2.2.1

Patient recruitment

The final Turkish version was tested on a consecutive series of patients at the Department of Otorhinolaryngology–Head and Neck Surgery and Department of Radiation Oncology (Hacettepe University, Faculty of Medicine) between September 2006 and February 2008. The study included patients newly diagnosed as having head and neck cancer. Patients with recurrent or second primary head and neck cancers were excluded from the study. Patients were required to speak and read Turkish. Eligible patients were invited to participate in the study and signed an informed consent form approved by the Faculty of Medicine Ethics Committee, Hacettepe University. The study was also approved by the ethics committee. Age, sex, tumor site, TNM tumor stage, histologic tumor type, and treatment data were recorded as the demographic and clinical characteristics of the study sample.

2.2.2

Instruments and procedures

The patients were asked to complete 3 sets of questionnaires: the first set was given 1 day before beginning treatment; the second set, 3 months after the completion of treatment, and the third set, 10 days after the second was completed. Because the acute effects of treatment typically diminish by 3 months, we administered the second set of questionnaires 3 months after completion of treatment to evaluate the impact of treatment. The third set was administered 10 days after the second set to measure test-retest reliability because 10 days was considered a sufficient interval to ensure that the patients would not remember their responses to the second set of questionnaires. None of the patients in the study underwent any treatment during this 10-day interim period.

The first and second sets of questionnaires included the Turkish EORTC QLQ-C30 and the Turkish UW-QOL. The third set included the Turkish UW-QOL only. The Turkish versions of UW-QOL–version 4 and EORTC QLQ-C30–version 2.0 were used in the study. Performance status was assessed and rated by a physician using the Karnofsky and ZEW performance scales, synchronous with the first and second sets of questionnaires.

2.2.3

Questionnaire scoring

Each item on the UW-QOL (version 4) is scored from 0 to 100. Higher scores indicate better quality of life. A “composite score” is obtained by calculating the mean of the 12 items. There are also 3 “global questions” concerning overall quality of life, which are analyzed separately. These 3 global questions are not used in the composite score .

The EORTC QLQ-C30 scoring manual was used to calculate the item scores of the EORTC QLQ-C30 version 2.0 . The EORTC QLQ-C30 is a 30-item questionnaire composed of both multi-item scales and single-item measures. These include 5 functional scales, 3 symptom scales, 1 global health status/quality of life scale, and 6 single items. A high score functional scales represents a high/healthy level of functioning, and a high score on the global health status/quality of life represents a high quality of life, but a high score on a symptom scale/item represents a high level of symptoms .

The Karnofsky score ranges from 100 to 0, where 100 is “perfect” health and 0 is death. Although the score has been described with intervals of 10, a physician may choose decimals if he or she feels a patient’s situation falls between 2 point scores .

The ZEW (Zubrod/ECOG/WHO) score ranges from 0 to 5, with 0 denoting perfect health and 5 denoting death .

2.2.4

Statistical analysis

Reliability was evaluated by measuring both the internal consistency (Cronbach α ) and test-retest reliability (interclass correlation coefficient [ICC]) after a 10-day interval devoid of interim treatment. Internal consistency and reliability of the multi-item scales were assessed using Cronbach α coefficient, a measure of the internal consistency of a psychometric instrument. As a rule of thumb, internal consistency is considered good if α approximates .70 but does not exceed .90, because values greater than 0.90 imply the presence of redundant items . Test-retest reliability was measured with ICC, which considers not just the strength of the correlation but also systematic variations . A test-retest reliability coefficient greater than 0.4 is considered acceptable to justify discriminative use of quality of life scales .

There are 3 forms of validity: content, criterion, and construct. Content validity of the Turkish UW-QOL was ensured qualitatively by the translation process, as the authors made sure that the translated questionnaire was appropriate for Turkish patients with head and neck cancer; it was also assured by the rigorous process of forward-backward translation. Criterion validity considers whether a questionnaire compares favorably to a criterion standard, but this is typically less relevant in quality of life research because there is no criterion standard measure of quality of life. Construct validity is present if a scale behaves according to hypothesized relationships. One of the methods for establishing construct validity is to compare a questionnaire’s score against other variables. For example, we hypothesized that the Turkish UW-QOL score would decrease as the T stage of the disease advanced. Another way of assessing the construct validity would be to statistically compare UW-QOL score with EORTC QLQ-C30 score. A statistically significant relationship between UW-QOL score and disease stage would indicate clinical validity of the translated questionnaire. In the present study, to establish construct validity of the Turkish UW-QOL, we evaluated the relationship between UW-QOL composite score and the following:

• 1.
  

  EORTC QLQ-C30 scores

  
  
• 2.
  

  ZEW Performance Scale score

  
  
• 3.
  

  Karnofsky Performance Scale score

  
  
• 4.
  

  T stage of the disease

  
  
• 5.
  

  Global UW-QOL score

Spearman ρ values were used to compare correlations between the variables. Kruskal-Wallis, Jonckheere-Terpstra, and χ ² tests were used for comparisons between groups. To compare preoperative and the postoperative variables, the McNemar test was used as a statistical method. All analyses were performed using SPSS v.12.0 for Windows (SPSS, Chicago, IL). The level of statistical significance was set at P < .05.

2.2.1

Patient recruitment

The final Turkish version was tested on a consecutive series of patients at the Department of Otorhinolaryngology–Head and Neck Surgery and Department of Radiation Oncology (Hacettepe University, Faculty of Medicine) between September 2006 and February 2008. The study included patients newly diagnosed as having head and neck cancer. Patients with recurrent or second primary head and neck cancers were excluded from the study. Patients were required to speak and read Turkish. Eligible patients were invited to participate in the study and signed an informed consent form approved by the Faculty of Medicine Ethics Committee, Hacettepe University. The study was also approved by the ethics committee. Age, sex, tumor site, TNM tumor stage, histologic tumor type, and treatment data were recorded as the demographic and clinical characteristics of the study sample.

2.2.2

Instruments and procedures

The patients were asked to complete 3 sets of questionnaires: the first set was given 1 day before beginning treatment; the second set, 3 months after the completion of treatment, and the third set, 10 days after the second was completed. Because the acute effects of treatment typically diminish by 3 months, we administered the second set of questionnaires 3 months after completion of treatment to evaluate the impact of treatment. The third set was administered 10 days after the second set to measure test-retest reliability because 10 days was considered a sufficient interval to ensure that the patients would not remember their responses to the second set of questionnaires. None of the patients in the study underwent any treatment during this 10-day interim period.

The first and second sets of questionnaires included the Turkish EORTC QLQ-C30 and the Turkish UW-QOL. The third set included the Turkish UW-QOL only. The Turkish versions of UW-QOL–version 4 and EORTC QLQ-C30–version 2.0 were used in the study. Performance status was assessed and rated by a physician using the Karnofsky and ZEW performance scales, synchronous with the first and second sets of questionnaires.

2.2.3

Questionnaire scoring

Each item on the UW-QOL (version 4) is scored from 0 to 100. Higher scores indicate better quality of life. A “composite score” is obtained by calculating the mean of the 12 items. There are also 3 “global questions” concerning overall quality of life, which are analyzed separately. These 3 global questions are not used in the composite score .

The EORTC QLQ-C30 scoring manual was used to calculate the item scores of the EORTC QLQ-C30 version 2.0 . The EORTC QLQ-C30 is a 30-item questionnaire composed of both multi-item scales and single-item measures. These include 5 functional scales, 3 symptom scales, 1 global health status/quality of life scale, and 6 single items. A high score functional scales represents a high/healthy level of functioning, and a high score on the global health status/quality of life represents a high quality of life, but a high score on a symptom scale/item represents a high level of symptoms .

The Karnofsky score ranges from 100 to 0, where 100 is “perfect” health and 0 is death. Although the score has been described with intervals of 10, a physician may choose decimals if he or she feels a patient’s situation falls between 2 point scores .

The ZEW (Zubrod/ECOG/WHO) score ranges from 0 to 5, with 0 denoting perfect health and 5 denoting death .

2.2.4

Statistical analysis

Reliability was evaluated by measuring both the internal consistency (Cronbach α ) and test-retest reliability (interclass correlation coefficient [ICC]) after a 10-day interval devoid of interim treatment. Internal consistency and reliability of the multi-item scales were assessed using Cronbach α coefficient, a measure of the internal consistency of a psychometric instrument. As a rule of thumb, internal consistency is considered good if α approximates .70 but does not exceed .90, because values greater than 0.90 imply the presence of redundant items . Test-retest reliability was measured with ICC, which considers not just the strength of the correlation but also systematic variations . A test-retest reliability coefficient greater than 0.4 is considered acceptable to justify discriminative use of quality of life scales .

There are 3 forms of validity: content, criterion, and construct. Content validity of the Turkish UW-QOL was ensured qualitatively by the translation process, as the authors made sure that the translated questionnaire was appropriate for Turkish patients with head and neck cancer; it was also assured by the rigorous process of forward-backward translation. Criterion validity considers whether a questionnaire compares favorably to a criterion standard, but this is typically less relevant in quality of life research because there is no criterion standard measure of quality of life. Construct validity is present if a scale behaves according to hypothesized relationships. One of the methods for establishing construct validity is to compare a questionnaire’s score against other variables. For example, we hypothesized that the Turkish UW-QOL score would decrease as the T stage of the disease advanced. Another way of assessing the construct validity would be to statistically compare UW-QOL score with EORTC QLQ-C30 score. A statistically significant relationship between UW-QOL score and disease stage would indicate clinical validity of the translated questionnaire. In the present study, to establish construct validity of the Turkish UW-QOL, we evaluated the relationship between UW-QOL composite score and the following:

• 1.
  

  EORTC QLQ-C30 scores

  
  
• 2.
  

  ZEW Performance Scale score

  
  
• 3.
  

  Karnofsky Performance Scale score

  
  
• 4.
  

  T stage of the disease

  
  
• 5.
  

  Global UW-QOL score

Spearman ρ values were used to compare correlations between the variables. Kruskal-Wallis, Jonckheere-Terpstra, and χ ² tests were used for comparisons between groups. To compare preoperative and the postoperative variables, the McNemar test was used as a statistical method. All analyses were performed using SPSS v.12.0 for Windows (SPSS, Chicago, IL). The level of statistical significance was set at P < .05.