Discussion
In Turkey, standardized QoL tool for the assessment of disease burden in CVID patients is lacking. In the current study which had high response rate and positive response from CVID patients we validated the Turkish version of CVID QoL questionnaire and its psychometric properties. It showed excellent reliability, good content validity and reproducibility.
Concerning reliability, our results revealed that all items had excellent internal consistency (>0.9) and 2 subscales, EF and RF exhibited good internal consistency as well (>0.7). These findings are in agreement with the results (0.82, 0.84) of the original Italian version 10 and similar ( 0.91, 0.77) to the Norwegian cultural adaptation study 19. The GSS subscale did not achieve the acceptable internal consistency. It consists of only 4 items, 2 of them are related to the diarrhea, 1 skin diseases and 1 dietary change and these 4 elements were not very related to each other. This may be one of the reasons for the low internal consistency. When we consider only two items (4 and 14) which directly deal with bowel symptoms, it exhibited good internal consistency similar to the findings of the Norwegian study 19. Another possible reason we considered was that our sample group was small to establish construct validity 25. In addition, cutaneous problems are not seen as often as gastrointestinal manifestations. Generally autoimmune skin problems and case based cutaneous diseases are seen 26-28. On the other hand, Ballow et al. developed and published a new disease specific tool for primary antibody deficiencies. It did not include any question about skin problems 29. Therefore, we may consider that dermatologic features do not have an important impact on QoL of CVID patients, but more comprehensive studies are necessary to indicate this. Additionally, similar to the findings from the index study and cultural adaptation 10, 19 test–retest reliability results indicated that the Turkish CVID QoL scale (CVID QoL TR) also exhibits excellent short-term stability. This indicated that outcomes from the CVID QoL TR were reproducible, supporting its potential use as a patient-reported outcome tool.
Content validity is the ability of a tool to determine the area of ​​interest and the conceptual definition of a structure25. During the determination of content validity, we found our CVI was acceptable. But the content validity ratio could not reach the value of 0.7 in 6 items for at the first stage. Minor changes were then made in the 6 items and the main structure was maintained. The last version of tool was approved. We considered that these findings contributed to the content validity.
Convergent validity assesses the extent to which a questionnaire/ tool measures what it is designed to measure 30. It is estimated by correlating its items with other validated questionnaires measuring the same or similar constructs. To examine the convergent validity of the CVID-QoL-TR, we used SF-36 as a comparative tool. SF-36 is a well-known general QoL scale, translated and validated in Turkish language and used in various diseases 31, 32. Good correlations were found between QoL global, EF and RF subscales of the CVID-QoL-TR with certain items of the SF-36. CVID QoL scores correlated strongly with both SF-36’s physical and mental health domains. Quinti et al. showed good convergent validity for the EF and RF subscales correlating with conceptually similar dimensions of SF-3610. Andersen et al. reported the similar findings with/to WHQOOL BREF19. Discriminant validity is a statistical concept assessing the ability of a tool/questionnaire to detect true differences and discriminate between the other tools or changes. It indicates that the two things/measure that should not be related are actually irrelevant 25. Our results showed that the QoL, EF, RF scores were higher in the patients complaining of more than one infection within 3 months before the study. Quinti et al reported that the frequency of infections both within 3 months and 12 months before the study had an impact on the quality of life. We did not observe this association within 12 months before the study. We can speculate that this might be related to the questionnaire seeking answers to questions about the last 3 months and 12 months is a longer duration to recall.
Factor analysis is a multivariate statistics that obtains to find a small number of conceptually significant new variables (factors, dimensions) by combining a large number of related variables intended to measure the same structure or a particular property33, 34 More accurate factor analyses stated that sample size should have at least 3-5 times more number of items 35. In the current study, we could not verify the factor analysis since our sample size did not have large number of participants. However, we could perform factor analysis for GSS and RF subscale because they have 4 and 9 items respectively. We observed GSS subscale item 4 and14 related with the bowel symptoms were distinguished from theitem 2 ‘dietary changes’ and 26 ‘skin symptoms. In RF subscale, item 11 ‘run out of medications’ and item 16 ‘as contagious’, item 6 ‘cough’ and item 25 ‘limitation of leisure activity’ were distinguished from the other items in the RF dimension. But EF and QoL had more items than 50-participant-group could verify these factors. Although factor analysis does not confirm 3 factors. Good correlation with SF 36, reliability, reproducibility and high response rate showed us that CVID QOL TR is a useful scale. We can believe that factor analysis can be re-evaluated as the instrument will be used in the future.
We observed that being female were negatively associated with QoL. This finding was similar to the information of other CVID QoL studies10, 16, 18. Receiving IVIG treatment was the second factor associated with poor QoL, it was also consistent with the previous studies10, 16. We observed better quality of life in the patient with more than 13 years of education similar to the findings of the Italian and Norwegian group. We did not observe any association between BMI, age and QoL in our study. It might be related to the ethnic differences. Generally, we could not compare directly the findings of our study between the Italian and Norwegian study groups. Our QoL scores did not normally distributed, but the Italian and Norwegian groups showed that their findings were normally distributed (Table 3.) Our study group achieved similar mean CVID QoL scores with Norwegian group while higher than Italian group. 43.7 % of all replies were 0 and our floor and ceiling effects showed better QoL. Since these effects were not evaluated in the index study so we could not compare the findings totally. Differences in the results of our study group between the other groups may be explained with the variation in the demographic features of the study groups. Our study group had higher proportion of male and younger patients as well as the education levels of our patients were lower than the participants of other study groups. Furthermore, it might be associated with low socioeconomic status or other cultural differences that could not be differentiated in the disease specific tools. CVID QoL instrument is a disease specific questionnaire. It could not measure the impact the social requirements or economic or psychologic situations. Finally, we believe it is not suitable for comparison.
Our study had some limitations. One of them was the low number of adult CVID patients included in the study though we are one of the largest centres in Turkey. Therefore, the analysis could not be properly done to verify the factor analyses and structural validity. Second, there is also the possibility of recall bias, since patients were asked to report on their health in the past 3 months.
In conclusion, CVID disease specific questionnaire is necessary to better evaluate the disease burden on the patients Our study indicated that Turkish version of CVID QoL questionnaire was a reliable, useful and valid instrument for the measuring of quality of life in CVID patients. It is recommended to investigate its stability by applying it to larger patient groups and further consideration on factor analysis. In addition to that, future evaluation of QoL in CVID either with this CVID QoL TR in other Turkish patients or also other translations to other languages can facilitate the improve the knowledge about CVID disease burden on individually.