Support Care Cancer DOI 10.1007/s00520-014-2468-9
ORIGINAL ARTICLE
Oral-health-related quality of life in patients with cancer: cultural adaptation and the psychometric testing of the Persian version of EORTC QLQ-OH17 Mir Saeed Yekaninejad & Amir H. Pakpour & Jyothi Tadakamadla & Santhosh Kumar & Seyed Hamzeh Mosavi & Bengt Fridlund & Andrew Bottomley & Neil K. Aaronson
Received: 27 May 2014 / Accepted: 1 October 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose To evaluate the validity, reliability, responsiveness to treatment, and gender invariance of the Persian version of the European Organization for Research and Treatment of Cancer Oral Health Questionnaire (QLQ-OH17) among Iranian cancer patients. Methods Cancer patients (n=729) from three oncology centers in Tehran and Qazvin were recruited. A forward and backward translation procedure was performed to develop a M. S. Yekaninejad Department of Epidemiology and Biostatistics, School of Public Health, Tehran University of Medical Sciences, Tehran, Iran A. H. Pakpour (*) Social Determinants of Health Research Center, Qazvin University of Medical Sciences, Shahid Bahounar BLV, Qazvin 3419759811, Iran e-mail: [email protected] A. H. Pakpour e-mail: [email protected] J. Tadakamadla : S. Kumar Population and Social Health Research Program, Griffith Health Institute, School of Dentistry and Oral Health, Griffith University, Gold Coast, Australia S. H. Mosavi Department of General Surgery, Hazrat-e-Rasoul Hospital, Iran University of Medical Sciences, Tehran, Iran B. Fridlund School of Health Sciences, Jönköping University, Jönköping, Sweden A. Bottomley Quality of Life Department, European Organization for Research and Treatment of Cancer, Brussels, Belgium N. K. Aaronson Department of Psychosocial Counseling, The Netherlands Cancer Institute, Amsterdam, The Netherlands
culturally acceptable version of Persian QLQ-OH17. Internal consistency and test–retest reliability of the QLQ-OH17 was assessed. In addition, convergent and discriminant validity, concurrent validity, construct validity, and known-groups validity were evaluated. The factor structure of the questionnaire was examined by exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). Responsiveness to change was measured in an independent sample of patients with head and neck cancer undergoing radiotherapy. Finally, factorial invariance of the QLQ-OH17 was assessed across gender. Results The Persian version of QLQ-OH17 showed good internal consistency (Cronbach’s alpha coefficients of 0.71– 0.83) and reliability on repeated administration (intraclass correlation coefficients of 0.85–0.94). Persian QLQOH17 exhibited the original four-factor structure. Patients who perceived good oral health and satisfaction with their mouth reported significantly better oralhealth-related quality of life (OHRQoL) than those who perceived poor oral health and dissatisfied with their mouth. Similarly, those who perceived a need for dental treatment reported significantly poorer OHRQoL than those who have not perceived any treatment need. Older patients, females, and those experiencing greater caries had poorer QoL than their comparative counterparts. All QLQ-OH17 subscales were correlated with QLQ-C30 subscales and global QoL. Both male and female patients with cancer interpreted items on the QLQ-OH17 in a similar manner. The QLQ-OH17 was found to be responsive to treatment in a sample of head and neck cancer patients. Conclusions The Persian version of QLQ-OH17 is a valid and reliable questionnaire for assessing OHRQoL in Iranian patients with various cancers. Keywords Translation . Oral health . Quality of life . Cancer . Psychometrics . Questionnaire
Support Care Cancer
Introduction According to the Global Burden of Cancer Study, cancer is the leading cause of deaths worldwide with 8.2 million deaths in 2012. More than 60 % of new cases in the world are accumulated in Africa, Asia, and Central and South America [1]. In Iran, the data from the Global Burden of Diseases, Injuries, and Risk Factors Study 2010 suggest that cancer is the second most common reason of death and more than 57 % of total deaths are caused by cardiovascular diseases and cancers [2]. WHO defines quality of life (QoL) as an individual’s perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards, and concerns [3]. Currently, the multidimensional concept of QoL which includes behavioral competence and health, perceived quality of existence, and psychological well-being [4] has become an increasingly important issue in cancer patients [4–7]. The importance of QoL in cancer patients becomes more pronounced with increasing survival rates which involves intermediate and late treatment effects of cancer that might affect physical, psychological, and social well-being [8]. The reason for patient-reported QoL outcomes gaining importance in cancer patients is because of the varied symptoms and functional disabilities experienced by cancer patients which cannot be evaluated using routine laboratory tests or imaging procedures [9]. In addition, cancer survivors might be subjected to oral complications which diminish their QoL [10, 11]. These oral complications involve alterations in saliva function and taste sensation, infections and necrosis of the jaw [11], mucosal inflammation, ulceration, and oral candidiasis and bleeding [11]. Moreover, oral cavity has been documented as the most common source of sepsis in immunosuppressed patients with cancer [12]. The European Organization for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQC30) is one of the most widely used cancer health-related quality of life (HRQoL) instrument that has been developed by EORTC. The QoL Group and QoL Department of EORTC are international leaders in the development of patientreported outcome measures in oncology [13, 14]. More recently, health care professionals from various disciplines at Oslo University in Norway have developed a supplementary module “EORCTC QLQ-OH17” to the EORCTC QLQ-C30 for oral and QoL assessment in cancer patients. This is the only cancer-specific OHRQoL instrument which is a brief assessment tool focusing on oral health problems and QoL in cancer patients [15]. QLQ-OH17 is currently undergoing international field testing for cross-cultural applicability, validity, and reliability. Responsiveness to treatment was also evaluated in this study as an outcome measure to be considered as useful must demonstrate responsiveness to treatment effects [16].
The current paper evaluates validity, reliability, responsiveness to treatment, and gender invariance of this module in Persian language among Iranian cancer patients.
Materials and methods Study population Cancer patients attending three oncology centers in Tehran and Qazvin during the period June 2013 to December 2013 were recruited. Patients were eligible for inclusion in the study if they were 18 years of age or older, histologically diagnosed with cancer, were able to read and speak Persian, had life expectancy of at least 4 weeks, and able to provide written consent. Patients who are too old and weak, having cognitive impairment, participating in an oral health program or clinical trial study, and those suffering with psychiatric disorders were considered for exclusion. The study was approved by the Ethics Committee of the Qazvin University of Medical Sciences. Written informed consent was obtained from all the patients. Instruments Questionnaires were administered to all the participants. In addition to EORTC QLQ-C30 and QLQ-OH17, questionnaire also consisted of questions on self-perceived oral health, selfperceived satisfaction with mouth, and perceived need of dental treatment. The EORTC QLQ-C30 The EORTC QLQ-C30 (version 3.0) is a generic measure to assess HRQoL among patients with cancer [13]. It consists of 30 items which covers five functional scales (physical functioning, role limitation, emotional, cognitive, and social), three symptom scales (fatigue, nausea or vomiting, and pain), six single items (dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties), and a global health and QoL scale. All items are rated on a four-point Likert scale while global health and QoL are rated on seven-point Likert scale. QLQ-C30 consists of both multi-item scales and single-item measures, and total score for each scale is calculated. A raw score by estimating the average of items in the scale is calculated which is later transformed linearly to get a standardized score, which ranges from 0 to 100. Higher scores in functional scale and global health status/QoL indicate better QoL while higher scores in symptom scale indicate higher symptomatology [17].
Support Care Cancer
The EORTC QLQ-C30 has been translated into several languages including Persian and has been found to be highly valid and reliable [18]. The EORTC QLQ-OH17 The EORTC QLQ-OH17 is a specific tool to assess OHRQoL in cancer patients [15]. The EORTC QLQ-OH17 is composed of four multi-item scales (pain/discomfort, xerostomia, eating, and information) and three single items (two items related to use of dentures and one on future worries). All these items are rated on four-point Likert-type scale ranging from 1 (not at all) to 4 (very much) except one item related to denture usage which has a dichotomous response (yes/no). Higher scores on pain/discomfort, xerostomia, and eating represent a high level of symptomatology or problems while higher scores on information indicate higher information level [15]. The scoring procedure for QLQ-OH17 is similar to that of EORTC QLQ-C30 which is described in the EORTC scoring manual. Translation procedure Translation from the English original EORTC QLQ-OH17 into Persian was done in according to the EORTC manual proposed by the EORTC QoL study group [19]. Permission from the EORTC QoL department for translating QLQ-OH17 into Persian was sought by the principal investigator (AP). The aim of the translation procedure was to produce a Persian version of the EORTC QLQ-OH17 which is conceptually equivalent to the original English version. The translation procedure was initiated with forward translation. In this step, two translators proficient in English, who were native Persians, independently translated QLQ-OH17 from English to Persian. Later, a single reconciled version was prepared by the principal investigator (AP) and the translators through resolving the discrepancies between the translations. This interim Persian version was then translated back into English by two native English speakers who were blinded to the original English version. Back translation was conducted independently by the translators. The principal investigator (AP) compared the English translations with the original version to ensure consistent translation. In the next step, the interim translated version was pilot-tested on 18 patients (mean age=54.12, range=29–61) with various types of cancers (head and neck, breast, lymphoma/leukemia, and gastrointestinal). A structured interview was administrated to check if the items were confusing, upsetting, and difficult in terms of understanding or wording. It was found that all the patients were able to read and understand the items with no difficulty. Therefore, no specific change was made in the final version. At final stage, the piloted Persian version of the EORTC QLQ-OH17 was administrated on 678 Iranian patients with cancer for psychometric assessment.
Procedure Eligible patients were invited to attend an introduction session to explain study aims. Afterwards, the patients were asked to sign a consent form and complete the questionnaires in clinic rooms. Demographic information (age, educational status, marital status, and occupation) and clinical information were also collected via the patients’ medical records. Two calibrated dentists conducted oral examinations in the same clinics. A total of 678 patients were subjected to oral examination to record caries experience using decayed, missing, and filled teeth (DMFT) index [20], and WHO criteria was followed for recording caries [21]. The questionnaire was re-administrated to all the patients whose condition was stable (no expected clinical change) after 2 weeks from baseline assessment. For evaluating the responsiveness of the instrument to treatment, an independent sample of head and neck cancer patients were selected from the same hospitals in Tehran and Qazvin with similar inclusion and exclusion criteria. Only those patients that were newly diagnosed with head and neck cancer and were supposed to undergo radiotherapy were considered for inclusion. They were asked to complete the questionnaire at baseline and 2 months after the initiation of radiotherapy. Statistical analysis Statistical analyses were performed using statistical software PASW Statistics 18 (SPSS Inc., Chicago, IL) and LISREL 8.80. Response distributions were assessed for floor (the proportion of patients with the lowest possible scores) and ceiling effects (the proportion of patients with the highest possible scores). Floor or ceiling effects are considered to be present if more than 15 % of patients scored the lowest or highest scores on the scale, respectively [22]. The reliability of the QLQOH17 was evaluated using internal consistency and test–retest reliability at repeated administrations with a 2-week interval. Internal consistency was assessed using Cronbach’s alpha coefficient, and Cronbach’s alpha values of 0.7 or more were considered to be acceptable [23]. Test–retest reliability of the QLQ-OH17 was evaluated by calculating the intraclass correlation coefficient (ICC), and an ICC ≥0.70 was considered to be acceptable [23]. Convergent and discriminant validity of the QLQ-OH17 was evaluated using multi-trait scaling analysis [24]. The multi-trait scaling analysis was conducted to examine correlations between items and hypothesized scales (corrected for overlap) [25, 24]. Pearson’s correlation coefficients were used for multi-trait analyses in order to deflate skewness, kurtosis, and categorization of variables. Item convergent validity was accepted if item-own scale correlation was equal or higher than 0.40 [24]. It was also hypothesized that item-own scale correlation was higher than item-other scale (twice the standard error) correlation [24].
Support Care Cancer
To the best of our knowledge, there is no gold standard existing to evaluate the OHRQoL in cancer patients. Therefore, to evaluate concurrent validity of the QLQOH17, a series of multiple linear regression analyses with four subscales of the QLQ-OH17 as dependent variables and subjective perceptions of patients as independent variables was conducted. It was hypothesized that patients who are dissatisfied with their mouth, perceive poor oral health, and need for dental treatment report poor OHRQoL. In addition, ability of the QLQ-OH17 to differentiate between subgroups of patients was evaluated using independent t test. To evaluate factor structure of the QLQ-OH17, the whole sample was randomly split into two separate groups. The two groups were comparable in terms of demographic and clinical variables. An exploratory factor analysis (EFA) with principal component analysis (PCA) and Varimax rotation was conducted on the first half (n=339) to explore underline factor structure of this tool. The suitability of the EFA was evaluated using two indices including the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and the Bartlett’s test of sphericity. KMO higher than 0.7 and statistically significant Bartlett’s test of sphericity were considered to check if the data are factorable. A confirmatory factor analysis (CFA) was then performed on the second half (n=339) to ensure clear distinctions between the factors. The model parameters were estimated by the weighted least squares with polychoric correlations and asymptotic covariance matrices. Model fit was evaluated using various indices including chi-squared statistic, Bentler’s comparative fit index (CFI), non-normed fit index (NNFI), the root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Model fit was considered to be acceptable when χ2 was non-significant (p < 0.05), RMSEA was ≤0.08, SRMR was ≤0.08, NNFI was ≥0.90, and CFI was ≥0.90. In order to investigate whether the QLQ-OH17 is measuring same constructs across gender groups, a multiple-group CFA analysis using total sample was conducted. A two-level hierarchical procedure as proposed by Milfont and Fischer was used. At the first level, it is considered that patients belonging to both the gender groups conceptualize the constructs in the same way (configural invariance) while at the second level, all the factor loadings are assumed to be equal across the gender groups (metric invariance). Differences in CFI (ΔCFI ≤0.01) and NNFI (ΔNNFI ≤0.01) were used to demonstrate factorial invariance across groups. Responsiveness of the QLQ-OH17 instrument to clinical changes was tested using paired t tests (two-tailed). Effect sizes (Cohen’s d statistic) were computed for assessing the magnitude of differences over time.
Results A total of 729 patients with cancer were initially assessed for eligibility in the study. Seven hundred one patients meeting the inclusion criteria were invited to participate, of which 678 agreed to participate with a response rate of 97 %. Table 1 demonstrates the demographic and clinical characteristics of the study population. Six of ten participants (61.2 %) were women, and the mean age was 54.42 years (SD=13.9). Most of the patients (36.0 %) were educated until “secondary school” followed by illiterate group who constituted more than a quarter of the study population. Breast and colorectal cancers were the two most common types of cancers observed in the study population. Table 2 demonstrates that all the subscale and single-item scores of Persian QLQ-OH17 were normally distributed. No floor or ceiling effects were observed for any of the QLQ-OH17 subscales and single items. Internal consistency of the QLQ-OH17, as assessed by Cronbach’s Table 1 General characteristics of the study population Age in years mean (SD) DMFT mean (SD) Gender Male Female Educational status Illiterate Primary school Middle school Secondary school College Marital status Single
54.42 (13.90) 10.11 (6.25) n (%) 263 (38.8 %) 415 (61.2 %) n (%) 189 (27.9 %) 98 (14.5 %) 63 (9.3 %) 244 (36.0 %) 84 (12.4 %) n (%) 58 (8.5 %)
Married Widowed/divorced Occupational status Employed Unemployed Type of cancer Breast Stomach Colorectal Lymphoma/leukemia Head and neck Multiple myeloma Cervical Ovarian Renal cell/kidney Prostate Other
456 (67.3 %) 164 (24.2 %) n (%) 327 (48.2 %) 351 (51.8 %) n (%) 193 (28.5 %) 97 (14.3 %) 81 (12.0 %) 49 (7.2 %) 38 (5.6 %) 36 (5.3 %) 33 (4.9 %) 31 (4.6 %) 29 (4.3 %) 22 (3.2 %) 69 (10.2 %)
DMFT decayed, missing, and filled teeth index
Support Care Cancer Table 2 Descriptive statistics for the Persian version of QLQ-OH17
Pain/discomfort Xerostomia Eating Information Worried about future Worn dentures Ill-fitting denture
Number of forms
Mean (SD)
Floor N (%)
Ceiling (%) Cronbach’s alpha
Normality (Kolmogorov–Smirnov test)
663 671 677 673 675 655 541
49.08 (23.29) 45.15 (20.80) 52.78 (18.71) 27.43 (23.5 %) 33.29 (19.2) 50.32 (50.1) 47.14 (22.14)
8 (4.1) 9 (1.3) 17 (2.5) 159 (23.5) 48 (7.1) – 21 (3.1)
3 (1.5) 88 (13.0) 68 (10.0) 11 (1.6) 82 (12.1) – 76 (11.2)
0.123 0.201 0.112 0.102 0.09 – 0.142
alpha statistic, ranged from 0.71 to 0.83. Table 3 depicts that the ICCs for the subscales of the QLQ-OH17 ranged from 0.85 (eating) to 0.94 (xerostomia). Multi-trait scaling analysis provided support for scaling assumptions for all dimensions as depicted in Table 4. Itemscale correlations exceeded 0.40 for all the items, and no scaling errors were seen. Moreover, all QLQ-OH17 subscales were significantly correlated with QLQ-C30 subscales and global QoL scores (Table 5). Pain/discomfort and eating subscales were significantly related to all the subscales of EORTC QLQ-C30 while xerostomia did not significantly correlate with social functioning and dyspnea. Information was not related to physical functioning, social functioning, pain, fatigue, nausea or vomiting, and financial impact subscales of QLQ-C30. Table 6 presents the results from multiple linear regression analyses that were conducted to assess concurrent validity of the QLQ-OH17. The results indicated that patients who perceived good oral health and satisfaction with their mouth reported significantly better OHRQoL than those who perceived poor oral health and dissatisfied with their mouth. Similarly, those who perceived a need for dental treatment reported significantly poorer OHRQoL than those who have not perceived any treatment need. Female and older patients generally reported significantly poorer OHRQoL than male and younger patients for each QLQ-OH17 subscale. Moreover, patients with greater caries experience (higher DMFT index scores) reported significantly poorer OHRQoL for all the QLQ-OH17 subscales than those with lower caries experience (Table 7). Principal components analysis was conducted on the combined 14 items (excluding single items) to assess factor structures. Bartlett’s test of sphericity and KMO as criteria for EFA Table 3 Test–retest reliability of the Persian version of QLQ-OH17
Scale
ICC (95 % CI)
Pain/discomfort Xerostomia Eating Information
0.876 (0.855–0.894) 0.938 (0.928–0.947) 0.853 (0.828–0.875) 0.880 (0.859–0.897)
0.83 0.76 0.71 0.74 – – –
analysis was checked to verify factorability of the data. The results indicated that Bartlett’s test of sphericity was statistically significant (approximate χ 2 = 1,339.56 df = 91, p1 which accounted for 78.4 % of the variance in OHRQoL (Table 8). CFI of the Persian version of QLQ-OH17 using the four domain original version loadings of the 14 items resulted in a good fit: χ2 = 122.47, df = 71, p < 0.004, CFI = 0.97, RMSEA = 0.075, SRMR=0.057, and NNFI=0.96. Standardized regression weights for all items with their respective domains were
Table 4 Multi-trait scaling analysis of the subscales of Persian QLQOH17 Pain/ Xerostomia Eating Information discomfort Pain/discomfort Pain in gums Bleeding gums Pain in lips Problems with teeth Sore mouth Sores in mouth corners Xerostomia Dry mouth Sticky saliva Eating Sensitive mouth Taste change Problems with solid food Trouble enjoying meals Information Receiving any information about oral problems Satisfaction with information
0.718 0.785 0.876 0.610
0.247 0.187 0.323 0.123
0.262 0.424 0.352 0.348
0.036 0.151 0.048 0.165
0.783 0.676
0.335 0.147
0.386 0.111
0.142 0.180
0.241 0.349
0.722 0.931
0.382 0.411
0.063 0.018
0.067 0.297 0.148
0.241 0.241 0.311
0.766 0.815 0.876
0.087 0.097 0.142
0.151
0.099
0.884
0.144
0.095
0.123
0.150
0.945
0.190
0.013
0.178
0.943
>0.6 except two items (“problems with solid food” and “trouble enjoying meals”). The four factor structure of the QLQ-OH17 was tested across gender groups to evaluate measurement invariance. The results of multiple-group CFIs showed that both models (configural and metric invariance) yielded comparable fit in the factorial invariance tests across gender groups. At the first level, results indicated invariance of the factor structure of QLQ-OH17 across gender groups (configural invariance): χ2 =180.3, df=133, pMean value (n=211)
≤Mean value (n=467)
30.87 (18.72) 38.71 (21.32) 48.60 (23.95) 16.06 (7.58)
56.66 (27.38) 56.01 (19.59) 66.80 (24.62) 13.74 (7.33)
60.00 (27.38) 63.99 (23.30) 63.40 (21.81) 12.93 (7.71)
33.15 (27.59) 26.37 (16.32) 44.32 (19.04) 32.88 (26.24)
51.31 (20.72) 65.76 (24.92) 61.08 (24.09) 16.65 (6.11)
28.21 (25.38) 28.70 (18.61) 34.76 (17.86) 31.02 (12.23)
DMFT decayed, missing, and filled teeth index *Statistically significant according to Benjamini–Hochberg procedure for gender **Statistically significant according to Benjamini–Hochberg procedure for age categories †Statistically significant according to Benjamini–Hochberg procedure for DMFT
Support Care Cancer Table 8 Results obtained from exploratory factor analysis (n= 339) of QLQ-OH17
Item
Factor 1
Pain in gums Bleeding gums Pain in lips Problems with teeth Sore mouth
0.76 0.77 0.62 0.78 0.84
Sores in mouth corners Dry mouth Sticky saliva Sensitive mouth Taste change Problems with solid food Trouble enjoying meals Receiving any information about oral problems Satisfaction with information Eigenvalue Explained variance (%)
0.87
The difference in OHRQoL between the genders might be due to the difference in influences during the life course [34]. Similarly as anticipated, patients with higher DMFT reported poorer QoL. This is in accordance with a study from New Zealand [34] but in contrast to other studies on European population [35, 36]. These contrasting findings might be due to the cultural differences between the study regions. In addition to validity and reliability, sensitivity of the outcome measure to changes is an essential requirement [37]. We believe that assessing OHRQoL change in patients with a specific form of cancer undergoing a specific treatment is more clinically relevant than using a more heterogeneous sample; therefore, a homogenous group of cancer patients undergoing radiotherapy was included at least in the first instance. Additional studies are needed to investigate the responsiveness of the questionnaire in other patient groups undergoing different treatments. In accordance to the literature where poorer QoL has been reported in patients receiving radiotherapy over time [38–40], significantly poorer scores in all the subscales except the “information” subscale were
Table 9 Responsiveness of Persian QLQ-OH17 to radiotherapy in a subsample of Iranian head and neck cancer patients
Pain/discomfort Xerostomia Eating Information Worried about future Worn dentures Ill-fitting denture
Factor 2
Factor 3
Factor 4
0.77 0.76 0.62 0.72 0.58
33.25 %
19.55 %
12.98 %
0.60 0.93 0.92 12.58 %
observed in our subsample of head and neck cancer patients. This is due to the adverse effects of radiotherapy on oral tissues in head and neck cancer patients which include mucositis, decreased salivary function, infections, radiation caries, loss of taste, and radionecrosis of the jaw [41]. Although the sample size was large enough, the study is limited by the fact that we used subjective perceptions of patients to evaluate the concurrent validity as there is no gold standard OHRQoL instrument for cancer patients. Further, it is desirable to evaluate the measurement invariability of the instrument in relation to other known factors. In reference to the strengths of the study, we have assessed responsiveness to treatment and factorial invariance in addition to routine psychometric evaluations of validity and reliability. Along with the internal consistency, reliability on repeated administrations was also evaluated in this study. In conclusion, the results of the study indicated that Persian version of the QLQ-OH17 is a valid and reliable tool for assessing OHRQoL in Iranian patients with various cancers. QoL scores differed between the categories of self-perceived
Before treatment mean (SD)
During treatment mean (SD)
p value
Effect size
42.24 (19.90) 45.41 (20.67) 47.35 (18.10) 17.07 (12.65) 49.72 (17.33) 45.44 (18.78) 51.81 (24.16)
46.72 (20.73) 53.98 (22.23) 55.20 (21.49) 17.23 (12.89) 62.66 (20.30) 59.29 (22.71) 57.66 (28.97)
ORIGINAL ARTICLE
Oral-health-related quality of life in patients with cancer: cultural adaptation and the psychometric testing of the Persian version of EORTC QLQ-OH17 Mir Saeed Yekaninejad & Amir H. Pakpour & Jyothi Tadakamadla & Santhosh Kumar & Seyed Hamzeh Mosavi & Bengt Fridlund & Andrew Bottomley & Neil K. Aaronson
Received: 27 May 2014 / Accepted: 1 October 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose To evaluate the validity, reliability, responsiveness to treatment, and gender invariance of the Persian version of the European Organization for Research and Treatment of Cancer Oral Health Questionnaire (QLQ-OH17) among Iranian cancer patients. Methods Cancer patients (n=729) from three oncology centers in Tehran and Qazvin were recruited. A forward and backward translation procedure was performed to develop a M. S. Yekaninejad Department of Epidemiology and Biostatistics, School of Public Health, Tehran University of Medical Sciences, Tehran, Iran A. H. Pakpour (*) Social Determinants of Health Research Center, Qazvin University of Medical Sciences, Shahid Bahounar BLV, Qazvin 3419759811, Iran e-mail: [email protected] A. H. Pakpour e-mail: [email protected] J. Tadakamadla : S. Kumar Population and Social Health Research Program, Griffith Health Institute, School of Dentistry and Oral Health, Griffith University, Gold Coast, Australia S. H. Mosavi Department of General Surgery, Hazrat-e-Rasoul Hospital, Iran University of Medical Sciences, Tehran, Iran B. Fridlund School of Health Sciences, Jönköping University, Jönköping, Sweden A. Bottomley Quality of Life Department, European Organization for Research and Treatment of Cancer, Brussels, Belgium N. K. Aaronson Department of Psychosocial Counseling, The Netherlands Cancer Institute, Amsterdam, The Netherlands
culturally acceptable version of Persian QLQ-OH17. Internal consistency and test–retest reliability of the QLQ-OH17 was assessed. In addition, convergent and discriminant validity, concurrent validity, construct validity, and known-groups validity were evaluated. The factor structure of the questionnaire was examined by exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). Responsiveness to change was measured in an independent sample of patients with head and neck cancer undergoing radiotherapy. Finally, factorial invariance of the QLQ-OH17 was assessed across gender. Results The Persian version of QLQ-OH17 showed good internal consistency (Cronbach’s alpha coefficients of 0.71– 0.83) and reliability on repeated administration (intraclass correlation coefficients of 0.85–0.94). Persian QLQOH17 exhibited the original four-factor structure. Patients who perceived good oral health and satisfaction with their mouth reported significantly better oralhealth-related quality of life (OHRQoL) than those who perceived poor oral health and dissatisfied with their mouth. Similarly, those who perceived a need for dental treatment reported significantly poorer OHRQoL than those who have not perceived any treatment need. Older patients, females, and those experiencing greater caries had poorer QoL than their comparative counterparts. All QLQ-OH17 subscales were correlated with QLQ-C30 subscales and global QoL. Both male and female patients with cancer interpreted items on the QLQ-OH17 in a similar manner. The QLQ-OH17 was found to be responsive to treatment in a sample of head and neck cancer patients. Conclusions The Persian version of QLQ-OH17 is a valid and reliable questionnaire for assessing OHRQoL in Iranian patients with various cancers. Keywords Translation . Oral health . Quality of life . Cancer . Psychometrics . Questionnaire
Support Care Cancer
Introduction According to the Global Burden of Cancer Study, cancer is the leading cause of deaths worldwide with 8.2 million deaths in 2012. More than 60 % of new cases in the world are accumulated in Africa, Asia, and Central and South America [1]. In Iran, the data from the Global Burden of Diseases, Injuries, and Risk Factors Study 2010 suggest that cancer is the second most common reason of death and more than 57 % of total deaths are caused by cardiovascular diseases and cancers [2]. WHO defines quality of life (QoL) as an individual’s perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards, and concerns [3]. Currently, the multidimensional concept of QoL which includes behavioral competence and health, perceived quality of existence, and psychological well-being [4] has become an increasingly important issue in cancer patients [4–7]. The importance of QoL in cancer patients becomes more pronounced with increasing survival rates which involves intermediate and late treatment effects of cancer that might affect physical, psychological, and social well-being [8]. The reason for patient-reported QoL outcomes gaining importance in cancer patients is because of the varied symptoms and functional disabilities experienced by cancer patients which cannot be evaluated using routine laboratory tests or imaging procedures [9]. In addition, cancer survivors might be subjected to oral complications which diminish their QoL [10, 11]. These oral complications involve alterations in saliva function and taste sensation, infections and necrosis of the jaw [11], mucosal inflammation, ulceration, and oral candidiasis and bleeding [11]. Moreover, oral cavity has been documented as the most common source of sepsis in immunosuppressed patients with cancer [12]. The European Organization for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQC30) is one of the most widely used cancer health-related quality of life (HRQoL) instrument that has been developed by EORTC. The QoL Group and QoL Department of EORTC are international leaders in the development of patientreported outcome measures in oncology [13, 14]. More recently, health care professionals from various disciplines at Oslo University in Norway have developed a supplementary module “EORCTC QLQ-OH17” to the EORCTC QLQ-C30 for oral and QoL assessment in cancer patients. This is the only cancer-specific OHRQoL instrument which is a brief assessment tool focusing on oral health problems and QoL in cancer patients [15]. QLQ-OH17 is currently undergoing international field testing for cross-cultural applicability, validity, and reliability. Responsiveness to treatment was also evaluated in this study as an outcome measure to be considered as useful must demonstrate responsiveness to treatment effects [16].
The current paper evaluates validity, reliability, responsiveness to treatment, and gender invariance of this module in Persian language among Iranian cancer patients.
Materials and methods Study population Cancer patients attending three oncology centers in Tehran and Qazvin during the period June 2013 to December 2013 were recruited. Patients were eligible for inclusion in the study if they were 18 years of age or older, histologically diagnosed with cancer, were able to read and speak Persian, had life expectancy of at least 4 weeks, and able to provide written consent. Patients who are too old and weak, having cognitive impairment, participating in an oral health program or clinical trial study, and those suffering with psychiatric disorders were considered for exclusion. The study was approved by the Ethics Committee of the Qazvin University of Medical Sciences. Written informed consent was obtained from all the patients. Instruments Questionnaires were administered to all the participants. In addition to EORTC QLQ-C30 and QLQ-OH17, questionnaire also consisted of questions on self-perceived oral health, selfperceived satisfaction with mouth, and perceived need of dental treatment. The EORTC QLQ-C30 The EORTC QLQ-C30 (version 3.0) is a generic measure to assess HRQoL among patients with cancer [13]. It consists of 30 items which covers five functional scales (physical functioning, role limitation, emotional, cognitive, and social), three symptom scales (fatigue, nausea or vomiting, and pain), six single items (dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties), and a global health and QoL scale. All items are rated on a four-point Likert scale while global health and QoL are rated on seven-point Likert scale. QLQ-C30 consists of both multi-item scales and single-item measures, and total score for each scale is calculated. A raw score by estimating the average of items in the scale is calculated which is later transformed linearly to get a standardized score, which ranges from 0 to 100. Higher scores in functional scale and global health status/QoL indicate better QoL while higher scores in symptom scale indicate higher symptomatology [17].
Support Care Cancer
The EORTC QLQ-C30 has been translated into several languages including Persian and has been found to be highly valid and reliable [18]. The EORTC QLQ-OH17 The EORTC QLQ-OH17 is a specific tool to assess OHRQoL in cancer patients [15]. The EORTC QLQ-OH17 is composed of four multi-item scales (pain/discomfort, xerostomia, eating, and information) and three single items (two items related to use of dentures and one on future worries). All these items are rated on four-point Likert-type scale ranging from 1 (not at all) to 4 (very much) except one item related to denture usage which has a dichotomous response (yes/no). Higher scores on pain/discomfort, xerostomia, and eating represent a high level of symptomatology or problems while higher scores on information indicate higher information level [15]. The scoring procedure for QLQ-OH17 is similar to that of EORTC QLQ-C30 which is described in the EORTC scoring manual. Translation procedure Translation from the English original EORTC QLQ-OH17 into Persian was done in according to the EORTC manual proposed by the EORTC QoL study group [19]. Permission from the EORTC QoL department for translating QLQ-OH17 into Persian was sought by the principal investigator (AP). The aim of the translation procedure was to produce a Persian version of the EORTC QLQ-OH17 which is conceptually equivalent to the original English version. The translation procedure was initiated with forward translation. In this step, two translators proficient in English, who were native Persians, independently translated QLQ-OH17 from English to Persian. Later, a single reconciled version was prepared by the principal investigator (AP) and the translators through resolving the discrepancies between the translations. This interim Persian version was then translated back into English by two native English speakers who were blinded to the original English version. Back translation was conducted independently by the translators. The principal investigator (AP) compared the English translations with the original version to ensure consistent translation. In the next step, the interim translated version was pilot-tested on 18 patients (mean age=54.12, range=29–61) with various types of cancers (head and neck, breast, lymphoma/leukemia, and gastrointestinal). A structured interview was administrated to check if the items were confusing, upsetting, and difficult in terms of understanding or wording. It was found that all the patients were able to read and understand the items with no difficulty. Therefore, no specific change was made in the final version. At final stage, the piloted Persian version of the EORTC QLQ-OH17 was administrated on 678 Iranian patients with cancer for psychometric assessment.
Procedure Eligible patients were invited to attend an introduction session to explain study aims. Afterwards, the patients were asked to sign a consent form and complete the questionnaires in clinic rooms. Demographic information (age, educational status, marital status, and occupation) and clinical information were also collected via the patients’ medical records. Two calibrated dentists conducted oral examinations in the same clinics. A total of 678 patients were subjected to oral examination to record caries experience using decayed, missing, and filled teeth (DMFT) index [20], and WHO criteria was followed for recording caries [21]. The questionnaire was re-administrated to all the patients whose condition was stable (no expected clinical change) after 2 weeks from baseline assessment. For evaluating the responsiveness of the instrument to treatment, an independent sample of head and neck cancer patients were selected from the same hospitals in Tehran and Qazvin with similar inclusion and exclusion criteria. Only those patients that were newly diagnosed with head and neck cancer and were supposed to undergo radiotherapy were considered for inclusion. They were asked to complete the questionnaire at baseline and 2 months after the initiation of radiotherapy. Statistical analysis Statistical analyses were performed using statistical software PASW Statistics 18 (SPSS Inc., Chicago, IL) and LISREL 8.80. Response distributions were assessed for floor (the proportion of patients with the lowest possible scores) and ceiling effects (the proportion of patients with the highest possible scores). Floor or ceiling effects are considered to be present if more than 15 % of patients scored the lowest or highest scores on the scale, respectively [22]. The reliability of the QLQOH17 was evaluated using internal consistency and test–retest reliability at repeated administrations with a 2-week interval. Internal consistency was assessed using Cronbach’s alpha coefficient, and Cronbach’s alpha values of 0.7 or more were considered to be acceptable [23]. Test–retest reliability of the QLQ-OH17 was evaluated by calculating the intraclass correlation coefficient (ICC), and an ICC ≥0.70 was considered to be acceptable [23]. Convergent and discriminant validity of the QLQ-OH17 was evaluated using multi-trait scaling analysis [24]. The multi-trait scaling analysis was conducted to examine correlations between items and hypothesized scales (corrected for overlap) [25, 24]. Pearson’s correlation coefficients were used for multi-trait analyses in order to deflate skewness, kurtosis, and categorization of variables. Item convergent validity was accepted if item-own scale correlation was equal or higher than 0.40 [24]. It was also hypothesized that item-own scale correlation was higher than item-other scale (twice the standard error) correlation [24].
Support Care Cancer
To the best of our knowledge, there is no gold standard existing to evaluate the OHRQoL in cancer patients. Therefore, to evaluate concurrent validity of the QLQOH17, a series of multiple linear regression analyses with four subscales of the QLQ-OH17 as dependent variables and subjective perceptions of patients as independent variables was conducted. It was hypothesized that patients who are dissatisfied with their mouth, perceive poor oral health, and need for dental treatment report poor OHRQoL. In addition, ability of the QLQ-OH17 to differentiate between subgroups of patients was evaluated using independent t test. To evaluate factor structure of the QLQ-OH17, the whole sample was randomly split into two separate groups. The two groups were comparable in terms of demographic and clinical variables. An exploratory factor analysis (EFA) with principal component analysis (PCA) and Varimax rotation was conducted on the first half (n=339) to explore underline factor structure of this tool. The suitability of the EFA was evaluated using two indices including the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and the Bartlett’s test of sphericity. KMO higher than 0.7 and statistically significant Bartlett’s test of sphericity were considered to check if the data are factorable. A confirmatory factor analysis (CFA) was then performed on the second half (n=339) to ensure clear distinctions between the factors. The model parameters were estimated by the weighted least squares with polychoric correlations and asymptotic covariance matrices. Model fit was evaluated using various indices including chi-squared statistic, Bentler’s comparative fit index (CFI), non-normed fit index (NNFI), the root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Model fit was considered to be acceptable when χ2 was non-significant (p < 0.05), RMSEA was ≤0.08, SRMR was ≤0.08, NNFI was ≥0.90, and CFI was ≥0.90. In order to investigate whether the QLQ-OH17 is measuring same constructs across gender groups, a multiple-group CFA analysis using total sample was conducted. A two-level hierarchical procedure as proposed by Milfont and Fischer was used. At the first level, it is considered that patients belonging to both the gender groups conceptualize the constructs in the same way (configural invariance) while at the second level, all the factor loadings are assumed to be equal across the gender groups (metric invariance). Differences in CFI (ΔCFI ≤0.01) and NNFI (ΔNNFI ≤0.01) were used to demonstrate factorial invariance across groups. Responsiveness of the QLQ-OH17 instrument to clinical changes was tested using paired t tests (two-tailed). Effect sizes (Cohen’s d statistic) were computed for assessing the magnitude of differences over time.
Results A total of 729 patients with cancer were initially assessed for eligibility in the study. Seven hundred one patients meeting the inclusion criteria were invited to participate, of which 678 agreed to participate with a response rate of 97 %. Table 1 demonstrates the demographic and clinical characteristics of the study population. Six of ten participants (61.2 %) were women, and the mean age was 54.42 years (SD=13.9). Most of the patients (36.0 %) were educated until “secondary school” followed by illiterate group who constituted more than a quarter of the study population. Breast and colorectal cancers were the two most common types of cancers observed in the study population. Table 2 demonstrates that all the subscale and single-item scores of Persian QLQ-OH17 were normally distributed. No floor or ceiling effects were observed for any of the QLQ-OH17 subscales and single items. Internal consistency of the QLQ-OH17, as assessed by Cronbach’s Table 1 General characteristics of the study population Age in years mean (SD) DMFT mean (SD) Gender Male Female Educational status Illiterate Primary school Middle school Secondary school College Marital status Single
54.42 (13.90) 10.11 (6.25) n (%) 263 (38.8 %) 415 (61.2 %) n (%) 189 (27.9 %) 98 (14.5 %) 63 (9.3 %) 244 (36.0 %) 84 (12.4 %) n (%) 58 (8.5 %)
Married Widowed/divorced Occupational status Employed Unemployed Type of cancer Breast Stomach Colorectal Lymphoma/leukemia Head and neck Multiple myeloma Cervical Ovarian Renal cell/kidney Prostate Other
456 (67.3 %) 164 (24.2 %) n (%) 327 (48.2 %) 351 (51.8 %) n (%) 193 (28.5 %) 97 (14.3 %) 81 (12.0 %) 49 (7.2 %) 38 (5.6 %) 36 (5.3 %) 33 (4.9 %) 31 (4.6 %) 29 (4.3 %) 22 (3.2 %) 69 (10.2 %)
DMFT decayed, missing, and filled teeth index
Support Care Cancer Table 2 Descriptive statistics for the Persian version of QLQ-OH17
Pain/discomfort Xerostomia Eating Information Worried about future Worn dentures Ill-fitting denture
Number of forms
Mean (SD)
Floor N (%)
Ceiling (%) Cronbach’s alpha
Normality (Kolmogorov–Smirnov test)
663 671 677 673 675 655 541
49.08 (23.29) 45.15 (20.80) 52.78 (18.71) 27.43 (23.5 %) 33.29 (19.2) 50.32 (50.1) 47.14 (22.14)
8 (4.1) 9 (1.3) 17 (2.5) 159 (23.5) 48 (7.1) – 21 (3.1)
3 (1.5) 88 (13.0) 68 (10.0) 11 (1.6) 82 (12.1) – 76 (11.2)
0.123 0.201 0.112 0.102 0.09 – 0.142
alpha statistic, ranged from 0.71 to 0.83. Table 3 depicts that the ICCs for the subscales of the QLQ-OH17 ranged from 0.85 (eating) to 0.94 (xerostomia). Multi-trait scaling analysis provided support for scaling assumptions for all dimensions as depicted in Table 4. Itemscale correlations exceeded 0.40 for all the items, and no scaling errors were seen. Moreover, all QLQ-OH17 subscales were significantly correlated with QLQ-C30 subscales and global QoL scores (Table 5). Pain/discomfort and eating subscales were significantly related to all the subscales of EORTC QLQ-C30 while xerostomia did not significantly correlate with social functioning and dyspnea. Information was not related to physical functioning, social functioning, pain, fatigue, nausea or vomiting, and financial impact subscales of QLQ-C30. Table 6 presents the results from multiple linear regression analyses that were conducted to assess concurrent validity of the QLQ-OH17. The results indicated that patients who perceived good oral health and satisfaction with their mouth reported significantly better OHRQoL than those who perceived poor oral health and dissatisfied with their mouth. Similarly, those who perceived a need for dental treatment reported significantly poorer OHRQoL than those who have not perceived any treatment need. Female and older patients generally reported significantly poorer OHRQoL than male and younger patients for each QLQ-OH17 subscale. Moreover, patients with greater caries experience (higher DMFT index scores) reported significantly poorer OHRQoL for all the QLQ-OH17 subscales than those with lower caries experience (Table 7). Principal components analysis was conducted on the combined 14 items (excluding single items) to assess factor structures. Bartlett’s test of sphericity and KMO as criteria for EFA Table 3 Test–retest reliability of the Persian version of QLQ-OH17
Scale
ICC (95 % CI)
Pain/discomfort Xerostomia Eating Information
0.876 (0.855–0.894) 0.938 (0.928–0.947) 0.853 (0.828–0.875) 0.880 (0.859–0.897)
0.83 0.76 0.71 0.74 – – –
analysis was checked to verify factorability of the data. The results indicated that Bartlett’s test of sphericity was statistically significant (approximate χ 2 = 1,339.56 df = 91, p1 which accounted for 78.4 % of the variance in OHRQoL (Table 8). CFI of the Persian version of QLQ-OH17 using the four domain original version loadings of the 14 items resulted in a good fit: χ2 = 122.47, df = 71, p < 0.004, CFI = 0.97, RMSEA = 0.075, SRMR=0.057, and NNFI=0.96. Standardized regression weights for all items with their respective domains were
Table 4 Multi-trait scaling analysis of the subscales of Persian QLQOH17 Pain/ Xerostomia Eating Information discomfort Pain/discomfort Pain in gums Bleeding gums Pain in lips Problems with teeth Sore mouth Sores in mouth corners Xerostomia Dry mouth Sticky saliva Eating Sensitive mouth Taste change Problems with solid food Trouble enjoying meals Information Receiving any information about oral problems Satisfaction with information
0.718 0.785 0.876 0.610
0.247 0.187 0.323 0.123
0.262 0.424 0.352 0.348
0.036 0.151 0.048 0.165
0.783 0.676
0.335 0.147
0.386 0.111
0.142 0.180
0.241 0.349
0.722 0.931
0.382 0.411
0.063 0.018
0.067 0.297 0.148
0.241 0.241 0.311
0.766 0.815 0.876
0.087 0.097 0.142
0.151
0.099
0.884
0.144
0.095
0.123
0.150
0.945
0.190
0.013
0.178
0.943
>0.6 except two items (“problems with solid food” and “trouble enjoying meals”). The four factor structure of the QLQ-OH17 was tested across gender groups to evaluate measurement invariance. The results of multiple-group CFIs showed that both models (configural and metric invariance) yielded comparable fit in the factorial invariance tests across gender groups. At the first level, results indicated invariance of the factor structure of QLQ-OH17 across gender groups (configural invariance): χ2 =180.3, df=133, pMean value (n=211)
≤Mean value (n=467)
30.87 (18.72) 38.71 (21.32) 48.60 (23.95) 16.06 (7.58)
56.66 (27.38) 56.01 (19.59) 66.80 (24.62) 13.74 (7.33)
60.00 (27.38) 63.99 (23.30) 63.40 (21.81) 12.93 (7.71)
33.15 (27.59) 26.37 (16.32) 44.32 (19.04) 32.88 (26.24)
51.31 (20.72) 65.76 (24.92) 61.08 (24.09) 16.65 (6.11)
28.21 (25.38) 28.70 (18.61) 34.76 (17.86) 31.02 (12.23)
DMFT decayed, missing, and filled teeth index *Statistically significant according to Benjamini–Hochberg procedure for gender **Statistically significant according to Benjamini–Hochberg procedure for age categories †Statistically significant according to Benjamini–Hochberg procedure for DMFT
Support Care Cancer Table 8 Results obtained from exploratory factor analysis (n= 339) of QLQ-OH17
Item
Factor 1
Pain in gums Bleeding gums Pain in lips Problems with teeth Sore mouth
0.76 0.77 0.62 0.78 0.84
Sores in mouth corners Dry mouth Sticky saliva Sensitive mouth Taste change Problems with solid food Trouble enjoying meals Receiving any information about oral problems Satisfaction with information Eigenvalue Explained variance (%)
0.87
The difference in OHRQoL between the genders might be due to the difference in influences during the life course [34]. Similarly as anticipated, patients with higher DMFT reported poorer QoL. This is in accordance with a study from New Zealand [34] but in contrast to other studies on European population [35, 36]. These contrasting findings might be due to the cultural differences between the study regions. In addition to validity and reliability, sensitivity of the outcome measure to changes is an essential requirement [37]. We believe that assessing OHRQoL change in patients with a specific form of cancer undergoing a specific treatment is more clinically relevant than using a more heterogeneous sample; therefore, a homogenous group of cancer patients undergoing radiotherapy was included at least in the first instance. Additional studies are needed to investigate the responsiveness of the questionnaire in other patient groups undergoing different treatments. In accordance to the literature where poorer QoL has been reported in patients receiving radiotherapy over time [38–40], significantly poorer scores in all the subscales except the “information” subscale were
Table 9 Responsiveness of Persian QLQ-OH17 to radiotherapy in a subsample of Iranian head and neck cancer patients
Pain/discomfort Xerostomia Eating Information Worried about future Worn dentures Ill-fitting denture
Factor 2
Factor 3
Factor 4
0.77 0.76 0.62 0.72 0.58
33.25 %
19.55 %
12.98 %
0.60 0.93 0.92 12.58 %
observed in our subsample of head and neck cancer patients. This is due to the adverse effects of radiotherapy on oral tissues in head and neck cancer patients which include mucositis, decreased salivary function, infections, radiation caries, loss of taste, and radionecrosis of the jaw [41]. Although the sample size was large enough, the study is limited by the fact that we used subjective perceptions of patients to evaluate the concurrent validity as there is no gold standard OHRQoL instrument for cancer patients. Further, it is desirable to evaluate the measurement invariability of the instrument in relation to other known factors. In reference to the strengths of the study, we have assessed responsiveness to treatment and factorial invariance in addition to routine psychometric evaluations of validity and reliability. Along with the internal consistency, reliability on repeated administrations was also evaluated in this study. In conclusion, the results of the study indicated that Persian version of the QLQ-OH17 is a valid and reliable tool for assessing OHRQoL in Iranian patients with various cancers. QoL scores differed between the categories of self-perceived
Before treatment mean (SD)
During treatment mean (SD)
p value
Effect size
42.24 (19.90) 45.41 (20.67) 47.35 (18.10) 17.07 (12.65) 49.72 (17.33) 45.44 (18.78) 51.81 (24.16)
46.72 (20.73) 53.98 (22.23) 55.20 (21.49) 17.23 (12.89) 62.66 (20.30) 59.29 (22.71) 57.66 (28.97)
