Contrary to popular belief, psychiatric diagnoses like major depressive disorder, generalised anxiety disorder, and schizophrenia are not always stable over time. When a person first presents with symptoms of a mental health disorder, even the most experienced clinician can find it challenging to make a reliable diagnosis.
Despite this knowledge and supporting research (e.g., de la Vega et al., 2018; Wood et al., 2021), psychiatric diagnoses are often treated like they are fixed, which can have detrimental consequences for patients, clinicians, and researchers. For example, if a patient is diagnosed with depression but over the course of a year this comes to be experienced as bipolar disorder, the patient might end up receiving care which is not effective. Not only is this potentially dangerous and a waste of the patient’s time, but it also uses up resources which are already stretched thin. There are clear practical and theoretical benefits to knowing more about the stability of psychiatric diagnoses.
This blog summarises a recent cohort study by Jørgensen et al. (2022), who sought to address this issue by mapping out subsequent diagnoses over a 10-year period following an initial psychiatric diagnosis at hospital admittance.
Jørgensen et al. (2022) analysed data on all patients experiencing their first contact with psychiatric hospitals in Denmark from 1995 to 2008. Individuals were included in the study if they were 18 years or older and experienced their first psychiatric diagnosis in that period, as long as that diagnosis was one of the 20 most frequent psychiatric diagnoses. This allowed for 10 years of follow-up for each patient, ending in December 2018.
The researchers used social sequence analyses with entropy measurements, starting with the initial diagnosis and assigning subsequent diagnoses in six-month increments. They used Cox proportional hazards regression models to assess the risk of receiving a subsequent diagnosis (at the one-cipher level, F0–F9) after each first-time diagnosis.
Diagnostic data were included for almost 185,000 people, with women (n = 107,820; 58%) outnumbering men (n = 77,129; 42%). Median age was 42.5 years, with a standard deviation of 18.5 years, and range of 18 to >100. Ethnicity was not recorded and therefore not available for analysis.
Over the 10 years of follow-up, almost half (46.9%) of all patients had at least one subsequent diagnosis that differed from their initial diagnosis. Across diagnoses, most diagnostic changes happened within the first 5 years.
The likelihood of receiving a psychiatric diagnosis different than the first diagnosis varied significantly across diagnostic categories, ranging from 23% to 87%. The highest variability occurred with persistent delusional disorders, mental and behavioural disorders due to multiple drug use and use of other psychoactive substances, and acute and transient psychotic disorders. The lowest variability occurred with eating disorders, non-organic sexual dysfunction, and somatoform disorders.
The following associations between first-time diagnoses and subsequent diagnoses had the highest adjusted hazard ratios (≥ 3.00):
- Mental and behavioural disorders due to use of cannabinoids x schizophrenia, schizotypal, and delusional disorders: 4.03 (p < .001)
- Manic episode x schizophrenia, schizotypal, and delusional disorders: 3.49 (p < .001)
- Acute and transient psychotic disorders x mood disorders: 3.03 (p < .001)
- Delusional disorder x intellectual disability: 5.90 (p < .001)
- Acute and transient psychotic disorders x intellectual disability: 4.70 (p < .001)
- Schizotypal disorder x disorders of psychological development: 3.14 (p < .001)
Clinicians interested in knowing more about the likelihood of specific subsequent diagnoses following specific initial diagnoses will want to consider the detailed table and figures within the article.
Whether or not a clinician is aware of the provisional nature of an initial psychiatric diagnosis, patients and their families often experience such diagnoses as being as solid and reliable as the diagnosis of a broken bone. However, the results of this study show that initial psychiatric diagnoses should be understood as provisional, with an almost 50% likelihood of change over time.
These findings provide important information for frontline psychiatric clinicians, with the authors concluding that,
This information could help clinicians to […] inform patients and families on the degree of diagnostic uncertainty associated with receiving a first psychiatric hospital diagnosis, as well as likely and unlikely trajectories of diagnostic progression.
Additionally, the researchers observed that sometimes the first diagnosis is an early manifestation of a developing illness, such as when schizotypal disorder or transient psychosis is a precursor to schizophrenia, or when a depressive episode is the first sign of bipolar. Finally, some disorders have an inherently higher likelihood of comorbidities, in the way that personality disorders increase the risk of substance abuse or depression.
Strengths and limitations
- This is a large, population-wide study, which gives it a breadth, depth, and reliability that very few research studies can claim.
- The findings have much practical utility, helping clinicians both in planning follow-up for newly diagnosed patients, and in informing patients and their families about the uncertainty and instability of the initial diagnosis.
- The study included only patients who were seen in hospitals. Most mild to moderate cases of anxiety, depression, and substance use disorder are treated in the primary care sector or municipal clinics, which means that the findings of the study apply only to more severe cases of these illnesses.
- There was a substantial discrepancy across sex of the participants: 58% were female and 42% were male. It would be interesting to know whether there are also sex differences in the likelihood of subsequent diagnoses.
- Ethnicity data was not available to the researchers, but this might also be valuable for the work of clinicians in communicating with patients and in planning follow-up after primary diagnoses.
- Physical illness was another factor not included in these analyses, which might further inform the diagnostic process. We know there are psychological effects of long-term physical conditions (read my Mental Elf blog on a paper by Carroll et al. (2022) here), and that serious physical conditions can increase the likelihood of suicide (as explored in my blog on a paper by Nafilyan et al. (2023)). Future studies might focus on the nature of the psychiatric diagnostic process in patients who also have a serious physical illness.
- The study’s results are described only broadly in The Lancet Psychiatry article, with the details concerning specific disease trajectories left for the reader to decipher in complex visual format via a table and figures. I found some of the information in the figures to be highly challenging in their complexity, and would love to see the detailed results explained verbally, or perhaps conveyed graphically but more simply. This could increase many readers’ comprehension of the findings, and thereby increase clinicians’ ability to put the results into practice.
- There is a rich vein of information here waiting to be analysed, synthesised, and rendered useful to busy practitioners in a number of specific areas. For example, for a clinician working mainly with clients with substance use disorders, what’s useful in the data generated in this study? This is not so much a limitation of the study, but rather an opportunity for further development of its findings.
Implications for practice
These findings inform clinicians about the instability of initial psychiatric diagnoses, both generally and how this might manifest differently across different diagnoses. This knowledge can help clinicians advise patients and their families about their diagnoses more thoughtfully and accurately, giving them a better understanding of the probable trajectory of their conditions.
The findings also emphasise the importance of clinicians putting in place a plan to follow up with patients, checking to see if there is a need to adjust diagnosis and/or treatment. This is especially important during the first 5 years after diagnosis, and for certain diagnostic categories which may need more frequent follow-ups, such as persistent delusional disorders, mental and behavioural disorders due to multiple drug use and use of other psychoactive substances, and acute and transient psychotic disorders.
The findings of this study can help clinicians recognise the provisional nature of their initial diagnoses. This can relieve them of some of the stress of determining exactly what’s happening with a specific patient, perhaps leading practitioners to view diagnosis as a process that takes several months or even years, rather than an instant labelling that can be based on conclusions from one event or hospital visit.
The results of this study are consistent with the conclusions of Caspi et al. (2021), a paper which I blogged about in 2021, in suggesting that mental disorders are best understood and treated when there’s more emphasis on the whole person, and less reliance on specific diagnoses. Perhaps, for example, we should describe psychiatric patients as “showing signs consistent with schizophrenia” rather than as “having schizophrenia” or (worse) “being a schizophrenic”.
Statement of interests
Jørgensen, T. S. H., Osler, M., Jorgensen, M. B., & Jorgensen, A. (2023). Mapping diagnostic trajectories from the first hospital diagnosis of a psychiatric disorder: a Danish nationwide cohort study using sequence analysis. The Lancet Psychiatry, 10(1), 12-20.
Carroll, S., Moon, Z., Hudson, J., Hulme, K., & Moss-Morris, R. (2022). An Evidence-Based Theory of Psychological Adjustment to Long-Term Physical Health Conditions: Applications in Clinical Practice. Psychosomatic Medicine, 84(5), 547-559.
Caspi, A., Houts, R. M., Ambler, A., Danese, A., Elliott, M. L., Hariri, A., … & Moffitt, T. E. (2020). Longitudinal assessment of mental health disorders and comorbidities across 4 decades among participants in the Dunedin birth cohort study. JAMA Network Open, 3(4), e203221-e203221.
de la Vega, D., Piña, A., Peralta, F. J., Kelly, S. A., & Giner, L. (2018). A review on the general stability of mood disorder diagnoses along the lifetime. Current Psychiatry Reports, 20, 1-10.
Matthews, D. (2021). Mental disorders start early and vary across the lifespan: it’s time to pay attention to the whole person, and less to the diagnosis #IoPPNfestival. The Mental Elf.
Matthews, D. (2022). A long-term physical health condition changes everything: therapeutic responses to psychological distress must change too #BABCP2022. The Mental Elf.
Matthews, D. (2023). Does a diagnosis of severe physical illness elevate suicide risk? The Mental Elf.
Nafilyan, V., Morgan, J., Mais, D., Sleeman, K. E., Butt, A., Ward, I., Tucker, J., Appleby, L., & Glickman, M. (2023). Risk of suicide after diagnosis of severe physical health conditions: A retrospective cohort study of 47 million people. The Lancet Regional Health – Europe, 25, 100562.
Wood, A. J., Carroll, A. R., Shinn, A. K., Ongur, D., & Lewandowski, K. E. (2021). Diagnostic stability of primary psychotic disorders in a research sample. Frontiers in Psychiatry, 12, 734272.