- Top of page
Objective: To assess the reliability and validity of a Spanish version of the LDQOL 1.0 (Liver Disease Quality of Life questionnaire).
Methods: Observational, cross-sectional study in Spanish patients awaiting liver transplantation (LT). Feasibility was assessed by analyzing administration times and missing responses. Ceiling and floor effects were calculated and reliability was tested by examining internal consistency (Cronbach's alpha). Convergent validity was tested by examining correlations between LDQOL disease-specific and Short Form health survey with 36 questions (SF-36) dimensions. Known groups' validity was tested by examining the LDQOL's capacity to discriminate between groups defined by etiology and Child–Turcotte–Pugh (CTP) scores.
Results: A total of 200 patients were included for analysis. Mean age (SD) was 52.6 (9.8) years and 73% of the sample were male. The most common indication for LT was liver cancer (34%). Mean (SD) time to complete the questionnaire was 35.8 minutes (21.2 minutes). Missing responses were highest on the dimensions of sexual functioning and symptoms of liver disease. Ceiling effects were over 20% on 7 of the LDQOL's 12 disease-specific scales. Cronbach's alpha coefficients were over 0.70 on all but 2 dimensions. Correlations between SF-36 and LDQOL disease-specific dimensions generally fulfilled the hypotheses, with 35 of the 40 highest and lowest correlations (87.5%) being in the expected direction. The LDQOL discriminated well between patients in CTP class A and C, and as hypothesized, hepatocarcinoma and alcoholic cirrhosis patients scored better on most dimensions than patients with hepatitis C virus or other etiologies.
Conclusions: The Spanish version of the LDQOL 1.0 has shown satisfactory reliability and validity.
- Top of page
Advanced liver disease impacts both patient survival and quality of life (QOL) , although several treatments, including liver transplants, have been shown to lead to substantial improvements on both of these outcomes [2–4]. The standardized measurement of patients' health-related QOL (HRQOL) therefore has an important role to play in the assessment and management of liver disease patients [5,6]. For that reason, several questionnaires have been developed to measure the impact of liver disease and its treatment on patients' HRQOL [7–9]; one of the most complete of the currently available instruments is the Liver Disease QOL Questionnaire (LDQOL 1.0) .
The LDQOL 1.0 was designed to be a disease-specific, self-report measure for patients with advanced, chronic liver disease . It includes the generic SF-36 health survey as well as 75 disease-specific items that measure HRQOL in 12 dimensions. Testing of the original US English version of the questionnaire showed the instrument to be reliable and valid for use in pretransplant liver disease patients and patients submitted for liver transplant evaluation. A short version of the LDQOL developed more recently has also proven to be responsive to changes over time as well as predicting mortality in patients with advanced liver disease [10,11].
A preliminary validation of an adapted Spanish version of the LDQOL was performed in posttransplant patients , but it has not been validated for use in pretransplant patients, which was the original target group for the instrument. It is important to validate instruments in different populations because the specific characteristics of those groups may affect psychometric performance. Likewise, when HRQOL questionnaires are adapted for use in different languages and cultural contexts, the new language version will require revalidation to ensure that adaptation has not affected instrument performance [13,14].
Earlier studies with the LDQOL found that there was only a low to moderate correlation between LDQOL 1.0 scores and clinician-derived indices of severity such as the Child–Turcotte–Pugh (CTP) and Model for End Stage Liver Disease (MELD) scores . Fewer studies have examined in detail the relationship between LDQOL scores and demographic and clinical variables such as age, gender, and etiology. More detailed knowledge of the influence of these variables on LDQOL scores would be helpful in designing future studies and in interpreting results.
The objectives of the present study are therefore to validate the Spanish version of the LDQOL 1.0 questionnaire in pretransplant liver disease patients and to analyze the impact of a range of clinical and sociodemographic variables on LDQOL 1.0 scores.
- Top of page
This study has shown the Spanish version of the LDQOL to be a reliable and valid instrument for use in Spanish patients with advanced chronic liver disease awaiting transplant. In an earlier study, we had examined the properties of the Spanish version in post-transplant patients and found that the questionnaire was also generally reliable in this group of patients . This instrument adds to disease-specific instruments already adapted for use in Spain, such as the Chronic Liver Disease Questionnaire .
Although the LDQOL is a relatively long questionnaire, the rate of missing responses was generally low, presumably at least in part because the questionnaire was interview-administered. The main exception was the dimension of sexual problems, which had a high rate of missing responses primarily because a relatively high proportion of the patients reported that they had not had sexual relations in the past 4 weeks. Some of the patients were also reluctant to respond to items in this dimension in the presence of an interviewer. The questionnaire also took a considerable time to administer (mean of 35.8 minutes), although it is a very complete instrument covering a wide range of relevant dimensions for liver disease and transplanted patients. It also incorporates the generic SF-36, a fact which should be borne in mind in studies where both generic and disease-specific measures of HRQOL are required. The short version of the LDQOL developed recently might provide a useful alternative where the long version is impractical .
The LDQOL showed substantial ceiling effects on a number of the disease-specific scales. This is likely because in a substantial proportion of the sample, the primary indication for transplant was HCC, and these patients have been shown to score relatively well on generic HRQOL measures . High ceiling effects on the loneliness and quality of social interaction dimensions could stem from a cultural effect because of the presence of strong social and family networks in Spain. Comments from the patients during cognitive debriefing also indicated that the stigma dimension may not be wholly relevant for the Spanish patients because many found the idea of being embarrassed or self-conscious about the disease unusual and alien. The dimensions of loneliness, hopelessness, stigma, sexual functioning, and sexual problems also showed the highest ceiling effects in the original version (18.7%, 12.3%, 18.2%, 24.3%, and 24.1% ), suggesting that the items included may not adequately capture the full range of effects and/or that the dimensions may not be as relevant to patients as some others. Such effects are important as they may affect the instrument's ability to reflect improvement over time.
With regard to the high ceiling effect seen on the sexual problems (women) scale, this may have been due to a reluctance to acknowledge problems in this area of their lives with an interviewer present. It is also possible that scale items were not relevant for the entire sample for the time period covered (4 weeks) because frequency of sexual intercourse has been shown to decrease in women with nonalcoholic liver disease . However, we were not able to ascertain whether this was the case in our sample, although this is certainly an aspect of using the LDQOL that would warrant further research.
In terms of reliability, the Spanish version of the LDQOL was generally satisfactory in the present study. The Cronbach's alpha values were over the recommended 0.70 threshold  for 8 of the 12 disease-specific dimensions, and close to 0.70 on the remaining dimensions. On the SF-36, the 0.70 threshold was met on all but one of the eight dimensions. This means the LDQOL is suitable for use at the group level; very few of the dimensions, however, had an alpha value of 0.90 or over so they would not be suitable for use with patients at an individual level. Interestingly, the four dimensions that showed the poorest results on Cronbach's alpha in the present study also had some of the lowest alphas on the original version, although they were all over the threshold of 0.70 except for quality of social interaction . Further work may be required to bolster reliability in these scales.
In the development of the original LDQOL, factor analysis was used to investigate the scale structure of that version . In the present study, as our aim was to determine whether the Spanish instrument would be sufficiently reliable and valid for use in research settings, we preferred to investigate whether the LDQOL dimensions formed coherent scales through the calculation of Cronbach's alpha. As all of the alphas for the disease-specific dimensions were very close to or above 0.70, which is the generally recommended threshold for use of this type of instrument at group level , we considered that to be a sufficient indication of reliability to suggest that the instrument could be used in research settings in Spain.
Likewise, it should be noted that all of the interviews were carried out by the same psychologist who was involved in the study from the outset (LJC), so interrater reliability was not an issue here. However, this would be an aspect that would warrant research in future studies.
With regard to convergent validity, we found that the LDQOL disease-specific dimensions correlated as expected with individual SF-36 dimensions and summary scores. Correlations were particularly strong between LDQOL 1.0 disease-specific dimensions and the MCS and SF-36 dimensions that measure psychological and social functioning, indicating the considerable focus on psychological and emotional aspects of the illness in the LDQOL 1.0. Only a small number of the hypotheses regarding the likely pattern of correlations were not met, and the deviations from the hypotheses were generally minor, suggesting good convergent validity for the Spanish measure.
On the other hand, correlations with the MELD score were generally low to moderate, although strongest with the physical functioning dimensions of the SF-36. This likely reflects the aspects measured by the MELD and is similar to findings observed with the original version of the instrument . Likewise, we found that the dimensions that best discriminated between groups defined by the CTP category were symptoms of liver disease, effects of liver disease, and sexual functioning, which also showed some of the strongest correlations with the CTP score in the original validation study , although correlations between clinician-derived measures and HRQOL have generally been found to be weak [15,29]. The fact that sexual problems did not discriminate well between the groups in our study may be due to the high number of missing responses.
Our analysis of the known groups' validity suggests that a number of dimensions on the LDQOL Spanish version distinguished adequately between the patients defined by clinical categories. In particular, the instrument was successful at discriminating between the patients defined by CTP class (see Fig. 1), although we also found that HCV was associated with considerably worse HRQOL on most of the LDQOL disease-specific dimensions than either HCC or LC ALC. Although these differences were only statistically significant on four dimensions (symptoms, effects of liver disease, concentration, and sleep), they are nevertheless in line with results from similar studies that showed that patients with HCV tend to have worse HRQOL than patients with other etiologies [30,31]. The present results also complement those found in previous studies that found differences between cirrhotic and noncirrhotic patients .
On the other hand, our analysis of the effect of sociodemographic variables showed that age and gender generally had only a small effect on LDQOL 1.0 scores. The poorer score on the social interaction dimension in men may be due to the disease leading to greater restrictions on social activities or a tendency to be more withdrawn, perhaps because of a greater impact on self-image though this has been little explored in the literature. The poorer score in men on the sexual functioning dimension does receive some support in the literature, which suggests that men with CTP grades B and C have significant sexual dysfunction and significant reduction of both total and free testosterone levels . This contrasts with findings in women that suggest that liver disease does not have a significant impact on sexual satisfaction or desire [27,34].
In general, the results of validity testing support the use of the instrument in future studies in Spain, although further testing is still required in larger, more heterogeneous samples, and in longitudinal studies to test the instrument's sensitivity to change. Likewise, additional information, such as the magnitude of a minimal clinically relevant difference on the different dimensions would help to improve interpretability .
The study had some limitations. Firstly, although the sample was relatively large, the study was performed in only one center and the sample included a high proportion of HCC and male patients. Both of these characteristics limit the generalizability of the results. The inclusion of a large number of HCC cases reflects the case mix on the waiting list at this center . It should also be noted that at least some of the impact on HRQOL is likely to be due to the underlying cirrhosis rather than the tumor itself, which tends to be asymptomatic. The questionnaire was administered by interview and the results might differ if the questionnaire was self-administered. Future studies should examine whether there are systematic differences in the scores using the two different modes of administration. Finally, we were not able to examine the instrument's sensitivity to change as these are results from a cross-sectional study. Analysis of this aspect is currently ongoing for the Spanish version of the LDQOL. Finally, we did not impute values for missing responses as the proportion of missing responses was low in most of the scales, and they were not concentrated in particular items within the scales. In the present study, the relatively small proportion of missing responses would not have overly biased results. However, if higher proportions of missing responses are observed or if they are more systematic, then other methods of imputing scores should be considered, such as simple mean imputation, or more sophisticated techniques such as regression imputation .
In conclusion, the Spanish version of the LDQOL 1.0 provides a reliable and valid means of measuring outcomes in patients with liver disease awaiting transplant. The fact that it is a very complete instrument means that administration time is considerable and the new, shorter version may prove more practical for much clinical research or clinical practice. The considerable ceiling effects on a number of dimensions also suggest its performance in longitudinal studies should be carefully evaluated.