Abstracts fra Bibliotek for Læger 2/2015

Originalartikel: Under Hensyn til den nuværende Situation …
- Retsmedicin i den politiløse tid 1944-1945
Hans Trier

Danish coroners and forensic pathologists in the period without police 1944-45

Bibl Læger 2015;207:92-121.

This paper investigates how Danish coroners, forensic experts and legal authorities of the state reacted to substitute the missing police during the last seven and a half months of the German occupation when the police was deported or went underground. Medical officers of health and forensic pathologists cooperated with public prosecutors and substituted the police by establishing themselves or involving other local officials as interrogators of witnesses in connection with the coroner‘s inquests. The Ministry of Justice reacted late and only after having been informed about an already established local solution. This solution was copied in a national circular two months after the police had disappeared, assigning the coronors and forensic pathologists extensive police authority. The level and quality of forensic activities was maintained fairly well, except for cases of »political murders«, where the gathered information was seemingly scantier. However, studies of literature and archives reveal that doctors and other employees at the Institute of Forensic Medicine actively gathered supplementary information about these murders, but kept some of it hidden and omitted from the official documents. It was handed over to the underground criminal police and sent to the reestablished police force after the liberation for the purpose of investigation.

Et billede fra min hverdag
Kirsten Boisen

Originalartikel: Forsøgssekventielle metaanalyser i systematiske oversigter
Jørn Wetterslev, Janus Christian Jakobsen & Christian Gluud

Trial sequential analysis of meta-analysis

Bibl Læger 2015;207:124-151.

Most systematic reviews with meta-analyses, including Cochrane systematic reviews, do not have sufficient statistical power to detect or refute even large intervention effects. Meta-analyses should accordingly often be regarded as interim-analyses on the way towards the required information size. The evaluation of meta-analyses has to compare the total number of randomised participants with the required meta-analytic information size considering statistical diversity. When the number of participants in a meta-analysis is less than required, based on a realistic intervention effect, the constant application of a traditional 95% confidence interval or a 5% statistical significance threshold will lead to too many false positive and false negative conclusions. The Lan-DeMets sequential monitoring boundaries in Trial Sequential Analysis offer adjusted confidence intervals and restricted thresholds for statistical significance when the diversity-adjusted required information size for the meta-analysis has not been reached. A Bayesian meta-analysis, using prior distributions for both the intervention effect and the statistical heterogeneity, may be even more reliable to decide whether an intervention effect is present or not. However, Bayesian meta-analyses also raise a number of difficulties with interpretation and until easy-to-use software programs for full Bayesian meta-analysis are accessible, trial sequential analysis represents a more assumption transparent analysis than the use of traditional meta-analysis with unadjusted confidence intervals and unadjusted thresholds for statistical significance.

Opdateret litteraturliste

Kvartalets genstand
Morten A. Skydsgaard

Originalartikel: Et klinisk etisk dilemma

- Case behandlet i en klinisk etisk komité:omsorgssvigt versus tvangsbehandling
Henriette Bruun

Case analysed in a clinical ethics committee

Bibl Læger 2015;207:154-168

Through the presentation of a concrete ethical dilemma, this article presents the praxis of a clinical ethics committee in Denmark. The dilemma was analysed in the clinical ethics committee of psychiatry in the Region of Southern Denmark. The dilemma is well known in psychiatry; it concerns the conflict between respect of autonomy and the risk of neglect. The patient is a young woman with schizophrenia and psychotic symptoms, which have re-emerged because of stress. The woman has no insight in her own illness and she does not want to receive antipsychotic medication for her psychotic symptoms. However, she agrees to meet with the community treatment team. How are they to act? Should they accept her refusal of antipsychotic medication, though they see it as the best treatment?This dilemma is analysed and described according to Beauchamps and Childress’ four fundamental principles: 1. the non-maleficence principle, 2. the beneficence principle, 3. the principle of autonomy and 4. the principle of justice. Common values in Danish psychiatry: respect, proficiency and responsibility are included in the description.

Originalartikel: Mendelsk randomisering
Chunsen Wu & Jørn Olsen

Mendelian randomization

Bibl Læger 2015;207:169-184.

Randomized controlled trials (RCTs) are often considered the best design to establish causal evidence but RCTs are not always feasible or ethical. The name “Mendelian randomization” is a method based on Mendel’s two laws of inheritance. Mendelian randomization was suggested by Karan in a letter to the editor of Lancer in 1986. Observational studies showed low serum cholesterol to be associated with an increased risk of cancer but it was not clear whether low serum cholesterol was causally associated with cancer risk or due to reverse causation. For example, early stages of cancer may lead to a decrease in serum cholesterol levels. If low serum cholesterol is causally associated with cancer risk, then genotypes associated with low serum cholesterol level should also be associated with a high cancer risk. If the association is based on reverse causation then the genotypes predicting a low level of serum cholesterol will not influence cancer risk. Results indicated this prediction to be true. Mendelian randomization can estimate causal association under three conditions. 1) Genetic variants are associated with the exposure. 2) Genetic variants are not directly associated with the outcome. 3) Genetic variants are not associated with other causes of the outcomes. Mendelian randomization is a "deconfounding method" and a method to avoid reverse causation. But Mendelian randomization also has limitations if genetic variants are directly linked to outcomes; if pleiotropic effects exist. Mendelian randomization is an important tool for epidemiologist researchers, especially in studies where a lifelong exposure is of interest. With the rapid development of genome wide association studies and a better understanding of what these genes do, there will be many more opportunities to apply this design in the future.