Rosenberg L, Slone D, Shapiro S, Kaufman D W, Miettinen O S
Am J Epidemiol. 1981 Jun;113(6):646-52. doi: 10.1093/oxfordjournals.aje.a113143.
In the course of evaluating the acute effect of caffeine-containing coffee on the risk of myocardial infarction (MI) among women 30-49 years of age, this study noted appreciable differences in coffee consumption among hospital patients admitted for illnesses not known to be related to coffee use. Among these potential controls, the coffee consumption of patients who had been compelled to hospital by conditions having an acute onset, such as fractures ("acute condition controls") was compared with that of patients admitted for chronic disorders ("chronic condition controls"). The proportion of drinkers of caffeine-containing coffee among 980 acute condition controls (65%) was significantly greater than that among 5835 chronic condition controls (58%) (p less than 0.01); by contrast, the proportion of drinkers of decaffeinated coffee was significantly greater among chronic condition controls (22%) than among acute condition controls (18%) (p less than 0.01). With regard to caffeine-containing coffee, other findings were that the frequency of drinking decreased with increasing contact with medical care in the year before admission; that patients who changed their consumption after discharge tended to give up coffee; and that the proportion of drinkers was greater among women in the community than among the hospital patients. These results suggest that relatively long-standing illness influences against drinking caffeine-containing coffee among women 30-49 years of age. They also suggest that there is an opportunity for overestimation of the relative risk of MI among coffee drinkers in hospital-based studies that use chronic condition controls.
