CONTENTS Part I. Peptic Ulcer Disease...........0....0 2000000 cee Lene eee 429 Introduction 20... cece eee e cect e cee nee cee, 429 Impact of Smoking and Smoking Cessation on Ulcer Occurrence ........... 429 Smoking and Gastrointestinal Physiology ...........0..0 0c cccuceeuee 429 Trends in Peptic Ulcer Disease 2.02.00. cc cece ee 430 Morbidity From Peptic Ulcers 200... cece eee e a eee 431 Mortality From Peptic Ulcers... 0.0 c ccc ceceees 432 Effects of Smoking on Ulcer Healing and Recurrence .................... 432 Healing of Duodenal Ulcers 2.000. ccc cece eee, 432 Recurrence of Duodenal Ulcers 2.00.00. 000.00 c cece ccc cee ceuee 433 Healing of Gastric Uleers 2... 06. ec ec ce eee ceee ees 440 Recurrence of Gastric Ulcers 2.0.0.0... ccc cece reece eee. 440 Summary 6... ccc cece eee n tense eeeterieees 44] Part II. Osteoporosis and Skin Wrinkling ..... ve ceees Soe eee eeees ress, 443 Osteoporosis . 0.0 e tbe eee e een eevee eeeeee, 443 Introduction 2.0... ccc cece cece eeeeeteeeees 443 Pathophysiologic Framework ........00.000 00 0c cece eee eeeecueecees 443 Bone Mineral Content in Smokers Compared With Nonsmokers ......... 444 Smoking as a Risk Factor for Osteoporotic Fractures ................... 449 Smoking Cessation and Osteoporosis and Fracture ..................-. 453 Summary... 0.0... e ete ne cence ecceeees 453 Skin Wrinkling 2.6.00 e eet ev eceeeyeees 453 Introduction 2.0.0... ec ec eet e eter t ee eneeeees 453 Pathophysiologic Framework .....00.00 0.000. c cece cece ccccccccueee 453 Smoking and Skin Wrinkling .......000 000.0. c cc cece cece cc eeee, 455 Smoking Cessation and Skin Wrinkling ....................0.0.-.0-. 456 Summary 2.6.0... e tenet te eteneeeeees 457 Conclusions ............... bet etetett eee eenes fete ette eee e teens 457 References 2.0.0... c cere cect eect veeeereeres 459 427 PART I. PEPTIC ULCER DISEASE Introduction Numerous studies have demonstrated the association between smoking and the occurrence of peptic ulcer disease. This association was noted in the 1964, 1971, and 1972 Surgeon General's Reports (US PHS 1964; US DHEW 1971, 1972). The 1979 Report stated that the evidence of an association between cigarette smoking and peptic ulcer was strong enough to suggest a causal relationship (US DHEW 1979). That Report concluded that cigarette smoking was associated with the incidence of peptic ulcer disease and with increased risk of dying from peptic ulcer disease; the evidence that smoking retards healing of peptic ulcers was regarded as highly suggestive. The 1989 Report (US DHHS 1989) stated that smoking cessation may reduce peptic ulcer incidence and is an important component of peptic ulcer treatment, even with the effective drug therapy presently available. This Section focuses on smoking cessation and the occurrence and course of peptic ulcer disease. Impact of Smoking and Smoking Cessation on Ulcer Occurrence Smoking and Gastrointestinal Physiology Kikendali, Evaul, and Johnson (1984) reviewed the effect of cigarette smoking on aspects of gastrointestinal physiology relevant to peptic ulcer disease. The literature available at the time of their review supported the following concepts. Chronic cigarette smokers have higher maximal acid output than nonsmokers. Smoking | cigarette or more has no consistent immediate effect on acid secretion. Smoking 1 cigarette immediately decreases alkaline pancreatic secretion and immediately results in a pronounced fall in duodenal bulb pH, especially in subjects with gastric acid hyper- secretion. Smoking has a variable effect on gastric emptying, depending on experimen- tal design. Smoking increases duodenogastric reflux. Smoking decreases gastric mucosal blood flow. Smoking during waking hours inhibits the antisecretory effects of a nocturnal dose of cimetidine, ranitidine, or poldine. Subsequent to this review, the-two latter concepts have been seriously challenged. Robert, Leung, and Guth (1986) found that neither nicotine nor smoking inhibited basal gastric mucosal blood flow in rats. Several investigators could not confirm that smoking antagonized the antisecretory effect of cimetidine or ranitidine (Deakin, Ramage, Williams 1988; Bianchi Porro et al. 1983; Bauerfeind et al. 1987). However, several of the findings from this earlier review (Kikendall. Evaul, Johnson 1984) have been confirmed by more recent reports. Parente and associates (1985) confirmed higher pentagastrin-stimulated acid secretion among chronic heavy smokers than among nonsmokers. Smokers also had higher basal serum pepsinogen-I levels. These differences were statistically significant and large enough to be of clinical importance. Higher maximal gastric acid secretory rates among smokers compared 429 with nonsmokers were also demonstrated by Whitfield and Hobsley (1985) in a study of 201 patients with duodenal ulcer. Additionally, Mueller-Lissner (1986) noted that chronic smokers who abstained from smoking for 12 hours had more duodenogastric bile reflux than nonsmokers and confirmed that smoking cigarettes acutely augments the already elevated rate of bile reflux. Quimby and coworkers (1986) reported that active smoking transiently decreased gastric mucosal prostaglandin synthesis. In summary. the known effects of smoking on gastroduodenal physiology provide multiple potential mechanisms for enhancement of an ulcer diathesis by active smoking. Several of the effects of smoking, most notably the inhibition of alkaline pancreatic secretion, the reduction of duodenal bulb pH, and the reduction of prostaglandin synthesis, are transient effects that could be reversed quickly by abstinence from smoking. Trends in Peptic Ulcer Disease During the past several decades. the rates of hospitalization for and mortality from peptic ulcer disease in the United States have declined dramatically (Kurata et al. 1983). Although changes in coding practices and/or diagnostic procedures could explain some of the decline, the trends in mortality from peptic ulcer have paralleled the decreasing prevalence of smoking. Kurata and coworkers (1986) studied trends in ulcer mortality and smoking in the United States between 1920 and 1980 and estimated that the portion of duodenal-ulcer-related mortality attributable to smoking was between 43 and 63 percent for men and 25 and 50 percent for women. In contrast, Sonnenberg (1986) concluded that smoking was not the main determinant of the birth cohort phenomenon of declining peptic ulcer mortality in the United Kingdom. This study descriptively compared the death rates for duodenal and gastric ulcer with the annual cigarette consumption in the United Kingdom according to birth cohorts and found a lack of correlation between ulcer mortality and cigarette consumption (Sonnenberg 1986). Thus, factors in addition to cigarette smoking may also underlie the recent trends in these indicators of peptic ulcer disease. Two factors that have received considerable attention in recent years are Helicobacter pylori gastritis (Graham 1989) and the use of nonsteroidal anti-inflammatory drugs (Griffin, Ray, Schaffner 1988). Martin and associates (1989), in an endoscopic study. found that smoking was a risk factor for peptic ulcer disease among patients who had Helicobacter pylori gastritis. Willoughby and colleagues (1986) found that smoking was associated with peptic ulcer disease among subjects with rheumatoid arthritis. most of whom were taking nonsteroidal anti-inflammatory drugs. Ehsanullah and colleagues (1988) and Yeomans and associates (1988) also showed an association of smoking with the acute gastric erosions and submucosal hemorrhages induced by these drugs. These studies demonstrated that smoking 1s associated with ulcer disease related to both Helicobacter pylori and nonsteroidal anti-inflammatory drugs. 430 Morbidity From Peptic Ulcers In an analysis of prospective cohort data on ulcer incidence in women from the National Health and Nutrition Examination Survey | Epidemiologic Followup Study. the relative risk for developing peptic ulcer was 1.3 among former smokers (95-percent confidence interval (CI), 0.7—2.9) and 1.9 among current smokers (95-percent CI. 1.2—2.6) compared with lifetime nonsmokers (Anda et al. 1990). In this study. former smokers were defined as persons who had smoked at least 100 cigarettes in their lifetime but who were not smoking at the time of the baseline interview. The mean length of followup in this cohort was 9 years. This analysis used the Cox proportional hazards mode! to adjust for the potential confounding effects of age. sex, socioeconomic status, regular aspirin use, alcohol intake, and coffee consumption. Ainley and associates (1986) surveyed the smoking behavior of 1.217 patients undergoing endoscopy. This study did not include “normal” or community controls as all patients had indications for endoscopy. Of the smokers, 11.9 percent had gastric ulcers. a diagnosis shared by 7.7 percent of ex-smokers (p<0.025) and 4.6 percent of never smokers (p<0.001). Of the smokers. 12.8 percent had duodenal ulcer compared with 6.8 percent of ex-smokers (p<0.01) and 6.! percent of never smokers (p<0.001). Ina study of nearly 6.000 Japanese men living in Hawaii (Stemmermann et al. 1989), 243 developed gastric ulcers and 99 developed duodenal ulcers in 20 years of followup. Gastric ulcer developed among 6.7 percent of current smokers compared with 3.8 percent of former smokers and 3.2 percent of lifetime nonsmokers (p<0.0001). Duodenal ulcer developed more often (p<0.0001) among current smokers than among former smokers or never smokers (2.7 vs. 1.4 vs. 0.9 percent. respectively). These three studies show that smokers are more likely than never smokers and former smokers to develop peptic ulcer disease. Two of the studies show higher frequencies among smokers for both duodenal and gastric ulcer. All three studies demonstrate that the risk of peptic ulcer for former smokers is between that for current smokers and for never smokers. The tendency of symptomatic smokers to stop smoking would bias the results of such studies toward reducing the apparent benefit of cessation (Chapter 2). These studies strongly suggest that the smoker’s risk of developing either gastric or duodenal ulcer is diminished after smoking cessation. In an early analysis of cross-sectional survey data among men aged 20 to 79 in Tecumseh, MI (Higgins and Kjelsberg 1967), the age-adjusted prevalences of self- reported peptic ulcer among nonsmokers (presumably never smokers), ex-smokers, and current smokers were 5.2, 8.0, and 7.1 percent. respectively. The definitions of smoking status were not presented, and the differences were not statistically significant. In this study, the prevalences of peptic ulcer among women who were nonsmokers, ex- smokers. or current smokers were 1.4, 1.5, and 2.8 percent. respectively: these differen- ces were reported as statistically significant between smokers and nonsmokers (Higgins and Kjelsberg 1967). Earlier studies such as this, which were conducted before the advent of endoscopy, had relatively poor diagnostic accuracy and may consequently have been biased toward underestimating the effects of smoking. Additional reports linked smoking to some of the complications of peptic ulcer disease. For example. 86 percent of 128 patients presenting with perforated duodenal 431 ulcer were cigarette smokers compared with 51 percent (p<0.01) of retrospectively matched controls (Smedley et al. 1988). Other reports noted that smokers comprised 87 percent (Heuman, Larsson, Norrby 1983) and 86 percent (Hodnett et al. 1989) of patients with perforated duodenal ulcers and 83 percent of males undergoing surgery for peptic ulcer (Ross et al. 1982). These latter studies were uncontrolled, and the high percentages of smokers have not been confirmed in some other surgical series. Never- theless, these latter studies support the findings of Smedley and associates (1988) and suggest that smokers with peptic ulcer who continue to smoke may be at greater risk for ulcer complications than nonsmokers. Mortality From Peptic Ulcers The American Cancer Society Cancer Prevention Study I (ACS CPS-I) found that the relative risk of mortality for peptic ulcer among men was 3.1 for current smokers (95-percent CI, 2.24.2) and 1.5 for former smokers (95-percent CI, 1.0-2.3) compared with lifetime nonsmokers (US DHHS 1989). In the U.S. Veterans Study, the duodenal ulcer mortality ratios for current and ex-smokers compared with never smokers were 3.2 and 1.8, respectively (Kahn 1966). Ex-smokers in this report were persons who stopped smoking for reasons other than physician’s orders but were otherwise not clearly defined. The mortality ratios for gastric ulcer among current and ex-smokers were 4.1 and 3.4. respectively. Although these differences in mortality were not statistically significant, the trends were similar to those in ACS CPS-I and supported the results of that study. Effects of Smoking on Ulcer Healing and Recurrence Healing of Duodenal Ulcers Numerous trials evaluating ulcer therapy have suggested that smoking adversely affects ulcer healing. Kikendall, Evaul, and Johnson (1984) reviewed the results of 18 studies that assessed the impact of smoking on healing of duodenal ulcers. In most of these studies, the percentage of healed ulcers was lower among current smokers than among nonsmokers (Table |). These studies were not explicitly designed to study smoking, and the nonsmoking category presumably included never as well as former smokers. When the data from these studies were subjected to meta-analysis, the percentage of healed ulcers was lower among smokers than among nonsmokers in patients treated with H2-blockers (p<0.0001) and in patients given placebo (p<0.0001) (Table 2). The median difference in percentage of subjects completely healed was 22 percentage points in favor of nonsmokers in groups treated with H2-blockers, 21.5 percentage points in groups receiving other active therapy, and 22 percentage points in groups receiving placebo. The data for groups receiving active therapy other than H2-blockers were not subjected to statistical analysis because the data were not homogeneous, but the data in Table 1 show that nonsmokers in most of these other treatment groups fared better than their smoking peers. Most trials published since this 432 1984 review show similar trends toward greater likelihood of healing of duodenal ulcers in nonsmokers. Recently, several reports have suggested that sucralfate (Lam et al. 1987) and misoprostol (Lam et al. 1986) may have particular value in treating duodenal ulcers among patients who smoke. Lam (1989) has compiled a list of six studies showing comparable duodenal ulcer healing rates for smokers and nonsmokers treated with sucralfate. Although a few studies offer contrary data (Van Deventer, Schneidman, Walsh 1985; Martin 1989), much of the evidence suggests that sucralfate heals duodenal ulcers in smokers and nonsmokers at comparable rates. The claim that the efficacy of prostaglandins for duodenal ulcer healing is unaffected by smoking is based on the results of a single study (Lam et al. 1986). The design of this study is unusual because patients who smoked were encouraged to abstain from smoking during the study; therefore, healing efficacy in smokers may have been due to the combined effects of misoprostol and smoking cessation. Other duodenal ulcer treatment trials (Bianchi Porro and Parente 1988: Brand et al. 1985; Nicholson 1985) showed improved healing among nonsmokers. Nicholson ( 1985) treated duodenal ulcer patients with 200 t1g misoprostol 4 times daily and documented healing in 73 of 138 smokers (53 percent) and 66 of 93 nonsmokers (71 percent, p<0.01). Thus, the evidence is tenuous at best that oral prostaglandins can overcome the adverse effects of smoking on the healing of duodenal ulcers. Other recently reported clinical trials are not systematically reviewed in this Chapter. Most of the recent trials that have analyzed the effects of smoking on duodenal ulcer healing show lower healing rates among smokers than among nonsmokers. In contrast to the numerous comparisons of duodenal ulcer healing rates among smokers and nonsmokers, only one study has examined specifically the effect of smoking cessation on duodenal ulcer healing (Hull and Beale 1985). In this study, 70 male smokers with duodenal ulcers were advised to stop smoking and were treated with cimetidine for 3 months. Those who stopped were no more likely than those who continued smoking to have healed their ulcers on endoscopic exam at 3 months (75 vs. 81 percent, respectively, not significant). Cimetidine treatment was then stopped. Three months later, 72 percent of those who quit smoking and 39 percent of smokers were ulcer-free at repeat endoscopy (p<0.05) (Hull and Beale 1985). Although these results require confirmation, the findings suggest either that some of the adverse effects of smoking on duodenal ulcer disease may persist for a few weeks after cessation of smoking or that cimetidine therapy may mitigate these effects. Recurrence of Duodenal Ulcers A number of prospective clinical trials of maintenance therapy for duodenal ulcer have assessed the impact of smoking on ulcer recurrence. In one of the larger trials (Sontag et al. 1984), 370 subjects with previously documented duodenal ulcer, who had no active ulcer at enrollment endoscopy. were randomized to placebo or cimetidine. Endoscopy was repeated at 6 and 12 months or whenever dyspepsia occurred during the 12 months of followup. In the placebo group. smokers were more likely than nonsmokers to experience recurrence (72 vs. 21 percent. p-blocker therapy Bianchi Porro et al. (1981) H2-blockers 4 76 66 36 86 <0.05 20 Korman et al. (1983) Ho-blockers 4-6 71 63 64 95 <0.01 32 Korman, Hansky et al. (1982) Ranitidine 4 13 62 12 100 <0.05 38 Hetzel et al. (1978) Cimetidine 6 43 36 43 80 NS -6 Korman et al. (1981) Cimetidine 6 10 50 1s 100 <0.05 50 Marks et al. (1980) Cimetidine 6 19 78 10 60 NS 18 Bardhan et al. (1979) Cimetidine 4 94 65 40 65 NS 0 Gugler et al. (1982) Cimetidine 8 34 64 16 94 <0.05 29 Gugler et al. (1982) Oxmetidine 8 35 71 14 93 NS 22 Korman, Hetzel et al. (1982) Oxmetidine 4 27 70 1S 87 NS \7 Korman, Hetzel et al. (1982) Cimetidine 4 28 68 13 92 <0.05 a4 TABLE 1.—Continued Patients with healed ulcers Sep Duration of Rx Smokers Nonsmokers Difference in Reference Drug (wk) Na oP N? ol p-value % healed Active therapy other than H2-blockers Bianchi Porro et al. (1980) Cimetidine or pirenzepine 4 63 71 27 81 NS 10 Sonnenberg et al. (1981) Cimetidine, pirenzepine, or 4 66 54 68 73 <0.05 19 placebo Barbara et al. (1979) Pirenzepine 4 16 69 28 43 NS -26 Vantrappen et al. (1982) Arbaprostil 4 68 65 14 79 NS 14 Peterson et al. (1977) Antacid 4 28 75 8 88 NS 13 Kormanet al. (1981) Antacid 6 13 39 12 67 <0.05 28 Marks et al. (1980) Sucralfate 6 20 90 9 67 NS -23 Nagy (1978) Carbenoxolone _ 11 55 10 80 NS 25 Young and St. John (1982) Carbenoxolone 6 14 50 6 83 NS 33 Lam et al. (1979) Antacid + sulpiride _— 17 59 34 91 <0.05 32 Lam et al. (1979) Placebo or sulpiride — IS 27 35 Sl NS 24 Massarrat and Eisenmann Antacid 8 56 48 24 715 <0.05 27 (1981) OCP TABLE 1.—Continued Patients with healed ulcers Duration of Rx Smokers Nonsmokers Difference in Reference Drug (wk) Ni Ge nN Al p-value & healed Placebo therapy Bianchi Porro et al. (1980) Placebo 4 55 3t 1S 53 NS 22 Nagy (1978) Placebo _ tl 25 1 30 NS 5 Young and St. John (1982) Placebo 6 Is 20 5 40 NS 20 Hetzel et al. (1978) Placebo 4 42 37 42 42 NS 5 Peterson et al. (1977) Placebo 4 25 32 13 69 <0.03 37 Vantrappen et al. (1982) Placebo 4 65 28 26 65 <0.05 37 Barbara et al. (1979) Placebo 4 25 28 10 50 NS 22 Korman, Hansky et al. (1982) Placebo 4 14 0 I 36 <0.05 36 Bianchi Porro et al. (1981) Placebo 4 62 24 20 50 <0.01 26 Bardhan et al. (1979) Placebo 4 33 24 13 38 NS 14 NOTE: NS=not statistically significant. “Natotal followed in smoking category. fh ; : . : : : 4 =percentage of total who experienced healed ulcers within specified time: p-values calculated by chi-square when not provided in paper. SOURCE: Kikendall, Evaul, Johnson (1984). TABLE 2.—Results of statistical analysis of pooled data from Table 1 Percentage healed Smokers Nonsmokers Test statistic N* %? N@ ge Z p-value All patient groups H2-blockers 449 70 278 90 7.1 <0.0001 Placebo 347 28 166 49 4.6 <0.0001 Subset of large patient groups H2-blockers 284 70 183 89 5.3 <0.0001 Placebo 149 29 88 Si 3.4 <0).0012 “N=total followed in smoking category. mo =percentage of total who experienced healed ulcers within specified time. SOURCE: Kikendall. Evaul. Johnson (1984). smokers receiving cimetidine were as likely to experience recurrence as nonsmokers receiving placebo, leading the authors to conclude that for smokers, quitting smoking may be more important in the prevention of ulcer recurrence than receiving cimetidine treatment (Sontag et al. 1984). Table 3 displays the results of similar prospective, controlled trials of the recurrence of duodenal ulcer identified in a literature search performed in March 1990. Trials or treatment groups with fewer than 12 smokers or 12 nonsmokers and reports that did not provide the raw data relative to smoking were omitted. Smokers had more recurrences than nonsmokers in every trial or every treatment group, regardless of the treatment (even surgery) and prophylactic therapy used to achieve healing. The difference was statistically significant in about half of the studies. The only study of larger size that failed to show even a nonsignificant advantage for nonsmokers was an Australian community-based study, not included in Table 3 because the requisite raw data were not published (Nasiry etal. 1987). This study differed from most of those listed in Table 3 in several ways, including larger numbers of exclusions. 41-percent withdrawals, primary reliance on symptoms rather than endoscopy to document recurrences, and lack of systematic effort to control the use of medications that may affect ulcer recurrence. Factors such as these may explain the disparate results. One trial listed in Table 3 found that incremental increases of cigarette consumption were significantly associated with greater risk of duodenal ulcer recurrence (Korman et al. 1983). Massarrat, Miiller, and Schmitz-Moormann (1988) and Piper, McIntosh and Hudson (1985) also found that the number of cigarettes smoked per day was a significant predictor for ulcer recurrence. Although these studies were designed to assess risk factors for recurrence of duodenal ulcer, the latter two studies are not listed in Table 3 because one did not present the necessary raw data (Massarrat. Miiller. Schmitz-Moormann 1988) and the other (Piper, McIntosh, Hudson 1985) had a study design that differed from that of the studies listed in Table 3. 8tr TABLE 3.—Recurrences of duodenal ulcer in smokers and nonsmokers in clinical trials Smokers Nonsmokers Followup Reference Prophylaxis (mo) N° a N° oP p-value Sontag et al. (1984) Cimetidine [2 186 34 114 18 <0.01 Bianchi Porro et al. (1982) Cimetidine 12 66 59 40 42 NS‘ Lauritsen et al. (1987) Ranitidine 12 48 33 21 19 NS Gibinski et al. (1984) Ranitidine 12 62 45 123 11 <0.005 Cerulli et al. (1987) Nizatidine 3 139 17 118 4 0.001 Brunner (1988) Roxatidine acetate 6 48 Ag 4i 20 <0.01 Lauritsen et al. (1987) Enprostil 12 52 65 I4 50 NS Sonnenberg et al. (1981) Various 12 33 52 33 33 NS Battaglia et al. (1984) Various 12 46 30 24 21 NS Paakkonen et al. (1989) Sucraltate 12 13 69 19 47 NS Bynum and Koch (1989) Sucralfate 4 58 45 64 39 NS Classen et al. (1983) Sucraltate 6 37 25 51 18 NS Graffner and Lindell (1988) Parietal cell vagotomy 60-168 190 24 116 7 <0.01 Rydning et al. (1982) Diet 6 55 69 18 39 <0.05 Sontag et al. (1984) Placebo 12 39 72 31 21 <0.001 Bolin et al. (1987) Placebo 12 13 85 13 77 NS Marks et al. (1989) Placebo 12 2] 95 12 67 <0.05 Paakkonen et al. (1989) Placebo 12 16 88 24 67 NS Bynum and Koch (1989) Placebo 4 50 si 67 30 <0.01 Classen et al. (1983) Placebo 6 39 62 45 4t NS Cerulli et al. (1987) Placebo 3 146 37 110 25 0.05 TABLE 3.—Continued Recurrences Smokers Nonsmokers Followup 7 7s ; Reference Prophylaxis (mo) N’ & N" & p-value Hallerback et al. (1987)" None 12 Hil 80 147 58 <0.001 Korman et al. (1983) None 12 45 84 60 S3 <0.01 Lam et al. (1987) None 24 60 100 178 14 <0.05 Lee, Samloff. Hardman (1985) None 4 58 69 49 45 <0.05 Koelz and Halter (1989) None 12 25 64 28 S0 NS NOTE: NS=not statistically significant. “Netotal followed in smoking category. “G=percentage of total who experienced recurrence within the specified time: p-values calcul “p5 yr and PCA, women aged 60-69 yr: smokers had lower PCA than nonsmokers" BM was significantly less in heavy smokers (220 cig/day) compared to never smokers in each age, sex strata” Mean BM compared to normal nondiabetics: Smoker <11 cig/day 9.3% less 11-15 cip/day 10.1% less >15 cig/day 12.7% less Nonsmoker 5.4% less Mean BM in smokers significantly less than mean BM in nonsmokers Stratifying by age and menopausal status, no difference in BM between smokers and nonsmokers No association of smoking and BM Comments Using the 40-49- year-old women as baseline, the 60-69- year-old smokers lost more PCA/yr since menopause than nonsmokers, but this finding was statistically significant only among nonobese women, no control for confounding Controlled tor age and sex only All subjects were diabetic, and findings may not generalize to all smokers; no control for confounders Controlled for age. race, sex, Menopausal status No control tor confounders OtP TABLE 5.—Continued Reference Meilstrém et al. (1982) Lindquist (1982) Sparrow et al. (1982) Rundgren and Mellstr6m (1984) Suominen et al, (1984) Johnell and Nilsson (1984) Population 357 men in a population-based study in Sweden 1.462 women ina population-based study in Sweden 341 men aged 40-80 followed for 3-5 yr 409 men and 559 women born in 1901-02 or 1906-07 from a population-based study in Sweden 142 men aged 3t-75 395 49-yr-old white women randomly selected from participants in a population-based study in Sweden Bone measurement Findings BM by DPA at heel BM by DPA at 3rd lumbar vertebrae PCA x ray of right 2nd metacarpal performed at baseline and 3-5 yr later BM by DPA at heel BM by yray attenuation in the calcaneums BM by y absorptiometry at the radius | cm and 6 cm proximal to the ulnar styloid BM lower in smokers vs. nonsmokers" Stratifying by age and menopausal status, no difference in BM between smokers and nonsmokers At baseline, no difference between PCA in smokers and nonsmokers: over the 3~5-yr period, smokers fost more PCA than nonsmokers" (B=-0.148, p=0.03) BM in women was 15-30% lower in smokers vs. nonsmokers* and in men 10-20% lower in smokers vs. nonsmokers”; no difference between ex-smokers and smokers BM in smokers lower than that in nonsmokers, but not statistically significant No association of smoking and BM in univariate or multivariate analysis Comments No control for confounders Controlled for age, race. sex, Menopausal status; dita may include that reported in Lindquist (1981) Controlled only for age Controlled for age, race. sex, weight, but not for menopausal status or estrogen use Multiple tests performed: controled for age only Controlled for age, race (white), sex, height, weight, age at menarche, menopausal status, number of children breast feeding. oral contraceptive use. physical activity, and calcium intake Ltr TABLE 5.—Continued Reference Jensen, Christiansen, Rodbro (FY8S) Sowers, Wallace. Lemke (1985) Cauley etal. (1986) Slemenda et al. (1987) McDermott and Witte (L988) Population 136 postmenopausal women volunteers from Sweden randomly assigned to different estrogen doses and followed for {yr 86 women volunteers from 2 rural communities in lowa 78 white postmenopausal women not on estrogen therapy 84 peri- and postmenopausal women evaluated every 4 mo tor 3 yri none on estrogen therapy 35 smokers (21 ppd for 214 yr and currently smoking) 35 nonsmokers (never smoked): matched for age, sex, weight. height, calcium intake, menopausal status, and estrogen use Bone measurement BM by SPA at distal radius performed at baseline and after ! yr of estrogen treatment BM by SPA at distal radius BM CT scan of the dominant radius at 30% of distance from wrist to elbow BM SPA at midshaft and distal radius BM SPA of midradius Findings At baseline, no difference in BM between smokers (smoked in prior 6 mo) and nonsmokers (no smoking in prior 6 mo); in 28 smokers treated with high doses estrogen, the mean % increase in BM was less than the mean & increase in 28 treated nonsmokers” No association of smoking and BM No association of smoking and BM in univariate analysis No association of smoking and BM overall in peri- and postmenopausal groups No association of smoking and BM Comments No control for confounders Small study with poor power: subjects were young, limiting generalizability: no contro! for confounders Small study with poor power: no contol of confounders Small study with poor power; no control of confounders Authors state that power to detect a 5% difference between groups at = 0.05 was >80% in both men and women, confounding controlled by matching 8hr TABLE 5.—Continued Reference Aloia et al. (198%) Picard et al. (1988) Bilbrey, Weix, Kaplan (1988) Stevenson et al. (1989) Slemenda et al. (1989) Population 26 menstruating white women volunteers 183 healthy French-Canadian women aged 40-50 1,069 women referred for osteoporosis screening to 18 centers in Tt States 284 healthy women (112 premenopausal, 172 postmenopausal) volunteers aged 21 68 84 peri- and postmenopausal women Bone measurement BM SPA of the radius and DPA of the spine BM by DPA of 2nd—4th lumbar vertebrae and by SPA of the distal radius BM by SPA of distal and midradius BM by DPA of temoral neck, Wards triangle. trochanteric region and 2nd—4th lumbar vertebrae BM by SPA of distal and midradius, DPA of lumbar spine Findings Smoking was associated with lower BM in the radius (p<0.01) and of the spine (p<0.03) No association of smoking with either BM of the lumbar vertebrae or distal radius No association of smoking with BM of radius In premenopausal women, correlation of ppd smoked and BM of vertebrae=—0).24"; no association at other sites; in postmenopausal women, no association of ppd smoked and BM at any site Significantly low BM in heavy smokers compared with nonsmokers, no difference in rates of change in BM between smokers and nonsmokers Comments Controlled for physical activity and height only No controt for confounders Controlled for menopausal status only Controlled for menopausal status (by design) and adjusted tor age and body mass index NOTE: PCA=percent cortical area: BM=bone mass: SPA=single photon absorptiometry, DPA=dual photon absorptiometry. ppd= pack s/duy. “p<0.05, Eleven other published studies reported no association between smoking and bone mineral content (Bilbrey, Weix, Kaplan 1988; Cauley et al. 1986: Johneill and Nilsson 1984; Lindergard 1981; Lindquist 1982; Lindquist et al. 1981: McDermott and Witte 1988; Picard et al. 1988; Slemenda et al. 1987: Sowers. Wallace, Lemke 1985; Stevenson et al. 1989). In addition, one study that found differences in bone mass between heavy smokers and nonsmokers reported no differences in longitudinally measured rates of bone loss (Slemenda et al. 1989). Some of these studies were small. and the findings of no association may be due to type II statistical errors, that is. the failure to find a true association (Cauley et al. 1986: Slemenda et al. 1987: Sowers, Wallace. Lemke 1985); other studies were large and had excellent statistical power (Bilbrey, Weix, Kaplan 1988; Johnell and Nilsson 1984: Lindquist 1982: McDermott and Witte 1988). One study evaluated the effect of smoking on bone mass among women taking estrogen (Jensen, Christiansen. Rodbro 1985). Among 56 postmenopausal women who underwent replacement therapy with high doses of estrogen for | year. the mean percentage increase in bone mass of the distal radius was 1.01 in 28 smokers compared with 2.58 in nonsmokers. This difference was statistically significant. Smoking as a Risk Factor for Osteoporotic Fractures Daniell (1976) reported that 76 percent of women with osteoporotic vertebral fractures smoked !0 cigarettes or more per day for 5 years or more, compared with 43 percent of controls with no vertebral fracture. Smoking is strongly associated with age. alcohol use, and, among some populations, use of exogenous estrogens. These are potentially strong confounders of the relationship between smoking and vertebral fracture, but Daniell’s comparison between cases and controls did not consider them. Since Daniell’s 1976 study, seven other case-control studies have examined the association between smoking and fracture of the hip or vertebrae (Table 6). Five of the seven case-control studies reported an increased risk of these osteoporotic fractures among smokers (Aloia et al. 1985: Cooper, Barker, Wickham 1988; Paganini-Hill et al. 1981; Seeman et al. 1983: Williams et al. 1982), and this association was Statistically significant in three of the studies (Aloia et al. 1985; Cooper, Barker, Wickham 1988: Williams et al. 1982). In the study by Williams and coworkers (1982), smokers were compared with obese nonsmokers, making it difficult to assess the independent associa- tion of smoking with the risk of osteoporotic fractures. A second analysis of smoking and the risk of hip or forearm fracture among the same subjects who were studied by Williams and colleagues (1982) showed no overall association of smoking and fractures (Alderman et al. 1986). In only two case-control studies were statistical adjustments made for age and exogenous estrogen use, which are potentially strong confounding variables; in both of these studies, there was no statistically significant association of smoking and fracture risk (Paganini-Hill et al. 1981: Kreiger et al. 1982: Kreiger and Hilditch 1986). In five cohort studies (Table 7). there was no increase in the risk of fracture among smokers (Farmer et al. 1989; Felson et al. 1988; Hemenway et al. 1988: Holbrook. Barrett-Connor, Wingard 1988, Jensen 1986). Three of these reports were based on 449 OS TABLE 6.—Summary of case-control studies of smoking and fractures Reference Daniell (1976) Seeman et al. (1983) Aloia et al. (1985) Population Cases: 38 women aged 40--69 with acute symptomatic vertebral fractures after minimal trauma Controls: 572 women outpatient volunteers aged 50-69 Cases: 10S men aged 44-84 with vertebral tractures Controls: 10S men aged 44-83 with Paget's disease matched for age and length of tollowup Cases: 5% white women (mean age 64.5) volunteers with vertebral fractures Controls: 58 white women volunteers matched for age Vertebral fractures Comparison Estimated relative risk 210 cig/day for 25 yr vs. less 42h Nonobese, nondrinking, nonsmokers vs. nonobese, nondrinking smokers with no underlying disease: aged <60 OB aged 60-69 1.6 aged 270 3.1 Smokers vs. nonsmokers 3.286 Comments No contro! for confounders: no statistical analysis One-third of the cases had it medical condition associated with bone loss; controls with Paget's disease may not be representative of men without vertebral fractures: design controls for age. obesity, and alcohol use Controlled for age only; multiple other risk factors examined using univariate tests