Background
It is widely acknowledged that recovery may be achieved even in subjects affected by serious mental illnesses. However, the dimensions to be included in the concept of recovery are still the object of ongoing debate [
1]. It is generally assumed that recovery will comprise both objective and subjective [
2] components, otherwise defined as clinical and personal domains [
3]; the objective component generally refers to clinical outcomes which are evaluated by means of operationally defined criteria; subjective recovery refers to the ongoing process of positive changes in an individual’s subjective experience of themselves as human beings [
1]. Clinical objective and personal subjective recovery are largely independent phenomena [
4] and both should be considered as targets for therapeutic interventions in schizophrenia and related disorders. Symptom remission represents the fundamental component underlying clinical recovery, together with improved functioning [
1], and is viewed as the main target for psychopharmacological interventions [
5]. Up until fairly recently, a univocal method for the assessment of remission was lacking; to this aim, a significant step forward was represented by publication of the Remission in Schizophrenia Working Group criteria (RSWGcr) [
6] which proved to be conceptually viable and easy to use both in clinical trials and clinical practice [
7].
According to RSWGcr [
6] clinical remission is based on a symptom severity criterion comprising eight items of the PANSS scale chosen as being the most diagnostic-specific for schizophrenia, and a duration criterion, thus excluding symptom domains not diagnostically relevant for the disorder. As PANSS scale provides ratings investigating not only symptom severity per se but also functional impairment, a score of “mild” or better (i.e. 3 points or less) at all eight “core” symptoms was considered sufficiently representative of a level of impairment consistent with symptomatic remission of the disorder [
7]. According to recent reviews [
8,
9] reported remission rates vary widely across studies (17-88%), likely due to use of symptom severity criterion alone in the majority of studies [
9]. A number of studies have demonstrated the validity of these remission criteria using two different approaches, namely comparison of different definitions of symptomatic remission and association of remission criteria with various outcome dimensions, mainly overall symptomatic status and functional outcome [
10]. However, several recent studies seem to highlight the potential limitations of severity criteria as currently conceived in predicting functioning and other important outcome variables. Indeed, a study aimed at investigating symptomatically remitted and non remitted patients demonstrated a significantly better level of functioning for remitted patients, although the latter continued to display significant areas of inadequate functioning, low levels of subjective wellbeing and moderate-severe emotional distress [
11]. Moreover, a recent study attempting to provide an ecological validation for the symptomatic remission criterion, showed how although remitted patients reported fewer positive symptoms, better mood states and partial recovery of reward experience, remission status was not related to functional recovery [
12]. Starting from these premises, and taking into account the need for further investigation into the validity of current criteria for symptomatic remission, the present study was devised to compare the efficacy of three different and increasingly “stringent” sets of criteria in evaluating remission by means of PANNS in relation to functional and cognitive status.
Discussion
The RSWG remission criteria have previously been compared with criteria proposed by other authors [
10]. Other studies have evaluated RSWGcr versus a modified version in terms of number of items included [
31] or cut-off scores at each core item [
32]. One study focused on the evaluation of accuracy of RSWGcr using PANSS total score as a “golden standard” [
33]. Finally, another study [
30] compared four definitions of remission based upon severity scores at SAPS and SANS instead of PANSS. To our knowledge, this is the first study to compare RSWGcr with their modified versions, by extending the number of items of PANSS used to evaluate remission.
In our sample 50% of subjects were in clinical remission according to RSWGcr [
6], a proportion that decreased significantly by approximately one third using PANSS-PNScr, and was halved when adopting PANSS-TScr. These results are in contrast with those of van Os et al. [
31] who found no substantial change in remission rates when including two PANSS items (namely, depression and suicidality) to the eight “core symptoms”. This discrepancy however may be explained considering that we adopted more stringent alternative criteria than those used by van Os et al. [
31]. Our results are somewhat similar to those obtained by Beitinger et al. [
32] who reanalysed data from six antipsychotic trials applying more stringent criteria with regard to cut-off scores used by RSWGcr; indeed, the frequency of remitted patients using the original RSWGcr was approximately 42% in both medium-term and long-term studies; using scores ≤ 2, remitted subjects were 16% and 13%, respectively; using a score of 1 they were respectively 3.4% and 5%. Based on these results the authors concluded that a choice of severity score ≤3 was a “realistic” choice, given that “more stringent thresholds yield remission frequencies that are not realistic”. In our study, not based on different scoring thresholds but rather on an extension of the number and type of PANSS items considered, a significant reduction of remission rates was obtained using alternative criteria, but which was not so marked as to be unrealistic. Moreover, clinical status evaluated by PANNS and CGI was invariably significantly better among remitted patients, independent of remission criteria adopted. It is noteworthy that only a few significant differences were detected in mean scores obtained at BACS between remitters and non remitters using SRWGcr; on the contrary, significant differences were detected in mean scores of almost all BACS subtests and MMSE between remitters and non-remitters using both PANSS-PNScr and PANSS-TScr, indicating a better neurocognitive functioning among patients judged as being in clinical remission according to the more selective criteria adopted in this study. This evidence seems to be of relevance, as cognitive performance is a strong predictor of functioning [
34], and the best levels of functioning were found among patients considered to be remitted according to the alternative criteria of remission.
To confirm validity of the alternative criteria, these were compared with RSWGcr, particularly to assess impact produced on functioning. Approx. 20% of subjects were found to be “functionally remitted”; when functioning was evaluated on the basis of clinical remission status, a substantial increase in rates of functional remission was observed, ranging from 32.1% in patients clinically remitted according to RSWGcr, to 42.1% among patients remitted according to PANSS-PNScr, with a peak of 53.8% among patients remitted according to PANSS-TScr. Thus, by broadening the number and type of PANSS items used to evaluate remission, the ability to identify well-functioning patients was markedly improved. Confirmation of this was obtained by evaluating the proportion of patients in clinical remission who were devoid of significant impairment at each single dimension of PSP; with regard to “socially useful activities”, this proportion increased from 32% using RSGWcr to 42% using PANSS-PNScr and 53.8% using PANSS-TScr. The rates of patients devoid of impairment in “social relationships” were 46%, 58.9% and 69.2%, respectively. Furthermore, 36% of patients viewed as remitted according to RSWGcr were in employment, as were 44% of individuals remitted according to PANSS-PNScr, and 46% of remitters according to PANNS-TScr. Even when taking into consideration, as pointed out by Lambert et al. [
10] that “functioning in schizophrenia…is probably influenced by other factors independent from remission status”, the results obtained are quite impressive, being achieved in the same set of patients, but with employment status clearly changing according to the way in which clinical remission is evaluated. Finally, as expected from previous studies [
11,
35], mean PSP scores obtained were unfailingly significantly higher among remitters than non-remitters, a finding that in our study was independent of the remission criteria, although the magnitude of differences in mean scores varied largely according to criteria adopted in evaluating clinical remission. Indeed, the effect sizes for remission evaluated by means of RSWGcr, PANNS-PNScr and PANSS-TScr were 1.06, 1.31 and 1.58 respectively for the ‘socially useful activities’ dimension, 0.73, 0.80 and 0.92 for ‘social relationships’, and 0.83, 1.01 and 1.45 for PSP total score; therefore, the use of more restrictive criteria to evaluate clinical remission is associated with a better assessment of how patients function in everyday life. Our findings also support the hypothesis that patients in remission according to more restrictive criteria display a better neurocognitive functioning, which may explain, at least in part, the improved vocational functioning of these patients [
36].
In order to better investigate whether the remission criteria proposed would reflect differences in outcome we performed a series of specific analyses. Comparing RSWGcr, PANSS-PNScr and PANSS-TScr on their ability to identify patients with better functional and cognitive outcomes, the assessment of sensitivity, specificity, predictive value, and ROC analysis showed that PANSS-PNScr is characterized by the best performances. Using regression analysis, only PANSS-TScr remission is a significant predictor of functioning, while all remission criteria used in this study predicted cognitive outcome. The general linear model analysis adopted to further investigate the effect of different clinical remission criteria demonstrates a significant effect of remission class both on functioning and cognition, with patients judged as remitted according to PANNS-TScr showing significantly higher scores than those of patients, both remitted and non remitted, according to RSWGcr, but not exceeding scores of patients non-remitted according to PANSS- PNScr. Overall, these results confirm that the best prediction of functioning and, at least in part, of cognition, is achieved using remission criteria based on the use of all items of PANSS, followed by criteria based on the use of positive and negative items of the same scale.
Prior to drawing conclusions, several limitations characterizing the present study should be considered. First, the sample size of the study was rather limited; second, it focused solely on chronic outpatients who referred to the centre over a specific period, thus excluding patients who had moved away, refused to continue treatment or no longer needed continuing care. Therefore, the findings emerging from the study should be applied only to chronic patients undergoing long-term treatment. Additionally, as sample heterogeneity is considered one of the main flaws of remission studies [
10], it should be taken into account how the present study included patients affected by both schizophrenia and schizoaffective disorders. Although remission rates observed were consistently, but not significantly, higher among patients with schizoaffective disorders, this should not detract from the relevance of our results, in view of the effective difference in remission rates between the two diagnostic groups independent of remission criteria adopted. The fact that the criterion of severity alone, without duration, was used in evaluating remission should be taken into account; indeed, this limitation prevented the drawing of any firm conclusions as to the validity of complete remission criteria. However, considering that remission studies generally demonstrate how use of the severity criterion alone is associated with higher remission rates [
10] compared to use of both the severity and duration criteria, it is to be expected that if the time component is taken into account, the rates of remission found should be even lower. There is however no reason why that the proportional lowering of rates found in this study as the severity remission criteria became more stringent should not be confirmed, even if the time component is adopted in evaluating remission.
Lastly, in evaluating predictive factors for functioning and cognitive status, other important factors (i.e. pre-morbid IQ and premorbid functioning) which may be significantly involved, were not taken into consideration. Even in the light of these limitations however, the evidence obtained would seem to be of interest.
As expected from longitudinal studies demonstrating a clear positive correlation between severity of psychopathology and levels of impairment in psychosocial functioning [
37], the present study confirmed the validity of severity remission criteria proposed by the RSWG, associated with a better symptomatologic and functional profile. The results obtained moreover lent further support to the findings of Van Os et al. [
7], who reported how the use of standardized remission criteria in schizophrenia “had the potential to improve documentation of clinical status in medical records, by providing an objective measure of illness course and treatment effect that is applicable to routine clinical care”. Moreover, our data indicate that the use of all items of Negative and Positive Scales of PANNS, and particularly of the entire PANSS scale, seem to be associated with a better identification of truly “remitted” patients, at least when taking into consideration a better personal, social and cognitive functioning as expression of remission. The use of these criteria does not imply a risk of achieving unrealistic results; indeed, the adopting of more restrictive severity criteria was not associated with a drastic reduction of remission rates. However, further studies should be undertaken to evaluate the extent to which use of the six-month duration criterion, in addition to the more restrictive severity criterion adopted in this study, may elicit a decrease in remission rates, particularly as remission studies evaluated according to the criteria of Andreasen et al. generally demonstrate that the use of both severity and duration criteria results in the finding of lower remission rates compared to the use of the severity criterion alone [
10]. Nonetheless, even taking into account the latter possibility, there is no reason to suggest that the conclusions of our study, and in particular performance of the different sets of PANSS-based remission criteria would not be confirmed even when taking into account the time component.
Acknowledgements
The authors wish to thank the other components of the Cagliari Recovery Study Group for their contribution to data collection: Davide Aru, Chiara Bandecchi, Elena Corda, Luca Deriu, Enrica Diana, Francesca Fatteri, Alice Ghiani, Alice Lai,Serena Lai, Lorena Lai, Tiziana Lepori, Raffaella Maccioni, Paola Milia, Valeria Perra, Sonia Pintore, Silvia Pirarba, Elisabetta Piras, Sara Piras, Laura Puddu, Rachele Pisu Randaccio, Lucia Sanna, Elisabetta Sarritzu, Manuela Taberlet, Cristina Tocco, Enrico Zaccheddu, and Ms Anne Farmer for language editing of the English version of the paper.
Competing interests
The authors declare that they have no competing interests.
Authors’ contribution
FP participated to the design and coordination of the study, and helped to draft the manuscript, MT participated to manage the data base of the study, to the statistical analysis and helped to draft the manuscript, MB participated to the statistical analysis and helped to draft the manuscript, RC participated to the study design and to the draft of the manuscript, BC conceived the study, participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.