A Method of Assessing Clinical States in Bacterial Pneumonia Part II

Abstract

In Part I, the criterion for the evaluation of chest X-ray films was provided, the correlation between various symptoms and X-ray film findings in bacterial pneumonia was investigated by factor analysis, and a tentative rating scale for overall assessment of chemotherapy was proposed.The present report deals with data newly obtained in 91 cases together with the data of the 72 previous cases. Reproducibility between the results of factor analyses on the previous and the present data was examined and analysis was again carried out on the total data. Cluster analysis was then performed on the total data to observe agreement with the results of the factor analysis and to confirm the classification structure of various symptoms.(1) Factor analysis of the initial severity of symptoms observed at the start of chemotherapy reveals the following classification structure, common to both the previous and the present data: group 1 (basic symptoms), including dyspnea, rales, cyanosis, cardiac failure and two items on the X-ray film (by each physician and by the central committee); group 2 (respiratory tract inflammation), including cough, character and volume of sputum ; and group 3 (systemic reaction), including fever, leucocytosis and chest pain. The fact that the erythrocyte sedimentation rate did not show a definite relation to the third group in the present study was the only point of disagreement with the previous results. The same classification as above was also derived from cluster analysis.(2) According to the changes in symptoms during two weeks of chemotherapy, it was mentioned in Part I that the result of factor analysis was almost the same as that of the initial severity except for the fact that the changes in the X-ray films correlated to the changes in groups 1 and 3 simultaneously with the status of X-ray film findings in the classification structure being characterized as a problem for further investigation. In the present report, cluster analysis was performed together with factor analysis adding new data to those already obtained. The common findings from these results can be summarized as follows: group 1 concerning basic symptoms (same as in group 1 of (1) except for the X-ray film findings), group 2 concerning respiratory tract inflammation (same as in group 2 of (1)), group 3 (fever and leucocytosis) and group 4 (X-ray film findings and erythrocyte sedimentation rate). However, the changes in X-ray films also correlated with group 1 so the problem previously mentioned still remains.(3) The change in each individual symptom was investigated during the twoweek period of chemotherapy. The “life table survival curve” method was applied in this investigation, except that “disappearance of symptom” was used in place of “death” in the life table . It was found that the symptoms which showed rapid disappearance and those which showed slow change became separated, corresponding to the characteristic classification into the above four groups: the symptoms in the first and the third groups showed the most rapid changes, followed in decreasing order by those in the second group and those in the fourth group.The, method, aggregating various symptoms in the above four groups, was applied to the analysis of a clinical trial to compare the effects of different antibiotics. It was illustrated that a fairly compact and reasonable summary could be given for the results of the trial since the method arranges various symptoms in a few characteristic groups and focuses on the alleviation of symptoms which is believed to be the final goal of chemotherapy.