Thread by @technocrypto

Back of napkin math shows that funding oxygen intervention for India is currently more effective than even top rated interventions from @GiveWell

Thread below.

@Effect_Altruism @MedCrisis @GiveIndia @VitalikButerin @juliagalef @KellyBEworks @robertskmiles @JaEsf

1/X

Hat tip to Dr. Rohin Francis for identifying the specific opportunity in his excellent video on the crisis India is currently facing:

https://twitter.com/MedCrisis/status/1387428737583550468

I've used numbers from covid.giveindia.org/healthcare-her… because they were very specific and can handle int'l donations easily.
2/X

They claim able to deploy funds within 1-2 weeks. Where details were lacking I checked with an MD I know who has been treating COVID in northern Canada to estimate the impact of various interventions, assuming effective triage and that each item is already a choke point.
3/X

My starting assumptions were:

Avg patient age 50 yo, giving post-survival life expectancy of ~25 years

70 ₹ to 1 USD

Avg 28 oxygen-days needed to save a patient

40 USD to add one year of life expectancy via Givewell top charities

4/X

All assumptions highly conservative. Exchange rate is pre-adjusted to cover payment processing and conversion. Patients are actually skewing younger right now. 4 weeks of oxygen to save just one patient actually assumes triage inefficiency. 40 USD is cheapest estimate given.
5/X

Next I took all listed items and calculated the number of days the equipment needs to be in usage to beat GiveWell cost effectiveness. Results:

Oxygen plant: 37 days
Oxygen concentrators: 22 days
Bipaps: 72 days
Ventilators: 600 days

For the first 3 these are rock solid.
6/X

For oxygen tanks there are recurring costs. If we assume all costs borne by the donor we get:

B type oxygen cylinder: 33 days
D type oxygen cylinder: no breakeven

However if refills are paid for locally we have to unadjust the exchange rate, giving:

B: 30 days
D: 180 days

7/X

If we were to further claim that local refill payments are unlikely to compete with charitable giving, and just remove them entirely, we get:

B type: 12 days
D type: 10 days

Note that this assumes that only one of plants or cylinders limits care, rather than both.
8/X

In summary:

Almost all interventions planned by this oxygen campaign will outperform GiveWell top charity recommendations given highly conservative assumptions about effectiveness, length of crisis, etc.

Ventilators and D type cylinders are the weakest interventions but...
9/X

...given the long service time of this equipment even these are likely to prove relatively effective in the event that all chokepoints for other interventions could be met. India also has a strong record of redeploying unneeded equipment to other countries for later crises.
10/X

Please check my math and support via ACH or credit card if you agree: covid.giveindia.org/healthcare-her…
There is ~600k USD of funding room left in this campaign. If met we can continue down Rohin's list. I welcome corrections regarding rates/assumptions/etc. My full working is below.
11/X

Here are the breakdowns for how I did each calculation. The full .ods spreadsheet can be downloaded from file.io/EDsyrhSJMDWA to check the underlying math. Images assume no oxygen cylinder refill cost to donors, but spreadsheet does not.
12/X

Effectiveness is hard to judge, I tried to estimate both the risk of a patient who would need that intervention and the amount of impact the intervention would have. Ventilators fair poorly because patients on them die up to half the time in spite of multi-week treatment.
13/X

The least obvious assumption I made is probably that D type oxygen tanks would be used for high flow rate treatment exclusively, so even though they are 4 times bigger they run out 4x as fast, while only being twice as effective at saving lives as B type cylinders are.
14/X

Oxygen concentrators come out looking really good in all-cost effectiveness, but of course it probably comes down to what is actually available and how quickly it can be produced/obtained.
15/X

An important question here is what the marginal impact of donation actually is. E.g. will the Indian government step in to sufficiently fund oxygen supply to the point where it is logistically rather than financially limited? I welcome any insight that anyone can offer here.
16/X

Note that the speed of government response is a major factor. If the government can't buy all available oxygen supply at these prices within ~5 weeks, then donating will still outperform it for some of the interventions, assuming they are locally available within 2 weeks.
17/X

It's a glaring omission, but for crypto donations see @CryptoRelief_ as well! Reliable org who has already deployed 1M USD worth of funds directly towards oxygen concentrators:

https://twitter.com/sandeepnailwal/status/1388813415309737986

I'm just trying to draw in the #EffectiveAltruism community as well!

18/X