Here's an example of how challenging it can be to pursue accuracy as it pertains to mental health.  Before getting into the example, a quick comment about the word "accuracy" itself:

I generally try to avoid the terms "truth" or "fact."  While I believe that it is possible to define "facts" and identify things that reasonably meet the criteria as fact, it still seems less helpful than talking about the pursuit of accuracy.  Talking about "truth" in my opinion, should be left entirely to the philosophers.  Accuracy on the other hand, is a term I like very much.  How do I define it?  I define accuracy as "an intentional effort to fully understand the scope and limits of what we can reasonably claim to know, and working only within those limits."

Today's example of the challenges inherent in pursuing accuracy comes through the subject the "Schizophrenia" diagnostic label and the search for a genetic origin.   Two pieces, written on the same subject of research with conclusions that could not be more opposite.  The first piece is written by psychiatrist Emily Means, and appears in her blog Evolutionary Psychiatry as well as on Psychology Today.  I am posting several excerpts from the article in an attempt to offer a fair summary:

Last month, a veil was lifted from the mysterious and devastating disease of schizophrenia in this article in the green journal (online advance release).
Because schizophrenia is so devastating and highly inherited, researchers have been searching for the schizophrenia “gene” for a long time. Finding the gene could mean more targeted treatments and aggressive early intervention in those who have the highest risk, so that perhaps the disease wouldn’t progress. What they’ve found so far in very large genome-wide association studies is not a single gene, but sets of different ones that together confer greatly increased risk of developing the disorder. It’s difficult to find these gene networks, because the ways genes interact is extremely complex, changes during development, and in many cases is poorly understood. For the first time, researchers were able to combine three large sets of data of folks from varying populations with and without schizophrenia that had not only information about the genes, but more specific information about patients’ symptoms.
Using an immense amount of data crunching, researchers found what small genetic differences were in common between different folks with symptoms of schizophrenia, then more precisely define which of these hundreds of risk genes were more or less risky. They found one set in 9 different people where every single person had schizophrenia (100% risk), and others that conferred far less risk…but there were 42 different sets of genes that had risk for schizophrenia of >70%. By crunching even more numbers for these high risk sets of genes, they found that some of these sets “grouped” together in what is called a genotypic network. Then they overlapped the data from the risky genetic networks with the symptoms of the sick individuals to come out with eight different distinct diseases, now all known as “schizophrenia.”
If this research holds true on reexamination, we might be taking the first steps toward isolating the pathology of a major psychiatric disorder, as infectious disease doctors did a hundred and fifty years ago with the development of germ theory. 
Wow.  So this momentous study (hyperlinked in the excerpt) has "lifted the veil" from what we label schizophrenia?  That would be nothing short of earth shattering, something that over a half a century of gene-finding expeditions has utterly failed to do.  Means goes on to double down on her assertion, by stating that the study's findings identified sets of genes that "greatly increased the risk of developing the disorder."  Greatly.  Means goes on to state that the study found one set in 9 (whole!) people where every single person had schizophrenia.  However, there seem to be a few critically important omissions in her article.

So now let's contrasts Means article with that of Joanna Moncreif, who writes about the exact same study on her blog:


The recent headlines were generated by a paper which reports a Genome Wide Association study in a combined sample consisting of tens of thousands of people diagnosed with schizophrenia and controls (4). Nine and a half million SNPs were examined. Out of these millions of SNPs, 128 were statistically significantly more common in people diagnosed with schizophrenia, using a ‘p’ value of 0.00001 to correct for the large number of tests conducted. These 128 SNPs were then mapped to 108 chromosomal regions or ‘loci’, most of which had never previously been thought to have anything to do with schizophrenia. The paper provides no odds ratios or other measure of the strength of the associations between the SNPs and having a schizophrenia diagnosis, but the lead researcher confirmed that each genetic loci was associated with an increase in the risk of schizophrenia of approximately 0.1% (5). Using a method dependent on multiple assumptions (Risk Profile Scores), the maximum amount of variation explained by a combination of the genes identified was 3.4% (4).
 The genetics of schizophrenia parallels findings in most common physical diseases. A small number of genes with substantial effects have been identified for a small number of conditions (breast cancer, Alzheimers and AMD). Other than this, the hugely expensive human genome project and all the research that has followed from it has failed to reveal that particular genes increase the risk of developing common and important diseases to any relevant degree (6).
 As in the latest schizophrenia studies, some genes appear to be slightly more common in people with conditions like Type 2 diabetes compared to those without, but having one of these genes will increase the risk of developing the disease only marginally. But the fact that people’s genetic risk varies around the average is simply tantamount to saying that everyone is different (6). We knew that anyway.
These results have lead some commentators to conclude that ‘genetic predispositions as significant factors in the prevalence of common diseases are refuted’ (6). Yet researchers and the media continue to report genetic studies as if they represent evidence that bodily diseases and behavioural or mental disorders are genetic conditions- that is conditions that are determined in an important way by detectable genetic variation. The recent schizophrenia study was reported in just such a way. ‘They’ve really managed to show that genetics plays a role’ said one schizophrenia researcher (7). The study authors claim their findings will help unravel the aetiology of the condition and lead to the development of effective and targeted drug treatments (1).
It is difficult to see that these findings will have any clinical implications for our understanding of schizophrenia or its treatment, however. Leaving aside the difficulty of defining mental abnormality or ‘schizophrenia,’ if the genes identified predict only about 0.1% of the risk, then whether you have one or two, a few or even a lot of the genes concerned tells you very little about your particular vulnerability for developing the condition.
Moncrief's article chooses to detail the specific methodology and the statistical significance of each finding, rather than gloss that over in order to arrive at a more hyperbolic conclusion.  Not only does Moncrief provide reference citation for claims that go beyond the scope of this single study, each of her statements about the study are a simple recitation of the findings themselves.  We can check to see whether or not Moncrief is being accurate by simply looking at the study.  And the most accurate thing to say is that, identifying a genetic predisposition one tenth of one percent is not only no lifting of any "veil," it is also of little significance.  If anything, the quest for genetic root causes for all sorts of diseases has been very discouraging in recent years, and its time that we start facing up to the reality that there may not be "silver bullet" genes (or gene sets) that are tied to diabetes, Alzheimer disease, or so-called illnesses of the mind.

Let me be clear, I would be absolutely thrilled if we were able to identify a gene or set of genes that was proven to be highly linked to a schizophrenia diagnosis.  Just like I would be thrilled if we could identify any sort of clear biomarkers or medical tests.  If we could demonstrate illness of the mind to be literal biological diseases "just like diabetes or cancer" then we would be able to utterly revolutionize treatment for the better.  We might even be able to "cure" mental illnesses.  If the day comes when we genuinely have a clear and coherent body of evidence that proves this, I will cheer.

But as of this moment in time, we don't have that.  And each year it looks less likely that we will ever find such a reductionist answer, because it looks increasingly likely that so called "mental illness" simply do not neatly fit a biomedical paradigm.  I repeat, if we have a breakthrough and the evidence proves a biomedical root for mental illness, I will be the first to celebrate.  But I believe in accuracy.  Accuracy means not acting like you know more then you know.  It means not drawing conclusions or making decisions based on what you want to be true or hope might one day be proven true.  It means making decisions based on the limits of what you know today.  That has been the ongoing failure in psychiatry and mental health - treatments are selected based on our beliefs that "one day we will be proven right" even though for the last half a century we've failed in that endeavor.  Now we must be held accountable for the damage our treatment paradigms have done to real human lives.