|
Article Excerpt
Epidemiological studies of the health effects of the use of alcohol have been widespread (see reviews in Corrao, Rubbiati, Bagnardi, Zambon & Poikolainen 2000; Holman, English, Milne & Winter 1996). They have often demonstrated protective effects from regular alcohol use, although there has been some reluctance to report such results from the perspective of responsible public health evidence. It is widely accepted that studies of clinical alcoholics report harms from drinking. Why do population studies tend to report only benefits?
There have been attempts to demonstrate the false nature of epidemiological studies that report protective effects. One important instance was the discovery that the inclusion of former drinkers in the reference group would bias upwards the harm associated with not drinking (the "sick quitters" hypothesis; Shaper, Wannamethee & Walker 1988). Therefore, most current studies exclude former drinkers, or at least interpret them separately. Other alternative hypotheses included that of social isolation as an artifactual mechanism (Skog 1996), which failed to survive empirical testing (Greenfield, Rehm & Rogers 2002; Murray, Rehm, Shaten & Connett 1999). A suggestion that personality might account for self-selection into protected and hazardous drinking categories (Lipton 1999) was similarly unconfirmed empirically (Murray, Barnes & Ekuma 2005). Further, there are analyses whose goal is to dispute the results of the majority of epidemiologic studies on alcohol (Fillmore, Kerr, Stockwell, Chikritzhs & Bostrom 2006).
On the other hand, much physiological evidence has accumulated to indicate that regular drinking is cardio-protective. The accumulated weight of studies of blood lipid effects (Suh, Shaten, Cutler & Kuller 1992), platelet aggregation (Rimm, Williams, Fosher, Criqui & Stamfer 1999), systemic inflammation (Imhof, Froelich, Brenner, Boeing, Pepys & Koenig 2001; Mukamal, Cushman, Mittleman, Tracy & Sisovick 2004) and so on make the protective effect of regular drinking plausible (see also Rehm, Room, Graham, Monteiro, Gmel & Sempos 2003 for a review).
Taken at face value, risk or hazard ratios (HRs) of usual daily drinking, with reference to non-drinking and with former drinkers set aside, are generally protective across the whole distribution of drinking. This finding applies particularly to the hazard of coronary heart disease (CHD) and also to all-cause mortality (and morbidity) and sometimes to other cardiovascular diseases (CVD). It applies to both men and women, although not equally.
The notion that alcohol consumption confers a health benefit has entered the public consciousness, at least in North America and Western Europe. I would argue that it was prevalent in the public consciousness even before we knew enough about the relationship to describe it with confidence. We probably still don't know enough about the subject of usual drinking to recommend it broadly to the public.
What is wrong with this picture? Clearly people suffer harm from drinking alcohol. But where is the harm reflected in population-level studies?
Part of the Answer
It turns out that if you assess the hazards associated with heavy episodic drinking (HED) in these studies (leaving aside the specific definition of HED for a moment), you can demonstrate a clear and statistically significant health hazard (Laatikainen, Maaninen, Poikolainen & Vartiainen 2003; Murray, Connett, Tyas, Bond, Ekuma et al. 2002). Another contribution on harm from drinking comes to us from the anthropological literature. It has been reported for decades now that a pattern of HED (sometimes called binge drinking) is associated with elevated levels of social and physical harms (Knupfer 1966). Binge drinking in this context refers to occasions when individuals consume elevated levels of alcohol. It is similar, but not identical, to drinking enough to feel drunk. There are relatively well-established survey questions that tap this behavior (Room 1990), and differences in the prevalence of HED have been demonstrated between cultures (Huijbregts, Feskens, Rasanen, Fidanza, Nissinen, Menotti & Kromhout 1997).
Until recently, designers of epidemiological surveys have not recognized a need to incorporate questions on the patterns of consumption of alcohol. There have been a few exceptions, each in studies with other limitations, and calls have been made for the collection of more suitable drinking data (Arria & Gossop 1998) and more explicit modeling of available data (Rehm et al. 2003). The design and conduct of more suitable epidemiologic studies could be an exercise that would consume some 15 years. In the meantime, investigators are struggling to adapt existing studies, using drinking variables that were designed before drinking pattern was regarded as an issue (Britton & Marmot 2004; McElduff & Dobson 1997; Mukamal, Conigrave, Mittleton, Camargo, Stamfer, Willet & Rimm 2003).
Definitions of heavy episodic drinking
It turns out that definitions of HED from the anthropological literature work fairly well in the context of...
|
