Alcohol Consumption and Emphysema

Alcohol Consumption and Emphysema To the Editor: The provocatively entitled article (Chat 1984; 85:372-77) by Drs. Pratt and Vollmer, reports a significant positive association of centrilobular emphysema (CLE) with smoking (p =0.009) and a similar negative association of CLE with alcohol consumption (p =0.008). Their statistical analyses suggest that alcohol is as protective against CLE as smoking is causativel This study merits thoughtful reconsideration. First, their statistical analyses and interesting theory may be based on flawed historic data. One can derive from the data presented in their Tables 2 and 3 that they classified eight of the 65 heavy alcohol-using men as never-smokers (65 - 57 =eight). In my experience, heavy alcohol users are almost invariably current smokers or ex-smokers; therefore, I suspect the information obtained retrospectively from medical records may be incomplete or inaccurate. As my impressions are anecdotal, I offer a friendly challenge to the authors or their colleagues to prospectively Interview two or three score heavy alcohol users on the wards or clinics in North Carolina while I do the same in Southern California, to see if we can find a single adult heavy alcohol user who has never been a smoker. Second, if the published smoking classifications were correct, there would be 31 never-smokers in the series (204 x 0.152 =31):one of 16 non-alcohol users had CLE (73 x 0.452 - 57 x 0.561 =1); one of the six slight-to-moderate alcohol users had CLE (66 x 0.485- 59 x 0.525 =one); and three of the eight heavy alcohol users had CLE (65 x 0.354 - 57 x 0.351 =three). In these 31 neversmokers the incidence of emphysema increased from 6 percent to 17 percent to 38 percent as tbe exposure to alcohol increased from none to slight-to-moderate to heavy. If we accept the classifications given, the pattern in this group suggests a causative rather than beneficial effect of alcohol. Third, both alcohol abuse and tobacco abuse are serious problems. I fear the peruser of Cheat and those promoting the sale of alcoholic beverages will accept the title, 'The beneficial effect of alcohol consumption on the prevalence and extent of centrilobular emphysema" as a statement offaet rather than an interesting speculation.

James E. Hansen, M.D., F.C.C.~ Professor ofMedicine, Harbor-UCLA Medical Center, UCLA School ofMedicine, TOfTance, California

To the Editor: Dr. Hansens letter reveals some rather peculiar ideas about, first, the meaning of statistical probability and, second, the reliability of data based on small numbers of cases. First, he suggests that, since the probability of the smoking effect on CLE was .009 and of the alcohol effect was .008, then alcohol is "as protective against CLE as smoking is causative." Probability data are not supposed to be used that way.Those values, of course, only show that the likelihood that the associations seen may have occurred merely by chance was only nine (or eight) in a thousand. That is, if there really were no association and we did the entire experiment 1,000 times, only nine (or eight) times would there have been a difference as large or larger than we found. The degree of causativeness or protectiveness is shown in any of the tables other than the one Dr. Hansen used (1able 6). They show, on average, that smokers who drink "heavily" are about half as likely to have CLE as the smokers who don't drink. The smoking drinkers still have a much

804

higher severity of CLE than nonsmokers. Thus, it appears that alcohol is not as "protective" as smoking is "causative," In paragraph 2, Dr. Hansen questions our smoking background information. We acknowledged in the paper that patients may not have told the truth, but we only counted a case as a nonsmoker if the clinical chart clearly stated that the patient claimed to be a nonsmoker. If there were no statement either way, we did not use the case at all. Since the alcohol effects were detectable whether we used the whole population, or the confessed smokers alone, we don't think this objection affects our observation. Other analyses of the data show that the background information is sufficiently valid to detect statistically significant differences between pipe-cigar smokers and the rest (see ref 28). In paragraph 3, Dr. Hansen does some calculations concerning the prevalence of CLE in our nonsmoking population and shows that 6 percent of the nondrinkers, and 38 percent of the heavy drinkers had CLE. His calculations are arithmetically correct. Then he suggests that drinking is causing CLE in the nonsmokers. Here he overlooks the small numbers of cases involved, 16 nondrinkers and eight heavy drinkers I Statistical analysis of these proportions shows that, although the difference in prevalence is 32 percent (38-6), the probability that this could have occurred by chance is well over o. OS. 10addition, the average measured extent of the emphysema was only 2 percent in the nonsmoker population. Thus, none of the nonsmokers who were "heavy" drinkers could have had a clinically significant amount of involvement even though the prevalence happened to be high. If one were willing to assume that all of the "heavy" drinkers were smokers, then the beneficial effect of alcohol would be even more striking. Finally, Dr. Hansen worries about the social effects of our observation, which he prefers to call a "speculation:' We used our last paragraph to state our view regarding how the information should be used. We emphasized that alcohol use was not being recommended for prophylaxis of emphysema. However, we did not think that our statistically valid observation should be suppressed simply because it might be misused. Philip C. Pratt, M.D., F.C.C.P.

Professor ofPathology, Dulce University MedicalCenter; Durham

Re-expansion of Refractory Atelectasis To the Editor: We enjoyed reading the article entitled "Be-expansion of Refractory Atelectasis Using a Bronchofiberscope with a Balloon Cuff" by Harada et al (Chest 1983; 84:725-28). We have been using this technique routinely for over six years and still find it safe and effective in treating refractory atelectasis. We also found that some patients may require a second treatment when the atelectasis recurs.'

J. Eugene Millen,

M.D.; and Frederick L. Glauser, M.D., F.C.C.~ Medical College ofVirginia and Veterans Adminlstration Medical Center, Richmond

REFERENCE 1 Millen JE, Vandree J, Glauser FL. Fiberoptic bronchoscopic balloon occlusion and re-expansion of refractory unilateral atelectasis. Crit Care Med 1978; 6:50

Communications to the EdItor