Use this URL to cite or link to this record in EThOS:
Title: Effect of household dynamics on risk of disease associated with household contact
Author: Freeman Chirwa, Tobias
ISNI:       0000 0001 3545 0405
Awarding Body: London School of Hygiene & Tropical Medicine
Current Institution: London School of Hygiene and Tropical Medicine (University of London)
Date of Award: 2001
Availability of Full Text:
Access from EThOS:
Access from Institution:
Household contact is known to be a risk factor for transmission of many infections, and the magnitude of such contact-associated risk is a classic measure of transmissibility. The risk associated with household contact may be underestimated due to misclassification of contact status of some individuals, particularly in cohort studies of diseases with long incubation period. Such studies generally begin with contact defined at a single point in time, thus a "snap-shot" of dynamic households. However, individuals change households, form new households, die, or migrate over time. Thus, some individuals who experienced household contact may be misclassified as non-contacts. Published analyses of Karonga Prevention Study (KPS) data have indicated that household contact with active paucibacillary or multibacillary leprosy as assessed during a survey carried out 1980-84 (LEP-1) imparted two- or fivefold increased risk of leprosy, respectively, compared to individuals not living in such households. This was as assessed in a second survey carried out 1986-89 (LEP-2). The current project began as an investigation of the implications of household dynamics on these measures of household contact associated risk, and evolved into a broad consideration of household dynamics, touching upon a variety of demographic and epidemiological issues. The approach included detailed analysis of KPS data and development of a simulation model of household dynamics tracing contact status over a period of time, to quantify contact status misclassification and estimate the "true" underlying rate ratios adjusting for this misclassification. Not even such a model captures all the selective household changes of a rural society, and there will be many unrecorded and unrecordable contacts. A total of 112886 individuals were interviewed in LEP-1, of whom about 85,000 were examined in LEP-2. 46% of this population was under 15 years of age. Procedures for smoothing the age distribution initially characterized by age heaping (a direct result of birth estimates) and for correcting for the under-ascertainment of infants, especially common in studies when reporting of birth dates is poor, are explained. The crude birth and death rate were estimated to be 49 births per 1000 persons and 10 deaths per 1000 persons respectively. Under-5 mortality was estimated to be about 250 deaths per 1000 live births. Estimates of mortality adjusted for age, sex and socio-economic factors show interesting patterns. Mortality was higher in north rather than south Karonga (rate ratio of 1.29, 95% Cl: 1.19, 1.38); and in those with estimated rather than precise years of birth (rate ratio of 1.14, 95% Cl: 1.03, 1.25). No significant differences in mortality were found between leprosy cases and noncases, 1.14 (95% Cl: 0.84, 1.54). The finding of significantly lower mortality among those with compared to those without a BCG scar, rate ratio of 0.70 (95% Cl: 0.64, 0.76) was surprising though we suspect that it reflects residual socio-economic confounding rather than a biological effect of the vaccine. The mean and median household size (6 42 and 5 respectively) were similar for the LEP-1 and LEP-2 surveys. 85% of heads of households were male with their mean age, 47 (s.d. 14.6) lower than that for females, 55 (s.d. 14.2). There was a very high rate of household change among 15-29 year olds with a higher rate for females in the lower part of this age band (approximately 63% over 5 years). A household dynamics model was constructed in order to simulate births, deaths, in- and out-migrations, marriages and movement between households on an annual basis. Its parameters are derived from the LEP and census data. The contact status misclassification rate is defined as the proportion of all individuals in contact with at least 1 index case in the simulations who were initially classified as non-contacts. The model results show high contact status misclassification in particular among the 15-29 year olds (largely a reflection of active household change). Improved estimates of contact associated leprosy risks showed higher rate ratios in young children than adults. Apart from attributing such results to intensity of contact and sharing of environmental factors with source cases, they are also consistent with (but do not confirm) genetic susceptibility. Apart from investigating household contact-associated risks of disease, the analysis of household dynamics in this thesis provides methods and baseline measures for understanding demographic and social pattern changes, of particular importance in this era of HIV.
Supervisor: Fine, P. Sponsor: WHO ; TDR
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
Keywords: Infection