Category Archives: Research Methods

What we almost know

Posted by on September 25, 2013 Comments Off on What we almost know

The impact of intestinal flora on health conditions is known, thus fecal transplants for numerous conditions. Last week I wrote about an experiment with mice verifying the impact on obesity.

As a consequence, there are numerous products being marketed as “probiotic.” We have no idea whether those products are helpful, harmful, or simply benign.Bacterial flora are an instance where we know some bacteria are essential, and we know some can be added beneficially, but we do not know which ones to add. There are promising studies with lactobacillus acidophilus shown: lactobacillus We almost know about GI flora, but not quite enough yet.

There are other topics where we are at a similar place. We know a lot about what mental illness is and isn’t. We know that increasing serotonin in the interstitial spaces of the brain helps with depression and that too much serotonin is associated with schizophrenia. As of this point, however, we don’t have imaging or blood tests or biopsies that will tell us who is mentally ill–we use written testing and observation not laboratory tests to diagnosis it. When physicians attempt to treat it, it is largely a matter of trial and error. We do not know beforehand whether a particular selective serotonin re-uptake inhibitor (SSRI), such as fluoxetine or venlafaxine will work with an individual, or if any SSRI will work. Some people will do better with buproprion, which uses a different mechanism. Others will do better with a serotonin–norepinephrine reuptake inhibitor (SNRI) such as duloxetine. Others will experience no clinical effect at all. We almost know about mental illness, but not quite enough yet.

These examples lead to a more general question about what we know and do not know. It is usually phrased as “nature vs. nurture,” but it is really genetics vs. environment. I am not sure if the the “versus” between them is appropriate–something else I do not know–as it is the relative interaction of the two rather than a false choice between them that is a more likely source of the truth about who we really are.

Researchers often conduct identical twin studies, controlling for genetic variation by comparing the environmental impact of twins with nearly identical DNA. Molecular computer graphic of DNA double helix However, the studies are limited: it is intriguing if they both smoke the same brand of cigarettes or like the same foods despite very different upbringings, but it does not neatly tell us which behaviors are genetic and which are environmental.

Similarly, we know that 3/4 of children of two bi-polar parents are likely to have bi-polar disorder, which seems to indicate a Mendelian genetic inheritance, but we only almost know about the inheritability of mental illness, not quite enough.

Even where we know that a trait is inherited, we often do not know what genes or constellation of genes are associated with a given, visible trait. We almost know about the human genome, but not quite enough yet.

As scientists or those of a scientific bent, we are obligated to say what we know and what we don’t know, being able to distinguish the difference. It is not always an easy distinction to make, but is central to our effort to know more, and eventually know enough.

How researchers confuse the public

Posted by on May 20, 2013 Comments Off on How researchers confuse the public

A nursing professor once told me how a graduate student came to her all excited. There was a population cohort dying from an epidemic of cardiac disease that had been ignored in the literature: women above age 85.

Now, while we can chuckle together about the naivete of the observation, consider it for a moment from the graduate student’s perspective:

1. Clearly the phenomenon was real as a high percentage of women over age 85 may well be dying of cardiac disease.
2. The red flag of gender bias stood out–it was women who were being victimized by this scourge.
3. A literature search turned up no one acknowledging the problem.

What was missing was the larger context: the twin facts that no one lives forever, and everyone eventually dies of something.

Sometimes more experienced researchers fail to provide the needed context. The following study came to my attention this week. Truth be known, I have not read the entire study, just the abstract and the accounts of those who have, so I will admit up front that this discussion may be unfair to the researchers.

The Canadian study is The long arm of parental addictions: The association with adult children’s depression in a population-based study.

It came to my attention in Trouble Coping with Parental Addiction

I am going to quote the abstract in full:

Parental addictions have been associated with adult children’s depression in several clinical and population-based studies. However, these studies have not examined if gender differences exist nor have they controlled for a range of potential explanatory factors. Using a regionally representative sample of 6268 adults from the 2005 Canadian Community Health Survey (response rate=83%), we investigated the association between parental addictions and adulthood depression controlling for four clusters of variables: adverse childhood experiences, adult health behaviors, adult socioeconomic status and other stressors. After controlling for all factors, adults exposed to parental addiction had 69% higher odds of depression compared to their peers with non-addicted parents (OR=1.69; 95% CI, 1.25–2.28). The relationship between parental addictions and depression did not vary by gender. These findings underscore the intergenerational consequences of drug and alcohol addiction and reinforce the need to develop interventions that support healthy childhood development.

The authors suggest that previous studies have not directly looked at gender differences of children of addicted parents. If so, then that is a clear contribution to the literature. But, the abstract and the descriptive article that I cited above go further: there is the clear implication that beyond a correlation or “relationship,” depression is the consequence of parental addiction.

Let’s consider two scenarios:

First, parents engage in substance abuse. The substance abuse causes them to act out, to neglect their children’s physical and emotional needs. The children grow up insecure and prone to depression. That is the clear implication of the abstract.

Second, a small but measurable percentage of the population inherit a tendency to mental illnesses (bi-polar, uni-polar depression, anxiety disorder, etc.) They self-medicate with both legal and illegal substances. They have children, whom they raise while self-medicating. A high percentage of their children inherit the tendency to mental illness, including uni-polar depression, and depressive cycles of bi-polar, showing a higher tendency toward those illnesses in the general population.

I find the second scenario more compelling. Indeed, attributing the problems of children to their parents because children follow their parents temporally is to my mind a classic case of the post hoc, ergo propter hoc logical fallacy in which one concludes that events following another event were caused by the previous event due to their proximity.

Runny noses do not cause colds, nor does “catching a chill,” whatever that means. Scientific studies are confusing enough to the general public, particularly through the filter of news media that do not look past the headline. We need researchers to be more cautious and guarded in their conclusions.

When we read that President Harry Truman wanted some “one-handed economists,” we can be sure that the economists were doing their job in informing him. We need the same of healthcare researchers.