Introduction

Cancer is one of the leading causes of mortality and remains a significant barrier to increasing life expectancy worldwide [1]. According to estimates from the International Agency for Research on Cancer, World Health Organization (WHO), approximately 19.3 million new cancer cases were reported in 2020 [1]. By 2040, it is projected that 28.4 million new cancer cases will occur, with countries that have a high Human Development Index (HDI) expected to experience the most significant increase in incidence [1]. Malaysia is classified as a high-HDI country, with 115,238 new cancer cases recorded between 2012 and 2016 [2]. Of these, 40.9% were detected at advanced stages, with the Johor state reporting the highest incidence rate [2].

Cancer is a chronic condition that progresses over the course of a patient’s life [3], and has become a concern for the entire family [4]. Cancer care has gradually shifted toward home care, with family members taking on the role of primary carers for patients [4]. Their responsibilities vary and may include assisting with the patient’s personal needs, maintaining cleanliness, managing finances, arranging care and services, organizing transportation, addressing problems and offering emotional support [5]. While providing cancer care can be a meaningful and fulfilling experience, caregivers often face significant challenges. These include conflicts between their social roles, limitations on their personal activities, strains in marital and family relationships and a decline in their physical health [6–8]. Such challenges can lead to substantial psychological effects [9] and in more severe cases, impair the caregivers’ overall quality of life (QOL) [4].

This concerning for caregivers of patients in advanced stages of cancer, as they are at an even higher risk of experiencing detrimental side effects [10,11]. The prolonged duration of illness and treatment, repeated phases of illness and hospital admissions, uncertain prognosis and increasing side effects from various treatment options further exacerbate caregivers’ burden [12,13]. The decline in QOL among cancer caregivers is often more pronounced in Asian culture, where filial virtue [14] and societal norms compel family members to take on the role of primary caregivers [15]. Notably, these negative impacts on caregivers can create a ripple effect, ultimately affecting the quality of care provided to cancer patients [16]. Therefore, it is of pressing need to assess the level and deficits in caregivers’ QOL to ensure appropriate support can be provided [14].

While numerous instruments have been developed to measure cancer patients’ QOL, only a limited number are available for their caregivers [17,18]. Among these, Weitzner’s [19,20] Quality of Life Index-Cancer (CQOLC) is the most widely used. The CQOLC is a multidimensional tool consisting of 35 items across 4 main factors, with well-tested psychometric properties [19,20]. The original CQOLC has been validated in multiple languages, including American English [20], French, Turkish and Chinese [21–23]. Although a Malay version is available, to the best of our knowledge, it has only been validated among caregivers of specific diseases such as HIV [24] and gastrointestinal cancer [3]. To date, no validation studies have been conducted among caregivers of general cancer patients in Malaysia. This represents a significant gap, as the lack of a validated tool for the wider population of cancer caregivers limits the ability to carry out research, systematically monitor caregiver well-being, and develop targeted support programs. Given the substantial challenges faced by cancer caregivers, having a reliable and validated assessment instrument is essential for understanding their needs and improving their quality of life. Therefore, we initiated this study to validate the Malay version of the CQOLC among caregivers of cancer patients, including assessments of internal consistency, test-retest reliability, and confirmation of the original factor structure, to provide a robust tool for use in both clinical and research settings.

Materials and methods

Results

Demographic characteristics

The characteristics of the participants are presented in Table 1. A total of 309 caregivers (99.68 %) of cancer patients were recruited for the study. The mean age of the participants was 44.00 14.22 years. The majority of participants were female (62.78 %), Malay (70.55 %), married (68.28 %) and had attained a secondary level of education (50.49 %). Spouses made up the largest group of caregivers (47.00 %), followed by children and parents of patients. On average, caregivers spent 75.50 50.60 hours per week providing care.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Participants’ socio-demographic characteristics (n = 309).

https://doi.org/10.1371/journal.pone.0335681.t001

Construct validity

This study reported a sampling adequacy of 0.90 from the Kaiser-Meyer-Olkin (KMO) measure and a significant Bartlett’s test of sphericity (p < 0.001). Exploratory Factor Analysis (EFA) using principle axis factoring with varimax rotation indicated that the MCQoL was valid. Varimax rotation was used because the constructs were uncorrelated and the rotation minimizes the number of items with higher factor loadings on each construct [27,28]. A total of four factors were extracted for the MCQoL, with eigenvalues ranging from 1.46 to 9.01, accounting for a total variance of 53.81%. Items with factor loadings of less than 0.40 were removed from the construct, as shown in Table 2. The three items removed were item numbers 4, 5, and 35, which corresponded to satisfaction with sexual functioning, maintenance of outside activities and family interest in caregiving, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Exploratory Factor Analysis of MCQoL (4-factor model).

https://doi.org/10.1371/journal.pone.0335681.t002

The EFA revealed that the Disruptiveness domain contained 9 items, with factor loadings for items 1, 2, 3, 13, 24, 26, 29, 30 and 32 ranging from 0.60 to 0.78. The Burden domain included 12 items, namely items 9, 11, 14, 15, 17, 18, 19, 20, 21, 25, 31 and 33 with factor loadings ranging from 0.44 to 0.69. The Positive Adaptation domain consisted of 8 items, with factor loadings for items 10, 12, 16, 22, 23, 27, 28 and 34 ranging from 0.59 to 0.78. Lastly, the Financial Concern domain had 3 items, with factor loadings for items 6, 7 and 8 ranging from 0.58 to 0.84. A summary of the exploratory factor analysis of the MCQoL is presented in Table 2.

We performed a model 1 of CFA of the 4-factor model of the MCQoL questionnaire, which allocated items to domains based on the findings from the EFA of this study. Specifically, items 1, 2, 3, 13, 24, 26, 29, 30 and 32 were assigned to the Disruptiveness domain; items 9, 11, 14, 15, 17, 18, 19, 20, 21 and 25 were allocated to the Burden domain; items 10, 12, 16, 22, 23, 27, 28 and 34 were designated for the Positive Adaptation domain; and items 6 to 8 were allocated to the Financial Concern domain. The model 1 of CFA of the MCQoL questionnaire revealed that the 4-factor model with 32 items did not fit the data well, with a chi-square value of p < 0.05, CFI = 0.829, TLI = 0.815, and RMSEA = 0.075 (Fig 1). Several revisions to the CFA model were undertaken to address areas of misfit. Specifically, items with modification indices greater than 3.84, standardized residuals above 2.58 and factor loadings below 0.50 were considered problematic. Problematic items were removed, or, where theoretically appropriate, correlating error between two items was applied. In total, eight items (numbers 9, 10, 19, 20, 25, 30, 31 and 32) were removed, while two pairs of error terms for items within the same factor were allowed to correlate. The final CFA model, comprising 24 items across four factors, demonstrated an acceptable fit: RMSEA criterion = 0.065, CFI= 0.904, Chisq/df (/df ratio) = 2.316 and moderate fit for TLI = 0.892. The final CFA model, presented in Figure 2, identified four distinct domains. The Disruptiveness domain comprised seven items (1, 2, 3, 13, 24, 26 and 29), with factor loadings ranging from 0.57 to 0.76. The Burden domain also included seven items, specifically, items 11, 14, 15, 17, 18, 21 and 33, with factor loadings between 0.51 and 0.80. The Positive Adaptation domain consisted of eight items (10, 12, 16, 22, 23, 27, 28 and 34), with loadings spanning 0.53 to 0.78. Finally, the Financial Concern domain contained three items (6, 7 and 8), with factor loadings ranging from 0.59 to 0.87 (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Model 1: Confirmatory Factor Analysis of MCQoL (4 factor model with 32 items).

https://doi.org/10.1371/journal.pone.0335681.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Final model: Confirmatory Factor Analysis of MCQoL (4 factor model with 24 items).

https://doi.org/10.1371/journal.pone.0335681.g002

Reliability

To assess the reliability of the final model of the MCQoL questionnaire, internal consistency was evaluated using Cronbach’s alpha coefficient (Table 3). The overall Cronbach’s alpha value for the MCQoL questionnaire was 0.88. All sub-scales demonstrated relatively high Cronbach’s alpha values, ranging from = 0.80 to 0.87. Additionally, test-retest stability was assessed as another measure of reliability. The intraclass correlation coefficient (ICC) for all constructs indicated fair to good stability, with values ranging from 0.37 to 0.77.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Reliability of final Confirmatory Factor Analysis of MCQoL (4 factor model with 24 items).

https://doi.org/10.1371/journal.pone.0335681.t003

Discussion

This study aimed to examine the validity and reliability of the Malay version of the CQoL (MCQoL). The original MCQoL is an adapted scale derived from the English CQOLC, an instrument designed to measure the disease-specific quality of life of cancer patients’ family members or caregivers. We obtained several key and novel findings. The final MCQoL demonstrated excellent internal consistency, with a Cronbach’s alpha of 0.88, which comparable to the original English CQOLC ( = 0.91) [20] and surpasses the reliability previously reported for Malaysian HIV/AIDS caregivers ( = 0.73 - 0.84) [24]. The alignment between the final model of MCQoL and the original English CQOLC likely stems from the instrument’s specific focus on cancer caregiving, whereas the lower reliability among HIV/AIDS caregivers may reflect differences in population and care context.

In terms of scale validity, EFA identified four factors for the final MCQoL, consistent with both the original MCQoL and the English CQOLC. Three items (4, 5, and 35) were removed due to low factor loadings of below 0.40 [28]. The elimination of Item 4, “Satisfaction with sexual functioning,” reflects deep-rooted cultural sensitivities in Malaysian society, where open discussions about intimate matters remain taboo, potentially affecting response validity. Similarly, the removal of Item 5, “Maintenance of outside activity” highlights cultural differences, particularly the Malaysian emphasis on collective family welfare over individual pursuits. In this context, family priorities often supersede personal interests, especially when caring for a loved one with cancer. If a patient resists treatment, relatives may appeal to familial obligations, encouraging them to “think about your parents or children.” In such life-and-death situations, discussing personal activities outside the family is often considered inappropriate or even selfish [36].

Although the factor structure of the MCQoL differs from international adaptations such as the Chinese and German versions, these variations likely reflect authentic cultural distinctions rather than measurement flaws. The redistribution of items across domains suggests that the conceptualization of caregiver quality of life is shaped by socio-cultural values, particularly in cultures with strong familial orientations. This underscores the importance of cultural adaptation in psychometric instrument development, highlighting the importance of addressing not only linguistic translation but also deeper cultural contexts that influence caregiving experiences and their assessment.

Interestingly, the items in the final MCQoL EFA model closely aligned with the subscales of the original MCQoL and the English CQOLC model [20]. The first factor included 9 out of the 32 items, with five items (1, 3, 24, 26 and 29) mapping directly to the “Disruptiveness” subscale in the original MCQoL and English CQOLC versions. Four additional items (2, 13, 30 and 32), not originally included in any subscale, also loaded onto this factor [20,24]. The second factor consisted of twelve items, all ten (9, 11, 14, 17, 18, 19, 20, 25, 31 and 33) from the “Burden” subscale of the earlier models, as well as item 21 (from “Disruptiveness”) and item 15 (previously unassigned). The third factor comprised of all seven items (10, 12, 16, 22, 27, 28 and 34) from the “Positive Adaptation” subscale of the original models, along with one additional item (23) (previously unassigned). Finally, the fourth factor included all 3 items (6, 7, and 8) from the “Financial Concern” subscale in the English CQOLC version [8], although this differed slightly from the original MCQoL, where only items 6 and 8 loaded in this domain [24]. This factor structure highlights both the conceptual consistency with theoretical models and meaningful refinement through cultural adaptation, ensuring the final MCQoL’s relevance to Malaysian caregiving experiences.

Noted that the initial CFA model 1 fit indices for MCQoL were mixed. While the TLI and CFI values were outside the acceptable range, the /df ratio and RMSEA met criteria for acceptable fit. To improve the CFA model 1, eight problematic items were removed. Four items were removed due to low factor loadings (items 9, 10, 20 and 31) and high residuals (items 19 and 25). Additionally, two items (30 and 32) were removed based on high modification indices. This approach aligned with the English CQOLC version, where items 30 and 32 were similarly not included in subscales [20]. Besides, two pairs of items within the same factors were allowed to have correlated errors in the model due to high modification indices. These correlations were theoretically justified and consistent with the original English CQOLC version [20].

In the final model, the first factor (Disruptiveness) included seven out of the twenty-four items: five items (1, 3, 24, 26 and 29) matched the original MCQoL and English CQOLC versions [20,24], and two items (2 and 13) were not previously assigned to specific subscales. The second factor (Burden) consisted of seven items, including five (11, 14, 17, 18 and 33) aligning with the original MCQoL and English CQOLC versions [20,24], one item (21) from “Disruptiveness” and another item (15) not originally assigned. The third factor (Positive Adaptation) comprised six items (10, 12, 16, 22, 27, 28 and 34) from the original models, plus one additional item (23) that did not align with the original four subscales. Finally, the fourth factor (Financial Concern) included all three items (6, 7 and 8) from the English CQOLC version [20].

The model fit indices for the final four-factor MCQoL, containing 24 items, indicated an acceptable fit. Specifically, the /df ratio was 2.32 (below the threshold of 3), the RMSEA was 0.06 (less than 0.08), and the CFI was 0.90 (above 0.90), demonstrating good alignment between the hypothesized structure and the sample data [37]. The CFA results confirmed the multidimensional nature of the MCQoL instrument. Notably, CFA was not performed for the original MCQoL or the English version of the CQOLC, while the German adaptation’s CFA reported poor model fit indices despite item removals [38]. These findings highlight the inherent challenges of achieving optimal model fit in cross-cultural validations, as differences in item interpretation and cultural context can affect factor structures. This study has several limitations that should be acknowledged. First, the use of single-center sampling restricts the generalizability of the findings, as participants were drawn from only one institution and may not represent the broader population. Second, convenience sampling was employed, which can introduce selection bias and limit the representativeness of the sample. However, this is the first study to date to validate the MCQoL questionnaire for caregivers of cancer patients in Malaysia. Therefore, it serves as an important stepping stone for future research, providing a foundation for larger-scale, multi-centre studies that can further confirm and extend these findings.

Conclusion

The final MCQoL scale demonstrated acceptable reliability and validity findings. The four-factor MCQoL model achieved a cumulative explained variance of 53.81%, and the internal consistency reliability was good for both the overall scale and its four domains. Based on these findings, the final MCQoL can be considered a reliable and valid instrument for assessing the quality of life of cancer caregivers in Malaysia. This scale holds potential for use by researchers and healthcare practitioners to better understand and address the challenges faced by this population.

Acknowledgments

The study team would like to express sincere gratitude to all study participants, study expert committee members and staffs in Clinical Research Centre (CRC) for the contribution for the success of the study. The study team would also like to thank the Director General of Health Malaysia for permission to publish this article.