Two-arm cluster randomized controlled trial located in Hala district, Pakistan to determine the impact of using Lady Health Workers (LHW) of National Program for Family Planning and Primary Health Care to diagnose and manage severe pneumonia with oral amoxicillin on treatment failure rates at day 6 among 2-59 month old children. LHWs in the control arm receive a refresher in standard pneumonia case management. LHWs in the intervention arm receive standard training that is enhanced to include training in the recognition of severe pneumonia and its home management with oral amoxicillin. Clusters are by Union Council (UC), administrator units consisting of 7 to 25 LHWs; each UC is randomized to either enhanced pneumonia case management with oral amoxicillin therapy (intervention) for severe pneumonia or standard case management and referral to the nearest health facility for treatment (control). Process indicators reflecting the LHW's ability to assess, classify and treat pneumonia in the intervention group and cost-effectiveness data is also being collected.

Primary Hypothesis:

Enhanced pneumonia case management and oral amoxicillin therapy for severe pneumonia delivered by LHWs in the community will result in a reduction in treatment failure among children 2 - 59 months of age with severe pneumonia who are treated by the LHW compared with those referred for care by the LHW.

Secondary Hypotheses:

1. The proportion of treatment failure, [persistence of lower chest indrawing (LCI) or need for second line treatment between day 3 and day 14], will be less in the intervention arm compared with the control arm.
2. LHWs can adequately assess, classify, and treat severe pneumonia in 2 - 59 month old children, and adequately recognize and refer children who present with danger signs during initial antimicrobial therapy.

Acute respiratory infection (ARI) is one of the leading causes of morbidity and mortality in children less than five years of age, and is responsible for over 2 million annual deaths in this age group. Of the 10.8 million global child deaths estimated in 2000, 34% were in South Asia and 21% of the total deaths were attributed to pneumonia. Inclusion of all deaths for which acute lower respiratory infections (ALRI) are an associated cause would increase these estimates by two to threefold. The ARI mortality burden is principally constituted by pneumonia. The median incidence of pneumonia in under-five children in developing countries is estimated to be 0.29 episodes (interquartile range 0.21-0.71) per child year. It is estimated that in the developing countries 150.7 million new pneumonia cases occur each year, of which 11-20 million (7-13%) are severe enough to require hospital admission. The region of the world with the highest incidence of pneumonia is South and South East Asia with an estimated 55.2 million episodes occurring in children under five years old. In the developed world, an estimated 14.7 million cases occur each year.

Recent trends indicate that Pakistan has begun to lag behind many regional countries in major indicators such as infant mortality rates (IMR). Currently, Pakistan has an IMR of 81/1000 live births and under-five mortality rate of 103/1000 live births. Annual number of child deaths in Pakistan is estimated at 565,000, giving Pakistan 4th ranking after India, Nigeria and China. Of more than 560,000 under-five deaths reported in Pakistan in 2001, 19% were due to pneumonia.

At the community level WHO guidelines recommend that trained health workers treat non-severe pneumonia with an oral antibiotic and refer children with severe pneumonia (i.e. lower chest wall indrawing [LCI] present) to a local health facility to be treated by injectable antibiotic (and oxygen, as required). A series of WHO sponsored randomized controlled trials and a subsequent meta-analysis of results demonstrated that these guidelines implemented by community-based health workers (CHWs) resulted in a 24-27% reduction in total child mortality and a 36-42% reduction in pneumonia mortality in these research settings.Pandey et al. have shown 28% reduction in total under-five mortality using locally trained community health workers to detect and treat pneumonia.

It has been previously shown that oral amoxicillin is an effective treatment for (WHO-defined) severe pneumonia in children. More recently a multi-centre randomized controlled trial (RCT) conducted in eight countries has shown that oral amoxicillin is as effective as injectable penicillin for the treatment of severe pneumonia in a hospital setting. For non-severe pneumonia two large multicentre RCTs in Pakistan and India have shown that three-days' treatment with oral amoxicillin therapy was equivalent to five-days' treatment. Non-adherence was the most important risk factor for treatment failure in these studies and was also significantly related to longer duration of therapy. In addition, it has been shown in studies from India and Bangladesh/Indonesia that the three-days' antibiotic therapy resulted in lower prevalence of resistant organisms 15 days after initiation of therapy, when compared with a five-days' therapy.

The non-severe pneumonia and severe pneumonia studies mentioned above were all hospital based clinical efficacy studies (several conducted by members of this project team) so that close monitoring and follow-up could be assured. There is now a need to demonstrate the effectiveness of these management strategies for severe pneumonia in the community when implemented by CHWs. If the findings of the hospital-based RCTs can be replicated in community settings then this may have a very significant impact on global policy in community-based management of ARI in young children. Children with very severe pneumonia or very severe disease (such as those who have danger signs) will continue to be referred for hospital treatment following current WHO IMCI guidelines. The use of oral antibiotic for treating severe pneumonia at the community level has many potential benefits. This case management strategy, if shown to be effective, could:

• simplify treatment of severe pneumonia;
• minimize the need for referral or hospitalization;
• decrease treatment administration costs;
• reduce transport, food and lost income costs for the family;
• reduce the pressure on already overburdened hospitals;
• reduce the risk of needle-associated complications such as needle-borne infections; and
• reduce the risk of nosocomial infection during hospital stay.

LHWs in Pakistan are very well placed in the community and, in Hala district comprise the mainstay of the primary health care system. LHWs are trained in standard ARI case management based on WHO's guidelines and the program has an active health education component. Management of pneumonia through standard case management, part of IMCI - an approach developed by the WHO and UNICEF - is the principal strategy for reducing ARI mortality.LHWs employing the IMCI approach classify the severity of respiratory infections by observing a child for the key signs of pneumonia: chest in-drawing and fast breathing. With the help of the refresher course recently developed by the LHW program - based on the IMCI algorithm - LHWs will be additionally trained in identifying and managing severe pneumonia at the community level and improve their skills in promotion of care-seeking for ANA.While it is hoped that these measure will markedly reduce overall childhood mortality and pneumonia related mortality in the community, the study being proposed here will measure the effect of enhanced case management of ANA by LHWs in reducing the clinical treatment failure rate among children 2-59 months of age.

DEFINITION OF TREATMENT FAILURE

Suspected treatment failure will be determined by the LHW on subsequent home visits after the child has been enrolled. The LHW will immediately contact the Lady Health Worker Supervisor (LHS) who will independently confirm each case of failure. The provision of clinical care (i.e. arranging referral) will not be delayed should there be a delay between the identification of suspected treatment failure and confirmation by the LHS.

Clinical Treatment Failure through day 5 for children 2-59 months will be defined as:

1. Clinical deterioration through day 5 or death occurring within 14 days of enrollment, prior to change in primary antibiotic therapy.
2. Persistence of fever (>or=100 F) and lower chest indrawing on day 3 (after 48 hours of initiation of therapy)
3. Fever(>or=100 F)or lower chest indrawing or appearance of any danger sign on day 6
4. Secondary treatment failure or relapse is defined as appearance of any danger sign or lower chest indrawing or fast breathing (>or= 50 breaths/min) between day 7 and 14.

Case management in Control Arm:

The standard of care and training for LHWs in the control clusters will be the current curriculum and training workshops. All LHWs in the control clusters will receive a refresher in management of ARI through LHW supervisors. They will continue to treat non-severe pneumonia and other cases of ARI with oral co-trimoxazole. Skills in recognition of severe pneumonia will be reinforced in the refresher course and LHW will be instructed to continue to evaluate and refer these cases to the Rural Health Center (RHC) or district general hospital for admission and treatment (with injectable chloramphenicol or ampicillin) as per current policy. Patients who refuse to go to any health facility for treatment (either public or private), will not be offered enhanced home treatment with amoxicillin but will be provided oral co-trimoxazole as per GoP policy. These children will not be included in final analysis for treatment failure, but will be recorded in the overall epidemiological data from the area. Oral cotrimoxazole is provided free of cost per GOP policy. Any compensation for research related injury including hospitalizations and transportation to and from the health clinic will be borne by the study. Subjects who develop signs of penicillin allergy will be treated with an antihistamine at no cost and transported to the local health facility at the cost of project.

Case management in Intervention Arm:

LHWs in the intervention clusters will have specific training to enhance their ability to evaluate and treat ARI. This will include post-training evaluation of their ability to assess and classify non-severe and severe pneumonia and manage non-severe pneumonia. This will consist of an improved training module for the LHWs (respiratory care component) with an amended treatment plan for severe pneumonia with oral amoxicillin. The enhanced respiratory training module for LHWs will include prompt recognition and referral of study treatment failures (including very severe disease) by day 3, based on objective criteria of fever and lower chest indrawing. In the intervention arm LHWs who diagnose severe pneumonia will not refer the patient to the nearest facility (RHC with admitting facility or district general hospital), but instead will offer treatment with oral amoxicillin (90 mg/kg/day divided BID) at home for 5 days with close follow-up. Additionally, these LHWs will continue to treat non-severe pneumonia entirely in the community with a 3 days course of oral amoxicillin as per standard of practice. The study drug (oral amoxicillin) is provided free of cost per GOP policy. Any compensation for research related injury including hospitalizations and transportation to and from the health clinic will be borne by the study.

For severe pneumonia, oral amoxicillin will be provided as a suspension at a total dose of 90 mg/kg per day for 5 days

Patients who vomit within a half an hour of oral treatment will receive a repeat dose of amoxicillin. This will be attempted three times over 30 minutes. If the patient fails three attempts they will be considered a treatment failure.

Sample Size Calculation - Primary Objective

The investigators have calculated a sample size that will allow us to test the following null hypothesis:

Null hypothesis: The proportion of children who fail treatment in the treatment group will not be greater than the proportion of those who fail in the control group.

The method of Hayes and Bennett21, which employs the following equations, was used to calculate the number of clusters. The sample size calculation was then repeated using PASS software using procedures for cluster randomized equivalency trial Because the number of clusters was fixed, the investigators calculated the sample size for the trial by solving for the number of participants required given a set of assumptions.

c =2 + (Zα/2 + Zβ)2 * [π0*(1-π0)/n +π1*(1-π1)/n+k2*(π02+π12)]/(π0-π1)2

where π0 = prop int, π1=prop control, n=number per cluster, k=coefficient of variation

Below the investigators detail the assumptions used in sample size calculations and provide justifications for these assumptions:

1. Failure proportion in control group: The investigators anticipate that the proportion of children 2-59 months who fail standard of care therapy by day 6 will be 15%. This is based on the 15% failure proportion found in the APPIS I trial.(11) A consensus among investigators determined that failure proportion will be lower in this trial because the definition of failure in APPIS I was overly conservative.
2. Failure proportion in intervention group: The investigators anticipate that the proportion of children who fail therapy by day 6 in the intervention arm will be 10%. The null hypothesis is that the proportion of children 2-59 months who fail treatment in the intervention group will not be greater than the proportion of those who fail in t control group. Based on a consensus of the investigators, if a difference of 5% or less is observed in treatment failure between the groups, then the two treatments will be considered equivalent. This equivalence was considered clinically equivalent.
3. Number of clusters: The investigators have based the sample size determination on a total of 18 available clusters (9 per arm).
4. Power: The investigators have used a power of 0.9, setting the Type II error level at 0.1, or a 10% chance of not being able to detect a difference in treatment failure if it truly existed. This was based on a consensus among the investigators. Power was set high because this was an equivalency trial, and more emphasis should be placed on making sure any lack of difference was not based on an underpowered study.
5. Alpha: Based on a consensus of the principal investigators, the null hypothesis will be rejected based on a one-sided test (because the study focuses only on whether the intervention is not equivalent to the control), with a p=0.05 level of significance.
6. Coefficient of variation (k): Inherent correlations between observations in homogeneous clusters are accounted for in cluster-randomized trials through the coefficient of variation. The investigators have used an estimate of 0.20 for the study. Hayes and Bennett21 note that k is typically less than 0.25 for health studies. In this case the outcome is the proportion of children failing treatment and the investigators therefore assume that variation in failure rates between clusters will not exhibit great variation. A coefficient of variation of 0.2 would be equivalent to no cluster in the control arm having a failure rate below 0.06 or above 0.18.

Based on these assumptions, the investigators conclude that they will need to enroll 290 cases of severe pneumonia per cluster among 9 clusters per arm (total 18 clusters) over two years, for a total of 5,233 cases of severe pneumonia.

Inclusion Criteria:

• Age between 2 to 59 months who present to LHWs with severe pneumonia.
• Informed consent given by a legal guardian.

Exclusion Criteria:

• Very severe disease.
• Persistent vomiting.
• Parental or caretaker refusal to participate in the study.
• Children currently being treated for non-severe pneumonia who advance to severe pneumonia.
• Suspected or known kerosene oil ingestion.
• Prior enrollment in the study within 2 weeks of last follow up
• Children with severe malnutrition . Children with severe diarrhea with signs of dehydration