Application of PANSS-6 in the assessment of severity and improvement in patients with schizophrenia: a clinical antipsychotic trial of effectiveness

Abstract

Background

The positive and negative syndrome scale (PANSS) has long been used in the assessment of patients with schizophrenia; however, its lengthy administration process, which can even exceed an hour, has led to interest in developing a shorter version. PANSS-6 was created as a more efficient and streamlined alternative, and in this study, we aimed to evaluate its application in assessing severity and improvement in patients with schizophrenia undergoing a multicenter, open-label study of paliperidone.

Methods

A total of 271 patients aged 18–60 years with a DSM-IV diagnosis of schizophrenia participated in the study. Over a 6-week observation period, patients received paliperidone monotherapy and underwent clinical assessments using PANSS. The study conducted a comprehensive analysis of internal consistency, the effectiveness of antipsychotic treatment, response and remission assessments, and early response prediction, comparing PANSS-6 with PANSS-30.

Results

PANSS-6 showed good reliability and was highly correlated with PANSS-30. It was effective in evaluating the severity of schizophrenia and monitoring the score reduction over time. We observed no significant difference in the mean effect sizes (ES) between PANSS-6 and PANSS-30. Furthermore, using PANSS-6, early response at week 2 was found to be predictive of ultimate response and remission at week 6.

Conclusion

In this study, PANSS-6 demonstrated good internal consistency and a strong correlation with PANSS-30. These findings support its value for assessing treatment outcomes and predicting early response in patients with schizophrenia undergoing antipsychotic therapy. The use of PANSS-6 may offer a practical and streamlined method for symptom monitoring in clinical settings.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12888-025-07600-x.

Introduction

The positive and negative syndrome scale (PANSS) was officially published in 1987 and has since become the gold standard for assessing psychotic symptoms through semi-structured interviews [1]. It has been translated into over 40 different languages to date [2]. Originally comprising 30 items, PANSS-30 is widely acknowledged as a valuable instrument for evaluating the severity of psychopathological symptoms, tracking treatment effectiveness, assessing response and remission, and measuring recovery in schizophrenia and other psychotic disorders. The implementation and scoring of PANSS-30 may require up to an hour or even more, which can burden both patients and evaluators/clinicians, making it impractical for routine use. Thus, researchers and clinicians have long sought to develop a shorter version of PANSS that can be used in both research and clinical settings. The development of a shorter yet comparably robust version of PANSS would represent a significant advancement in the field of psychiatry.

Compared to the general psychopathology subscale, the positive symptoms subscale and negative symptoms subscale (7 items each) collectively score better in distinguishing patients’ differences in symptom severity [3]. Furthermore, on average, these two subscales together are more accurate and representative of patient’s condition than the total score of 30 items, thus retaining 14 items to form a shorter PANSS-14 [3]. Kahn et al. incorporated 6 out of 7 positive symptom items, 6 out of 7 negative symptom items, and 7 out of 16 general psychopathology items to create the Mini-PANSS, which comprises 19 items [4]. Yamamoto et al. extracted two items each from the three subscales of the complete PANSS [delusions (P1), suspicion (P6), emotional withdrawal (N2), passive/apathetic social withdrawal (N4), tension (G4), and unusual thought content (G9)] to create the Brief PANSS [5]. PANSS-8 defines remission through the “core symptoms” of schizophrenia, namely delusions (P1), conceptual disorganization (P2), hallucinations (P3), blunted affect (N1), passive/apathetic social withdrawal (N4), lack of spontaneity and flow of conversation (N6), mannerisms and posturing (G5), and unusual thought content (G9) [6].

However, the structural validity, reliability, and responsiveness to change of the Mini-PANSS remain unclear; furthermore, it still requires a considerable amount of time and is inconvenient to administer. Similar to the Mini-PANSS, the structural validity and reliability of the Brief PANSS in patient samples have not yet undergone psychometric testing. Based on the Rasch measurement model, although PANSS-30, PANSS-14, and PANSS-8 did not show scalability (means that each item contributes distinct information to the assessment of severity, resulting in a meaningful total score of overall severity), a six-item version of PANSS (PANSS-6) was scalable and proposed [7]. The authors validate the construct of PANSS-6 by noting that it encompasses four out of the five symptoms outlined as criteria for schizophrenia in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), with the only exception being “Grossly Disorganized or Catatonic Behavior.”

The clinical application of PANSS-6 short version was validated in patients with schizophrenia to assess illness severity, evaluate its stability, and measure its sensitivity to changes in disease severity, and early improvement indicated by it can predict response or remission [8, 9]. PANSS-6 is associated with quality of life and is also used in patients with treatment-resistant schizophrenia [10, 11]. There is no significant difference between PANSS-30 and PANSS-6 measurements; cross-sectional symptom remission assessed via PANSS-6 aligns well with the definition of cross-sectional symptom remission by the Remission in Schizophrenia Working Group [12]. Moreover, PANSS-6 can serve as a substitute for PANSS-30 by utilizing early efficacy data from 2 weeks to predict treatment outcomes at 12 weeks [13].

The integration of PANSS-6 into routine clinical practice shall alleviate the burden on doctors, researchers, and patients [14]. PANSS-6 bridges the measurement gap between research and clinical care in schizophrenia [15]. Østergaard, et al. developed a brief semi-structured interview, called Simplified Negative and Positive Symptoms Interview (SNAPSI), which takes approximately 15–20 min to administer, and they devised a video-based training program for PANSS-6 ratings, which they tested on clinical staff with varying professional backgrounds and levels of clinical experience [16–18]. Condensing the assessment into six key items not only reduces the assessment burden but also streamlines the evaluation process without compromising quality. This facilitates quantified treatment based on assessments, enhancing schizophrenia management.

In summary, although the psychometric properties and sensitivity to change of the PANSS-6 have been well-established through large, retrospective analyses of randomized controlled trials (RCTs), its utility for prospective prediction of treatment outcomes within a structured treatment pathway requires further investigation. Building upon this foundation, the present study leverages data from a prospective, open-label trial of paliperidone monotherapy in acutely exacerbated Chinese inpatients. Our primary aims were: (1) to corroborate the scale’s reliability and convergent validity with the PANSS-30 within this specific trial context, and (2) to specifically evaluate whether early improvement on the PANSS-6 at Week 2 can effectively predict ultimate response and remission at Week 6.

Methods

Study design

This multicenter, open-label study of paliperidone was conducted at 9 research centers across China (Supplementary Table 1) between December 2021 and September 2023. The main purpose of the original research is to investigate the association between candidate gene polymorphisms and the efficacy and safety of oral paliperidone formulations in Chinese patients with schizophrenia. The study is a secondary analysis of the clinical data from the original research. The study has been registered with the National Medical Research Registration and Filing Information System at http://114.255.48.20 under filing number MR-11-22-001016. The Ethics Committee of Central Institutional Review Board approved this study (approval number RO76477SCH4072). All participants were informed of the study and provided signed informed consent.

Participants

The study enrolled 277 hospitalized patients aged 18–60 years who were diagnosed with schizophrenia according to the DSM-IV criteria. The inclusion criteria were as follows: patients experiencing an acute exacerbation phase, with a PANSS-30 total score ≥ 60 and patients with at least 3/7 positive symptom items with scores ≥ 4. The exclusion criteria were as follows: pregnant or lactating women, individuals planning pregnancy, patients with contraindications to paliperidone, and those with severe and unstable physical illnesses. Prior paliperidone treatment did not affect eligibility for enrollment; however, patients who had already initiated paliperidone therapy at the time of screening were excluded.

This multicenter clinical study had an observation period of 6 weeks, with follow-up visits scheduled every 2 weeks. During the observation period, patients received paliperidone oral formulation monotherapy (dosage, 3–12 mg/day), which was administered either once daily or twice daily. The tapering and discontinuation of prior medications were required to be completed within 1 week. To achieve an effective and stable dose, dose adjustments for the study drug were to be finalized within 2 weeks. After this initial adjustment period of 2 weeks, further dose modifications for paliperidone were generally not permitted. To ensure treatment adherence for included inpatients prescribed oral paliperidone, the nursing staff is responsible for administering medications and monitoring ingestion to verify that the prescribed doses are taken.

Clinical assessments

PANSS is widely used to evaluate symptoms in patients with schizophrenia [1]. PANSS-30, which comprises 30 items, assesses positive and negative symptoms and general psychopathology. PANSS gauges the severity of symptoms like hallucinations, delusions, disorganized thinking, blunted affect, social withdrawal, and lack of motivation. Each item is rated on a seven-point scale, with higher scores indicating greater symptom severity. PANSS offers comprehensive assessment of schizophrenia-related symptomatology and assists clinicians in monitoring changes in symptom severity over time. Prior to the study’s commencement, all evaluators participated in a unified training session on the PANSS. Subsequent reliability assessments consistently demonstrated that the inter-rater correlation coefficient for the PANSS total score remained above 0.80.

Meanwhile, PANSS-6 serves as a concise rating scale for measuring schizophrenia severity [7]. The items encompass delusions (P1), conceptual disorganization (P2), hallucinations (P3), blunted affect (N1), passive/apathetic social withdrawal (N4), and lack of spontaneity and flow of conversation (N6). Each symptom item is scored from 1 (absent) to 7 (extreme severity), culminating a total score ranging from 6 to 42. Sources of information for assessment include observational findings during interviews, reports from informants (such as caregivers or staff), and/or verbal accounts from the patients themselves. The assessment reference period is the most recent week. Scores of PANSS-6 were gathered from PANSS-30.

Outcomes

Treatment efficacy

The trajectory of PANSS-6 and PANSS-30 scores and the reduction in total scores over the 6-week study period were analyzed to estimate the treatment effect. To evaluate the sensitivity to change at the endpoint, effect size (d) was calculated. Effect size, a dimensionless measure, facilitates comparisons between scales using different metrics. To standardize PANSS-30 and PANSS-6 scores, they were transformed into T-scores with a mean of 50 and standard deviation of 10. This conversion method has been previously applied, using the following formula to compute the T-score:

The “speed of change” on PANSS-6 and PANSS-30 was used as the outcome measure in a previous study [11]. In the present study, the “speed of change” was defined as the change in total score per 2 weeks.

Response

The ultimate response was defined as a reduction of ≥ 40% in PANSS-30/PANSS-6 scores by week 6. Early improvement was defined as a reduction of ≥ 20% in PANSS-30/PANSS-6 scores by week 2 [9].

Remission

Symptom remission was defined according to Andreasen’s criteria, which included achieving a maximum score of 3 (indicating only mild presence) on eight specific symptoms assessed by PANSS-30 (items P1, P2, P3, N1, N4, N6, G5, and G9) [6], except for the criterion of maintaining this score for six months (criteria A) or having total scores below sixty-one (criteria B) [7]. In addition, remission as defined by PANSS-6 required a maximum score of 3 for each item (criteria A) or a total score of less than 14 across all six items (criteria B) [7].

Statistical analysis

Summary statistics were used to describe the data, which were analyzed using R software (version 4.3.2, R Core Group, Vienna, Austria) with data visualization done using the R package “ggplot2”.

Internal consistency

During each visit, we assessed the internal consistency of PANSS-30 and PANSS-6 using Cronbach’s alpha. An acceptable range for Cronbach’s alpha is typically between 0.70 and 0.95, indicating good consistency. Notably, scales with fewer items may show lower Cronbach’s alpha values.

Correlation between PANSS-6 and PANSS-30

We used Pearson’s correlation coefficient (r) to evaluate the agreement between PANSS-6 and PANSS-30 at each assessment. A strong correlation, as defined by Cohen [19], is reflected by an r value of ≥ 0.5.

Effectiveness of antipsychotic treatment

One-way repeated-measures analysis of variance (RM-ANOVA) was performed to explore the impact of treatment duration on PANSS-6 and PANSS-30 total scores. Effect size was calculated by comparing the mean difference between baseline and post-treatment (week 6) scores for each measure, divided by the standard deviation of the difference. A substantial effect size, with a d-value of 0.80 or above, suggests clinically significant improvements at the endpoint. We have applied the Bonferroni correction to account for multiple comparisons.

Response, remission, and predictive value of early response

Our objective was to investigate whether early response to antipsychotics at week 2 predicts eventual response and remission (criteria A and B) as assessed by PANSS-6 and PANSS-30 at week 6. We calculated the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for early improvement.

The total symptom reduction measured by PANSS-6/PANSS-30 at week 2 was entered into the receiver operating characteristic (ROC) curve analysis to assess sensitivity and specificity levels and establish corresponding cutoff scores. The area under the curve (AUC) was used to quantify diagnostic accuracy in distinguishing responders from nonresponders and remitters from nonremitters. The AUC value ranges from 0 to 1. Typically, an AUC of 0.5 signifies no discriminative ability, and an AUC between 0.5 and 0.7 suggests some discriminative ability. Furthermore, an AUC between 0.7 and 0.9 indicates good discriminative ability, and an AUC between 0.9 and 1 denotes excellent discriminative ability [20].

Results

Demographic and clinical characteristics

A total of 271 patients diagnosed with schizophrenia were included in the analysis. Among them, 248 patients completed the week 2 follow-up, 216 patients completed the week 4 follow-up, and 184 patients reached the week 6 endpoint. The average age of the patients was 32.34 ± 10.46 years, with males accounting for 42.44% of the sample. Regarding marital status, 24.35% of the patients were married, and 32.70% had a college education or higher. The mean age of onset of schizophrenia was 25.68 ± 9.28 years, with 31.37% of the patients experiencing their first episode. The average duration of illness was 79.48 ± 79.43 months, and the patients had an average of 2.12 ± 1.97 hospitalizations. In addition, 22.14% of the patients had a family history of mental disorders. The raw scale scores for symptom severity for each time point are independently detailed in Table 1.

Table 1.

Results of the raw scale score, cronbach’s alpha, pearson’s correlation coefficient (r), and T-score

	Baseline	Week 2	Week 4	Week 6
PANSS-6, mean ± SD	20.68 ± 5.01	16.90 ± 4.96	13.78 ± 4.61	12.72 ± 4.25
Cronbach’s alpha	0.54	0.66	0.71	0.70
PANSS-30, mean ± SD	81.85 ± 15.62	67.05 ± 15.09	57.39 ± 14.66	53.18 ± 13.17
Cronbach’s alpha	0.78	0.84	0.87	0.86
Pearson coefficient (r)	0.67**	0.81**	0.85**	0.84**
PANSS-6 T-score, mean ± SD	50.00 ± 10.00	42.46 ± 9.90	36.23 ± 9.20	34.11 ± 8.48
PANSS-30 T-score, mean ± SD	50.00 ± 10.00	40.52 ± 9.65	34.34 ± 9.39	31.65 ± 8.43

PANSS: the positive and negative syndrome scale; SD: standard deviation; **: P < 0.005

Reliability and validity

The reliability of PANSS-6 was assessed using Cronbach’s alpha, with a minimum value of 0.54 observed at baseline, indicating acceptable internal consistency. Subsequent visits showed an increase in the Cronbach’s alpha coefficient, as shown in Table 1. The validity of the PANSS-6 scale was determined by assessing the correlation between PANSS-6 and PANSS-30 total scores. At each visit, the correlation between PANSS-6 and PANSS-30 total scores consistently exceeded 0.65 (Table 1).

Effectiveness of antipsychotic treatment

The results of RM-ANOVA showed a significant main effect of treatment time on both PANSS-6 total scores [F (3, 174) = 214.30, P < 0.001, ηp² = 0.79] and PANSS-30 total scores [F (3, 174) = 227.47, P < 0.001, ηp² = 0.80]. Post-hoc pairwise comparisons using Bonferroni correction further revealed significant differences in mean scores at weeks 2, 4, and 6 compared to baseline scores, as shown in Table 2.

Table 2.

Week 6 effect sizes and 6-week RM-ANOVA for PANSS total scores

		Mean difference	Std error	P/d
PANSS-6	Week 2	-1.14	0.26	< 0.001
	Week 4	-6.98	0.31	< 0.001
	Week 6	-8.42	0.34	< 0.001
	Effect size	0.87	1.32	1.65
PANSS-30	Week 2	-15.98	0.99	< 0.001
	Week 4	-25.06	1.10	< 0.001
	Week 6	-30.34	1.20	< 0.001
	Effect size	0.92	1.33	1.63

PANSS: the positive and negative syndrome scale

Bonferroni correction to account for multiple comparisons

Table 1 shows T-scores for PANSS-6 and PANSS-30, illustrating the trajectory of these scores over the 6-week acute treatment period with paliperidone. This progression is visually depicted in Fig. 1a, showing the changes in PANSS-6 and PANSS-30 T-scores throughout the treatment. The reduction in T-scores at weeks 2, 4, and 6 is shown in Fig. 1b, emphasizing the positive impact of the treatment on symptom severity. Moreover, the mean effect sizes of the PANSS-6 and PANSS-30 did not differ significantly (z = − 0.027, p = 0.979), as shown in Table 2.

Fig. 1.

(a) PANSS-30 T-score vs. PANSS-6 T-score at baseline and at weeks 2, 4, and 6. (b) T-score reduction during the 6 weeks of treatment

Paliperidone administration resulted in significant alleviation of symptom severity over a 6-week treatment period, as evidenced by PANSS-6 and PANSS-30 scores. The speed of change, calculated as -2.79 points per 2 weeks (95% CI: -3.04, -2.55) for PANSS-6 and − 10.32 points per 2 weeks (95% CI: -11.13, -9.43) for PANSS-30, indicated consistent alleviation of symptoms. Item-level analyses of the “speed of change” with both PANSS scores is shown in supplementary materials.

Response, remission, and early response prediction

By week 6, 92/184 patients (50.0%) met the criteria for ultimate response based on PANSS-6, whereas 71/184 patients (38.59%) met the criteria based on PANSS-30. At week 2, 78/180 patients (43.33%) met the criteria for early response according to PANSS-6, and 72/180 patients (40.00%) met the criteria based on PANSS-30.

At week 6, 113/184 patients (61.41%) met criteria A and 104/184 patients (56.52%) met criteria B for remission according to PANSS-6. Conversely, 112 /184 patients (60.87%) met criteria A and 138 /184 patients (75.00%) met criteria B for remission based on PANSS-30.

Meanwhile, we established a cutoff score for cross-sectional remission using the current data. Patients with a CGI-S score of 3 (mildly ill) showed a PANSS-6 mean score of 11.84. Consequently, we defined remission based on PANSS-6 as a total score of < 12, corresponding to a PANSS-30 total score of ~ 51 (refer to Supplementary materials). At week 6, 82/184 patients (44.57%) met criteria for remission according to PANSS-6, whereas 78/184 patients (42.39%) met criteria for remission based on PANSS-30.

The predictive value of early response at week 2 for response and remission at week 6 is shown in Table 3. The percentage reductions of PANSS-6 and PANSS-30 at week 2 were analyzed using ROC curves to determine which measure more accurately predicted response and remission at week 6 (Fig. 2). An AUC for early PANSS-6/PANSS-30 improvement at week 2 of 0.6 to 0.8 would suggest some discriminative ability. The AUC for early PANSS-6/PANSS-30 improvement at week 2 was > 0.8, indicating good discriminative ability in predicting the response at week 6.

Table 3.

Predictive accuracy of week 2 early improvement for week 6 outcomes with PANSS

Outcome	Early improvement	Sensitivity	Specificity	PPV	NPV
Response	PANSS-6	0.66	0.81	0.78	0.70
	PANSS-30	0.72	0.81	0.71	0.81
Remission a	PANSS-6	0.55	0.72	0.73	0.54
b	PANSS-30	0.45	0.74	0.85	0.30
c	PANSS-6	0.51	0.70	0.74	0.46
d	PANSS-30	0.51	0.78	0.79	0.49
e	PANSS-6	0.59	0.69	0.62	0.67
f	PANSS-30	0.64	0.78	0.69	0.74

PANSS: the positive and negative syndrome scale; PPV: positive predictive value; NPV: negative predictive value

^a PANSS-6 total scores of < 14

^b PANSS-30 total scores of < 61

^c A maximum score of 3 in each item (P1, P2, P3, N1, N4, and N6)

^d A maximum score of 3 in the 8 items (P1, P2, P3, N1, N4, N6, G5, and G9)

^e PANSS-6 total scores of < 12

^f PANSS-30 total scores of < 51

Fig. 2.

Receiver operating characteristic curves. Note: Sensitivity indicates the proportion of actual positives correctly identified, specificity indicates the proportion of actual negatives correctly identified, the best cutoff is the threshold value that maximizes the sum of sensitivity and specificity, and AUC (Area Under the Curve) represents the overall performance of the model

Discussion

This study showed that PANSS-6 has strong internal consistency, a high correlation with PANSS-30 scores, and comparable sensitivity to PANSS-30 in detecting symptom improvement. PANSS-6 indicated early response at week 2 effectively predicted ultimate response and remission by week 6. PANSS-6 also showed no significant difference in effect sizes compared to PANSS-30, supporting its validity as a streamlined alternative for clinical and research use. Meanwhile, the study’s design—a 6-week, multicenter trial with standardized paliperidone dosing—ensured methodological rigor. The inclusion of patients in acute exacerbation enhanced clinical relevance, as this population represents a priority for rapid symptom monitoring. By restricting dose adjustments to the first 2 weeks, the design minimized confounding from variable dosing while reflecting real-world titration practices. However, this design choice may underestimate real-world applicability as clinicians often titrate doses beyond 2 weeks, particularly in refractory cases [21, 22].

The findings align with prior research advocating for abbreviated PANSS versions to reduce assessment burden without sacrificing validity; to this end, PANSS-6 emerges as a simplified tool with significant potential for clinical monitoring [23]. The internal consistency Cronbach’s alpha of PANSS-6 (minimum 0.54 at baseline, increasing over time) met acceptable thresholds for reliability, which is consistent with earlier studies reporting scalability and dimensionality of PANSS-6 [7]. However, our study adds nuance by demonstrating improved reliability over time (α increasing to 0.75 by week 6), suggesting that symptom stabilization enhances measurement precision, a previously unreported phenomenon. The high correlation between PANSS-6 and PANSS-30 (r > 0.65) corroborates findings of Kølbæk et al., who reported similar concordance (r = 0.25–0.86) [8, 11, 24]. The focus of PANSS-6 on DSM-5-aligned symptoms (e.g., delusions and disorganization) [25] may explain its stronger alignment with PANSS-30 compared to other abbreviated scales. This consistent pattern of strong correlations across visits further supports the validity of PANSS-6 as a reliable measure for assessing psychiatric symptoms.

The RM-ANOVA tests showed a significant multivariate effect of treatment time. Within the context of this study, the statistically significant pattern of differences across the four time points is consistent with a true change in symptomatology over time, as captured by both the PANSS-6 and PANSS-30. The results suggest that antipsychotic treatment significantly affected clinical severity over the four time points. Specifically, there were significant improvements in PANSS-6 total scores from baseline to week 2, week 4, and week 6. The trajectories of changes in disease severity as measured by PANSS-6 and PANSS-30 are very similar and parallel during the 6 weeks of treatment. The comparable effect sizes (ES) between PANSS-6 and PANSS-30 are shown in Table 2. These findings suggest that both measures had similar responsiveness to the treatment and were comparably effective in capturing symptom changes over the course of the study.

The higher percentage of patients meeting the criteria for ultimate response based on PANSS-6 (50.00%) compared to PANSS-30 (38.59%) suggests that the former may be more sensitive to detecting improvements in symptomatology over the course of treatment. This could be attributed to the focus of PANSS-6 on core symptoms, allowing for a more rapid assessment of treatment efficacy. However, it is important to note that the choice of the assessment scale may also influence the observed response rates. PANSS-6, with its focus on a subset of symptoms, may overlook improvements or declines in other areas that are captured by PANSS-30. Therefore, whereas the PANSS-6 provides a quicker, more focused assessment of core symptoms, the PANSS-30 offers a more comprehensive evaluation of overall symptomatology, which is advantageous for capturing the broader spectrum of improvements associated with remission. Criteria A and B have been previously used to assess the predictive role of baseline self-reported cognitive function on symptomatic remission in schizophrenia [26]. Interestingly, although the percentage of patients meeting remission criteria A (PANSS-6: total score < 14; PANSS-30: total score < 61) was similar between PANSS-6 (61.41%) and PANSS-30 (60.87%), the proportion meeting criteria B, which identified symptom remission using Andreasen’s criteria, differed notably (75.00% vs. 56.52%). PANSS-30 identified a higher percentage of patients meeting criteria B for remission compared to PANSS-6, suggesting that the comprehensive PANSS-30 scale may be more sensitive in capturing a broader range of symptom improvements associated with remission. These findings underscore the importance of considering the choice of assessment scale in evaluating treatment outcomes, particularly regarding the definition of remission.

The predictive power of early PANSS-6 response predicted week 6 response (AUC = 0.81) in schizophrenia. Our results further validate the utility of PANSS-6 in this context, offering comparable predictive accuracy as PANSS-30 but with fewer items. Notably, the remission criteria of PANSS-6 (total score < 12) showed strong concordance with PANSS-30-based remission. However, the definition of response and remission in schizophrenia varies across different studies and may differ even among different subpopulations, such as treatment-resistant patients [27]. Consequently, they may not be entirely comparable, making like-for-like comparisons challenging. Although further studies employing structured interviews such as SNAPSI are needed to confirm its predictive validity, our findings position the PANSS-6, alignment with PANSS-30, as a promising tool worthy of evaluation in future research on optimized treatment algorithms. For example, it has been proposed to reallocate non-responders to alternative therapies by week 2 [28], which could ultimately help improve outcomes and reduce drug exposure.

Notably, PANSS-6 addresses the critical need for an efficient, quantifiable tool in schizophrenia care. Its brevity enables seamless integration into routine clinical workflows. For research, PANSS-6 reduces trial costs and dropout rates, particularly in low-resource settings where lengthy assessments are impractical [14]. The PANSS-6, when administered via SNAPSI, shows significant clinical utility in assessing core schizophrenia symptoms across inpatient and outpatient settings. Validated through multiple studies [15–18, 29], PANSS-6 administered via SNAPSI shows strong agreement with gold-standard PANSS-30-derived ratings [e.g., intraclass correlation coefficients (ICCs) of 0.75 for outpatients [29] and 0.77 for inpatients [18], confirming its reliability for symptom severity measurements. The structured, time-efficient SNAPSI (15–25 min) [15] streamlines data collection while maintaining psychometric rigor, addressing the “measurement gap” between research and clinical care [15]. Standardized training programs further enhance accuracy and inter-rater reliability, with post-training analyses revealing 72% of staff achieving acceptable deviations from gold-standard PANSS-6 ratings (≤ 6 total score difference), particularly among clinicians (91% accuracy) [17]. Improved inter-rater agreement (ICC = 0.74 across diverse raters [16]) and responsiveness (Spearman’s ρ = 0.55–0.70 for symptom changes [18, 29]) underscore the adaptability of PANSS-6 to real-world practice. Collectively, PANSS-6 paired with SNAPSI supports measurement-based care by enabling efficient, reliable symptom monitoring, facilitating timely clinical decisions, and bridging the translation of research insights into routine schizophrenia management.

The study by the previous Chinese validation [24] and the present study differ in several key aspects regarding the application of PANSS-6. Firstly, the study populations varied significantly. The prior research included a mixed cohort of 216 inpatients (both acute and chronic) diagnosed using ICD-10 from a single hospital. In contrast, the present study focused on a more homogeneous group of 271 acutely ill inpatients diagnosed under DSM-IV, with stricter symptom severity entry criteria. Secondly, the study design and treatment context also differed. The previous validation was a real-world observational study with flexible, clinician-determined treatments, similar to another study conducted under naturalistic conditions [30]. Our study, however, was a structured, multicenter clinical trial where all participants received standardized paliperidone monotherapy, allowing for a more controlled assessment of drug efficacy. There are differences between observational studies and RCTs [31]. Thirdly, a major distinction lies in the analytical approach. While both studies evaluated the accuracy of PANSS-6 in identifying treatment outcomes, the present study placed a greater emphasis on prediction analysis. We specifically investigated whether early improvement (at week 2) could predict ultimate response and remission (at week 6) using ROC analysis, a method not employed in the earlier work. In summary, the previous Chinese validation reinforced the practical utility of PANSS-6 in routine care, whereas our study provides stronger evidence for its predictive validity and sensitivity to change within a standardized clinical trial framework. The findings are complementary, supporting the scale’s use across different clinical and research contexts.

This study builds upon the foundational work of several large studies, which robustly established the psychometric properties and drug-placebo sensitivity of the PANSS-6 through large-scale, retrospective analyses of pooled data, placebo-controlled trials [12, 23]. While these prior analyses provided crucial evidence for the scale’s validity in the context of randomized controlled trials (RCTs), our study offers a complementary, clinically-oriented validation. The novelty of our work lies in its prospective assessment of the PANSS-6 within a dedicated, open-label clinical trial of paliperidone monotherapy conducted in a Chinese inpatient population during an acute exacerbation phase. This design provides a bridge between highly controlled RCTs and real-world clinical practice. Furthermore, whereas Hieronymus et al. demonstrated the principle of early prediction (e.g., at Week 1), our study specifically operationalizes this by prospectively validating that early improvement (≥ 20% reduction) on the PANSS-6 at Week 2 predicts ultimate response and remission at Week 6 using ROC analysis, offering a concrete, actionable timepoint for clinicians.

According to the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) guidelines, PANSS-6 shows sufficient structural validity (high-quality evidence), responsiveness (high-quality), and criterion validity (high correlation with PANSS-30) for assessing core schizophrenia symptoms [32]. However, evidence for internal consistency (low-quality) and content validity (very low-quality) remains limited, and key domains like cross-cultural validity and test–retest reliability remain unexamined. Although its short administration time (15–20 min) and feasibility render it practical for clinical trials, PANSS-6 is classified under Category B (“potential for recommendation but requires further research”) because of incomplete symptom coverage (e.g., cognitive deficits) and reliance on non-independent PANSS-30 data. The clinical case implemented PANSS-6 in measurement-based care and made symptom measurement feasible [33]. Future studies should prioritize validating subscale unidimensionality and expanding cross-cultural psychometric evaluations under COSMIN standards [32]. Furthermore, a study found no differences in symptom severity changes over time when using PANSS-6 with and without informant information [34]. PANSS-6 has been expanded to various application scenarios, such as functioning and quality of life [10, 35–39].

Limitations

Several limitations of this study should be acknowledged. First, the data were not collected specifically to validate PANSS-6, which may affect the generalizability of the results. It is acknowledged that deriving PANSS-6 scores from PANSS-30 ratings may introduce circular bias. This limitation is inherent in the method of score derivation and should be considered when interpreting the results. Second, the sample comprised only acutely ill schizophrenia patients, and thus, the findings may not apply to patients in non-acute or chronic stages. Third, the remission criteria were based on a single snapshot assessment, not accounting for symptom fluctuations over time. Fourth, the study did not assess inter-rater reliability or the performance of PANSS-6 in non-hospitalized or chronic populations. Fifth, we did not conduct a sex-specific analysis. Future studies should consider incorporating sex-stratified analyses to better understand the applicability of PANSS-6 across different sexes. In addition, although the attrition rate (32% at week 6) is consistent with dropout rates reported in similar trials [40], it risks survivorship bias by excluding dropouts from analysis. Sensitivity analyses comparing completers and dropouts would strengthen conclusions. Finally, the reliance on PANSS-30-derived PANSS-6 scores, rather than independent assessments, may have inflated correlations between the two scales. The SNAPSI interview was not utilized to support consistency in ratings, which may impact the reliability of the rating process. Future research should consider incorporating standardized interview protocols to enhance rating consistency.

Conclusion

In conclusion, this study upholds PANSS-6 as a reliable, valid, and efficient tool for schizophrenia assessment in the acute phase. Its alignment with the PANSS-30, coupled with its demonstrated predictive power for early response and concordance with remission criteria, supports its utility as a valuable instrument for both research and clinical practice. By reducing assessment burden without sacrificing validity, PANSS-6 bridges the gap between rigorous research protocols and real-world care; integration of PANSS-6 into routine clinical practice holds promise for improving patient care, facilitating timely treatment decisions based on quantifiable data, and ultimately advancing personalized treatment strategies in schizophrenia. Future studies should continue to validate its utility across diverse treatment settings and patient subgroups, ensuring its broad applicability in clinical practice.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PANSS

The positive and negative syndrome scale

ES

Effect sizes

PANSS-6

Six-item version of PANSS

DSM-5

Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition

SNAPSI

Simplified Negative and Positive Symptoms Interview

RM-ANOVA

Repeated-measures analysis of variance

PPV

Positive predictive value

NPV

Negative predictive value

ROC

Receiver operating characteristic

AUC

Area under the curve

ICCs

Intraclass correlation coefficients

COSMIN

COnsensus-based Standards for the selection of health Measurement Instruments

Author contributions

YW, ZL: conceptualization, methodology, writing-original draft preparation. TL, YS: statistical analysis. ZZ, FL, ZF, YZ: investigation, data curation. FD, CW: writing-review and editing. QB: study design, writing-review and editing, supervision, project administration, funding acquisition. All authors have read and approved the final version of the manuscript.

Funding

This study was supported by the Beijing Hospitals Authority Clinical Medicine Development of special funding support (ZLRK202335).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The Ethics Committee of Central Institutional Review Board approved this study, which followed the Declaration of Helsinki. All participants were informed of the study and provided signed informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.