Introduction

In recent years, there is an increasing interest in the measurement of health-related quality of life (HRQoL) of populations suffering from chronic diseases. In the case of HIV infection, the introduction of highly active antiretroviral therapy (HAART) changed the disease’s natural course, marking a considerable decline in disease mortality, in the incidence of opportunistic infections, the time of hospitalization, and the time interval until progression to AIDS [1]. With the use of antiretroviral drugs, AIDS is considered to be a chronic disease, and the clinicians’ efforts are focused in the accomplishment of long-term suppression of viral load, and the patients’ compliance to treatment which in some instances is low due to adverse drug reactions [2]. Under these circumstances, the recognition of strategies which will maximize this population’s HRQoL is inevitable as is the appreciation of the diseases aspects in patients’ lives when framing therapeutic strategies [3, 4].

Numerous studies have been conducted on HIV+ patients’ HRQoL, particularly in relation to medical (i.e. disease stages, viral burden, disease outcome, and therapeutic regimens) and social or demographic variables. The initial discord about the use of generic or disease-specific instruments for the measurement of HRQoL in chronic conditions [5] was followed by the development of disease-specific tools. In HIV-infected patients, most HRQoL instruments are multifactorial subjective scales that aim mainly in detecting the impact of the infection and its therapy in certain dimensions of everyday life and its possible changes during the disease’s natural course or the application of pharmaceutical regimens [5, 6]. The MOS-HIV instrument is a brief and comprehensive tool of HRQoL measurement in HIV+ and AIDS patients and contains 35 questions grouped into 10 health dimensions of HRQoL: general health (GH), physical functioning (PF), mental health (MH), role functioning (RF), social functioning (SF), vitality (VT), pain (PN), cognitive functioning (CF), health distress (HD), health transition (HT), and quality of life (QL) [7, 8]. The scores obtained for each dimension allow two global summary scores to be obtained (i.e. PHS: physical health summary score and MHS: mental health summary score). For the PHS score, the PF, PN, and RF scale scores contributed most strongly. For the MHS score, the MH, HD, QL, and CF scales contributed most strongly. The VT, GH, and SF scales contributed to both factors. The summary scores were transformed to t-scores with a mean of 50 and a standard deviation of 10 and were standardized to make it possible to compare between scales with different response formats, ranging from 0 to 100, with higher scores indicating better HRQoL [8, 9].

The MOS-HIV has been shown to be internally consistent, to correlate with concurrent measures of health (e.g. symptom scales), to discriminate among distinct groups (e.g. patients in different disease stages), to predict future outcomes (e.g. prediction of increased hazard of death in HIV+ individuals with CD4+ <100 cells/μl) and to be responsive to changes over time [8–10]. No research has been carried out on the assessment of HRQoL of HIV/AIDS patients in Greece, mainly due to the great lack of measurement instruments in Greek that have been developed or adapted against agreed scientific criteria and attributes. The present study aims at evaluating the reliability and validity of the translated in the Greek language MOS-HIV questionnaire in an urban population of HIV-infected individuals of different disease stages.

Methods

Participants and data collection

The sample consisted of HIV-positive men and women, of various disease stages, patients of the clinic of Infectious Diseases of a tertiary hospital in Athens, Greece. The patients, after being informed about the study’s content and goals, were asked to participate during their regular follow-up visit at the clinic. The inclusion criteria were age over 18 years and known HIV infection—with positive Elisa and Western Blot tests—for at least 3 months. Among the 185 randomly selected and eligible individuals, a number of 154 gave written informed consent to participate in the study (response rate 83.2%). The patients completed the self-reported questionnaire, during their visit at the clinic, usually before their scheduled doctor’s appointment. Additional information about socio-demographic data was also collected. Clinical data regarding HIV infection status were obtained from their medical files.

Instrument

The MOS-HIV was developed in response to the need for a comprehensive, rapid assessment of HIV/AIDS-specific HRQoL. Since 1997, it has been used in clinical trials [8] and cross-sectional and cohort studies (i.e. from PROMETHEUS [11] and PISCES study groups [12] to other investigators of HRQoL in HIV [2, 13], and it has numerous validated translations [14–19]). This questionnaire shares the same common pool of items with the SF-36 of the Medical Outcomes Study, but some additional domains of HRQoL measures were added to produce a 30-item form of the MOS-HIV [7]. In subsequent efforts to increase reliability of the initial MOS-HIV, 35 questions were included to assess 10 dimensions (or scales) of health, which included GH (5 items), PN (2 items), PF (6 items), RF (2 items), SF (1 item), MH (5 items), VT (4 items), HD (4 items), CF (4 items), QL (1 item), and HT (1 item) [8]. To score the MOS-HIV, the raw item scores of each scale were summed and transformed into a 0–100 scale, permitting comparisons between the constructs of with different response categories. The questionnaire takes about 5 min to complete. A higher score indicated better HRQoL.

Translation

The MOS-HIV used in this study was translated into Greek from the original questionnaire using international guidelines [20–23] and the internationally accepted translation method (i.e. the standard linguistic validation process) suggested by The Mapi Research Institute [24]. The standard linguistic validation process started with the conceptual analysis of the original instrument’s items. Upon agreement, two professional translators, native speakers of the Greek (i.e. target) language and fluent in the English (i.e. source) language, undertook independent forward translations into the target language. A reconciled version of the instrument was developed, and a backward translation of this reconciled version back into the original language was performed by a professional translator. The back-translation and the original one were compared, and any discrepancies between them led to changes to the reconciled translation in the Greek language. The clinic’s medical doctors participating in the research study reviewed this latest version and gave their feedback. At the next stage, the questionnaire was administered to a small group of the clinic’s patients who volunteer to take part at the cognitive debriefing phase in order to assess clarity and comprehension of the items. After this final feedback, the final Greek version was produced. The items and the categories of the Greek translation of MOS-HIV were in the same grouping and order, as in the original English version.

Socio-demographic and HIV/AIDS-related parameters

Social and demographic elements regarding age, sex, ethnicity (i.e. Greek/non-Greek), education (i.e. first, second, or third level of education), monthly income, current employment status, reception of disability benefits, and use of intravenous drugs were answered by the patients in a separate sheet. HIV/AIDS-related information that included a recent CD4+ cell count, the Karnofsky Performance Status, the disease stage (asymptomatic HIV, symptomatic HIV, or AIDS), the time since diagnosis, and the existence of other infections or chronic conditions that may influence their HRQoL was abstracted from their recent medical files.

Statistical analysis

Internal consistency reliability was determined by the calculation of Cronbach’s α coefficient. Scales with reliabilities equal to or greater than 0.70 were considered acceptable [25]. The floor and ceiling effects were also tested based on the percentages of scores at extremes of the scaling range in order to provide evidence that the scale has adequate variability in a range that is appropriate to its intended use [25]. Subscales are intended to contribute to the definition of a unique construct (i.e. HRQoL) yet at the same time to reflect distinct constructs. For this reason, convergent validity was tested through intercorrelations among the scales belonging to the same questionnaire. The values for intercorrelation (Pearson’s r) between scales should be between 0.40 and 0.80 [25]. Discriminant construct validity was evaluated by analyzing the association between scores and CD4+ cell count and disease stages. Based on previous literature [7, 14, 15, 26–28], it was anticipated that subjects with low CD4+ cell count would report lower values on the MOS-HIV scales. We also hypothesized that asymptomatic HIV+ patients would have the highest mean MOS-HIV scale scores, symptomatic HIV+ patients would have intermediate mean MOS-HIV scale scores, and AIDS patients would have the lowest mean MOS-HIV scale scores. We expected differences between disease stages would be greater for MOS-HIV scales that mainly measure physical health status (i.e. PHS, VT, PN, RF, and GH).

Analysis of variance (ANOVA) was used to compare mean scales between the three groups of patients (asymptomatic, symptomatic, and AIDS). In order to control for multiple testing, Bonferroni correction was used. The MOS-HIV scales were tested for their discriminative ability using receiver operating characteristic (ROC) curves. The overall performance of the ROC analysis was quantified by computing area under the curve (AUC). An area of 1 indicated perfect performance, while 0.5 indicated a performance that was not different than chance.

Results

Sample characteristics are presented in Table 1. Participants were generally of male gender (76.6%) and had Greek nationality (89.6%). A 50.6% was in asymptomatic disease stage. The mean CD4+ cell count was 516.2 cells/mm3, and 86.4% of the subjects had CD4+ cell count more than 200 cells/mm3. HIV-1 RNA level had a medial of 49.5 copies/mL.

Table 1 Demographic and clinical characteristics of the HIV—infected subjects
Full size table

Table 2 lists the summary of scales’ scores and floor and ceiling effects. Scores were generally well distributed over the range of possible values for the single scales. Floor effects ranged from 0.6% for CF, PN, MH, MHS, and PHS to 22.7% for RF scales. Ceiling effects ranged from 0% for MHS and PHS to 46.1% for PN.

Table 2 Descriptive characteristics of the MOS-HIV scales
Full size table

Internal consistency reliability was good for each scale, with the α value always greater than 0.80. All the MOS-HIV scales exceeded the minimum reliability standard of 0.70 (Table 3). PF and HD had the greatest reliability coefficient, equal to 0.87 and 0.88, respectively.

Table 3 Internal consistency reliability (Cronbach’s α) and intercorrelations of MOS-HIV scales
Full size table

Convergent and discriminant construct validity at the scale level were tested to determine whether the scales reflect different constructs of HRQoL. Correlations among MOS-HIV scales were all significant. The convergent validity analysis revealed that most of the correlations between pairs of scales were within the acceptable range (0.40–0.80). Lower correlations were observed between the HT scale and the scales for HD, CF, PN, GH, and VT, as identified also in another study [15]. PN scale had a low correlation (<0.30) with the RF and CF scales. The GH scale showed acceptable correlations with all scales except for CF, RF, and HT scales.

We measured discriminant construct validity for each scale (i.e. the level the scale scores could discriminate according to severity of the disease), using as validation variables CD4+ count and stage of disease severity [29]. We expected lower scale scores with more severe disease and with low level of CD4+ count [7, 26–28, 30].

Table 4 presents the mean scale scores for asymptomatic, symptomatic, and AIDS patients. Scores on GH, PF, PN, and PHS scales were significantly lower for AIDS patients compared to asymptomatic ones. Furthermore, scores on PF and PN were also significantly lower for AIDS patients compared to symptomatic HIV+ ones.

Table 4 Comparisons of scores on MOS-HIV scales between the different disease stages
Full size table

Table 5 indicates the ability of the MOS-HIV scales to discriminate between subjects that are characterized by differences in severity of disease, as defined by CD4+ cell count. All scales except for RF and HT could discriminate well subjects with CD4+ cell count less than 200 cells/mm3 and more than 200 cells/mm3. The AUC ranged from 0.67 for MH scale to 0.82 for GH scale.

Table 5 Discriminative ability of MOS-HIV scales on the basis of clinical characteristics
Full size table

Discussion

The Greek translation of the MOS-HIV instrument was found to be suitable and appropriate to assess HRQoL of HIV-infected individuals residing in Greece. These data provide evidence for applicability in patients in different stages of disease severity, as recorded in other similar international studies [14, 16, 25].

All scales exceeded the minimum reliability criterion of a Cronbach’s α coefficient over 0.70. The validity of this instrument was also supported by the observed relationship of HRQoL scale scores and CD4++ lymphocyte count and disease stages.

The scales for MH, HD, and HT displayed well-balanced distributions, with a slight skew toward higher scores for the PF, RF, SF, and PN scales. The ceiling effect was observed in the scales of SF and PN. This phenomenon is well explained by the fact that half of the patients (50.6%) were in asymptomatic HIV disease stage, and most of them were not receiving antiretroviral treatment at the time of the survey. Results of prior studies have also demonstrated that ceiling effects in the PF, RF, SF, CF, PN, and HT and in patients with more advanced disease stage, this was pronounced in the CF, PN, and HT scales [30]. In addition, the RF scale of the MOS-HIV has been found to have floor effects in patients with advanced HIV disease [16, 17, 28, 30].

All scales displayed satisfactory convergent validity except for RF scale. In particular, this scale showed correlations below the accepted level, except for the VT scale. It is possible that the translated version of one of the items may not have functioned as well as in the original English version, a finding also reported in the Italian psychometric trial [15]. Schiffano et al. [15] argued that the items on the RF subscale may not be adequate for describing the conditions of persons with AIDS, who are more likely to be incapable of conducting normal everyday activities. The SF scale also had acceptable correlation coefficients with respect to all scales, and the GH scale correlated well with all scales except for CF and HD.

Previous studies have shown that the MOS-HIV scales define, through a principal factor analysis, two correlated dimensions: a physical health dimension and a mental-emotional dimension [8, 9, 14]. As expected, scales that assess physical health, including PF and PN, were more strongly associated with severity of HIV disease [29]. We also observed that the above scales could discriminate well between individuals in different disease stages, as seen in numerous other studies of the MOS-HIV [7, 8, 26, 28, 30, 31]. The most significant comparisons in mean scores with the analysis of variance could be made between asymptomatic HIV and AIDS patients in the areas of GH, PF, and PN scales. Asymptomatic HIV+ patients exhibited much higher scores in the domains of GH, PF, and PN compared to symptomatic HIV+ and to AIDS patients.

Better HRQoL scores were reported to be seen in the HIV-infected individuals with higher CD4+ cell counts [16, 17, 32]. We also observed that individuals with more CD4+ had higher HRQoL scores particularly in the domains related to the physical health as described by Wu et al. [7], and other international validation studies [18] have previously reported. On the contrary, Paton et al. [19] had found that the differences between the disease stages and correlations with CD4+ cell counts extended to nearly all the subscales of Physical and Mental Health. In the Italian and UK English samples, CD4+ cell count was associated with physical and social/role-related facets of HRQoL [14].

Our study also supports the absence of a significant association between scores on Mental Health domains and stages of HIV infection, a common conclusion of some previous studies [8, 33]. In general, we would expect AIDS patients to report with worse HRQoL than patients with asymptomatic or symptomatic HIV infection, the pattern observed for most of the comparisons in this study. However, the unpredictable course of HIV infection allows an AIDS diagnosis on the basis of an opportunistic infection that subsequently may resolve, rendering the patient asymptomatic. In this case, an AIDS patient might be experiencing less somatic pain than many symptomatic HIV+ individuals. Similarly, AIDS patients’ psychological and social sense of well-being may not be as poor as for patients who have recently been diagnosed as seropositive [15].

In interpreting our results, some important limitations of the study must be taken into account. The analysis of scores variability over time was not performed, and the results of our study demonstrate the need to further investigate the performance of the scales among persons with HIV/AIDS, especially with regard to their responsiveness to changes over time. Longitudinal studies also need to be undertaken to assess the impact of introduction of ΗAART on disease progression and changes in HRQoL.

Our results compared favorably to those from previous studies, leading to the conclusion that the Greek translation of MOS-HIV questionnaire represents a useful tool for measuring data on the HRQoL, inclusively among Greek-speaking patients with AIDS. However, changes in some of the subscales, such as role functioning, might improve performance.