Re: Ensuring the integrity of clinical practice guidelines: a tool for protecting patients
The underlying problem is that clinical guidelines are very susceptible to bias because the conventions of evidence for agreeing to whom a treatment or other intervention should apply are weak. This results in a subjective ‘free-for-all’ with over-diagnosis, over-treatment, under-diagnosis and under-treatment [1].
This can be overcome by analysing clinical trial results carefully by applying evidence-based ‘triage’ to identify those groups in the trial (1) who improve without active treatment, (2) who are helped by treatment and (3) those who fail to be helped. Having identified the evidence-based treatment indication criteria, these can be gathered together to form diagnostic criteria. This idea, illustrated by the analysis of a large international clinical trial [2], was described at the recent conference supported by the BMJ on ‘Preventing over-diagnosis’ hosted by Dartmouth College in New Hampshire [1].
Sensitivity, specificity and the resulting overall likelihood ratio were used originally when radar was invented and they are also helpful when analysing the performance of screening tests. However, these indices are unhelpful for assessing tests for differential diagnosis. In this setting it is findings associated with a short list of differential diagnoses that are needed and also findings that occur commonly in some of these differential diagnoses and rarely (or never by definition) in others [3].
Until these issues are addressed, clinical guidelines will be highly subject to bias. The same mistakes could also be made all over again about 'stratified' or 'personalised' medicine and progress held back.
References
1. Llewelyn D E H, Garcia-Puig, J. How different urinary albumin excretion rates can predict progression to nephropathy and the effect of treatment in hypertensive diabetics. JRAAS, 2004, 5; 141-145.
2. Moynihan R, Glasziou P, Woloshin S, Schwartz L, Santa J, Godlee, F. Winding back the harms of too much medicine, BMJ 2013;346:f1271.
3. Llewelyn H, Ang AH, Lewis K, Abdullah A. The Oxford Handbook of Clinical Diagnosis, 2nd edition. Oxford University Press, Oxford, 2009, pp754 -760.
Rapid Response:
Re: Ensuring the integrity of clinical practice guidelines: a tool for protecting patients
The underlying problem is that clinical guidelines are very susceptible to bias because the conventions of evidence for agreeing to whom a treatment or other intervention should apply are weak. This results in a subjective ‘free-for-all’ with over-diagnosis, over-treatment, under-diagnosis and under-treatment [1].
This can be overcome by analysing clinical trial results carefully by applying evidence-based ‘triage’ to identify those groups in the trial (1) who improve without active treatment, (2) who are helped by treatment and (3) those who fail to be helped. Having identified the evidence-based treatment indication criteria, these can be gathered together to form diagnostic criteria. This idea, illustrated by the analysis of a large international clinical trial [2], was described at the recent conference supported by the BMJ on ‘Preventing over-diagnosis’ hosted by Dartmouth College in New Hampshire [1].
Sensitivity, specificity and the resulting overall likelihood ratio were used originally when radar was invented and they are also helpful when analysing the performance of screening tests. However, these indices are unhelpful for assessing tests for differential diagnosis. In this setting it is findings associated with a short list of differential diagnoses that are needed and also findings that occur commonly in some of these differential diagnoses and rarely (or never by definition) in others [3].
Until these issues are addressed, clinical guidelines will be highly subject to bias. The same mistakes could also be made all over again about 'stratified' or 'personalised' medicine and progress held back.
References
1. Llewelyn D E H, Garcia-Puig, J. How different urinary albumin excretion rates can predict progression to nephropathy and the effect of treatment in hypertensive diabetics. JRAAS, 2004, 5; 141-145.
2. Moynihan R, Glasziou P, Woloshin S, Schwartz L, Santa J, Godlee, F. Winding back the harms of too much medicine, BMJ 2013;346:f1271.
3. Llewelyn H, Ang AH, Lewis K, Abdullah A. The Oxford Handbook of Clinical Diagnosis, 2nd edition. Oxford University Press, Oxford, 2009, pp754 -760.
Competing interests: No competing interests