Clinical validation of ratings on the six-item Positive and Negative Syndrome Scale obtained via the Simplified Negative and Positive Symptoms Interview

Introduction

In the treatment of schizophrenia, assessment and quantification of the severity of symptoms using rating scales is essential for optimal, ideally measurement-based, care (Correll et al., 2011). Among rating scales employed in schizophrenia, the 30-item Positive and Negative Syndrome Scale (PANSS-30) introduced in 1987 remains the most widely used tool to measure symptom severity and efficacy in antipsychotic drug trials (Kay et al., 1987). Due to the many items, the interview dedicated to the PANSS-30 – the comprehensive Structured Clinical Interview for PANSS (SCI-PANSS) – can be burdensome and time-consuming to complete (Kay et al., 1992b). This has precluded widespread use of the PANSS-30 in clinical practice and created a serious measurement gap between research and clinical practice (Correll et al., 2011). To bridge this gap, efforts have been directed towards developing shorter and psychometrically valid subscales of the PANSS-30. As part of these efforts, we have previously assessed the scalability (also referred to as unidimensionality) and transferability of the PANSS-30 and two proposed shorter versions – the PANSS-14 (Santor et al., 2007) and the PANSS-8 (Andreasen et al., 2005) – using item response theory analysis (Østergaard et al., 2016). Scalability requires that each item in a rating scale provides unique information regarding the latent trait (here, the severity of schizophrenia symptoms), while transferability refers to the consistency of scalability over time (Bech, 2012). Psychometrically, in order to allow the use of sum scores of individual items in a rating scale as a meaningful measure of the overall illness severity, both scalability and transferability must be present (Bech, 2012). In the study of acutely exacerbated patients with schizophrenia, we found that the PANSS-30, PANSS-14 and PANSS-8 were not scalable. However, when removing two items (G5 = mannerism and posturing and G9 = unusual thought content) from the PANSS-8, the remaining six items (PANSS-6: P1 = delusions, P2 = conceptual disorganization, P3 = hallucinatory behaviour, N1 = blunted affect, N4 = passive/apathetic social withdrawal and N6 = lack of spontaneity and flow of conversation) constituted a scale that was scalable, transferable and able to separate the efficacy of antipsychotic agents from that of placebo (Østergaard et al., 2016). Subsequently, we assessed the psychometric properties of the PANSS-6 among outpatients with chronic (Østergaard et al., 2018a) and treatment-resistant schizophrenia (Østergaard et al., 2018b) based on a reanalysis of data from different phases of the CATIE study (Lieberman, 2007; McEvoy et al., 2006). Again, we found that the PANSS-6 was scalable, whereas the PANSS-30 was not. Furthermore, the PANSS-6 was able to identify symptom remission according to the Andreasen et al. criteria, that is, a score of mild or less for each item in the PANSS-8 (Andreasen et al., 2005), with very high accuracy (Østergaard et al., 2018a, 2018b). Since the PANSS-8 is based on the DSM-V diagnostic criteria of schizophrenia, the close relationship between the PANSS-6 and the PANSS-8 suggests high content validity of the PANSS-6.

However, an important limitation of our prior studies of the PANSS-6 is that the PANSS-6 data were extracted from studies using the PANSS-30 and hence are based on a much more comprehensive and time-consuming assessment of psychopathology than is required to rate the PANSS-6. Therefore, it remains unknown whether obtaining sufficient information for a valid PANSS-6 rating is possible via a brief and focused interview, which is a prerequisite for the feasibility of using the PANSS-6 in clinical practice. Recently, we developed such a brief interview that focuses specifically on extracting information regarding the six PANSS-6 items, namely the Simplified Positive And Negative Symptoms Interview (SNAPSI; Kay et al., 1992a; Kølbæk et al., 2018; Østergaard et al., 2017). In order to subject the SNAPSI + PANSS-6 combination to clinical validation, we designed a study addressing the following three primary research questions:

Methods

Results

Sample characteristics

A total of 82 patients consented to participate in the study. Five were either discharged before completion of the baseline rating or withdrew consent. Hence, the final sample comprised 77 inpatients with schizophrenia (M_age = 35.1 ± 11.7 years, male = 57.1%, paranoid schizophrenia = 75.3%). The median duration of admission (n = 55) was 25 days (interquartile range (IQR) = 18–44). For eight participants, the baseline and end-point assessments took place during separate admissions. Most of the participants (93.5%) were treated with a second-generation antipsychotic, 20.8% with a first-generation antipsychotic and 66.2% with two or more antipsychotics. Among patients treated with antipsychotics, 45.5% received concomitant antidepressants. The mean PANSS-30 total score at baseline was 81.2 ± 15.9 (range = 48–126). The mean values for the individual PANSS-30 items can be found in Supplemental Table S2. Further participant characteristics are listed in Table 1.

OPEN IN VIEWER

Table 1.

Demographic and clinical characteristics of the 77 participants.

	Total (N = 77)	Male (N = 44)	Female (N = 33)	p-Value
Age (years), M ± SD	35.1 ± 11.7	35.6 ± 12.3	34.3 ± 10.9	0.64
Diagnosis (ICD-10)
Paranoid schizophrenia (F20.0)	58 (75.3%)	35 (79.5%)	23 (69.7%)	0.32
Other schizophrenia diagnosis (F20.x)	19 (24.7%)	9 (20.5%)	10 (30.3%)
First contact to psychiatric services (years), median (interquartile range)	11.0 (13.0)	9.0 (12.0)	13.5 (13.3)	0.29
Illness duration (years), median (interquartile range)	8.5 (11.0)	8.5 (11.3)	9.0 (10.0)	0.71
Number of outpatient contacts, median (interquartile range)	14.0 (25.0)	13.0 (20.0)	16.5 (38.5)	0.14
Number of psychiatric bed days since 1st contact to psychiatric services, median (interquartile range)	229.0 (493.0)	205.5 (407.8)	262.0 (583.0)	0.11
Drugs for psychosis
Aripiprazole	33 (42.9%)	15 (34.1%)	18 (54.6%)	0.07
Chlorprothixene	8 (10.4%)	–	–	–
Clozapine	12 (15.6%)	5 (11.4%)	7 (21.2%)	0.23
Olanzapine	30 (39.0%)	17 (38.6%)	13 (39.4%)	0.95
Paliperidone	11 (14.3%)	–	–	–
Quetiapine	28 (36.4%)	16 (36.4%)	12 (36.4%)	1.00
Risperidone	7 (9.1%)	–	–	–
Zuclopenthixol	6 (7.8%)	–	–	–
Other	7 (9.1%)	–	–	–
Drugs for depression
Selective serotonin reuptake inhibitors	20 (26.0%)	11 (25.0%)	9 (27.3%)	0.82
Serotonin noradrenaline reuptake inhibitors	7 (9.1%)	–	–	–
Tricyclic antidepressants	6 (7.8%)	–	–	–
Others	6 (7.8%)	–	–	–
Drugs for relapse prevention
Antiepileptics	10 (13.0%)	–	–	–
Lithium	<5	–	–	–
Drugs for insomnia
Benzodiazepines	29 (37.7%)	11 (25.0%)	18 (54.6%)	0.01
Zolpidem	16 (20.8%)	7 (15.9%)	9 (27.3%)	0.22
Zopiclone	<5	–	–	–
Stimulants
Central stimulants or atomoxetine	<5	–	–	–
Electroconvulsive therapy
Electroconvulsive therapy	<5	–	–	–
Coercion
Coercive psychiatric detention	12 (15.6%)	6 (13.6%)	6 (18.2%)	0.66
Coercive electroconvulsive therapy	<5	–	–	–
Coercive feeding	<5	–	–	–
Coercive medication	<5	–	–	–
Other coercion treatment	9 (11.7%)	–	–	–

The SNAPSI informant interview was conducted prior to the SNAPSI participant interview for almost all ratings (97%, n = 124). The mean duration of the SNAPSI and SCI-PANSS interviews was 18.3 ± 6.8 and 49.2 ± 12.8 (n = 127) minutes, respectively. Three severely ill participants were unable to complete the full SCI-PANSS baseline interview at once, so these were conducted and completed over two to three consecutive days. The mean duration of the SNAPSI at the end-point interview (n = 50) was significantly shorter when the same rater interviewed the participant at both baseline and end point compared to when the raters differed (14.7 ± 6.4 vs. 20.6 ± 8.9 minutes, p = 0.012).

Criterion validity

The ICC between the PANSS-6 total scores obtained by the SNAPSI and the PANSS-30-derived PANSS-6 total scores obtained using the SCI-PANSS was 0.77 (95% confidence interval (CI) 0.62–0.85). Table 2 lists the ICCs for the six individual PANSS-6 items ranging from 0.45 (N1 = blunted affect) to 0.89 (P3 = hallucinatory behaviour). The relative mean difference between the PANSS-6 total scores obtained using the SNAPSI and the PANSS-30-derived PANSS-6 total scores was –1.22 ± 2.8 points (n = 127).

OPEN IN VIEWER

Table 2.

Criterion validity of PANSS-6 ratings obtained via SNAPSI using PANSS-30-derived PANSS-6 ratings obtained via SCI-PANSS as a reference (n = 127).

	M (SD)	Range	ICC	95% CI	F	p-Value
P1 delusions	4.0 (1.4)	1–7	0.69	0.58–0.77	5.66	<0.001
P2 conceptual disorganization	2.5 (1.3)	1–6	0.70	0.60–0.78	5.75	<0.001
P3 hallucinatory behaviour	3.8 (2.0)	1–7	0.89	0.84–0.92	16.40	<0.001
N1 blunted affect	3.4 (1.2)	1–6	0.45	0.31–0.58	2.68	<0.001
N4 passive/apathetic social withdrawal	2.8 (1.2)	1–6	0.63	0.49–0.73	4.73	<0.001
N6 lack of spontaneity and flow of conversation	2.0 (1.4)	1–6	0.62	0.49–0.72	4.48	<0.001
PANSS-6 total score	18.5 (4.3)	9–30	0.77	0.64–0.85	8.98	<0.001

PANSS-6: six-item version of the Positive and Negative Syndrome Scale; SNAPSI: Simplified Negative and Positive Symptoms Interview; PANSS-30: 30-item version of the PANSS; SCI-PANSS: Structured Clinical Interview for PANSS; SD: standard deviation; ICC: intraclass correlation coefficient; CI: confidence interval.

Construct validity

Spearman’s correlation coefficient of the PANSS-6 ratings obtained via the SNAPSI and the PANSS-30 total score and the PANSS-30-derived subscale scores via the SCI-PANSS (positive subscale and negative subscale) ranged between 0.57 (PANSS-30) and 0.69 (the PANSS-6-derived three-item positive subscale via the SNAPSI vs. the PANSS-30-derived seven-item positive subscale via the SCI-PANSS; p < 0.001). Figure 1 illustrates the correlations between all comparisons of the PANSS-6 via the SNAPSI and the PANSS-30 via the SCI-PANSS total scores and subscale scores. The squared Spearman rank correlation coefficients ranged between 0.32 and 0.56, indicating that 32% and 56% of the variance in the PANSS-30 and the PANSS-8, respectively, was explained by the PANSS-6 ratings via the SNAPSI.

OPEN IN VIEWER

Figure 1.

Spearman’s correlation coefficients between the six-item version of the Positive and Negative Syndrome Scale (PANSS-6) via the Simplified Negative and Positive Symptoms Interview (SNAPSI) and the 30-item version of the PANSS (PANSS-30) via the Structured Clinical Interview for PANSS (SCI-PANSS) total scores and subscale scores (n = 127).

Responsiveness (sensitivity to change)

A total of 27 participants were not assessed at the second time point (primarily because they were discharged before the second assessment could be carried out). The median duration between the two assessments (n = 30) was 26 days (IQR = 13–160 days). For eight of these, the baseline and end-point assessments took place on different admissions. The median duration between baseline and end-point assessment if taking place within the same admission (n = 22) was 18.5 days (IQR = 12–29 days). The correlation between changes in baseline–end-point PANSS-6 and PANSS-30-derived PANSS-6 total scores for the 50 participants who were assessed at both time points is shown in Figure 2 (Spearman’s ρ = 0.70, p < 0.001). Spearman’s correlation coefficient of the baseline-end-point PANSS-6 ratings obtained using the SNAPSI and the PANSS-30 total score and subscale scores (PANSS-8, positive subscale and negative subscale) ranged between 0.49 (negative subscale) and 0.74 (PANSS-8; p < 0.001). The relative mean deviation between baseline–end-point changes of PANSS-6 total scores and PANSS-6 derived from PANSS-30 total scores (n = 50) was 0.70 ± 3.0.

OPEN IN VIEWER

Figure 2.

Spearman’s correlation coefficients for baseline–end-point changes between the PANSS-6 via the SNAPSI and the PANSS-30 via the SCI-PANSS total scores and subscale scores (n = 50).

Discussion

This is the first study to validate PANSS-6 ratings obtained via the SNAPSI against PANSS-30-derived PANSS-6 ratings obtained via the comprehensive SCI-PANSS. The main findings were: (a) the criterion validity, that is, the correlation between the PANSS-6 ratings obtained via the SNAPSI and the PANSS-30-derived PANSS-6 ratings obtained via the SCI-PANSS, was good; (b) the content validity, that is, the correlation between the PANSS-6 ratings obtained via the SNAPSI and the eight PANSS-30-derived items defining remission (PANSS-8) obtained via the SCI-PANSS, was good; (c) the responsiveness (sensitivity to change) of the PANSS-6 ratings obtained via the SNAPSI, that is, the correlation between the change in the PANSS-6 ratings via the SNAPSI and the change in the PANSS-30-derived PANSS-6 ratings, was good and; (iv) the SNAPSI could be completed in <15 minutes if the rater had prior knowledge of the patient being rated.

The good psychometric properties of the PANSS-6 have already been established (Lin et al., 2018; Østergaard et al., 2016, 2018a, 2018b). However, the primary limitation of these studies was that the PANSS-6 ratings were extracted from the PANSS-30 ratings via the SCI-PANSS. The present study was designed specifically to address this limitation, and the findings provide robust evidence that valid PANSS-6 ratings can be obtained via the brief and focused SNAPSI. The correlation between the PANSS-6 ratings obtained using the SNAPSI and the PANSS-30-derived PANSS-6 ratings is similar to the level of test–retest reliability of the PANSS-30 where correlation coefficients ranging from 0.60 to 0.80 have been found at the subscale level (Kay et al., 1987). Thus, the information necessary for the PANSS-6 ratings can be obtained by the SNAPSI, although this focused interview is considerably briefer than the SCI-PANSS. Specifically, the mean duration of the SNAPSI was 18.3 minutes (compared to 49.2 minutes for the SCI-PANSS) and significantly briefer at the end point (M = 14.7 minutes) if the same rater had interviewed the participant at baseline. For clinical practice, this finding suggests that the SNAPSI can be completed in even less time by an interviewer with an established rapport and prior knowledge of the patient’s psychopathology. Also, if administered towards the end of a session that includes other interviews or a more unstructured conversation, less time may be required to complete the SNAPSI, since this would allow for a more targeted interview.

The estimates of validity for the individual PANSS-6 items were highest for the positive items. Notably, the correlation between the three-item positive subscale of the PANSS-6 via the SNAPSI and the seven-item positive subscale of the PANSS-30 via the SCI-PANSS was lower than the correlation for the individual positive items (delusions, conceptual disorganization and hallucinatory behaviour). A plausible explanation for this finding may lie in the overlap in content of the seven items in the positive subscale of the PANSS-30. For instance, delusions may be rated on three of these seven items (i.e. P1 = delusions, P5 = grandiosity and P6 = suspiciousness/persecution). Hence, a delusion with both grandiose and paranoid content (e.g. being surveilled by foreign intelligence agencies for having developed a brilliant technology that can instantly double the speed of the Internet worldwide) will contribute to a high rating on all three of these items (P1, P5 and P6), which is not necessarily meaningful (redundancy). Such redundancy is less likely to be an issue for the PANSS-6, which only contains P1 and not P5 and P6.

Notably, the inter-rater reliability of the PANSS-6 via the SNAPSI raters was lower than that of the PANSS-30 via the SCI-PANSS raters. This is likely due to the fact that the PANSS-30 raters (the gold standard reference in the present study) received extensive training and were certified by the PANSS Institute, whereas the PANSS-6 via the SNAPSI raters only received minimal training (Kølbæk et al., 2018). While this difference may have led to conservative estimates of the validity of the PANSS-6 ratings obtained via the SNAPSI, it likely strengthens the generalizability of the results to most clinical settings, where only basic training will be feasible. The level of agreement between the six individual SNAPSI-derived and PANSS-30-derived PANSS-6 item scores is equivalent to that reported in our PANSS-6 via SNAPSI inter-rater reliability test, where the lowest item-level inter-rater reliability was observed for N1 = blunted affect (ICC = 0.52), N6 = lack of spontaneity and flow of conversation (ICC = 0.45) and P2 = conceptual disorganization (ICC = 0.49; Kølbæk et al., 2018). However, for the latter two items, the agreement between the SNAPSI-derived and the PANSS-30-derived scores was higher in the present study (ICC = 0.62 and 0.70, respectively), possibly due to more variance in symptom severity. Thus, the low inter-rater reliability of P2 and N6 in our inter-rater reliability test was likely a result of sampling bias (n = 12 participants). Notably, the variance of N1 in the present study was also low. Thus, the poorer correlation between the PANSS-6 ratings via the SNAPSI and the PANSS-30-derived PANSS-6 ratings via the SCI-PANSS may also be explained by reduced variance in symptom severity in the study population. Nonetheless, the relatively lower reliability found for N1 = blunted affect and N6 = lack of spontaneity and flow of conversation is quite consistent with the PANSS-30 literature (Igarashi et al., 1998; Lindström et al., 1994; Von Knorring and Lindström, 1992; Yehya et al., 2016). Furthermore, the reliability for these two items was also among the poorest in the PANSS-30 via SCI-PANSS inter-rater reliability test in the present study of the two PANSS-certified PANSS-30 raters (Supplemental Table S1). Taken together, there appears to be a general need for improving the evaluation of N1 = blunted affect and N6 = lack of spontaneity and flow of conversation. Potential ways to accomplish this goal include refining the anchoring points for the two negative symptom items in question, optimizing interviews/interview technique, or training raters more thoroughly with regard to the observation of the targeted psychopathology.

The correlation between the baseline–end-point changes in the PANSS-6 total scores obtained using the SNAPSI and the baseline–end-point changes in the PANSS-30-derived PANSS-6 total scores was good. This finding demonstrates that the PANSS-6 ratings obtained via the brief SNAPSI are sensitive to change in the core symptom severity during the treatment of schizophrenia. Relatedly, we also found a strong correlation between the PANSS-6 total scores obtained using the SNAPSI and the PANSS-8 total scores, suggesting that the PANSS-6 explains a very large fraction of the variance in the PANSS-8, which represents the items of the remission criteria (Andreasen et al., 2005) and the core diagnostic criteria of schizophrenia, thereby confirming the content and construct validity of the PANSS-6 via the SNAPSI approach (American Psychiatric Association, 1994; WHO, 1993). The correlation between the PANSS-6 total scores obtained using the SNAPSI and the PANSS-30 total score obtained using the SCI-PANSS was lower compared to our previous studies (Lin et al., 2018; Østergaard et al., 2016, 2018a, 2018b). This result is likely due to the fact that the PANSS-6 was extracted from the PANSS-30 ratings in the previous studies and thereby not affected by inter-rater reliability and test–retest reliability as in the present study where the ratings were conducted by independent raters on the basis of independent interviews. Furthermore, 14 of the PANSS-30 items are rated partly (12 items) or solely (2 items) based on informant data. For the PANSS-6 rating via the SNAPSI, the informant information was collected by the SNAPSI informant interview, whereas for the PANSS-30 rating, this information was collected using the IQ-PANSS and often supplemented by an unstructured interview of the informant. Unfortunately, the informant for the SNAPSI and the IQ-PANSS was not always the same staff member. These methodological differences and the fact that the informant providing the information for the PANSS-6 ratings and the PANSS-30 ratings differed in more than a third of the ratings may have contributed to discrepant ratings.

In the present study, the baseline–end-point PANSS-6 total score mean change was minimal. This apparent lack of improvement in the core symptoms of schizophrenia during inpatient treatment was somewhat surprising. Notably, the PANSS-6 via the SNAPSI exclusively measures core symptoms of schizophrenia (six out of the eight symptom items in the remission criteria defined by the Expert Schizophrenia Working Group based on the diagnostic symptom criteria for schizophrenia; Andreasen et al., 2005). However, in some research/clinical settings, a broader assessment of symptoms may be required – for instance using the SCI-PANSS and the PANSS-30. In this regard, it is important to keep in mind that factor analyses of the PANSS-30 have quite consistently failed to replicate a common underlying factor structure across samples (Van Der Gaag et al., 2006), while shorter versions of the PANSS may be more robust across samples (Lindenmayer, 2017). In other words, choosing psychometric tools for a given sample is not straightforward. We believe that the PANSS-6 via the SNAPSI represents a good ‘basic measure’ that will be informative in most clinical settings and in many research projects. Although not being core symptoms of schizophrenia, depression and anxiety symptoms are very common in schizophrenia (Emsley et al., 1999), and the presence/severity of suicidality also often determines the decision to admit or discharge patients. In the present study, the minimal change on the PANSS-6 may be explained be admission due to other causes than psychotic exacerbation. For example, a relatively large proportion of the participants were treated with antidepressants, and the mean depression and anxiety scores were high. If these symptoms are of interest as outcome measures, neither the PANSS-6 nor the PANSS-30 has proven to have adequate psychometric properties to measure these symptoms. For the measurement of depressive symptoms among individuals with psychotic disorders, scales with the ability to discriminate depressive symptoms from negative symptoms, such as the Calgary Depression Scale for Schizophrenia (Lako et al., 2012), are recommended. Likewise, for measuring anxiety, the Scale of Anxiety Evaluation in Schizophrenia (Smith et al., 2019), which is currently the most widely evaluated scale in the population, would be preferable. Ideally, in order to facilitate time-friendly clinical implementation of measurement-based care, future studies should aim at developing brief versions of scales specifically measuring symptoms of depression, anxiety and other syndromes that often accompany core schizophrenia symptoms. Since time is limited in most clinical settings, the use of self-report scales measuring symptoms and other important treatment outcomes, such as medication side effects and quality of life (Correll et al., 2011; Palmier-Claus et al., 2012; Schouby Bock et al., 2020; Suzuki, 2011; Topp et al., 2015), may further facilitate shared decision making and measurement-based care.

Several limitations should be considered when interpreting the results of this study. First and foremost, in most cases, our raters had no prior knowledge of the participants’ psychiatric history or psychopathology (aside from the SNAPSI informant interview). This is likely to have prolonged the duration of the SNAPSI compared to the time needed to conduct the SNAPSI in clinical practice. However, the lack of the PANSS-6 raters’ familiarity with the patients before the interview could also be considered as a strength of this study, as it underlines that sufficient information for the PANSS-6 rating can be obtained using the SNAPSI without prior knowledge of the patient. When the same rater conducted both the baseline and end-point PANSS-6 rating, the duration of the interview was substantially reduced. This length of the SNAPSI (approximately 15 minutes) will probably translate well to clinical practice. Second, the participants’ schizophrenia diagnoses were not confirmed via structured diagnostic interviews. However, prior studies have shown that the validity of the diagnoses of schizophrenia assigned as part of standard clinical practice at psychiatric hospitals in Denmark is high (Jakobsen et al., 2005; Uggerby et al., 2013). Third, our study was carried out only among inpatients with schizophrenia. Although we have no reason to expect that the PANSS-6 ratings obtained via the SNAPSI would be less reliable and valid in other settings (e.g. outpatient clinics), this should be subjected to future testing. Finally, depending on the research questions and areas of individual need, in both research and clinical practice, the PANSS-6 ratings might need to be accompanied by ratings on measures of other constructs that are relevant in relation to the care of individuals with schizophrenia, such as depression (Addington et al., 1996), anxiety (Llorca et al., 2014), cognition, medication side effects (Lingjaerde et al., 1987; Waddell and Taylor, 2008), level of functioning and quality of life (Topp et al., 2015).

In conclusion, we find that the PANSS-6 ratings obtained using the brief SNAPSI are valid measures of core symptoms of schizophrenia and their changes over time. We therefore believe that the combination of the SNAPSI and the PANSS-6 represents a viable, effective and clinically feasible solution for the core symptom assessment in the clinical care of patients with schizophrenia. The seven raters who conducted the SNAPSI and rated the PANSS-6 had varying levels of clinical experience and received sufficient training with minimal use of time, which suggests that our findings translate well to most clinical settings. With regard to use in clinical practice, we recommend that patients are rated with the PANSS-6 at intervals that match the clinical setting and the severity of illness (more often among acutely ill inpatients than among stable outpatients). Furthermore, whenever changes in medication occur, pre and post measurements using the PANSS-6 obtained via the SNAPSI and a valid side-effect measure (Lingjaerde et al., 1987; Schouby Bock et al., 2020) are critical to monitor treatment response. In clinical research, the PANSS-6 obtained via the SNAPSI may serve as a psychometrically sound and brief alternative to the PANSS-30. Future studies should explore the validity and utility of the combination of the SNAPSI and the PANSS-6 in other treatment settings.

Supplemental Material