Skip to content →

Bad Blood: Review

Bad Blood by John Carreyrou was the most unputdownable book that I read in 2018.

The book reveals that the truth is sometimes stranger than fiction.  Reading the astounding details on Theranos, once a revered Silicon valley unicorn and a health care startup, I had to constantly remind myself that many narrated incidents did take place, and the names are real people that walk among us in flesh and blood today.

The indelicate machinations of the principal actors in the book are comparable to the over-the-top villainy that we read in airport thrillers and potboilers. Some of them are composites of miscreants in Robin Cook’s medical thrillers and John Grisham’s legal thrillers.1 The victims carry the quiet fatalism of the characters that come under harm in Le Carre novels – decent, honest people placed in precarious situations and then silenced, displaced, or demolished in their careers.

A movie is in the works. It will be an excellent screenplay, I think — sort of a cross between Spotlight and Michael Clayton. But, read the book for the true details.

Bill Gates in his review writes,

Some of the details he shares are—for lack of a better word—insane. Holmes would invite prospective investors to the lab, so they could get their blood tested on a Theranos machine. The device had been programmed to show a really slow progress bar instead of an error message. When results didn’t come back right away, Holmes sent the investors home and promised to follow up with results.

As soon as they left, an employee would remove the blood sample from the device and transfer it to a commercial blood analyzer. Her investors got their blood tested by the same machines available in any lab in the country, and they had no idea.

Details are indeed insane.  I have been following Carreyrou’s articles in WSJ’s since the heyday of Theranos.  I would say it was possible to be skeptical or bearish about the halo around the founder and the aspirational stories of “saving” people.  However, it truly beggared belief that the high valuation of Theranos was built by sustained acts of chicanery.

It is particularly disappointing that the investors did very little due diligence. There held the money, and arguably it was in their interest to ask probing questions. Far from probing questions, their actions were a weird mixture of motivated reasoning, learned helplessness, and an unwillingness to be open to dis-confirming ideas that came from the field.

In the face of such low odds and high stakes, Carreyrou’s pursuit is an ultimate act of brilliant journalism. His dogged pursuit of truth revealed that the Theranos house was built on a pack of lies.  Lies that were not just sophisticated non-denial denials and overstatements, but deliberate errors of commission and misdirection. There was simply no product. The service was simply nowhere near its promise. John Carreyrou deserves our admiration.2

In the end, the Theranos episode is a tragic story with many victims, ruined careers, and dashed hopes.  Ian Gibbons’s career was a heart-rending tattered legacy, slowly teetering to a tragic end.

I was heartened by those who chose the right action for various reasons. Particularly admirable are the whistleblowers, Tyler Schultz and Erika Cheung, then young 20-year olds, who despite their extraordinarily different backgrounds and life-journeys, made the same principled decision. They still seek to make a difference. Such talent and honor are what makes us hopeful of future improvements in health care delivery.

I can’t speak credibly about the Silicon Valley culture bubble or the lack thereof, but I wanted to focus on a couple of interesting Operational findings -particularly related to due diligence (and the lack thereof in the case of Theranos).

Flowtime vs. Throughput Confusion

[pg. 167] Flowtime of a testing process is often based on biochemistry, and cannot be re-engineered. This is often why commercial machines that run multiple blood tests are bulky. We need space and bulk for high throughput, as faster throughput is achieved by running multiple tests in parallel with more blood. Theranos’s objective was a thinly sliced design with very little blood that meant low-throughput — which in turn meant longer wait times for consumers.  To sum up, smaller product size meant longer processing times, as there is not much parallel processing of tests.

A smaller blood sample (by volume) meant that the tests are susceptible to “local” chemical variations in the blood, which means the tests needed to highly reliable and more accurate. (In the next paragraph, I will talk a bit about the precision of tests.).

It was amazing to me that this “first-order” tradeoff was not studied by investors.

Coefficient of variation

Given the natural variation in samples and the importance of the test results, it is important for the testing to be precise. To measure the precision of test results, various readings are conducted on the sample, and the coefficient of variation  (CV) of the data is calculated. The coefficient of variation measures the extent of random variation around an expected value.

The higher the value is, the more random the process, and less precise one single reading is.  This is a basic concept that we teach in Core stats and Core Operations classes. At Theranos, to measure the CV of test results from Edison machines, they [Theranos] dropped the extreme values and took only the median values into consideration [Pg. 192]. Of course, this approach will result in a lower CV, which Theranos subsequently reported as a measure of the precision of their tests. This action is just wrong in every dimension that I can think of.

It is astounding how this “untested” device showed up in the Walgreens’ retail supply chains without being tested by a third party.

We definitely dodged a bullet as a society through an act of careful journalism. It is a reminder that journalism is a public good worth cherishing.

Notes:

1. Sunny Balwani. Who’s this person? As a desi male, I have come across my share of characters of that exact caricature — party shirts with folded sleeves, buttons open, chain dangling on a hairy chest, heavy cologne, and smooth talk.  I am dumbfounded that none of the directors registered a flag. One day, we will know more about what motivated Sunny Balwani and Elizabeth Holmes.

2. George Polk Award.  I thought the editors at WSJ stood steadfast without buckling under what I saw as immense pressure. WSJ organization also preserved the necessary wall between the news division and opinion section, because their differences were astoundingly dissonant.

If you like what you read, you can subscribe to the OWL Blog for occasional newsletters. No spam. I respect your privacy.

Subscribe to My Newsletter

(Roughly) Weekly Emails. I respect your privacy.

Published in Books Life