European Journal of Oncology Nursing 18 (2014) 52e57
Contents lists available at ScienceDirect
European Journal of Oncology Nursing journal homepage: www.elsevier.com/locate/ejon
Psychometric properties of the Persian version of the Mishel’s Uncertainty in Illness Scale in Patients with Cancer Moosa Sajjadi a, b, Maryam Rassouli a, c, *, Abbas Abbaszadeh d, Hamid Alavi Majd e, Kazem Zendehdel f a
Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Tehran, Iran Gonabad University of Medical Sciences, Gonabad, Iran Pediatric Nursing Department, Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Vali-e Asr Street, Niyayesh Cross, Tehran, Iran d Department of Nursing, Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Tehran, Iran e Department of Biostatistics, School of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran f Cancer Research Center, Cancer Institute of Iran, Tehran University of Medical Sciences, Tehran, Iran b c
a b s t r a c t Keywords: Uncertainty in illness MUIS-A Psychometrics Cancer Confirmatory factor analysis
Purpose: Uncertainty is a major component in the illness experiences which extraordinarily can affect the psychological adjustment and the illness outcomes. Uncertainty in illness is defined as inability to define the illness-related events to the illness or disability in predicting the illness outcomes. The present study aimed to translate the Persian version of Uncertainty in Illness Scale (MUIS-A) and to investigate its psychometric properties on patients with cancer. Method: In this methodological study, validation of the Persian version of MUIS-A was performed in Iran on 420 cancer patients attending two major hospitals in Tehran, Iran. The scale was translated into Persian and back translated into English and revised according to editorial comments of the scale designers. Then, content and face validity, construct validity, internal consistency reliability and stability of the Persian version were measured. Data were analyzed using SPSS version 16 and LISREL 8.5. Results: Mean of the participants MUIS-A score was 90.1 (16.8). Confirmatory factor analysis confirmed validity of the whole instrument and its four subscales. The consistency of the instrument with a threeweek interval was r ¼ 0.91. Cronbach’s alpha was 0.89 for the whole scale of 32 MUIS-A items and a ¼ 0.58e0.86 for its four factors. Conclusions: The Persian version of the MUIS-A has good psychometric properties. It can be used to assess uncertainty in illness in Iranian patients with cancer. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Uncertainty is a natural part in the illness experience and can occur at all stages of disease including diagnosis, treatment and prognosis of the disease. Uncertainty in illness occurs when the patient is unable to determine the meaning of illness-related events; therefore, it is considered a major cause of psychological stress to the patients (Mishel, 1997a, 2013), that can leave major impacts on the psychological adaptation and outcomes of the disease (Neville, 2003). Uncertainty in illness is defined as the “inability to determine the meaning of illness-related events when * Corresponding author. Pediatric Nursing Department, Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Vali-e Asr Street, Niyayesh Cross, Tehran, Iran. Tel.: þ98 21 88655372; fax: þ98 21 88202521. E-mail address: [email protected] (M. Rassouli). 1462-3889/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejon.2013.09.006
the patient cannot determine the value of events or cannot accurately predict the disease outcome due to the lack of sufficient cues” (Mishel, 1988). Uncertainty occurs when the individuals cannot form a cognitive framework for understanding their status and interpreting illness related events (Bailey et al., 2011; Mishel, 1990, 2013). Therefore, in most cases, it is accepted as a major stressor that most people seek to reduce and to learn methods to cope with (Neville, 2003). Chronic diseases are a long term and intense source of psychological stress that influences all aspects of patients’ life and reduces their daily activities and quality of life (Flemme et al., 2005). Furthermore, various studies have been conducted on chronic patients such as patients with cardiovascular diseases, dialysis, hepatitis, liver transplantation, AIDS and Alzheimer’s whose results indicate that uncertainty exists in these patients and can lead to reduction in quality of life and the ability to cope with
M. Sajjadi et al. / European Journal of Oncology Nursing 18 (2014) 52e57
illness (Bailey et al., 2009; Brashers et al., 2004; Flemme et al., 2005; Lasker et al., 2010; Madar and Bar-Tal, 2009; Mauro, 2010; Sodowsky, 2012). Cancer, as one of the chronic diseases, is the second leading cause of death in the West (Shaha et al., 2008). It is the third cause of death in Iran after heart diseases and car accidents. Based on the recent statistics, the rate of cancer incidence in Iran is about 107 people in 100,000 people (more than 80,000 people considering Iran’s 75million population). It is predicted that with regard to the raise of environmental pollution, increasing the elderly (aging community) and population growth, coming decades will witness an increasing trend in cancer so that it becomes a principle health problem in Iran (Rasaf et al., 2012). Today, advances in cancer diagnosis and treatment technology has led to an increase in patients’ survival, and one of the important points in the care of these patients is to notice and improve the quality of life (Stewart et al., 2010b). From the diagnosis time to the onset of treatment, patients with cancer suffer from high levels of emotional tension (Chen et al., 2010). Due to the complexities of treatment and symptoms, patients experience some levels of uncertainty about illness which originates from problems related to the inability to perform daily activities, inadequate treatment, and concerns about the relapse of the disease (Haisfield-Wolfe et al., 2012). Various studies have been conducted in the field of uncertainty in illness in different types of cancer in other countries (mainly Western countries) (Detprapon et al., 2013; Haisfield-Wolfe et al., 2012; Harrow et al., 2008; Kurita et al., 2013; Parker et al., 2013; Shaha et al., 2008; Sherman and Simonton, 2010; Stewart et al., 2010a; Woodgate and Degner, 2002). The ultimate goal of these studies is to help the patients (or their families) to effectively manage uncertainty so that its negative effect on the patient’s psychological adaptation is reduced (Stewart et al., 2010b). Nurses are frequently in contact with the patient, thus they are in the best position to reduce uncertainty in patients by providing information and promoting patients’ understanding of health (Madar and Bar-Tal, 2009). In order to conduct interventions to reduce uncertainty in patients with cancer and to measure effectiveness of these interventions, a scale is needed to measure it. There are various scales for uncertainty assessment in different patients and groups (Lin et al., 2012; Mishel, 1997b; Pai et al., 2007; Stewart et al., 2010a). One of the convenient and widely used scales is Mishel Uncertainty in Illness Scale-Adult form (MUIS-A). MUIS-A is a valid and reliable scale and has been used in several studies. This instrument contains 32 items on Likert scale from 1 (strongly disagree) to 5 (strongly agree) points and occurs in four dimensions of ambiguity, complexity, inconsistency, and unpredictability. The instrument scores between 32 and 160, and earning more points shows greater uncertainty (Bailey et al., 2009; Mishel, 1997b). MUIS-A has a good internal consistency with Coronbach’s alpha of 0.87 for the whole instrument, 0.86 for the subscales of ambiguity (13 items), 0.81 for complexity (7 items), 0.78 for inconsistency (7 items), and 0.65 for unpredictability (5 items). This scale has been translated into different languages and its validity and reliability have been studied (Mishel, 1997b). Despite the importance of the concept of uncertainty in illness in cancer patients, no valid and reliable tool exists in this regard in Iran. Therefore, this study was conducted with the aim of translation and psychometric properties of the Persian Version of “Mishel Uncertainty in Illness Scale” in cancer patients. Methods Procedures The present study is a methodological research (LoBiondoWood et al., 2006) through which the MUIS-A scale is translated
53
and its Persian version has been validated among patients with cancer. The study population was all the cancer patients attending the clinics and oncology wards of Imam Khomeini Hospital (Iran’s largest cancer center) and Taleghani Hospital in Tehran. Given these centers accept patients from all over the country, results of the study can be better generalized to the society. Convenient sampling was performed in order to achieve a sufficient sample size to conduct a confirmatory factor analysis. The inclusion criteria were willingness to participate in research, cancer diagnosis by the oncologist, being aware of one’s illness, being Iranian, the ability to understand and speak Persian, the minimum age of 21 years, and absence of other serious illnesses and no history of serious mental disorders (such as schizophrenia). Type and stage of cancer and treatment type were not the inclusion criteria of this study. Based on these criteria, eventually 420 individuals were chosen. Scale translation In this study, translation and validation of the instrument was performed based on the suggested model by Wild et al. (2005). After obtaining written permission from the original designer of the scale, the original scale was translated into Persian by two people fluent in English and Persian. Then the two translations were compared and the final version was prepared with a few modifications. Then, the final translated version was given to two people fluent in both languages of English and Persian (one native English speaker), who were not in contact with the first people to translate it back into English. Then, the back-translation was contrasted by a supervisor with translations of the original scale in terms of similarity of translations, and some minor revisions were made in statements of the Persian version. Afterward, validation process of the translated scale was performed by assessing face and content validity, construct validity (using confirmatory factor analysis), reliability, and internal consistency. Content validity In order to examine the content validity index (CVI), the translated scale was given to ten people (a specialist in clinical psychology, an oncologist, two psychiatric nurses, two oncology nursing instructors, four nursing assistant professors with experience in instrument development) to review that and provide their correctional comments. They also investigated and affirmed face validity of the translated scale. Then, the scale was given to 10 patients with cancer to express their idea about simplicity of use, and understandability of words and phrases. The Persian version was finalized without much change in its sentences. Data collection To collect the data, after choosing eligible patients and explaining the study objectives and methods, the informed consent was obtained from them. Next, the questionnaire including demographics (age, sex, educational level, marital status, occupation, place of residence, type of illness, type of treatment, time of diagnosis) and MUIS-A scale were given to them to complete. Completion of the questionnaire took about 20e25 min. For illiterate people, the questionnaire was read and patients’ answers were marked by the researcher. Investigating validity of the MUIS-A construct and the model fitness was conducted by confirmatory factor analysis (CFA) using LISREL statistical software version 8.5.
54
M. Sajjadi et al. / European Journal of Oncology Nursing 18 (2014) 52e57
Construct validity CFA is a structural equation modeling technique which is used to determine the goodness of fit between a hypothesized model and the data obtained from study samples (Kline, 2010). Maximum likelihood algorithm was used to evaluate the fitness of the model. There are several fit indices for deciding on the goodness of fit of the model and it is recommended that different indices be used (Brown, 2006; Seo et al., 2004). In this study, indices of fit chisquare, root mean error of approximation (RMSEA), goodness-offit index (GFI), comparative fit index (CFI), adjusted goodness-of-fit index (AGFI), and standardized root mean square residual (SRMR) were used. One of the most used indexes is chi-square. Because this index is related to the sample size, ratio of chi-square to degrees of freedom is used, where values less than 2 indicate a good fit of the model. Another important index is the RMSEA, where values less than 0.08 are acceptable and less than 0.05 indicates good fit of the model. Researchers are advised to report the confidence interval (Kline, 2010; Seo et al., 2004). Another important index is SRMR, whose acceptable value is 0.06 or less (Brown, 2006; Kline, 2010). Appropriate values are >0.9 for CFI and GFI, and >0.85 AGFI (Bentler, 1990; Helsen et al., 2013).
Reliability Internal consistency and reliability were measured using SPSS software version 16. Scale internal consistency was measured by Cronbach’s alpha for the total scale and each subscale separately.
The best method used to evaluate the internal consistency is Cronbach’s alpha (Polit and Beck, 2013). The Cronbach’s alpha of about 0.7 is sufficient, and of >0.80 indicates high internal consistency of the tool (Polit and Beck, 2013). The reliability was measured by Pearson’s or Spearman’s correlation coefficients in testeretest on 15 patients with a three-week interval. Usually, the time interval of 2e3 weeks is appropriate between two tests (Hawthorne et al., 2011; Polit and Beck, 2013).
Results A total of 420 questionnaires were analyzed. Demographic characteristics of the study sample are shown in Table 1. Content validity index was measured on relevance, clarity, and simplicity of the translated scale as 0.98, 0.95, and 0.96, respectively. Table 2 shows mean and standard deviation of the overall scores of MUIS-A and its dimensions. In order to have a better comparison between the present scale and the original one, data of the present study were compared to the combined multiple studies on patients with cancer (Mishel, 1997b). The results show that mean of the overall score of uncertainty in illness in the present study (90.1) is greater than that of the combined sample of cancer patients proposed by Mishel (79.5). Coefficient of Cronbach’s alpha was 0.89 for the overall scale, and 0.58e0.86 for its dimensions (Table 2). The correlation coefficient testeretest reliability for MUIS-A was 0.91 (p < 0.001). Correlation coefficients for subscales of ambiguity, complexity, inconsistency, and unpredictability were respectively 0.90, 0.74, 0.83, and 0.60.
Table 1 Demographic and clinical characteristics of patients with cancer (n ¼ 420).
Age (year) Time since diagnosis (months)
Sex Marital status
Education level
Occupation
Time since diagnosis (months)
Type of cancer
Metastasis Kind of treatment
Female Male Single Married Widowed or divorced Illiterate Primary and Guidance school Diploma University degree Self-employed Employee Housewife Unemployed Other 24 Breast Colon and rectum Gastric/esophagus Hematologic Uterus/ovarian Prostate Lung Skin Other Yes No Chemotherapy Chemotherapy and surgery Chemo and radiotherapy All of them Other
Mean
SD (range)
46.4 18.5
13.9 (21e79) 24.2 (3e168)
N
%
217 203 58 331 31 87 151 107 75 76 34 174 35 103 166 163 91 103 99 36 72 25 5 21 5 54 143 251 105 189 26 90 10
51.7 48.3 13.8 78.8 7.3 20.7 36 25.5 17.9 18.1 8.1 41.4 8.3 24.1 39.5 38.8 21.7 24.5 23.6 8.6 17.1 6 1.2 5 1.2 12.9 36.3 63.7 25 45 6.2 21.4 2.4
M. Sajjadi et al. / European Journal of Oncology Nursing 18 (2014) 52e57
55
Table 2 Scores for MUIS-A scale and its dimensions in Iranian patients with cancer and it’s comparison with Mishel’s data bank.
Iranian patients with cancer (N ¼ 420) Mishel’s combined data for cancer patients (N ¼ 761)
Ambiguity mean (SD) e a
Complexity mean (SD) e a
Inconsistency mean (SD) e a
Unpredictability mean (SD) e a
Total mean (SD) e a
41.8 (9.4)e0.86
15.1 (3.2)e0.58
18.8 (4.2)e0.71
14.4 (3.5)e0.65
90.1 (16.8)e0.89
32.3 (9.3)e0.87
16.5 (5)e0.80
14.9 (4.7)e0.80
15.8 (3.7)e0.67
79.5 (16.9)e0.90
Construct validity To investigate construct validity and provide the most appropriate model for the MUIS-A, confirmatory factor analysis was used. To this end, full samples of cases were randomly split into two approximately equal groups. 200 subjects were considered as calibration sample and 220 subjects were considered as evaluation sample. Calibration samples were used to check the models (2 and 4 factors) and do possible revisions. Evaluation sample was used to validate the appropriate model. Two-dimensional model of the scale did not have good fit indices, but the four-dimensional model had better fit indices. The chi-square, RMSEA, SRMR and AGFI indices confirm the fit indices of the model (Table 3). By assessing the scale items, it was determined that the item 32 “The seriousness of my illness has been determined” had a weak item-scale correlation coefficient (r ¼ 0.02), therefore, item 32 was removed. According to the suggestion of modification of fit indices in confirmatory factor analysis, transferring item 8 “I do not know when to expect things will be done to me” (from the ambiguity to the inconsistency) and item 28 “treatment which is being done to me has a known probability of success” (from the complexity to the unpredictability) improved the fit of the model. Therefore, the fourdimensional model of MUIS-A with 31 items (ambiguity 12 items, complexity 5 items, inconsistency 8 items and unpredictability 6 items) is a model with the best fit indices. CFI and GFI indices are acceptable (Bentler, 1990) and other indices such as the ratio of chisquare to degree of freedom, RMSEA, AGFI and SRMR indicate a good fit of the model (Brown, 2006; Helsen et al., 2013; Kline, 2010). In order to further investigate validity of the construct, the relationship between education level and uncertainty in illness rate was used. In this study, a strong inverse correlation was found between uncertainty and the level of education (r ¼ 0.47, p < 0.001) in the sense that patients with higher level of education had less uncertainty. Discussion Uncertainty in illness is a concept that can have broad impact on cancer patients and its investigation in this group of patients is of
great importance. Despite the importance of the issue, this problem has not been studied in Iran that can be attributed to lack of proper tools in Iran. Therefore, this study was conducted with the aim of translating and validating the Persian version of the uncertainty in illness scale in patients with cancer. In this study, the MUIS-A scale was translated into Persian and its psychometric properties and factor structure were evaluated. When a tool is translated or used in another culture or society, it is essential to evaluate its psychometric properties because that may not be appropriate in the new society (Hillen et al., 2013; Michaeli Manee, 2011). In this study, content validity, construct validity, and reliability of this tool were assessed. Based on the findings, the Persian version of MUIS-A generally has good reliability and validity and its dimensions’ construct is similar to that of the original one (Mishel, 1997b). In this study, it has been shown that with regard to content validity index of 0.96, the Persian version of MUIS-A has a good content validity. Polit and Beck (2013) suggest that the CVI value of 0.9 or more should be considered as a standard for content validity of the scales (Polit and Beck, 2013). Confirmatory factor analysis of MUIS-A was conducted using a sample of 420 cancer patients. Goodness-of-fit indices indicated that the original model (32-item and 4-dimensional) of Persian version of the uncertainty in illness scale is appropriate, but for better fit indices and to enhance the fit indices some changes are needed to be made in the original scale. In confirmatory factor analysis, some suggestions are given by Lisrel to modify and improve the model fit indices. If they are justifiable logically and theoretically, they can be applied (Kline, 2010). In the Persian version, item 32 “The seriousness of my illness has been determined” from complexity dimension was deleted due to its low correlation coefficient of item-scale. In validation of the Swedish version of this scale for the cardiac patients, it was indicated that this item has no significant correlation with the overall score of the scale and complexity subscale. Furthermore, items 10 and 22 were excluded from the Swedish version (Hallberg and Erlandsson, 1991). Another study was conducted in Iran to validate the scale of uncertainty in illness e family form. This scale assesses uncertainty from patients’ family’s perspective. The original version of this scale has 31 items and is
Table 3 Confirmatory Factor analysis Fit indices for the different MUIS-A version in Iranian patients with cancer.
c2 (df), p MUIS-A, MUIS-A, MUIS-A, MUIS-A, a b c d e g f
2 4 4 4
factorsa factorsa (original) factorsa (modified) factorsb (modified)
657.4 767.4 660.7 652.3
Calibration sample (N ¼ 200). Validation sample (n ¼ 220). Root mean error of approximation. Comparative fit index. Goodness-of-fit index. Standardized root mean square residual. Adjusted goodness-of-fit index.
(349), (458), (428), (428),
p p p p
< < <
Contents lists available at ScienceDirect
European Journal of Oncology Nursing journal homepage: www.elsevier.com/locate/ejon
Psychometric properties of the Persian version of the Mishel’s Uncertainty in Illness Scale in Patients with Cancer Moosa Sajjadi a, b, Maryam Rassouli a, c, *, Abbas Abbaszadeh d, Hamid Alavi Majd e, Kazem Zendehdel f a
Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Tehran, Iran Gonabad University of Medical Sciences, Gonabad, Iran Pediatric Nursing Department, Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Vali-e Asr Street, Niyayesh Cross, Tehran, Iran d Department of Nursing, Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Tehran, Iran e Department of Biostatistics, School of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran f Cancer Research Center, Cancer Institute of Iran, Tehran University of Medical Sciences, Tehran, Iran b c
a b s t r a c t Keywords: Uncertainty in illness MUIS-A Psychometrics Cancer Confirmatory factor analysis
Purpose: Uncertainty is a major component in the illness experiences which extraordinarily can affect the psychological adjustment and the illness outcomes. Uncertainty in illness is defined as inability to define the illness-related events to the illness or disability in predicting the illness outcomes. The present study aimed to translate the Persian version of Uncertainty in Illness Scale (MUIS-A) and to investigate its psychometric properties on patients with cancer. Method: In this methodological study, validation of the Persian version of MUIS-A was performed in Iran on 420 cancer patients attending two major hospitals in Tehran, Iran. The scale was translated into Persian and back translated into English and revised according to editorial comments of the scale designers. Then, content and face validity, construct validity, internal consistency reliability and stability of the Persian version were measured. Data were analyzed using SPSS version 16 and LISREL 8.5. Results: Mean of the participants MUIS-A score was 90.1 (16.8). Confirmatory factor analysis confirmed validity of the whole instrument and its four subscales. The consistency of the instrument with a threeweek interval was r ¼ 0.91. Cronbach’s alpha was 0.89 for the whole scale of 32 MUIS-A items and a ¼ 0.58e0.86 for its four factors. Conclusions: The Persian version of the MUIS-A has good psychometric properties. It can be used to assess uncertainty in illness in Iranian patients with cancer. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Uncertainty is a natural part in the illness experience and can occur at all stages of disease including diagnosis, treatment and prognosis of the disease. Uncertainty in illness occurs when the patient is unable to determine the meaning of illness-related events; therefore, it is considered a major cause of psychological stress to the patients (Mishel, 1997a, 2013), that can leave major impacts on the psychological adaptation and outcomes of the disease (Neville, 2003). Uncertainty in illness is defined as the “inability to determine the meaning of illness-related events when * Corresponding author. Pediatric Nursing Department, Nursing & Midwifery School, Shahid Beheshti University of Medical Sciences, Vali-e Asr Street, Niyayesh Cross, Tehran, Iran. Tel.: þ98 21 88655372; fax: þ98 21 88202521. E-mail address: [email protected] (M. Rassouli). 1462-3889/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejon.2013.09.006
the patient cannot determine the value of events or cannot accurately predict the disease outcome due to the lack of sufficient cues” (Mishel, 1988). Uncertainty occurs when the individuals cannot form a cognitive framework for understanding their status and interpreting illness related events (Bailey et al., 2011; Mishel, 1990, 2013). Therefore, in most cases, it is accepted as a major stressor that most people seek to reduce and to learn methods to cope with (Neville, 2003). Chronic diseases are a long term and intense source of psychological stress that influences all aspects of patients’ life and reduces their daily activities and quality of life (Flemme et al., 2005). Furthermore, various studies have been conducted on chronic patients such as patients with cardiovascular diseases, dialysis, hepatitis, liver transplantation, AIDS and Alzheimer’s whose results indicate that uncertainty exists in these patients and can lead to reduction in quality of life and the ability to cope with
M. Sajjadi et al. / European Journal of Oncology Nursing 18 (2014) 52e57
illness (Bailey et al., 2009; Brashers et al., 2004; Flemme et al., 2005; Lasker et al., 2010; Madar and Bar-Tal, 2009; Mauro, 2010; Sodowsky, 2012). Cancer, as one of the chronic diseases, is the second leading cause of death in the West (Shaha et al., 2008). It is the third cause of death in Iran after heart diseases and car accidents. Based on the recent statistics, the rate of cancer incidence in Iran is about 107 people in 100,000 people (more than 80,000 people considering Iran’s 75million population). It is predicted that with regard to the raise of environmental pollution, increasing the elderly (aging community) and population growth, coming decades will witness an increasing trend in cancer so that it becomes a principle health problem in Iran (Rasaf et al., 2012). Today, advances in cancer diagnosis and treatment technology has led to an increase in patients’ survival, and one of the important points in the care of these patients is to notice and improve the quality of life (Stewart et al., 2010b). From the diagnosis time to the onset of treatment, patients with cancer suffer from high levels of emotional tension (Chen et al., 2010). Due to the complexities of treatment and symptoms, patients experience some levels of uncertainty about illness which originates from problems related to the inability to perform daily activities, inadequate treatment, and concerns about the relapse of the disease (Haisfield-Wolfe et al., 2012). Various studies have been conducted in the field of uncertainty in illness in different types of cancer in other countries (mainly Western countries) (Detprapon et al., 2013; Haisfield-Wolfe et al., 2012; Harrow et al., 2008; Kurita et al., 2013; Parker et al., 2013; Shaha et al., 2008; Sherman and Simonton, 2010; Stewart et al., 2010a; Woodgate and Degner, 2002). The ultimate goal of these studies is to help the patients (or their families) to effectively manage uncertainty so that its negative effect on the patient’s psychological adaptation is reduced (Stewart et al., 2010b). Nurses are frequently in contact with the patient, thus they are in the best position to reduce uncertainty in patients by providing information and promoting patients’ understanding of health (Madar and Bar-Tal, 2009). In order to conduct interventions to reduce uncertainty in patients with cancer and to measure effectiveness of these interventions, a scale is needed to measure it. There are various scales for uncertainty assessment in different patients and groups (Lin et al., 2012; Mishel, 1997b; Pai et al., 2007; Stewart et al., 2010a). One of the convenient and widely used scales is Mishel Uncertainty in Illness Scale-Adult form (MUIS-A). MUIS-A is a valid and reliable scale and has been used in several studies. This instrument contains 32 items on Likert scale from 1 (strongly disagree) to 5 (strongly agree) points and occurs in four dimensions of ambiguity, complexity, inconsistency, and unpredictability. The instrument scores between 32 and 160, and earning more points shows greater uncertainty (Bailey et al., 2009; Mishel, 1997b). MUIS-A has a good internal consistency with Coronbach’s alpha of 0.87 for the whole instrument, 0.86 for the subscales of ambiguity (13 items), 0.81 for complexity (7 items), 0.78 for inconsistency (7 items), and 0.65 for unpredictability (5 items). This scale has been translated into different languages and its validity and reliability have been studied (Mishel, 1997b). Despite the importance of the concept of uncertainty in illness in cancer patients, no valid and reliable tool exists in this regard in Iran. Therefore, this study was conducted with the aim of translation and psychometric properties of the Persian Version of “Mishel Uncertainty in Illness Scale” in cancer patients. Methods Procedures The present study is a methodological research (LoBiondoWood et al., 2006) through which the MUIS-A scale is translated
53
and its Persian version has been validated among patients with cancer. The study population was all the cancer patients attending the clinics and oncology wards of Imam Khomeini Hospital (Iran’s largest cancer center) and Taleghani Hospital in Tehran. Given these centers accept patients from all over the country, results of the study can be better generalized to the society. Convenient sampling was performed in order to achieve a sufficient sample size to conduct a confirmatory factor analysis. The inclusion criteria were willingness to participate in research, cancer diagnosis by the oncologist, being aware of one’s illness, being Iranian, the ability to understand and speak Persian, the minimum age of 21 years, and absence of other serious illnesses and no history of serious mental disorders (such as schizophrenia). Type and stage of cancer and treatment type were not the inclusion criteria of this study. Based on these criteria, eventually 420 individuals were chosen. Scale translation In this study, translation and validation of the instrument was performed based on the suggested model by Wild et al. (2005). After obtaining written permission from the original designer of the scale, the original scale was translated into Persian by two people fluent in English and Persian. Then the two translations were compared and the final version was prepared with a few modifications. Then, the final translated version was given to two people fluent in both languages of English and Persian (one native English speaker), who were not in contact with the first people to translate it back into English. Then, the back-translation was contrasted by a supervisor with translations of the original scale in terms of similarity of translations, and some minor revisions were made in statements of the Persian version. Afterward, validation process of the translated scale was performed by assessing face and content validity, construct validity (using confirmatory factor analysis), reliability, and internal consistency. Content validity In order to examine the content validity index (CVI), the translated scale was given to ten people (a specialist in clinical psychology, an oncologist, two psychiatric nurses, two oncology nursing instructors, four nursing assistant professors with experience in instrument development) to review that and provide their correctional comments. They also investigated and affirmed face validity of the translated scale. Then, the scale was given to 10 patients with cancer to express their idea about simplicity of use, and understandability of words and phrases. The Persian version was finalized without much change in its sentences. Data collection To collect the data, after choosing eligible patients and explaining the study objectives and methods, the informed consent was obtained from them. Next, the questionnaire including demographics (age, sex, educational level, marital status, occupation, place of residence, type of illness, type of treatment, time of diagnosis) and MUIS-A scale were given to them to complete. Completion of the questionnaire took about 20e25 min. For illiterate people, the questionnaire was read and patients’ answers were marked by the researcher. Investigating validity of the MUIS-A construct and the model fitness was conducted by confirmatory factor analysis (CFA) using LISREL statistical software version 8.5.
54
M. Sajjadi et al. / European Journal of Oncology Nursing 18 (2014) 52e57
Construct validity CFA is a structural equation modeling technique which is used to determine the goodness of fit between a hypothesized model and the data obtained from study samples (Kline, 2010). Maximum likelihood algorithm was used to evaluate the fitness of the model. There are several fit indices for deciding on the goodness of fit of the model and it is recommended that different indices be used (Brown, 2006; Seo et al., 2004). In this study, indices of fit chisquare, root mean error of approximation (RMSEA), goodness-offit index (GFI), comparative fit index (CFI), adjusted goodness-of-fit index (AGFI), and standardized root mean square residual (SRMR) were used. One of the most used indexes is chi-square. Because this index is related to the sample size, ratio of chi-square to degrees of freedom is used, where values less than 2 indicate a good fit of the model. Another important index is the RMSEA, where values less than 0.08 are acceptable and less than 0.05 indicates good fit of the model. Researchers are advised to report the confidence interval (Kline, 2010; Seo et al., 2004). Another important index is SRMR, whose acceptable value is 0.06 or less (Brown, 2006; Kline, 2010). Appropriate values are >0.9 for CFI and GFI, and >0.85 AGFI (Bentler, 1990; Helsen et al., 2013).
Reliability Internal consistency and reliability were measured using SPSS software version 16. Scale internal consistency was measured by Cronbach’s alpha for the total scale and each subscale separately.
The best method used to evaluate the internal consistency is Cronbach’s alpha (Polit and Beck, 2013). The Cronbach’s alpha of about 0.7 is sufficient, and of >0.80 indicates high internal consistency of the tool (Polit and Beck, 2013). The reliability was measured by Pearson’s or Spearman’s correlation coefficients in testeretest on 15 patients with a three-week interval. Usually, the time interval of 2e3 weeks is appropriate between two tests (Hawthorne et al., 2011; Polit and Beck, 2013).
Results A total of 420 questionnaires were analyzed. Demographic characteristics of the study sample are shown in Table 1. Content validity index was measured on relevance, clarity, and simplicity of the translated scale as 0.98, 0.95, and 0.96, respectively. Table 2 shows mean and standard deviation of the overall scores of MUIS-A and its dimensions. In order to have a better comparison between the present scale and the original one, data of the present study were compared to the combined multiple studies on patients with cancer (Mishel, 1997b). The results show that mean of the overall score of uncertainty in illness in the present study (90.1) is greater than that of the combined sample of cancer patients proposed by Mishel (79.5). Coefficient of Cronbach’s alpha was 0.89 for the overall scale, and 0.58e0.86 for its dimensions (Table 2). The correlation coefficient testeretest reliability for MUIS-A was 0.91 (p < 0.001). Correlation coefficients for subscales of ambiguity, complexity, inconsistency, and unpredictability were respectively 0.90, 0.74, 0.83, and 0.60.
Table 1 Demographic and clinical characteristics of patients with cancer (n ¼ 420).
Age (year) Time since diagnosis (months)
Sex Marital status
Education level
Occupation
Time since diagnosis (months)
Type of cancer
Metastasis Kind of treatment
Female Male Single Married Widowed or divorced Illiterate Primary and Guidance school Diploma University degree Self-employed Employee Housewife Unemployed Other 24 Breast Colon and rectum Gastric/esophagus Hematologic Uterus/ovarian Prostate Lung Skin Other Yes No Chemotherapy Chemotherapy and surgery Chemo and radiotherapy All of them Other
Mean
SD (range)
46.4 18.5
13.9 (21e79) 24.2 (3e168)
N
%
217 203 58 331 31 87 151 107 75 76 34 174 35 103 166 163 91 103 99 36 72 25 5 21 5 54 143 251 105 189 26 90 10
51.7 48.3 13.8 78.8 7.3 20.7 36 25.5 17.9 18.1 8.1 41.4 8.3 24.1 39.5 38.8 21.7 24.5 23.6 8.6 17.1 6 1.2 5 1.2 12.9 36.3 63.7 25 45 6.2 21.4 2.4
M. Sajjadi et al. / European Journal of Oncology Nursing 18 (2014) 52e57
55
Table 2 Scores for MUIS-A scale and its dimensions in Iranian patients with cancer and it’s comparison with Mishel’s data bank.
Iranian patients with cancer (N ¼ 420) Mishel’s combined data for cancer patients (N ¼ 761)
Ambiguity mean (SD) e a
Complexity mean (SD) e a
Inconsistency mean (SD) e a
Unpredictability mean (SD) e a
Total mean (SD) e a
41.8 (9.4)e0.86
15.1 (3.2)e0.58
18.8 (4.2)e0.71
14.4 (3.5)e0.65
90.1 (16.8)e0.89
32.3 (9.3)e0.87
16.5 (5)e0.80
14.9 (4.7)e0.80
15.8 (3.7)e0.67
79.5 (16.9)e0.90
Construct validity To investigate construct validity and provide the most appropriate model for the MUIS-A, confirmatory factor analysis was used. To this end, full samples of cases were randomly split into two approximately equal groups. 200 subjects were considered as calibration sample and 220 subjects were considered as evaluation sample. Calibration samples were used to check the models (2 and 4 factors) and do possible revisions. Evaluation sample was used to validate the appropriate model. Two-dimensional model of the scale did not have good fit indices, but the four-dimensional model had better fit indices. The chi-square, RMSEA, SRMR and AGFI indices confirm the fit indices of the model (Table 3). By assessing the scale items, it was determined that the item 32 “The seriousness of my illness has been determined” had a weak item-scale correlation coefficient (r ¼ 0.02), therefore, item 32 was removed. According to the suggestion of modification of fit indices in confirmatory factor analysis, transferring item 8 “I do not know when to expect things will be done to me” (from the ambiguity to the inconsistency) and item 28 “treatment which is being done to me has a known probability of success” (from the complexity to the unpredictability) improved the fit of the model. Therefore, the fourdimensional model of MUIS-A with 31 items (ambiguity 12 items, complexity 5 items, inconsistency 8 items and unpredictability 6 items) is a model with the best fit indices. CFI and GFI indices are acceptable (Bentler, 1990) and other indices such as the ratio of chisquare to degree of freedom, RMSEA, AGFI and SRMR indicate a good fit of the model (Brown, 2006; Helsen et al., 2013; Kline, 2010). In order to further investigate validity of the construct, the relationship between education level and uncertainty in illness rate was used. In this study, a strong inverse correlation was found between uncertainty and the level of education (r ¼ 0.47, p < 0.001) in the sense that patients with higher level of education had less uncertainty. Discussion Uncertainty in illness is a concept that can have broad impact on cancer patients and its investigation in this group of patients is of
great importance. Despite the importance of the issue, this problem has not been studied in Iran that can be attributed to lack of proper tools in Iran. Therefore, this study was conducted with the aim of translating and validating the Persian version of the uncertainty in illness scale in patients with cancer. In this study, the MUIS-A scale was translated into Persian and its psychometric properties and factor structure were evaluated. When a tool is translated or used in another culture or society, it is essential to evaluate its psychometric properties because that may not be appropriate in the new society (Hillen et al., 2013; Michaeli Manee, 2011). In this study, content validity, construct validity, and reliability of this tool were assessed. Based on the findings, the Persian version of MUIS-A generally has good reliability and validity and its dimensions’ construct is similar to that of the original one (Mishel, 1997b). In this study, it has been shown that with regard to content validity index of 0.96, the Persian version of MUIS-A has a good content validity. Polit and Beck (2013) suggest that the CVI value of 0.9 or more should be considered as a standard for content validity of the scales (Polit and Beck, 2013). Confirmatory factor analysis of MUIS-A was conducted using a sample of 420 cancer patients. Goodness-of-fit indices indicated that the original model (32-item and 4-dimensional) of Persian version of the uncertainty in illness scale is appropriate, but for better fit indices and to enhance the fit indices some changes are needed to be made in the original scale. In confirmatory factor analysis, some suggestions are given by Lisrel to modify and improve the model fit indices. If they are justifiable logically and theoretically, they can be applied (Kline, 2010). In the Persian version, item 32 “The seriousness of my illness has been determined” from complexity dimension was deleted due to its low correlation coefficient of item-scale. In validation of the Swedish version of this scale for the cardiac patients, it was indicated that this item has no significant correlation with the overall score of the scale and complexity subscale. Furthermore, items 10 and 22 were excluded from the Swedish version (Hallberg and Erlandsson, 1991). Another study was conducted in Iran to validate the scale of uncertainty in illness e family form. This scale assesses uncertainty from patients’ family’s perspective. The original version of this scale has 31 items and is
Table 3 Confirmatory Factor analysis Fit indices for the different MUIS-A version in Iranian patients with cancer.
c2 (df), p MUIS-A, MUIS-A, MUIS-A, MUIS-A, a b c d e g f
2 4 4 4
factorsa factorsa (original) factorsa (modified) factorsb (modified)
657.4 767.4 660.7 652.3
Calibration sample (N ¼ 200). Validation sample (n ¼ 220). Root mean error of approximation. Comparative fit index. Goodness-of-fit index. Standardized root mean square residual. Adjusted goodness-of-fit index.
(349), (458), (428), (428),
p p p p
< < <
