One of the common complaints I hear about mental illness diagnoses is that they are “unscientific,” based upon a bunch of subjective symptoms that are arbitrary. People who dismiss mental illness as not being “real” say that unlike medicine, the mental health profession doesn’t have laboratory tests, biopsies or meaningful imaging tests.
I would suggest, however, that the mental illness diagnostic reference manual, the DSM-5, is actually a good compromise based upon our current but limited knowledge of mental illness and its underlying causes. Moreover, most people’s understanding of medical diagnosis is often unrealistic and doesn’t take into account the messy reality.
How do mental illness diagnoses compare to more traditional medical diagnoses?
Most people who seem to come at this question also seem to misunderstand how traditional medical diagnoses are made in the hospital or other healthcare settings. I often run across the belief that most medical diagnoses are made by a laboratory or blood test, or a biopsy, and (in conjunction with certain other physical signs) the medical tests are often conclusive and tells doctors exactly what is wrong with the patient.
But if you talk to practicing physicians, you find that the reality is not nearly so neat and clean. Medical diagnoses can be just as messy and as complex as human beings themselves. (In fact, an entire popular medical television series, House, was based on this very premise.)
Some medical diagnoses are indeed neat and clean — and fairly easy to make. If you’ve broken your arm, an x-ray will help the doctor determine exactly what kind of break it is, where it occurs, and through such data, determine how best to set your arm to ensure it heals properly.
But then some diagnoses we take for granted — such as the common cold — don’t actually have any medical or laboratory test to confirm their existence. Doctors can order a set of tests to look for signs that your body is battling something, but those tests often can’t illuminate what exactly that something is. Only other subjective symptoms that the patient describes can help to do that.
Even then, doctors can still scratch their heads and only be able to narrow the possibilities — not always settling on a clear, single diagnosis.
Of course, you don’t have to take my word for it. The medical literature is strewn with thousands of research studies examining the reliability of thousands of different medical condition diagnoses. My take from reading a random selection of these is that interrater reliability for many medical diagnoses is only fair, but specialists have better reliability numbers than others do (as long as the condition is something covered by their specialty).1
But it’s not just that diagnoses themselves are hard to agree upon. Most times, doctors don’t have enough data in order to even make an accurate diagnosis. Fink et al. (2009) summarized the problem as such:
Only 10% of the results of consultations in primary care can be assigned to a confirmed diagnosis, while 50% remain “symptoms” and 40% are classified as “named syndromes” (“picture of a disease”).
Moreover, less than 20% of the most frequent diagnoses account for more than 80% of the results of consultations. This finding, confirmed empirically during the last fifty years, suggests a power law distribution, with critical consequences for diagnosis and decision making in primary care.
Those are some eye-opening statistics. And that’s just for medicine.
Furthermore, as the medical diagnostic coding system — the ICD 10 — has gotten larger and more complex, the ability to accurately code diagnoses has declined (see Stausberg et al., 2008 for example). It is simply incorrect to assume that most medical diagnosis is easy and reached with a blood or laboratory test. In the real world, medical diagnosis is just as complex, subjective, and messy as mental illness diagnosis.
Are Mental Disorder Diagnoses Any Better?
In a word, no. And some might rightfully argue that mental health professionals’ interrater reliability levels are even lower for mental disorders. That would be a fair criticism, especially since so many professionals of varying experience levels can actually make a mental disorder diagnosis (from a clinical social worker, psychiatrist, or pediatrician, to a family doctor, nurse practitioner, or general physician, among many, many more).
But having acknowledged as much doesn’t mean such labels or symptom constellations (if you prefer not to refer to them as diagnoses) are without purpose or value. Just as medical diagnoses help inform a physician’s treatment options, so too do mental disorder diagnoses.
For instance, it can be dangerous to prescribe an antidepressant to someone with bipolar disorder, as it could help bring about a manic or hypomanic state. That is valuable information to have if you’re the prescribing doctor.
I empathize with the fact that mental disorder diagnoses are more sociological and psychological constructs than most medical diagnoses. But to devalue mental illness diagnoses based upon the mistaken belief that medical diagnoses are so much easier is to minimize how complex and hard medical diagnoses are in the real world. And to devalue mental diagnoses in the first place seems to miss the ultimate purpose of labeling these things — to help people who are suffering and in pain.
I don’t think you always need a label to do that well, but I also don’t think such labels harm people as much as some critics suggest. They may be no more accurate than medical diagnoses, but they are only there to help inform treatment and research (and receive reimbursement from insurance companies). Nobody should stop and think that mental illness diagnoses define a person any more than any other single characteristic of that person would.
Fink, W., Lipatov, V. & Konitzer, M. (2009). Diagnoses by general practitioners: Accuracy and reliability. International Journal of Forecasting, 25, 784-793
Stausberg, J., Lehmann, N., Kaczmarek, D., & Stein, M. (2008). Reliability of diagnoses coding with ICD-10. International Journal of Medical Informatics, 77, 50-57
- “Interrater reliability” is the correlational measure of how much two different people would agree on a diagnosis, given the same symptom descriptions. The higher the reliability, the more likely one can infer that the diagnosis is fairly well-understood, described and recognized. [↩]