Jim’s prediction of the DESiGN trial results & why he won’t believe ’em

DESiGN (DEtection of Small for Gestational age Neonate) is a cluster randomised trial comparing the Perinatal Institute’s GAP/GROW protocols (click here) with standard care for detection of small for gestational age babies.

There are good things about it. It was registered here ISRCTN 67698474 on 2 November 2016, the protocol was published in March 2019 (click here) and the triallists include some very big names, Matias C. Vieira, Sophie Relph, Andrew Copas, Andrew Healey, Kirstie Coxon, Alessandro Alagna, Annette Briley, Mark Johnson, Deborah A. Lawlor, Christoph Lees, Neil Marlow, Lesley McCowan, Louise Page, Donald Peebles, Andrew Shennan, Baskaran Thilaganathan, Asma Khalil, Jane Sandall, and Dharmintra Pasupathy.

But DESiGN won’t tell us anything about the benefits or harms of customisation itself.

DESiGN is comparing units using GROW customised software with units using an unspecified local population chart. The intervention group are also receiving training in measurement techniques and in how to respond to abnormal values. Many control units are apparently not even using a chart for fundal height measurement. Instead fundal heights are “approximated to the gestational period” in that the number of centimetres is expected to approximate to the gestational age in weeks (± 2–3 cm). This rule aligns with no fundal height chart ever.

At best DESiGN will tell us that it’s better to train people to measure fundal height properly, plot it on a chart and act on the results, than to not train them, let them ignore fetal size altogether, measure it by palpation alone or tape measure, plot the values they get on any old chart, or approximate them to an erroneous (± 2–3 cm) simplification, and act on the result according to whim. The former will likely be better, but it will hardly prove the benefit of customisation.

Perhaps my subheading above is unfair. Perhaps DESiGN is simply a pragmatic evaluation of GAP/GROW against current real world practice, with no implications for the customised chart debate. If so, the triallists will hear no more from me.

But the protocol suggests that at least some of the authors hope to draw implications about the value of customisation. In that case, I look forward to reminding them that it won’t.

I’ve not seen any results, so my date-stamped prediction (click here or Jim’s DESiGN prediction) is data independent.  We shouldn’t have to wait long.  According to the protocol the last sites were randomised in July 2017 with outcome data to be collected till 31 November 2018. According to the registry the intention to publish date is now 28 February 2020.

For more on AsPredicted.org click here.  For the customised v standard chart debate click here.

Jim Thornton