StatsMiniBlog: Surrogate, proxy or process?

20140205-091454.jpgWhen we look at treatments for children’s ill health, we tend to be stuck in a pleasant dilemma. We normally want to use treatments to stop kids dying, and to make them better quicker – but it’s uncommon for children to die, and they’ll often get better anyway. So we end up using alternative outcome measures; a proxy, surrogate or process measure. Do you know what the difference is?

A proxy outcome refers to something we measure in an attempt to get at an intangible – so quality of life measures attempt to capture the truth of how someone’s life is and affects them. You’re often in a state where there are a few ways of trying to understand the intangible, so you might find a number of different measures being used to triangulate onto the ‘truth’ of the unmeasured.

A surrogate measure is something which the study folk believe is intimately related to the outcome we really want to measure, because it’s too rare, or takes to long to occur. Some surrogates appear to be robust; looking at the amount of minimal residual disease in acute lymphoblastic leukaemia really really does relate to survival. Others aren’t as good though – examples might be bone mineral density in preventing fractures from osteoporosis (this was shown to be incorrect when fluoride was used – great at increasing BMD and even better at increasing fracture risk), or length of time on a ventilator and neurodevelopmental outcome after the neonatal period.

A process measure is where we’re just seeing what’s been done, and this may, or may not, have any relationship with the outcome. Take duration of hospitalisation and … well … almost anything.

It’s important to know what is being measured, what you want to really know (What would Jack want?) and how those things are related. Understanding these things can get us closer to making clinical research applicable, meaningful and replicable.

– Archi

(Visited 1,356 times, 1 visits today)