The failure of decades of basic science research to reveal any specific biological or psychological marker that identifies a psychiatric diagnosis is well recognised. Unlike the rest of medicine, which has developed diagnostic systems that build on an aetiological and pathophysiological framework, psychiatric diagnostic manuals such as DSM 5 (American Psychiatric Association, APA, 2013) and ICD-10 (World Health Organization, WHO, 1994) have failed to connect diagnostic categories with aetiological processes. Thus, there are no physical tests referred to in either manual that can be used to help establish a diagnosis. This lack of scientific progress connected to diagnostic groupings is a problem for research from a variety of perspectives, including biological research, where leading research groups are abandoning the use of current diagnostic constructs (Marneros & Akiskal, 2007; Owen, O’Donovan, Thapar, & Craddock, 2011). Despite the belief that psychiatric disorders have a clear genetic loading, molecular genetic research is failing to uncover any specific genetic profile for any disorder. Possible genetic abnormalities appears to account for only a small percentage of causal factors, and whatever genetic contribution has been found crosses diagnostic categories rather than having a distinct profile for each diagnostic category (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013).

The one notable exception to the lack of aetiological organisation in diagnostic systems is ‘post traumatic stress disorder’ (PTSD), which implies that trauma leads to a particular and identifiable constellation of symptoms. However, there is a substantial body of evidence which finds that in the full spectrum of diagnoses in psychiatry, including psychosis, there is a greater likelihood of experiencing trauma and abuse (Bebbington et al., 2004; Escher, Romme, Buiks, Delespaul, & van Os, 2002; Goodman, Rosenberg, & Mueser, 1997; Greenfield, Strakowski, Tohen, Batson, & Kolbrener, 1994; Honig, Romme, Ensink, Pennings, & de Vries, 1998; Morrison, Frame, & Larkin, 2003; Mueser et al., 1998; Read, Agar, Argyle, & Aderhold, 2003; Varese et al., 2012).

If we were to apply the standards found in the rest of medicine, then the validity of a diagnostic construct depends on the extent to which it represents a naturally occurring category. If it can ‘carve nature at its joints’, then there should be some identifiable properties beyond symptoms or behaviours, in those who have the diagnosis that can distinguish them from those who don’t. The failure of basic science research to reveal any specific biological marker for psychiatric diagnoses means that current psychiatric diagnostic systems do not share the same scientific security, of belonging to a technological model developed by research grounded in the natural sciences, as the rest of medicine. The attempted solution of continuing to spend the bulk of mental health research time and effort trying to correct this deficit by relentlessly searching for evidence of biological correlates continues to deliver nothing scientifically or clinically useful. Our failure to find biological correlates should not necessarily be seen as weakness. Instead, we should view this failure as an opportunity to review the dominant paradigm in order to develop one that better fits the evidence.

The frequency with which patients are given more than one diagnosis raises a concern about the specificity of diagnostic categories. Widespread co-morbidity (making more than one diagnosis in order to encompass patients’ problems) indicates basic deficiencies in our understanding of the natural boundaries of even the most severe conditions we are diagnosing (Anckarsäter, 2010; Middleton, 2008; Van Os, 2003a,b). It is also common to find the ‘dominant’ diagnosis changing in any individual, almost exclusively on a subjective rather than empirical (such as physical test results) basis.

Unlike in the rest of medicine, where the reason for the patient's symptoms is clarified by a diagnosis, psychiatric diagnoses serve only as descriptors that do not have the power of explanation. Thus, when a clinician claims that a patient is ‘really’ depressed, or has ADHD, or has bipolar disorder, or whatever, not only are they trying to turn something based on subjective opinion into something that appears empirical, but they are engaging with the process of reification (that is, turning something subjective into something ‘concrete’). The problem with turning concepts into something that appears as if it exists as a fact in the natural world is that it can cause ‘tunnel vision’ for all concerned; a dominant story that limits alternative, more functional possibilities for any individual. Thus, if someone believes ADHD is a ‘real’ disorder that exists in their brains and is potentially lifelong, that person and those who know them may come to act according to this belief, thus helping to fulfil its prophecy.

There is also a poor correspondence between levels of impairment and having the required number of symptoms for many psychiatric diagnoses. Literature reviews and field trials to examine clinical significance criteria were not included in the preparation of DSM-IV. Thus, many below the threshold for a diagnosis have higher levels of impairment than those above, with many who reach the cut off for a diagnosis having relatively low levels of impairment (Angold, Costello, Farmer, Burns, & Erkanli, 1999; Kendler, 1999; Simonoff et al., 1997; Van Os, 2003a,b).

Reliability is the extent to which clinicians can agree on the same diagnoses when independently assessing a series of patients. Improving reliability of psychiatric diagnoses was hastened after critics pointed out that many of the common diagnoses were meaningless because of poor levels of agreement between psychiatrists about key symptoms. Rosenhan's (1973) study spurred on new attempts to ‘standardise’ diagnostic practice after he demonstrated that psychiatrists were often unable to discriminate between sane and psychotic people. Formal diagnostic systems like DSM and ICD attempted to address these problems by imposing diagnostic agreement on the profession through the use of standardised diagnostic criteria through symptom checklists.

However, analysis of the studies involved in developing the first diagnostic manual that took this approach of ‘operationalising’ diagnosis through the checklist of symptoms approach (DSM-III) found no diagnostic categories for which reliability was uniformly high. Ranges of reliability for major diagnostic categories were found to be broad, and in some cases ranged the entire spectrum from chance to perfect agreement, with the case summary studies (in which clinicians are given detailed written case histories and asked to make diagnoses–an approach that most closely approximates what happens in clinical practice) producing the lowest reliability levels (Kirk & Kutchins, 1994). No studies of the reliability of DSM as a whole when used in natural clinical settings have shown uniformly high reliability, with many finding reliability ratings that are not that different from those in the pre-operationalised criteria studies (Kirk & Kutchins, 1992, 1994; Williams et al., 1992). In DSM 5 field trials (Freedman et al., 2013), the kappa coefficients (a statistical measure of inter-rater agreement), were uniformly poor with some common diagnoses such as Major Depressive Disorder and Generalized Anxiety Disorder achieving having levels of inter-rater agreement little better than chance.

The technological paradigm that dominates is predicated on the assumption that the technical aspects of medical and psychological care are of primary importance, and that making diagnoses allows rational choice of the correct technical intervention.

However, the evidence finds that ‘common factors’, such as developing a strong therapeutic alliance, are much more important than therapeutic technique in determining outcomes (Duncan, Miller, Wampold, & Hubble, 2009; Wampold, 2001). For example, studies have shown that most of the specific features of Cognitive Behaviour Therapy (CBT) can be dispensed with, without adversely affecting outcomes (Jacobson et al., 1996; Longmore & Worrell, 2007).

Meta-analyses have concluded that the majority of variance in outcome as a result of therapy is accounted for by extra-therapeutic factors, whilst the quality of the therapeutic alliance accounts for most of the within-therapy variance in treatment outcome, and is up to seven times more influential in promoting change than treatment model used (Duncan, Miller, & Sparks, 2004; Wampold, 2001). This is also evident in ‘real life’ clinical encounters. For example, in a review of over 5,000 cases treated in a variety of National Health Service settings in the UK, only a very small proportion of the variance in outcome could be attributed to psychotherapeutic technique, as opposed to non-specific effects such as the therapeutic relationship (Stiles, Barkham, & Mellor-Car, 2008).

The importance of non-specific factors is also found when using psychoactive drug treatments. The evidence supports the view that pharmacological agents are best conceptualised as inducing particular psychological states (rather than correcting chemical imbalances) that is not specifically related to diagnosis, but is nonetheless the basis for their usefulness (Moncrieff, 2009). This reflects clinical practice where the few categories of psychoactive medications used in psychiatry are used in a non-diagnosis specific way. For example, Selective Serotonin Reuptake Inhibitors (SSRI) are claimed to be efficacious in conditions as disparate as borderline personality disorder, depression, generalised anxiety disorder, obsessive compulsive disorder, anorexia nervosa, bulimia, panic disorder, social phobias and so forth. As a psychoactive substance, SSRIs would appear to do ‘something’ to the mental state, but that something is not diagnosis specific. Like alcohol, which will produce inebriation in a person with schizophrenia, obsessive compulsive disorder, depression or someone with no psychiatric diagnosis, SSRIs will also impact individuals in ways that are not specific to diagnosis. Similarly, major tranquilisers (misnamed ‘anti-psychotics’) have also been advocated for the treatment of depression, anxiety disorders, bipolar affective disorder, personality disorders, as well as schizophrenia–a list that contains considerable overlap with that found for SSRIs.

The therapeutics of many psychiatric drug treatments also relies more on non-specific factors than disease-specific therapeutic effects. For example, it is generally assumed that drugs marketed as ‘antidepressants’ work through their pharmacological effects on specific neurotransmitters in the Central Nervous System, reversing particular states of ‘chemical imbalance’. However, the evidence points to placebo effects being more important than any neuro-pharmacological ones. Thus, several meta-analyses have concluded that most of the benefits from ‘antidepressants’ can be explained by the placebo effect, with only a small amount of the variance (and for most a clinically insignificant amount), attributable to the drug (Kirsch et al., 2008; Turner & Rosenthal, 2008). Thus, having a good relationship with the prescribing doctor is a stronger predictor of a positive response to an ‘anti-depressant’ than just taking the drug regardless of who prescribes it (Sparks, Duncan, & Miller, 2008).

The lack of treatment specificity is not limited to the more common and less severe presentations. Although drugs marketed as ‘antipsychotic’ are often claimed to reverse a biochemical imbalance in psychotic patients, no such imbalance has been demonstrated. Furthermore, academics have long been aware of a perplexing finding. Research carried out by the World Health Organisation (WHO) over the course of 30 years and starting in the early 1970s, found that patients outside the United States and Europe have much lower relapse rates, and are significantly more likely to have made a ‘full’ recovery and show lower degrees of impairment when followed up over several years, despite most having limited or no access to ‘anti-psychotic’ medication (Hopper, Harrison, Janka, & Sartorius, 2007). It seems that the regions of the world with the most resources to devote to mental illness–the best technology, medicines, and the best-financed academic and private-research institutions–had the most troubled and socially marginalised patients. Furthermore, meta-analyses of RCTs investigating the effectiveness of first and second generation neuroleptic drugs have found that, at best, the improvements seen are disappointingly limited (Bola, Kao, & Soydan, 2012; Lepping, Sambhi, Whittington, Lane, & Poole, 2011).

Unlike the rest of medicine, no overall improvement in prognosis has been demonstrated in Europe and North America over the past century for those diagnosed with a mental disorder. Some studies indicate the opposite - that compared to the past, there are now more patients who have developed chronic conditions. For example, in 1955, there were around 350,000 adults in the US state and county mental hospitals with a psychiatric diagnosis. During the next three decades when community based psychiatry became more prevalent, the number categorised as disabled from mental illness rose to 1.25 million. By 2007 the number of people categorised as disabled mentally ill grew to more than 4 million adults. Similarly, the numbers of youth in America categorised as having a disability because of a mental condition leapt from around 16,000 in 1987 to 560,000 in 2007 (Whitaker, 2010). A similar dramatic rise in numbers on disability support due to a mental illness, has been seen in the UK (Timimi, 2014). If the current national models of mental health service delivery were effective, we would not see this evidence of steadily worsening long term outcomes in parallel with steadily increasing expenditure.

As mentioned, long-term outcomes for serious mental disorders are worse in more industrialised than developing countries. The World Health Organisation's international outcome in schizophrenia studies found that after two years, about two thirds of the patients in less developed countries were doing well compared to only a third of the patients in the developed countries. The researchers concluded that being in a developed country was the strongest predictor of not attaining complete remission (Jablensky, 1992).

One problem with medical-model diagnostic approaches is that many of the diagnoses are conceived as conditions that are genetic and lifelong in nature, where the best one can hope for is gaining some control over symptoms (through, for example, use of medications). As such, psychiatric diagnoses can foreclose meaning by transforming a range of experiences, and possible meanings that can be applied to these experiences, into a narrow disease framework, limiting the cultural imagination to particular expectations.

Prognosis for those with mental disorders is also further hampered by the stigma associated with the medical model (Sayce, 2000). Nearly all studies that have looked at public attitudes towards mental illness have found that the medical model for mental illness (mental illness is an illness like any other illness) is associated with increased negative attitudes, greater fear of patients, and a greater likelihood of wanting to avoid interacting with them (Angermeyer, Holzinger, Carta, & Schomerus, 2011; Read, Haslam, Sayce, & Davies, 2006).

Similar findings emerge in personal stories of those diagnosed with a ‘mental illness’. Many of these stories describe how some users of mental health services felt stigmatised and marginalised by a psychiatric diagnosis (Romme, Escher, Dillon, Corstens, & Morris, 2009; Sayce, 2000). Being labelled with a chronic ‘genetic’ condition such as ‘schizophrenia’ interferes with a person's identity and biography. Indeed, the presence of ‘insight’ (meaning accepting you are mentally ill and need medical treatment) in schizophrenia has been found to be associated with lower self-esteem, despair, and hopelessness (Bassman, 2000) and is negatively correlated with emotional well-being, economic satisfaction, and vocational status (Hasson-Ohanon, Kravetz, Roe, David, & Weiser, 2006; Leff & Warner, 2006; Warner, 2010). Thus, accepting the medical-model attitude to diagnosis can bring expectations of a gloomy outlook with lifelong dependency on psychiatric treatment and reduced chances of a good recovery. For some, therefore, rejecting the diagnosis may be understood as a positive way of coping with the implications of the diagnosis for personal identity.

For the last few decades, Western mental-health institutions have been pushing the idea of ‘mental-health literacy’ on the rest of the world. A new global campaign for greater ‘recognition’ of mental illnesses in the non-industrialised world has developed, which assumes that ICD/DSM descriptions are universally applicable categories (World Health Organisation [WHO], 2010). Cultures are viewed as becoming more ‘literate’ about mental illness the more they adopt Western biomedical conceptions of diagnoses like depression and schizophrenia. In the process of doing this we not only imply that those cultures that are slow to take up these ideas are in some way ‘backward’, but we also export disease categories and ways of thinking about mental distress that were previously uncommon in many parts of the world. Thus, conditions like depression, post-traumatic stress disorder, and anorexia appear to be spreading across cultures, replacing indigenous ways of viewing and experiencing mental distress and opening up lucrative new markets for Western drug companies (Summerfield, 2008; Watters, 2009).

The idea of the individual as the locus of the self is a relatively recent Western invention and such a framework creates the psychological pre-conditions necessary for accepting the ‘atomised’ social worlds that have been created. Yet mental wellbeing seems closely connected to social and economic factors. Several international studies have concluded that the greater the inequality (in economic and social resources) in any society, the poorer is the mental health of that society (Friedli, 2009; James, 2007; Pickett, James, & Wilkinson, 2006; Pickett & Wilkinson, 2010).

A subtler source of impact on cultural beliefs is due to psychiatric diagnoses inadvertently setting standards for ‘normality’, by categorising what emotional and behavioural traits and experiences should be considered ‘disordered’. As the criteria for diagnoses are arrived at by subjective judgments rather than objective evidence (being literally voted in or out of existence by committees), they will have an automatic bias toward the cultural standards found in economically dominant societies (who also tend to control what counts as ‘knowledge’ globally). This sets in motion a diagnostic system vulnerable to institutional racism in the dominant societies and colonialism in others, as other standards of normality will, at least to some extent, come to be viewed as ‘primitive’, ‘superstitious’ etc., and their populations will be viewed as needing to be (psycho)educated (Timimi, 2012). As a result, then, for the majority of the world, all manner of complex somatic/emotional complaints have to be re-categorised, spiritual explanations have to be denounced, parenting practices viewed as oppressive, and so on.

Thus imposing Western medical-model DSM/ICD-style psychiatry on non-Western populations, risks: the undermining of existing cultural strategies for dealing with distress, more not less stigma, and the imposition of an individualistic approach that may marginalise family and community resources, and diverting attention from social injustice.

Diagnostic thinking has had a significant impact on service provision and public and professional beliefs about mental distress. As a result of popularising the diagnostic systems, it is widely argued that a significant proportion of the population suffers from mental illness, that this amounts to a significant economic burden, and that there is a strong case for investing in improved mechanisms of detection and treatment for these disorders. Across several surveys in the industrialised nations, only about a third of those identified as suffering a mental-health problem (according to DSM/ICD criteria) sought, or were interested in seeking, professional help (Andrews, Issakidis, & Carter, 2001; Kessler, McGonagle, & Zhao, 1994). This has been interpreted as unsatisfactory case detection, provision and treatment, due to public and professional ignorance. However, there is little evidence to support the idea that popularising mental health diagnoses, convincing professionals and the public about the high prevalence of mental disorders, and convincing policy makers of the need to diagnose and treat more people, benefits the mental health of the society.

In order to increase rates of diagnosis and treatment, a variety of campaigns have been undertaken. For example, in the UK the Royal College of Psychiatrists and Royal College of General Practitioners launched their ‘Defeat Depression’ campaign in the early 1990s (Paykel & Priest, 1992). It was intended to raise public awareness of depression, reduce stigma, train general practitioners in recognition and treatment, and make specialist advice and support more readily available. Evaluations of this campaign found no evidence that it led to any significant improvements in clinical outcomes, but was associated with a rapid increase in antidepressant prescribing (Croudace, Evans, & Harrison, 2003; King, Davidson, & Taylor, 2002; Thompson, Kinmonth, & Stevens, 2000).

Similarly research evaluating the ‘Beyond Blue’ campaign in Australia, which aims to increase awareness about depression, demonstrated an increase in medical model beliefs about depression and an increase in rates of diagnosis of major depression between 1998 and 2008. It also noted a decrease in mental health related quality of life over this decade, and that having mental health literacy (i.e., a belief that depression was a medical diagnosis requiring medical treatment) that was categorised as poor or fair, was a significant protective for major depression (Goldney, Eckert, Hawthorn, & Taylor, 2010).

Unlike other areas of public health, mental health in those societies with the most developed services appears to be the poorest. In such societies, ‘epidemics’ of psychiatric diagnoses (e.g., Attention Deficit Hyperactivity Disorder, autism, depression, bipolar disorder) have only emerged and become popularised in recent years. Whilst there are complex political, social and cultural reasons for this, they are in part due to popularisation of diagnoses and its subsequent influence on beliefs about the nature of personhood and the meanings attached to states of distress.

For a diagnostic system to establish itself as scientifically useful, its categories should be able to ‘carve nature at its joints’. For a diagnostic system to establish itself as clinically useful it should show that use of diagnostic labels aids treatment decisions in a way that impacts on outcomes. As reviewed above there is little evidence to support either position. There is much evidence to suggest that instead, they can cause significant harm. The only evidence-based conclusion therefore is that formal psychiatric diagnostic systems like ICD and DSM should be abolished.

We can and should do better. We have all the evidence we need to work on re-organising our approaches locally, nationally, and internationally to develop services that use evidence-based paradigms and can reduce the amount of harm DSM/ICD has caused, at the same time as improving outcomes. Paradigms that draw on the existing evidence for what improves outcomes and that incorporates the views of those who matter most–service users–can easily be developed and implemented. The message from this research is that services can improve outcomes by concentrating on developing meaningful relationships with service users that fully include them in decision-making processes (Bracken et al., 2012) and that incorporate service user feedback to help with delivering flexible treatment models where there is regular testing of whether or not a particular intervention is improving outcomes for that service user (Duncan, 2012; Timimi, Tetley, Burgoine, & Walker, 2013). International service user-led movements, such as the ‘recovery’ movement, that focus on the inclusion of people in recovery from mental-health problems as collaborators in research, service development and treatment model development, provide good examples of how this evidence can be developed to change institutional culture (New Freedom Commission on Mental Health, 2003; Shepherd, Boardman, & Burns, 2008).

The real gift of psychiatry to medicine is an understanding of the person in their context leading to an integrated whole-person model of healthcare. Psychiatry has to sit at the confluence of a variety of disciplinary discourses (sociology, anthropology, psychology, philosophy, biology, politics etc.), and it is this broader understanding of the person and their wellbeing that psychiatry brings. By importing the diagnostic model from general medicine, we end up miss-selling and under-utilising the unique skills the profession of psychiatry brings to healthcare by the ‘dumbing down’ of what we do into simplistic, diagnosis-driven protocols that have more to do with successful consumer culture marketing than with science. Changing to more evidence-compatible paradigms is now long overdue.