Psychometric Evaluation of the Thai Version of the Burdened Caregiver Scale for Schizophrenia and Co-occurring Methamphetamine Use: A Pilot Study

Subjects

Data were collected from 142 caregivers of individuals with schizophrenia and co-occurring methamphetamine use at psychiatric hospitals. The inclusion criteria were as follows: (1) being the primary caregiver of a person of a family member diagnosed with schizophrenia or schizoaffective disorder according to the DSM-V criteria,¹⁰ without compensation for caregiving, and responsible for accompanying the patient to outpatient services; (2) having served as the primary caregiver for at least one year, as identified by individuals with schizophrenia and co-occurring methamphetamine use; (3) being at least 20 years of age; and (4) consenting to participate in the study.

Procedure

Over four weeks, identified inpatients, who met the criteria of a diagnosis of schizophrenia with co-occurring methamphetamine use and were 18–60 years old, were selected. A psychiatric nurse asked them to name their primary caregiver. The researcher asked them if we could contact the primary caregiver. When they agreed, and when the caregiver met the inclusion criteria, the self-report scale was collected and completed by the researcher teams’ caregivers or interviews.

Ethical approval

Ethical approval for the study was obtained from the Ethics Review Committee for Research Involving Human Research Participants (COA No. 284/2560). The scopes, risks, and benefits of this study for the subjects were explained. Before data collection, written consent was obtained directly from the PCPSs Participation was voluntary, and participant anonymity and confidentiality were guaranteed.

Data collection

The data collected included the following:

Scale development

Development of the TIBSCSM occurred in two phases: a) qualitative and b) quantitative. Item generation occurred during the qualitative phase, and item reduction and the validation process happened during the quantitative phase.¹¹ The two steps involved collecting data from different subjects.

Item generation: a qualitative approach

Item generation occurred in two steps. First, the content was derived from the questionnaire from face-to-face, semi-structured interviews performed by the researcher. Briefly, discussions addressed the impact and burden of care based on caregiver stress theory, a middle-range theory. Second, the objective was to predict caregiver stress and its outcomes from demographic characteristics, an objective burden in caregiving, stressful life events, social support, and social roles.⁷ The determined wording of question stems and the range of response options until saturation by 30 caregivers’ interviews. The researcher performed content analysis. Third, the researcher identifies 32 questions from this interview process. These items were answered using a five-point Likert scale, defined as: 1: never/not at all; 2: rarely/a little; 3: sometimes/somewhat; 4: often/a lot; and 5: always/very much.

30 caregivers were requested to remark on any part of the scale (i.e., content, wording, response choices) that they considered inappropriate or merited improvement. Ambiguous items and those that were misinterpreted or infrequently answered were withdrawn or rephrased, leading to a preliminary scale that contained 32 items. Lastly, preliminary interviews were conducted with the caregivers to ensure that the scale was a true reflection of the caregivers’ understanding and confirmed content validity. The second round of interviews with caregivers guaranteed its face validity.

Item reduction and validation of the TIBSCSM: a quantitative approach

The item reduction process comes from the results of statistical analyses and the steering committee’s expertise. Item response theory and classical test theory conduct by Statistical approaches.¹² Both metrological properties and their impact on the final scale’s content, taking into account the items’ meanings, were discussed and then removed items. The researcher retained items to produce the final version of the TIBSCSM in more robust and psychometrically sound solutions. The researcher tested for construct validity, reliability, and some aspects of external validity in the last version.

Convergent validity was evaluated by examining correlations between the TIBSCSM and two measures of caregivers’ quality of life: the Thai version of the Schizophrenia Caregiver Quality of Life Questionnaire (S-CGQoL-Thai Scale) and the Thai version of the World Health Organization Brief Quality of Life Scale (WHOQOL–BREF–THAI). Construct validity defines the construct to be assessed by the scale and measures its construct’s internal structure and the theoretical relationship of its item and subscale scores. It was evaluated using principal component factor analyses with varimax rotation¹³ to define the number of independent items and dimensions.

Sample characteristics

The mean age of the 142 participants was 47.1 years old (SD = 8.4). They were predominantly female (73.9%), married (63.4%), and about half had completed at least a bachelor’s degree (50.7%) and worked in agriculture (57.8%). Additionally, almost all of them (83.1%) had insufficient income, were healthy (61%), and had been the caregiver for more than five years (48%). Regarding medical history, about one-third of them (35.1%) used universal healthcare coverage, and more than half of them had no medical illness (61%) ( Table 1).

Exploratory factor analysis

The first-order factors loading = 0.5–0.8 and the variance of first-order factors loading were explained by a latent variable of second-order factors of the four constructs ranking 66.3% – 74.0%. Besides, factor loading of TIBSCSM, factor score, and R2 range between = 0.6–0.8, of TIBSCSM.

The second-order factors analysis of measurement model of TIBSCSM shows factor loading completely standardized solution of second-order factors in the high level ( Figure 1). The latent variables can explain the variance of construct of physical function, self-esteem, role and social enjoyment, and relationship satisfaction of 97.2%, 99.5%, 97.0% and 92.5%, respectively. The average variance extracted (AVE) of 0.948 and composite reliability (CR) of 0.987 ( Table 3).

Figure 1. Scree plot graph of TIBSCSM scale.

Validity

Content validity: The TIBSCSM, with 32 items, showed a content validity index = 1.

Convergence validity: The TIBSCSM correlated with the S-CGQoL-Thai Scale and the WHOQOL–BREF–THAI. Factor analysis showed good construct validity, Kaiser–Meyer–Olkin (KMO) = 0.9, Bartlett’s Test of Sphericity, χ² = 5248.5, df = 496, p < 0.001. For the content validity, the scale shows CVI = 1 and S-CVI = 0.86.

Reliability

Internal consistency: Cronbach’s alpha showed high internal consistency reliability (α = 0.9).

The corrected item total correlation ranged from 0.5–0.8. The total variance explained was 64.9%, which is excellent. Also, the extract construct of factor analysis was the following: physical function, self-esteem, role and social enjoyment, and relationship satisfaction ( Figure 1).

The mean, standard deviation, internal consistency, and reliability of various question sets: a) Physical Function (6 items: Item 1, Item 10, Item 14, Item 24, Item 29, Item 32): Mean: 52.53, S.D. =16.75, Internal Consistency: Ranges from 0.75 to 0.86, and Reliability = 0.96, b) Self-Esteem (Mastery) (9 items: Item 15, Item 16, Item 17, Item 18, Item 20, Item 21, Item 22, Item 30): Mean: 46.03, S.D. = 14.42, Internal Consistency ranges from 0.73 to 0.83, Reliability = 0.95, c) Role Enjoyment (11 items: Item 2, Item 3, Item 4, Item 5, Item 12, Item 13, Item 23, Item 25, Item 26, Item 27, Item 28): Mean = 50.04, S.D. = 13.60, Internal Consistency ranges from 0.71 to 0.84, Reliability = 0.98, d) Relationship Satisfaction (6 items: Item 6, Item 7, Item 8, Item 9, Item 11, Item 31): Mean = 49.96, S.D. = 15.62, Internal Consistency ranges from 0.67 to 0.81, and reliability = 0.86. This comprehensive data is summarized in Table 4.

Correlation testing between the 32 items found that the correlation coefficient of all 32 questions was in the range 0.3–0.8, and the correlation coefficient of the internal items of each construct were medium to high in the positive correlation with statistical significance at 0.05 ( Table 2, Figure 2).

Figure 2. Second-order of Impact and burden care model.

Implications for clinical practice and further research

This study highlights the necessity of tailoring interventions to distinct caregiver burden domains. Caregivers experiencing physical strain may benefit from structured respite care programs, while those struggling with self-esteem and emotional distress require psychological support.^2,7,22 The identified dimensions reflect caregivers’ diverse experiences, underscoring the need for multidisciplinary approaches to caregiving support.⁹

Additionally, caregiver burden is intensified by financial strain, role confusion, and social stigma, reinforcing the necessity of targeted interventions to alleviate these challenges.²³ Previous studies indicate that caregiver stress significantly impacts patient outcomes, particularly in cases of co-occurring schizophrenia and substance use disorders.⁴ Therefore, psychiatric nurses and mental health professionals can utilize the TIBSCSM to develop specialized training, psychoeducation programs, and policy initiatives that enhance caregiver resilience.⁵

Further methodological refinement using variance-based structural equation modeling²⁰ could be beneficial in future studies to enhance the measurement precision of caregiver burden assessment tools. Recent advancements in factor analysis techniques, as discussed by Kock (2019),²⁴ may also contribute to improving the construct validity of psychometric instruments designed for caregiver burden assessment.