% of CSF isolates.16 It appears that this serotype has a predilection for causing meningitis in the very young. This observation is consistent with the view that there are important epidemiologic differences between pneumococcal serotypes, in both age and disease distribution.17 The serotypes usually associated with pediatric carriage and disease and most frequently associated with penicillin resistance (serotypes 6A, 6B, 14, 19F and 23F) were isolated infrequently in this study. The pneumococcus must be considered in any case of serious infection in a young infant in a developing country, particularly if signs of meningitis are present. The findings of this study support the continued investigation of maternal immunization with 23-valent pneumococcal polysaccharide vaccine as a strategy for controlling neonatal infections.7, 8 Earlier this century in industrialized countries, S. pyogenes was an important cause of puerperal sepsis. In the US during the 1930s and 1940s, S. pyogenes was a leading cause of neonatal sepsis and meningitis.18 In our study 10% of the S. pyogenes isolates were detected in the first week of life and thus could have been attributed to maternal infection. 42% of umbilical swabs in Papua New Guinea showed S. pyogenes carriage.12 Recent reports from industrialized countries suggest that there has been a global resurgence of S. pyogenes disease.19 It is not clear whether this is truly global, or whether S. pyogenes has always been an important neonatal pathogen in developing countries, and its resurgence is limited to industrialized countries. Our study does not help to clarify this. The virtual absence of S. agalactiae (group B Streptococcus), which is the most important neonatal pathogen in industrialized countries, was striking. Another study from The Gambia found that rectovaginal carriage rates for S. agalactiae in Gambian women in labor were similar to those reported in the US, and significant infant carriage was demonstrated in the same study.20 Studies from other developing countries, particularly southern Africa, have shown that in those settings S. agalactiae is an important neonatal pathogen, mainly causing early onset neonatal disease (K Klugman, personal communication).21, 22 The absence of S. agalactiae in this study was not the result of small numbers of infants being seen during the first week of life given that 360 (8% of all infants enrolled and evaluated) were seen in the first week, compared with 505, 419, 389 and 384 in the 2nd, 3rd, 4th and 5th weeks, respectively. However, serious illness is much more common during the first week of life,1 so we can assume that many early sepsis cases did not reach health facilities. The importance of Salmonella spp. as a pathogen in this age group is an important finding of this study. The association of this pathogen with malnutrition is striking. This pathogen must be considered in any sick young infant in a developing country, especially if malnutrition is also present. Most of the isolates of Staphylococcus aureus were from the Gambia (69%) which was undergoing an epidemic of scabies at the time of the study. Scabies in very young infants typically presents with pustular eruptions on the palms and soles which are almost always secondarily infected. This may have been the source of many of the staphylococcal infections in our study. Historically there is great variation in neonatal staphylococcal infection by country and time, although most documented outbreaks have been hospital-associated. Staphylococcal bacteremia in neonates correlates with skin disease and tends to be self-curing. In this study the case-fatality rate in infants who were blood culture-positive for Staphylococcus aureus was 21%. This study supports the view that it is reasonable to include an antistaphylococcal penicillin such as cloxacillin in the initial therapy of neonatal sepsis in infants with pustular skin sores or in communities where many infants contract scabies. The finding of Salmonella spp. in 10% (17 of 167) of positive blood isolates is troubling as infection with this species of bacteria is not well covered with the penicillin-gentamicin combination. Ampicillin, chloramphenicol or third generation cephalosporins would be required to cover Salmonella spp. The case-fatality rate of young infants with clinical signs suggestive of serious bacterial infection and a positive blood culture was 30%, despite hospital admission for standard treatment (usually with benzylpenicillin or ampicillin and gentamicin) in well-functioning hospitals. The number of deaths despite hospital care points to the importance of prompt care-seeking, early detection of cases and effective treatment and high-lights the overwhelming importance of prevention. Overall clinical symptoms were not particularly helpful in distinguishing among infections caused by different organisms. Microbiology data are rarely available to guide patient care in developing country hospitals. Therefore empiric treatment should be adequately effective against the likely pathogens. The studies described in this paper represent the largest prospective study of early infant infections in developing countries. The findings suggest that in developing countries initial therapy of neonatal sepsis with ampicillin and gentamicin is suitable, covering most of the likely Gram-positive and Gram-negative organisms. The main problem with this regimen is the poor coverage of Staphylococcus aureus, which is now usually resistant to penicillin. The growing problem of penicillin-resistant S. pneumoniae makes this combination unsuitable for the empiric therapy of neonatal meningitis in many areas, which should be treated initially with a third generation cephalosporin such as cefotaxime once the diagnosis has been confirmed by lumbar puncture. This would also be a suitable first line antibiotic in this age group where it is available, although poor antistaphylococcal activity is of concern. If skin sepsis suggests likely staphylococcal disease, initial therapy should include an antistaphylococcal agent such as cloxacillin, but if this is used to replace ampicillin, the reduced efficacy of the combination against Salmonella spp. and S. pneumoniae should be kept in mind. If a child has not improved within 48 h, or if deterioration is evident within that time, treatment should be changed to second line antibiotic therapy, although cost will limit this in many developing country settings. Where possible this should be guided by bacteriology. However, if this is not available the addition of an antistaphylococcal agent such as cloxacillin to ampicillin and gentamicin would constitute a logical second line. Where the initial regimen included an antistaphylococcal agent, the second line therapy should strive to improve the Gram-negative cover by the use of cefotaxime if available, or chloramphenicol. The use of chloramphenicol in young infants has been associated in the past with serious adverse effects caused by the high doses used. If appropriate doses and dose intervals are used and they are calculated and administered with great care, chloramphenicol succinate can be safely administered intramuscularly to infants in this age group. Better treatment of seriously ill young infants in developing countries is urgently needed. In most cases this can still be achieved with relatively inexpensive antibiotics. However, this study has shown that, even with adequate treatment, there is a high mortality. Prevention must be central to efforts to control this problem. ACKNOWLEDGMENTS The investigators thank the medical and nursing staff of the participating hospitals, the bacteriology laboratory staff at each of the study centers and most importantly the families of young infants who participated in the study.

Duplicate vaginal swabs were collected from 100 women, and comparisons were made between an in-house broth-agar culture system and a commercially available kit, the Mycoplasma IST kit (bioMérieux), for the detection of Mycoplasma hominis and Ureaplasma urealyticum. There was good agreement between the two systems for detection of the genital mycoplasmas in terms of sensitivity, with values of > 92% being obtained. In terms of specificity, the mutual comparisons were less favorable, though specificity values of > 72% were obtained. Statistically there was no significant difference in the performance of the two tests (P < 0.1 for both M. hominis and U. urealyticum). While the broth-agar culture system was considerably less expensive than the kit, the Mycoplasma IST kit provided additional information on antibiotic susceptibilities and had the advantages of a shelf life of up to 12 months and not requiring the preparation of culture media. The prevalences of colonization obtained for M. hominis and U. urealyticum were extremely high in this randomly selected group of women from periurban and rural settlements in the Eastern Highlands of Papua New Guinea, being > or = 70% for M. hominis and > or = 78% for U. urealyticum. colonization with both genital mycoplasmas simultaneously was also very common, with > or = 60% of women being colonized by both M. hominis and U. urealyticum.

Acute lower respiratory tract infection (ALRI) is the major cause of death among children in Papua New Guinea. This longitudinal study reports the bacteriologic findings for children observed in their hamlets. A total of 1,449 nasal swab specimens from 158 children less than 5 years of age who were studied intensively for 18 months were examined. Non-serotypable strains of Haemophilus influenzae were isolated from 91% of specimens, and serotypable strains were isolated from 35% (8% H. influenzae type b) of specimens. All children had acquired Streptococcus pneumoniae by the age of 3 months. The most frequently occurring serotypes of S. pneumoniae were 6, 19, and 23. Children more frequently carried invasive pneumococci during an episode of ALRI than when they were healthy. Also, children more frequently carried serotypable strains of H. influenzae during the 2 weeks preceding an episode of ALRI than when they were healthy. Between-children analyses showed that children who were susceptible to attacks of ALRI and those who were not susceptible had similar rates of carriage of bacteria.