Current psychiatric diagnostic practice is based around groups of symptoms reported by patients. This type of ‘diagnostics’ started in the 1980s in the United States, where the Diagnostic and Statistical Manual of Mental Disorders III (DSM-III) was developed under the leadership of the late Robert Spitzer.

The goal of the symptom-based diagnostic method was to improve reliability, as the pre-DSM-III diagnostics were descriptive and varied substantially across countries. An example of this diagnostic variability was the fact that the concept of schizophrenia was at the time much broader in the United States than in Europe (Kendell, Reference Kendell1975). The general idea of the DSM-III was to standardize diagnoses and thus improve the investigation of the etiology and treatment of psychiatric disorders.

One of the (unintended) consequences of the development of DSM-based diagnoses has been the fact that the number of psychiatric diagnoses has increased from dozens at the end of the 1970s up to currently almost 300 diagnoses in DSM-5. This drastic increase in the number psychiatric diagnoses has been strongly criticized by Allen Frances (Reference Frances2014), the central figure in the development of DSM-4. In his opinion, the ever-increasing number of psychiatric diagnoses ‘medicalizes’ many normal life problems and forms of behavior, which often serves the interests of pharmaceutical companies more than patients. Furthermore, TV advertising of psychotropic drugs in certain countries, can serve to influence both patients and prescribers. With all this in mind Kendler and Parnas (Reference Kendler and Parnas2014) hold the view that the medical model of psychiatric disorders is only one way to understand the core of the psychiatric disorders.

Dr Nassar Ghaemi (Reference Ghaemi2013) of Harvard University has claimed that up to nine out of ten of DSM-5 psychiatric diagnoses are ‘scientifically unproven’. This kind of critique has naturally raised strong opposition among the developers of DSM-5 diagnostics. On the one hand, some psychiatrists, following the principle of Occam’s razor, suggest that clinical practice is better served by fewer number of diagnostic concepts. On the other hand, some psychiatrists hold the view that increasing the number of psychiatric diagnoses leads to better clinical practice and psychiatric research.

It is quite likely that, in this dispute, Robert Spitzer would have voted for the principle of Occam’s razor. This supposition of mine is based on an informal discussion that I shared with him many years ago in which he criticized the need of clinical psychiatrists to reach quickly a ‘specific’ diagnosis. According to Spitzer, the most usual diagnosis in the initial phase of the psychiatric should be an unspecific or ‘not otherwise specified’ (NOS) diagnosis, that is, psychosis NOS or depression NOS. He argued that in most cases the less specific diagnoses are good enough for the initial clinical assessment and the effort to reach a ‘too specific’ or ‘exact’ diagnosis decreases the diagnostic reliability. In the same way, David Brendel (Reference Brendel2006) has pointed out that the diagnosis in the intial phase is always a hypothesis that should be tested and corrected during the subsequent phases of treatment. A ‘quasispecific’ diagnosis made too quickly can easily become a dogmatic prison. It is also often forgotten that the circumstance of normal clinical practice usually differs substantially from randomized treatment studies.

In the worst case scenario, the effort to reach a ‘precise’ symptom-based diagnosis prevents the psychiatrist viewing his or her patient as a ‘whole’ person with a unique life experience and can easily lead to unnecessary medicalization. This danger may be further heightened, if (as is the case too frequently in modern medicine) the patient’s doctor changes with high frequency.