1) Coping and Adaptation Processing Scale-Short-Form (CAPS-SF)

The CAPS, based on Roy's adaptation theory [3], is a self-report measure developed to assess the coping adaptation process of individuals in stressful situations. Each item is a short statement about how an individual responds to a crisis or extremely difficult event. The original questionnaire had 47 items with 5 sub-factors: resourceful and focused, physical and fixed, alert processing, systematic processing, and knowing and relating. The meaning of each sub-factor is as follows. Resourceful and focused is a major dimension of the construct of coping and adaptation processing that reflects behaviors using self and resources. It concentrates on expanding input, being inventive, and seeking outcomes. Physical and fixed is the dimension that highlights physical reactions and the input phase of handling situations. Alert processing is the dimension of coping and adaptation that uses behaviors that represent both the personal and physical self. It focuses on all three levels of processing: input, central, and output. Systematic processing describes personal and physical strategies used to take in situations and methodically handle them as a dimension of coping. Finally, knowing and relating is the dimension in which the strategies reflect the use of self and others, memory, and imagination. Cronbach's α coefficient for 47 items was .94 during the development, while Spearman-Brown split-half reliability scores for the five subscales were: .84 for resourceful and focused, .84 for physical and fixed, .80 for alert processing, .74 for systematic processing, and .78 for knowing and relating [3]. The 47-item CAPS was shortened to the 15-item version by Roy et al. [11] through item analysis based on IRT. The original 15-item CAPS-SF had satisfactory Cronbach's alpha coefficients (.82), and the preliminary validity of the 15-item CAPS-SF was confirmed. Face validity was confirmed in that items of the CAPS-SF are based on the MRT of coping and adaptation processing (i.e., including all adaptive modes and all types of cognitive processes). Concurrent validity was examined in an intervention study of 50 participants with spinal cord injuries. The scores on the CAPS-SF correlated with a value of .38 with Quality of Life (QoL) measure [11]. In addition, divergent validity was demonstrated in a correlational study of a sample of 35 patients with other neurologic disabilities. The CAPS-SF has a negative correlation of −.39 with self-reports of cognitive deficits, specifically concentration and memory difficulties [11]. The CAPS-SF uses a Likert scale format with response choices ranging from 4 (always) to 1 (never). Three items are reverse scored (items 5, 13, and 14). The possible range of scores is 15~60, with a high score indicating a greater capacity for using effective coping strategies [11].

2) Functional Assessment Cancer Therapy-General (FACT-G)

To assess criterion validity, the Korean version of the FACT-G (version 4), developed by Cella et al. [15] and validated for Korean patients with cancer by Kim et al. [16], was used after obtaining permission through the Functional Assessment of Chronic Illness Therapy Measurement System. The FACT-G is a measure that has globally established validity and reliability. This self-report scale measures both functional aspects and multidimensional domains of QoL such as physical, social/family, and emotional aspects that patients with cancer can experience in their course of treatment. Based on previous research suggesting that the use of effective coping strategies positively influences QoL [7, 8, 11, 17], we c onsider Q oL a n appropriate c oncept for evaluating the CAPS-SF's criterion validity.

This tool comprises 25 items on the QoL of patients with cancer, classified into 4 sub-factors: six physical well-being items, six social/family well-being items, six emotional well-being items, and seven functional well-being items, scored on a 5-point Likert scale ranging from 0 “absolutely no” to 4 “always.” The total range of scores is 0~100 points, where a higher score means better QoL. Cronbach's α when developing the Korean version was .90 [16]. Cronbach's α in the present study is .91 for the overall questionnaire, .89 for physical well-being, .87 for social/family well-being, .88 for emotional well-being, and .89 for functional well-being sub-factors.

3) Participants' general and clinical characteristics

Seven items were used to assess gender, age, marital status, education level, occupation, type of diagnosis, and duration of disease.

1) Translation and back translation process

Permission to translate and use the CAPS-SF was obtained via email correspondence with its developer, Roy. The translation and cultural adaptation process was based on Brislin's translation model [18, 19] and Roy and Chayaput's translation and cultural adaptation process [20]. Forward translation was conducted first, followed by the first panel discussion, back translation, a second panel discussion, a review of the back translation by the developer of the instrument, a content validity test, a pilot test, and final completion.

Forward translation of the measure from English to Korean was performed by two nursing professors proficient in both Korean and English. The first panel discussion was conducted to review translation accuracy, sentence structure, and similarity or congruence of meaning between the original, forward-, and translated items. This expert panel comprised the first two translators, an expert in English literature with a doctorate, two nursing professors with experience in scale development, and an oncology nurse. The translated scale was back translated into English by two bilingual English literature professors. Mutual independence was maintained between forward and back translators. The second panel discussion was conducted to review simplicity of terminology and similarity or congruence of meaning between original, forward-, and back-translated items. This expert panel comprised the two forward translators and back translators, an expert in English literature with a doctorate, two nursing professors with experience in scale development, and an oncology nurse. Only basic revisions were made such as switching the order of phrases in the 15 items. The translated CAPS-SF retained the same scale format, sequence, and item numbers as the original measure, and was confirmed by Roy, its developer.

2) Content validity

To test the content validity of the translated CAPS-SF items after the forward and back translation, a group of eight oncology nursing experts was formed: three adult nursing professors, one nursing professor in charge of the university's cancer nursing department, two nursing professors with experience in scale development, one oncologist, and one oncology nurse. They rated each preliminary scale item using a 4-point Likert scale: 1 “very inadequate,” 2 “inadequate,” 3 “adequate,” and 4 “very adequate.” The Content Validity Index (CVI) was computed as the proportion of experts rating an item as 3 or 4 [21]. The resulting itemlevel CVI was .80 or higher for each item; therefore, all 15 items were included in the final questionnaire.

3) Pilot test

Prior to the main investigation, a pilot test of the translated CAPS-SF was conducted to confirm that the scale's word difficulty level, sentence comprehension, and organization were appropriate, and that there were no difficulties associated with responding. The pilot test was conducted April 13~17, 2015. Participants were 30 patients with cancer visiting the hematology-oncology outpatient clinic of the C National University H Hospital. The questionnaire took eight minutes to complete. The results of the pilot test showed that eight participants (26.7%) found the meaning of Item 13, “Tend to become ill,” difficult to understand, while 14 participants (46.7%) responded that it was confusing, which prompted sentence revision. For Item 8, “Am more effective under stress,” five participants (16.7%) found the meaning of the sentence difficult, five (16.7%) found the meaning confusing, and six participants (20.0%) reported being disconcerted when responding, which also called for sentence revision. Therefore, after expert panel discussion, Item 13 was revised as “Tend to lie sick in bed” and Item 8 as “Can solve problems more effectively under stress.” The second pilot test was conducted on April 20, 2015 with 10 cancer patients using the preliminary scale with revised items (Items 8 and 13). The results of the pilot test showed agreement that there were no difficulties in understanding and responding to all items in the questionnaire. Cronbach's α was .85 for all items in the second pilot test.

1) Factor analysis

Previous studies have not performed factor analysis of the 15- item CAPS-SF and could not confirm the factor structure. Therefore, EFA was performed to confirm the factors and structure of the use of the instrument with cancer patients. The KMO to determine whether the 15 items used in the present study were adequate for EFA was .80, which exceeds the standard value of .60 [14]. Bartlett Sphericity verifies the null hypothesis–the correlation matrix of the variables is an identity matrix–which implies adequacy for factor analysis [22]; Bartlett Sphericity was χ²=724.76 (p<.001), which is adequate for factor analysis. EFA was performed using varimax rotation, an orthogonal rotation method in principal component factor analysis.

First, the 15-item CAPS-SF was reported on a unidimensional scale at the time of development, without performing the factor analysis. Therefore, in this study, EFA was performed setting the factor number to 1. However, the factor loadings for the 15 items ranged from .39 (Item 8) to .67 (Item 6), not satisfying the criterion of .50 or above in 5, 8, 10, 14, and 15 items. In addition, the explanatory power of total variance was 29.9%, which did not meet the standard value of 60.0%.

Therefore, we performed the rotation again based on the scree plot and eigenvalue of 1 and above for the factor extraction. As a result, four factors with an eigenvalue of 1.0 or above were extracted, which explained 60.6% of the total variance. The factor loadings for the 15 items ranged from .52 (Item 15) to .86 (Item 14), satisfying the criterion of .50 or above in all items [14] (Table 3).

Table 3
Factor Loading from Exploratory Factor Analysis for KCAPS-SF in Four Sub-factors Model

• Download Table

CFA was performed on the four sub-factors extracted in the EFA. T he standardized regression weight for e ach item w as .50~.85, exceeding the criterion of .50 [22] for all items. Results of the goodness of fit tests for the 15 items and four sub-factors of the KCAPS-SF were χ² (p)=115.67 (.013), Normed χ²=1.38, GFI=.92, SRMR=.02, RMSEA=.05, TLI=.94, and CFI=.95 (Table 4). That is, Normed χ² was below 3 [14]; GFI, TLI, and CFI above .90 [14, 22]; and SRMR and RMSEA below .05 [14, 22], indicating good model fit (Table 4).

Table 4
Confirmatory Factor Analysis for KCAPS-SF

• Download Table

Convergent validity assesses consistency in measuring a construct across items in the scale. In this study, since the critical ratio was 4.49~8.03 (p<.001), satisfying the criterion of ≥1.96 [22]; AVE was .62~.72, satisfying the criterion of ≥.50 [22]; and CR was .83~.90, satisfying the ≥.70 criterion [22], convergent validity was confirmed for the 15-item KCAPS-SF (Table 4). Discriminant validity of items is the degree to which sub-factors differ from each other. Items demonstrate discriminant validity of the AVE of a sub-factor is greater than the squared correlation coefficient between sub-factors [22]. Correlation coefficients between sub-factors ranged .23~.70, and since the squared value of the greatest coefficient, (.70), was .49, the AVE of each sub-factor (range .62~.72) was confirmed to be greater. Therefore, the low correlation between sub-factors of the scale indicates that independence is maintained between the sub-factors and discriminant validity was achieved (Table 4).

2) Factor naming

Factors were named according to the meaning of high-factor loading items in each sub-factor and by referencing the factor names of each sub-factor in the original 47-item CAPS [3]. The first factor included five items such as “Gather as much information as possible to increase my options” and “Try to get more resources to deal with the situation” and was named “Resourceful and focused coping” because it reflects an individual coping strategy of systematically gathering resources and information in order to understand and solve a problem. The eigenvalue of “Resourceful and focused coping” was 2.79 with 18.6% of its variance explained, and factor loadings of items ranged .52~.74. The second factor included three items such as “Give up easily too often” and “Tend to lie sick in bed” and was named “Physical and fixed coping” because it reflects the initial stages of coping involving the management of physical responses and stress. Its eigenvalue was 2.46 with 16.4% variance explained, and factor loadings of items ranged .79~.86. The third factor included four items such as “Take strength from spirituality or the successes of courageous people” and “Use humor in handling the situation,” and was named “Positive and knowing-based coping” because it reflects the attempt to solve a problematic situation by gathering relevant knowledge for effective coping, including one's own or others' successful experiences, the use of humor, and generating as many solutions as possible. The eigenvalue of “Positive and knowing-based coping” was 2.10 with 14.0% of variance explained, and item factor loadings ranged .56~.85. The fourth factor included three items such as “Can solve problems more effectively under stress” and “Call the problem what it is and try to see the whole picture” and was named “Alert processing coping” because it reflects alert, careful, and thoughtful coping and analysis of a problem. Its eigenvalue was 1.72 with 11.5% variance explained, and item factor loadings ranged .60~.74 (Table 3). The procedures of factor naming of the sub-factors in the KCAPS-SF and the names of each sub-factor were sent to Roy, the instrument's developer through an email and the results were confirmed by Roy.