unfriendly takeover

1
1116 Unfriendly takeover SiR,—The commentary by Dr Andrew Herxheimer (March 21, p 732) on the evolution of the Pharmaceuticals Unit (PHA) in the European regional office of WHO (WHO/EURO) underlines a concern that is shared by many of those who have worked closely with PHA as either national representatives or individuals. We were recently (November, 1991) advisers two "Meeting to discuss reorientation of the Pharmaceuticals Programme" at WHO/EURO. The discussions focused on the future of the PHA and its relation with the Quality of Care and Technologies assessment unit. Our perception was that WHO/EURO is facing major economic and strategic difficulties. The document developed during the meeting and formally approved by the director of WHO/EURO included a clear plea to maintain high competence at PHA, both clinical pharmacological and pharmaceutical. A well- defined agenda for action was established, which welcomed a broader interplay between the two units whenever possible. As external advisers, we felt that organisational changes should not interfere with the professional integrity of PHA activities. The intention to broaden the perspectives of its activities, interventions, and evaluations beyond drug-based ones was agreed, and short- term as well as long-term strategies were indicated. No doubt the heavy pressures on WHO/EURO to assist emergencies in Eastern and Central Europe have diverted and confounded longer term priorities, such as education and research. In our opinion, one of the crucial roles of WHO/EURO is to stimulate development of Europe-wide collaboration. Hence the urgent need to fill the post of regional adviser to pharmaceuticals and to provide auxiliary staff and resources. We believe it unlikely that WHO/EURO can solve the present drug emergency unless its emphasis shifts to a scientifically based clinical pharmacological and epidemiological approach. This change is probably the only way to ensure that appropriate advice is given and to facilitate communication to develop reliable reference groups for promoting sound national drug policies. Institute of Pharmacological Research "Mario Negri", 20157 Milan, Italy GIANNI TOGNONI P. K. M. LUNDE Statistical correction for measurement imprecision SIR,-Your recent editorial (March 7, p 587) emphasised that imprecise measurement of a risk factor tends to dilute the observed relation between that risk factor and the risk of a disease; this notion has also been discussed by MacMahon and coworkers.1 They proposed that a statistical correction for the "regression dilution bias" should be included in the presentation of data from prospective epidemiological studies. Phillips and Davey Smith2 have noted that statistical control for a confounding factor is not perfect when there is measurement imprecision in the confounder. They have presented a method of statistical correction in a multivariate modeJ.2 Neither MacMahon et al nor Phillips and Davey Smith make a distinction between the measurement imprecision and the true (biological or behavioural) intrasubject variability over time. Both of these sources of intrasubject variation reduce the correlation between a single measurement of a risk factor and the frequency of that risk factor in a cohort. Measurement imprecision can be assessed by short-term repeat measurements. Remeasurements of a risk factor after several years are influenced by both measurement imprecision and true intrasubject variability, which is dependent on the tracking of that risk factor. While statistical correction for measurement imprecision may be warranted, statistical adjustment of the relation between a risk factor and the risk of a disease (eg, in relative risk) may be questionable. If a risk factor truly varies in an individual over time, the probability of disease associated with that risk factor (pathophysiological impact) may also vary over time, or be less than if the risk factor remained constant. Theoretically, this could be the case for blood pressure and serum triglycerides in coronary heart disease. Both blood pressure and blood triglycerides vary in a circadian pattern, as well as seasonally and over the life-span of an individual. Blood pressure varies according to physical strain and reactivity to emotional stimuli. Blood triglyceride concentrations, and to some extent blood pressure, also varies according to the fasting state of an individual. Data from the seven countries study suggest that changes in blood pressure might increase the risk of coronary heart disease, even when allowing for a given blood pressure If this were true, then a correction based on the intrasubject variability of blood pressure could erroneously exaggerate the strength of association between blood pressure and risk. Studies with repetitive measurements of blood lipids over time should investigate the impact of the variability of lipids on the risk of coronary heart disease. I suggest that both the assessment of true intrasubject variability and the measurement imprecision by means of short and long term repeat measurements should be a part of all epidemiological studies, and that a statistical correction in both bivariate and multivariate analysis should be applied for measurement imprecision, but only with great caution for the true intraindividual variability. Research Institute of Public Health, University of Kuopio, Box 1627, 70211 Kuopio, Finland JUKKA T. SALONEN 1. MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heart disease. Part 1, prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet 1990; 335: 765-74. 2. Phillips AN, Davey Smith G. How independent are "independent" effects? Relative risk estimation when correlated exposures are measured imprecisely. J Clin Epidemol 1991; 44: 1223-31. 3. Farchi G, Capocaccia R, Verdecchia A, Menotti A, Keys A. Risk factor changes and coronary heart disease in an observational study. Int J Epidemiol 1981; 10: 31-40. Retractor design and the lingual nerve SiR,—Mrs Brahams (March 28, p 801) draws attention to an important complication of one of the most frequent operations in oral surgery-removal of wisdom teeth. Although there may be room for improvement in retractor design to prevent lingual nerve damage, the number of patients affected might be reduced by adhering more closely to agreed criteria for surgical intervention. Indications for removal of wisdom teeth have been the subject of a National Institutes of Health (NIH) consensus conference:’ these are, recurrent pericoronitis, follicular cyst, caries not amenable to restorative procedures, internal or external resorption, and periodontal disease. However, no prospective studies have identified the proportion of lower wisdom teeth actually removed where surgery, according to these criteria, was not indicated. We therefore evaluated treatment decisions made by six oral surgeons in National Health Service hospitals for 28 consecutive male and 44 female patients, aged 15-44 years (mean 25), referred for lower wisdom tooth assessment. Patients were interviewed and examined by an independent assessor immediately after treatment decisions had been made. The assessor recorded the presence or absence of local disease on standard proformas, and treatment decisions were then evaluated by a panel of two further independent assessors, according to NIH consensus criteria. Of the 139 wisdom teeth examined, 56 were erupted, 78 partly erupted, and 5 fully erupted. 29 patients had been scheduled for surgery under general anaesthesia and 37 under local anaesthesia; 6 patients were not scheduled for surgery. 53 teeth did not meet criteria for surgery, of which 39 (74%) had nevertheless been scheduled for removal; if a single episode of pericoronitis is regarded as an indication for removal, then a further 12 wisdom teeth scheduled for surgery met consensus criteria and the remaining 27 did not. These findings suggest that of the 125 teeth that were scheduled for removal, 22% could have been left in situ. 14 patients who were scheduled to have both lower wisdom teeth removed under local anaesthesia (surgery usually done at two separate outpatient visits) could, if criteria had been adhered to more strictly, have avoided the removal of disease-free contralateral teeth. Although preventive measures have been responsible for striking decreases in dental disease, it would be a pity if such strategies included prophylactic removal of wisdom teeth-a procedure that gives rise to complications like any other surgical operation. There seems to be scope for reducing the numbers of wisdom teeth that are removed,

Upload: pkm

Post on 03-Jan-2017

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Unfriendly takeover

1116

Unfriendly takeoverSiR,—The commentary by Dr Andrew Herxheimer (March 21,

p 732) on the evolution of the Pharmaceuticals Unit (PHA) in theEuropean regional office of WHO (WHO/EURO) underlines aconcern that is shared by many of those who have worked closelywith PHA as either national representatives or individuals.We were recently (November, 1991) advisers two "Meeting to

discuss reorientation of the Pharmaceuticals Programme" at

WHO/EURO. The discussions focused on the future of the PHAand its relation with the Quality of Care and Technologiesassessment unit. Our perception was that WHO/EURO is facingmajor economic and strategic difficulties. The document developedduring the meeting and formally approved by the director ofWHO/EURO included a clear plea to maintain high competence atPHA, both clinical pharmacological and pharmaceutical. A well-defined agenda for action was established, which welcomed abroader interplay between the two units whenever possible. Asexternal advisers, we felt that organisational changes should notinterfere with the professional integrity of PHA activities. Theintention to broaden the perspectives of its activities, interventions,and evaluations beyond drug-based ones was agreed, and short-term as well as long-term strategies were indicated.No doubt the heavy pressures on WHO/EURO to assist

emergencies in Eastern and Central Europe have diverted andconfounded longer term priorities, such as education and research.In our opinion, one of the crucial roles of WHO/EURO is tostimulate development of Europe-wide collaboration. Hence theurgent need to fill the post of regional adviser to pharmaceuticalsand to provide auxiliary staff and resources.We believe it unlikely that WHO/EURO can solve the present

drug emergency unless its emphasis shifts to a scientifically basedclinical pharmacological and epidemiological approach. This

change is probably the only way to ensure that appropriate advice isgiven and to facilitate communication to develop reliable referencegroups for promoting sound national drug policies.Institute of Pharmacological Research"Mario Negri",

20157 Milan, Italy

GIANNI TOGNONIP. K. M. LUNDE

Statistical correction for measurementimprecision

SIR,-Your recent editorial (March 7, p 587) emphasised thatimprecise measurement of a risk factor tends to dilute the observedrelation between that risk factor and the risk of a disease; this notionhas also been discussed by MacMahon and coworkers.1 Theyproposed that a statistical correction for the "regression dilutionbias" should be included in the presentation of data from

prospective epidemiological studies. Phillips and Davey Smith2have noted that statistical control for a confounding factor is notperfect when there is measurement imprecision in the confounder.They have presented a method of statistical correction in a

multivariate modeJ.2Neither MacMahon et al nor Phillips and Davey Smith make a

distinction between the measurement imprecision and the true(biological or behavioural) intrasubject variability over time. Both ofthese sources of intrasubject variation reduce the correlationbetween a single measurement of a risk factor and the frequency ofthat risk factor in a cohort. Measurement imprecision can beassessed by short-term repeat measurements. Remeasurements of arisk factor after several years are influenced by both measurementimprecision and true intrasubject variability, which is dependent onthe tracking of that risk factor. While statistical correction formeasurement imprecision may be warranted, statistical adjustmentof the relation between a risk factor and the risk of a disease (eg, inrelative risk) may be questionable.

If a risk factor truly varies in an individual over time, theprobability of disease associated with that risk factor

(pathophysiological impact) may also vary over time, or be less thanif the risk factor remained constant. Theoretically, this could be thecase for blood pressure and serum triglycerides in coronary heartdisease. Both blood pressure and blood triglycerides vary in acircadian pattern, as well as seasonally and over the life-span of an

individual. Blood pressure varies according to physical strain andreactivity to emotional stimuli. Blood triglyceride concentrations,and to some extent blood pressure, also varies according to thefasting state of an individual. Data from the seven countries studysuggest that changes in blood pressure might increase the risk ofcoronary heart disease, even when allowing for a given bloodpressure If this were true, then a correction based on the

intrasubject variability of blood pressure could erroneouslyexaggerate the strength of association between blood pressure andrisk. Studies with repetitive measurements of blood lipids over timeshould investigate the impact of the variability of lipids on the risk ofcoronary heart disease.

I suggest that both the assessment of true intrasubject variabilityand the measurement imprecision by means of short and long termrepeat measurements should be a part of all epidemiological studies,and that a statistical correction in both bivariate and multivariateanalysis should be applied for measurement imprecision, but onlywith great caution for the true intraindividual variability.Research Institute of Public Health,University of Kuopio, Box 1627,70211 Kuopio, Finland JUKKA T. SALONEN

1. MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heartdisease. Part 1, prolonged differences in blood pressure: prospective observationalstudies corrected for the regression dilution bias. Lancet 1990; 335: 765-74.

2. Phillips AN, Davey Smith G. How independent are "independent" effects? Relativerisk estimation when correlated exposures are measured imprecisely. J ClinEpidemol 1991; 44: 1223-31.

3. Farchi G, Capocaccia R, Verdecchia A, Menotti A, Keys A. Risk factor changes andcoronary heart disease in an observational study. Int J Epidemiol 1981; 10: 31-40.

Retractor design and the lingual nerveSiR,—Mrs Brahams (March 28, p 801) draws attention to an

important complication of one of the most frequent operations inoral surgery-removal of wisdom teeth. Although there may beroom for improvement in retractor design to prevent lingual nervedamage, the number of patients affected might be reduced byadhering more closely to agreed criteria for surgical intervention.Indications for removal of wisdom teeth have been the subject of aNational Institutes of Health (NIH) consensus conference:’ theseare, recurrent pericoronitis, follicular cyst, caries not amenable torestorative procedures, internal or external resorption, and

periodontal disease. However, no prospective studies haveidentified the proportion of lower wisdom teeth actually removedwhere surgery, according to these criteria, was not indicated. Wetherefore evaluated treatment decisions made by six oral surgeons inNational Health Service hospitals for 28 consecutive male and 44female patients, aged 15-44 years (mean 25), referred for lowerwisdom tooth assessment.

Patients were interviewed and examined by an independentassessor immediately after treatment decisions had been made. Theassessor recorded the presence or absence of local disease onstandard proformas, and treatment decisions were then evaluatedby a panel of two further independent assessors, according to NIHconsensus criteria. Of the 139 wisdom teeth examined, 56 wereerupted, 78 partly erupted, and 5 fully erupted. 29 patients had beenscheduled for surgery under general anaesthesia and 37 under localanaesthesia; 6 patients were not scheduled for surgery. 53 teeth didnot meet criteria for surgery, of which 39 (74%) had neverthelessbeen scheduled for removal; if a single episode of pericoronitis isregarded as an indication for removal, then a further 12 wisdomteeth scheduled for surgery met consensus criteria and the

remaining 27 did not.These findings suggest that of the 125 teeth that were scheduled

for removal, 22% could have been left in situ. 14 patients who werescheduled to have both lower wisdom teeth removed under localanaesthesia (surgery usually done at two separate outpatient visits)could, if criteria had been adhered to more strictly, have avoided theremoval of disease-free contralateral teeth. Although preventivemeasures have been responsible for striking decreases in dentaldisease, it would be a pity if such strategies included prophylacticremoval of wisdom teeth-a procedure that gives rise to

complications like any other surgical operation. There seems to bescope for reducing the numbers of wisdom teeth that are removed,