DUP is generally considered a significant predictor of outcome, particularly for schizophrenic disorders [
24]. Indeed, a longer DUP has been associated with a worse clinical and psychosocial outcome, frequently interpreted as a consequence of a more intense and rapid progression of the neurodegenerative process in the first years of untreated illness. This findings lends support to the importance of the early treatment of schizophrenia, repeatedly associated with a better prognosis [
25‐
28]. However, the predictive value of DUP has been acknowledged mainly for cases at onset and with regard to the short- and medium-term periods [
29], whilst its role in predicting long-term prognosis remains to be clarified [
30].
In this study DUP proved to be an important predictor of clinical and functional outcome of schizophrenia even in the very long term, as attested by data showing the association between a shorter DUP and higher frequency of more favorable course of illness, reduced rates of hospitalization and better overall functioning in the cohort of chronic schizophrenic outpatients studied. As previoulsy pointed out, a limited number of studies have provided data on the impact of DUP in the long-term (i.e follow-up of 10 years or more) with very few studies reporting data regarding very long term outcome (15 yrs or more); moreover, evidence provided by these studies is somewhat contradictory. Some studies reported a significant relationship between longer DUP and worse outcome [
12,
16,
31,
32]. In particular, a study involving a sample of 58 patients with schizophrenia followed for a mean period of 15 years, demonstrated a reduction in overall functioning and a more severe psychopathology among patients with a DUP of more than 1 year compared to those with a DUP of less than 6 months [
16]. On the contrary, several studies present in literature yelded opposite results [
15,
33‐
35]. In particular, a study of clinical and psychosocial outcomes in 55 adolescent patients with schizophrenia after a follow-up period of 15 years failed to demonstrate any predictive power of DUP on the different outcome measures considered [
33].Moreover, in a multicenter follow-up study of 349 patients, throughout the first two years of disease DUP did not appear to be predictive of levels of social disability observed in patients fifteen years later [
15]. Differences in methodology regarding the study design (i.e prospective vs retrospective), methods (i.e.standardized vs not standardized techniques for evaluating DUP), sampling (i.e first onset vs chronic or mixed cases; early onset vs not early onset) may explain the above mentioned conflictings results. However, in view therefore of the lack of published studies of both a prospective or retrospective nature on the impact of DUP in the very long-term (16 yrs or more) , the results obtained in the present survey of schizophrenic patients followed for an average of 25 years ( range 16- 33 yrs) would appear to be of particular relevance, considering that the demonstration of very long-term negative consequences of a prolonged delay in starting treatments tend to confirm the potential relevance of early treatment even on the long-term outcome of psychotic disorders [
36]. However, in evaluating these results, the possibility that a longer DUP may be related to severity of the illness, and thus may be a marker than a determinant of outcome cannot be ruled out [
9]. Moreover, the above mentioned lack of prospective or retrospective data in literature relating to DUP over such a long-term follow-up period implies the need for particular caution in interpreting the findings of the study. Indeed, further studies should be undertaken in an attempt to confirm the findings here reported . Likewise, caution should be applied in drawing conclusions from the findings of this study in view of its several limitations. The first limitation is represented by the retrospective nature of the data collected. This limitation is, at least in part, counterbalanced by the relatively good quality of data obtained from clinical records, which were stored by means of a structured data recording system in use in the university community mental health centre where the present study was performed. However, the lack of use of a specifically structured interview in collecting data pertaining to important aspect such as age at onset, age at first treatment, premorbid personality, and in particular DUP should be acknowledged. Similarly, the fact that only 22% of patients studied were at their first episode when first seen should be underlined. Consequently, data pertaining to onset of ilness were collected for the majority of patients after an average three years of illness, a time frame sufficient to favour recall biases both from patients and caregivers. Secondly, we acknowledge that this is a small study, with a limited number of cases examined, partly due to the selection criteria applied, including the exclusion of cases such as schizoaffective disorders, considered in other studies. This therefore considerably limits the power of the results obtained, although the inclusion in the sample only of cases for which reliable follow-up data were available probably enhanced the quality of the study. Third, the fact that the study considers a very selected sample of patients, namely only those who were in contact with the centre without interruptions for periods of six months or more should be acknowledged; patients who had died, moved away, refused to stay in treatment due to any reason, including scarce insight and/or very severe illness, or not having further need of continuing care, were not considered in out study. Therefore, what emerges from the present study should be applied only to patients who remain in long term treatment. Fourth, it is an acknowledged fact that a wide range of means of classifying “short” and “long” DUP is described in literature, (i.e. less than or exceeding 3, 6 or 12 months) and the method used in this study (DUP less than or exceeding 12 months) is not necessarily the best. Moreover, the dichotomization of DUP addresses the problem of skewness in duration of untreated psychosis, although is confounded by the loss of a great deal of information. On the other hand, the approach of dichotomizing DUP in “short” and “long” on the basis of a median split used in our study is one of the most conservative methods used applied to address these issues. Moreover, it should be pointed out that in regression analyses performed, DUP was introduced both as a continuous and a dichotomous variable, confirming in any case its value as explanatory variable predicting outcomes. Fifth, the involvement of only some potential confounding factors focusing on the relationship between DUP and outcome was excluded, including gender, familiarity for mental illness and age at onset, whilst others, such as premorbid personality, mode of onset, diagnostic subtyping could not be ruled out. However, regarding this point, it should be underlined that several recent reviews and meta-analyses reported the significance of DUP as a predictor of outcome, independent of these confounding variables [
3,
4].
In conclusion, the findings of future studies investigating large patient cohorts over an extended period of time will undoubtedly provide confirmation as to whether or not DUP is indeed a reliable predictor of outcome even in the very long-term, as indicated by the present study. In any case, the data presented here give further support to the hypothesis whereby early intervention, associated with an appreciable effect on the fate of patients, is capable of altering the prognosis of schizophrenia even in the long term.