Transition to schizophrenia

In our study the most common diagnostic change of first-episode ATPD was to schizophrenia (12.6%) and in general to schizophrenia spectrum disorders (22.3%). According to our data, one in eight patients with first-episode ATPD will be diagnosed with schizophrenia over an average of 1.7 years. Moreover, one in five of those who developed schizophrenia were younger than 20 years. Our data are consistent with previous studies that found the predominant pattern of diagnostic change from first-episode ATPD was to schizophrenia spectrum disorders. ^{Reference Singh, Burns, Amin, Jones and Harrison20,Reference Castagnini, Bertelsen and Berrios21,Reference Chang, Pang, Chung and Chan28,Reference Moller, Jager, Riedel, Obermeier, Strauss and Bottlender29} Interestingly, some earlier reports appear to indicate the type of ATPD subsyndromes may have some bearing on subsequent diagnostic revisions, with the ‘polymorphic’ subgroups being more likely to transition to affective disorders and the ‘non-polymorphic’ subgroups being more likely to be later reclassified as schizophrenia. ^{Reference Sajith, Chandrasekaran, Sadanandan Unni and Sahai26,Reference Chang, Pang, Chung and Chan28,Reference Salvatore, Baldessarini, Tohen, Khalsa, Sanchez-Toledo and Zarate30} It is worth noting that some previous studies reporting high stability rates for ATPD, included samples of patients with acute polymorphic psychotic disorder only. ^{Reference Abe, Otsuka and Kato24,Reference Sajith, Chandrasekaran, Sadanandan Unni and Sahai26} It would therefore appear that the presence of polymorphous (cycloid) features may predict a relatively more stable and more reliable diagnostic construct than a schizophrenia-like clinical picture, which may instead represent a prodromal phase of a schizophrenia spectrum disorder. Although there may be some evidence to the contrary, ^{Reference Singh, Burns, Amin, Jones and Harrison20} this view has been recently supported by data from the McLean-Harvard International First-Episode Project, which found after a 2-year follow-up that 67% of patients retained their initial diagnosis of acute polymorphic psychotic disorder with and without schizophrenic symptoms (F23.0 and F23.1) and only 28.6% of patients maintained their initial diagnosis of acute schizophrenia-like psychotic disorder. ^{Reference Salvatore, Baldessarini, Tohen, Khalsa, Sanchez-Toledo and Zarate30} Furthermore, Castagnini et al found that ATPD subgroups with schizophrenic symptoms (F23.1 and F23.2) were more prevalent in men and also showed an earlier age at onset than other ATPD subgroups, which the authors suggested may point to a close kinship with schizophrenia of these subgroups. ^{Reference Castagnini, Bertelsen and Berrios21} Based on this evidence, it has been anticipated that in the upcoming revision of ICD-10 the ATPD subgroups with schizophrenic symptoms may be collapsed into the ‘other primary psychotic disorders’ category and only the pure polymorphous subtype may be retained in the new definition of ATPD. ^{Reference Gaebel, Zielasek and Cleveland31}

Of all the recorded diagnostic changes, schizophrenia is the mental disorder with the worst prognosis and poorest long-term global functioning. Thus, the identification of features delineating those whose condition will convert to schizophrenia from the remainder of those with a first-episode ATPD could enable implementation of early-intervention strategies and subsequent amelioration of prognosis in this group of patients. In our sample, analysis of variance of diagnostic subgroup characteristics revealed male gender and younger age at first presentation to be significantly associated with diagnostic transition to schizophrenia. A non-significant trend for longer first admission to hospital in association with later development of schizophrenia was also found. Previous research indicated that poor premorbid social adaptation ^{Reference Marneros, Pillmann, Haring, Balzuweit and Bloink13} and early recurrence of psychosis ^{Reference Suda, Hayashi and Hiraga22} may be other useful predictors of future development of schizophrenia. Furthermore, as far as the phenomenology of ATPD is concerned, negative symptoms, ^{Reference Jager, Bottlender, Strauss and Moller14} rapidly changing clinical features and anxiety ^{Reference Marneros, Pillmann, Haring, Balzuweit and Bloink32} during the acute psychotic episode have been found to be discriminating factors for schizophrenia and ATPD respectively. Even though the presence of these early predictors should arouse clinical suspicion of later transition to schizophrenia, longitudinal follow-up remains the only reliable method to elucidate diagnosis. For this reason, of particular clinical relevance was the survival analysis regarding this subgroup of patients. Kaplan-Meier estimates of the transition risks were 80% at 2.8 years and 90% at 4.6 years. The mean time to change to a diagnosis of schizophrenia in our sample was 1.7 years. Men turned out to switch sooner than women (1.5 v. 2.2 years). Of particular note, in our sample, is that the chance of earlier development of schizophrenia was 74% higher when duration of first hospital admission exceeded 14 days, 68% higher if age at onset was younger than 30 and 62% higher for male gender. Taken together, the above findings suggest that young age at first presentation, male gender and non-rapidly remitting psychotic symptoms in the context of the emergence of clinical features resembling ATPD may be an early manifestation of an underlying schizophrenic disease process and that close monitoring of patients with first-episode ATPD with such characteristics over a follow-up period of 3 years or more is warranted in order to confidently exclude a diagnosis of schizophrenia.

Incidence

Our study found an overall incidence rate for patients admitted to Scottish hospitals with first-episode ATPD of 4.1 per 100 000 population. The range of the annual incidence rates was between 3.3 and 4.7 per 100 000 population. Previously, Singh et al ^{Reference Singh, Burns, Amin, Jones and Harrison20} found an annual incidence rate of 3.9 per 100 000 in the population of Nottingham, and Castagnini et al ^{Reference Castagnini, Bertelsen and Berrios21} found an annual incidence rate of 9.6 per 100 000 in the Danish population using data drawn from the Danish psychiatric central register. Both these previous studies used incidence cohorts, which also included patients treated in out-patient services. It can be argued that because of the lack of data from Scottish out-patient clinics our incidence rate may be a slight underestimation of the overall incidence rate of ATPD in the Scottish population, but our figures are in line with those of Singh et al, and the typically florid and dramatic nature of ATPD psychopathology means that only a minority are likely to be successfully managed in the community without requiring admission to hospital. Hence, our incidence rate is likely to represent an accurate estimate of the incidence rate of ATPD in the Scottish population. The much higher rates in the Danish population of Castagnini et al ^{Reference Castagnini, Bertelsen and Berrios21} may reflect diagnostic habits regarding the concept of reactive psychosis, previously commonly used in Denmark and then incorporated into the paradigm of ATPD in ICD-10, possibly resulting in an overestimation of its incidence rate. It is also worth noting that the stability of the incidence of new-diagnosis ATPD that we observed over the 13-year recruitment period suggests a consistency of diagnostic practice across time in psychiatrists practising in Scotland, which further supports the validity of our results.

Gender

In our sample we found a slight male preponderance with a male/female ratio of 1.19. This compares with a reduced prevalence of ATPD in women reported by Castagnini et al with a male/female ratio of 0.93. ^{Reference Castagnini, Bertelsen and Berrios21} Singh et al instead found a marked male preponderance with a male/female ratio of 1.87. ^{Reference Singh, Burns, Amin, Jones and Harrison20} This latter finding appeared to be accounted for by a higher rate of false positive diagnosis of ATPD in men then in women. In our sample too, men were less likely to retain their initial diagnosis than women, so that if the ‘true’ incidence rate of ATPD was to be considered there would be a female preponderance. Indeed, even though so far clinical consensus has been that ATPD mainly affects females, ^{Reference Marneros1} a number of previous studies showing female preponderance have either focused exclusively on the acute polymorphic psychotic disorder (F23.0) subcategory ^{Reference Suda, Hayashi and Hiraga22,Reference Sajith, Chandrasekaran, Sadanandan Unni and Sahai26} or have examined a cohort of individuals where this subcategory was overrepresented. ^{Reference Jorgensen, Bennedsen, Christensen and Hyllested18} Hence, it would appear that gender ratio may be a reflection of the relative proportion of ATPD subcategories in the sampled population. It is thus not possible to draw any definite epidemiological conclusions on gender ratio in ATPD from our study because of the lack of ATPD subcategories in our data-set.

Age at onset

Our findings of an early-middle adulthood onset are in line with the previous studies of ATPD. Our mean age at onset (37.4 years) is akin to those previously reported by Jorgensen et al (37 years), Marneros et al (35.8 years) and Castagnini et al (42.2 years). ^{Reference Jorgensen, Bennedsen, Christensen and Hyllested18,Reference Marneros, Pillmann, Haring, Balzuweit and Bloink19,Reference Castagnini, Bertelsen and Berrios21} We found a significantly higher mean age at onset in women than in men (41.8 v. 33.8 years) and a reversal of the pattern of age distribution with a higher proportion of women than men presenting after 40 years of age. Previously, Castagnini et al had found a similar later mean age at onset of ATPD in women (46.2 v. 37.8 years) and a similar reversal of the pattern of incidence rates by gender groups over 50 years. ^{Reference Castagnini, Bertelsen and Berrios21} This latter finding is consistent with the evidence that psychotic symptoms tend to emerge later in women, perhaps owing to a protective effect exerted by oestrogens via their neuromodulatory and neuroprotective activity on the brain. ^{Reference Boerma, van der Stel, van Amelsvoort, Linszen and de Haan33} Also our data seem to confirm that average age at onset of ATPD is higher than that of schizophrenia. ^{Reference Marneros, Pillmann, Haring, Balzuweit and Bloink13,Reference Salvatore, Baldessarini, Tohen, Khalsa, Sanchez-Toledo and Zarate30}

Diagnostic stability

The overall diagnostic stability rate in our cohort was estimated at 53.9% over approximately 4 years, including patients presenting with recurrent (7.6%) and mono-episodic (46.3%) ATPD. This is similar to that previously found by Marneros et al ^{Reference Marneros, Pillmann, Haring, Balzuweit and Bloink19} (54%), whose follow-up examinations took place on average 10.2 years after the first episode, but differs from those reported by Singh et al ^{Reference Singh, Burns, Amin, Jones and Harrison20} (34%) and Castagnini et al ^{Reference Castagnini, Bertelsen and Berrios21} (39%), whose observation period lasted for 3 years and 6 years respectively. Of particular note is the observation that high stability rates for ATPD, ranging from 63 to 100%, have been reported in studies conducted in low- and middle-income countries. ^{Reference Amini, Alaghband-rad, Omid, Sharifi, Davari-Ashtiani and Momeni23,Reference Thangadurai, Gopalakrishnan, Kurian and Jacob25,Reference Sajith, Chandrasekaran, Sadanandan Unni and Sahai26} One interpretation of these results has been that the concept of ATPD may be applied differently to different cultural groups on account of cultural variations in the manifestation of its typical psychopathology.

Little is known about the temporal pattern of diagnostic shift in ATPD. In our research, due to the staggered entry of patients into the study we had to accommodate varying lengths of follow-up time among different patients. Kaplan-Meier analysis with censoring was used to factor in these missing data and it revealed a mean time to any diagnostic shift from first-episode ATPD of 7.6 years. Furthermore, the spectrum of observed diagnostic shifts appears to be encompassing a wide variety of psychiatric disorders as it is shown in Fig. 1. Hence, although it is possible our ATPD diagnostic stability rate is a slight over-estimate as it may include misdiagnosed mental disorders or non-specific early manifestation of other psychiatric vulnerabilities, it could also be argued that changes in diagnosis after 6 or 7 years of first-episode ATPD reflect independent risks of other underlying psychiatric problems rather than an evolving disorder that first presented with ATPD. Although our data would appear to suggest that the general ICD-10 diagnostic criteria for first-episode ATPD are of limited clinical utility, lacking as they are much in the way of predictive diagnostic specificity, it is still possible that additional criteria for specific ATPD subcategories such as the polymorphous subtype may instead reliably identify longitudinally stable clinical syndromes of benign prognosis and outcome.

Recurrence rate

In accordance with existing data our study also revealed that 1 in 13 (7.6%) of these individuals with acute and rapidly remitting psychoses will experience recurrences of similar psychotic episodes. The presence of a recurrent pattern with complete inter-episodic recovery in a proportion of patients with ATPD has been long recognised. Earlier reports found rates of a recurrent course in ATPD ranging from 10 to 50% ^{Reference Singh, Burns, Amin, Jones and Harrison20,Reference Castagnini, Bertelsen and Berrios21,Reference Thangadurai, Gopalakrishnan, Kurian and Jacob25,Reference Jager, Hintermayr, Bottlender, Strauss and Moller34} but this broad range could be accounted for by differences in sampling methods and study design. Interestingly, a chronic course, which is somewhat at odds with the concept of acute transient psychoses, has also been described. ^{Reference Moller, Jager, Riedel, Obermeier, Strauss and Bottlender29,Reference Jager, Bottlender, Strauss and Moller35} Of note, Marneros et al found no significant difference in the frequency of relapses between ATPD and schizophrenia. ^{Reference Marneros, Pillmann, Haring, Balzuweit and Bloink13} These findings challenge the Kraepelinian dichotomy of functional psychoses and are more in keeping with several European psychiatric traditions as to the existence of psychotic syndromes with a favourable and relatively benign outcome. Unfortunately, the operationalised diagnostic criteria used in the current international taxonomies seem to be unable to reliably identify these individuals. In our sample neither gender nor age at first presentation appeared to be parameters significantly associated with this subgroup of patients. However, a non-significant trend for shorter length of first admission was observed.

Limitations

Although it is possible that the steady decline of psychiatric in-patient beds recorded across Scotland over the recruitment period and the potential shortening of duration of hospital admission following the introduction of home intensive treatment services in some Scottish NHS Boards may have biased the statistics on ATPD incidence rates and average length of hospital stays, it is questionable whether such changes in healthcare provision may have significantly had an impact on the clinical management of short-lived and typically florid psychoses. No information regarding deaths occurring during the follow-up period could be extracted from our data and the effect this might have had on our findings could not be estimated. Diagnostic accuracy is a common problem with administrative data since clinical diagnoses cannot be verified and validated. It is unlikely that this constitutes a major problem since uniformity of psychiatric training across Scotland and our large national cohort will have minimised the impact on the results of any diagnostic discrepancies. Moreover, the naturalistic design of the study has precluded the analysis of any confounding effect that potential therapeutic interventions may have exerted on the observed outcomes. Unfortunately, due to the nature of the data it is also impossible to exclude any previous presentation with psychotic symptoms to mental health services outside Scotland and any subsequent relapse that may have been managed in the out-patient clinic or may have lead to hospital admission outside Scotland. The lack of additional coding regarding ATPD subcategories does not permit any consideration on the clinical utility of the ‘polymorphous’ variety of ATPD. Also, information is missing on whether onset of symptoms was abrupt (<48 h) or acute (>48 h but <2 weeks), with the former having been reported to be associated with higher diagnostic stability. ^{Reference Sajith, Chandrasekaran, Sadanandan Unni and Sahai26}

Implications

According to a synthesis of our and other available data, ATPD requiring hospital admission is a rare mental disorder with an incidence of about 4 per 100 000 per year, which typically presents in early-middle adulthood. The gender ratio is more or less equal, although women appear to be more likely to retain their initial diagnosis of ATPD than men. As a diagnostic concept, ATPD remains inadequately validated and of limited longitudinal diagnostic stability. Around one in two people initially diagnosed with ATPD will change to a different diagnosis and one in eight will eventually develop schizophrenia. Hence, acute onset and full recovery within 1-3 months do not seem to be always associated with long-term favourable outcome. Nonetheless, almost a half of people presenting with ATPD may only have a single episode. There also appears to be a group of people who retain the diagnosis and present with recurrent episodes of illness, lending support to the view that some still poorly defined acute and rapidly remitting psychoses are not merely prodromal phases of affective or schizophrenic psychoses but a separate clinical entity, which does not fit into the traditional Kraepelinian dichotomy of functional psychoses. In this seemingly heterogeneous population of ATPD, there seems to be clinical (longer first admission to hospital) and demographic (young age at onset and male gender) variables that are associated with increased risk and earlier development of schizophrenia. Monitoring over a follow-up period of around 3-5 years may therefore be warranted in order to confirm or exclude a diagnosis of schizophrenia.