Abstract
Patient involvement in routine outcome monitoring (ROM) is vital for collaborative treatment and understanding differences between patient and clinician perspectives. This study assessed the agreement between patient and clinician ratings using the Instrument for Forensic Treatment Evaluation (IFTE) and its self-report version (IFTE-S) in high-security forensic psychiatric centers in Flanders, Belgium. Results showed concordance in domains like protective behavior and resocialization skills, but patients rated their problematic behavior as less severe than clinicians did. Item-level analyses revealed a more differentiated pattern, identifying more concordance on observable items and discordance in subjective or stigmatized behaviors. Discrepancies remained stable over the first 15 months of treatment and were not influenced by primary psychiatric diagnosis. The study underscores the importance of addressing such differences in treatment to enhance shared understanding, promote feedback-informed care, and support patient engagement. Further longitudinal research is needed to confirm these findings and improve ROM practices in forensic settings.
Plain language summary
In forensic psychiatric treatment, it is important that patients are actively involved in evaluating their own treatment progress. One way to do this is through routine outcome monitoring (ROM), where patients and clinicians regularly rate how treatment is going. Comparing these perspectives can help improve communication, trust, and collaboration. However, little research has explored how much patients and clinicians agree on these evaluations in high-security forensic settings. This study examined how closely patients and clinicians agreed when using the Instrument for Forensic Treatment Evaluation (IFTE) and its self-report version (IFTE-S). The study included patients living in both high-security forensic psychiatric centers in Flanders, Belgium, and followed them during the first 15 months of treatment. Overall, patients and clinicians showed good agreement when rating protective skills (such as coping skills and responsibility for the offense) and resocialization skills (such as work, finances, and daily activities). However, patients consistently rated themselves as having fewer problematic behaviors (such as substance use or manipulative behavior) than clinicians did. When looking more closely at individual questions, agreement was higher for behaviors that are easy to observe, while disagreement was more common for behaviors that are private, sensitive, or linked to stigma or legal consequences. The level of agreement did not change over time during the first 15 months of treatment and did not differ between diagnostic groups. This suggests that early in high-security treatment, differences in perspective may remain stable and are not strongly influenced by diagnosis. These findings highlight the importance of openly discussing differences between patient and clinician views during treatment. Addressing these differences may improve understanding, strengthen collaboration, and support more meaningful and feedback-informed care in forensic psychiatry.
Keywords
Introduction
Routine Outcome Monitoring in Forensic Psychiatry
In Belgium, forensic psychiatric centers (FPCs) are specialized high-security secure hospitals that provide mental health assessment and forensic psychiatric treatment. Contrary to other countries, only individuals who have committed crimes but are deemed Not Guilty by Reason of Insanity (NGRI) due to the presence of a mental disorder can be admitted in an FPC. Two FPC were implemented in 2014 and 2017 in Flanders, Belgium following recommendations of the European Court of Human Rights (ECHR) that criticized the Belgian state for detention of internees in unsuitable facilities and solicited the government to take structural measures (ECHR 2016, No. 113/2018). Patients in FPC have high security needs and high risk profiles and are placed in FPC through a court order. In contrast to less secure hospitals, placement in an FPC is mandatory, which implies that neither the FPC nor the patient can refuse placement. Most common primary psychiatric diagnoses identified in a previous study were psychotic disorders and personality disorders, while six out of ten patients were diagnosed with a personality disorder and substance misuse as comorbid disorders (Jeandarme et al., 2022). The primary objectives of treatment within these settings are to stabilize and, where possible, ameliorate psychiatric symptoms, with the overarching aim of reducing the risk of recidivism in order to facilitate transfer to a less secure setting (Jeandarme et al., 2020). When the first FPC was implemented, a treatment duration of 3 to 4 years was anticipated. In 2020, the mean length of treatment was 2.8 years (Jeandarme et al., 2022). However, this study found that one fifth of the population was already in treatment for more than 5 years and it was anticipated that treatment duration would prolong in the coming years.
To evaluate the effectiveness of treatment and the extent to which these goals are reached, it is essential to conduct systematic assessments of both the risk of recidivism and broader treatment progress, e.g. needs and security assessments. A systematic review regarding outcome measures in forensic mental health services highlighted substantial variation in outcome instruments used in forensic mental health services and emphasized the importance of careful instrument selection (Ryland et al., 2021). Despite the large number of instruments potentially available, evidence for their use as outcome measures in forensic mental health services measured with the COnsensus-based Standards for health Measurement INstruments remained very limited. Instruments need to demonstrate good psychometric properties relevant to how they are used in practice (Ryland et al., 2021).
Over the past decades, numerous structured risk assessment instruments have been developed for this purpose, with the Historical Clinical Risk Management-20 (HCR-20; Douglas et al., 2014) being among the most widely applied internationally. Such instruments are designed to systematically evaluate specific risk factors and criminogenic needs during treatment and predict the risk of recidivism over time. However, comparatively less attention has been directed toward Routine Outcome Monitoring (ROM) instruments that assess broader treatment progress and patient-centered outcomes, such as personal, psychological, and social factors beyond risk factors (Goethals & Van Marle, 2012). At least half of the instruments identified in the review of Ryland et al. (2021) focused primarily on risk assessment and management. Shinkfield and Ogloff (2014) examined the applicability of existing instruments as potential ROM measures within forensic psychiatry, and identified six instruments as potentially valuable. Among these, the Short-Term Assessment of Risk and Treatability (START; Webster et al., 2009) is notable for its emphasis on the risk of aggression, while also offering insight into patients’ potential for positive treatment response. Additionally, the Camberwell Assessment of Need Forensic Version (CANFOR; S. Thomas et al., 2003) provides an assessment of the needs of forensic psychiatric patients. Also worth mentioning here are the DUNDRUM-3 and 4 scales that cover a broad range of functional domains including Functioning, Recovery, Risk, and Placement Pathway. These latter scales are complimentary to, but different from, violence risk assessment and have been validated in high secure settings on both mental illness and personality disorder pathways (McCullough et al., 2020). More recently, Ryland et al. (2021) identified the 10 instruments most frequently occurring within the literature that are also multidimensional and forensic specific. Half of the 10 instruments assessed were developed primarily as risk assessments. However, the use of commonly used risk assessment instruments as outcome measures in isolation may lead to an unbalanced view of progress, as they do not include important outcomes such as quality of life and social functioning. Furthermore, the underlying evidence for their use in this way is weak (Ryland et al., 2021). As noted by Schuringa (2020), relying solely on traditional risk assessment tools presents some further limitations. Many instruments include static historical factors, which are unchangeable through time or therapeutic intervention. Furthermore, the measurement scales commonly employed (typically ordinal scales ranging from three to five points) are insufficiently sensitive to detect subtle but clinically meaningful changes over short intervals. Consequently, these instruments are not optimally suited to the continuous monitoring of dynamic changes during treatment.
In response to the abovementioned limitations and the scarcity of ROM instruments specifically designed for forensic treatment progress, Schuringa et al. (2014) developed the Instrument of Forensic Treatment Evaluation (IFTE). Development began with a literature search and a Dutch translation of the Atascadero Skills Profile (ASP; Vess, 2001), which was piloted alongside the clinical items of the HKT-30. Because these items showed substantial overlap, the team proceeded using the 14 dynamic (clinical) risk items from the HKT-EX/HKT-R as the backbone for a ROM instrument and expanded them with additional ASP and newly developed items. This resulted in 22 items, including both risk factors and broader treatment outcomes, across three domains: protective behavior (e.g., problem insight, treatment cooperation), problematic behavior (e.g., impulsive behavior, antisocial behavior), and resocialization skills (e.g., balanced day time activities, work skills). To increase sensitivity to small changes in short time periods, the IFTE adopted a 17-point scale by adding intermediate response options between the five HKT-R anchor points. The instrument is typically administered on a 6-monthly basis and is designed to be scored by multiple members of the multidisciplinary treatment team, ideally comprising at least three professionals, such as psychologists, nurses, therapists, and psychiatrists, to reflect patient behavior across a variety of clinical contexts and circumstances. Prior research in a Dutch population of 232 high-security forensic patients reported moderate to good inter-rater reliability, with an ICC of >0.60 for all items, and good test-retest reliability, with α > .62–.89. The factor analysis largely confirmed the three-factor structure (Schuringa et al., 2014). A recent systematic review by Dhillon and Bennett (2024) supported the psychometric adequacy of the IFTE. Longitudinal findings further indicated that patients who initially exhibited lower scores on protective behavior and resocialization skills, alongside higher scores on problematic behavior, showed significant improvements in treatment outcomes over a 3-year follow-up period (van der Veeken et al., 2018). Moreover, prior research indicated that the IFTE outperforms unstructured clinical judgment in detecting clinical change. In a study conducted within a Dutch high-security FPC, the main clinician assessed whether patients had worsened, remained unchanged or improved, which was compared to the difference between two consecutive IFTE team scores. Additionally, change of inpatient violence was measured by the difference of the presence and/or absence of violent acts between both measurements. Findings revealed that clinical judgment tended to overestimate positive behavioral changes, whereas changes calculated with the IFTE more accurately reflected actual shifts in inpatient violence, underscoring the instrument’s greater alignment with observed clinical realities (Schuringa et al., 2021).
Patient Involvement in Routine Outcome Monitoring
An important function of ROM instruments lies not only in providing the multidisciplinary team with nuanced insights into treatment progress, leading to potential adjustments of treatment goals, but also in communicating these findings to patients (van den Brink et al., 2015). Providing feedback and engaging patients in dialog regarding their treatment progress could strengthen the relationship between patients and clinicians by promoting active patient involvement, transparency and shared decision-making (Carlier & Eeden, 2017). In addition to clinician-led feedback, significant advancement can be achieved by systematically involving patients in the evaluation of their own treatment progress with structured self-assessment instruments. This approach allows for a comprehensive appraisal of outcomes by integrating both clinical observations and patients’ perspectives within the therapeutic process. However, in the systematic review of Ryland et al. (2021) only two instruments included a patient reported scale, that is the CANFOR and the Dangerousness, Understanding, Recovery, and Urgency Manual (DUNDRUM, Kennedy et al., 2010). The literature highlights several potential benefits of incorporating patient self-assessment (Carlier & Eeden, 2017; van den Brink et al., 2015). First, it may enhance patients’ understanding of treatment goals and their practical implications. Second, it can capture patient’s subjective experience of their treatment needs, which may not always be observable to clinicians. Moreover, self-assessment has been shown to foster a sense of agency and responsibility for their treatment, thereby promoting engagement, and compliance by positioning patients as active participants in their treatment. Finally, dedicating time to discuss self-reported outcomes and valuing the patient’s perspective can further strengthen the therapeutic relationship. However, clinicians must stay cautious, as patients may underreport or overreport symptoms to influence legal outcomes (van den Brink et al., 2015)
In accordance with this patient-centered approach, a self-assessment version of the IFTE was developed to capture the patients’ perspectives on their own functioning and various aspects of their treatment over the previous six months: the Instrument for Forensic Treatment Evaluation – Self-report (IFTE-S; Schuringa et al., 2016). The IFTE-S can be characterized as a practice-driven extension of the IFTE rather than an independent assessment tool. Through the IFTE-S, patients are afforded the opportunity to self-evaluate their functioning, and these self-ratings are subsequently compared with the assessments of the multidisciplinary treatment team. This comparative process is intended to promote active patient engagement in the evaluation process and facilitate shared decision-making in clinical forensic practice. In this way, the instrument offers the potential to systematically integrate the patient’s perspective into forensic treatment evaluation, however, no empirical studies have yet investigated its application. In terms of structure, the IFTE-S mirrors the original instrument in that it comprises the same 22 items and employs the same 17-point scale. To ensure comprehensibility and accessibility for patients the wording of the items has been adapted. This approach enables direct comparison between self-report and clinician-rated assessments. However, as Ryland et al. (2021) stated, reframing equivalent items from the patient’s perspective has the disadvantage that certain outcome areas may only meaningfully be rated by patients.
Concordance Between Clinician and Patient Assessment
A central element in patient–clinician discussions of treatment progress concerns the extent of agreement and disagreement between their respective evaluations. Exploring both concordance and discordance in assessment may enhance mutual understanding and foster a more collaborative and nuanced treatment goal setting. For clinicians, such discussions can refine treatment planning, while for patients they can clarify the practical implications of treatment goals. Additionally, concordance might not only be important regarding patient–clinician communication but may also have prognostic value. For example, patient–clinician concordance on the DUNDRUM-3 and 4 regarding program completion and recovery has been found to correlate with conditional discharge (Davoren et al., 2015).
These considerations have led researchers to examine the degree of concordance between clinician and patient assessments of treatment progress. Recently, Luigi et al. (2025) conducted a meta-analysis and systematic review on shared assessment and management in forensic psychiatry, drawing on instruments such as the DUNDRUM -3 and 4, the Sex Offender Treatment Intervention and Progress Scale (SOTIPS, McGrath et al., 2013), the Historical Clinical Future-Revised (HKT-R, Spreen et al., 2014), and the Camberwell Assessment of Need–Forensic and Health of the Nation Outcome Scales–Secure (CANFOR-S, Thomas et al., 2008). Their findings suggest that clinicians tend to identify more needs and report less progress compared to patients across different instruments. However, the authors caution against firm conclusions, highlighting the substantial between-study heterogeneity and the limited number of studies available for each instrument. The reviewers further indicated that patient–clinician concordance was found to be relatively stronger for domains such as impulsivity and hostility in the meta-analysis, whereas lower agreement was reported in risk-focused measures compared to needs-focused measures in the narrative review. Luigi et al. (2025) suggested the need for further research, given that the relative stability of patients’ mental health profiles in some of the studies of the meta-analysis may have influenced observed levels of agreement. For example, a SOTIPS study on impulsivity was conducted in a prison treatment setting and no psychiatric institution (Lasher et al., 2015), while another HKT-R study on impulsivity and hostility included the lowest percentage of patients with psychotic disorders in the review (6%; Horst et al., 2023). These findings raise the possibility that underlying psychopathology may moderate patient–clinician concordance. However, to date no studies in forensic psychiatric settings have explicitly addressed this question.
Differences in concordance have empirically been observed across levels of security. Discordance between patients and clinicians was reported in both high- and low-security settings. Interestingly, higher levels of concordance on the DUNDRUM-3 and 4 scales have been found as patients approached discharge and transitioned to lower security settings (Davoren et al., 2015; Habets & Jeandarme, 2021; Lam et al., 2023; Luigi et al., 2025). These findings suggest that progression through the treatment trajectory, possibly characterized by a more optimistic outlook and increased opportunities for patient–clinician dialog may contribute to a more shared understanding of treatment progress, potentially reflecting a broader process of recovery. In light of this, Luigi et al. (2025) emphasized the importance of considering factors such as security level and treatment stage, when examining the degree of concordance between patient and clinician assessments.
Aim of the Current Study
The present study aims to extend existing research by examining patient–clinician concordance on the IFTE and its self-report version IFTE-S within high-security forensic settings. Three primary research objectives are addressed. First, the study investigates the overall degree of concordance between IFTE and IFTE-S scores. In line with previous findings, it is hypothesized that patients will rate their own functioning more favorably than clinicians (Luigi et al., 2025). Second, the study explores how concordance evolves during the first 15 months of treatment. Based on earlier research, it is expected that levels of agreement will increase over time as treatment progresses (Davoren et al., 2015; Lam et al., 2023). Third, the study examines whether the primary psychiatric diagnosis influences the degree of concordance between patient and clinician assessments of treatment progress. Given the lack of prior empirical evidence in this area, no specific hypotheses are formulated for this objective.
Methods
Participants
The IFTE was initially introduced at FPC Antwerp and FPC Ghent (Flanders, Belgium) through a pilot project conducted between April and September 2021, involving two treatment units in FPC Antwerp and three in FPC Ghent. Following this pilot phase, focus group interviews were held with multidisciplinary team members, including nurses, treatment coordinators, therapists, and psychiatrists, to identify uncertainties and address practical challenges related to the instrument and its implementation. During the first half of 2022, the IFTE was gradually implemented across both centers. All staff members received training prior to the phased roll-out, ensuring consistent application of the instrument within the treatment process.
For the present study, data were initially drawn from 487 patients with at least one IFTE assessment before October 4th, 2024. However, because the IFTE was introduced across all treatment stages, most patients did not have an assessment at the point of admission. To ensure comparability over time, only patients with IFTE assessments at minimum two different treatment phases, one of which had to be proximate to admission, were included. This selection process resulted in a final study sample of 122 patients.
The sample consisted of 116 men and six women. The mean age at admission was 40.02 years (SD = 13.6, range = 20–83). Of these patients, 63 patients resided at FPC Antwerp and 56 at FPC Ghent, while three patients transferred between the two centers during treatment. According to the Diagnostic and Statistical Manual of Mental Disorders 5 (DSM-5; American Psychiatric Association, 2013), 34.4% of patients had a primary diagnosis of a personality disorder, 32.8% were diagnosed with a psychotic disorder, and the remaining 32.8% had another primary psychiatric disorder (i.e., 7.4% with paraphilic disorder, 5.7% with behavior disorder, 5.7% with neurobiological developmental disorder, 4.9% with substance use disorder, 4.9% with intellectual disability, 1.6% with neurocognitive disorder, 1.6% with mood disorder, and 0.8% with trauma- and stress-related disorders). Furthermore, 28.7% of patients were identified as having a comorbid personality disorder, and 5.7% were diagnosed with a comorbid psychotic disorder. Among individuals whose primary diagnosis was a personality disorder, 16.7% also presented with a comorbid personality disorder. Of those primarily diagnosed with a psychotic disorder, 40.0% had a comorbid personality disorder. However, the difference in comorbidity rates between these groups did not reach statistical significance (p = .064). Additionally, the groups did not differ significantly in age at admission.
A subset of the total patient group had an additional IFTE-S (n = 75). No statistically significant differences were observed in age at admission, FPC setting, primary or comorbid psychiatric disorders between patients who completed the IFTE-S and those who did not.
Measures
The IFTE is a ROM instrument for treatment evaluation, comprising 14 clinical indicators from the HKT-R, supplemented by eight additional indicators relevant to treatment in a forensic setting. These 22 indicators are grouped into three domains: protective behavior, problematic behavior, and resocialization skills. The protective behavior domain includes eight items: Problem insight, Cooperation with treatment, Responsibility for the crime, Coping skills, Medication Use, Skills to prevent drug use, Skills to prevent physically aggressive behavior, and Skills to prevent sexual transgressive behavior. The problematic behavior domain comprises nine items: Psychotic symptoms, Recent drug use, Impulsive behavior, Antisocial behavior, Hostility, Violation of rules and conditions, Antisocial network, Sexual transgressive behavior, and Manipulative behavior. Finally, the resocialization skills domain contains five items: Common social skills, Self-care skills, Labor skills, Balanced daytime activities, and Financial skills.
All items are rated on a 17-point scale with five anchor points. In the current study, internal consistency was assessed using Cronbach’s alpha across the three measurement points. For each time point, reliability coefficients ranged from .746 to .765 for protective factors, from .652 to .703 for problematic behavior, and from .666 to .787 for resocialization factors.
The IFTE-S is the self-reported version of the IFTE and includes the same 22 items, utilizing the identical 17-point scale. However, the wording of the items has been adapted to enhance clarity and accessibility for patients. An example of this adaption is illustrated in Figure 1, which presents the wording of the “responsibility for the crime” item in both the IFTE and IFTE-S. In the current study, Cronbach’s alpha for the three time points ranged from .645 to .790 for protective factors, from .646 to .694 for problematic behavior, and from .690 to .765 for resocialization factors.
Item 10 of the IFTE and IFTE-S regarding taking responsibility for the crime.
Procedure
The IFTE was administered at admission, typically after approximately three months of treatment to allow for sufficient observation of patient behavior. Thereafter, the instrument was completed at 6-month intervals. Scoring was performed independently by multiple members of the treatment team. Completion of the IFTE is mandatory for the treatment coordinator of the ward, the patient’s primary individual counselor (nurse or therapeutic assistant), and one additional nurse or therapeutic assistant from the same ward. If the patient is engaged in therapy, the therapist who has worked most intensively with the patient during the preceding period is required to complete the instrument. The same procedure applies to occupational and creative therapists. Psychiatrists and patients are also invited to complete the IFTE and IFTE-S, respectively, though their participation is optional. Notably, 75 patients agreed to complete the questionnaire on at least one assessment occasion. Scoring is based exclusively on the behavior that raters have personally observed during the preceding period. Crucially, raters are instructed not to consider underlying psychopathology that may explain the behavior. This emphasis on directly observed behavior ensures that ratings reflect concrete, observable indicators of functioning and reduces the need for time-consuming case file reviews. The average duration required to complete the IFTE is approximately 10 min.
The collected data are subsequently used to calculate average scores and inter-rater agreement at each time point, as well as to assess both statistical and clinical changes. Using this standardized methodology, clinicians can estimate progress in clinical risk factors based on the patient’s behavior across various situations. The results can then inform the formulation of future treatment goals (Schuringa et al., 2018).
For this study, there were on average three raters per assessment of the IFTE. The first assessment occurred 3.7 months (SD = 2.8, range = 0.4–9.8) after admission, the second assessment 6.3 months (SD = 1.8, range = 3.6–14.7) later and the third assessment 6.7 months (SD = 1.9, range = 3.5–14.7) later. In total, there were 122 IFTE and 40 IFTE-S at Time 1, 122 IFTE, and 50 IFTE-S at Time 2, and 96 IFTE, and 45 IFTE-S at Time 3.
The present study received approval from the local ethics committee of FPC Antwerp and FPC Ghent on February 28, 2023. Only routinely collected, anonymized data were utilized, and all data processing procedures adhered to the requirements set forth by the General Data Protection Regulation (GDPR).
Statistical Analyses
Preliminary statistical analyses were conducted utilizing IBM SPSS Statistics, version 25. Mean scores of the clinicians IFTE scores were calculated for each time point. Multivariate outliers were detected and subsequently excluded, resulting in the removal of 18 data points. To address missing data, Full Information Maximum Likelihood (FIML) estimation was employed, thereby maximizing the use of available data (Stoel & Garre, 2011). The assumption of normality was evaluated using the Shapiro-Wilk test for the smaller IFTE-S sample and the Kolmogorov–Smirnov test for the larger IFTE sample, with a significance threshold of p < .05 indicating deviation from normal distribution.
To assess differences between clinician-rated domain and item IFTE scores and patient self-reported IFTE-S scores, paired samples t-tests were employed for variables exhibiting a normal distribution, whereas Wilcoxon signed-rank tests were utilized for variables that did not meet the assumption of normality. Only patients with both an IFTE and IFTE-S were included in these analyses. Subsequently, linear mixed-effects models incorporating the interaction between rater (patient vs. clinician) and time (assessment at Time 1, Time 2, or Time 3) were employed to examine whether differences between patient and clinician domain ratings varied across assessment points. Additionally, linear mixed-effects models with an interaction between rater (i.e., patient and clinician) and primary diagnosis (i.e., personality disorder, psychotic disorder, and other disorders) were utilized to determine whether the differences in domain ratings depended on the primary diagnosis of the patient. Prior to the analysis, the statistical assumptions of linearity, normality of residuals, and homoscedasticity were evaluated, and confirmed to be unviolated. To control for multiple comparisons, a Bonferroni correction was applied at each model (p ≤ .002).
Results
Concordance on the IFTE and IFTE-S
Table 1 presents the differences between clinicians and patient ratings on the IFTE (-S) across all time points. Convergence between rater scores was observed for the protective behavior and resocialization skills domains at all time points. Similar concordance was found for the following individual items: skills to prevent drug use, physically aggressive behavior and sexual transgressive behavior, labor skills, balanced daytime activities, and financial skills. Additionally, on two of the three measurement occasions, no significant differences were found between clinician and patient ratings for medication use, responsibility for the offense, and coping skills. Although the problematic behavior domain did not show full concordance across raters at all time points, several specific problematic behavior items did demonstrate concordance at two or three time points, including violations of rules, and conditions, antisocial behavior, psychotic symptoms, sexual transgressive behavior, impulsive behavior, and hostility.
Differences Between IFTE and IFTE-S Scores Based on Paired Simple t-test or Wilcoxon Signed Rank Test.
Significant p-value based on Bonferroni correction: p < .002.
Paired samples t-test.
Wilcoxon signed-rank test.
Lack of concordance for problematic behavior items was primarily reflected in significant differences for manipulative behavior, antisocial network, and recent drug use. Additionally, some significant discrepancies were also observed in the protective behavior and resocialization skills domains, particularly for problem insight, cooperation with treatment, common social skills, and self-care skills. Across all items showing significant discordance between clinician and patient ratings, patients consistently perceived themselves as possessing stronger skills and exhibiting fewer problematic behaviors than reported by clinicians.
Change in Concordance on the IFTE and IFTE-S Over Time
Linear mixed models were conducted with the three domains of the IFTE(-S) as dependent variables. Parameter estimates are shown in Table 2 and graphically depicted in Figure 2. The main effect of time on protective behavior was not significant. A significant main effect of rater was observed, indicating that clinicians reported lower levels of protective behavior compared to patients. However, no significant time × rater interaction was detected. Regarding the problematic behavior domain, the main effect of time was significant, indicating a reduction in problematic behavior over time. A significant main effect of rater was also found, with clinicians reporting higher levels of problematic behavior than patients. However, the time × rater interaction was not significant. Lastly, the main effect of time on resocialization skills was not significant. No significant main effect of rater emerged, nor was the time × rater interaction significant.
Parameter Estimates of Linear Mixed Models regarding Rater and Time, With the Three Domains of the IFTE(-S) as Dependent Variables.
Change in concordance on IFTE and IFTE-S scales throughout the first 15 months of treatment.
Differences in Concordance on the IFTE and IFTE-S Depending on Primary Diagnosis
Nine linear mixed models were conducted with protective behavior, problematic behavior and resocialization skills as dependent variables at Time 1, Time 2 and Time 3. The main effect of primary diagnosis in each of the nine models was not significant. However, a significant main effect of rater was observed in the models of protective behavior and problematic behavior at each time point, indicating that clinicians reported lower levels of protective behavior and higher levels of problematic behavior compared to patients. However, no significant primary diagnosis × rater interaction was detected, indicating that the discrepancy between patient and clinician ratings did not vary by diagnostic group at each time point. No follow-up pairwise comparisons were therefore conducted. Parameter estimates of the nine models are shown in Table 3.
Parameter Estimates of Linear Mixed Models Regarding Rater and Primary Diagnosis, With the Three Domains of the IFTE(-S) as Dependent Variables at Each Time Point.
Discussion
The present study examined patient–clinician concordance on the IFTE and its self-report counterpart, the IFTE-S, during the first 15 months of treatment at two high-security forensic settings. Consistent with previous findings suggesting that patients tend to present themselves in a more favorable light than clinicians (Luigi et al., 2025), results revealed significant discrepancies within the problematic behavior scale. Specifically, patients consistently reported lower levels of problematic behavior than clinicians across all time points. In contrast, no significant differences emerged between patient and clinician ratings in the protective behavior and resocialization skills scales. This pattern may be explained by the different nature of the constructs assessed. Skills and competencies are generally more concrete, externally observable, and verifiable by both patients and clinicians. As Lasher et al. (2015) suggested, clients may find it easier to evaluate tangible and observable behaviors. Conversely, problematic behaviors often involve more sensitive or concealed actions that may be more susceptible to social desirability and self-report biases, thereby increasing the likelihood of patient–clinician discordance in this domain. However, risk-sensitive professional bias may also play a role (Markham, 2025), as clinicians carry legal and safety responsibilities, which may result in a blame culture and predispose them to adopt a defensive practice and lower thresholds for perceiving behavior as problematic.
When examining individual items, the observed concordance for several protective behaviors and resocialization skills may reflect their visibility in both therapeutic and daily contexts. For instance, progress in skills to prevent drug use (e.g., through drug screenings), physically or sexually aggressive behavior (e.g., incident reports), and practical domains such as financial management, employment, or engagement in structured daytime activities, is typically observable, and documented within the FPC. Somewhat unexpectedly, concordance was also found for medication use and coping skills. However, adherence to prescribed medication and the implementation of coping strategies are routinely monitored, discussed, and reinforced during treatment sessions, which may reduce discrepancies in perception between patients and clinicians. Lasher et al. (2015) reported comparable concordance regarding emotion management in a prison setting and attributed this to the centrality of emotional regulation within therapy, which may foster at least a basic internalization of these therapeutic discussions and promote greater emotional awareness over time. This interpretation aligns with the current findings: while clinicians initially rated patients as having significantly poorer coping skills at admission, patient–clinician agreement emerged at the 6- and 12-month assessments, suggesting increasing convergence in understanding and evaluation as treatment progressed. A comparable pattern was observed for responsibility for the crime, for which a similar explanation could be valid. Responsibility-taking is a core theme in forensic treatment and is addressed explicitly and repeatedly in offense-focused interventions, which could foster a shared understanding between patients and clinicians.
In contrast, discordance emerged for certain protective behaviors, notably problem insight, and cooperation with treatment. These constructs reflect more internal and subjective processes that depend on patients’ self-awareness and the quality of the therapeutic relationship, as noted by Luigi et al. (2025). Differences in scoring may also arise from different evaluation criteria: clinicians may rate patients lower due to limited engagement or poor participation in group therapy sessions, whereas patients may base their self-evaluations on concrete or isolated indicators of cooperation, such as timely completion of assignments (Lasher et al., 2015). Within the resocialization domain, patient–clinician discordance was observed for social skills and self-care skills, consistent with findings reported by ter Horst et al. (2023). These constructs might be shaped by social or cultural norms, leading to different interpretations between patients and clinicians. Patients may perceive themselves as more competent in these areas, whereas clinicians may identify deficits based on broader behavioral observations.
Within the problematic behavior domain, discordance was particularly pronounced for items such as antisocial network, manipulative behavior, and recent drug use. These behaviors may be underreported by patients due to their stigmatizing or potentially legal implications, or because they occur outside direct clinical observation. Both Lam et al. (2023) and Lasher et al. (2015) reported low patient–clinician agreement regarding the patient’s network, with the latter authors suggesting that clinicians may assign less favorable scores if patients do not explicitly demonstrate efforts to engage with prosocial contacts within the forensic setting.
In contrast, concordance was observed for problematic behavior items such as antisocial behavior, sexual transgressive behavior, violation of rules, and conditions, psychotic symptoms, impulsivity, and hostility. This pattern likely reflects the more overt and clinically salient nature of these behaviors, and thus more systematically documented and discussed during treatment. Regarding impulsivity and hostility, Luigi et al. (2025) proposed that concordance observed in Lasher et al. (2015) and ter Horst et al. (2023) was facilitated by the relative stability of their study populations. However, the present sample of high-security forensic patients in their first 15 months of treatment cannot be considered a stable population, suggesting that alternative factors may account for the observed concordance. Notably, significant patient–clinician discordance did emerge at the 15-month assessment for both problematic behaviors, highlighting the dynamic nature of these behaviors over the course of treatment.
When examining the evolution of concordance over the first 15 months of treatment, the results did not support expectations derived from previous research (Davoren et al., 2015; Lam et al., 2023). Specifically, differences between patient and clinician ratings did not change significantly over time for any of the IFTE(-S) scales. One plausible explanation for this discrepancy lies in the characteristics of the populations studied. Prior studies were conducted in lower-security settings and often closer to patients’ discharge, periods typically associated with greater treatment stabilization, increased therapeutic engagement, and improved self-awareness. By contrast, the present study focused exclusively on the first year of high-security treatment, which represents the initial phase of a substantially longer and more intensive therapeutic trajectory. Accordingly, the findings suggest that the early stage of high-security treatment may be too premature to expect meaningful shifts in patient–clinician concordance regarding broader treatment progress.
Regarding psychopathology, no significant variation in patient–clinician concordance across diagnostic groups was observed for any IFTE(-S) scale at any measurement point. These findings suggest that psychopathology may not be a primary determinant of concordance in the early stages of treatment. One possible explanation is that patients across diagnostic profiles may share similar patterns of self-report bias, such as underreporting of risk-related behaviors or presenting themselves in a socially desirable manner, which may diminish variability between groups. However, several contextual factors warrant consideration when interpreting these findings. The high-security forensic setting is characterized by strict routines, intensive monitoring, and structured risk management, all of which are implemented to ensure safety and reduce variability in patient behavior within these secure contexts (Kennedy, 2022). As such, these institutional features may reduce observable variability that might otherwise differentiate diagnostic groups on routine outcome measures and patient–clinician concordance. Moreover, the first 15 months of treatment represent a formative phase with ongoing diagnostic refinement, limited therapeutic alliance, and relatively generic interventions while clinicians are still developing a comprehensive understanding of the patient. Patient–clinician concordance across diagnostic groups may become more noticeable as treatment progresses and interventions become increasingly individualized. Finally, the current analysis relied solely on primary diagnoses and did not account for comorbidity, which is highly prevalent in forensic psychiatric populations and may conceal meaningful effects of diagnostic differences on patient-clinician concordance. As noted previously, in the current sample, 40.0% of patients with a primary diagnosis of a psychotic disorder also met criteria for a comorbid personality disorder. Additionally, among individuals with a primary diagnosis of a personality disorder, 16.7% were diagnosed with a comorbid personality disorder. Such overlap may have influenced the observed results. Future research should therefore replicate these findings in less restrictive settings, later stages of treatment, and with diagnostic models that incorporate comorbid conditions to more fully determine whether psychopathology influences patient–clinician concordance.
Clinical Implications
The observed discrepancies between patient and clinician evaluations, particularly in the domain of problematic behavior, highlight the importance of explicitly addressing different perceptions in treatment. Within the FPCs, IFTE scores are routinely discussed with patients during individual treatment plan meetings, where IFTE factors are explicitly linked to patients’ treatment goals and differences between clinician-rated IFTE scores and patient-rated IFTE-S scores are considered. However, the present findings suggest that these discrepancies may require more systematic and targeted clinical attention. Open discussion of these inconsistencies may foster therapeutic alliance, enhance self-insight, and reduce socially desirable responding, especially in domains linked to stigma and institutional or legal consequences. In these cases, incorporating collateral sources of information (e.g., behavioral observations, incident reports, urine screenings, or staff feedback) may support more balanced assessment and help patients understand the rationale behind clinician ratings. Additionally, discrepancies in internal and subjective constructs (e.g., problem insight, cooperation with treatment) suggest that interventions may need to more explicitly target self-reflective abilities. Interventions such as mentalization-based treatment (Bateman & Fonagy, 2019), schema-focused approaches (Young et al., 2006), or guided self-reflective exercises (e.g., journaling) could support patients in expressing internal states and integrating feedback into their self-evaluation.
Conversely, patient–clinician differences may also provide clinicians with valuable insights into the patient’s subjective experience, contextual interpretations, or internal processes. Engaging patients in collaborative discussions about these differences may therefore not only enhance mutual understanding but also prompt clinicians to critically reflect on their own perspective. Such reflections could counteract tendencies toward overly cautious or defensive clinical interpretations, ultimately contributing to more nuanced and person-centered evaluations.
The findings also underscore the value of feedback-informed treatment and transparent psychoeducation regarding assessment criteria. Rather than assuming shared evaluation frameworks, clinicians should clarify how ratings are determined, which behaviors are considered relevant, and why certain scores may differ from patient expectations. Importantly, this process should be dialogical rather than corrective. Clinicians’ evaluations are not inherently more accurate or valid, as patient perspectives may be shaped by subjective experiences, cultural norms, or personal effort that is not externally visible. Systematically exploring how patients understand their own progress may prevent misinterpretation and ensure that treatment planning is collaborative rather than paternalistic. To facilitate this, clinicians may benefit from specialized training that increases their awareness of the institutional power structures that can influence their judgments, as well as the contextual and personal factors that shape patients’ self-assessments. This reflective practice can ultimately enhance the therapeutic alliance and ensure that treatment remains patient-centered and responsive to individual needs.
Overall, the early treatment phase appears to be characterized by stable discordance of perceptions in problematic behaviors and stable concordance in protective behaviors and resocialization skills. However, item-level differences are evident across all three domains, suggesting that expectations of rapid alignment between patients and clinicians may not be realistic in high-security settings. Treatment programs may find it helpful to view concordance as a long-term therapeutic objective rather than a quick indicator of progress, and to adjust milestones accordingly. Instead of treating concordance in treatment progress as a goal in itself, it may be more useful to approach it as a means of fostering mutual understanding, supporting a feedback-informed treatment process where collaborative dialog enhances both parties’ perspectives.
Finally, given the lack of differences across diagnostic groups, treatment planning should not assume that disorder type predicts patient–clinician concordance. Instead, attention may be better directed toward contextual and relational factors such as the treatment stage, institutional constraints, the quality of therapeutic alliance, and the visibility of specific behaviors.
Strengths, Limitations and Future Research
The current study represents the first evaluation of patient–clinician concordance using the IFTE and IFTE-S instruments, thereby addressing a notable gap in the literature regarding the application of ROM-tools to assess broader treatment progress, including patient-centered outcomes. Rather than limiting the analyses to the three overarching domains, this investigation also examined individual item concordance. While domain-level findings offer valuable insights into overall therapeutic advancement, the item-level analyses yielded important nuances, enabling treatment providers to identify specific areas warranting greater emphasis within feedback-informed care.
Despite these strengths, several limitations warrant consideration. First, the number of assessment time points was limited to the first 15 months of treatment, and the sample size remained modest. Future research would benefit from extending the observation period to encompass the entire duration of high-security treatment and, ideally, the full continuum of care, including transitions to lower-security settings. Such longitudinal approaches would enable a more comprehensive understanding of concordance dynamics between patients and clinicians throughout the therapeutic trajectory and could examine whether concordance evolves during stabilization, intervention phases, and reintegration.
Furthermore, patient participation in the completion of the IFTE-S was voluntary, leading to missing data, as not all patients completed the instrument or not at every assessment interval. Nonetheless, the data obtained remain highly valuable for analysis and interpretation, given the general paucity of patient data in forensic psychiatric research and the application of an analytical approach that accommodates missing data.
Moreover, the study examined diagnostic groups based solely on primary psychiatric diagnoses. Given the high prevalence of comorbidity in forensic populations, future research should account for comorbid conditions and examine additional patient characteristics that may influence concordance, thereby providing a more nuanced account of the factors shaping treatment perceptions. Relatedly, future analyses may benefit from incorporating predictors, such as treatment motivation, therapeutic alliance, or cultural background, to more comprehensively identify determinants of concordance.
A further limitation concerns the measurement properties of the IFTE-S. Although its clinical utility was demonstrated, the instrument requires additional psychometric evaluation to strengthen its utility for routine outcome monitoring in forensic psychiatric settings. Future research should address this need by employing larger sample sizes to ensure robust and generalizable findings. Finally, the study relied primarily on self-report and clinician ratings, both of which are susceptible to bias. Future research could integrate objective behavioral indicators (e.g., incident records, biometric markers) to triangulate findings. Incorporating mixed-method approaches may further clarify the subjective meaning behind discrepancies and enhance the clinical application of concordance measures.
Conclusion
This study provides the first evaluation of patient–clinician concordance on the IFTE and IFTE-S instruments within high-security forensic psychiatric treatment. Overall, concordance was observed for protective behavior and resocialization skills, whereas the problematic behavior domain exhibited significant discrepancies, possibly reflecting the greater sensitivity and stigma associated with these behaviors. Item-level analyses further highlighted specific areas of concordance, such as impulsivity and hostility, and discordance, such as problem insight, and cooperation with treatment, offering a more nuanced understanding of patient–clinician perceptions. Concordance did not differ significantly across diagnostic groups, suggesting that contextual factors, treatment phase, and behavioral observability may have a greater influence on concordance than primary psychiatric diagnosis. Similarly, no significant changes in concordance were observed over the first 15 months of treatment, indicating that early treatment may be insufficient to foster agreement in perceptions of treatment progress. These findings underscore the importance of explicitly addressing patient–clinician agreement and discrepancies, fostering collaborative evaluation within treatment and feedback-informed care.
Footnotes
Ethical Considerations
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of FPC nv. on February 28th, 2023.
Consent to Participate
The Ethics Committee waived the need for patient consent for the collection, analysis and publication of the retrospectively obtained and anonymised data for this non-interventional study.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: No author had financial benefit or interest that has arisen from the direct applications of the research. However, Sophie Verschueren, Ruben van den Ameele and Inge Jeandarme are employed by FPC Antwerpen.
Data Availability Statement
The datasets generated during and/or analyzed during the current study are not publicly available due to the sensitive nature of the data but are available from the corresponding author on request for appropriate purposes.