Tobacco Documents Online

186 NX813 q.1 PONE ATHEROSCLEROSZS 97 , (OlELSEVZER 501 IRELAHD LTD ZR iil ELSEVIER Atherosclerosis 129 (1997) 41-48 .... ATH EROSCLEROSIS Ig Ig Risk factors associated with the development of peripheral arterial disease in smokers: a case-control study Janet T. PowelP,*, Richard J. Edwards", Phillip C. Worrella, Peter J. Franksa, Roger M. Greenhalgha, Neil R. Poulter~'b aDepartment of Surgery, Charing Cross and Westminster Medical School, Fulham Palace Road, London W6 8RF, UK bDepartment of Epidemiology and Public Health, UCLMS, I-19 Torrington Place, London WCIE 6BT, UK Received 4 June 1996; revised 9 October 1996; accepted 4 November 1996 lt_ l Abstract Purpose and method: A hospital based case-control study was designed to investigate what aspects of smoking and what co-factors of smoking are associated with the development of peripheral arterial disease (PAD). Cases were 291 smokers, newly referred with PAD, and controls were 828 age and sex matched smokers without PAD. Results: Reported recent tobacco usage was similar in cases and controls but total tobacco exposure was associated with the risk of PAD--adjusted odds ratios (ORs) increasing with tertile of pack-years smoked to reach 1.63 (95% CI, 1.11-2.39; P = 0.011), for the highest tertile ( > 48 pack-years) compared with smokers in the lowest tertile (< 31 pack-years). Cases reported smoking significantly lower tar and nicotine yield cigarettes than controls, but tended to inhale more deeply, and had significantly higher plasma concentrations of cotinine. ORs for PAD were significantly and independently increased by systolic blood pressure > 160 mmHg (8.1 (5.2-13.0); P < 0.0001), history of hypertension (2.4 (1.5-3.2); P = 0.0003) and apolipoprotein B > 0.9 g/1(3.8 (2.3-7.6); P - 0.008). Conclusions: Increased total exposure to tobacco and the ability to smoke tobacco in a way which maximises nicotine yield are associated with increased risk of smokers developing PAD. There is no evidence that smoking low tar cigarettes reduces this risk, whereas both hypertension (particularly systolic) and high levels of apolipoprotein B, increase this risk. © 1997 Elsevier Science Ireland Ltd. Keywords: Peripheral atheros~lerosis; Smoking; Low tar cigarettes; Apolipoprotein B; Systolic hypertension 1. Introduction Smoking has been identified consistently as the major risk factor for the development of peripheral arterial disease (PAD), and over 90% of patients with symp- tomatic PAD having a chronic smoking history [1-4]. Nevertheless the importance of other cardiovascular risk factors, genetic background or the particular facets of the smoking habit that differentiate those smokers who do and do not develop PAD is not clearly estab- lished. Reported data concerning systolic and diastolic blood pressure (BP), glucose intolerance and lipids, as *Corresponding author. Tel: +44 181 8467312; fax: +44 181 8467330. risk factors for PAD, are inconsistent, although the data concerning diabetes and plasma fibrinogen are more consistent [4]. However the effect of smoking to increase plasma fibrinogen concentrations varies ac- cording to particular fibrinogen genotypes [5] and it is also possible that the dietary modifications associated with smoking influence the development of cardiovas- cular disease [6]. Although there is consistent evidence that the risk of developing lung cancer is directly re- lated to the tar yield of cigarettes [7,8], the evidence linking tar yields to the development of cardiovascular disease is limited and controversial [9-12]. Since more than 30% of adults over 40 years of age smoke and therefore are potentially at risk of develop- ing PAD [13], it is important to try to identify what 0021-9150,97/517.00 ~ 1997 Elsevier Science Ireland Ltd. All rights reserved. PII S0021-91 50( 96106012-I

42 J.T. Powell et al. / Atherosclerosis 129 (1997) 41-48 co-factors determine the development of symptomatic PAD. Therefore the present case-control study was designed to compare the prevalence of details of smok- ing habit and various cardiovascular risk factors in two populations of smokers, those with symptoms of PAD referred for diagnosis and surgical opinion (cases) and other hospital patients without symptomatic PAD (con- trols). 2. Subjects and methods Cases comprised consecutive new referrals to the Vascular Surgical Service at Chafing Cross Hospital for evaluation of presumed PAD, who reported smoking in :the previous 4 weeks (current smokers). Between 1988 and the end of 1992 there were 654 new referrals of whom 35 (5%) had never smoked, 289 (44%) had stopped smoking more than 4 weeks ago and 330 (50%) were current smokers (potential cases): of these, 319 cases agreed to take part in the study. By means of systematic screening of suitab!e clinics and wards over the same period, 961 potential controls who also re- ported smoking in the previous 4 weeks, were identified, of whom 899 agreed to take part in the study. Of these, 556 (62%) were recruited from outpatient clinics (ortho- paedic, ophthalmology, dermatology and urology) at Charing Cross and St Mary's. Hospitals and 343 (38%) from Chafing Cross inpatient wards (short-stay, urol- ogy, general surgery and orthopaedic). Controls were chosen to ensure that the age-sex distribution broadly matched that among the cases. All data were recorded on a study questionnaire which was partly self-adminis- tered and was not returned by 22 potential cases and 39 controls. Cases and controls reporting a history of emphysema or of cancer of the lung, upper respiratory tract, mouth, lip, stomach, oesophagus or bladder were excluded. Controls who reported positively to a standard claudi- cation questionnaire [14] were excluded. Subjects were not excluded from the study if they refused consent for a blood sample. Following application of these exclu- sion criteria 291 cases (192 males) and 828 controls (536 males) remained eligible for inclusion in the study. Of the 291 cases, 255 (88%) had intermittent claudication and 36 (12%) had critical ischaemia. Intermittent clau- dication was defined from the history and Clinical ex- amination in the presence of an ankle/brachial systolic pressure index of < 0.8. Critical ischaemia is defined as either persistently recurring ischaemic rest pain requir- ing regular adequate analgesia for more than 2 weeks with an ankle pressure of < 50 mmHg, or ulceration of the foot or toes, with an ankle systolic pressure of < 50 mmHg [15]. Of the 828 controls 266were recruited from ortho- paedic outpatients. 116 from urology outpatients, 110 from opthalmology outpatients, 72 from dermatology outpatients, 142 were in-patients for herniorrhaphy, 83 were urology in-patients and 39 orthopaedic in-pa- tients. Past history of diabetes, hypertension and abnormal blood lipids with details of current medication and a history of parents or siblings dying from cardiovascular disease before the age of 60 were recorded in the questionnaire. For each mode of tobacco use (manufac- tured cigarettes, rolls-ups, pipes and cigars) subjects were asked for the age at which they started smoking, whether they had stopped and when, and maximum amounts smoked per day between ages 15-30, 31-50, 51 and above and during the last year. Roll-up cigarette smokers were asked to give amounts smoked per day. as either number of cigarettes smoked per day or ounces of tobacco (1/2 ounce of tobacco =20 cigarettes). Manufactured cigarette smokers were asked for the length of time they had smoked filter tipped cigarettes and roll-up cigarette smokers, whether they inserted filters into their roll-ups. Manufactured cigarette smok- ers were asked for the exact names of the brands they had smoked in the previous 5 years and for how long they had smoked each. Pack years were calculated for manufactured and roll-up cigarette smokers by multi- plying duration of smoking by the number of cigarettes smoked per day divided by 20, for the 3 age bands, and these values were then added together. Average yields of tar, carbon monoxide.and nicotine per cigarette were coded from estimates given by the Laboratory of the Government Chemist for June 1989. Newer brands were coded according to later estimates. Cigarette smokers were asked for usual length of cigarette smoked (greater than half, about half or less than half) and usual depth of inhalation (mouth and nose only, back of throat, lungs moderately or lungs deeply). All subjects were asked how many times they had smoked in the previous week, the previous day and on the day of their interview. Social class was assessed on the basis of current and past employment, the subject's last occupation was used if currently retired or unemployed and spouse's occupa- tion was used for married or widowed female subjects. Average weekly exercise output was evaluated and specific dietary information, including fat and salt in- take were recorded, using a validated questionnaire [16]. For the first 500 subjects, heights and weights were measured but thereafter, only subjects' self-reported height and weight were available for use in analyses. Routine clinic pulse and BP readings were recorded. Non-fasting blood samples were obtained by antecu- bital venepuncture using EDTA anticoagulated tubes from 235 cases and 382 controls. Carboxyhaemoglobin was measured on an IL-282 carboximeter (coefficient of variation: 3.2%), plasma thiocyanate by an automated

ular the ects ette ices es). the ttes [I ted ok- hey ~or tlff- ttes ere':|- the :tte I ~ 2tte , all) dy. ~cd lay ~ed 2ts. tire J.T. Powell et aL/ Atherosclerosis 129 (1997~ 41-48 43 colourimetric assay (coefficient of.variation: 9.5%) [17] and plasma cotinine by gas liquid chromatography (coefficient of variation: 18%) [18]. Total cholesterol, fibrinogen and apolipoprotein (B) were measured as previously described (coeffic.ients of variation: 2.7, 5.4 and 4.9% respectively) [19]. Apolipoprotein(a) was mea- sured by radioimmunoassay (Pharmacia, UK) accord- ing to the manufacturer's instructions (coefficient of variation: 5.0%). 3. Statistical methods Analyses were carried out using SPSSx computer software [20]. To assess the strength of association between each variable and the disease, odds ratios (ORs) with 95% CI were calculated. Continuous vari- ables were categorised into tertiles. All CI for OR's were calculated using the legit method for large sam- ples. Because smoking habit has previously been identified as the pivotal risk factor for PAD, an initial logistic regression was conducted to compare different mea- sures of smoking to determine which was most closely associated with PAD and hence should be used to adjust other case-control comparisons. A stepwise backwards elimination procedure was used (P= 0.10 for terms to drop out of the model). Variables consid- ered were pack years, age started smoking, inhalation into the lungs, mode of smoking, and years smoking filter-tipped cigarettes as a proportion of total years smoking. Pack years could only be calculated meaning- fully for manufactured and roll-up cigarette smokers and hence this analysis (which requires complete data for each subject) excluded pipe and cigar smokers (n = 24 cases and 105 controls). Pack-years was identified as the most significant smoking related factor differentiat- ing cases from controls. Multiple logistic regression was then used to identify other variables adjusted for age, sex, diabetes and pack-years history of smoking, associ- ated with the disease. To evaluate whether continuous variables were associated in a linear fashion with PAD, they were categorised into quartiles and log-odds were .plotted against the median of each quartile. Since no variables exhibited linear relationships, all continuous variables were treated in the multiple logistic regression as unordered categorical variables distributed in tertiles. 4. Results 4. I. General characteristics of cases and controls The mean ages of cases and controls were 64.8 (interquartile range 58-71) years and 63.7 (interquartile range 56-71) years, respectively. The mean body mass index of cases was 24.4 (interquartile range 22-27) kg/mz compared with 24.0 (interquartile range 22-27) kg/m2 in controls. A history of diabetes was signifi- cantly more common among cases than controls (12 versus 5% respectively, P < 0.001). Cases also reported a history of hypertension and hyperlipidaemia more frequently than did controls, with 99 cases (34%) re- porting hypertension compared with 166 controls (20%), P < 0.001 and 29 cases (10%) reporting hyperlip- idaemia compared with 30 controls (4%), P < 0.001. In keeping with the significantly higher reported rates of hypertension and diabetes, and the presence of PAD among cases, the use of medications for hypertension and diabetes, anticoagulants and aspirin was more fre- quent among cases than controls (data not shown). No other medications were used differentially by cases and controls. There was no difference in socioeconomic class between cases and controls. Almost half (48%) of cases and controls were in socioeconomic class III, with 28% being in higher social classes in "both groups. 4.2. Smoking habits and their validation by smoking markers Levels of risk associated with various aspects of reported smoking habit are shown in Table 1. Similar proportions in each group, (about three-qua.rters), smoked manufactured cigarettes and a further one sixth smoked hand-rolled cigarettes. The reported number of cigarettes smoked in the last year was similar in cases and controls although there was a statistically non-sig- nificant trend for cases to have started smoking at a younger age and to have smoked for longer than con- trols. Hence an increased cumulative pack-year history was more sighificantly associated with the development of PAD (P = 0.011) than pack-year history in earlier years of life (P -- 0.05). The depth of smoke inhalation tended to be greater among cases than controls, (P= 0.16). Only half the women, cases and controls, reported lung inhalation. Among men the depth of smoke inhalation was greater among cases than controls (P = 0.009). There were no significant differences between cases and controls in use of filtered cigarettes (98% of all those smoking manu- factured cigarettes) or the amount of cigarette smoked (more than half the cigarette smoked by 89% of cases and 86% of controls). Nearly all of the manufactured cigarette smokers (96%) could provide the name of the brand used most recently and 88% could remember what they smoked .5 years ago. The tar and nicotine yield of manufactured cigarettes derived from the brands reportedly used by cases were significantly lower than those used by controls .both at the time of the interview and 5 years previously (Table 2). During the 5 years before presentation 28% of cases and 25% of controls had switched cigarette brands. Although a 2063633588

Table l Tobacco usage J.T. Powell et al. / Atherosclerosis 129 (1997) 41-48 Cases (n = 291) Controls (n = .828) ORa (95% CI) Type of tobacco use at interview Manufactured cigarettes.onls~ 218 (75%) 609 (74%) Roll-up cigarettes only 49 (17%) 108 (13%) Pipe only 8 (3%) 48 (6%) Cigars only 11 (4%) 40 (5%) Mixture 5 (2%) 17 (2%) P=0.185 Depth of inhalation Mouth and nose 53 (18%) 182 (22%) 1.0 Back of throat 49 (17%) 143 (17%) 1.15 (0.73-1.82) Lungs moderate 128 (44°'4) 324 (39%) 1.43 (0.98-2.11) Lungs deep 59 (20%) 173 (21%) 1.25 (0.80-1.96) ; 289 822 P = O. 159 bNumber/day (last year) < 15 97 (36%) 268 (37%) " 1.0 15-24 121 (45%) 297 (4t%) 1.36 .(0.98, 1.90) 25+ 49 (18%) 152 (21%) 1.06 (0.70, 1.63) 267 717 P = 0.522 bAge started (years) < 16 92 (35°'4) 221 (31%) 1.0 16-19 93 (35°/'0) 235 (33%) 0.94 (0.66, 1.34) 20+ 80 (30%) 253 (36%) 0.75 (0.52, 1.08) 265 709 P ~ 0.119 Years smoked <40 69 (26%) 224 (34%) 1.0 40-49 101 (38%) 238 (34%) 1.22 (0.79, 1.90) 50+ 95 (36%) 227 (32%) 1.35 (0.77, 2.36) 265 709 P = 0.283 bTotal pack years smoked <31 72 (27%) 253 (35°,4) '1.0 31-48 86 (32°/'0) 237 (33%) 1.23 (0.84, 1.79) 49+ 107 (40%) 220 (31%) 1.63 (t.11, 2.39) 265 709 P = 0.011 ~Pack )'ears smoked before 31 )'ears of age < 11 79 (30%) 258 (36°'4) 1.0 l 1 - 16 101 (38%) 238 (34%) 1.69 (1.09-2.75) 17 + 85 (32%) 213 (30%) 1.25 (0.76- 2.23) 265 709 P = 0.05 Tabh I'~~ Ttlr 9 14 Tar 14 .Vic, .Vic~ OR are adjusted for age and sex, P-values are 2'.2 for trend. For subjects smoking cigarettes (manufactured or roll-up) at interview, 267 cases, 717 controls. greater proportion of cases than controls switched to low tar cigarettes (8% and 3% respectively), this differ- ence was not significant. Biochemical smoking markers varied with age, sex and depth of inhalation. After adjustment for these factors, there was a tendency for the risk of PAD to be associated with increased levels of carboxyhaemoglobin and plasma thiocyanate and a significant association with increasing concentrations of plasma cotinine (P = 0.006) (Table 2). Since recent cigarette consumption was similar among cases and controls, the observation that cases, who reported smoking lower nicotine yield cigarettes, had increased concentrations of plasma co- tinine was parad.oxical. The role of recall bias to ex- plain this paradox was investigated by comparing reported recent smoking habit with objective smoking markers in cases and controls. Of subjects who reported not smoking in the 24 h immediately prior to interview, 9/17 (53%) cases and 5/9 (56%) controls had cotinine concentrations (tl.2 -~ 18 h) in the range indicative of

J.T. Powell et al. / Atherosclerosis 129 (1997) 41-48 Table 2 Level of tobacco smoke component for cigarette brand and objective markers of smoking among users of manufactured cigarettes 45 Cases (n = 218) Controls (n = 609) ORa (95% CI) I. ili !| i| I | I- il Tar level (mg/cigarette) current cigarette <9 78 (38%) 181 (31%) 1.0 9-13 65 (32%) 15I (26%) 1.0I (0.67, 1.52) 14+ 62 (30%) 253 (43%) 0.57 (0.38, 0.86) 205 585 P = 0.005 Tar level (mg/cigarette) cigarette smoked 5 years ago <9 65 (35%) 168 (31%) 1.0 9-13 50 (27%) 126 (23%) 1.00 (0.69, 1.70) 14+ 69 (38%) 246 (46%) 0.76 (0.50, 1.15) 184 540 P = 0.17 Nicotine level (mg/cigarette) current cigarette <0.8 52 (25%) 119 (20%) 1.0 0.8-1.1 77 (38%) 181 (31%) 1.01 (0.65, 1.56) 1.2+ 76 (37%) 285 (49%) 0.63 (0.41, 0.97) 205 585 P = 0.017 Nicotine level (mg/cigarette) cigarette smoked 5 years ago <0.8 65 (35%) 173 (32%) 1.0 0.8-1.1 58 (32%) 135 (25%) 1.08 (0.68, 1.55) 1.2+ 51 (33%) 232 (43%) 0.65 (0.44, 1.06) 184 540 P -- 0.20 bWhole blood carboxyhaemoglobin (%) <2.7 40 (20%) 86 (29%) 1.0 2.7-4.4 77 (39%) 98 (33%) 1.70 (1.03, 2.81) 4.5+ 79 (40%) 110 (37%) 1.62 (0.97, 2.70) 196 294 P ~ 0.08 bPlasma cot#~ine (nmol/l) < 300 300-699 700 + b Plasma ~hiocyanate (.umol/l) < 100 100-150 150+ 40 (21%) 99 (34%) 1.0 80 (41%) 110 (34%) 1.78 (1.09, 2.91) 75 (38%) 85 (29%) 2.05 (1.24, 3.40) 195 294 P = 0.006 49 (25%) 101 (35%) 1.0 80 (41%) 95 (330/0) 1.89 (1.17, 3.05) 67 (340/0) 98 (33%) 1.58 (0.96, 2.62) 196 294 P = 0.08 :ld tag ~ All OR are adjusted for age and sex and additionally for depth of inhalation for the smoking markers, the P-value is obtained from Z'- for trend analysis. ~ Only those with a blood sample taken on the day they attended hospital are used in these analyses, where the OR has been adjusted for age, sex and depth of inhalation. recent smoking, whereas for those who admitted smok- ing in the past 24 h 168/9 (99%) cases and 283/285 (99"/,) controls had continine concentrations in the -smoking range. Even for the shorter half-life marker carboxyhaemoglobin (q ,_ 4-5 h), 5/17 (29%) cases and 3i9 (33%) controls, had levels indicative of very recent smoking. Hence, assuming the accurate discriminant powers of the smoking markers, inaccurate reporting of very recent smoking was similar in cases and controls. 4.3. Cardiovascular risk factors In order to evaluate the risks of developing PAD in association with several standard cardiovascular risk factors, the ORs of developing PAD using separate logistic regression analyses run for each variable, unad- justed and adjusted for age, sex, diabetes and pack years, are presented in Table 3. No increased risk of developing PAD was associated with ethnic origin,

46 J.T. Powell et al. / Atherosclerosis 129 (1997) 41-48 Table 3 Risk factors for peripheral arterial disease in smokers OR (95% CI) P-value ~Adjusted OR (95% CI) bP-value Systolic blood pressure (mmHg) < 140 1.0 1.0 140-159 2.91 (1.81, 4.66) 2~69 (1.66, 4.38) 160+ 8.07 (4.85, 13.49) <0.0001 7.63 (4.45, 13.07) <0.0001 167 cases, 355 dontrols 1.0 1.0 1.10 (0.68, 1.80) 1.08 (0.66, 1.79) 2.08 (1.33, 3.30) 0.001 1.95 (1.22, 3.11) 0.004 Diastolic blood pressure (mmHg) Cholesterol (mmol/l) Apolipoprotein B (g/l) Apolipoprotein(a) (mg/dl) Fibrinogen (g/l) <75 75-84 85+ 167 cases, 355 controls <5.2 5.2-6.4 6.5+ 205 cases, 382 controls <0.70 0.70-0.89 0.9+ <I0 10-39 40+ <5.1 5.1-6.0 6.1+ 1.0 1.0 1.63 (1.03, 2.59) 1.89 (1.16, 3.08) 2.10 (1.33, 3.33) 0.002 2.82 (1.70, 4.67) 1.0 1.0 2.72 (1.65, 4.48) 2.82 (1.68, 4.76) 4.06 (2.50, 6.61) . <0.0001 4.45 (2.65, 7.45) 204 eases, 382 controls 1.0 0.78 (0.50, 1.20) 1.19 (0.74, 1.81) 204 cases, 379 controls 1.0 1.51 (0.96, 2.37) 1.85 (1.21, 2.87) 197 eases, 374 controls <0.0001 <0.0001 1.0 0.71 (0.46, 1.12) 0.396 1.25 (0.81, 1.92) 0.24 1.0 1.32 (0.83, 2.11) 0.005 1.66 (1.05, 2.62) 0.02 l./ll l, [ [ l i Adjusted for age, sex, diabetes and pack years of smoking. Z'z test or chi square for trend (3 categories). social class, age started smoking, type of tobacco product used, or family history of arterial disease. A complete data set was only available for 413 subjects (141 cases and 272 controls) who smoked manufactured cigarettes, largely because blood samples were not obtained from all subjects. However there w, ere no differences between those who gave a blood sample and those who did not with respect to age, sex, pack-year history of smoking, other smoking habits, body mass index and BP. After adjustment for estab- lished risk factors for PAD--age, sex, diabetes and pack-year history of smoking--multiple logistic regres- sion including 21 variables identified the following 3 independent factors as being associated with an in- creased risk of PAD: systolic BP (OR = 8.1 (95% CI, 5.2-13.0), (P < 0.0001)); history of hypertension (OR = 2.4 (95% CI 1.5-3.2), (P = 0.003)) and apolipo- protein B (OR = 3.8 (95% CI 2.3-7.6), (P = 0.008)). 5. Discussion Smoking has been established consistently as the major risk factor for PAD [I-4]. Therefore by recruit- ing only smokers, differences in the details of the smoking habit between cases and controls, and a clearer picture of the other risk factors for the develop- ment of PAD in smokers were established. The smok- ing habits of both cases and controls (number of cigarettes currently smoked) was similar to that re- ported for age-sex matched groups in the general population [13]. The results of this hospital based case- control study indicate that whilst an increased total exposure to tobacco products (pack year history) and nicotine absorption were associated with a significant increase in the risk of developing symptomatic PAD, there was no indication that smoking lower tar yield cigarettes decreased the likelihood of developing symp- tomatic PAD. Although all cases with PAD were newly referred, the developing problem may have been evident for a considerable time prior to consultation with a specialist. Hence the disease could have prompted a recent switch to a lower tar brand cigarette, on the assumption that this was a safer form of smoking. Indeed, inspection of the cigarette brand smoked 5 years previously indicates that there had been slightly more switching to lower tar brand cigarettes among the cases. However, even 5 years prior to presentation when most cases did not suffer PAD symptoms, cases reported smoking cigarette

i | of ,tal tnd ]I ~] .tnt ed, ]l _--. H I" a iSto tes _il • tte H J.T. Powell et al. / Atherosclerosis 129 (1997) 41-48 47 brands which had lower tar and nicotine content than controls (Table 2). It remains possible that cases were more prone to recall bias, since it has long been recog- nised that patients with smoking related disorders may be deceptive about their current smoking habit [19,21]. However, recent recall bias was similar in cases and controls, with objective smoking markers identifying similar proportions of covert smokers in both cases and controls who claimed not to have smoked in the 24 h prior to interview. Therefore the clear trends for cases to have increased levels of all three smoking markers (Table 2) is indicative that cases may have smoked their cigarettes differently, perhaps dragging the cigarette down to the butt to maximise the nicotine yield. The risk of PAD was more strongly associated with plasma cotinine concentration than with carboxyhaemoglobin or thiocyanate concentrations, which reflect the metabolism of two gaseous products of tobacco com- bustion, carbon monoxide and hydrogen cyanide, re- spectively. This may indicate that nicotine, the precursor of cotinine, is associated more closely with vascular injury than carbon monoxide and hydrogen cyanide and that smoking a cigarette to maximise nicotine yield is a particularly dangerous habit. The information collected on cardiovascular risk fac- tors other than smoking was less complete than much of the smoking data (Table 3). Nevertheless there were clear indications that, as for coronary heart disease, hypertension and hyperlipidaemia are important risk factors for the development of PAD in smokers. After adjustment for age, sex, diabetes, and pack years of smoking, the 3 principal co-factors of smoking involved in the development of PAD were increased systolic BP, history of hypertension and increased concentrations of apolipoprotein B. The 7-8 fold increase in odds ratio for those with systolic blood pressures > 160 mmHg is in keeping with prospective studies which indicate that systolic is a better predictor of future cardiovascular events than diastolic pressure [22]. The significance of systolic BP and a history of hypertension as indepen- dent risk factors was re-emphasised in the stepwise logistic regression analysis of the 413 smokers with a completed data set for all variables. The aetiological relationship between lipid profiles and PAD has been likened to that between lipid profiles and coronary heart disease [23], which is consistent with the observa- tion that over one half of PAD patients die from coronary heart disease [24-26]. A significant associa- tion between serum triglycerides and PAD has been observed in many studies, but this association has rarely been found to be independent [27,28]. The results of the logistic regression analysis in this study which identified apolipoprotein B, rather than chol,esterol, as an independent risk factor for the development of PAD are compatible with apolipoprotein B acting as a marker of both cholesterol-rich and triglyceride-rich lipoproteins. The failure of fibrinogen to emerge as an independent risk factor in the logistic regression analy- sis may reflect the fact that smoking is a major determi- nant of plasma fibrinogen concentrations; both cases and controls were smokers and data was adjusted for pack-years. This study, which recruited both cases and controls from hospital patients, cotfld be subject to criticism. The presence of controls in hospital could hage been determined in part by the effects of smoking and hence the controls could have over-represented those with smoking-related illness. The controls' diagnoses were selected to avoid this problem. In addition, since the cases were referred to hospital for evaluation and diag- nosis of their PAD, the results may not be applicable to the general population. However, the major irripact on quality of life and health service resources associated with PAD is confined largely to those cases severe enough to be referred to hospital. 6. Conclusion These data do not support the hypothesis that those smoking low tar cigarettes are less likely to suffer PAD. Rather our evidence indicates that patients with PAD know how to smoke their cigarettes in such a manner as to maximise the nicotine yield. This smoking tech- nique appears to be a major determinant of developing PAD among smokers. However, it also seems likely that control of hypertension and optimal lipid profiles are likely to effect an important reduction in the mor- bidity attributable to PAD in the community. Acknowledgements This work was supported by research grants from the Tobacco Products Research Trust and the British Heart Foundation. We thank past research assistants on this project for their contribution. These include I. Corrie, C. Harris and M. Beksinska. We also thank all those consultants who gave permission for us to study their patients. 2063633592 References [1] Juergens JL, Barker NW, Hines EA. Arteriosclerosis obliterans: review of 520 cases with special reference to pathogenic and prognostic factors. Circulation 1960;21:188-195. [2] Lord JW. Cigarette smoking and atherosclerosis occlusive dis- ease. JAMA 1965;191:249-251. [3] Hughson WG, Mann Jl, Garrod A. Intermittent claudica- tion.:prevalence and risk factors. BMJ 1978;1:1379-1381. [4] Fowkes FGR. Epidemiology of atherosclerotic arterial disease in the lower limbs. Eur J Vasc Surg 1988;2:283-291.

48 J.T. Powell et al. / Atherosclerosis 129 (1997) 41-48 [5] Behague I, Poirir O, Nicaud Vet al./~ Fibrinogen gene polymor- phisms are associated with plasma fibrinogen and coronary artery disease in patients with myocardial infarction. Circulation 1996:93:440-449. [6] Margetts BM, Jackson AA. Interactions between people's diet and their smoking habits: the dietar~ and nutritional survey of British adults. Br Med J 1993;307:1381-1384. [7] Surgeon General. The health consequences of smoking. The changing cigarette. Rockville, Maryland, USA: US Department of Health and Human Services, 1989. [8] International Agency for Research on Cancer. Tobacco smok- ing. IARC Monograph Evaluating Carcinogen Risk to Humans, 1985 pp. 37-421. [9] Kaufman DW, Helmrich SP, Rosenberg L, Miettenen OS, Shapiro S. Nicotine and carbon monoxide content of cigarette smoke and the risk of myocardial infarction in young men. New Engl J Med 1983;308:409-413. • [10] Pettiti DB, Friedman GD. Cardiovascular and other diseases in smokers of low yield cigarettes. J Chron Dis 1985;48:581-588. [I 1] Palmer JR, Rosenberg L, Shapiro S. Low yield cigarettes and the risk of nonfatal myocardial infarction in women. New Engl J Med 1989;320:1569-1573. [12] Negri E, Franzosi MG, La Vecchia C, Santoro L, Nobili A, Tognoni G. Tar yield of cigarettes and risk of acute myocardial infarction. Br Med J 1993;306:1567-1570. [13] Colhoun H, Prescott-Clarke P, Dong W, Hedges B, Lampe F, Taylor A. In: Colhoun H and Prescott-Clarke P (eds). Health Survey for England 1994. London: HMSO, 1996. [14] Rose GA, Blackburn H. Cardiovascular survey methods. World Health Organization (Geneva) Monograph series No. 56: 1968. [15] European working group on critical leg ischaemia. Second Eu- ropean Consensus document on chronic critical leg ischaemia. Circulation 1991;84(Suppl IV):1-26. [16] Yarnell JWG, Felicity AM, Milbank JE, Swectman PM, Walker CL. A short dietary questionnaire for use in an epidemiologieal survey: comparison with weighed dietary records. Hum Nutr Appl Nutr 1983;37A:103-112. [17] Butts WC, Kuehmenan M, Widdowson GM. Automated method for determining serum thiocyanate to distinguish smok- ers from non-smokers. Clin Chem 1974;20:1344-1348. [18] Feyerabend C, Russell MAH. Rapid gas-liquid chromatographic determination of cotinine in biological fluids. Analyst 1980; 105:998-1001. [19] Wiseman S, Kenchington G, Dain R et al. Influence of smoking and plasma factors on patency of femoropopliteal vein grafts. Br Med J 1989;299:643-646. [20] SPSS Incorporated. SPSS Reference Guide. Chicago: SPSS lnc, 1990. [21] Sillet R, Wilson MB, Malcolm RE, Ball KP. Deception among smokers. Br Med J 1978;65:197-200. [22] Rutan GH, Kuller LH, Neaton JD, Wentworth DH, McDonald RH, McFate Smith W. Mortality associated with diastolic hy- pertension and isolated systolic hypertension among men screened for the Multiple Risk Factor Intervention Trial. Circu- lation t988;77:504-514. [23] Leng GC and Fowkes FGR. Lipids: Epidemiology: In Fowkes FGRE (ed). Epidemiology of Peripheral Vascular Disease. Lon- don: Springe~r-Verlag, 1991. [24] Jelnes R, Gaardsting O, Hougaard Jensen K, Baekgaard N, Tonnesen KH, Schroeder T. Fate in intermittent claudication: outcome and risk factors. BMJ 1986;293:1137-1139. [25] Davey-Smith G, Shipley M, Rose G. Intermittent claudication, heart disease risk factors and mortality: the Whitehall Study. Circulation 1990;82:1925--'1931. [26] Criqui MH, Langer RD, Fronek Aet al. Mortality over a period of 10 years in patients with peripheral arterial disease. New Engl J Med 1992;326:381-386. [27] Fowkes FGR, Housley E, Riemersma RA et al. Smoking, lipids, glucose intolerance and blood pressure as risk factors for periph- eral atherosclerosis compared with isehaemic heart disease in the Edinburgh Artery Study. Am J Epidemiol 1992;135:331-340. [28] Greenhalgh RM, Lewis B, Rosengarten DS, Mevart I, Calnan JS, Martin P. Serum lipids and lipoproteins in peripheral vascu- lar disease. Lancet 1971;ii:947-950. ! ! l [