Unintended Consequences

We all know the story of the Titanic, at least everyone who has seen the movie – which I think is everyone. The boat carried 2,224 passengers and crew but only had lifeboats for 1,178 people. Of course, the ship sank after hitting an iceberg and as you might predict from this mismatch, 1,514 people toied. Well, at least died that day. The sinking of the Titanic killed another 844 people just over three years later. How, you ask?

Fewer people know about the SS Eastland. After the sinking of the Titanic, there was an unmeasured and knee-jerk response to it around the world.  Such unbalanced responses usually follow tragedies, especially if the tragedies appear preventable. Often, good and needed reform comes from these responses – but not always and usually not only good and needed reform. In the United States, in response to the sinking of the Titanic, President Woodrow Wilson signed into law the Seamen’s Act in 1915 which, among other things, required that all ships greater than 100 tons be outfitted with enough lifeboats to carry every passenger.

Thus, the SS Eastland was retrofitted and several large lifeboats were added to the ship, making it increasingly top-heavy. Ship owners and others had opposed many parts of the law and even cautioned about the problems of making ships too heavy, but just two years after the sinking of the Titanic, famous politicians like William Jennings Bryan and Woodrow Wilson rode the tragedy to great political capital, satiating a public who cried out for safety. On July 24, 1915, the newly retrofitted vessel set out from the Chicago River with 2,572 passengers on board. Within a few minutes of leaving the dock, the top-heavy vessel rolled over and sank to the river floor, killing 844 people.

This was an unintended consequence: a measure taken to save lives ended up killing people. We start out with good intentions but don’t foresee all of the potential impacts of our actions. Sometimes we make things better, and sometimes we make things worse. In 2008, Airbus redesigned the cabin of the A380 airplane to reduce external noises in order to improve the passenger experiences. This was a nice idea. They sound-proofed the cabin to limit engine and wind noises. But the now quieter cabin accentuated other internal noises, like flushing toilets and people snoring, and actually resulted in more unhappy passengers.

I’m sure you can think of all sorts of examples of unintended consequences, but let me give a couple more.

In 1958, Chairman Mao Zedong of Communist China, initiated the Four Pests Campaign as part of the Great Leap Forward program. The Great Leap Forward initiative was a centrally-planned program designed by the Chinese government with the intention of improving the lives of the Chinese people through a program of industrialization, agricultural collectivization, and other improvements. The Four Pests Campaign itself was designed to eliminate rats, flies, mosquitoes, and sparrows.

Sparrows had been identified because they notoriously ate grain seeds, destroying statistically significant portions of the yearly grain production. This decreased yield threatened the Chinese people with malnourishment due to the decreased agricultural production. The central planners determined that eliminating the pests, particularly the sparrow, would have a positive benefit for production and food supply, and, therefore, the Chinese people.

Chinese farmers and others throughout the nation were enlisted to destroy sparrow nests, breaks eggs, and kill the birds in any way possible, including denying them rest while flying by scaring them, literally causing them to fly to death (due to exhaustion). The strategy worked and the sparrow was nearly made extinct in China.

By 1960, though, scientists realized that rice production, rather than increasing, had actually decreased. It turns out the the sparrow served an important and unrecognized role in eating insects which fed off of the grains. With the sparrows gone, the insects thrived and ate massive amounts of grains. The result of this ecological imbalance was the Great Chinese Famine, which led to the death of 20-45 million due to starvation.

Surely, we in the US are immune to such inanity.

In the 1940s and 1950s, scientists and environmentalists in the United States raised awareness regarding DDT, a then common pesticide, and its potential harms to animal and human health. In response to the popular book Silent Spring, published in 1962, President John F. Kennedy commissioned an investigation into the dangers of DDT, and this investigation concluded that DDT usage should end. By 1971, the EPA held hearings about the safety of the pesticide. Internal EPA data had indicated that DDT was not a threat, but due to public outcry stirred on by Silent Spring, six months of public hearings were held. At the end of these hearings, the EPA banned the use of DDT. Most of the world, including the World Health Organization, followed suit in the years to come.

The WHO, prior to banning DDT, had previously used the pesticide to fight malaria throughout the world. After its invention in 1939, DDT was used to eradicate malaria as a public health problem in many nations, including the US. Malaria had been a significant problem in the southern United States, with about 15,000 cases annually. The CDC was founded in 1946 specifically to combat malaria, and was stationed in the South, in Atlanta, since malaria was then confined to the thirteen southern states. Starting in 1947, the CDC sprayed 6.5 million homes – inside and out – with DDT to kill mosquitoes that served as the main vector. They used airplanes to spray massive amounts of land and swamp areas as well. This program essentially eliminated malaria in just 5 years. In the US and every other country where DDT was used, death from malaria had been virtually eliminated.

Of course, 25 years after malaria was no longer a public health problem, the US led the way to eliminate DDT, with a public no longer mindful of how bad malaria had once been. Rachel Carson, the author of Silent Spring, had blamed most cancers on the pesticide and predicted that within a short period of time, nearly 100% of people would be affected by cancers due to DDT. The idea was appealing to Carson, who at the time was being treated for metastatic breast cancer. She used the kind of pseudo-science that is still used every day today to make such ridiculous claims, convincing readers that causation equals correlation. She wrote in Silent Spring: 

Trout in both eastern and western parts of the United States were affected; in some areas practically 100 per cent of the trout over three years of age developed cancer. […] The story of the trout is important for many reasons, but chiefly as an example of what can happen when a potent carcinogen is introduced into the environment of any species. Dr. Hueper has described this epidemic as a serious warning that greatly increased attention must be given to controlling the number and variety of environmental carcinogens. ‘If such preventive measures are not taken,’ says Dr. Hueper, ‘the stage will be set at a progressive rate for the future occurrence of a similar disaster to the human population.

Her bold claim, made at the same time that prominent scientists were claiming that cigarette smoking was not carcinogenic, of course proved to be untrue. In fact, the Dr. Wilhelm Hueper she mentioned, her mentor, claimed that evidence linking cigarettes to cancer was created by the pesticide industry as part of a cover-up. Eventually these false claims and activity led to worldwide bans or voluntary non-use of DDT.

You might note from these graphs that cigarette smoking and longer lifespans are the largest contributors to cancer rates. The rise in cancer that she observed had been due to cigarette smoking. The main premise of the book was false, as had been the claim of decreasing bird populations.

With malaria free to travel due to the DDT bans, it is estimated than more than 50 million people died of malaria who might otherwise have lived, making Carson partly responsible for more deaths than almost any other single person in the 20th century. For her work, she received the Presidential Medal of Freedom. Despite the bad science behind the book, which has become more obvious over the decades since its publication, the book is still revered by environmental groups. It was named one of the top 25 science books of all time by Discover magazine. Funny how dogma works. (If you are mad at me right now, read this or this or this).

These bad outcomes are all, of course, unintended consequences. Saving grains from pests, making airplane rides more pleasant, making sure people don’t drown for want of a lifeboat, and preventing the cancer epidemic, are all noble causes. But the pursuit of such good made things worse in each case.

There are three sorts of unintended consequences. Sometimes there is an unexpected benefit; sometimes there is an unexpected drawback; and sometimes, as in these above examples, there is a perverse result. Virtually every intervention that we use medicine has all three of these associated with it.

Intuition (and good intention) often fail due to unintended consequences. One of the main problems with anecdotal medicine is that it is short-sighted in that the focus on improving one thing comes at the neglect of understanding what else is potentially affected. For example, it might be true that lowering blood sugars is associated with improved microvascular outcomes in diabetics. It makes sense that getting a diabetic’s blood sugars as close to normal as possible is a good thing. It’s intuitive. A study of a drug might show that it works well at producing good glycemic control; it might even show that it reduces the risk of kidney disease or eye disease. Those are good intentions. If one focused solely on these outcomes, then the drug might be declared a success and hailed as the next new miracle advancement.

However, if these lower blood sugars come at the cost of higher all-cause mortality due to increased cardiovascular disease, then this unintended consequence would more than negate any positives. It is not enough that something makes sense, nor is it enough that it has some good benefits; we have to consider the net impact of any intervention.

Use of surrogate markers magnifies the effect. Surrogate markers cut the observational period of a study short. If you take a cholesterol drug, what you would hope for is that it reduces your chance of death, improves your quality of life, or both. It takes a long time to study this, so commonly a drug will look at a surrogate maker or surrogate endpoint – in this case, it might be cholesterol levels. In a relatively short time, a study can show that lipids are lowered if an investigational drug is taken. We are then left to assume that this will result in lower mortality later. But it may not; it may actually increase mortality (as in the case of too-tight blood sugar control in diabetes). The lack of direct confirmation of the real outcome is the main problem with surrogate markers; but a secondary problem is that since the period of observation is cut short then unintended outcomes aren’t necessarily observed and therefore aren’t taken into account. Studies that showed increased cardiac mortality with intense blood sugar control wouldn’t have shown that if the study only lasted 6 months and was ended at some surrogate endpoint, like improved HgA1c levels.

Visceral over-reactions often create disproportionate responses that increase the risk of an unintended consequence. There was likely a good reason why the SS Eastland didn’t carry a bunch of lifeboats: it wasn’t designed to do so. A long history of ship building and nautical experiences informed the practice of having fewer lifeboats on ships like the Eastland, but all that information meant nothing when a handful of non-sea-faring politicians decided to do what “just made sense” rapidly. Often, these type of knee-jerk reactions happen when something emotionally and viscerally negative happens – even when the proposed solution might not have prevented the outcome. An obstetrical unit which suffers a spontaneous uterine rupture of a woman with an unscarred uterus in labor might have such a reaction; this exceedingly rare event is likely unpreventable, but someone who doesn’t understand the science might blame oxytocin. If the patient had been on 20 mu/min of oxytocin, then the leaders of the unit might suggest that patients not be allowed to have more than, say, 12 mu/min, even though there is no scientific evidence that this matters. The unintended consequence might be a dramatically higher rate of cesarean delivery on many thousands of other patients over several years and the eventual death of a woman from a pulmonary embolism associated with cesarean delivery. In other words, some times the best thing to do when something bad happens is nothing – unless the evidence is clear that a change in practice is indicated.

Medical examples of unintended consequences could fill several volumes. The unscientific fat-phobia fad started in the 1970s led to the current obesity epidemic (people ate less fat because people told them that fat would make them fat and so they ate more simple carbohydrates and became fatter). The introduction of pain scores into clinical medicine and patient satisfaction scores contributed to the opioid epidemic (the patient’s subjective idea of how much he hurts was treated like an objective vital sign that then demanded pain medication for fear, on the part of a physician, of a low patient satisfaction score).

A recent example is the fallout from the Women’s Health Initiative Study (WHI). The media, led by Oprah Winfrey, created a frenzy about the safety of hormone replacement therapy after the initial results of WHI were released. Women stopped their hormones in droves, worried that they would die of cancer or heart attacks or strokes if they dared continue. One paper reported that as many as 91,610 women died premature deaths due to unnecessarily stopping their estrogen therapy in the ten years that followed the media scare stirred up by the WHI. I don’t believe for one second that Oprah was trying to do anything but help women when she scared them into stopping their hormones, but the tens of thousands of deaths that followed were an unintended consequence.

So could these stories have been prevented? Are unintended consequences just a natural part of progress? Yes and no.

The Wilson administration was warned that adding extra lifeboats would increase the risk of boat-toppling, but these warnings were unheeded because it was felt to be politically expedient to jam-through the new legislation.

Chairman Mao’s plan of killing sparrows was obviously wrong to many then living scientists, but the culture of his government forbade dissent and contrarian ideas.

Rachel Carson, if anything, was used by the tobacco industry to serve as a red herring. There was strong scientific evidence that cigarette smoking was leading to the increase in cancers since the 1950s; Carson and her mentor ignored this evidence because they practiced agenda-driven science and repeatedly ignored any evidence that didn’t support their flawed world view. There wasn’t then and has never been scientific evidence that DDT was carcinogenic to humans. Even the EPA internal data said that it was not harmful. But Presidents Kennedy and Nixon courted the votes of the anti-DDT, environmentalist groups and pushed through fear-mongering, biased panels that dealt with theater and vote-pandering, not science. By the way, DDT is now finally being used again to fight malaria after decades of pro-DDT activism.

When the WHI released their interim findings in 2002, there were no statistically significant data that indicated that estrogen increased the risk of breast cancer, heart disease, stroke, or death. But the authors of the paper didn’t do enough to clarify this fact, and reporters, who love sensationalism and scary stories (like Silent Spring) ran with it and, in the process, convinced millions of women to stop their estrogen for no scientific reason.

These are not cases of ‘We once thought this and later learned we were wrong.’ Most unintended consequences are not of this sort. Most are just the result of bad analysis and bad science. A new paper this week, already reported on by 52 news outlets, claims that women who work night shift or do heavy lifting are more likely to be infertile; at least that’s how the data is being presented. It doesn’t actually say that all, but instead uses surrogate markers and non-controlled data to highlight some potential trends. But the paper is already being used to promote the idea that less activity will improve fertility. When a woman dies of a pulmonary embolism in her attempt to get pregnant due to unwarranted inactivity, that unintended consequence could have been prevented.

Good science could have prevented all of these unintended consequences. Yet, good science can’t prevent them all. I don’t know what the Airbus engineers could have done differently.