hey there,
since im a new assistant in psychiatry, ive been struggling when it came to guidelines .. actually .. ok, this is going to sound so ignorant coming out of a newbie in the field but seriously NONE OF THEM sounds right !! i mean, the ICD and DSM diagnostic criteria seems so broad to me and my mind cant seem to accept them lol .. ahhhh i dont think ive explained myself well enough and im having hard time doing that,
my point is, i feel like we need to have strong neuroscience basics in which we should build the diagnosis on ..
ive been going through kaplan book of psychiatry and a friend of mine suggested i should read fish psychopathology but none of these books could satisfy my thirst for knowledge ..
where can i get scientific FACTS in which i could understand mental illnesses better and thus be able to diagnose !!
and to be honest, DSM seems unrealiable .. i think the ICD sounds more clinical oriented but again, it doesnt draw a clear cut line for diseases ..
ahhhh im so confused
😛 

you are conflating different terms here: issues or reliability, validity and utility of diagnoses. As a general rule DSM-IV diagnoses are very reliable. In the 1970s there was an important study called the UK/US diagnostic study which found that psychiatrists in New York were diagnosing schizophrenia far more commonly than were psychiatrists in London. The reason being American psychodynamic psychiatry had a much more fluid concept of what was schizophrenia than did European kraepelinian psychiatry. (Also we had lithium in Europe and no one was using lithium in the US). As a result of this and other challenges to the reliability of psychiatric disorders e.g. the Rosenhan study, the DSM-III working group was set up with the aim of operationalising diagnostic criteria for mental disorders. As a result if you are trained in making DSM-IV diagnoses then you can reliably make most diagnoses. Reliability means that if a patient presents to 10 different doctors they will more or less make the same diagnosis. DSM diagnoses are slightly more reliable than ICD-10 as it is a bit more of a cook-book approach. The DCR-10 is a particular reliable guide to making ICD-10 diagnoses and is used in research.
The problem is most psychiatric diagnoses are not valid (exceptions are Alzheimer's dementia, Lewy Body dementia etc). There is no validity to diagnoses such as schizophrenia, bipolar disorder, major depressive disorder and so on. These diagnoses are NOT diseases (or if they are there is no evidence to support this). They currently have no external validity in reality, and certainly no biological validity. Most of the biomarkers from research are very weak, with low rates of replication, and in particular a lack of specificity. Many people working in psychiatric genetics have now admitted that the main problem with doing psychiatric research is it is based on the assumption that 'schizophrenia' and so on are real diseases, and universal concepts. However, many of the candidate genes for schizophrenia are also candidate genes for bipolar disorder or major depression or panic disorder... and about 1% of cases of schizophrenia have been attributed to variable large chromosomal microdeletions (including 22.q11). What we are probably looking at with schizophrenia, bipolar and so on are different diseases with a similar manifestations. That is these disorders are syndromes that we have out of necessity lumped into imperfect categories. This highlights the limitations of the categorical approach to diagnosis.
The final issue with diagnoses are how useful are they? As a clinician my primary interest is in how useful my diagnoses are to my patients and I think many diagnoses are useful, but I could probably do without most of the diagnoses in DSM-IV.
If we take the example of bipolar disorder it is useful to me in the following ways:
1. making the diagnosis guides me with treatment - so I would start with mood stabilisers and then therapy for bipolar, but would start with DBT or MBT for borderline PD and avoid medication altogether; similarly I may prescribe antidepressants for a patient with major depression, but I would be concerned about destabilising or worsening illness with antidepressant monotherapy in bipolar
2. although there can be harm from giving someone a 'label' many of my patients find it useful. it means they don't feel alone with their suffering, there are support groups, and it means they think i understand the problem and can help
3. diagnoses allow access to benefits/welfare and protection under disability discrimination
4. DSM-IV diagnoses, because they are reliable if not valid, give us a common language for us to communicate with other colleagues. But they are also important in the legal setting as well. as one psychiatrist commented upon discussing whether we should abandon the diagnosis of schizophrenia altogether: "what would expert witness say when giving evidence? that a patient has the illness formerly known as schizophrenia?" Also diagnoses are important in terms of billing and getting reimbursements from insurance companies. I don't agree with this, but that is how it currently is.
So to answer your question: are DSM-IV/ICD-10 diagnoses useful - many of them are but most of them are probably BS. are they reliable - yes when made by an experienced clinician and especially if using the SCID or SCAN (semistructured interviews). are they valid - NO.