Trial of applied regression discontinuity to treatment with noradrenaline
Background:
For many treatments and interventions in intensive care there is a scarcity of evidence regarding their implementation in practice. Most of these treatments will not be amenable to study with a randomised controlled trial. The lack of clear evidence leads to widespread variation in how these treatments are implemented. Some of these treatments are implemented around arbitrary thresholds for starting or changing treatment.
E.g. hypotensive patients may be commenced on a vasopressor such as noradrenaline. There is limited evidence to suggest what is the correct blood pressure to commence this treatment at, or what blood pressure to aim whilst on treatment.
It is accepted practice throughout many intensive care units that a target blood pressure to aim for is a mean arterial pressure value of above or equal to 65 mmHg.
If clinical behaviour clusters around specific cut off thresholds then these can be utilised to apply local randomisation to observations around the cut point. Using a regression discontinuity design the treatment effect of noradrenaline applied at the threshold value can be estimated.
Aims:
- Identify if clusters of behaviour exist with regards to threshold blood pressure values and noradrenaline treatment.
- Identify if a regression discontinuity design can be applied to the treatment threshold to estimate the treatment effect.
- Identify if there is variation in outcome with alteration in treatment threshold value.