Antibiotic Efficacy in Pneumonitis Following Paraffin (Kerosene) Ingestion in Children

The average of 100 children per annum attending Red Cross War Memorial Children's Hospital 9RCWMCH) with the diagnosis of kerosene ingestion would give a sample of 200 children over a two-year period, with 100 patients in each group. From a postulated secondary infection rate of 15 to 50% for children not receiving an antibiotic, a midway point of 25% estimated the treatment failure rate in the placebo group. With no information available on the treatment failure rate in the active group, failure rates of 10% and 5% were arbitrarily applied. With 25% and 5% treatment failure rates for placebo and active groups respectively, at a level of significance of α = 0.05 a sample size of 100 per group gives a power of 0.98 and with failure rates of 25% and 10% a power of 0.80.

Statistical analysis was done using IBM SPSS Version 20 (SPSS Inc., Chicago, IL, USA). Categorical variables are expressed as n (%) and continuous variables as median (interquartile range (IQR)). A P value of ≤ 0.05 was considered significant for all situations. For categorical variables, Fischer's exact test was used for small samples or less frequent occurrences. Chi-Square testing was applied for larger samples or more frequent occurrences. Mann-Whitney or Kruskal-Wallis tests were used for ordinal and continuous variables. Significant correlation between factors and covariates (Spearman's rank coefficient) favoured univariate analysis over binary logistic regression modelling to determine potential risk factors for treatment failure. Continuous variables were categorised for clinical relevance or logistic regression testing. In some instances, specific clinical parameters or reported symptoms were not recorded or the presence or absence of a risk factor was unknown. The missing values, unknown factors and the flow of patient follow-up account for totals not always adding up to the full number of study participants. In the results for Day 3 and Day 5 post-ingestion, the denominator used to calculate proportions for reported symptoms included those patients who attended and who were telephone interviewed, whereas the denominator for clinical signs was only the patients who attended.

The primary outcome measure was treatment failure, as reported. Secondary outcome measures were length of hospital stay, reported symptoms (cough, shortness of breath, wheeze and fever) and clinical signs (respiratory rate, flaring, recessions, grunting, wheeze, crepitations, temperature and altered mental status) at follow-up at Day 3 and 5 post-ingestion for placebo and active groups. Further investigation explored the role of confounding conditions (upper respiratory tract infection, active Mycobacterium tuberculosis infection) and risk factors for treatment failure or delayed resolution (vomiting post-ingestion, household smoking contact, HIV exposure status, prior respiratory history, young age etc). Secondary outcome measures, confounding conditions and risk factors are not reported in this Clinical Trials format, but are reported in the PI's Master's thesis. (Balme KH. The efficacy of prophylactic antibiotics in the management of pneumonitis following paraffin (kerosene) ingestion in children [Master's thesis]. [Cape Town]: University of Cape Town; 2013. 113 p.)