MAJOR ARTICLE. Downloaded from by guest on 07 January 2019

Transcription

1 MAJOR ARTICLE Comparison of Oral Amoxicillin with Placebo for the Treatment of World Health Organization Defined Nonsevere Pneumonia in Children Aged 2 59 Months: A Multicenter, Double-Blind, Randomized, Placebo-Controlled Trial in Pakistan Tabish Hazir, 1 Yasir Bin Nisar, 1 Saleem Abbasi, 1 Yusra Pervaiz Ashraf, 1 Joza Khurshid, 1 Perveen Tariq, 3 Rai Asghar, 3 Asifa Murtaza, 2 Tahir Masood, 4 and Sajid Maqbool 4 1 ARI Research Cell, Children's Hospital, Pakistan Institute of Medical Sciences, and 2 Department of Pediatrics, Federal Government Services Hospital, Islamabad; 3 Department of Pediatrics, Rawalpindi General Hospital, Rawalpindi; and 4 Children's Hospital, Lahore, Pakistan Background. World Health Organization (WHO) acute respiratory illness case management guidelines classify children with fast breathing as having pneumonia and recommend treatment with an antibiotic. There is concern that many of these children may not have pneumonia and are receiving antibiotics unnecessarily. This could increase antibiotic resistance in the community. The aim was to compare the clinical outcome at 72 h in children with WHO-defined nonsevere pneumonia when treated with amoxicillin, compared with placebo. Methods. We performed a double-blind, randomized, equivalence trial in 4 tertiary hospitals in Pakistan. Nine hundred children aged 2 59 months with WHO defined nonsevere pneumonia were randomized to receive either 3 days of oral amoxicillin (45mg/kg/day) or placebo; 873 children completed the study. All children were followed up on days 3, 5, and 14. The primary outcome was therapy failure defined a priori at 72 h. Results. In per-protocol analysis at day 3, 31 (7.2%) of the 431 children in the amoxicillin arm and 37 (8.3%) of the 442 in placebo group had therapy failure. This difference was not statistically significant (odds ratio [OR],.85; 95%CI, ; P 5.60). The multivariate analysis identified history of difficult breathing (OR, 2.86; 95% CI, ; P 5.027) and temperature.37.5 C 100 F at presentation (OR, 1.99; 95% CI, ; P ) as risk factors for treatment failure by day 5. Conclusion. Clinical outcome in children aged 2 59 months with WHO-defined nonsevere pneumonia is not different when treated with an antibiotic or placebo. Similar trials are needed in countries with a high burden of pneumonia to rationalize the use of antibiotics in these communities. Clinical Trials. NCT Pneumonia is one of the major causes of death in children,5 years of age in developing countries, causing.2 million deaths annually [1]. To reduce the number of pneumonia-related deaths in communities where Received 17 January 2010; accepted 17 November Correspondence: Tabish Hazir, MBBS, ARI Research Cell, Children Hospital, Pakistan Institute of Medical Sciences, Islamabad, Pakistan (tabishhazir@ hotmail.com). Clinical Infectious Diseases 2011;52(3): Ó The Author Published by Oxford University Press on behalf of the Infectious Diseases Society of America. All rights reserved. For Permissions, please journals.permissions@oup.com /2011/ $37.00 DOI: /cid/ciq142 qualified personnel and diagnostic facilities are often not available, the World Health Organization (WHO) developed case management guidelines based on simple clinical signs followed by empirical treatment [2]. Children with fast breathing only are classified as having nonsevere pneumonia requiring oral antibiotics at home, and children who have lower chest indrawing (LCI) and/ or general danger signs (GDS) are classified as having severe pneumonia and very severe pneumonia, respectively. Both groups require referral to hospital and injectable antibiotics. These guidelines have shown to effectively reduce acute respiratory illness (ARI) related mortality by 34% [3] in communities where they have been successfully implemented. Comparison of Oral Amoxicillin with Placebo d CID 2011:52 (1 February) d 293

2 However, there are certain concerns regarding these guidelines. Many children classified as having nonsevere pneumonia (cough and fast breathing) have neither clinical pneumonia as assessed by physicians nor pneumonia apparent on chest radiographs [4, 5]. Use of the sign of fast breathing alone for prescribing antibiotics may not be the most judicious from a public health perspective and may lead to the spread of antibiotic resistance. Fast breathing may be caused by conditins other than pneumonia. Respiratory rate in children is affected by many factors, such as fever. Although WHO guidelines recommend that respiratory rate must be counted when a child is afebrile, calm, and preferably sleeping, in real life situations, such practice is not always possible. Another longstanding unaddressed issue is the inclusion of many children with wheeze in this algorithm. Data from both developed and developing countries show that wheeze is more common in viral than bacterial infections, particularly respiratory syncytial virus infection [6212]. A Cochrane metaanalysis of infants with bronchiolitis found no difference in clinical outcome with antibiotics, compared with placebo [13]. WHO guidelines do not make a distinction between viral and bacterial pneumonia, because controversy still exists concerning the relative importance of these pathogens. A reliable clinical distinction between bacterial and viral infection has yet to be established, and this issue has been problematic for the program managers, senior researchers, and technical experts and remains to be answered. These children continue to receive antibiotics because of the concern that it may not be safe to do otherwise. To develop effective guidelines for initial antibiotic treatment, it is essential to have information on the likely causative organism and the pattern of antibiotic resistance. Such studies are expensive and require expertise, which is often lacking in the routine setting of most of the developing countries. For the aforementioned reasons, there is a need to further elaborate the role of antibiotics in children with cough and fast breathing. To address this issue, we conducted this randomized, double-blind, placebo-controlled equivalence trial of oral amoxicillin versus placebo in children aged 2 59 months with WHO-defined nonsevere pneumonia, taking into account all the necessary safety precautions for their well being. METHODOLOGY Study Design and Patients This double-blind, randomized, placebo-controlled trial was performed in 4 hospitals in 3 cities in Pakistan (Islamabad, Lahore, and Rawalpindi). Children aged 2 59 months with cough, difficult breathing, or both were screened for fast breathing. Children were classified using standard WHO algorithm for ARI (Table 4) [2]. Children classified as having nonsevere pneumonia were included in this study. Children who had underlying chronic illness, had a history of >3 episodes of wheeze or acute bronchial asthma, and had used any antibiotic in appropriate doses during the previous 48 h were excluded. Verbal informed consent was obtained from parents or caregivers. The trial was approved by the institutional ethical review boards of the Pakistan Institute of Medical Sciences, Islamabad, and other participating institutions. Procedures Enrolled children were randomly assigned to receive either oral amoxicillin (15 mg/kg every 8 h) or placebo with similar color and consistency in the same volume and were masked with regard to taste and smell to the greatest extent possible (Novartis Pharma). The treatment was continued for 3 days. The randomization scheme was generated by a computer program at ARI Research Cell, Islamabad, with uneven blocks of 4, 6, and 8. Self-adhesive sticking labels with unique identification numbers were prepared in Islamabad. A copy of the randomization list with unique identification numbers was given to a health professional not associated with the study who randomized the study drugs. Drug assignment was concealed from patients, parents, and study personnel. The code was broken after the data analysis. The first dose of the drug was given under supervision by the mother or caregiver in the facility with use of a syringe. Moreover, oral salbutamol and paracetamol were given when required. Study physicians counseled caregivers on the recognition of danger signs and urged them to return to the health facility at any time if the condition of the child deteriorated. All guardians were provided with the cell phone numbers of study personnel for an unscheduled visit. Hypotheses The primary hypothesis was that the therapy outcome in those who received amoxicillin should be no different from that in those who received placebo when used for 3 days for the treatment of nonsevere pneumonia in children aged 2 59 months. The study outcome was treatment failure by day 3 (Table 5). The secondary outcomes were treatment failure by day 5 (cumulative treatment failure) and relapse. Relapse of disease was defined as development of any sign of pneumonia 6 14 after fast breathing had initially returned to normal. Clinical resolution was defined as either return of respiratory rate to the normal age-specific range or an improvement by day 3 (Table 5). Assessment of Patients Respiratory rate was counted twice for 1 min each in 5 min, preferably when the child was quiet, feeding, or asleep. For assessment, the mean value of the 2 readings, rounded to 2 significant figures, was used. If the difference between the 2 readings was >5 breaths per min, a third reading was taken. The mean was then taken from the 2 readings having a difference of,5 breaths per min. According to WHO protocol, 294 d CID 2011:52 (1 February) d Hazir et al.

3 Figure 1. a wheezy child was given 3 cycles of salbutamol nebulization 15 min apart and was reassessed for fast breathing. To assess the respiratory rate objectively, febrile children were given paracetamol, although it was not always possible to record the respiratoryrateinchildrenwhentheydidnothavefever.all study personnel responsible for data collection were trained for 3 days in standard ARI patient management and in study methods. A senior pediatrician at each site supervised all procedures Follow-up of Patients The study physician assessed every child on day 3 (first followupvisit), day 5 (second follow-up visit), and day 14 (third follow-up visit). Treatment was stopped after 3 days if a child was judged to have clinical resolution defined a priori (Table 5). The study personnel assessed patients who did not come for followup at the specified date on the subsequent day at the child s home. If the child was not available on that day, one more attempt was made to contact them before they were judged to be lost to follow-up. The patient was judged to be compliant if she Trial Profile. or he had taken 80% of the required dose since the previous visit. Drug containers were returned at follow-up visits, and the remaining drug was measured using specially designed compliance cards. Therapy Change Oral amoxicillin was changed (or started in placebo group) to oral chloramphenicol if the child was declared as treatment failure by day 3, day 5, or at relapse. Children who developed severe pneumonia or very severe pneumonia and/or disease or who did not improve after 48 h of oral chloramphenicol therapy were referred for inpatient treatment with the appropriate injectable antibiotic. Oral cefixime was used for 7 days when injectable therapy was not possible. All children were followed up until completely well. Statistical Analysis The sample size was estimated on the basis of equivalence (defined a priori) of a difference of <5% in the treatment failure rates between 2 study groups by day 3. We used a standard Comparison of Oral Amoxicillin with Placebo d CID 2011:52 (1 February) d 295

4 Table 1. Baseline Characteristics of All Patients in the 2 Treatment Groups (n 5 873) Table 1. (Continued) Baseline characteristics Amoxicillin (n 5 431) Placebo (n 5 442) Baseline characteristics Demographic indicators Sex Amoxicillin (n 5 431) Placebo (n 5 442) Male 281 (65.2%) 276 (62.4%) Age, months Mean (1 SD) 11.1 (610.2) 12.0 (611.2) Median (IQR) 8.0 ( ) 8.0 ( ) Age categories, months (34.6%) 145 (32.8%) (31.1%) 145 (32.8%) (34.3%) 152 (34.4%) History Duration of illness, days Mean (6 S.D) 2.8 (61.7) 2.8 (61.9) Median (IQR) 2.0 ( ) 2.0 ( ) Duration of illness categories <3 days 224 (52.0%) 232 (52.5%) >4 days 207 (48.0%) 210 (47.5%) Fever 322 (74.7%) 333 (75.3%) Cough 426 (98.8%) 438 (99.1%) Difficult breathing 381 (88.4%) 399 (90.3%) Vomiting 52 (12.1%) 52 (11.8%) Diarrhea 22 (5.1%) 18 (4.1%) Breastfeeding 308 (71.4%) 310 (70.1%) Breastfeeding, by age 2 5 months 128/149 (86.0%) 122/145 (84.1%) 6-11 months 105/134 (78.4%) 122/145 (84.1%) months 75/148 (50.6%) 66/152 (43.4%) History of wheeze 21 (4.9%) 16 (3.6%) Signs on examination Wheezing 235 (54.5%) 221 (50.0%) Wheezing, by age 2 5 months 77/149 (50.3%) 71/145 (49.0%) 6-11 months 75/134 (56.0%) 74/145 (51.0%) months 83/148 (56.1%) 76/152 (50.0%) Temperature, o F Mean (6 SD) 99.2 (6.9) 99.1 (6.7) Median (IQR) 99.0 ( ) 99.0 ( ) Temperature,100 0 F 273 (63.3%) 303 (68.6%) >100 0 F 158 (36.7%) 139 (31.4%) Respiratory rate, breaths/min, by age 2 11 months Mean (1 SD) 58.9 (64.5) 58.9 (64.0) Median (IQR) 59.0 ( ) 59.0 ( ) Respiratory rate categories,60 breaths/min 164/283 (58.0%) 165/290 (56.9%) >60 breaths/min 119/283 (42.0%) 125/290 (43.1%) months Mean (6 S.D*) 54.3 (66.9) 52.9 (65.8) Median (IQR ) 55.0 ( ) 53.0 ( ) Respiratory rate categories Less than 50 42/148 (28.4%) 52/152 (34.2%) breaths/min More than (71.6%) 100/152 (65.8%) breaths/min Abbreviations: IQR, interquartile range; SD, standard deviation. formula for detection of equivalence 2N 5 4(Z a 1 Z b ) 2 p(1-p)/ d 2 [14]. Results of a previous study from the same center reported a 6% treatment failure rate with oral amoxicillin for nonsevere pneumonia among children aged 2 59 months [15]. With 1-sided a of.05, power of 90%, and anticipated loss to follow-up of 8%, the estimated sample size was 422 children per group, rounded to 450 children per group. Data were recorded using autocopy forms. One copy was sent to the coordinating center for data entry. Data were double entered, and validated using Epi Info software, version 6.04 (Centers for Disease Control and Prevention, CDC Atlanta). Epi Info, version 6.04, and SPSS, version 13.0 (SPSS) were used for analysis. The primary outcome was analyzed by intention to treat, with other analyses done per protocol. For the primary outcome, proportion of treatment failure and success for the 2 regimens were estimated and compared. Baseline characteristics of all children in 2 treatment arms, by primary outcome, were presented as frequencies and percentages. As a secondary objective, baseline characteristics between cumulative treatment success and failure groups by day 5 were also compared using the v 2 test for categorical variables and Student s t test for continuous variables. Univariate analysis of categorical variables included estimates of odds ratios (ORs) and 95% confidence intervals (CIs). For Student s t test, only P values are reported in results. Furthermore, a multivariate model was constructed to look for determinants of cumulative treatment failure (day 5) by forward stepwise logistic regression in all patients. All significant variables from univariate analysis were included in the model. Moreover, some continuous variables, such as age, duration of illness, and respiratory rate, were categorized to make our results relevant to the management of pneumonia. Statistical significance was considered at the 5% level. RESULTS Enrollment began in May Pilot was completed in August From September 2005 through June 2008, a total of 900 children were enrolled in this study at the 4 study sites: Children s Hospital and Federal Government Services Hospital, Islamabad; Rawalpindi General Hospital, Rawalpindi; and 296 d CID 2011:52 (1 February) d Hazir et al.

5 Table 2. Primary Outcome on Day 3 and Cumulative Treatment Failure by Day 5 Outcome Amoxicillin Placebo Difference (95% CI) OR (95% CI) P value Intention-to-treat analysis Treatment failure by day 3 50/450 (11.1%) 45/450 (10.0%) 1.1 ( ) 1.13 ( ).59 Cumulative treatment failure by day 5 77/450 (17.1%) 86/450 (19.1%) 22.0 ( ) 0.87 ( ).44 Per-protocol analysis Treatment failure by day 3 31/431 (7.2%) 37/442 (8.3%) 21.1 ( ) 0.85 ( ).52 Cumulative treatment failure by day 5 58/431 (13.5%) 78/442 (17.6%) 24.1 ( ) 0.73 ( ).09 Relapse by day 14 3/373 (.8%) 5/364 (1.4%) -.6 (0.1.9) 0.58 ( ).46 Abbreviations: CI,confidence interval. Children s Hospital, Lahore. Four hundred fifty children each were randomly allocated to the amoxicillin and placebo arms. After exclusion of 19 patients (2 bottles broken, 7 discontinued intervention, and 10 lost to follow-up) from the amoxicillin group and 8 patients (lost to follow-up) from the placebo group, 873 children completed the treatment (431 with oral amoxicillin and 442 with placebo) and were successfully followed up until day 14 (Figure 1). Baseline characteristics of the 2 groups are shown in Table 1. No statistically significant differences were observed for baseline characteristics between the 2 groups. Five hundred fifty-seven children (63.8%) were male, and 294 children (33.7%) were aged 2 5 months. In the infant group (<12 months), 477 children (83.2%) were breastfed, and 471 (53.9%) had wheeze at the time of enrollment. At day 3 of follow-up, treatment failed in 31 children (7.2%) in the amoxicillin group and in 37 (8.3%) in the placebo group (Figure 1). This difference was not statistically significant (OR,.85; 95% CI, ; P 5.52). At day 5, 373 of 400 children in the amoxicillin group and 364 of 405 children in the placebo group were declared to have had treatment success (Figure 1). Therefore, the cumulative treatment failure rate by day 5 in the amoxicillin group ( 13.5% [58 children]) and in the placebo group (17.6% [78 children]) was not statistically significant (OR,.73; 95% CI, ; P 5.09). According to intention to treat analysis, the primary outcome between the 2 groups on day 3 did not differ significantly, compared with per-protocol analysis (OR, 1.13; 95% CI, ; P 5.44) (Table 2). At the day 14 follow-up visit, 3 of 373 children in the amoxicillin group and 5 of 364 children in placebo group had experienced relapse; however, the difference was not statistically significant (difference, -.6; OR,.58; 95% CI, ; P5.69). On day 3, the antibiotic was changed (amoxicillin group) or started (placebo group) in 68 children. The most common reason for change or start of antibiotic was development of LCI (n 5 45). The number of children who developed LCI in the amoxicillin (n 5 23) and placebo (n 5 22) groups was similar. Similarly, 8 children in the amoxicillin group and 9 children in the placebo group developed WHO-defined general danger signs, specifically, lack of ability to drink. From day 3 through day 5, an additional 29 children in the placebo group and 20 in the amoxicillin group developed LCI and 7 more children in each group developed general danger signs. A total of 26 enrolled children; 14 in the placebo group and 12 in the amoxicillin group needed hospitalization. All children who failed therapy initially were treated successfully with second- or thirdline antibiotics. No deaths were reported among enrolled children in this study. The comparison of baseline characteristics of children in the cumulative treatment failure and treatment success groups isshownintable3.bivariateanalysesshowedthattherapy failure was statistically associated with duration of illness,3 days (OR, 1.58; 95% CI, ; P 5.015), history of fever (OR, 1.67; 95% CI, ; P 5.030), difficult breathing (OR, 3.55; 95% CI, ; P 5.004), history of vomiting (OR, 1.76; 95% CI, ; P 5.02), history of diarrhea (OR, 2.45; 95% CI, ; P 5.01), presence of wheeze (auscultatory and audible; OR, 1.58; 95% CI, ; P 5.03), body temperature.37.5 C F (OR, 2.22; 95% CI, ; P 5.001), and respiratory rate.50 breaths per minutes in children aged months (OR, 2.46; 95% CI, ; P 5.03). In logistic regression analysis of all patients, cumulative treatment failure was associated with history of difficult breathing (OR, 2.86; 95% CI, ; P 5.027) and presence of temperature F measured at the time of enrollment (OR, 1.99; 95% CI, ; P ). DISCUSSION In the present study, we revealed that therapy failure rates among children aged 2 59 months with WHO-defined nonsevere pneumonia who were treated with amoxicillin were equivalent to such rates among similar children receiving placebo. Current guidelines are based on the concern that pneumonia causes death in more children in the developing countries than any other disease, and in situations lacking access to laboratory Comparison of Oral Amoxicillin with Placebo d CID 2011:52 (1 February) d 297

6 Table 3. Comparison of Cumulative Therapy Success and Failure (by Day 5) by Baseline Characteristics of All Patients (n 5 873) Baseline characteristics Therapy Success (n 5 737) Therapy Failure (n 5 136) OR (95% CI) P value Demographic indicators Gender Male 473 (64.2%) 84 (61.8%) 0.90 ( ).59 Age, months Mean (6S.D) 11.8 (610.7) 10.5 (610.8).228 Median (IQR) 8.0 (4.0 to 15.0) 7.0 (4.0 to 12.0) Age categories, months (33.4%) 48 (35.3%) 0.81 ( ) (31.5%) 47 (34.6%) 0.78 ( ) (35.1%) 41 (30.1%) 1.0 History Duration of illness, days Mean (6S.D) 2.9 (61.8) 2.6 (61.6).078 Median (IQR) 2.0 ( ) 2.0 ( ) Duration of illness <3 days 372 (50.5%) 84 (61.8%) 1.58 ( ).015 >4 day 365 (49.5%) 52 (38.2%) 1.0 Fever 543 (73.7%) 112 (82.4%) 1.67 ( ).03 Cough 728 (98.8%) 136 (100.0%).19 Difficult breathing 649 (88.0%) 131 (96.3%) 3.55 ( ).004 Vomiting 80 (10.9%) 24 (17.6%) 1.76 ( ).02 Diarrhea 28 (3.8%) 12 (8.8%) 2.45 ( ).01 Breastfeeding 518 (70.3%) 100 (74.1%) 1.17 ( ).44 Breastfeeding, by age 2 5 months 216/246 (87.8%) 34/48 (70.8%) 0.34 (.15.75) months 184/232 (79.3%) 43/47 (91.5%) 2.80 ( ) months 118/259 (45.6%) 23/41 (56.1%) 1.53 ( ).21 History of wheeze 30 (4.1%) 7 (5.1%) 1.84 ( ).43 Signs on examination Wheezing 372 (50.5%) 84 (61.8%) 1.58 (1.07 to 2.35).01 Wheezing, by age 2 5 months 112/246 (45.5%) 36/48 (75.0%) 3.59 (1.70 to 7.69) months 118/232 (50.9%) 31/47 (66.0%) 1.87 (.93 to 3.80) months 142/259 (54.8%) 17/41 (41.5%) 0.58 (.28 to 1.19).11 Temperature Mean (6S.D) 99.1 (6.8) 99.4 (61.0).001 Median (IQR) 99.0 ( ) 99.8 ( ) Temperature categories,100 0 F 508 (68.9%) 68 (50.0%) 1.0 >100 0 F 229 (31.1%) 68 (50.0%) 2.22 (1.51 to 3.27).001 Respiratory rate, breaths/min, by age 2 11 months Mean (6S.D) 58.8 (64.2) 59.9 (64.6).015 Median (IQR) 59.0 (56.0 to 61.0) 59.0 (57.0 to 62.5) Respiratory rate, breaths/min,60 273/478 (57.1%) 58/90 (64.4%) 1.0 >60 205/478 (42.9%) 32/90 (35.6%) 0.73 (.45 to 1.20) months Mean (6S.D) 53.3 (66.2) 55.7 (67.5).026 Median (IQR) 53.0 ( ) 57.0 ( ) Respiratory rate, breaths/min,50 87/259 (33.6%) 7/41 (17.1%) 1.0 >50 172/259 (66.4%) 34/41 (82.9%) 2.46 (.99 to 6.35).03 Abbreviations: CI,confidence interval; IQR, interquartile range; SD, standard deviation. 298 d CID 2011:52 (1 February) d Hazir et al.

7 Table 4. World Health Organization Classification of Acute Respiratory Illness in Children Presenting with Cough and/or Difficult Breathing No pneumonia (cough and cold) Nonsevere pneumonia Severe pneumonia Very severe disease Table 5. Clinical outcome Clinical resolution by day 3 Study Outcome Definitions Treatment failure by day 3 Treatment failure from days 4 to 14 Clinical resolution from days 4 to 14 Respiratory rate, breaths/minute,50 (infants 2 11 months),40 (children months) No lower chest indrawing Respiratory rate, breaths/ minute.50 (infants 2 11 months),40 (children months) No lower chest indrawing Lower chest indrawing with or without rapid breathing Unable to drink, convulsions, abnormally sleepy or difficult waking, stridor in calm child or clinically severe malnutrition Study definitions Respiratory rate* less by 10 breaths per minute compared to baseline reading, OR Respiratory rate* less than the age specific range. *without lower chest indrawing or general danger signs Developed lower chest indrawing or any of the General danger signs Presence of fast breathing, lower chest indrawing or general danger signs after clinical resolution on day 3. All other children investigations, tachypnea can be used as a surrogate for the diagnosis of pneumonia. There are no randomized, placebocontrolled trials comparing the usefulness of antibiotics in children classified as having nonsevere pneumonia. Our study is, to our knowledge, the first double-blinded, randomized, placebo-controlled trial providing scientific evidence performing similar trials in diverse settings of developing countries that have a high burden of pneumonia. The results of these studies, if similar, could form the basis for changing the current WHO recommendations. This will further rationalize the use of antibiotics in one of the common childhood illnesses, thereby reducing pressure on antimicrobial resistance, which is becoming a real problem for the program in developing countries [16, 17]. Our study had a high proportion of wheezers. Because of the well-known association of wheeze with viral pneumonias, particularly RSV bronchiolitis [629], it is likely that this high incidence of wheeze is actually a reflection of changing epidemiology of pneumonia in developing countries. Current WHO guidelines take cognizance of this fact and, to avoid overcategorization of pneumonia and rationalize the use of antibiotics in children with audible wheeze, recommend giving 2 cycles of rapidly acting inhaled bronchodilators at 15-min intervals before reassessment for antibiotic administration. This recommendation has shown to be useful [18], but there are still some unresolved issues. Data show that 70% of the children with auscultatory wheeze do not have audible wheeze [18] and, therefore, are being excluded from a trial of bronchodilators before assessment. Moreover, bronchodilators are not effective in bronchiolitis [19, 20]. Our results provide evidence that can be used for developing a more pragmatic management algorithm for children with wheeze. There could be a concern that many children with fast breathing may have bacterial pneumonia and, therefore, will be denied the benefit of antibiotic therapy. Of interest, that rate of subsequent deterioration in both arms suggests that giving amoxicillin to these children with fast breathing has no protective effect on the subsequent course of illness. Greater emphasis, therefore, is needed for closer follow-up and better counseling of the caregivers of these children, so that the children at high risk can be identified in time and the appropriate treatment is initiated. Moreover, further study is needed on the characteristics of these children to better understand the reasons for failure. Determinants of subsequent deterioration can then be used in a model to identify children at high risk at the time of presentation. There are certain strengths and limitations of our study. The loss to follow-up was only 2%. No deaths were reported in our study because of the improved training of the study staff in acute respiratory illness case management and close follow-up at home if the patient failed to visit at the health facility. This again highlights the importance of close monitoring and good followup. Our study did not include microbiologis analysis, thereby adding little to the existing knowledge of the prevalent etiologic agents in developing countries. Because this randomized controlled trial was not intended to evaluate the effectiveness of antibiotics in possible bacterial pneumonia but was an evaluation of performance of current WHO acute respiratory illness case management guidelines, performing any investigation before antibiotic prescription would have diverted us from the purpose of our analysis. In this randomized controlled trial, WHO guidelines were followed in exactly the same way as they are being practiced in the community. We conclude that our results strongly suggest that the benefit of antibiotics in children 2 59 months of age who are have WHO-defined nonsevere pneumonia is questionable. Further research from developing countries is needed to validate the results of this study. If validated, the use of antibiotics in WHO current guidelines can be further rationalized. Comparison of Oral Amoxicillin with Placebo d CID 2011:52 (1 February) d 299

8 Acknowledgments We thank the participants in the study, without whom this study would not have been possible; the study physicians (Dr Sobia Riaz, Dr Sajid Randhawa, and Dr Alamzeb) and other junior paramedical staff, for their assistance; Novartis Pharma Pakistan, which supplied the study drugs for this trial; and Shamim A. Qazi, Dr Donald Thea, Dr LeAnne Fox, and Dr. Patricia Hibberd, for their valuable technical advice. Financial support. This work was supported by ARI Research Cell, Children Hospital, Pakistan Institute of Medical Sciences, Islamabad, Pakistan. Potential conflicts of interest. All authors: no conflicts. References 1. Rudan I, Boschi-Pinto C, Biloglav Z, et al. Epidemiology and etiology of childhood pneumonia. Bull World Health Organ 2008; 86: World Health Organization. Technical basis for the WHO recommendations on the management of pneumonia in children at first level health facilities. Geneva: World Health Organization, (WHO/ ARI/91.20). 3. Sazawal S, Black RE. Effect of pneumonia case management on mortality in neonates, infants, and pre-school children: a meta-analysis of community-based trials. Lan Infect Dis 2003; 3: Hazir T, Nisar YB, Qazi SA, et al. Chest radiography in children aged 2 59 months diagnosed with non-severe pneumonia as defined by World Health Organization: descriptive multi-centre study in Pakistan. BMJ /bmj Mulholland EK, Simoes EF, Costales MOD, et al. Standardized diagnosis of pneumonia in developing countries. Pediatr Infect Dis J 1992; 11: Forgie IM, O Neill KP, Lloyd-Evans N, et al. Etiology of acute lower respiratory tract infections in Gambian children: I. Acute lower respiratory tract infections in infants presenting at the hospital. Pediatr Infect Dis J 1991; 10: Selwyn BJ. on behalf of the coordinated data group of BOSTID researchers. The epidemiology of acute respiratory tract infections in young children: comparison of findings from several developing countries. Rev Infect Dis 1990; 12(suppl 8):s Nunez RM, Duque J, Henriquez CW, et al. Viral and chlamydial etiology of acute infections of the lower respiratory tract in Colombian children. Pediatr Infect Dis J 1988; 7: Sobeslarky O, Sebikari, Hardland PSEG, et al. The viral etiology of acute respiratory tract infections in children in Uganda. Bull World Health Organ 1997; 55: Cherian T, Simoes EA, Steinhoff MC, et al. Bronchiolitis in tropical south India. Am J Dis Child 1990; 144: Weber MW, Dackour R, Usen ST, et al. The clinical spectrum of respiratory syncitial virus disease in the Gambia. Pediatr Infect Dis J 1998; 17: Videla C, Carballal G, Misirlian A, et al. Acute lower respiratory infections due to respiratory syncytial virus and adenovirus among hospitalized children from Argentina. Clin Diag Virol 1998; 10: Spurling GK, Fonseka K, Doust J, et al. Antibiotics for bronchiolitis in children. Cochrane Database Syst Rev CD Friedman LM, Furburg CD, DeMets DL. Fundamentals of clinical trials. St. Louis: Mosby, Hazir T, Qazi SA, Nisar YB, et al. Comparison of standard versus double dose of amoxicillin in the treatment of non-severe pneumoniainchildrenaged2 59months: a multi-centre, double blind, randomized controlled trial in Pakistan. Arch Dis Child 2007; 92: Hazir T, Qazi SA, Nisar YB, et al. Can WHO therapy failure criteria for non-severe pneumonia be improved in children aged 2 59 months. Int J Tuberc Lung Dis 2006; 10: Qazi SA. Antibiotic strategies for developing countries: an experience with Acute Respiratory Infections in Pakistan. Clin Infect Dis 1999; 28: Hazir T, Qazi S, Nisar YB, et al. Assessment and management of children aged 1 59 months presenting with wheeze, fast breathing, and/or lower chest indrawing; results of a multi-centre descriptive study in Pakistan. Arch Dis Child 2004; 89: Kellner JD, Ohlsson A, Gadomski AM, et al. Bronchodilators for bronchiolitis. Cochrane Database Syst Rev CD Flores G, Horwitz RI. Efficacy of beta2-agonists in bronchiolitis: a reappraisal and meta-analysis. Pediatrics 1997; 100: d CID 2011:52 (1 February) d Hazir et al.