The CANFOR Scale for the Assessment of Needs of Forensic Psychiatric Patients: Preliminary Report on the Polish Version of the Scale and Its Reliability

Abstract

Background:

The constantly growing number of patients in psychiatric wards (including forensic wards) is a serious social, medical, and economic problem. Therefore, it is necessary to study the needs of these patients. The aim of the study was to adapt and test the reliability of the Polish version of the scale for the assessment of needs of forensic psychiatric patients—Camberwell Assessment of Need—Forensic Version (CANFOR).

Methods:

The scale was translated into Polish and then back-translated. The study was conducted among 60 patients of basic and enhanced security level forensic psychiatry wards at the Provincial Specialist Mental Health Hospital in Poland.

Results:

When analyzing the reproducibility of the responses in terms of assessing the reliability of the CANFOR questionnaire, perfect agreement of 100% was obtained in 22 areas of need. Similarly, in the repeatability study of the CANFOR questionnaire after 2 weeks, perfect agreement—100%—was obtained in 24 areas of need. In the correlation analysis of the CANFOR with another measurement tool—LSP, significant values at the P < .05 level were obtained when measuring several variables.

Conclusions:

The study resulted in high reliability of the Polish version of the CANFOR scale. It is advisable to continue research in the area of psychometric properties of the scale, including the validity of the tool.

Main Points

• The Polish version of the CANFOR questionnaire meets the credibility criteria.

• The Polish version of the CANFOR questionnaire has satisfactory initial psychometric properties.

Introduction

Most current mental health approaches are based on identifying the needs of service recipients, that is, people who benefit from services. The World Health Organization, in a 2007 report,¹ clearly emphasized the role of national institutes of public health in assessing the needs of patient populations. The concept of needs is very popular in the planning and evaluation of medical care in Western Europe.²

In connection with the reform of psychiatric care and the postulated optimization of inpatient treatment, it is necessary to examine the needs of patients treated in hospital wards (including forensic wards), as their ever-increasing number constitutes a serious social, medical, and economic problem. If people with mental illness are treated properly, their stay in inpatient facilities is shortened, thereby significantly reducing treatment costs. In addition, if we try to take proper care of these people and identify their needs, we give them the opportunity to recover or improve their functioning in their own environment. The concept of universal help and homogeneous clinical needs is now being replaced by goal-oriented, individually tailored treatment standards.³ It is assumed that if the goal of assistance is optimal health, the determinant of effectiveness should be the identified individual needs of the patient and medical services adapted accordingly. The needs assessment is intended to improve the quality of mental health services provided.^2,4

Patients in forensic psychiatry wards, as recipients of mental health services, are in a particularly vulnerable position. On the one hand, these are persons to whom certain provisions concerning the group of recipients of health services apply; on the other hand, they are perpetrators who have committed criminal acts, but the proceedings have been discontinued due to insanity. For this reason, they do not serve their sentences in prisons but are sent for compulsory treatment in forensic psychiatric wards.

In Poland, the execution of protective measures is regulated by the Executive Penal Code. Article 200 of the EPC defines the concept of a psychiatric facility and includes a division into facilities with basic, enhanced, and maximum security conditions. The minister responsible for health issues appoints a psychiatric committee for protective measures. The court, after consulting the psychiatric commission for protective measures, determines the type of facility for the perpetrator by issuing an appropriate ruling. Article 201 § 1a of the EPC defines in detail the tasks of the psychiatric commission for protective measures, which include:

1. Issuing opinions for competent courts or other authorized institutions on the admission, discharge, or transfer of perpetrators against whom a detention order has been issued in psychiatric institutions.

2. Analysis of available documentation, including medical documentation.

3. Analysis of information on the number of available places in psychiatric institutions intended for the execution of protective measures.

4. Visit and assessment of psychiatric facilities with basic, enhanced, and maximum security conditions, where the protective measure is performed.

Release from a closed healthcare facility is made by court order—the head of the psychiatric facility in which the security measure is performed, pursuant to Article 203 § 1 of the EPC, is obliged, at least every 6 months, to send an opinion on the health condition of the perpetrator placed in this institution and on the progress in treatment or therapy. Such an opinion should be sent without delay if the perpetrator’s continued stay in the institution is not necessary.

According to evidence-based medicine, identification of needs should be carried out using standardized psychometric tools.^5,6,7 Two tools are most commonly used in clinical-scientific research:⁸ Medical Research Council’s Needs for Care Assessment Schedule⁹ and Camberwell Assessment of Need (CAN).⁴

The CAN is a tool incorporating responses from the patient and the professional that was developed by the Social Psychiatry Unit of the Institute of Psychiatry at the Royal College of London to assess the needs of people with psychotic disorders. The questionnaire takes about 20-30 minutes to complete, does not require specialist training, focuses on 22 problem areas, and the response scale includes: no need, met need, unmet need, understood as a serious problem existing independently of the adequate help provided. There are now many different versions of CAN, used to assess the needs of different populations, including forensic psychiatric patients—Camberwell Assessment of Need—Forensic Version (CANFOR).^10,11 The tool is available also in clinical (CAN-C), research (CAN-R), or shortened versions (CANSAS; Short Appraisal Schedule). It could be used also for the assessment of the Needs of Older People (CANE; Camberwell Assessment of Need for the Elderly),¹² and for adults with developmental and intellectual disability (CANDID).¹³ Research on the needs of psychiatric patients, carried out with the use of the CAN questionnaire, confirms its usefulness in terms of planning mental health services and strategies for providing them.^14,15,16

The CANFOR is a questionnaire to survey the needs of forensic psychiatry patients staying in prisons, remand centers, and forensic psychiatric wards with all three degrees of security—maximum, enhanced, and basic.^10,11 This tool has obtained satisfactory reliability indicators. Inter-rater reliability was high for staff (0.998) and service users (0.991). Test–retest reliability was also satisfactory: for staff (0.852) and for service users (0.795).¹⁷

When constructing the scale for forensic psychiatry patients, the requirements that the CAN met were retained. It was assumed that the forensic scale should, like the original, have appropriate psychometric properties, be relevant and reliable, be fillable in 30 minutes, be usable by a wide range of professionals, be easy to learn and use without special training, be suitable for routine clinical practice and research, and be applicable to a wide range of audiences.¹⁷

There are 3 versions of this scale:¹⁰

1. Camberwell Assessment of Need—Forensic Clinical Version (CANFOR-C).

2. Camberwell Assessment of Need—Forensic Research Version (CANFOR-R).

3. Camberwell Assessment of Need—Forensic Short Version (CANFOR-S).

Each includes 25 modules, identifying needs in the areas of: accommodation, food, housekeeping, self-care skills, daily activities, physical health, psychotic symptoms, health and treatment information, psychological distress, safety to self, safety to others, alcohol use, drug use, social contact, close relationships, sex life, childcare, basic education, telephone skills, transport, obtaining financial support, money management, treatment process, risk of sexual offenses, and arson. The questions included in the different versions of the questionnaire identify specific patient needs and record the level of assistance received in meeting these needs from various sources. Informal sources include family and friends, while formal sources include hospital staff, social care, and associations of various kinds. Each module of the CANFOR-R version that was used in the study described in this thesis is divided into 4 sections, which respectively address: (1) an assessment of the patient’s need fulfillment, (2) an assessment of the degree of help to fulfill the need that the patient receives from informal sources, (3) an assessment of the degree of help to fulfill the need that the patient receives from formal sources, and (4) an assessment of the overall degree of satisfaction with the help that the patient receives to fulfill the need from formal and informal sources. Section 1 is assessed using a 5-point scale where 0 indicates no need, 1—a met need; 2—an unmet need; 8—not applicable; and 9—not known. Sections 2 and 3 are also assessed using a 5-point scale that is defined differently, depending on the degree of help the patient receives in meeting the need: 0—no help; 1—low help; 2—medium help; 3—high help; and 9—not known. Section 4 (satisfaction with the help received) is assessed using a shortened, 3-point scale: 0—not satisfied; 1—satisfied; and 9—not known.

The aim of the study was to adapt and test the properties of the reliability of the Polish version of the scale for the assessment of needs of forensic psychiatric patients—CANFOR. Nowadays, the only existing adaptations of the CANFOR scale come from Portugal, Italy, and Spain.^18,19,20 In Poland, until now, there has been no scale examining the needs of patients in psychiatric forensic wards. Approval was obtained from Professor Stuart Thomas to translate the scale for validation and adaptation in Poland.

The usefulness of the CANFOR scale is usually also tested on the basis of correlation with the results of such tools as the Life Skills Profile (LSP) and the Global Functioning Assessment Scale (GAF), the scales of which have confirmed psychometric properties.^17,18,19,20 It was decided to assess the initial correlations of the Polish version of the CANFOR scale with the data obtained from the questionnaire based on the LSP scale, which was already being already used.^17,18,19,20 Life Skills Profile is a scale of life skills as the name suggests. Life Skills Profile assesses the patient’s functioning in 5 areas: independent care (in other words—self-care), non-turbulence, social contact, communication, and responsibility. There are 3 versions of the scale—39, 20, and 16 points. The first one was used in our examination.

Material and Methods

Sixty patients aged 43.9 ± 13.4 years (mean ± SD) from basic and enhanced security level forensic psychiatry wards were recruited to test the initial version of the tool in Poland. The basic sociodemographic properties of the study group are presented in Table 1.

Table 1.

Sociodemographic Data

Feature (Variable)
Age
Mean ± SD	43.9 ± 13.4
Min-max	22-65
Length of stay (years)
Mean ± SD	2.1 ± 1.4
Min-max	1-7
Gender	n	%
Women	2	3.33
Men	58	96.67
Diagnosis
1. Schizophrenia, schizotypal, and delusional disorders	50	83.33
2. Mood disorders	6	10.00
3. Personality disorders	4	6.67
Type of offence
Homicide	8	13.33
Violence	46	76.67
Sexual offence	1	1.67
Other	5	8.33
Degree of security
Basic	40	66.67
Enhanced	20	33.33

n, frequency; SD, standard deviation; Min, lowest value; Max, highest value.

Inclusion criteria for study group:

• 18-65 years of age;

• informed consent for participation in the study; and

• stay in a forensic psychiatric ward due to a protective measure.

Exclusion criteria:

• intellectual disability; and

• inability to read and write.

All patients participating in the study were informed about the aim and method of the study and gave written consent. The data were collected under the supervision of a psychiatrist and psychologist who were members of the research team. The study was approved by the Bioethics Committee of Wroclaw Medical University, Poland, Poland (Approval no.: KB – 730/2011, Date: 2011).

Procedure

The study was conducted in the first half of 2013 at the Provincial Specialist Mental Health Hospital in Lubiąż, Poland, in wards with basic and enhanced security levels.

All materials were prepared in Polish, translating the text according to WHO recommendations, i.e., with back-translation and linguistic consultation,²¹ keeping in mind the very important issue of reflecting the conceptual intention of the authors of the tool, and not only reflecting the exact content of the original.²²

With this in mind, the English version of the scale was translated into Polish. After obtaining the Polish version during consultations with a team of specialists in psychiatry and psychology, it turned out that particular points of the questionnaire were adequate to the Polish reality and there was no need to modify them. The scale was then back-translated. The final version of the scale was psychometrically tested for reliability.

Statistical Analysis

The reliability of the CANFOR scale was tested by means of a concordance study between 2 investigators performing the study under identical conditions (so-called inter-rater) and a concordance study in a repeated study after a specified time (2 weeks) using the test–retest method. Cronbach’s α coefficient was used to assess the coherence of the test over time.

Calculated Cohen’s κ statistics (κ), Spearman’s rank correlation coefficient values, and the percentage of concordant choices for both investigators were used to assess inter-observer reliability.

Calculated Cohen’s κ statistics (κ), Spearman’s rank correlation coefficient values, and the percentage of concordant choices for both studies were used to assess the repeatability between 2 studies conducted on 2 dates (T0) by the same observer (test–retest reliability).

The Shapiro–Wilk test was used to assess the normality of distributions. Spearman’s rho correlation coefficient was used to analyze the interdependence of the CANFOR scale with other LSP tools.

When starting the statistical analysis, it was assumed that:

1. In testing the reliability of the CANFOR questionnaire, the κ coefficient was high, that is, between 0.81 and 1.00.

2. In a relevance study, there is a negative correlation of the CANFOR scale with the subscales of the LSP scale—patients with greater life skills have lower needs.^17-19

The calculations used the statistical software packages STATISTICA and MS Excel spreadsheet. A P value of <.05 was considered statistically significant.

Descriptive statistics of the data are presented with n (%) and, for non-normalized variables, are shown as “median (min-max),” and for normalized variables, are shown as “mean ± SD.”

Results

A total of 60 patients from basic and enhanced security wards (40 and 20 patients respectively), including 2 women and 58 men, were eligible for and participated in the study. The subjects were aged between 22 and 65 years. The length of stay in an inpatient health care facility averaged 2 years and 1 month (with a range of 1-7 years). The most common diagnosis was a disorder from the group (F20-F29) (schizophrenia, schizotypal, and delusional disorder)—50 people, followed by mood disorders (F30-F39)—6 people, and personality disorders (F60-F69)—4 people. The classification of disorders was based on the criteria of the International Statistical Classification of Diseases and Health-Related Problems—Tenth Revision (ICD-10). The respondents committed the following criminal acts: 8 people—homicide, 46—violent crimes, 1—sexual offense, and 5—other types of acts, such as theft and driving under the influence of alcohol. These data are presented in Table 1.

The effect of gender was not analyzed due to the presence of only 2 women in this study. For example, in Romeva’s study there were 87 men and only 3 women among the subjects. In contrast, in the Portuguese study by Talin,¹⁹ the number of men was 108 and women 35.

Assessment of Reliability and Reproducibility of the CANFOR-R Questionnaire

In the assessment of the reliability of the CANFOR questionnaire during the analysis of inter-rater reliability, that is, assessing at the same time by 2 independent researchers, perfect agreement was obtained in 22 areas of need. In 2 modules: “Health and treatment information” and “Childcare,” concordance was very good. In 1 module, “Sexual offences”—good—and specifically: the module “Information on health status and treatment,” the concordance according to the assessment of the staff is 89.7%, κ = 0.881, and according to the assessment of the patients, the concordance is 89.5%, κ = 0.897. The module “Childcare” in the assessment of staff achieved a concordance of 91.1%, κ = 0.903, in both the assessment of staff and patients. Lastly, in the module “Sexual offenses,” the concordance in the staff assessment is 84.3%, κ = 0.649, and in the patients’ assessment, it is also 84.3%, κ = 0.649. These data are presented in Table 2.

Table 2.

Inter-rater Reliability

CANFOR-R Area of Need	Carers (n = 60)		Patients (n = 60)
	Concordance (%)	Cohen’s kappa κ	Concordance (%)	Cohen’s kappa κ
1. Apartment	100.0	1.000	100.0	1.000
2. Meals/food	100.0	–	100.0	–
3. Housekeeping	100.0	1.000	100.0	1.000
4. Self-service	100.0	–	100.0	–
5. Daily activity	100.0	1.000	100.0	1.000
6. Physical health	100.0	1.000	100.0	1.000
7. Psychotic symptoms	100.0	1.000	100.0	1.000
8. Information on health and treatment	89.7	0.881	89.5	0.897
9. Psychological stress	100.0	1.000	100.0	1.000
10. Personal safety	100.0	1.000	100.0	1.000
11. Safety to others	100.0	1.000	100.0	1.000
12. Alcohol	100.0	1.000	100.0	1.000
13. Drugs/medication	100.0	1.000	100.0	1.000
14. Social contacts	100.0	1.000	100.0	1.000
15. Close relations	100.0	1.000	100.0	1.000
16. Sexual life	100.0	1.000	100.0	1.000
17. Childcare	91.1	0.903	91.1	0.903
18. Primary education	100.0	1.000	100.0	1.000
19. Telephone	100.0	–	100.0	–
20. Transport	100.0	–	100.0	–
21. Money	100.0	–	100.0	–
22. Benefits	100.0	1.000	100.0	1.000
23. Treatment	100.0	1.000	100.0	1.000
24. Sexual offences	84.3	.649	84.3	0.649
25. Arson	100.0	1.000	100.0	1.000

All Cohen’s κ are significant at the P < .001 level (except for needs no. 2, 4, 19, 20, and 21 because the test could not be performed).

On the other hand, in the test of the reproducibility of the answers to the CANFOR questionnaire after 2 weeks (test–retest reliability), perfect concordance—100% was obtained in 24 areas of need, and in the one “Sexual offenses”—good (Table 3), that is, needs assessment (staff) concordance 84.3%, κ = 0.649, needs assessment (patients): concordance 84.3%, κ = 0.649.

Table 3.

Analysis of Reproducibility of Responses after 2 Weeks (Test–Retest Reliability)

CANFOR-R Area of Need	Carers (n =60)		Patients (n = 60)
	Concordance (%)	Cohen’s Kappa	Concordance (%)	Cohen’s Kappa
1. Apartment	100.0	1.000	100.0	1.000
2. Meals/food	100.0*	–	100.0*	–
3. Housekeeping	100.0	1.000	100.0	1.000
4. Self-service	100.0*	–	100.0*	–
5. Daily activity	100.0	1.000	100.0	1.000
6. Physical health	100.0	1.000	100.0	1.000
7. Psychotic symptoms	100.0	1.000	100.0	1.000
8. Information on health and treatment	100.0	1.000	100.0	1.000
9. Psychological stress	100.0	1.000	100.0	1.000
10. Personal safety	100.0	1.000	100.0	1.000
11. Safety to others	100.0	1.000	100.0	1.000
12. Alcohol	100.0	1.000	100.0	1.000
13. Drugs/medication	100.0	1.000	100.0	1.000
14. Social contacts	100.0	1.000	100.0	1.000
15. Close relations	100.0	1.000	100.0	1.000
16. Sexual life	100.0	1.000	100.0	1.000
17. Childcare	100.0	1.000	100.0	1.000
18. Primary education	100.0	1.000	100.0	1.000
19. Telephone	100.0*	1.000	100.0*	1.000
20. Transport	100.0*	1.000	100.0*	1.000
21. Money	100.0*	1.000	100.0*	1.000
22. Benefits	100.0	1.000	100.0	1.000
23. Treatment	100.0	1.000	100.0	1.000
24. Sexual offences	84.3	0.649	84.3	0.649
25. Arson	100.0	1.000	100.0	1.000

Perfect repeatability (100%) was achieved in 24 areas of need and good in one (“Sexual offenses”).

The coherence of the test over time was assessed at an equally high Cronbach’s α level (from 0.649 to 1.0).

Interaction of the CANFOR-R with Other Measuring Tool

In the Polish correlation analysis of the CANFOR-R with another measurement tool (Table 4), significant values at the P < .05 level were obtained only in the case of several variables. A negative correlation was obtained between the total number of needs identified by carers and the total number of needs identified by patients based on the CANFOR scale and the scores obtained in the “mastery” domain of the LSP scale, respectively: ρ = −0.353, P = .006 and ρ = −0.385, P = .002; the number of needs met as defined by patients and the domain “restraint”: ρ = −0.284, P = .028; the number of unmet needs as determined by carers and the domain “restraint”: ρ = −0.278, P = .032; the number of unmet needs as determined by patients and the domain “restraint”: ρ = −0.301, P = .019. A negative correlation was also obtained in the comparison between the number of needs met identified by carers and the number of needs met identified by patients and the domain “social contacts,” respectively: ρ = 0.505, P = .000 and ρ = 0.518, P = 0.000.

Table 4.

Spearman’s Correlation Coefficients (ρ) of CANFOR-R Questionnaire Scores with Those of the LSP Questionnaire (Significant Values at P < .05 Are Marked).

	LSP
	Self-care	Restraint	Social contacts	Communication	Responsibility
LP-O
ρ	0000	−0.353	0.058	−0.012	0.011
P	−.997	.006	.661	.928	.931
LP-P
ρ	0.059	−0.385	0.066	−0.027	0.019
P	.653	.002	.618	.836	.885
LPZ-O
ρ	−0.101	−0.237	0.505	−0.036	−0.074
P	.443	.068	.000	.782	.573
LPZ-P
ρ	−0.081	−0.284	0.518	−0.048	−0.071
P	.539	.028	.000	.717	.589
LPN-O
ρ	0.057	−0.278	−0.205	−0.007	0.057
P	.663	.032	.115	.958	.664
LPN-P
ρ	0.106	−0.301	−0.198	−0.016	0.061
P	.418	.019	.130	.902	.643

LPN-O, number of unmet needs identified bycarers; LPN-P, number of unmet needs identified by patients; LP-O, total number of needs identified by carers; LP-P, total number of needs identified by patients; LPZ-O, number of needs met identified by carers; LPZ-P, number of needs met identified by patients; P, level of test significance; ρ, Spearman’s correlation coefficient.

Discussion

In terms of assessing the reliability of the CANFOR questionnaire when analyzing the inter-rater reliability, very high concordance was obtained in 22 areas. Similarly, the UK study had equally high results: with regard to the assessment of the total number of needs in the staff assessment, a concordance κ = 0.998 was obtained, while in the patients’ assessment of κ = 0.991, with regard to the unmet needs, respectively: 0.972 and 0.985.¹⁷ In contrast, low concordance was obtained in Castelletti’s Italian study¹⁸ for the total number of needs, assessed only by staff, in the area “Information on health and treatment” κ = 0.00, while unmet needs in the areas “Information on health and treatment” κ = 0.15, “Alcohol” κ = 0.00, “Drugs” κ = 0.00, and “Arson” κ = −0.12.

In a test of the test–retest reliability of the CANFOR questionnaire after 2 weeks, the level of concordance in the assessment of needs as assessed by staff was 84.3%, κ = 0.649, while the level of concordance in the assessment of needs as assessed by patients was also 84.3%, κ = 0.649. In the UK study, high concordance was obtained for patients κ = 0.795 and for staff κ = 0.852.¹⁸ Again, lower concordance was reported in Castelletti’s study,¹⁸ which the authors believe may be due to a discrepancy in the professional experience of the 2 researchers. In the Portuguese study, slightly lower concordance (assumed level of agreement between researchers κ = 0.445) was reported in 8 areas (κ = 0.182-0.374): “Housekeeping,” “Information on health and treatment,” “Safety to others,” “Social contacts,” “Close relations,” “Childcare,” “Telephone,” and “Arson.”¹⁹

Several factors may have contributed to the high-reliability coefficient obtained in the Polish study. The literature on the subject draws attention to the following aspects concerning the testing of the reliability of estimation scales. One of these is the accuracy of the measurements, which does not always depend on the tool itself but often on the skill and/or experience of the raters. The accuracy of ratings can be improved by increasing the number of competent judges who rate or by subjecting them to training in the skills of rating scales.²³ The original CANFOR manual was used to prepare for the correct conduct of the study: Camberwell Assessment of Need-Forensic Version. A needs assessment for forensic mental health service users, (published by the Royal College of Psychiatrists),¹¹ which provides detailed instructions for collecting interviews studied by both competent judges (a psychologist and a specialist psychiatrist). Other formal training was not provided, as suggested in the work on the validation of the original version of the scale.¹⁸ Differences in the assessment of the three domains between researchers can be attributed to the “leniency” error or the halo effect.

In the context of the “leniency” error, we can divide competent judges into 2 groups: a group of lenient estimators and a group of strict estimators. This type of error is also called prosecutorial or lawyers’ attitude toward the study. This occurs when a study is conducted on people known to the researchers—some patients were known to the competent judges conducting the study, and this may have influenced the results.

The halo effect is a very common type of error. Under the influence of the halo effect, judges succumb to the general impression of the subjects and, as a result, their judgments become inaccurate or divergent.²³ The errors described above did not affect the overall assessment of the reliability of the scale, which can be considered very good.

The negative correlation between the LSP and CANFOR scales, in studies by other authors, was more obvious than in our study.

In Spain, in terms of diagnostic accuracy, a negative correlation was obtained between LSF in all subscales (0.01, P < .05; of which in 2 subscales, the correlation was statistically weak P > .05) compared to needs assessed based on the CANFOR.¹⁴ However, this study took into account the total number of needs on the part of staff as well as patients without considering the division of needs into general, met, and unmet needs when rating the accuracy.

In the United Kingdom, on the other hand, the total number of needs assessed by the CANFOR scale was compared with the GAF scale and the 5-item life needs scale (specifically constructed for the study). The total number of needs had a statistically significant relationship with the GAF scale (τ = −0.27, n = 52, 95% CI 0.07-0.469), and there was a positive correlation between the total number of needs and the results of the 5-item life needs scale (τ = 0.37, n = 52, 95% CI 0.19-0.55).¹⁷ However, this study investigated the relationship between the needs assessed by staff and the GAF scale; Kendall’s tau rather than Pearson’s coefficient was used to determine the correlation.

The number of examined people (n = 60) is comparable with numbers used in other examinations of this kind. Although the number of subjects was higher in Portugal (n = 143) and Spain (n = 90), it was slightly lower in Thomas’ study¹⁸ (53 subjects) and even lower in Castelletti’s study¹⁹ in Italy (50 subjects). It is, of course, useful to model large sample sizes to minimize random error, remember to interpret the obtained accuracy coefficients carefully, and take into account the dimensions of the patient’s insight into his or her own functioning in subsequent studies.

A CANFOR-type tool for forensic psychiatry patients has not yet been available in Polish. Research on the psychometric parameters of diagnostic questionnaires is associated with numerous methodological controversies. In the analyses discussed here, for example, those related to the selection of the study group, competent judges, and the choice of the reference tool. In the Polish examination, a negative correlation was obtained between LSP and the needs assessed based on the CANFOR scale only in 2 domains: restraint and social contacts. What is more, the correlation analyses were mostly moderate, and in the case of the relationship between the number of needs met identified by patients from CANFOR and the restraint domain of LSP, the relationship turned out to be weak. Therefore, it is necessary to test the validity of the tool further, involving the methods of convergent and divergent validity assessments, as well as factor analysis. It is worth continuing research that includes a deeper analysis of psychometric parameters, taking into account a larger group of respondents from other forensic psychiatric centers, including maximum security wards.

Study Limitations

The main limitation of the research was the limited number of subjects, although the number of subjects in other projects regarding the evaluation of the CANFOR tool's properties was similar.

Further research will focus on increasing the group of respondents. Standard deviations close to the mean values were observed in some of the variables, which may call into question the normality of the distributions.

The accuracy of the described CANFOR scale requires confirmation in further research. Enlargement of the examined group and more analyses of different kinds of validity are needed.

The results of the study indicate that the reliability of the Polish version of the CANFOR questionnaire is very high—the scores given by two independent researchers at the same time are in the good or very good range (from 0.649 to 1.0).²⁴ In the range of diagnostic accuracy, a negative correlation was obtained between LSP and the needs assessed based on the CANFOR scale only in two domains: restraint and social contacts. More research is needed on the psychometric accuracy of the tool.

Funding Statement

The authors declared that this study has received no financial support.

Data Availability Statement

All data are available upon reasonable request.