Abstract
The purpose of this study was to translate the Diabetic Foot Ulcer Scale‐Short Form (DFS‐SF) into Korean and evaluate its psychometric performance in patients with diabetic foot ulcers (DFUs). The DFS‐SF was translated into Korean using translation guidelines from the World Health Organization. The Korean version of DFS‐SF (DFS‐SF‐K) went through the full linguistic validation process and was evaluated in 320 Korean patients with DFU. For this study, Cronbach's alpha was used to evaluate the reliability of the DFS‐SF‐K. Exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and known‐group validity were conducted to examine construct validity. Correlation of the DFS‐SF‐K with SF‐36 was used to test concurrent validity of the DFS‐SF‐K. Cronbach's alpha was >0.70 for all scales. Factor loadings of the 29 items on the six subscales ranged from 0.82 to 0.95. The six‐subscale model was validated by CFA (x 2/df = 4.55, P < 0.001, RMR = 0.06, GFI = 0.73, NFI = 0.90, TLI = 0.91, CFI = 0.92, RMSEA = 0.10). Spearman's correlations and known‐groups comparisons supported construct validity. The newly translated DFS‐SF‐K may be used to assess the impact of QoL with DFUs in Korean patients.
Keywords: DFS‐SF, diabetic foot ulcer, Korean, QoL, validity
1. INTRODUCTION
According to the survey Diabetes in Korea, 44.8% of patients who had a foot dissection in 2003 had diabetes.1 The incidence rate of foot dissection was 10.1 times higher among patients with diabetes than those without diabetes. Meanwhile, the incidence rate of foot ulcers was 7.8 times higher among patients with diabetes compared with those without diabetes. Such relative risks of foot complications were found to be much higher than those of kidney and eye complications arising from diabetes, demonstrating that foot complications are more serious than other diabetic complications.1 In addition, when patients had their feet dissected for diabetic foot diseases, the number of days of health care use increased 1.6 to 2.1 times more than when non‐diabetic patients had their feet dissected. When patients had foot ulcers, the number of days of healthcare use increased 1.4 to 2.0 times. The mean annual total healthcare expense for foot dissection and foot ulcers increased 2.0 to 2.3 times and 1.7 to 2.3 times more for those with diabetes than for those without diabetes.1, 2
These complications affect the lives of patients in many ways, not just financially, as limitations in movement caused by foot dissection require adaptation to a new lifestyle.3, 4 Many studies suggest that diabetic foot greatly affects patients' physical health.4, 5, 6, 7, 8, 9, 10 Several studies reported that diabetic foot has a negative effect on patients' daily, social, leisure, and family lives.3, 11 In addition, another study showed that it not only affects physical health but also has psychological and emotional effects.12 In other words, diabetic foot affects patients' quality of life (QoL) in various ways.
Understanding patients' QoL can lead to more successful treatment. The interdisciplinary treatment plan should consider not only clinical factors but also patients' experiences and perceptions.13 Furthermore, when medical professionals take a patient‐oriented approach, assessing patients' experiences and understanding and how they have been treated for diabetes and its complications, they could give prescriptions and education tailored to individual patients, which would greatly help patients manage their diabetic feet preventatively.14
Generic and disease‐specific tools are available for measuring QoL. Disease‐specific tools could provide better information, including details about physical, emotional, and social functions related to the characteristics of the disease.15 Tools designed to specifically assess the QoL of patients with diabetic foot include the Diabetic Foot Ulcer Scale (DFS), the Diabetic Foot Ulcer Scale‐Short Form (DFS‐SF), and NeuroQoL. The DFS Scale was developed based on a semi‐structured interview and focus group of diabetic foot patients and their family members. Regarding wound severity and healing, this tool shows internal consistency, reliability, validity, and sensitivity.16 Its shortened version, DFS‐SF, also shows a similar level of robustness and sensitivity.17 The DFS‐SF has shown a statistically significant correlation with SF‐36 and is user friendly in that patients could use it more often and easily in the hospital as it consists of only 29 questions. NeuroQol is a tool that measures the QoL of patients with diabetes complicated by peripheral neuropathy and diabetic foot ulcer (DFU). However, it shows weak sensitivity to ulcer severity.18
Therefore, the DFS‐SF could be regarded as an appealing instrument in assessing the QoL of patients with diabetic foot. Thus, many countries have translated the DFS‐SF into different languages and use these translated versions.19, 20, 21 However, no Korean version is yet available.
This study aimed to translate the DFS‐SF into the Korean language, to apply the DFS‐SF to Korean patients and consider the applicability and contribution of the tool to assess the QoL of patients with diabetic foot, and to verify the validity and reliability of the DFS‐SF‐K—the Korean version of DFS‐SF.
2. MATERIALS AND METHODS
2.1. Design
This was a methodological study designed to translate the tool to assess the QoL of patients with diabetic foot (DFS‐SF), the tool developed by Bann et al,17 into the Korean language and verify the DFS‐SF‐K's validity and reliability.
2.2. Instruments
2.2.1. Diabetic Foot Ulcer Scale‐Short Form
The original DFS is a foot ulceration‐specific instrument that consists of 58 items grouped into 11 domains as follows: leisure, physical health, daily activities, emotions, non‐compliance, family, friends, positive attitude, treatment, satisfaction, and financial.16 The DFS‐SF, which was used in this study, is its shorter version and has 29 items rated on a 5‐point Likert‐type scale. The items are assessed using six categories as follows: leisure (five items), dependence/daily life (five items), negative emotions (NEs) (six items), physical health (five items), worry about ulcers (WUC)/feet (four items), and burden from ulcer care (four items).17 Domain scores are based on the sum of all items associated with that domain (raw item scores are reverse coded when necessary), and the scores per dimension are transformed to a scale of 0 to 100, with higher scores corresponding to better QoL. The original English‐language versions of both instruments have been psychometrically tested and demonstrated good validity and reliability.16, 17
2.2.2. SF‐36
To verify criterion validity, this study obtained consent from the author to use the QoL scale. SF‐36 is composed of 26 questions divided into eight subcategories (PF, RP, BP, VT, SF, RE, MH, and GH).22 In the scoring of the items in SF‐36 from 0 to 100, weighted values were assigned because the response scale varied depending on the question.23 Then, the scores in the questions applicable to each of the eight subcategories were added and averaged before they were used as scores for each subcategory. A higher score means a better health status. The instrument was translated into Korean.24 It was found that it has high internal consistency reliability and convergent and discriminative validities.
2.3. Study process
2.3.1. Development of a provisional Korean DFS‐SF
The scale was translated in the following order: forward translation, expert panel, back translation, pretesting, and final version, in accordance with the guidelines of the World Health Organization.25 First, consent was obtained from Mapi Research Trust (Lyon, France) for this study to validate the Korean DFS‐SF. For forward translation, one wound specialist physician, two wound specialist nurses, and two translators translated the original scale into Korean. Word‐for‐word translation was avoided in favour of concise and clear words and sentences that ordinary people could easily understand. The expert panel, established to compare disagreements and culturally different expressions between the original and translated versions, consisted of those involved in forward translation, a bilingual nursing professor, a Korean language major, and a researcher working as a wound specialist nurse. The panel raised questions about words and expressions, suggested alternatives, and revised the translated version by making simple revisions, such as changing the passive voice, tense, and order of words and phrases. One wound specialist physician and translators fluent in English back‐translated the translated version without referring to the original English version. Similarities in sentence structure and meaning of the back‐translated questions were reviewed by the researcher and the nursing professor before the final version was completed. The translated DFS‐SF was used to assess 10 patients with DFUs from June 1 to 15, 2016. The mean age of the participants for pretesting was 65.23 ± 8.55 years. The purpose of the pretesting was to assess the clarity, easiness of understanding, adequacy of the questionnaire form of the DFS‐SF‐K, and the time taken to respond to the questionnaire. According to the pretesting results, about 15 to 20 minutes were adequate for completing the questionnaire. Participants had no difficulty ticking off the questions in the self‐administered questionnaire, and feedback on the words that ordinary people might find difficult to understand, such as exudate, was incorporated by inserting an explanation behind the word, such as in “exudate (pus or discharge around the wound).”
2.3.2. Validity and reliability
Content validity verification: To verify the content validity of the DFS‐SF, an expert group of five wound specialist nurses and one nursing professor was organised to validate the content. Each question was evaluated using the 4‐point Likert scale to calculate the content validity index (CVI) as follows: very valid (4 points), valid (3 points), not valid (2 points), and not valid at all (1 point). The group also evaluated the readability of the tool while they verified its content validity and provided feedback about the problematic questions that they thought needed to be revised or were difficult to understand.
Construct validity verification: To verify construct validity, we performed exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). By doing so, we verified the goodness of fit, convergent validity, and discriminant validity of the model and analysed the known‐groups technique.
Criterion validity verification: To verify criterion validity, we used SF‐36—which has internationally proven validity and reliability as a tool to measure QoL—and identified its correlations with each subcategory in the DFS‐SF‐K.
Reliability verification: To verify the reliability of the DFS‐SF‐K, internal consistency was analysed using Cronbach's alpha.
2.4. Sample and data collection
After acquiring research approval from the institutional review board (IRB) of the hospital to which the authors were affiliated, the study participants included were patients who satisfied the following inclusion criteria: (a) native Korean, (b) able to understand Korean, (c) has current or healed DFU, and (d) does not have psychiatric illness. Data were collected from patients with a diabetic foot who visited as an outpatient or were admitted and treated at the Diabetic Wound Center of Korea University Guro Hospital between July 1 and December 26, 2016. The questionnaire was administered by the researcher, and the participants were asked to fill out the consent form for research participation before they responded to the questionnaire. The questionnaire clearly stated that its results shall not be used for purposes other than research, that the participants may cancel their participation in the research at any time during the research period, and that their information shall be kept confidential and their names shall remain anonymous. With regards to the sample size, the number of participants required for EFA should be at least 10 times higher than the number of variables, generally ≥100. The size required in a sample for CFA is usually 100 to 200. In this regard, 360 copies of the questionnaire were distributed in the hospital, and 320 copies were used for final analysis data, excluding 40 copies with insufficient responses.
2.5. Ethical considerations
This study was granted approval by the IRB of Korea University Guro Hospital. The principles set out in the Declaration of Helsinki, as well as national and local guidelines for research, were followed.
2.6. Statistics
Data collected in this study were analysed using IBM SPSS Statistics for Windows and SPSS AMOS, version 20.0 (IBM Corp., Armonk, NY, USA). The general characteristics of the participants were analysed and expressed as frequencies, percentages, and standard deviations.
The expert group verified the content validity of the tool with the 4‐point Likert scale and calculated the CVI. The construct validity was verified using EFA and CFA. EFA used a principal component analysis and the varimax rotation. The Kaiser–Meyer–Olkin (KMO) test and Bartlett's sphericity test were performed to determine whether the collected data were adequate for factor analysis. KMO is a scale that represents correlations between variables; the correlation becomes more adequate when it is closer to 1.0. The Bartlett sphericity test was used to prove the null hypothesis that the correlation matrices of the variables are identity matrices and to verify whether the diagonal line is 1 and the other lines are 0. In general, when KMO is ≥0.60, and the Bartlett sphericity test is P < 0.05, data are considered adequate for factor analysis. By using CFA AMOS 22.0, the normed fit index (NFI), Tucker–Lewis fit index (TFI), and comparative fit index (CFI) were used to confirm the relative goodness of fit, while the goodness‐of‐fit index (GFI) and root‐mean‐square error of approximation (RMSEA) were used to confirm the absolute goodness of fit. For the known‐groups technique, the independent t test was used. To verify criterion validity, the correlations between SF‐36 and each of the subcategories in the DFS‐SF‐K were analysed. To validate the overall reliability of the tool and each of the subcategories of the tool, Cronbach's alpha was calculated.
3. RESULTS
3.1. General characteristics of the participants
Among the 320 participants, the men accounted for 63.12% (Table 1). Patients' mean age was 67.75 years, and 51.88% of the patients were treated with their caregiver. The mean period since they were diagnosed with diabetes was 18.40 years, and 38.44% of the patients had an experience of previous amputation. Of the patients, 93.44% were diagnosed with neuropathies as diabetic complications, 50.63% with chronic kidney disease (CKD), 34.96% with chronic heart failure (CHF), and 62.50% with ophthalmologic disease.
Table 1.
Demographic and clinical characteristics of study participants (N = 320)
| Demographic and clinical characteristics | N |
|---|---|
| Age (y) | 67.75 ± 14.90 |
| Gender (male: female) | 202:118 |
| Partner involved in care | 166 (51.88%) |
| Duration of diabetes | 18.40 ± 8.20 |
| Previous lower extremity amputation | 123 (38.44%) |
| TcpO2 < 30 mm Hg | 177 (55.31%) |
| Toe Pressure < 30 mm Hg | 123 (38.44%) |
| Current ulcer: healed ulcer | 189:131 |
| Comorbidities | |
| Heart failure | 162 (50.63%) |
| End‐stage renal disease | 111 (34.96%) |
| Diabetic polyneuropathy | 299 (93.44%) |
| Visual impairment | 200 (62.50%) |
3.2. Validity of the Korean DFS‐SF
3.2.1. Content validity
The mean CVI in the DFS‐SF was 0.98, and the CVI of each question was evaluated to be ≥0.8. Among the 29 questions, the CVI of 1.0 was the most common with 18 questions (62.1%), followed by 0.96 with 9 questions (31.0%) and 0.92 and 0.88 each with 1 question (3.4%).
3.2.2. Construct validity
Exploratory factor analysis
The KMO measure of the sampling adequacy was 0.90, higher than the baseline of 0.80. In the Bartlett sphericity test, the chi‐square value was found to be 16 655.11, the degree of freedom 406, and the P‐value, <0.001, which verified the adequacy for factor analysis.
The primary component analysis was used for factor analysis, and the varimax rotation was used to analyse exploratory factors. The results indicated that the number of factors with an eigenvalue of ≥1 was six, and the cumulative explanatory power of the six factors was 92.43% (Table 2). The first factor was leisure with five questions (16.16% in explanatory power); the second factor was physical health (PH) with five questions (16.65% in explanatory power); the third factor was dependence/daily life (D/DL) with five questions (16.22% in explanatory power); the fourth factor was NE with six questions (18.61% in explanatory power); the fifth factor was WUC with four questions (12.26% in explanatory power); and the sixth factor was burden from ulcer care with four questions (12.53% in explanatory power).
Table 2.
Factor loading from exploratory factor analysis for DFS‐SF‐K
| Item no. | Description | F1: L | F2: PH | F3: D/DL | F4: WU/F | F5: NE | F6: BUC |
|---|---|---|---|---|---|---|---|
| D4G | Depressed because you are not able to do what you wanted | 0.892 | |||||
| D4I | Angry that this has happened to you | 0.864 | |||||
| D4B | Frustrated by others doing things for you | 0.864 | |||||
| D4J | Frustrated because you have difficulty getting about | 0.862 | |||||
| D4C | Frustrated because you not able to do what you wanted | 0.848 | |||||
| D4A | Angry because you are not able to do what you wanted | 0.819 | |||||
| D2E | Pain during night | 0.951 | |||||
| D2C | Had difficulty sleeping | 0.946 | |||||
| D2A | Felt fatigued | 0.945 | |||||
| D2B | Felt drained | 0.944 | |||||
| D2D | Pain while walking or standing | 0.944 | |||||
| D3D | Spend more time planning daily life | 0.925 | |||||
| D3B | Depend on others to do household chores | 0.922 | |||||
| D3E | Felt doing anything took longer than would have liked | 0.919 | |||||
| D3C | Depend on others to get out of the house | 0.911 | |||||
| D3A | Depend on others to look after you | 0.906 | |||||
| D1D | Made you choose different kind of holiday | 0.934 | |||||
| D1A | Stopped from engaging in recreational activities | 0.933 | |||||
| D1E | Had to spend more time planning leisure activities | 0.926 | |||||
| D1C | Stopped from getting away for a holiday | 0.914 | |||||
| D1B | Changed kinds of recreational activities | 0.911 | |||||
| D5C | Bothered by appearance of ulcer | 0.908 | |||||
| D5A | Bothered by having to keep weight of foot ulcer | 0.895 | |||||
| D5B | Bothered by amount of time involved in caring for ulcer | 0.887 | |||||
| D5D | Bothered by having to depend on others for care of ulcer | 0.868 | |||||
| D4H | Worried about getting ulcers in future | 0.857 | |||||
| D4E | Worried that you may have to have an amputation | 0.854 | |||||
| D4F | Worried about injury to feet | 0.851 | |||||
| D4D | Worried that ulcer will never heal | 0.842 | |||||
| Variance (%) | 16.16 | 16.65 | 16.22 | 18.61 | 12.26 | 12.53 | |
| Accumulation (%) | 16.16 | 32.81 | 49.03 | 67.64 | 79.90 | 92.43 | |
Abbreviations: BUC, bothered by ulcer care; D/DL, dependence/daily life; L, leisure; NE, negative emotion; PH, physical health; WU/F, worried about ulcers/feet.
Confirmatory factor analysis
By using AMOS 20.0, a CFA was performed on the DFS‐SF‐K questions. The goodness of fit of the model was analysed, and the results showed the following values: χ 2 = 1666.48, P < 0.001, χ 2/df = 4.64, RMR = 0.06, GFI = 0.73, NFI = 0.90, TLI = 0.91, CFI = 0.92, and RMSEA = 0.10 (Figure 1).
Figure 1.
Confirmatory factor analysis. F1, leisure; F2, physical health; F3, dependence/daily life; F4, worried about ulcers/feet; F5, negative emotion; F6, bothered by ulcer care
Convergent validity means that if various methods are used to measure a certain construct, the correlation between the measures should be high. In other words, it refers to the validity of the explanatory power of the observed variables for the latent variables. When verifying convergent validity, the average variance extracted (AVE) should be ≥0.5, while the construct reliability (CR) should be ≥7. The AVE for each subcategory of the DFS‐SF‐K were as follows: “leisure,” 0.78; “physical health,” 0.87; “dependence/daily life,” 0.81; “negative emotion,” 0.86; “worry about ulcers/feet,” 0.75; and “burden from ulcer care,” 0.78 (Table 3). All AVEs were > 0.5. For CR, the values were as follows: “leisure,” 0.95; “physical health,” 0.97; “dependence/daily life,” 0.96; “negative emotion,” 0.97; “worry about ulcers/feet,” 0.92; and “burden from ulcer care,” 0.93. As all CR values were > 0.7, CR was regarded as having convergent validity. Discriminative validity means that the correlation between the measures obtained from the different methods on constructs should be low. In other words, it is a type of validity to verify whether no redundancy or similarity exists between constructs or whether discrimination exists between them. When discriminative validity is verified, the AVE should be higher than the squared correlation coefficient. When the AVE on the diagonal line is higher than the squared coefficient of a correlation between the latent variables at the bottom of the diagonal line, discriminative validity between the two variables is ensured (Table 4).
Table 3.
Convergent validity of DFS‐SF‐K
| Standardisedestimate | Estimate | S.E. | C.R. | P | SMC | CR | AVE | |
|---|---|---|---|---|---|---|---|---|
| D1A ← F1 | 0.97 | 1.00 | 0.94 | 0.95 | 0.78 | |||
| D1B ← F1 | 0.93 | 0.96 | 0.03 | 37.43 | <0.001 | 0.86 | ||
| D1C ← F1 | 0.92 | 0.86 | 0.02 | 36.10 | <0.001 | 0.85 | ||
| D1D ← F1 | 0.96 | 1.08 | 0.02 | 46.32 | <0.001 | 0.93 | ||
| D1E ← F1 | 0.95 | 1.03 | 0.02 | 42.87 | <0.001 | 0.91 | ||
| D2A ← F2 | 0.97 | 1.00 | 0.94 | 0.97 | 0.87 | |||
| D2B ← F2 | 0.97 | 0.95 | 0.02 | 48.02 | <0.001 | 0.94 | ||
| D2C ← F2 | 0.98 | 0.94 | 0.02 | 51.23 | <0.001 | 0.95 | ||
| D2D ← F2 | 0.97 | 0.98 | 0.02 | 46.83 | <0.001 | 0.93 | ||
| D2E ← F2 | 0.98 | 1.06 | 0.02 | 51.04 | <0.001 | 0.95 | ||
| D3A ← F3 | 0.99 | 1.00 | 0.97 | 0.96 | 0.81 | |||
| D3B ← F3 | 0.96 | 0.87 | 0.02 | 53.40 | <0.001 | 0.93 | ||
| D3C ← F3 | 0.97 | 0.92 | 0.02 | 58.62 | <0.001 | 0.94 | ||
| D3D ← F3 | 0.89 | 0.87 | 0.03 | 32.18 | <0.001 | 0.78 | ||
| D3E ← F3 | 0.89 | 0.96 | 0.03 | 33.41 | <0.001 | 0.80 | ||
| D4A ← F4 | 0.95 | 1.00 | 0.90 | 0.97 | 0.86 | |||
| D4B ← F4 | 0.94 | 1.00 | 0.03 | 36.13 | <0.001 | 0.89 | ||
| D4C ← F4 | 0.96 | 0.99 | 0.03 | 39.25 | <0.001 | 0.92 | ||
| D4G ← F4 | 0.96 | 1.07 | 0.03 | 39.90 | <0.001 | 0.93 | ||
| D4I ← F4 | 0.97 | 1.06 | 0.03 | 41.74 | <0.001 | 0.94 | ||
| D4J ← F4 | 0.96 | 1.17 | 0.03 | 40.01 | <0.001 | 0.93 | ||
| D4D ← F5 | 0.92 | 1.00 | 0.85 | 0.92 | 0.75 | |||
| D4E ← F5 | 0.92 | 1.00 | 0.04 | 28.92 | <0.001 | 0.84 | ||
| D4F ← F5 | 0.90 | 0.93 | 0.03 | 27.29 | <0.001 | 0.81 | ||
| D4H ← F5 | 0.96 | 1.04 | 0.03 | 33.53 | <0.001 | 0.92 | ||
| D5A ← F6 | 0.90 | 1.00 | 0.81 | 0.93 | 0.78 | |||
| D5B ← F6 | 0.96 | 1.26 | 0.04 | 30.15 | <0.001 | 0.91 | ||
| D5C ← F6 | 0.93 | 1.10 | 0.04 | 27.48 | <0.001 | 0.86 | ||
| D5D ← F6 | 0.91 | 1.17 | 0.05 | 26.15 | <0.001 | 0.83 |
Table 4.
Discriminant validity of DFS‐SF‐K
| L | PH | DD | WU | NE | BUC | |
|---|---|---|---|---|---|---|
| L | 0.78a | |||||
| PH | 0.08b | 0.87a | ||||
| DD | 0.04b | 0.12b | 0.81a | |||
| WU | 0.11b | 0.08b | 0.23b | 0.86a | ||
| NE | 0.00b | 0.02b | 0.14b | 0.34b | 0.75a | |
| BUC | 0.01b | 0.05b | 0.08b | 0.20b | 0.19b | 0.78a |
Abbreviations: BUC, bothered by ulcer care; D/DL, dependence/daily life; L, leisure; NE, negative emotion; PH, physical health; WU/F, worried about ulcers/feet.
AVE.
Coefficient of determination (r 2).
Known‐groups validity
The mean DFS‐SF‐K scores of the group with DFUs (n = 189) and the groups without DFUs (n = 131) were 195.96 ± 100.55 and 421.16 ± 91.24, respectively (Table 5). The mean score of the group with DFUs was statistically significantly higher than that of the group without DFUs (t = 20.82, P < 0.001). In all the subcategories, the mean score of patients with DFUs was statistically significantly higher than that of the patients whose wounds had healed, which ensured known‐groups validity.
Table 5.
Comparison of DFS‐SF‐K scores between current and healed DFUs
| DFS‐SF‐K | Groups | t | P | |
|---|---|---|---|---|
| Current DFU(n = 189) | Healed DFU (n = 131) | |||
| M ± SD | M ± SD | |||
| L | 28.892 ± 30.30 | 62.82 ± 39.58 | 8.27 | <0.001 |
| PH | 39.26 ± 33.60 | 78.66 ± 30.53 | 10.89 | <0.001 |
| D/DL | 31.67 ± 24.21 | 61.03 ± 32.53 | 8.78 | <0.001 |
| WU/F | 28.26 ± 21.27 | 70.99 ± 29.16 | 14.34 | <0.001 |
| NE | 37.26 ± 38.84 | 77.54 ± 41.10 | 8.82 | <0.001 |
| BUC | 30.60 ± 32.57 | 70.10 ± 40.34 | 9.30 | <0.001 |
| DFS‐SF‐K | 195.96 ± 100.55 | 421.16 ± 91.24 | 20.82 | <0.001 |
Abbreviations: BUC, bothered by ulcer care; D/DL, dependence/daily life; L, leisure; NE, negative emotion; PH, physical health; WU/F, worried about ulcers/feet.
3.2.3. Criterion validity
To verify criterion validity, correlations between DFS‐SF‐K and SF‐36 were analysed (Table 6). The results showed that the overall DFS‐SF‐K had a positive correlation with the overall SF‐36 (r = 0.68, P < 0.001), and each of the subcategories also showed a positive correlation (r = 0.20–0.54, P < 0.001).
Table 6.
Correlations of DFS‐SF‐K and SF‐36
| Measures | DFS‐SF‐K | SF‐36 | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| L | PH | D/DL | WU/F | NE | BUC | PF | RP | BP | GH | VT | SF | RE | MH | ||
| D | L | 1.00 | 0.44** | 0.38** | 0.43** | 0.47** | 0.53** | 0.50** | 0.46** | 0.44** | |||||
| F | PH | 0.29** | 1.00 | * | 0.30** | 0.27** | 0.30** | 0.33** | 0.25** | 0.24** | 0.26** | 0.28** | |||
| S | D/DL | 0.21** | 0.34** | 1.00 | 0.31** | 0.24** | 0.31** | 0.33** | 0.35** | 0.27** | 0.27** | 0.35** | |||
| ı | WU/F | 0.34** | 0.29** | 0.48** | 1.00 | 0.45** | 0.38** | 0.49** | 0.53** | 0.37** | 0.54** | 0.43** | 0.52** | ||
| S | NE | 0.03 | 0.13** | 0.38** | 0.58** | 1.00 | 0.31** | 0.36** | 0.34** | 0.38** | 0.20** | 0.37** | 0.39** | 0.41** | |
| F | BUC | 0.11 | 0.23** | 0.28** | 0.45** | 0.44** | 1.00 | 0.35** | 0.36** | 0.40** | 0.37** | 0.30** | 0.31** | 0.34** | 0.41** |
| ı | |||||||||||||||
| K | |||||||||||||||
Abbreviations: BP, bodily pain; BUC, bothered by ulcer care; D/DL, dependence/daily life; GH, general health; L, leisure; MH, mental health; NE, negative emotion; PF, physical functioning; PH, physical health; RE, role emotional; RP, role physical; SF, social functioning; VT, vitality; WU/F, worried about ulcers/feet.
P < 0.01.
3.3. Reliability of the DFS‐SF‐K
Cronbach's alpha for the reliability of all the 29 questions was 0.95, while that for each of the subcategories ranged from 0.96 to 0.99 (Table 7).
Table 7.
Reliability for DFS‐SF‐K
| Dimensions | Items | Corrected item‐total correlation | Cronbach's α |
|---|---|---|---|
| L | Q1A | 0.97 | 0.98 |
| Q1B | 0.97 | ||
| Q1C | 0.98 | ||
| Q1D | 0.97 | ||
| Q1E | 0.97 | ||
| PH | Q2A | 0.99 | 0.99 |
| Q2B | 0.99 | ||
| Q2C | 0.98 | ||
| Q2D | 0.99 | ||
| Q2E | 0.98 | ||
| D/DL | Q3A | 0.97 | 0.98 |
| Q3B | 0.97 | ||
| Q3C | 0.97 | ||
| Q3D | 0.97 | ||
| Q3E | 0.97 | ||
| WU/F | Q4D | 0.95 | 0.96 |
| Q4E | 0.95 | ||
| Q4F | 0.95 | ||
| Q4H | 0.94 | ||
| NE | Q4A | 0.98 | 0.99 |
| Q4B | 0.98 | ||
| Q4C | 0.98 | ||
| Q4B | 0.98 | ||
| Q4I | 0.98 | ||
| Q4J | 0.98 | ||
| BUC | Q5A | 0.95 | 0.96 |
| Q5B | 0.94 | ||
| Q5C | 0.94 | ||
| Q5D | 0.95 | ||
| Total | 0.95 | ||
Abbreviations: BUC, bothered by ulcer care; D/DL, dependence/daily life; L, leisure; NE, negative emotion; PH, physical health; WU/F, worried about ulcers/feet.
4. DISCUSSION
This study was conducted to adapt the DFS‐SF—which is widely used worldwide—for the Korean population and to examine its validity and reliability and thus contribute to the expansion of studies on the QoL of patients with DFUs. The strengths of the DFS‐SF are that it can measure disease‐specific items in terms of QoL as a tool tailored to patients with DFUs, it saves time in acquiring information owing to its shorter form, and it reduces the burden on patients. The saved time helps to better identify patients by allowing the use of multiple instruments, based on which various fields could assess patients and provide interventions.
This study had experts evaluate correlations between questions for the content validity of the DFS‐SF‐K and then calculated the CVI, which shows the degree of agreement among experts. An instrument is considered to have no content validity if the CVI is ≤0.5 and to have high content validity if the CVI is ≥0.8.26 In this study, the mean CVI was high at 0.98, confirming that the DFS‐SF‐K consisted of the valid contents to measure the QoL of patients with DFU.
When a tool is developed based on a theory, and the factor structure has already been confirmed in a factor analysis intended to verify construct validity, a CFA is more appropriate to use to verify the validity of the tool than an explanatory factor analysis.27 All the studies on the adaptation of the DFS‐SF, however, present the limitation that CFA alone is not enough to verify validity.28 Hence, this study performed both an explanatory factor analysis and a CFA.
By using an explanatory factor analysis, the constructs of the DFS‐SF‐K were examined, and the results showed that the scale was made up of the contents that measured six constructs.
CFA is used more often than explanatory factor analysis as an important approach to verify construct validity. While explanatory factor analysis analyses factors divided into a set of items, CFA verifies the hypothesis that each item belongs to a specific factor.27 This study examined the factor structure of the DFS‐SF‐K by using CFA, and the results indicated that six‐factor models were the most adequate.
A GFI should not be sensitively affected by the sample size and should prefer a more concise model that fits well with data.29 The relative GFI is an index that shows how well a theoretical model explains data in comparison with the worst model and includes NFI and CFI. The non‐normed fit index (NNFI) is one of the oldest fit indices. If a perfect model is 10 compared with the base model, when a theoretical model is as good as 9 compared with the base model, the NNFI shows a value of 0.9. If it is ≥0.9, the goodness of fit is considered good enough. The NNFI is not markedly affected by the size of a sample, and it is regarded as a good fit index because its value increases as fewer model errors occur and is more concise. On the other hand, as the errors in the NFI are reduced as the sample size increases, some scholars argue that, if possible, it should not be used.29
In this regard, the NFI was 0.90 in this study, and the NNFI (TLI), a more important relative index, was 0.91. Therefore, the relative fit of the model is regarded as good. The absolute fit index does not assess the goodness of fit of a theoretical model in a relative way by comparing it with those of other models. Rather, it assesses the theoretical model in an absolute way by determining how well the model fits the data.29 The index includes GFI and RMSEA. The GFI should be at least ≥0.70. A value of 0.90 means that the goodness of fit of the model is optimal. Meanwhile, the RMSEA is not affected by the sample size and considers model errors and conciseness at the same time. Although its lower limit is 0, it does not have an upper limit, and a lower number means better goodness of fit. The goodness of fit is good if the RMSEA is 0.50, decent if it is <0.80, and normal if it is ≤0.10.29 In this study, the GFI was 0.73, while the RMSEA was 0.10. Although they did not reach their optimal levels, they were found to be acceptable and could be regarded as relatively fit. The root mean square residual (RMR) does not have an absolute baseline for the level of acceptance, but it is considered more fit as it gets closer to 0.26 In this study, the RMR was 0.06, and its goodness of fit could be considered relatively good.
To ensure the validity of CFA, this study examined correlations between factors and measured variables using standardised factor loadings, CR, and AVE. The results showed that the measured items had consistency in measuring constructs, and the independence between the factors was maintained. Accordingly, the convergent and discriminative validities of each factor in the DFS‐SF‐K were verified. These results suggest that the DFS‐SF‐K is a tool that measures the constructs of the L, PH, D/DL, NE, WU/F, and BUC of patients with diabetic foot well.
The known‐groups technique, another method to verify construct validity, compares the results of the groups that are expected to have differences in constructs because of theoretically well‐known attributes and verifies whether the theory shows expected differences.26 This is based on the hypothesis of previous studies that patients with diabetic foot and wounds would score lower than the patients whose wounds have healed.3, 10, 30 When this study compared the DFS‐SF‐K scores of the patient group that currently had ulcers and the patient group that had healed ulcers, the mean score of the group with ulcers was statistically significantly higher, and the results were the same in all the subcategories. These results support the finding of other studies that foot ulcers would greatly affect the QoL of diabetic patients.10, 31
SF‐36, which was used as a criterion tool to verify criterion validity, showed correlation coefficients ranging from 0.20 to 0.54 with the subcategories of the DFS‐SF‐K. As such, SF‐36 showed that all the correlations were statistically significant, thereby proving criterion validity. Its results were similar to those of other studies where the DFS‐SF was translated to other languages.
In relation to reliability verification, reliability coefficients of 0.90, 0.80, 0.70, and ≤0.50 are generally interpreted as “highly reliable,” “good,” “moderate,” and “not acceptable for use,” respectively.32 In this study, Cronbach's α of the DFS‐SF‐K ranged from 0.96 to 0.99, while the mean reliability coefficient was high at 0.95. The DFS‐SF‐K is considered to have a high internal consistency as a tool to measure the QoL of patients with diabetic foot. Given that the reliability coefficient of the original English tool was 0.93, that of the translated Chinese tool was 0.90, and that of the adapted Greek tool was 0.94, the DFS‐SF‐K shows a higher level of reliability than the tools reported in other similar studies.17, 19, 20, 21
As demonstrated earlier, the DFS‐SF‐K, whose validity and reliability were verified in this study, is composed of the same questions as the original tool, without any questions added or removed. As such, the scale could be useful for comparing the QoL of Korean patients with diabetic foot with that of patients in other countries. Furthermore, various types of validity were examined to verify and enhance the validity of the DFS‐SF‐K. This study not only performed a CFA, which was not performed in previous studies, but also explained fit indices in a CFA in various ways. Therefore, this study is expected to serve as a basic reference for developing and adapting the DFS‐SF in a new language and to help nursing researchers who will attempt to develop a tool with a high level of validity in the future.
In addition, as the validity and reliability of each of the subcategories were verified, this study will contribute to understanding the QoL of individual patients by determining the score in each subcategory and identifying which subcategory is relatively insufficient for Korean patients with diabetic foot.
This study has the following limitations: First, as this study was conducted among patients in a single general hospital, it could not include patients in more diverse environments. Second, the participants of this study were selected based on convenient extraction, so the study's limitation is that it does not represent the target population. This would be a limitation to generalising the results of this study. Nonetheless, the validity and reliability of the DFS‐SF‐K were verified as adequate in this study. Thus, the results of this study could help to understand the QoL of Korean patients with DFUs in the future and be actively used in related studies.
5. CONCLUSION
By translating the DFS‐SF, the tool developed by Bann et al,17 to assess the QoL of patients with diabetic foot, in accordance with the circumstances of Korean patients, and verifying its validity, reliability, and sensitivity, this study was conducted to allow other studies to assess the QoL of Korean patients with diabetic foot in the future. After translation, the validity and reliability of the DFS‐SF‐K were verified, and the results showed that its validity and reliability for Korean patients with diabetic foot were high and adequate, respectively. Accordingly, the DFS‐SF‐K could be used in the future to develop and study specific interventions to improve the QoL of Korean patients with diabetic foot and would ultimately contribute to the adaptation of patients with diabetic foot.
CONFLICTS OF INTEREST
The authors have no conflicts of interest to report.
Lee Y‐N. Translation and validation of the Korean version of the Diabetic Foot Ulcer Scale‐Short Form. Int Wound J. 2019;16(Suppl. 1):3–12. 10.1111/iwj.13025
REFERENCES
- 1. Task Force Team for Basic Statistical Study of Korean Diabetes Mellitus . Report of task force team for basic statistical study of Korean diabetes mellitus: diabetes in Korea 2007. Seoul, South Korea: Korean Diabetes Association; 2008. [Google Scholar]
- 2. Park IB, Baik SH. Epidemiologic characteristics of diabetes mellitus in Korea: current status of diabetic patients using Korean health insurance database. Korean Diabetes J. 2009;33:357‐362. [Google Scholar]
- 3. Goodridge D, Trepman E, Embil JM. Health‐related quality of life in diabetic patients with foot ulcers: literature review. J Wound Ostomy Continence Nurs. 2005;32:368‐377. [DOI] [PubMed] [Google Scholar]
- 4. Brod M. Pilot study—quality of life issues in patients with diabetes and lower extremity ulcers: patients and care givers. Qual Life Res. 1998;7:365‐372. [DOI] [PubMed] [Google Scholar]
- 5. Mazlina M, Shamsul A, Jeffery F. Health‐related quality of life in patients with diabetic foot problems in Malaysia. Med J Malaysia. 2011;66:234‐238. [PubMed] [Google Scholar]
- 6. Yekta Z, Pourali R, Ghasemi‐Rad M. Comparison of demographic and clinical characteristics influencing health‐related quality of life in patients with diabetic foot ulcers and those without foot ulcers. Diabetes Metab Syndr Obes. 2011;4:393‐399. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Raspovic KM, Wukich DK. Self‐reported quality of life and diabetic foot infections. J Foot Ankle Surg. 2014;53:716‐719. [DOI] [PubMed] [Google Scholar]
- 8. Ribu L, Hanestad BR, Moum T, Birkeland K, Rustoen T. A comparison of the health‐related quality of life in patients with diabetic foot ulcers, with a diabetes group and a nondiabetes group from the general population. Qual Life Res. 2007;16:179‐189. [DOI] [PubMed] [Google Scholar]
- 9. Garcia‐Morales E, Lazaro‐Martinez J, Martinez‐Hernandez D, Aragón‐Sánchez J, Beneit‐Montesinos J, González‐Jurado M. Impact of diabetic foot related complications on the health related quality of life (HRQol) of patients—a regional study in Spain. Int J Low Extrem Wounds. 2011;10:6‐11. [DOI] [PubMed] [Google Scholar]
- 10. Goodridge D, Trepman E, Sloan J, et al. Quality of life of adults with unhealed and healed diabetic foot ulcers. Foot Ankle Int. 2006;27:274‐280. [DOI] [PubMed] [Google Scholar]
- 11. Fejfarová V, Jirkovská A, Dragomirecká E, et al. Does the diabetic foot have a significant impact on selected psychological or social characteristics of patients with diabetes mellitus? J Diabetes Res. 2014;2014:371938. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Ribu L, Wahl A. Living with diabetic foot ulcers: a life of fear, restrictions, and pain. Ostomy Wound Manage. 2004;50:57‐67. [PubMed] [Google Scholar]
- 13. Ribu L, Hanestad BR, Moum T, Birkeland K, Rustoen T. Health‐related quality of life among patients with diabetes and foot ulcers: association with demographic and clinical characteristics. J Diabetes Complications. 2007;21:227‐236. [DOI] [PubMed] [Google Scholar]
- 14. Siersma V, Thorsen H, Holstein PE, et al. Health‐related quality of life predicts major amputation and death, but not healing, in people with diabetes presenting with foot ulcers: the Eurodiale study. Diabetes Care. 2014;37:694‐700. [DOI] [PubMed] [Google Scholar]
- 15. Hogg F, Peach G, Price P, Thompson M, Hinchliffe R. Measures of health‐related quality of life in diabetes‐related foot disease: a systematic review. Diabetologia. 2012;55:552‐565. [DOI] [PubMed] [Google Scholar]
- 16. Abetz L, Sutton M, Brady L, McNulty P, Gagnon DD. The diabetic foot ulcer scale (DFS): a quality of life instrument for use in clinical trials. Prac Diabetes Int. 2002;19:167‐175. [Google Scholar]
- 17. Bann CM, Fehnel SE, Gagnon DD. Development and validation of the diabetic foot ulcer scale‐short form (DFS‐SF). Pharmacoeconomics. 2003;21:1277‐1290. [DOI] [PubMed] [Google Scholar]
- 18. Vileikyte L, Peyrot M, Bundy C, et al. The development and validation of a neuropathy‐ and foot ulcer‐specific quality of life instrument. Diabetes Care. 2003;26:2549‐2555. [DOI] [PubMed] [Google Scholar]
- 19. Hui LF, Yee‐Tak Fong D, Yam M, Yuk Ip W. Translation and validation of the Chinese diabetic foot ulcer scale‐short form. Patient. 2008;1:137‐145. [DOI] [PubMed] [Google Scholar]
- 20. Kontodimopoulos N, Veniou A, Tentolouris N, Niakas D. Validity and reliability of the Greek version of the diabetic foot ulcer scale—short form (DFS‐SF). Hormones (Athens). 2016;15:394‐403. [DOI] [PubMed] [Google Scholar]
- 21. Sobol E, Macioch T, Krakowiecki A, Mrozikiewicz‐Rakowska B, Kasprowicz M, Hermanowski TR. Translation and validation of the polish diabetic foot ulcer scale ‐ short form (Dfs‐sf). Value Health. 2015;18:A615‐A616. [Google Scholar]
- 22. Ware JE Jr, Sherbourne CD. The MOS 36‐item short‐form health survey (SF‐36): I. conceptual framework and item selection. Med Care. 1992;30:473‐483. [PubMed] [Google Scholar]
- 23. Ware JE, Kosinski M, Dewey JE, Gandek B. SF‐36 Health Survey: Manual and Interpretation Guide. Lincoln, RI: Quality Metric; 2000. [Google Scholar]
- 24. Koh SB, Chang SJ, Kang MG, Cha BS, Park JK. Reliability and validity on measurement instrument for health status assessment in occupational workers. Korean J Prev Med. 1997;30:251‐266. [Google Scholar]
- 25. World Health Organization . Process of Translation and Adaptation of Instruments. Geneva, Switzerland: WHO; 2009. [Google Scholar]
- 26. Lee E, Lim N, Park H, et al. Nursing Research and Statistical Analysis. Seoul, South Korea: Soomoonsa; 2009. 207 p. [Google Scholar]
- 27. Harrington D. Confirmatory Factor Analysis. Oxford, UK: Oxford University Press; 2009. [Google Scholar]
- 28. Kyunghee L, Sujin S. Validity of instrument development research in Korean nursing research. J Korean Acad Nurs. 2013;43:697‐703. [DOI] [PubMed] [Google Scholar]
- 29. Hong S. The criteria for selecting appropriate fit indices in structural equation modeling and their rationales. Korean J Clin Psychol. 2000;19:161‐177. [Google Scholar]
- 30. Vileikyte L. Diabetic foot ulcers: a quality of life issue. Diabetes Metab Res Rev. 2001;17:246‐249. [DOI] [PubMed] [Google Scholar]
- 31. Spanos K, Saleptsis V, Athanasoulas A, et al. Factors associated with ulcer healing and quality of life in patients with diabetic foot ulcer. Angiology. 2016;68:242‐250. [DOI] [PubMed] [Google Scholar]
- 32. Bae B. Structural Equation Modeling with Amos 19: Principles and Practice. Seoul, South Korea: Chungram Books; 2011. [Google Scholar]