- Email: [email protected]
PREOPERATIVE PREDICTION OF POSTOPERATIVE DEEP-VEIN THROMBOSIS
1015 settle in the alveoli (whence they would be removed by alveolar clearing mechanisms) or be exhaled. Sidestream smoke particles, however, if agglomerated, would be more likely to settle in the bronchi. Differences in breathing patterns of passive and active smokers may also be important. Despite the burgeoning evidence of the harmfulness of passive smoking, I tend to the view that the main case against public smoking is an aesthetic one, as with other antisocial habits. However, given the formidable chemistry of sidestream smoke, it seems inconceivable that this form of pollution is not inimical to the human organism, and I remain of the view that much of the injurious effect of passive smoking awaits discovery.9 While healthy scepticism, such as Lee’s, is never out of place, we would be wise to recall the medical attitude to smoking of seventy-five years ago: "The evidence that it is to any great degree injurious to the mature adult is much less strong ... If it is so harmful, why do not physicians see more of its ill effects?" 10 50 Thong Lane, Gravesend, Kent DA 12 4LD
fibrinogen. We have shown that the sensitivity of the radioisotopic technique diminishes progressively with time after operation6 and that
a
false conclusion may be recorded if thrombosis has been
merely delayed rather than totally prevented. In other words, if the onset of thrombosis is delayed by a few days the isotopic technique may fail to detect it altogether. This is especially relevant to patients on heparin, since heparin significantly delays the onset of thrombosis.7,8 It might have been better if Lowe et al. had used phlebography to confirm the absence of DVT in patients on heparin prophylaxis. G. -
PREOPERATIVE PREDICTION OF POSTOPERATIVE DEEP-VEIN THROMBOSIS
SIR,-Dr Lowe and colleagues (Feb. 20, p. 409) have deriveda discriminant function to predict patients at high risk of deep-vein thrombosis (DVT) after gastrointestinal surgery. They used two clinical variables, age and % overweight for age, sex, and height (%
MW). Lowe et al. chose the cut-off point in their first study to identify 90% of patients who had acquired DVT. It seems that they chose the cut-off point by visual inspection of the scatter diagram. Such a technique forced them to alter the cut-off point from 175 in study 1 to 170 in study 2. In a similar series of investigations McNicol and colleagues, who studied gynaecological patients, chose their cut-off point also by inspection apparently and found that it was also not3 reproducible from one study! to the next.2 The cut-off point can and should be determined statistically for it to be reproducible from one study to the next by the same investigators as well as by others. We have done a study similar in design to that of Lowe et al. to assess the predictive value of clinical variables as well as some blood tests.4We cannot confirm the findings of Lowe et al. and wish to highlight differences between their results and ours. Firstly, we found that the frequency of DVT in general surgical patients was only 18% (10/56) as against the 37 - 5% (39/104) reported by Lowe et al. Furthermore, none of the preoperative tests of coagulation, platelet function, and fibrinolysis nor any clinical variable including age, % MW, blood loss, or length of operation showed a significant difference between the group means of patients who had DVT and of those who did not. However, a prognostic index based on two tests (activated partial thromboplastin time and 3-h fibrin digestion teSt5) was obtained, and the cut-off point was determined statistically to minimise the misclassification of both groups of patients. This index gave a successful group allocation rate of 59%. If the cut-off point had been chosen by inspection only, the success rate of the index would have been improved to 71%, at the cost of reproducibility. We found age and % MW to be very poor discriminants: the risk factors for DVT in patients undergoing major abdominal surgery in Australia are presumably no different from those in similar patients in the U.K. We wish to strike a cautionary note on the absolute reliance Lowe et al. placed in the isotopic diagnosis of DVT by l2sl-labelled
are
concerned with
perils of second-hand smoking.
New Sci
our
analysis,
in the Oslo
Study (Dec.
12, p. 1303), for separation of effect of antismoking and diet
on
incidence of acute myocardial infarction (AMI). Certainly, our numbers were small, and the results should be interpreted with care: there are indications but not proof of cause-and-effect mechanisms. On the other hand, if we had not tried to do any quantitation of each factor we would also have been criticised. The only way we could find out anything with the computer programs available to us was by introducing the change factors as if they were baseline variables, even though we knew that such a procedure could be hazardous. However, the changes happened quickly and persisted, and they were monitored individually throughout the study. Thus, we believe our results to be fairly correct for the lipid factor. Somewhat more caution, though, should be exercised when interpreting the fairly negative conclusion1 about cigarette smoking. More details have now been published.’ As Professor Kringlen notes (Feb. 27, p. 504) the intervention group patients attended twice as often as did the control patients, and there may have been a group-specific favourable psychological effect. However, the difference in AMI incidence between intervention and control groups was not statistically associated with such a group-specific factor and could explain little if any of the between-group difference in AMI incidence. Changes in serum cholesterol explained about 10 of the 17 cases prevented and cigarette smoking, at best, 4 cases. Thus, little is left over for a psychological effect. The probability of a type II error is great because of the small difference (17 cases) but the data support only weak suggestions of a psychological effect beyond what can be explained by changes in serum cholesterol and in cigarette smoking. One might argue that a psychological feeling of well-being induced by the more frequent contact for the men in the intervention group lowered serum cholesterol independently of diet. This seems unlikely, however. The existence of a higher level of psychosocial wellbeing in the intervention group is questionable since adherence to follow-up was slightly better in the control group; also, a questionnaire about anxiety, answered anonymously, was answered by "yes" by 27 and 9 men in the intervention and control groups; respectively. Furthermore a common argument against recommending a cholesterol-lowering diet is the possible anxiety induced by the feeling of being a "risk individual". Medical Outpatient Clinic, Life Insurance Companies’ Institute for Medical Statistics, Ullevaal Hospital,
Oslo, Norway 9. Stock SL. The
-
CHOLESTEROL, SMOKING, AND THE OSLO STUDY SIR,-In your issue of Feb. 27 (p. 503) Dr Vandenbroucke and Dr Hofman
SHERRIDAN L. STOCK
.
ALI SHAH TEHSEEN Z. DHALL D. PAL DHALL
Vascular Laboratory, Woden Valley Hospital, Woden, A.C.T. 2606, Australia
INGVAR HJERMANN INGAR HOLME PAUL LEREN
1980; October 2: 10-13.
10. Editorial. The effect of tobacco in health and disease. JAMA 1905; (Sept 16): 855-56. 1. Clayton JK, Anderson JA, McNicol GP. Preoperative prediction of postoperative deep vein thrombosis. Br Med J 1976; ii: 910-12. 2 Crandon AJ, Peel KR, Anderson JA, Thompson V, McNicol GP. Postoperative deep vein thrombosis: Identifying high-risk patients. Br Med J 1980; 281: 343-44. 3 Armitage P. Statistical methods in medical research. Oxford: Blackwell Scientific Publications, 1971. 4. Dhall TZ, Shah GA, Ferguson IA, Ardlie NG, Dhall DP. Preoperative blood tests in prediction of postoperative deep vein thrombosis. Thromb Res (in press). 5 Hickman JA, Gordon-Smith IC. Timed fibrin digestion: a simplified technique for the measurement of the fibrinolytic activity of the blood. J Clin Pathol 1972; 25: 191-93.
6. Shah
GA, Dhall TZ, Ferguson IA, Dhall DP. Detection of deep vein thrombosis by 125I-labelled fibrinogen scanning technique: a methodological study. Thromb Res
1980, 18: 101-12.
Hampson WGT, Harris FC, Lucas HK, et al. Failure of low dose heparin to prevent deep vein thrombosis after hip replacement arthroplasty. Lancet 1974; ii: 795-97. 8. Sagar S, Nairn D, Stamatakis JD, et al. Efficacy of low-dose heparin in prevention of extensive deep-vein thrombosis in patients undergoing total hip replacement. 7.
Lancet 1976; i: 1151-54. 1. Holme I. On the separation of the intervention effects of diet and antismoking advice on the incidence of major coronary events in coronary high risk men: The Oslo Study. J Oslo City Hosp 1982; 32: 31-54.
fibrinogen. We have shown that the sensitivity of the radioisotopic technique diminishes progressively with time after operation6 and that
a
false conclusion may be recorded if thrombosis has been
merely delayed rather than totally prevented. In other words, if the onset of thrombosis is delayed by a few days the isotopic technique may fail to detect it altogether. This is especially relevant to patients on heparin, since heparin significantly delays the onset of thrombosis.7,8 It might have been better if Lowe et al. had used phlebography to confirm the absence of DVT in patients on heparin prophylaxis. G. -
PREOPERATIVE PREDICTION OF POSTOPERATIVE DEEP-VEIN THROMBOSIS
SIR,-Dr Lowe and colleagues (Feb. 20, p. 409) have deriveda discriminant function to predict patients at high risk of deep-vein thrombosis (DVT) after gastrointestinal surgery. They used two clinical variables, age and % overweight for age, sex, and height (%
MW). Lowe et al. chose the cut-off point in their first study to identify 90% of patients who had acquired DVT. It seems that they chose the cut-off point by visual inspection of the scatter diagram. Such a technique forced them to alter the cut-off point from 175 in study 1 to 170 in study 2. In a similar series of investigations McNicol and colleagues, who studied gynaecological patients, chose their cut-off point also by inspection apparently and found that it was also not3 reproducible from one study! to the next.2 The cut-off point can and should be determined statistically for it to be reproducible from one study to the next by the same investigators as well as by others. We have done a study similar in design to that of Lowe et al. to assess the predictive value of clinical variables as well as some blood tests.4We cannot confirm the findings of Lowe et al. and wish to highlight differences between their results and ours. Firstly, we found that the frequency of DVT in general surgical patients was only 18% (10/56) as against the 37 - 5% (39/104) reported by Lowe et al. Furthermore, none of the preoperative tests of coagulation, platelet function, and fibrinolysis nor any clinical variable including age, % MW, blood loss, or length of operation showed a significant difference between the group means of patients who had DVT and of those who did not. However, a prognostic index based on two tests (activated partial thromboplastin time and 3-h fibrin digestion teSt5) was obtained, and the cut-off point was determined statistically to minimise the misclassification of both groups of patients. This index gave a successful group allocation rate of 59%. If the cut-off point had been chosen by inspection only, the success rate of the index would have been improved to 71%, at the cost of reproducibility. We found age and % MW to be very poor discriminants: the risk factors for DVT in patients undergoing major abdominal surgery in Australia are presumably no different from those in similar patients in the U.K. We wish to strike a cautionary note on the absolute reliance Lowe et al. placed in the isotopic diagnosis of DVT by l2sl-labelled
are
concerned with
perils of second-hand smoking.
New Sci
our
analysis,
in the Oslo
Study (Dec.
12, p. 1303), for separation of effect of antismoking and diet
on
incidence of acute myocardial infarction (AMI). Certainly, our numbers were small, and the results should be interpreted with care: there are indications but not proof of cause-and-effect mechanisms. On the other hand, if we had not tried to do any quantitation of each factor we would also have been criticised. The only way we could find out anything with the computer programs available to us was by introducing the change factors as if they were baseline variables, even though we knew that such a procedure could be hazardous. However, the changes happened quickly and persisted, and they were monitored individually throughout the study. Thus, we believe our results to be fairly correct for the lipid factor. Somewhat more caution, though, should be exercised when interpreting the fairly negative conclusion1 about cigarette smoking. More details have now been published.’ As Professor Kringlen notes (Feb. 27, p. 504) the intervention group patients attended twice as often as did the control patients, and there may have been a group-specific favourable psychological effect. However, the difference in AMI incidence between intervention and control groups was not statistically associated with such a group-specific factor and could explain little if any of the between-group difference in AMI incidence. Changes in serum cholesterol explained about 10 of the 17 cases prevented and cigarette smoking, at best, 4 cases. Thus, little is left over for a psychological effect. The probability of a type II error is great because of the small difference (17 cases) but the data support only weak suggestions of a psychological effect beyond what can be explained by changes in serum cholesterol and in cigarette smoking. One might argue that a psychological feeling of well-being induced by the more frequent contact for the men in the intervention group lowered serum cholesterol independently of diet. This seems unlikely, however. The existence of a higher level of psychosocial wellbeing in the intervention group is questionable since adherence to follow-up was slightly better in the control group; also, a questionnaire about anxiety, answered anonymously, was answered by "yes" by 27 and 9 men in the intervention and control groups; respectively. Furthermore a common argument against recommending a cholesterol-lowering diet is the possible anxiety induced by the feeling of being a "risk individual". Medical Outpatient Clinic, Life Insurance Companies’ Institute for Medical Statistics, Ullevaal Hospital,
Oslo, Norway 9. Stock SL. The
-
CHOLESTEROL, SMOKING, AND THE OSLO STUDY SIR,-In your issue of Feb. 27 (p. 503) Dr Vandenbroucke and Dr Hofman
SHERRIDAN L. STOCK
.
ALI SHAH TEHSEEN Z. DHALL D. PAL DHALL
Vascular Laboratory, Woden Valley Hospital, Woden, A.C.T. 2606, Australia
INGVAR HJERMANN INGAR HOLME PAUL LEREN
1980; October 2: 10-13.
10. Editorial. The effect of tobacco in health and disease. JAMA 1905; (Sept 16): 855-56. 1. Clayton JK, Anderson JA, McNicol GP. Preoperative prediction of postoperative deep vein thrombosis. Br Med J 1976; ii: 910-12. 2 Crandon AJ, Peel KR, Anderson JA, Thompson V, McNicol GP. Postoperative deep vein thrombosis: Identifying high-risk patients. Br Med J 1980; 281: 343-44. 3 Armitage P. Statistical methods in medical research. Oxford: Blackwell Scientific Publications, 1971. 4. Dhall TZ, Shah GA, Ferguson IA, Ardlie NG, Dhall DP. Preoperative blood tests in prediction of postoperative deep vein thrombosis. Thromb Res (in press). 5 Hickman JA, Gordon-Smith IC. Timed fibrin digestion: a simplified technique for the measurement of the fibrinolytic activity of the blood. J Clin Pathol 1972; 25: 191-93.
6. Shah
GA, Dhall TZ, Ferguson IA, Dhall DP. Detection of deep vein thrombosis by 125I-labelled fibrinogen scanning technique: a methodological study. Thromb Res
1980, 18: 101-12.
Hampson WGT, Harris FC, Lucas HK, et al. Failure of low dose heparin to prevent deep vein thrombosis after hip replacement arthroplasty. Lancet 1974; ii: 795-97. 8. Sagar S, Nairn D, Stamatakis JD, et al. Efficacy of low-dose heparin in prevention of extensive deep-vein thrombosis in patients undergoing total hip replacement. 7.
Lancet 1976; i: 1151-54. 1. Holme I. On the separation of the intervention effects of diet and antismoking advice on the incidence of major coronary events in coronary high risk men: The Oslo Study. J Oslo City Hosp 1982; 32: 31-54.