Unusual clustering of diseases and manifestations in pediatric admissions of a tertiary care center.

BACKGROUND Cluster investigations has been an important tool in investigations of diseases. While clustering of diseases or a manifestation in community can cause great public alarm, similar unexplained clustering in hospital admissions has baffled clinicians. OBJECTIVES (a) To determine whether perceived unexplained clustering of diseases and manifestations among hospital admissions is real or significant. (b) To look for possible explanations of such clustering. DESIGN One-year prospective observational and two-year retrospective study. MATERIAL AND METHODS Cases admitted under pediatric and neonatal services (excluding rooming in babies) from September 2002 to August 2003 formed material for prospective and cases admitted for 2 years prior for retrospective study. Whenever an unexplained cluster of disease/manifestation was perceived, case definition was finalized and cases recorded. Retrospective cluster analysis was done for some significant unexplained clusters detected prospectively. These clusters case rate were analyzed in relation to age, sex, climatic conditions, etiology of case, etc. to detect any correlation. Pearson correlation coefficient, chi-square test, centroid method and Z test of proportion were used for statistical significance. RESULTS Eight unexplained clusters were perceived in prospective study. 3 (hyperbilirubinemia in exchange range hypocalcemic seizures and vesicoureteric reflux) proved statistically not significant. 3 were small clusters of uncommon diseases / manifestations (biliary atresia, cardiomyopathy, and acral gangrene). Two large significant clusters, major non-traumatic bleeding manifestations (MNTBM) and acute renal failure (ARF) occurred to which retrospective study was extended. Tmax and Tmin (average maximum and minimum daily temperature) had statistically significant positive correlation with ARF (gamma = + 0.83, P <0.001 for Tmax and gamma =+ 0.56, P = 0.002 for Tmin) and negative correlation with MNTBM (gamma = - 0.34, P = 0.040 for Tmax and gamma = - 0.59, P <0.001 for Tmin). Barometric pressure has significant negative and positive correlation with MNTBM and ARF respectively (gamma =+ 0.57, P <0.001 for MNTBM and gamma = -0.45,P = 0.006 for ARF). The clusters also had significant positive correlations with female sex and age under 1 year (Z = 2.48, P <0.001,chi2=13.83, P<0.001 for sex and age of MNTBM and Z =3.11, P <0.001, chi2 =10.85, P <0.001 forage and sex of ARF cases respectively). Three small clusters and a small subgroup of MNTBM(subcutaneous nodules as manifestations of bleeding disorders) occurred predominantly under one year and different sexes were involved. CONCLUSIONS Several significant unexplained clustering were noted among hospital admissions. There was significant correlation with climatic conditions, age and sex. Larger, longer and multicentric studies in different geographical areas are required to investigate more plausible but complex biological phenomenon and associations related to diseases or manifestations. Cluster awareness has diagnostic and management implications for clinician as it also helps in early recognition of disease outbreak and dissemination of information and hospital staff to be prepared to handle increased number of cases and its treatment.