[SystemSafety] Muddled thinking punctures plan for British ventilator

Olwen Morgan olwen at phaedsys.com
Fri May 22 21:55:12 CEST 2020

There are ventilators and ventilators. AFAI can make out from the 
numbers I've looked at, around half of the people who need top-of-the 
line ventilators will die anyway - there's only so much you can do in 
the face of a cytokine storm. Others will do perfectly well on the much 
simpler CPAP devices. Of the three approaches that Dewi mentions, 
ramping up production of CPAP machines is probably the most effective 
/single/ measure. As regards the other two options:

* option 1 cannot be effective in the short-to-early-mid-term because of 
the /systemically raised global demand/ that a pandemic causes - this 
one will take time to produce results yet could even, in the longer 
term, result in a glut on the ventilator market (probably good news for 
ICUs in China ;-)

* option 3 is likely to take longer to produce results than option 2.

I say this after having worked a few years ago for a manufacturer of 
ventilators and would point out "production lines" did not really exist 
for that company. Ventilators are simply not mass-produced as cars are. 
If you wanted to scale up production, you simply needed more floor space 
filled with assembly benches and basic electronic test equipment, and 
more workers to do assembly and, above all, stringent testing. My guess 
is that the critical factor in such a scale-up would be engineering 
staff with the requisite QA and testing skills for medical devices (such 
testing is not particularly easy or cheap to automate). That being the 
case, one would not, IMHO, want them to be diverted to the development 
of new designs.

And this is to take a first-world-centric view. Globally, the real time 
bomb is in Africa. Slower spread of the disease in Africa is, AFAI am 
aware, likely to be due to the demographic profile (heavily skewed 
towards having young populations) and possibly - just possibly - the 
effect of a warmer climate on virus transmission. On the other hand, 
even in these conditions and with the limited testing capability of most 
African states, it is clear that the R number exceeds 1 and therefore 
that the number of infections will grow exponentially and eventually 
overwhelm health systems that have nowhere near first-world 
capabilities.The worst-case effects of this in the near-coastal parts of 
the Sahel are pretty ghastly (see the stats and maps in /New Scientist/ 
and /The //Economist/). Massive final excess-death counts are far from 

And while I'm on death counts, the current US death toll exceeds 95,000, 
which is greater than the total of US military deaths in Korea and 
Vietnam combined. At this rate the number will easily rise to 117,000, 
which exceeds the total of US military deaths in WW1 - of which, 
ironically, more than half were due to Spanish Flu!

Finally, after the above eclectic ramble, a simple control-chart-like 
technique provides an insight into whether the epidemic is coming under 
control. Plot a three-day moving average of new cases against existing 
cases, both variates on logarithmic scales. While the spread is 
exponential, this will give a straight line. As the epidemic comes under 
control, new cases fall precipitously below the straight line - as has 
been seen in China, South Korea and, AFAI am aware Taiwan. There's a 
good video of this on the /Vox/ YouTube channel and I daresay someone 
more adept at statistics than I am could turn this sort of thing into a 
proper control chart with warning and confirmation limits.



On 20/04/2020 10:23, Dewi Daniels wrote:
> I've only had a quick glance at the Twitter thread, but I've read the 
> government rebuttal.
> The FT article suggests that it was naive to try to develop new 
> designs and that it would have been better to focus on manufacturing 
> existing designs under licence. The government rebuttal points out 
> that the government's strategy focused on three pillars:
> 1. procuring more devices from existing manufacturers overseas
> 2. scaling up production of existing ventilator suppliers
> 3. working with industry to design and manufacture new devices.
> This three pillar approach seems very sensible to me.
> I think the FT article underestimates the difficulty of setting up a 
> new production line to manufacture an existing design under license. 
> The intent of the third pillar is presumably to design simple devices 
> that serve a limited purpose, but which can be manufactured quickly 
> and cheaply.
> If some of the suppliers chose to create simple, low cost designs 
> that satisfied the bare minimum function, and these designs proved to 
> be unsuitable, surely the fault lies with the specification and not 
> with the implementation? The FT article seems to suggest that 
> suppliers with experience of medical devices would have known that the 
> requirements were incomplete. I think that greater care in specifying 
> the requirements would have avoided nugatory work. Nevertheless, I 
> don't think the problems with some of the early designs invalidate the 
> approach as claimed by the FT article; rather, the specification 
> should be updated, which is exactly what's happened.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.techfak.uni-bielefeld.de/pipermail/systemsafety/attachments/20200522/e750a441/attachment.html>

More information about the systemsafety mailing list