Seattle is apparently not the only Effective Altruism group to talk about doing volunteering meetings, only to remember that the traits that make volunteering useful are almost antithetical to the traits that make it fun and doable to for a group on a drop-in basis. I am kind of hoping that blood donation can bridge that gap. So here’s my math on how effective donating blood is. The Red Cross estimates a single donation can save three people, but what they mean is “a single donation can go to three different people.” To get the actual value we need to see how many units of blood were donated and how many deaths they prevented.
The most recent data I could find was the 2011 National blood Collection and Utilization Survey Report (PDF), which couldn’t make it harder to do this kind of calculation if it tried. They were extremely loose with what “unit” referred to, so I’m going to stick with the whole and red blood cell transfusions, so my numbers are consistent. There were 15,721,000 units collected, of which 14,589,000 were deemed usable. 13,785,000 were used, of which 37,000 were directed to a specific patient, and 65,000 were self-donations, which are less effective for various reasons. The collections numbers don’t call out general vs. specific donations and the numbers are small, so I’ll just use the total number used. If some blood donations are also generating plasma and white cells in addition to the red blood cells counted here, that would only increase effectiveness.
A single donation is one pint. Health and Human Services fails to define what they mean by unit, but it appears to mean “whatever you get from one donation after some filtering“, so let’s assume it’s 1:1. The average recipient receives 2.75 units. If you assume each person who received a transplant would otherwise die (supported by this sourceless FAQ), that means each donation saves ~1/3 of a life (discounting for donations that are rejected). Using GiveWell’s $5,000/life number, that’s still equivalent to donating $1,667. That is overstating the case, because some portion of recipients (I can’t find out how many) have diseases like sickle cell anemia that require chronic transfusions, and the fair thing is to count their lifetime transfusion count, not their per treatment count. To get an upper bound I’ll use the Red Cross’s number that a car accident victim can use up to a 100 pints of blood, which means each donor saves 1% of that life, which is equivalent to $50 to an extremely effective charity.
But the question isn’t “what is the average value of donated blood?” but “what is the marginal value of your potential donation?” I can’t find any direct numbers for this, but we have the following evidence:
- Very little blood is thrown out.
- People are spending lots of time and money developing artificial blood substitutes. Despite this there are no generally accept substitutes for blood’s oxygen-carrying capacity.
- The Red Cross spends a lot of time and money harassing people to donate. They called my parents’ house for years after my one donation (I’m O-).
- Some blood is able to reach the “too old” state, but then used to ill effect, indicating lumpy supply or demand. Unless you can predict demand spikes you should use the average efficacy. If you can predict demand spikes, there are probably more effective things to do with that power.
So I’m just going to use the average effectiveness as the marginal effectiveness for now.
What are the costs to the donor of donating? The one time I donated it was high because I slept for the next two days. If you’re my friend Elena who went into shock after donating, it cost you days and several thousand dollars in ER visits. So it is probably not worth it for either of us to donate. But for a typical person with no side effects, it’s plausibly useful. If it’s replacing work time, then effectiveness depends on their hourly wage. Multiple websites list the time to donate as 60-90 minutes, which translates to a minimum psuedowage of $33/hour and a maximum of $1667. The average hourly American wage is $24/hour, although I would estimate the average wage of people earning to give as somewhat higher than that. So that’s extremely plausible on its face. But if the time isn’t coming out of work, and is made rewarding to the participant, blood donation is hugely effective. This suggestions that an event that induced people to donate without replacing work would be effective, more so if it could be made into a positive experience. So a blood donation event could be a huge win for an EA event.
[Side note: if you decide to do this yourself, I would recommend donating anywhere but the Red Cross if at all possible. I’m going to try for Bloodworks NW, because if I get enough people they will send a truck and we can make it an actual party]
Aceso Under Glass 
My understanding (from some old LessWrong thread about easy health stuff) is that [cis] men also see health benefits from regular blood donation. Apparently [something I’ve forgotten] builds up in blood, and menstruating keeps it less of a problem from people who do that.
I *think* this thing I forgot was about iron: http://www.medicaldaily.com/why-donating-blood-good-your-health-246379
I like this post, but I think the conclusion is wrong because the assumption that the impacts of marginal contributions are the same as average ones is false. Amongst other reasons, my impression is that blood banks frequently start emergency drives when demand spikes, and I doubt that they’re failing to take advantage of opportunities to save lives for anything between 3 and 100 units. In particular, this 2006 Slate story says that the Red Cross is paid something like $150 for each unit: http://www.slate.com/articles/news_and_politics/explainer/2006/09/the_business_of_blood.html
If the marginal benefits were anywhere near as big as you’re estimating (~1% of a life) the blood is radically underpriced. Although that’s conceivable, my understanding is that the Red Cross is compensated for costs rather than in a market, which if anything would suggest that their costs are “too high” relative to the real marginal costs of supplying blood like that. Another way to put this is that common estimates in the US value a statistical life at ~$5 million, so you’re implying that the social value of a marginal blood donation is ~$50,000. That seems implausible – if that were the case, it would be worth paying for far more outreach (a fulltime person could justify their salary just by recruiting one donor a year!), etc. The level of outreach you describe (and I’ve seen) seems much closer to something where the social value is <$100 (since the Red Cross is also being compensated for numerous other expenses with that ~$150: http://www.reddit.com/r/todayilearned/comments/19ejp7/til_the_red_cross_sells_the_blood_that_people/).
I think the right conceptual approach for an EA to decide whether to donate is to figure out what the marginal recruitment cost for a bloodbank is and then donate if you would if you had no impact but someone paid you that much. However, that approach assumes perfect altruism (i.e. taking any action where the benefits to someone else exceeds the cost to you); most EA actions (like donations to top charities) have vast multipliers on the benefits to others versus the cost to you.
Marginal vs. average cost was definitely the most challenging part of this. This was originally written for the context of “is blood donation better than volunteering as an EA social gathering?”, for which the benefit bar is much lower, but there appears to be a fair amount of interest so I’m trying to get answers from my local blood bank. Until then, a couple of points:
I believe $150 is low for the cost of blood. This survey (http://www.ncbi.nlm.nih.gov/pubmed/21174480) has the price for red blood cells and platelets to medical facilities and patients as $US210.74 ± 37.9 and the mean charge to the patient (n = 167) was $US343.63 ± 135. for plasma it’s $US60.70 ± 20 and $US533.90 ± 69. There’s no number for platelets. I’m unclear if that’s red cells and platelets combined or if it’s that much each, but that gives a range for total value-to-patients(‘-insurance) of a pint of whole blood at $US673.35 – $US1020.1.
My understanding is Red Cross probably runs a profit on blood sales and spends it on disaster relief, although no one is sure because their books are so bad. Blood Works NW reports that they would make a small profit if operating at capacity, but on average lose money and are sustained by monetary donations. This suggests patients or hospitals are capturing most of the utility.
The blood used to treat mass disaster victims is already stocked when the disaster happens (http://www.nydailynews.com/life-style/health/donating-blood-disaster-best-time-weeks-months-article-1.1318535). The banks opportunistically use disasters to buff up their stores, and it’s frequently wasted (so my estimate for post disaster donation effectiveness would be much lower).
Blood banks report an incredibly small buffer (http://www.americasblood.org/stoplight.aspx) (22% with 1-2 day supply as I’m writing this). I’m skeptical that it’s actually that low, but believe that there’s room for more. The report I linked to in the post showed a vanishingly small amount of usable blood being thrown out.
I’m waiting to talk to the blood bank on this, but it sounds like the result of not enough blood isn’t “car crash victims left to bleed out” but “urgent but not emergency surgeries postponed”, and I’ll bet there’s also “doctors become more conservative with blood”. The cost of that is harder to calculate.
The red cross and possibly other banks *do* spend money soliciting people to donate and making it easy to do so (e.g. blood mobile). If I can get numbers on that I will share them. The complication there is that I suspect frequent harassment makes people more likely to quit giving altogether, so giving blood now may lead to someone else giving more blood in the long run.
There is some evidence that purchased blood is more likely to contain infections, some undetectable, than donated blood (http://www.ncbi.nlm.nih.gov/pubmed/12437514). You could get around this the way sperm banks do, by testing donors again when the window period is closed, but that is much more expensive and lowers the useful lifespan of the blood.
*Someone* has to donate the first and last pints of blood, and it seems weird that they get different amounts of utility credit for it. The math is also a nightmare. I’d rather look at room-to-give, possibly with multiple categories (e.g. blood necessary for emergencies and chronic recipients, blood necessary for urgent surgeries, blood that lets doctors be less conscious of how much blood they are using), and assign utilities to each of them. The fact that we use low quality blood and have so little waste suggests to me we’re still in a category where blood is highly useful, if not *as* useful as the first pint of blood to a car crash victim.
Good find on the better cost data: http://www.ncbi.nlm.nih.gov/pubmed/21174480
Even taking the high end sum of roughly $1,000 for a unit of blood (which I would guess is 10-50 times the marginal cost of recruitment for a blood donation), your stated lower bound on the effectiveness of blood donations is 50x the implied social benefits, which I find hard to square. What do you think explains the lack of social investment in blood acquisition if the benefits are as big as you suggest?
I’m trying to get better numbers, either on recruitment costs or the investment in artificial blood research, so I can answer that more precisely.
Speculatively: paying cash outright is heavily stigmatized, due to perception it lowers donor quality (testing isn’t a complete answer because there’s a window where infections are undetectable). There is also a perception paid donation supplants voluntary donation. Whether these are actually true is irrelevant, the perception of truth is enough that blood centers don’t want to pay outright. They are experimenting with other incentives, up to waving $75 court fines or a day off of work (in Italy, and there’s no info on the wages of the donors)
I tried the following thought experiment: Assume the worst happens, all blood must be purchased, and the supplier must be retested at a later date before the blood is released. Sperm donors do that, and the rate is 35-50 per donation, plus ~$25 on showing up for retest. At those rates payment would be a fraction of the cost of blood, so paying everyone is a viable option. Unfortunately blood doesn’t last long enough for that to be useful. It looks like there’s some tech that extends that almost indefinitely (http://thebloodconnection.org/products-services/laboratory-services/frozen-blood-program/), it looks like the same tech that is used in auologous blood donations, which is markedly inferior to fresh blood. One option might be a tiered system, where paid donations must be iced and and the donor retested, and unpaid donations needn’t. That might also avoid the replacement effect, because unpaid donors get a glow from their blood being more valuable than unpaid donors.
In conclusion: I have no problem concluding the Red Cross is just dumb, but I’m working on getting actual numbers to prove it either way.
A large RCT on the impact of giving $5-$15 gift cards to blood donors in Ohio calculated marginal costs of a unit of blood between $10 and $66: http://economics.yale.edu/sites/default/files/files/Workshops-Seminars/Labor-Public/macis-110929.pdf (last row of Table 12)
Huh, I assumed the cost was much higher. I’m still confused why they let themselves run that small a margin for error if more is so cheap, but it does look like they could get more if they wanted.