☀️     🌓

Prescribing Advice for GPs

An NHS Prescribing Advisers' Blog

Publication Bias

Studies that are published are commonly positive about the study drug with very few showing a negative outcome. This is because of Publication Bias.

Publication Bias occurs when only certain information, namely positive trial results, is published in journals.

This is an age old tactic as many scientists, upon getting a poor result, blame the experiment (or study) and set off to conduct the study again until they get the expected answer. However, in a well constructed and properly randomised and blinded study this should not be the case and the result perhaps should be made available.

For example, the results of the Serevent Multicentre Asthma Research Trial (SMART) have not been published despite the fact that the study was started in 1996. Action by the Food and Drugs Administration has brought about labelling changes in the United States following safety concerns raised by the study.

Share 'Publication Bias' by emailShare 'Publication Bias' on FacebookShare 'Publication Bias' on TwitterShare 'Publication Bias' on MastodonShare 'Publication Bias' on LinkedInShare 'Publication Bias' on reddit

Manipulate Graphical Data

Data represented in graphs can be manipulated by altering the axes of the graph. There are many ways the data can be altered including:

  • using different scales on graphs that are side by side
  • using log scales to compress the differences
  • using amputated axes to exaggerate the differences
  • origins of graphs not starting at zero

By using these tricks a graphical representation of the data can be altered to make it visually more impressive. The distance between lines on the same graph can be enhanced or reduced.

Look out for enhancements where a demonstration of difference would be expected by the reader, for example in the primary outcome of a study. Look out for reductions where similarities between treatments would be expected by the reader, for example adverse effects.

Share 'Manipulate Graphical Data' by emailShare 'Manipulate Graphical Data' on FacebookShare 'Manipulate Graphical Data' on TwitterShare 'Manipulate Graphical Data' on MastodonShare 'Manipulate Graphical Data' on LinkedInShare 'Manipulate Graphical Data' on reddit

Poor Comparators

One of the easiest ways to show that one drug is effective compared to another is to choose a poor comparator. This can be done in several different ways:

  • compare with a low or sub-therapeutic dose to enhance the effects of the study drug
  • compare with a pharmacologically poor molecule to enhance the effects of the study drug
  • compare with a high dose to minimise the side effects of the study drug
  • compare against a drug or dose not used in current clinical practice

When reading a study ask yourself, "Is this a valid comparison, would I have chosen to use this comparator in the study?"

For example, in the ASCOT Study the comparator arm studied doses of atenolol 100mg and bendroflumethiazide 1.25mg in Hypertension. Atenolol 100mg is no more effective than atenolol 50mg and bendroflumethiazide is used at a starting dose of 2.5mg in the UK.

Share 'Poor Comparators' by emailShare 'Poor Comparators' on FacebookShare 'Poor Comparators' on TwitterShare 'Poor Comparators' on MastodonShare 'Poor Comparators' on LinkedInShare 'Poor Comparators' on reddit

Relative vs. Absolute Benefit

Published studies will often quote risk reductions and almost without exception these are relative risk reductions rather than actual risk reductions. This is because relative risk reductions are usually numerically greater than the actual risk reduction. Larger numbers will look and sound more impressive.

For example, in the FIT 1 study hip fracture rates were decreased from a baseline risk of 2.2% to 1.1% in the treatment arm. This is an absolute risk reduction of 1.1% or a relative risk reduction of 50% - which sounds better to you?

Share 'Relative vs. Absolute Benefit' by emailShare 'Relative vs. Absolute Benefit' on FacebookShare 'Relative vs. Absolute Benefit' on TwitterShare 'Relative vs. Absolute Benefit' on MastodonShare 'Relative vs. Absolute Benefit' on LinkedInShare 'Relative vs. Absolute Benefit' on reddit

Generalised findings

Evidence from trials should be interpreted carefully to ensure that conclusions and recommendations are not applied to groups of patients who were not represented in the trials.

It is inappropriate to extrapolate the findings of a study to a patient group or type who were not present in the study.

For example, some studies of depression are undertaken in severely depressed patients. The effects seen in these trials should not be extrapolated to cases of minor depression as the actual clinical effects in these patients were not covered in the study.

Share 'Generalised findings' by emailShare 'Generalised findings' on FacebookShare 'Generalised findings' on TwitterShare 'Generalised findings' on MastodonShare 'Generalised findings' on LinkedInShare 'Generalised findings' on reddit

Newer Posts »

Prescribing Advice for GPs is powered by ClassicPress.
Connect to our RSS or Atom Feeds.