“…Of the sixteeen LDC studies, 10 are case-control designs (two mortality studies), 5 cohort studies (all morbidity), and one a case-fatality study. In contrast to the relatively robust definitions of ALRI, the measurement of exposure has relied in almost all studies on proxies, including the type of fuel and stove (Cerqueiro et al, 1990;Collings et al, 1990;Johnson & Aderele, 1992;Mtango et al, 1992;Victora et al, 1994;Shah et al, 1994;O'Dempsey et al, 1996, Wesley & Loening, 1996Lopez-Bravo et al, 1997), whether the child stays in the smoke during cooking (Kossove, 1982, Mtango et al, 1992, Victora et al, 1994, reported hours spent near stove (Kossove, 1982;Pandey et al, 1989), and whether the child is carried on the mother's back during cooking (Campbell et al, 1989;Armstrong & Campbell, 1991;de Francisco et al, 1993). Apart from Ezzati's study (see below), only one previous study made direct measurements of pollution (particulates) and exposure (COHb) in a subsample (Collings et al, 1990).…”