Articles. Funding Bill & Melinda Gates Foundation grant to WHO.

Transcription

1 Simplified antibiotic regimens compared with injectable procaine benzylpenicillin plus gentamicin for treatment of neonates and young infants with clinical signs of possible serious bacterial infection when referral is not possible: a randomised, open-label, equivalence trial African Neonatal Sepsis Trial (AFRINEST) group, Antoinette Tshefu, Adrien Lokangaka, Serge Ngaima, Cyril Engmann, Fabian Esamai, Peter Gisore, Adejumoke Idowu Ayede, Adegoke Gbadegesin Falade, Ebunoluwa A Adejuyigbe, Chineme Henry Anyabolu, Robinson D Wammanda, Clara L Ejembi, William N Ogala, Lu Gram, Simon Cousens Summary Background WHO recommends hospital-based treatment for young infants aged 0 59 days with clinical signs of possible serious bacterial infection, but most families in resource-poor settings cannot accept referral. We aimed to assess whether use of simplified antibiotic regimens to treat young infants with clinical signs of severe infection was as efficacious as an injectable procaine benzylpenicillin gentamicin combination for 7 days for situations in which hospital referral was not possible. Methods In a multisite open-label equivalence trial in DR Congo, Kenya, and Nigeria, community health workers visited all newborn babies at home, identifying and referring unwell young infants to a study nurse. We stratified young infants with clinical signs of severe infection whose parents did not accept referral to hospital by age (0 6 days and 7 59 days), and randomly assigned each individual within these strata to receive one of the four treatment regimens. Randomisation was stratified by age group of infants. An age-stratified randomisation scheme with block size of eight was computer-generated off-site at WHO. The outcome assessor was masked. We randomly allocated infants to receive injectable procaine benzylpenicillin (group A, reference group); injectable gentamicin and oral amoxicillin for 7 days (group B); injectable procaine benzylpenicillin gentamicin for 2 days, then oral amoxicillin for 5 days (group C); or injectable gentamicin for 2 days and oral amoxicillin for 7 days (group D). Trained health professionals gave daily injections and the first dose of oral amoxicillin. Our primary outcome was treatment failure by day 8 after enrolment, defined as clinical deterioration, development of a serious adverse event (including death), no improvement by day 4, or not cured by day 8. Independent outcome assessors, who did not know the infant s treatment regimen, assessed study outcomes on days 4, 8, 11, and 15. Primary analysis was per protocol. We used a prespecified similarity margin of 5% to assess equivalence between regimens. This study is registered with the Australian New Zealand Clinical Trials Registry, number ACTRN Findings In Kenya and Nigeria, we started enrolment on April 4, 2011, and we enrolled the necessary number of young infants aged 7 days or older from Oct 17, 2011, to April 30, At these sites, we continued to enrol infants younger than 7 days until March 29, In DR Congo, we started enrolment on Sept 17, 2012, and continued until June 28, We randomly assigned 3564 young infants to either group A (n=894), group B (n=884), group C (n=896), or group D (n=890). We excluded 200 randomly assigned infants, who did not fulfil the predefined criteria of adherence to treatment and adequate follow-up. In the per-protocol analysis, 828 infants were included in group A, 826 in group B, 862 in group C, and 848 in group D. 67 (8%) infants failed treatment in group A compared with 51 (6%) infants in group B (risk difference 1 9%, 95% CI 4 4 to 0 1), 65 (8%) in group C ( 0 6%, 3 1 to 2 0), and 46 (5%) in group D ( 2 7%, 5 1 to 0 3). Treatment failure in groups B, C, and D was within the similarity margin compared with group A. During the 15 days after random allocation, 12 (1%) infants died in group A, compared with ten (1%) infants in group B, 20 (2%) infants in group C, and 11 (1%) infants in group D. An infant in group A had a serious adverse event other than death (injection abscess). Interpretation The three simplified regimens were as effective as injectable procaine benzylpenicillin gentamicin for 7 days on an outpatient basis in young infants with clinical signs of severe infection, without signs of critical illness, and whose caregivers did not accept referral for hospital admission. Funding Bill & Melinda Gates Foundation grant to WHO. Published Online April 2, S (14) See Online/Comment S (15) See Online/Articles S (14) Department of Community Health, Kinshasa School of Public Health, Kinshasa, DR Congo (A Tshefu PhD, A Lokangaka MD, S Ngaima MD); Departments of Pediatrics and Maternal Child Health, Schools of Medicine and Public Health, University of North Carolina, Chapel Hill, NC, USA (C Engmann MB BS); Department of Child Health and Paediatrics, School of Medicine, Moi University, Eldoret, Kenya (Prof F Esamai PhD, P Gisore MMed [Ped]); Department of Paediatrics, College of Medicine, University of Ibadan, and University College Hospital, Ibadan, Nigeria (A I Ayede MB BS, A G Falade MD); Department of Paediatrics and Child Health, Obafemi Awolowo University, Ile-Ife, Nigeria (Prof E A Adejuyigbe FMCPaed, C H Anyabolu FWACP); Department of Paediatrics, Ahmadu Bello University Teaching Hospital, Ahmadu Bello University, Zaria, Nigeria (Prof R D Wammanda FWACP, Prof W N Ogala FMCP); Department of Community Medicine, Ahmadu Bello University Teaching Hospital, Ahmadu Bello University, Zaria, Nigeria (C L Ejembi FNMCP); Department of Infectious Disease Epidemiology, Faculty Published online April 2,

2 of Epidemiology and Population Health, London School of Hygiene & Tropical Medicine, London, UK (L Gram MSc, S Cousens DipMathStat) Correspondence to: Prof Fabian Esamai, Department of Child Health and Paediatrics, School of Medicine, College of Health Sciences, Moi University, Eldoret 30100, Kenya fesamai2007@gmail.com Introduction Neonatal infections, including pneumonia, sepsis, and meningitis, cause about 0 6 million deaths in neonates every year. 1 Clinical differentiation of these conditions is difficult. WHO s Integrated Management of Childhood Illness (IMCI) guidelines classify neonates and young infants aged 0 59 days who are not feeding well and have fever, low body temperature, fast breathing, severe chest indrawing, convulsions, and no spontaneous movements as having possible serious bacterial infection. 2 In 2012, an estimated 6 9 million possible serious bacterial infections occurred in neonates worldwide. 3 WHO recommends that young infants with possible serious bacterial infection be admitted to hospital and treated with a combination of injectable gentamicin and benzylpenicillin or ampicillin for at least 7 10 days. 4 The effectiveness of this treatment in reduction of risk of death is well accepted among health practitioners. However, inpatient care is not feasible or acceptable for many families in resource-poor settings because of financial, cultural, or transportation constraints. 5 7 Research in Bangladesh 6 and India 8 has shown that neonates with signs of severe infection can be treated with injectable gentamicin in combination with oral co-trimoxazole or injectable procaine benzyl penicillin in the community when hospital referral was not accepted by families. The results of a subsequent study from Pakistan 7 showed that the efficacy of a procaine benzyl penicillin gentamicin combination was much higher than that of co-trimoxazole gentamicin. 7 No data are available from Africa. Although the studies in Asia have provided proof of principle, the proposed treatment is difficult to deliver in programme settings. Simplified regimens that combine oral and intramuscular anti biotics or include a switch from injectable to oral antibiotics after the first 2 3 days of treatment might be more straightforward than standard regimens to deliver at the outpatient level in resource-poor health systems. Simplified regimens would increase access to treatment for many ill young infants who are currently unable to receive it, and could contribute to reductions of neonatal mortality. Thus research is needed to inform guidelines on the management of ill neonates and young infants for situations in which referral is not possible, both in Africa and worldwide. We report a trial done in three African countries to compare the efficacy of three simplified antibiotic regimens with that of the injectable procaine benzylpenicillin gentamicin com bination for treatment of young infants with clinical signs of severe infection (clinical severe infection), but without signs of critical illness (convulsions, unconsciousness, or inability to feed at all) or with only rapid breathing, whose parents did not accept referral to hospital. Methods Study design This study is an individually randomised, multicentre, open-label equivalence trial, in which all sites followed the same protocol and contributed to the results. The methods of this study have been described in detail previously. 9,10 Briefly, the study was done at five sites, one each in DR Congo and Kenya, and three in Nigeria (Ibadan, Ile-Ife, and Zaria). In DR Congo, we did the study in rural areas of North and South Ubangi districts, in the province of Equateur. The population of the study area was about people. The study included 30 health areas, each with a health centre. Health centres are primary level facilities that provide curative and health promotion services. Each health centre was staffed by one trained nurse, who provided treatment to ill young infants and oversaw the activities of community health workers, who mainly engage in preventive and promotional activities. In Kenya, we did the study in rural areas in the counties of Busia, Bungoma, and Kakamega in western Kenya. The population of the study area was about people. We selected 34 clusters for the study; each cluster had one dispensary or health facility staffed by two health workers who provided treatment to ill young infants and supervised the activities of community health workers. In Nigeria, we did the study at three sites in Ile-Ife, Ibadan, and Zaria. In all three sites, community health extension workers did community activities and provided treatment to ill young infants. The Ile-Ife centre was in the Ife Central and East Local Government Areas (Ile-Ife community), a semi-urban community in Osun State, in southwestern Nigeria (estimated population people). The Ibadan study centre was in Ido and Lagelu Local Government Areas of Oyo State in the southwestern part of Nigeria. The population was mainly rural with some peri-urban areas, and the study population under surveillance was about The Zaria study centre was in the Zaria Local Government Area, located in the northern part of Kaduna State in the North West geo political zone of Nigeria. It is composed of more than 100 urban, peri-urban, and rural settlements (estimated population ). The trial protocol was approved by the local institutional review boards at each site, the WHO Ethical Review Committee, and the ethics committee of the London School of Hygiene & Tropical Medicine. The trial followed the Council for International Organizations of Medical Sciences and Good Clinical Practice guidelines. We assessed infants for this study only if their families refused to accept hospital referral, which is the current standard of care. We obtained written and witnessed informed consent before enrolling eligible infants whose families refused referral. An independent data safety monitoring board monitored the study. We reported all serious adverse events, including death, inability to pass urine for more than 12 h, anaphylactic reaction, severe dehydration caused by diarrhoea, or disseminated or severe rash to WHO within 48 h of their occurrence. We also gave this information to the data safety monitoring board, 2 Published online April 2,

3 institutional review boards, and WHO Ethical Review Committee once per year. Enrolled infants whose condition worsened were provided rescue therapy (intramuscular ceftriaxone for 7 days) if the families still refused hospital referral. Participants In Kenya and DR Congo, community health workers identified births in the study areas, within 24 h of delivery if possible. In Nigeria, community health extension workers and traditional birth attendants identified births in the same timeframe. Community health extension workers in Nigeria receive 2 years of professional training and are permitted to give injections, whereas community health workers in Kenya and DR Congo have only a few months of training to provide disease prevention and health promotion advice to families. The community health workers or community health extension workers collected the birth and household data and did postnatal home visits on the day of birth (day 1) and days 3, 7, 14, 21, 28, 35, 42, 49, and 60 to promote health of mothers and babies. If the community health workers in DR Congo and Kenya identified a young infant with a sign of infection, they accompanied the family to the health centre where the study nurse was located, if the family agreed. If the community health extension workers in Nigeria identified a young infant with signs of infection, they called the study enrolment nurse to the infant s home. The enrolment nurses screened young infants at the health centre or dispensary in DR Congo and Kenya, and at home in Nigeria. Nurses advised the parents of young infants with clinical signs of severe infection to take the infant to hospital. If the parents did not accept referral, nurses assessed the infant for inclusion and exclusion criteria. The inclusion criteria were: age 0 59 days, any sign of clinical severe infection (stopped feeding well defined as poor feeding on observation, movement only when stimulated, severe chest indrawing, and axillary temperature 38 0 C or <35 5 C), parents did not accept or could not access referral level care, parents gave consent to participate in the study. The exclusion criteria were: critically ill (characterised by the presence of any of the following signs: unconsciousness, convulsions, unable to feed at all, apnoea, unable to cry, cyanosis, dehydration, bulging fontanelle, major congenital malformations inhibiting oral antibiotic intake, active bleeding requiring transfusion, surgical conditions needing hospital referral, and persistent vomiting defined as vomiting after three attempts to feed the baby within 30 min), very low weight (<1500 g at the time of presentation), and hospital admission for illness in the past 2 weeks or previously enrolled in the study. Nurses explained the study to the parents of the eligible infants. We then sought written informed consent for study participation from the parents. Randomisation and masking At enrolment, we stratified young infants aged 0 59 days with clinical signs of severe infection by age (0 6 days and 7 59 days) and we individually randomly assigned them within these strata to receive one of the four treatment regimens. We aimed to enrol a third of the overall sample in the group aged 0 6 days to contribute to pooled analysis with similar studies in Asia. An off-site person at WHO, who was not associated with the study, prepared randomisation lists. They generated randomisation lists for each site, for each of the two age strata, in a 1:1 ratio in blocks of eight using Stata 10. To conceal allocation, we sealed treatment allocation codes on a folded piece of card Panel 1: Intervention and comparison treatment regimens Reference treatment regimen Group A: injectable gentamicin once per day and injectable procaine benzylpenicillin once per day for 7 days (14 injections in total). Intervention treatments regimens Group B: injectable gentamicin once per day and oral amoxicillin twice per day for 7 days (seven injections in total). Group C: injectable gentamicin once per day and injectable procaine benzylpenicillin once per day for 2 days, thereafter oral amoxicillin for 5 days (four injections in total). Group D: injectable gentamicin once per day and oral amoxicillin twice per day for 2 days, thereafter oral amoxicillin twice per day for 5 days (two injections in total). Antibiotic dosages Injection procaine benzylpenicillin in a dose of units/kg once per day intramuscularly. One million units vial was reconstituted in 5 ml sterile water. Injection gentamicin in the range 4 0 mg/kg once per day IM in the first week of life and 7 5 mg/kg once daily intramuscularly thereafter. A 2 ml with 40 mg per ml preparation was used. Oral amoxicillin in suspension in a dose of 100 mg/kg per day (less than 2 kg are given 75 mg/kg per day), divided into 2 equal doses. Powder form in a bottle was reconstituted with sterile water to make 100 ml, providing 250 mg per 5 ml. Panel 2: Per-protocol analysis criteria Requirements to be included in the per-protocol analysis were that the young infants should have received adequate outcome assessment and were fully or partly adherent according to the following definitions. Adequacy of independent outcome assessment Independent outcome assessments were completed on days 4, 8, 11, and 15. Independent outcome assessments were completed on days 4 and on any of the days 8, 11, or 15. Treatment adherence and clinical follow-up Fully adherent: received 100% of doses of scheduled antibiotics on all 7 days or by the time of treatment failure if treatment failure occurred. Partly adherent: received 100% of scheduled antibiotics on the first 3 days of therapy or by the time of treatment failure, and at least 50% of all scheduled doses of each antibiotic on days 4 7 or by the time of treatment failure; and has not received any non-study injectable antibiotic before day 8 assessment (unless given because of treatment failure), or any non-study oral antibiotic on days Published online April 2,

4 pregnancies identified * infants under follow-up in two sets of sequentially numbered opaque coloured envelopes, one colour for each age stratum. Outcome assessment nurses were unaware of the infant s treatment allocation. The treatment allocation code remained concealed until after informed consent was obtained and the young infant enrolled in the study. In Kenya and Nigeria, the enrolment nurse obtained the treatment allocation by phoning a centrally based person holding the randomi sation envelopes, with the allocation confirmed by text message. In DR Congo, mobile phone connections identified as showing a sign of illness 6968 infants referred by community health workers and seen by study nurse came directly to study nurse 894 allocated treatment A 7129 cases of possible serious bacterial infection identified 4665 referred by community health workers and seen by study nurse 2464 taken directly to study nurse 6557 screened by study nurse for inclusion in the study 4301 referred by community health workers and seen by study nurse 2256 taken directly to study nurse 3600 infant with clinical severe infection 3564 enrolled and randomised 66 protocol violations 828 included in the analysis Figure: Trial profile *Includes ten triplets and 1235 twins. 884 allocated treatment B 58 protocol violations 826 included in the analysis 1935 stillbirths 7554 moved away for delivery 196 died before follow-up by community health workers 1277 enrolment ended before delivery 2117 accepted referral 2068 study nurse not seen for other reason mild or no illness 166 poor feeding since birth, underweight, eye infection, severe dehydration or severe jaundice, but no possible serious bacterial infection 572 accepted referral 2785 fast breathing only 172 critically ill 21 refused enrolment 11 infants taken to hospital 4 admission to hospital in last 2 weeks 896 allocated treatment C 34 protocol violations 862 included in the analysis 890 allocated treatment D 42 protocol violations 848 included in the analysis were unreliable, so we provided each health facility with blocks of eight envelopes for each age group, which were replenished regularly. Procedures The reference treatment regimen was injectable gentamicin and procaine benzylpenicillin for 7 days (14 injections, group A), which was compared with three simplified antibiotic regimens: injectable gentamicin and oral amoxicillin treatment for 7 days (seven injections, group B); injectable procaine benzylpenicillin gentamicin for 2 days, then oral amoxicillin for 5 days (four injections, group C); and injectable gentamicin once per day for 2 days and oral amoxicillin for 7 days (two injections, group D; panel 1). The rationale for the choice of antibiotic regimens has been described previously. 11 Panel 1 shows the target doses of the different antibiotics. We provided all treatments for 7 days. A treatment nurse at an outpatient health facility near the infant s home gave injections in DR Congo and Kenya, and a community health extension worker gave injections at the homes of enrolled infants in Nigeria. The treatment nurse gave the first dose every day when oral antibiotic was scheduled. The mother gave the second dose of oral antibiotic per day, every day. The treatment nurse or community health extension worker examined each baby for clinical improvement or deterioration before giving treatment. All study personnel had standardised training. Community health workers or community health extension workers, and their supervisors received training in WHO and UNICEF s Caring for the Newborn at Home course. Study nurses and their supervisors attended a Young Infant IMCI course. All study personnel received training on study-specific pro cedures as appropriate. We did exercises to standardise assessment of clinical signs and manage ment for com munity health workers, community health extension workers, nurses, and supervisors. We did supervisory and monitoring visits regularly. We also did data-based monitoring. Refresher training sessions were held every 3 6 months. Details have been reported previously. 11 An independent outcome assessment nurse visited enrolled infants at home to assess study outcomes on days 4, 8, 11, and 15 after the day of enrolment, or on any day if the treatment nurse or community health extension worker judged that the infant s clinical condition had deteriorated. The primary outcome was treatment failure. Outcomes The primary outcome was treatment failure by the day 8 post-enrolment visit. Treatment failure was defined as any one of: death, clinical deterioration (admission to hospital, emergence of any sign of critical illness at any time or a new sign of clinical severe infection after the day 2 visit, or re-emergence of a sign of clinical severe 4 Published online April 2,

5 infection on day 4 after initial disappearance), no improvement in clinical condition by day 4 (if a single sign of clinical severe infection existed at enrolment, persistence of the sign; if multiple signs existed at enrolment, persistence of more than one sign), infant not cured by day 8 (persistence of any clinical sign of severe infection on day 8 of enrolment), or development of a serious adverse event other than death that was thought to be related to the study antibiotics (eg, organ failure, anaphylactic reaction, severe diarrhoea, or disseminated and severe rash). The secondary outcomes were death between days 9 and 15 after enrolment, Procaine benzylpenicillin and (group A, n=894) Oral amoxicillin and intramuscular (group B, n=884) Procaine benzylpenicillin and gentamicin for 2 days for 5 days (group C, n=896) Gentamicin and oral amoxicillin for 2 days for 5 days (group D, n=890) Age (days) Mean (SD) 18 3 (16 3) 17 9 (15 4) 17 6 (15 6) 18 0 (16 1) <7 293 (33%) 279 (32%) 300 (33%) 288 (32%) (67%) 605 (68%) 596 (67%) 602 (68%) Sex Male 486 (54%) 463 (52%) 454 (51%) 498 (56%) Female 408 (46%) 421 (48%) 442 (49%) 392 (44%) Weight for age Z score Mean (SD) 0 9 (1 3) 0 9 (1 3) 0 9 (1 3) 0 9 (1 3) < (18%) 137 (15%) 160 (18%) 154 (17%) (82%) 747 (85%) 736 (82%) 736 (83%) Respiratory rate (breaths per min) Mean (SD) 64 1 (17 1) 64 7 (17 0) 65 3 (17 5) 64 4 (17 8) < (41%) 346 (39%) 342 (38%) 345 (39%) (24%) 223 (25%) 201 (22%) 230 (26%) (18%) 156 (17%) 182 (20%) 156 (18%) (9%) 89 (10%) 99 (11%) 86 (10%) (4%) 44 (5%) 46 (5%) 43 (5%) (3%) 26 (3%) 26 (3%) 30 (3%) Temperature ( C) Mean (SD) 37 4 (1 0) 37 4 (1 1) 37 3 (1 1) 37 4 (1 1) < (4%) 44 (5%) 58 (6%) 49 (6%) (48%) 429 (49%) 443 (49%) 424 (48%) (45%) 380 (43%) 370 (41%) 379 (43%) (2%) 31 (4%) 25 (3%) 38 (4%) Stopped feeding well 142 (16%) 130 (15%) 147 (16%) 159 (18%) Movement only on stimulation 21 (2%) 33 (4%) 22 (2%) 23 (3%) Severe chest indrawing 392 (44%) 384 (43%) 393 (44%) 384 (43%) Number of presenting signs (87%) 775 (8%) 789 (88%) 766 (86%) (11%) 101 (11%) 95 (11%) 106 (12%) 3 11 (1%) 7 (1%) 12 (1%) 18 (2%) 4 1 (<1%) 1 (<1%) 0 (0%) 0 (0%) Maternal age (years) Mean 26 3 (6 0) 26 3 (5 %) 25 9 (5 8) 26 1 (5 8) < (12%) 101 (11%) 115 (13%) 117 (13%) (85%) 753 (85%) 760 (85%) 743 (83%) Not known 29 (3%) 30 (3%) 21 (2%) 30 (3%) Place of birth Home 343 (38%) 356 (40%) 363 (41%) 371 (42%) Health facility 431 (48%) 415 (47%) 411 (46%) 420 (47%) Others 117 (13%) 110 (12%) 122 (14%) 98 (11%) Not known 3 (<1%) 3 (<1%) 0 (0%) 1 (<1%) (Table 1 continues on next page) Published online April 2,

6 Procaine benzylpenicillin and (group A, n=894) Oral amoxicillin and intramuscular (group B, n=884) Procaine benzylpenicillin and gentamicin for 2 days for 5 days (group C, n=896) (Continued from previous page) At least one antenatal clinic attendance Yes 850 (95%) 834 (94%) 845 (94%) 842 (95%) No 41 (5%) 46 (5%) 50 (6%) 47 (5%) Not known 3 (<1%) 4 (<1%) 1 (<1%) 1 (<1%) Number of livebirths (24%) 213 (24%) 214 (24%) 208 (23%) (40%) 329 (37%) 344 (38%) 334 (38%) (36%) 339 (38%) 337 (38%) 347 (39%) Not known 3 (<1%) 3 (<1%) 1 (<1%) 1 (<1%) Maternal education (years) Mean (SD) 7 2 (4 5) 7 1 (4 5) 7 3 (4 5) 7 3 (4 4) No formal school attendance 155 (17%) 161 (18%) 158 (18%) 151 (17%) < (56%) 506 (57%) 511 (57%) 504 (57%) (26%) 212 (24%) 222 (25%) 232 (26%) Not known 5 (1%) 5 (1%) 5 (0 6%) 3 (<1%) Cooking place and fuel Indoor with solid fuel 392 (44%) 381 (43%) 381 (43%) 380 (43%) Outdoor with solid fuel 193 (22%) 186 (21%) 200 (22%) 192 (22%) No solid fuel 309 (35%) 317 (36%) 315 (35%) 318 (36%) Data are mean (SD) or n (%). Table 1: Baseline characteristics Gentamicin and oral amoxicillin for 2 days for 5 days (group D, n=890) relapse (after disappearance of all signs of clinical severe infection by day 8, emergence of any sign of critical illness, or severe infection between days 9 and 15 after enrolment), and adherence to the allocated treatment between days 1 and 8. We referred all enrolled infants who failed therapy to a hospital again. The outcome assessment nurse offered injectable ceftriaxone rescue therapy for 7 days to those infants whose parents refused referral. Statistical analysis We designed the study to have high power to show that the experimental treatments were as efficacious as the reference treatment, if the true treatment failure rates in the experimental and reference arms were the same. We anticipated a failure rate of 10% in the reference group and set the equivalence margin at 5% ie, our criterion to infer that an experimental treatment was as efficacious as the reference treatment was that the upper 95% confidence limit for the risk difference (experimental minus reference) should be less than 5%. We selected a sample size of 900 per group to provide 90% power. We assumed that roughly 85% of infants would receive adequate treatment and assessment, which would give 760 young infants that could be analysed in each treatment group for the primary per-protocol analysis. We used standard case report forms to obtain the data at all sites. We double-entered data into an SQL database at each site. We sent cleaned copies of the database from all the sites to the central data coordination centre at the London School of Hygiene & Tropical Medicine every month for quality checks. We did the analysis using Stata The primary analysis was per protocol (ie, restricted to infants who met predetermined criteria for adequate treatment and follow-up; panel 2), which is a more conservative analysis than an intention-to-treat analysis for equivalence studies. We calculated the difference in the risk of treatment failure between the two groups with a 95% confidence interval. We did secondary analysis to investigate adverse events, including death and other serious outcomes. We also did an intention-to-treat analysis. We did a single, planned interim analysis in a blinded manner and it was reviewed by the data safety monitoring board after roughly 66% of participants were enrolled and treated. The data safety monitoring board recommended continuation of the study. This study is registered with the Australian New Zealand Clinical Trials Registry, number ACTRN Role of the funding source The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication. 6 Published online April 2,

7 Results We enrolled 3564 young infants with clinical signs of severe infection (figure) at the five study sites (574 infants in DR Congo; 1192 infants in Kenya; 632 infants in Ibadan, 647 infants in Ile-Ife, and 519 infants in Zaria). In Kenya and Nigeria, we started enrolment on April 4, 2011, and we enrolled the necessary number of young infants aged 7 days or older from Oct 17, 2011, to April 30, At these sites, we continued to enrol infants younger than 7 days until March 29, In DR Congo, we started enrolment on Sept 17, 2012, and continued until June 28, The figure shows the numbers allocated to each treatment group. The baseline characteristics of the four groups were similar (table 1). A third (1160 [33%] of 3564 infants) of enrolled infants were younger than 7 days. 611 (17%) infants had weight for age Z score less than 2. The most common presenting signs were temperature 38 C Procaine benzylpenicillin and (group A, n=894) Oral amoxicillin and intramuscular gentamicin for 7 days (group B, n=884) Procaine benzylpenicillin and gentamicin for 2 days followed by oral amoxicillin for 5 days (group C, n=896) Received all treatment doses as per protocol 827 (93%) 835 (94%) 868 (97%) 863 (97%) Did not receive all doses, but met per-protocol analysis criteria 10 (1%) 5 (1%) 5 (1%) 2 (<1%) Did not meet per-protocol analysis criteria for treatment* 57 (6%) 44 (5%) 23 (3%) 25 (3%) Follow-up by independent outcome assessor Received all independent outcome assessment visits 811 (91%) 803 (91%) 840 (94%) 826 (93%) Did not receive all independent outcome assessment visits, 36 (4%) 41 (5%) 31 (3%) 37 (4%) but met per-protocol analysis criteria Did not meet per-protocol analysis criteria for assessment 47 (5%) 40 (5%) 25 (3%) 27 (3%) Included in per-protocol analysis (met both treatment and assessment criteria) 828 (93%) 826 (93%) 862 (96%) 848 (95%) Gentamicin and oral amoxicillin for 2 days followed by oral amoxicillin for 5 days (group D, n=890) Data are n (%). *One infant was enrolled in group B, but given treatment according to group A on day 1; one other infant was enrolled in group B, but given treatment according to group D; 13 infants were enrolled with poor feeding since birth despite this not being an inclusion sign. Table 2: Treatment adherence and follow-up of enrolled infants Procaine benzylpenicillin and (group A, n=828) Oral amoxicillin and intramuscular gentamicin for 7 days (group B, n=826) Procaine benzylpenicillin and gentamicin for 2 days followed by oral amoxicillin for 5 days (group C, n=862) Gentamicin and oral amoxicillin for 2 days for 5 days (group D, n=848) First week after enrolment Treatment failure 67 (8%) 51 (6%) 65 (8%) 46 (5%) Risk difference 1 9% ( 4 4 to 0 1) 0 6% ( 3 1 to 2 0) 2 7% ( 5 1 to 0 3) Death 10 (1%) 8 (1%) 17 (2%) 10 (1 %) Appearance of a sign of critical illness 5 (1%) 5 (1%) 8 (1%) 11 (1 %) Appearance of a new sign of serious 7 (1%) 6 (1%) 4 (<1%) 4 (<1%) infection Serious adverse effect other than death 1 (<1%) 0 (0%) 0 (0 %) 0 (0%) Hospital admission 2 (<1%) 2 (<1%) 4 (<1%) 1 (<1%) No improvement in clinical condition 34 (4%) 24 (3%) 24 (3%) 17 (2 %) by day 4 Reappearance of inclusion sign 7 (1%) 5 (1%) 7 (1%) 3 (<1%) between days 5 and 8 Presence of inclusion sign on day 8 1 (<1%) 1 (<1%) 1 (<1%) 0 (0%) Withdrawal * 49/894 (5%) 23/884 (3%) 13/896 (2%) 11/890 (1%) Risk difference 2 9% ( 4 7 to 1 1) 4 1% ( 5 8 to 2 4) 4 2% ( 5 9 to 2 6) Second week after enrolment Death 1 (<1%) 2 (<1%) 0 1 (<1%) Serious adverse effect other than death Relapse 8 (1 %) 7 (1%) 2 (<1%) 7 (1%) First and second week after enrolment Death 12/894 (1%) 10/884 (1%) 20/896 (2%) 11/890 (1%) Data are n (%) or risk difference (95% CI). *Per-protocol withdrawals excluded. Percentage calculated out of number of infants who had not failed treatment. Table 3: Per-protocol analysis Published online April 2,

8 or more (1648 [46%] infants) and lower chest indrawing (1553 [44%] infants). 453 (13%) infants had two or more signs. Most (2029 [57%] of 3564) mothers had less than 12 years of education and 440 (12%) were aged younger than 20 years. Among all enrolled infants, 3364 (94%) fulfilled our treatment adherence and follow-up assessment criteria for inclusion in the per-protocol analysis (table 2). We identified treatment failure in 67 (8%) infants in group A, 51 (6%) infants in group B (risk difference 1 9%, 95% CI 4 4 to 0 1), 65 (8%) infants in group C ( 0 6%, 3 1 to 2 0), and 46 (5%) infants in group D ( 2 7%, 5 1 to 0 3; Table 3). We saw deterioration in clinical condition, death, or appearance of a serious adverse effect other than death in 25 (3%) infants in group A, compared with 21 (3%) infants in group B, 33 (4%) infants in group C, and 26 (3%) infants in group D. More parents withdrew their infants from group A (49 [5%] of 894 infants) than did parents in the other three groups, with 23 (3%) of 884 infants in group B, 13 (1%) of 896 infants in group C, and 11 (1%) of 890 infants in group D being withdrawn. The baseline characteristics of infants who were withdrawn from the study did not differ from those who did not withdraw. During the second week after enrolment, 24 (1%) of 3564 infants relapsed. Tests for interaction with respect to treatment failure did not show any evidence of effect modification by age, sex, weight, or maternal age and education (data not shown). In the intention-to-treat analysis, risk of treatment failure was similar in all groups (table 4). Treatment failure was observed in 72 (8%) of 894 infants in group A, 52 (6%) of 884 infants in group B (risk difference 2 2%, 95% CI 4 5 to 0 2), 67 (7%) of 896 infants in group C ( 0 7%, 3 1 to 1 8), and 48 (5%) of 890 infants in group D ( 2 7%, 5 0 to 0 3). We saw only one serious adverse effect other than death an infant allocated to group A developed an injection abscess. Discussion We identified that, for treatment of neonates and young infants with clinical signs of severe infection, the three simplified antibiotic regimens were as efficacious as the reference treatment of daily injectable procaine benzylpenicillin. Additionally, we noted improved adherence to the simplified regimens compared with the reference treatment. The simplest regimen, which included only two injections and oral antibiotic treatment, had both the lowest treatment failure rate and the highest adherence to treatment. Deterioration of clinical condition or death occurred in a small proportion of infants in all four groups. The most common reason for treatment failure in all treatment groups was persistence of clinical signs on day 4 after enrolment. Only one previous randomised trial 7 has compared the procaine benzylpenicillin gentamicin combination with a simple oral co-trimoxazole injectable gentamicin regimen, which was significantly less efficacious than Procaine benzylpenicillin and (group A, n=894) Oral amoxicillin and intramuscular (group B, n=884) Procaine benzylpenicillin and gentamicin for 2 days for 5 days (group C, n=896) Gentamicin and oral amoxicillin for 2 days for 5 days (group D, n=890) First week after enrolment Treatment failure 72 (8%) 52 (6%) 67 (8%) 48 (5%) Risk difference 2 2% ( 4 5 to 0 2) 0 6% ( 3 1 to 1 9) 2 7% ( 5 0 to 0 3) Death 10 (1%) 8 (1%) 18 (2 %) 10 (1%) Appearance of a sign of critical illness 6 (1%) 5 (1%) 8 (1%) 11 (1%) Appearance of a new sign of serious 7 (1%) 6 (1%) 5 (1%) 4 (<1%) infection Serious adverse effect other than death 1 (<1%) 0 (0%) 0 (0%) 0 (0%) Hospital admission 4 (0 5%) 3 (<1%) 4 (<1%) 1 (<1%) No improvement in clinical condition 36 (4%) 24 (3%) 24 (3%) 18 (2%) by day 4 Reappearance of inclusion sign 7 (1%) 5 (1%) 7 (1%) 4 (<1%) between days 5 and 8 Presence of inclusion sign on day 8 1 (<1%) 1 (<1%) 1 (<1%) 0 (0%) Second week after enrolment Death* 1 (<1%) 2 (<1%) 0 (0%) 1 (1%) Serious adverse effect other than death* 0 (0%) 0 (0%) 0 (0%) 0 (0%) Relapse* 8 (1%) 7 (1%) 2 (<1%) 7 (1%) First and second week after enrolment Death* 12 (1%) 10 (1%) 20 (2%) 11 (1%) Data are n (%) or risk difference (95% CI). *Percentage calculated out of number of infants who had not failed treatment. Table 4: Intention-to-treat analysis 8 Published online April 2,

9 the procaine benzylpenicillin gentamicin combination for treatment of infants with signs of severe infection. 7 Based on this finding, we did not include this regimen in our study. Our results from three African countries show for the first time that simplified antibiotic regimens are as efficacious as the reference treatment for young infants with signs of clinically suspected severe infection, in situations where referral is not possible (panel 3). The study has several strengths. First, it was a prospective, community-based, multicentre randomised trial with a large sample size in African populations that represented diverse settings in east, central, and west Africa. Second, we achieved high rates of follow-up in all study sites. Third, we noted very good adherence to treatment in all groups. Fourth, we implemented the study with good quality assurance and standardisation procedures in place, details of which have been published previously. 14 The health workers were trained to identify signs of clinical deterioration and their skills to do so are borne out by the low mortality rate. Fifth, we did the study in a programme setting within existing health systems at each site, which will enable dialogue with policy makers. We made use of existing health centres and groups of community health workers and nurses. Finally, the baseline characteristics were similar in all treatment arms, suggesting successful randomisation. These strengths support high internal and the external validity of the results from this study. Some limitations also existed. First, the trial was not fully blinded. Treatment allocation was open to the parents and the treating health workers because it was deemed to be unethical to give placebo injections to young infants. Although the outcome assessors were not actively informed about the infant s treatment allocation to reduce the risk of assessment bias, assessors could have obtained this information by questioning parents. Second, we only used clinical criteria to make our diagnosis, with no confirmatory microbiological or other supportive laboratory data. Parents of included infants did not accept hospital referral and we were unable to set up laboratory services at the first level health facilities. Moreover, the current standard of clinical practice, even in the best hospitals, is that neonates are treated for sepsis based on clinical signs. 15 The current methods for bacterial diagnosis have poor sensitivity, especially at the community level. 16 An ongoing large, multicentre causation study in Bangladesh, India, and Pakistan could provide some important data for the study of sepsis in the near future. Finally, our sample size did not have enough power to do an age group-specific subanalysis, but this type of analysis will be possible when the results of our study are combined with two similar ongoing studies in Bangladesh and Pakistan. Our findings suggest that young infants with clinical signs of severe infection can be effectively treated and managed on an outpatient basis when referral to a hospital is not possible or acceptable to families. Low-income and middle-income countries with a high burden of neonatal infections have problems with referral systems, among Panel 3: Research in context Systematic review We searched PubMed for various combinations of the terms infant * or infant or newborn* or newborn or neonat* or neonatal or infant, newborn and sepsis or septic or bacteremia or bactaeremia or bacter* or bacterial infection or infection or infections or infection* or pneumonia or meningitis or omphalitis or acute lower respiratory infection* or acute lower respiratory infections or acute respiratory infection or bacterial pneumonia or bacterial meningitis and community or community acquired or community based or community-based or community-acquired-infections and treatment. We focused mainly on studies in English and published since From Jan 1, 1980, to April 30, 2009, seven studies in eight reports were described in two systematic reviews. 12,13 Most studies had methodological inadequacies and used different delivery methods to deliver treatment, so the results were not comparable. WHO, USAID, and Saving Newborn Lives held a consultation to review unpublished data to assess the effectiveness of simplified antibiotic regimens for management of possible serious bacterial infection where referral was not accepted in low-income and middle-income country settings. The consultation recommended that well designed trials were needed to obtain information about the effectiveness of simpler antibiotic regimens where referral was not accepted. Interpretation The AFRINEST trial was designed to overcome drawbacks of earlier studies and determine the effectiveness of management of possible serious bacterial infection in young infants in Africa. Our findings show that simplified antibiotic regimens are effective and can be delivered in primary health-care settings where referral is not accepted by families. other health-system challenges. 6 8,17 19 High case fatality can result because of delay in appropriate and timely care seeking. 20,21 Ill infants whose families do not accept referral would benefit from prompt treatment near their homes, which would potentially reduce neonatal mortality rates. Contributors AT, AL, SN, CE, FE, PG, AIA, AGF, EAA, CHA, RDW, CLE, and WNO did the study and collected data. LG and SC did the primary analysis and all authors contributed to the interpretation of data and preparation of the manuscript. Declaration of interests SC and LG received funding support for data management from WHO. All other authors declare no competing interests. Acknowledgments The study was funded by a grant from the Bill & Melinda Gates Foundation to WHO, which sponsored the study. We thank the families who participated in these studies, the project field staff for their hard work and dedication, and the physicians and administration at every site who provided facilities for the conduct of the study. We also thank Rajiv Bahl, Nigel Rollins, Sachiyo Yoshida, and Shamim Ahmad Qazi for their assistance with coordination of the study. The following have contributed to the study and we therefore acknowledge their substantial contributions: David Muyodi, Millsort Kemboi, Judith Sitti, Evelyn Etemesi, Lydia Chebet, Enock Nyambane, Sylvia Nyabera, Moses Sitati, Edwin Kirwa, Priscah Mosol, and Francisca Lagat in Kenya; Elijah Afolabi Bamgboye, Ayotunde Olajide, Nathanael Afolabi, Bukola Adeniji, Ronke Egunjobi, and Olaide Olakehinde in Ibadan, Nigeria; Olufolake Akano, Christiana Olufunke Olagoke, Tosin Ogunsola, Aminat Subair, Jumoke Awoyemi, and Olusanjo Oyedokun in Ile-Ife in Nigeria; and Comfort H Garba, Joshua D Hyellashilni, and Mary Olothi in Zaria, Nigeria. References 1 United Nations Inter Agency Group for Child Mortality Estimation. Levels and Trends in Child Mortality. New York: UNICEF, WHO, The World Bank, UNDP, Published online April 2,

10 2 WHO and UNICEF. Integrated management of childhood illness (IMCI) Chart Booklet. Geneva: World Health Organization, Seale AC, Manu AA, Nair H, et al, for the PSBI Investigator Group. Estimates of possible severe bacterial infection in neonates in sub-saharan Africa, South Asia, and Latin America for 2012: a systematic review and meta-analysis. Lancet Infect Dis 2014; 14: WHO. Pocket book of hospital care for children: guidelines for the management of common illnesses with limited resources, 2nd edn. Geneva: World Health Organization, Bhandari N, Bahl R, Bhatnagar S, Bhan MK. Treating sick young infants in urban slum setting. Lancet 1996; 347: Baqui AH, El-Arifeen S, Darmstadt GL, et al. Effect of communitybased newborn-care intervention package implemented through two service-delivery strategies in Sylhet district, Bangladesh: a cluster-randomised controlled trial. Lancet 2008; 371: Zaidi AK, Tikmani SS, Warraich HJ, et al. Community-based treatment of serious bacterial infections in newborns and young infants: a randomized controlled trial assessing three antibiotic regimens. Pediatr Infect Dis J 2012; 31: Bang AT, Bang RA, Baitule SB, et al. Effect of home-based neonatal care and management of sepsis on neonatal mortality: field trial in rural India. Lancet 1999; 354: AFRIcan NEonatal Sepsis Trial Group. Simplified regimens for management of neonates and young infants with severe infection when hospital admission is not possible: study protocol for a randomized, open-label equivalence trial. Pediatr Infect Dis J 2013; 32 (9 suppl 1): S AFRIcan NEonatal Sepsis Trial Group. Treatment of fast breathing in neonates and young infants with oral amoxicillin compared with benzyl penicillin-gentamicin combination: study protocol for a randomized, open-label equivalence trial. Pediatr Infect Dis J 2013; 32 (9 suppl 1): S Zaidi AK, Baqui AH, Qazi SA, et al. Scientific rationale for study design of community-based simplified antibiotic therapy trials in newborns and young infants with clinically diagnosed severe infections or fast breathing in South Asia and sub-saharan Africa. Pediatr Infect Dis J 2013; 32 (9 suppl 1): S Sazawal S, Black RE, for the Pneumonia Case Management Trials group. Effect of pneumonia case management on mortality in neonates, infants, and preschool children: a meta-analysis of community-based trials. Lancet Infect Dis 2003; 3: Zaidi AK, Ganatra HA, Syed S, et al. Effect of case management on neonatal mortality due to sepsis and pneumonia. BMC Public Health 2011; 11 (suppl 3): S Wall SN, Mazzeo CI, Adejuyigbe EA, et al. Ensuring quality in AFRINEST and SATT: clinical standardization and monitoring. Pediatr Infect Dis J 2013; 32 (9 suppl 1): S Niermeyer S. Neonatal sepsis. In: Berman S, ed. Pediatric decision making, 4th edn. Denver: Mosby, 2003: Darmstadt GL, Saha SK, Choi Y, et al. Population-based incidence and etiology of community-acquired neonatal bacteremia in Mirzapur, Bangladesh: an observational study. J Infect Dis 2009; 200: Chowdhury EK, Arifeen SE, Rehman M, et al. Care at first-level facilities for children with severe pneumonia in Bangladesh: a cohort study. Lancet 2008; 372: Siddiqi S, Kielmann A, Khan M, et al. The effectiveness of patient referral in Pakistan. Health Policy Plan 2001; 16: Owais A, Sultana S, Stein AD, Bashir NH, Awaldad R, Zaidi AK.Why do families of sick newborns accept hospital care? A community-based cohort study in Karachi, Pakistan. J Perinatol 2011; 31: Sutrisna B, Reingold A, Kresno S, Harrison G, Utomo B. Care-seeking for fatal illnesses in young children in Indramayu, west Java, Indonesia. Lancet 1993; 342: Kallander K, Hildenwall H, Waiswa P, Galiwango E, Peterson S, Pariyo G. Delayed care seeking for fatal pneumonia in children aged under five years in Uganda: a case-series study. Bull World Health Organ 2008; 86: Published online April 2,