Clinical and laboratory predictors of invasive diarrhoea in children less than five years old.

Five hundred and thirty-three children < 5 years of age with acute watery diarrhoea were followed for appearance of blood in their stools to evaluate various clinical and laboratory predictors of dysentery. The statistical analysis was performed in two steps. Firstly, clinical and laboratory variables were evaluated as 'predictors' for isolation of invasive bacterial enteropathogens from stool culture. Secondly, all the variables, including isolation of invasive bacterial enteropathogens, were analyzed for predicting bloody diarrhoea. Presence of mucus (sensitivity = 57.4%, specificity = 83.1%, positive predictive value = 24.8%, negative predictive value = 95.3%, p = 0.0000) was the only significant clinical predictor of bloody diarrhoea. Polymorphonuclear leucocytes (PMN) > 10/hpf (sensitivity = 55.3%, specificity = 92.4%, positive predictive value = 41.3%, negative predictive value = 95.5%, p = 0.0000) and red blood cells, regardless of their number, (sensitivity = 78.7%, specificity = 93%, positive predictive value = 52.1%, negative predictive value = 97.8%, p = 0.0000) on light microscopy of fresh stool samples were two significant laboratory findings. Isolation of invasive bacterial enteropathogens was also a significant predictor of bloody diarrhoea (sensitivity = 19.1%, specificity = 89.5%, positive predictive value = 15%, negative predictive value = 92% p = 0.0077) but ranked lower than the other 3 variables. Our results indicated that it is the severity of invasion caused by offending enteropathogens, clinically manifesting as mucoid stools and/or presence of PMN or RBC, which may be a better indicator of the invasive process and subsequent development of dysentery rather than mere isolation of invasive enteropathogens from the stools.