The association of the persecutory ideation questionnaire with clinically-relevant and other outcomes: the moderating role of confidence

One of the most used self-administered instruments to assess persecutory delusions is the Persecutory Ideation Questionnaire (PIQ). Individual differences in PIQ scores are important because they predict the severity of symptoms associated with psychosis-related disorders. The current research demonstrates that PIQ is associated with two new outcomes: Satisfaction with life (Studies 1 and 2) and therapy length needed for hospital discharge (Study 2). Most relevant, we introduce meta-cognitive confidence in one’s scale responses as a construct capable of improving the predictive validity of the PIQ. Across two studies, participants from the general population (Study 1) and from a clinical sample (Study 2) completed the PIQ and then reported the confidence in their responses. As expected, the PIQ was associated with satisfaction with life in both cases and duration of therapy required to receive hospital discharge for the clinical sample. Most importantly, confidence further moderated the extent to which the PIQ scores were linked with both outcomes, with greater consistency between the PIQ and the dependent measures obtained for those with higher confidence. Therefore, asking a single item about the confidence associated with responses to the PIQ enhances the association of PIQ scores and relevant consequences across domains.

Previous research has focused on variables that are associated with PIQ scores, many of which are assumed to be antecedents [17][18][19] .Relatively less research exists focusing on other associations with PIQ scores that are assumed to be consequences of PIQ.The few examples available in the literature are centered on the instrument's ability to predict the severity of related disorders symptoms 16,20,21 .Although there has been some prior studies introducing positive treatments in clinical samples with paranoid delusions 22,23 , these studies did not analyze the association between the PIQ and positive mental health (e.g., satisfaction with life).Moreover, to the best of our knowledge, there is no previous research regarding the association of the PIQ with therapy length-related outcomes.
On the one hand, life satisfaction is a relevant positive outcome because it is associated with lowered risks of all-cause and natural-cause mortality 24 and other relevant outcomes such as lower suicide attempts 25 .Moreover, according to the Complete State Model of Health 26 clinical evaluations and interventions should not only be focused on detecting and reducing psychopathological symptoms, but also in increasing quality of life.Satisfaction with life has been used as one of the more important indicators to examine the subjective quality of life for people experiencing serious health concerns, and probably is the most adequate and appropriate construct to do it 27 .For example, Wu and Wu 28 found evidence that satisfaction with life is a valid measure of quality of live for patients with schizophrenia.
On the other hand, therapy length is also an important outcome to examine in this context because it is related to the cost-effectiveness associated with treatments 29 .Understanding how long therapy lasts and when a treatment has reached a good enough level is an important criterion not only for economic purposes but also for its clinical implications 30 .The dominant perspectives about how to predict therapy length are the dose-effect model 31 and the good-enough level (GEL) model 32 (see 33 for a comparison between both models).The GEL model predicts that patients remain in therapy until they, in conjunction with their therapist, determine that they have improved sufficiently-to the good-enough level.Therefore, on average the amount/length of treatment reflects treatment response and indicates how malleable patients' symptoms are, rather than being the driving force of treatment response as in the dose-effect model.Thus, the GEL model predicts that patients who respond best to treatment receive low doses of treatment, whereas patients who respond worst are the ones that need more time of treatment 30 .
As previously mentioned, paranoid delusional thinking is often linked with feelings of meta-cognitive confidence.That is, people have initial thoughts (primary cognitions) and they can subsequently think about those thoughts (secondary cognition or meta-cognition).Following a primary thought, people can also generate other thoughts that occur at a second level and play a critical role in delusional disorders 7 .Meta-cognition refers to these second order thoughts, or thoughts about one's primary thoughts.Therefore, thought confidence is a metacognition because it involves thinking about the validity of initial cognition.Previous literature about thought confidence conducted with a general population suggests that mental constructs predict judgments and behaviors better when people have higher (vs.lower) confidence in their thoughts 34 .According to Self-Validation Theory (SVT) 35 , confidence can moderate the correspondence between mental contents and judgments, including, in this case, the link between traits and behavior.
Research on SVT has revealed that confidence can apply to any mental construct.This work has shown that mental contents are especially predictive of judgment and behavior the greater the perceived validity that individuals have in their thoughts.These perceptions of validity can be assessed easily by asking people to rate the confidence or certainty they have in their responses to any inventory.Perceptions of validity have been useful in moderating the effects of individual differences in different domains (e.g., political ideology 36 , group processes 37 , health behaviors 38 ).As an illustration of this body of research guided by the SVT framework, some research 38 has shown that responses of participants within the general population to a porn scale were associated with porn consumption to a greater extent for people with high (vs.low) confidence in their responses to the scale.In another illustration 39 , research has shown that participants' trait aggressiveness in a general population sample was more predictive of actual aggressive behavior as individuals' confidence in their own responses to the aggressiveness inventory increased.The logic behind these findings is that people think about their traits along with the associated confidence when deciding how to act.In other words, mental contents are especially predictive of behavior the greater the perceived validity that individuals have in their thoughts.For example, people might think how much porn they like to consume or how aggressive they are before deciding to act and to the extent that they have confidence in that thought about how they are, they will act more in accord with it when the relevant situation comes.
To address the role of confidence in understanding PIQ outcomes, the current work had two core goals.The first is to analyze the relationship between the PIQ and satisfaction with life, and the link between the PIQ and therapy length to obtain hospital discharge.Second, an important goal is to test whether confidence in responses to the PIQ can moderate these relationships in the same way that confidence has moderated the impact of other individual differences on postulated outcomes.Study 1 examined to what extent confidence linked to the responses given to the PIQ can help to explain the conditions under which the relationship between the scale and satisfaction with life will be stronger.Study 2 tested this question in a clinical sample diagnosed with persecution delusions.Also, Study 2 included a behavioral measure related to paranoid delusions, the duration of therapy required to receive hospital discharge.

STUDY 1
The goals of Study 1 were to test the extent to which PIQ served to detect positive mental health in the general population, and to examine whether confidence in scale responses could strengthen the predictive validity of the inventory.First, participants answered the PIQ, and then reported the extent to which they were confident about their responses to the instrument (meta-cognitive confidence).These two variables and their interaction

Predictor variables
Persecutory ideation questionnaire (PIQ) We used the Spanish version of the "Persecutory Ideation Questionnaire" (PIQ) 16 previously validated by Fonseca-Pedrero and colleagues 41 .The PIQ is a 10-item instrument that includes statements about paranoid thinking such as: "People mean to do and say things to annoy me" or "Some people try to steal my ideas and take credit for them." The original PIQ and the Spanish version have been evaluated in previous studies 16,41 using clinical and general population samples, showing in both cases adequate internal consistency, unidimensional factorial validity, and criterion validity as a measure of paranoid style focused on persecutory ideation.Participants answered using a 4-point scale (0 = completely false to 3 = completely true).In our study we computed the mean of the 10 items of the PIQ, therefore the maximum possible score is 3, and the minimum score is 0. Higher scores indicate more persecutory ideation.The mean of the PIQ in Study 1 (M PIC ) was 0.99 (SD = 0.65).This value was equivalent to the mean values observed for this scale in other studies conducted with general population samples.For example, in the original study of the PIQ, the mean value of the general sample was 0.91 (range 0-4) (SD = 0.60) 16 .Subsequent studies have obtained similar mean values (e.g., Fernyhough et al. 42 : M PIC = 1.01,SD = not reported; Jones & Fernyhough 43 : M PIC = 0.87, SD = 0.68; Jones et al. 17 : e-questionnaire M PIC = 0.72, SD = 0.51, pen and pencil questionnaire M PIC = 0.82, SD = 0.63 or Van Dongen et al. 44 : M PIC = 0.33, SD = 0.33).In the present study, Cronbach's α value for the PIC scale was 0.93.In a median split, for people with low confidence scores Cronbach's α value was 0.92 and for people with high confidence scores it was 0.93.

Confidence
Participants indicated the confidence they had in their responses to the PIQ by answering the following item: "How confident are you in the responses you just gave to the scale?"(1 = "Extremely unconfident" to 9 = "Extremely confident").Thus, higher scores on this item indicate greater confidence (M TC = 6.33,SD = 1.99).This measure of confidence was identical to the one used by in prior research [36][37][38][39] .In previous SVT research conducted with a general population, confidence was measured with this question, and this item was found to moderate the predictive validity of other inventories relevant to group behavior and to moderate the association of other mental constructs unrelated to the present domain.

Satisfaction with life
The Spanish adaptation of the Satisfaction with Life Scale (SWLS) 40 was used in this study.The scale has 5-item that includes statements about life satisfaction such as: "In most ways my life is close to my ideal" or "The conditions of my life are excellent." Both the original SWLS scale and the Spanish version of this instrument have shown good psychometric properties in previous studies 40,45 in terms of reliability and unidimensional factorial validity.Participants answered using a 5-point scale (1 = strongly disagree to 5 = strongly agree) with higher values indicating more favorable evaluations of life.In the present study we computed the mean of the 5 items of the scale.Therefore, the maximum possible score is 5 and the minimum is 1, with higher indicating scores more life satisfaction.Cronbach's α value for the SWLS scale was 0.91 (M WB = 3.35, SD = 0.98).

Ethics approval
The study was approved by the ethics committee of the "Universidad de Castilla -La Mancha" (UCLM) and the Hospital General Universitario de Ciudad Real ("Comité Ético de Investigación Clínica HGUCR-UCLM", No. 01/2020/C-305).It was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

Results
The data for skewness of all the variables included in Study 1 ranged from 0.08 to − 0.73, and Kurtosis ranged from 0.26 to − 1.07.Therefore, according to recommended criteria 46 (maximums of 2 for skewness and 7 for kurtosis), the variables in our studies follow a normal distribution.First, we conducted a preliminary analysis to examine the relationships among the three variables included in the study using Pearson correlations.A significant negative correlation was observed between the PIQ and satisfaction with life, r(198) = −0.17,p = 0.01.Moreover, the correlation between confidence and satisfaction with life was positive and significant, r(198) = 0.18, p = 0.01.Finally, the correlation between the PIQ and confidence was also positive and significant, r(198) = 0.17, p = 0.02.Satisfaction with life was then subjected to a hierarchical regression analysis using the PROCESS add-on for SPSS (model 1) 47 .PROCESS is a computational procedure for SPSS and SAS that implements moderation or mediation analysis as well as their combination in an integrated conditional process model.We introduced the predictor variables (reported PIQ and confidence) at the first step, then added a computed interaction term at the second step.Since the predictors were continuous variables (i.e., PIQ and confidence) they were mean-centered prior to analyses.Finally, age and sex were included as covariates.As expected, results of the regression analysis showed a significant main effect of the PIQ on satisfaction with life, B = −0.35,t(194) = −4.09,p = 0.001, 95% CI −0.51, −0.18, indicating that as the PIQ increased, satisfaction with life decreased.The main effect of confidence on satisfaction with life was also significant, B = 0.21, t(194) = 3.17, p = 0.002, 95% CI 0.08, 0.35, revealing a positive association between confidence and satisfaction with life.

Discussion
The results of Study 1 provided evidence of the predicted link between PIQ and satisfaction with life.Scoring high in PIQ was associated with lower life satisfaction.As noted, this outcome is important because it has been associated with a wide range of psychological and health outcomes 24,25 .
More uniquely, the effect of the PIQ on satisfaction with life measure was further moderated by confidence.As predicted, scores on persecutory delusions were negatively associated with satisfaction with life to a higher extent as participants reported higher levels of confidence in their responses to the scale.This result suggests that researchers interested in analyzing the effect of paranoid persecutory delusions on wellbeing could benefit from considering confidence.This result is also convergent with previous research that proposed rigid confidence as a harmful characteristic of paranoid thought in general population 48 .www.nature.com/scientificreports/Finally, it is worth noting that although the PIQ scores informed by participants of this first study conducted in general population were relatively low in absolute terms, the critical question is to what extent PIQ scores will be significantly lower in a general population sample compared to a clinical sample with persecutory delusions.In the next study we will employ a clinical population to analyze the capability of the PIQ to discriminate between both samples.

STUDY 2
The first goal of Study 2 was to test the generalizability of the obtained results from a general population to a clinical sample.The second goal of this study was to test, for the first time, the possible association of the PIQ with an observed behavior, the length of the therapy required to be discharged.The third goal was examining the extent to which confidence could improve the link between PIQ and the new outcome.Again, we expected a negative relationship between PIQ and satisfaction with life, and that this relationship would be stronger as confidence increases.Regarding therapy length, we hypothesized that the PIQ would be positively related with the amount of time required to be discharged, and this relationship would also be moderated by confidence such that the impact of PIQ on discharge time would be greater as confidence increased.Psychotic Disorder (N = 5) were recruited in the "Hospital General Universitario de Ciudad Real" (HGUCR).All participants presented persecutory paranoid delusions.Participants' age ranged from 18 to 76 (M age = 39.95,SD = 13.45).Patients were hospitalized and were in a post-acute or stable phase of their disorders.Given the difficulty of accessing participants with persecutory delusions willing to sign the informed consent, we proposed to collect the maximum number of patients possible during a time interval of 18 months.This number of participants allowed us to detect a two-way interaction effect size of f 2 = 0.13 with 0.8 power.

Procedure
All candidates received a letter describing the research project based on developing inventories and the voluntary nature of participation.Also, they were informed that the decision to participate (or not) in the study would not influence the clinical assistance received.The study was approved by the ethics committee of the "Universidad de Castilla-La Mancha" (UCLM) and the HGUCR (Comité Ético de Investigación Clínica HGUCR-UCLM).Those who agreed to participate were informed that all collected information in the study was confidential and anonymous.As soon as patients were stabilized and were in conditions to respond to questions, they were asked for their permission to participate in the study.Participants who volunteered to participate completed the information consent, answered the PIQ, reported their confidence in their responses to this instrument, and answered the Life Satisfaction Scale, in that order, as soon as possible within the first 48 hours after admission.The metacognitive therapy began just after the completion of the tests, and therefore the treatment could not influence the previously reported responses to the PIQ or confidence questions.

Predictor variables
Persecutory ideation questionnaire (PIQ) Persecutory delusions were measured using the same scale 16 as in Study 1.In the present study, Cronbach's α value for the PIC scale was 0.94 (M PIC = 1.23,SD = 0.88).The mean score on the PIQ in Study 2 was within the same range of mean values observed in previous studies conducted with clinical samples.For example, in the original study of the PIQ the mean value for the clinical sample was 1.34 (range 0-4) (SD = 0.91) 16 .Other studies found similar results (e.g., Dudley et.al. 49 Time 1: M PIC = 1.38,SD = 1.00,Time 2: M PIC = 1.33,SD = 1.01;Van Dongen et.al. 44 : M PIC = 1.37,SD = 1.08;Van Dongen et al. 19 : M PIC = 1.48,SD = 1.18).In a median split, for people with low confidence scores Cronbach's α value was 0.95 and for people with high confidence scores was 0.93.

Satisfaction with life
We measured participants' life satisfaction using the same scale as in Study 1 (SWLS) 40,45 (M WB = 2.95, SD = 0.92).In the present study, Cronbach's α value for the SWLS scale was 0.81.

Therapy length required to be discharged
During hospitalization, all patients received psychological therapy based on the Well's metacognitive model 50 .Specifically, the adaptation of this therapy for patients with schizophrenia proposed by Morrison and colleagues 51 was used.As proposed by these authors, the duration and number of sessions varied based on the individual case formulation and as a function of improvement 51 .Therefore, longer periods of time receiving treatment before discharge indicated slower recovery and a release delay.The total length of therapy for each patient was registered in minutes.On average, patients received 561 min of metacognitive therapy before discharge (SD = 54 min).www.nature.com/scientificreports/indicating that greater confidence was associated with slower recovery and release delay (i.e., more Therapy length).Although speculative, according to the GEL Model this can be interpreted as a case in which higher reports of confidence indicates more resistance to change and a poor response to psychological treatment 30,32 .More importantly, the predicted interaction between the PIQ and confidence was significant, B = 0.24, t(54) = 2.20, p = 0.03, 95% CI 0.02, 0.46.As illustrated in Fig. 3, for participants scoring higher in confidence (+ 1SD), the PIQ was positively associated with Therapy length, B = 0.81, t(54) = 5.71, p < 0.001, 95% CI 0.53, 1.00.For those scoring lower in confidence (− 1SD), there was a weaker but significant relationship between the PIQ and Therapy length, B = 0.35, t(54) = 2.15, p = 0.04, 95% CI 0.02, 0.65.Therefore, having higher confidence in the PIQ answers was associated with PIQ better predicting slower recovery and more need for further therapy than having lower confidence.

Discussion
Patients' scores on the PIQ were overall negatively associated with their satisfaction with life.This finding conceptually replicated the results observed in the first study for the general population sample and provides convergent evidence of the predictive validity of the PIQ across different populations.Most relevant to the present concerns, the association between responses on the PIQ and life satisfaction was greater as reported confidence increased.Therefore, we successfully replicated the moderation effect from Study 1 and extended it to a clinical population.
Beyond replicating results for satisfaction with life, this second study further extended the predictive validity of PIQ to other relevant outcomes.As predicted, higher scores on the PIQ were related to greater therapy length before being discharged.These results speak to the clinical relevance of this instrument.Furthermore, consistent with SVT, confidence in the responses to the PIQ moderated this relationship.That is, the higher the confidence levels in the PIQ, the stronger the relationship between the PIQ and the duration of therapy.
Finally, participants in the clinical sample reported relatively low levels of persecutory ideation, although the mean value was within the same range observed in previous studies conducted with hospitalized patients 19,44 .One possible explanation for the relatively low values in the PIQ (in absolute terms) is that clinical participants were in a post-acute or relatively stable phase of their disorders.More critically for the present research is that the mean values of the PIQ were significantly different for the clinical and general population samples, supporting the ability of this instrument to discriminate between those two types of populations.

General discussion
Across two studies the results supported our hypothesis that the PIQ is a valid measure to predict relevant new outcomes across different types of samples.Therefore, the current research supports the predictive validity of the PIQ 16 by showing the instrument's ability to be negatively associated with a measure of positive mental health (i.e., life satisfaction) and clinically relevant behaviors (i.e., amount of therapy time required before discharge).Furthermore, confidence moderated the effects of informed persecutory paranoid delusions on satisfaction with life, both for general and clinical populations, and therapy length required to be discharged for a clinical population.For a general population sample (Study 1), we predicted and found that informed persecutory delusions were inversely related to satisfaction with life to a greater extent as confidence increased.Study 2 replicated this www.nature.com/scientificreports/result and extended it to patients diagnosed with disorders linked to persecutory paranoid delusions.Among other implications, this is important because it is the first time that the SVT framework has been applied to clinical settings and samples.Importantly, the PIQ, as a self-report instrument, has some limitations related, for example, to being susceptible to individual differences in self-awareness 52 or to so-called response factors (e.g., correction in judgments 53 ).For this reason, in Study 2 we also examined the instrument's ability to predict a relatively more objective outcome (i.e., amount of therapy time required before discharge).Previous research on the PIQ mainly focused on other factors associated with the scores, and the ability of the instrument to detect symptoms severity, but there are no prior studies analyzing the measure's capacity to predict outcomes relevant to the treatment.Of course, besides its well-established importance 33 , therapy length, like any other objective measure, has its own potential limitations of interpretation.For example, although the GEL model predicts that small doses of treatment are related to relatively fast rates of change, it could be that some patients are relieved earlier because they respond aggressively to the therapist 54 .In those possible cases, shorter times should not be interpreted as a positive outcome.Given that any measure is open to interpretation, future studies could benefit from using multiple assessments of both an objective and subjective nature to reduce the potential room for interpretation associated with each of them.
Considering that the PIQ was associated both with positive health and therapy length, our results are not only relevant to the general public but also especially relevant for clinical researchers.Moreover, in an applied context, adding a single item to the PIQ to measure confidence in responses enhances the predictive capacity of the instrument.Also, this confidence measure is easy for patients to answer 36,55 and does not significantly increase the time needed to complete the questionnaire.It thus provides a simple and efficient way to separate metacognitive confidence from first-order delusional thoughts (i.e., responses to the PIQ inventory).
In view of our results, future research could explore the extent to which higher confidence (including pathological confidence in clinical populations, e.g., I am confident that people are plotting against me so I must use my thoughts more) coming right after the PIQ is causing participants to report less satisfaction with life or to need more therapy time before discharge, or whether the questionnaire is predicting more or less accurately when meta-cognitive confidence is high (vs.low) (e.g., higher confidence increases precision in accessing mental content related to delusions).Future studies could also examine the extent to which confidence can moderate the predictive validity of other inventories typically associated with pathological doubt, such as the Obsessive-Compulsive Inventory-Revised (OCI-R) 56 or the Clark-Beck Obsessive Compulsive Inventory (CBOCI) 57 .
In addition to the limitations mentioned for the PIQ itself, there are other limitations to consider despite the overall importance of these results.First, the designs of both studies were correlational and there could therefore be alternative interpretations based on the involvement of third variables.It could be argued that instead of confidence increasing the relation between persecutory delusional thoughts and therapy length, a third (unmeasured) variable is related to both confidence and therapy length and is responsible for the observed relationships.Therefore, future research should manipulate confidence 31 in addition of using a measurement approach.Nonetheless, the practical applications of measuring confidence to enhance predictive ability remain.Second, the sample size of the clinical population (Study 2) could limit the generalizability of the results.Although a wide time frame was planned for data collection, the final sample size was relatively small.This is because participants' recruitment was difficult since patients with persecutory delusions are usually reluctant to sign an informed consent.In fact, previous studies with patients who suffer persecution delusions and schizophrenia tend to have similar or even smaller sample sizes 58 .To examine the statistical power of Study 2, we conducted a post-hoc analysis using G*Power.The results indicated that Study 2 (N = 60; Cohen's f 2 = 0.05) showed low statistical power (0.42).Thus, it is important to replicate this study.
Finally, it is important to mention that this research assumes that confidence is linked to high perceived validity.Therefore, we expect that the results obtained in these studies will be replicated when confidence is associated with high validity (e.g., being right), an association that tends to occur in paranoid delusions 3 .However, if people with paranoid delusions linked confidence to low validity (e.g., patients were aware of their mental rigidity), a different pattern of results might be obtained.Thus, the meaning of confidence is proposed to moderate the impact that confidence in one's personality has on subsequent behavior 59 .In fact, changing the meaning of confidence is a target of different metacognitive therapies for paranoid delusions (Metacognitive Training) 60 .
Keeping meaning constant, another avenue for future research has to do with specifying when confidence validates and when doubt could potentially validate.For example, previous research has documented a possible connection between low levels of confidence and extreme compensatory behaviors 61 .In this sense, if paranoid delusions had a compensatory nature, one would expect that relatively low (vs.high) levels of confidence would be linked with more predictive power of the PIQ.However, the nature of paranoia justifies our findings 48 , that is, a high (rather than low) confidence in the PIQ enhances prediction of behavior.
To conclude, our results indicate that measuring meta-cognitive confidence can increase the association between the PIQ's and satisfaction with life and therapy length both in clinical and general population contexts.

Figure 1 .
Figure 1.Study 1. Satisfaction with life as a function of the Persecutory Ideation Questionnaire (PIQ) and confidence.

Figure 3 .
Figure 3. Study 2. Therapy length required to be discharged as a function of Persecutory Ideation Questionnaire (PIQ) and confidence.