671
Conference
Medicine and the Law
European AIDS definition
Scientific and legal standards of proof in environmental personal injury cases
Sept 16 and 17, 1991, the European Centre for the Epidemiological Monitoring of AIDS convened a meeting of the European Community Project Management Group (PHG) on AIDS epidemiology to discuss a possible revision of the AIDS case-definition used for epidemiological surveillance. Representatives from the twelve countries, from Norway, Sweden, and Switzerland, from the Commission of the EC, and from the WHO Regional Office for Europe were present. This discussion was prompted by a proposal from the US Centers for Disease Control (CDC) to revise the AIDS case-definition, and a representative from CDC presented the rationale for the proposed change, which will shortly be published in Morbidity Mortality Weekly Report (MMWR) for implementation on April 1, On
1992. The 1987 definition would be extended to include all patients with CD4 counts below 200/tl, irrespective of clinical manifestations. Although some groups in the USA recognised that the change would result in easier access to health care and other services, the CDC’s main argument is that the proposal would permit better assessment of the number of HIV-infected individuals who need intensive medical care and are at highest risk of serious opportunistic illnesses. The European countries’ decision not to support this change is explained in a statement endorsed by representatives of the fifteen surveillance systems (plus that
of Austria). Five reasons
were
given:
(1) The extensive but varied coverage for medical care and other social benefits in Europe makes the US and European issues of very different. (2) Since all diseases included in the 1987 definition severely affect patients’ wellbeing, the vast majority of these patients will seek health care and will probably be diagnosed as AIDS cases, so completeness of reporting can be assessed quite easily. A definition based on CD4 counts would depend both upon the coverage of HIV testing of persons at risk for infection and upon the proportion of known HIV-infected individuals who have CD4 counts monitored. The completeness of reporting of the number in the total population of HIV-infected persons who have a CD4 count below 200/pl would be very difficult to estimate. (3) Those with ready access to lymphocyte phenotying would be over-represented if the CDC definition were applied, which would distort the contribution of the various risk groups to the picture of the epidemic. Interpretation of trends would be subject to bias, and linkage with data collected under the present definition would be access to care
impossible. (4) Under the proposed CDC definition symptom-free HIVinfected people would be labelled as "AIDS", which would carry psychological and social consequences for them. (5) CD4 cell counting is not yet well standardised. The
working group agreed that monitoring of CD4 is essential for staging HIV disease and for clinical care. However, the group was unanimous in deciding against the adoption of the CDC revision in Europe. The group did agree that surveillance of AIDS cases alone is not sufficient to give a precise picture of the epidemic and that additional surveillance systems should be considered. counts
European Centre for the Epidemiological Monitoring 94410 Saint-Maurice, France
R. A. ANCELLE PARK, Secretary, on behalf of of AIDS,
the EC PHG on AIDS
Epidemiology
Advances in understanding of the cellular mechanisms of disease and of the role of mutagenic agents have created the potential for a complex, but challenging, new area of environmental personal injury cases. The Sellafield leukaemia cases marked the beginning of this era but other cases-eg, legal actions on aplastic anaemia and congenital abnormalities ("CHARGE syndrome") allegedly caused by pesticide exposure of children and/or their parents-are in the pipeline. For diseases which rarely arise spontaneously and in which specific mutagens have been implicated, such as mesothelioma and hepatic angiosarcoma, causation in a legal sense should be easy to establish, but for other diseases a combination of epidemiological, biological, and dosimetry evidence is required to prove causation. The scientists who will be evaluating the strength of such evidence will have to do so in a legal context and that raises the controversial question of standard of proof. The legal standard of proof is far less stringent than scientists are used to. Success in a civil action depends only on "the balance of probabilities", whereas scientific proof more closely resembles the "beyond reasonable doubt" of a criminal case, being based on 95% probability or better. Thus analysis of scientific data by 95% confidence intervals may be inappropriate in legal cases. A higher than 5% probability that an association is due to chance should not in itself prevent a judge from finding an association "proved". In other words associations and observations which are not "statistically significant" may nevertheless be material in law. Statisticians are likely to be employed extensively in the reanalysis of data for the purposes of legal cases. Instead of experts choosing to exclude scientific evidence which is not "statistically significant", all evidence capable of proving a fact with degree of certainty greater than 51% should be made available to the court. The scientific evidence will not be distorted provided greater weight is attached to evidence of a higher certainty. It follows that a plaintiffs lawyers should be cautious about the use of wider confidence limits; the court will invariably have to balance inconsistent scientific evidence, and a finding of low certainty is likely, in the absence of supporting evidence, to be disregarded if there is contradictory evidence of higher certainty. Statistical significance in the context of an environmental or industrial exposure will often be expressed as a relative risk (RR) whose dependence on the lower 95% confidence limit exceeds 1. 51 % proof, on the other hand, equates neatly with a relative risk of 2-ie, the plaintiff must show that exposure to, say, a mutagen more than doubled his risk of contracting the disease in question, and it would not matter if the 95% CI for that RR included 1. The courts-and therefore those who assist them as expertswill have to consider the justification for or desirability of different permutations of confidence interval and RR—eg, an RRof 4 coupledwith a63% CI oroneof2’2and 95% CI. Another legal growth area is likely to be collective actions by groups of individuals who wish to avoid a risk of personal injury. In these cases the legal standard of proof will be even lower than 51 %. The strength of the evidence required will vary according to the severity of the risks and the circumstances, including any beneficial aspects of the activity in question. For example, local residents might
Conference
Medicine and the Law
European AIDS definition
Scientific and legal standards of proof in environmental personal injury cases
Sept 16 and 17, 1991, the European Centre for the Epidemiological Monitoring of AIDS convened a meeting of the European Community Project Management Group (PHG) on AIDS epidemiology to discuss a possible revision of the AIDS case-definition used for epidemiological surveillance. Representatives from the twelve countries, from Norway, Sweden, and Switzerland, from the Commission of the EC, and from the WHO Regional Office for Europe were present. This discussion was prompted by a proposal from the US Centers for Disease Control (CDC) to revise the AIDS case-definition, and a representative from CDC presented the rationale for the proposed change, which will shortly be published in Morbidity Mortality Weekly Report (MMWR) for implementation on April 1, On
1992. The 1987 definition would be extended to include all patients with CD4 counts below 200/tl, irrespective of clinical manifestations. Although some groups in the USA recognised that the change would result in easier access to health care and other services, the CDC’s main argument is that the proposal would permit better assessment of the number of HIV-infected individuals who need intensive medical care and are at highest risk of serious opportunistic illnesses. The European countries’ decision not to support this change is explained in a statement endorsed by representatives of the fifteen surveillance systems (plus that
of Austria). Five reasons
were
given:
(1) The extensive but varied coverage for medical care and other social benefits in Europe makes the US and European issues of very different. (2) Since all diseases included in the 1987 definition severely affect patients’ wellbeing, the vast majority of these patients will seek health care and will probably be diagnosed as AIDS cases, so completeness of reporting can be assessed quite easily. A definition based on CD4 counts would depend both upon the coverage of HIV testing of persons at risk for infection and upon the proportion of known HIV-infected individuals who have CD4 counts monitored. The completeness of reporting of the number in the total population of HIV-infected persons who have a CD4 count below 200/pl would be very difficult to estimate. (3) Those with ready access to lymphocyte phenotying would be over-represented if the CDC definition were applied, which would distort the contribution of the various risk groups to the picture of the epidemic. Interpretation of trends would be subject to bias, and linkage with data collected under the present definition would be access to care
impossible. (4) Under the proposed CDC definition symptom-free HIVinfected people would be labelled as "AIDS", which would carry psychological and social consequences for them. (5) CD4 cell counting is not yet well standardised. The
working group agreed that monitoring of CD4 is essential for staging HIV disease and for clinical care. However, the group was unanimous in deciding against the adoption of the CDC revision in Europe. The group did agree that surveillance of AIDS cases alone is not sufficient to give a precise picture of the epidemic and that additional surveillance systems should be considered. counts
European Centre for the Epidemiological Monitoring 94410 Saint-Maurice, France
R. A. ANCELLE PARK, Secretary, on behalf of of AIDS,
the EC PHG on AIDS
Epidemiology
Advances in understanding of the cellular mechanisms of disease and of the role of mutagenic agents have created the potential for a complex, but challenging, new area of environmental personal injury cases. The Sellafield leukaemia cases marked the beginning of this era but other cases-eg, legal actions on aplastic anaemia and congenital abnormalities ("CHARGE syndrome") allegedly caused by pesticide exposure of children and/or their parents-are in the pipeline. For diseases which rarely arise spontaneously and in which specific mutagens have been implicated, such as mesothelioma and hepatic angiosarcoma, causation in a legal sense should be easy to establish, but for other diseases a combination of epidemiological, biological, and dosimetry evidence is required to prove causation. The scientists who will be evaluating the strength of such evidence will have to do so in a legal context and that raises the controversial question of standard of proof. The legal standard of proof is far less stringent than scientists are used to. Success in a civil action depends only on "the balance of probabilities", whereas scientific proof more closely resembles the "beyond reasonable doubt" of a criminal case, being based on 95% probability or better. Thus analysis of scientific data by 95% confidence intervals may be inappropriate in legal cases. A higher than 5% probability that an association is due to chance should not in itself prevent a judge from finding an association "proved". In other words associations and observations which are not "statistically significant" may nevertheless be material in law. Statisticians are likely to be employed extensively in the reanalysis of data for the purposes of legal cases. Instead of experts choosing to exclude scientific evidence which is not "statistically significant", all evidence capable of proving a fact with degree of certainty greater than 51% should be made available to the court. The scientific evidence will not be distorted provided greater weight is attached to evidence of a higher certainty. It follows that a plaintiffs lawyers should be cautious about the use of wider confidence limits; the court will invariably have to balance inconsistent scientific evidence, and a finding of low certainty is likely, in the absence of supporting evidence, to be disregarded if there is contradictory evidence of higher certainty. Statistical significance in the context of an environmental or industrial exposure will often be expressed as a relative risk (RR) whose dependence on the lower 95% confidence limit exceeds 1. 51 % proof, on the other hand, equates neatly with a relative risk of 2-ie, the plaintiff must show that exposure to, say, a mutagen more than doubled his risk of contracting the disease in question, and it would not matter if the 95% CI for that RR included 1. The courts-and therefore those who assist them as expertswill have to consider the justification for or desirability of different permutations of confidence interval and RR—eg, an RRof 4 coupledwith a63% CI oroneof2’2and 95% CI. Another legal growth area is likely to be collective actions by groups of individuals who wish to avoid a risk of personal injury. In these cases the legal standard of proof will be even lower than 51 %. The strength of the evidence required will vary according to the severity of the risks and the circumstances, including any beneficial aspects of the activity in question. For example, local residents might
