Measuring quality of life among people living with HIV: a systematic review of reviews – Health and Quality of Life Outcomes

Figure 1 shows the PRISMA flowchart. The electronic database search identified 544 papers. After removal of duplicates, 278 abstracts were subjected to review. After applying our inclusion and exclusion criteria, 27 articles were obtained and subjected to full text review. Of those, 9 met the inclusion criteria. One further study was identified through the reference list search.

Fig. 1

PRISMA Flow diagram showing paper selection process

Full size image

Overview of the published systematic reviews

Characteristics of the reviews are summarised in Table 2. The search periods for the published systematic reviews addressed the period 1995–2013 (6 studies did not record the dates of the search and one study did not record the start date). Four reviews were restricted to abstracts/papers published in the English language [3, 10,11,12], one was restricted to papers published in English or Spanish [17], one had no language restrictions [18] and the remaining 4 reviews did not state whether or not the review was restricted by language [4, 13, 19, 20]. The number of articles reviewed ranged from 26 to 852 (six did not specify the number of articles reviewed). The number of generic HRQoL measures identified by the reviews ranged from 0 to 23 (median = 7) and the number of HIV specific measures identified ranged from 1 to 18 (median = 11).

Table 2 Summary of reviews identified

Full size table

Quality of the reviews

The results of the quality assessment [16] are shown in Table 3. Eight of the ten reviews clearly stated a well formulated research question. Four of the ten reviews had predefined and specified their inclusion and exclusion criteria, and 4 had conducted a comprehensive search including multiple scientific literature databases and manual searches of references. None of the reviews stated that at least two researchers had conducted an independent review of titles, abstracts and full text review and indicated how disagreements between the reviewers were resolved. No reviews included a quality assessment. Two reviews listed the studies they included along with descriptions of their key characteristics. None of the reviews assessed publication bias. The criterion ‘was heterogeneity assessed’ [16] was not applicable because none of the reviews included a meta-analyses.

Table 3 Quality assessment applied to the identified reviews

Full size table

Selection of measures for further scrutiny

Generic measures

Twenty-nine generic HRQoL measures were reviewed against the inclusion criteria (Fig. 2). Of those, 10 were not comprehensive, 2 were not self-administered, 7 took longer than 10 min to complete/ had 40 or more items and 1 had no normative data and were therefore excluded. Nine generic measures met the inclusion criteria and were retained for further analysis. These were the COOP/WONCA charts [21], EQ-5D [22, 23], FLZM Questions on Life Satisfaction [24], HUI [25], McGill Quality of life questionnaire [26], SF-12 [27], SF-20 [28], SF-36 [29,30,31] and WHOQOL-BREF [32, 33].

Fig. 2

Flow diagram showing selection of generic measures

Full size image

HIV-specific measures

Twenty-three HIV-specific HRQoL measures were reviewed against the inclusion criteria (Fig. 3). Of those, 3 were not comprehensive, 9 took longer than 10 min to complete/ had 40 or more items and there was no evidence of patient input in the development of 4 of the measures. Seven HIV-specific measures therefore met the inclusion criteria and were retained for further analysis. These were the ACTG SF-21 [34], HIV-QL31 [35], MOS-HIV [36], MQoL-HIV [37], PROQOL-HIV [38], WHOQOL HIV-BREF [39] and HIV-SQUAD [40].

Fig. 3

Flow diagram showing selection of HIV-specific measures

Full size image

Overview of selected generic HRQoL measures

Properties of each of the generic HRQoL measures are summarised in Table 4.

Table 4 Qualities and psychometric properties of the selected generic measures extracted from the reviews

Full size table

The COOP/WONCA charts contain 6–9 items including the following domains: physical fitness, social activities, feelings, change in health, daily activities and overall health. Each item has five response options, presented pictorially. They are available in several languages and have been found to be acceptable to patients, however very little psychometric data were available in the reviews and they have not been widely used with people with HIV. Therefore, insufficient evidence is available to determine the suitability of this instrument for use in HIV research or clinical practice.

The EQ-5D consists of 5 items encompassing five dimensions of quality of life (mobility, self-care, usual activities, pain/discomfort, anxiety/depression) and an optional visual analogue scale to rate overall health. It takes approximately 1 min to complete and is available in multiple languages. General population preference weights have been derived for many countries, enabling the use of the measure in economic analyses. The reviews provided evidence of construct and convergent validity, as well as responsiveness to treatment initiation, the development of opportunistic infections and adverse effects with small to medium effect sizes [4, 10, 12, 13]. The measure has been frequently used in research with people with HIV, and several authors recommended it for use in this population [4, 10, 13]. However, problems with ceiling effects were noted [4, 20] therefore the use of the scale in individuals with early, asymptomatic HIV infection was not recommended [4, 19]. Wu et al. (2013) recommended the use of EQ-5D alongside an HIV-specific measure (the MOS-HIV) in order to obtain HIV-specific quality of life alongside this utility measure [10].

Health utilities Index [25]

The Health Utilities Index is available in two versions (HUI mark 2 (HUI2) and HUI mark 3 (HUI3)), which consist of 15/16 items respectively and assess 7 domains – vision, ambulation, dexterity, emotion, cognition, hearing, speech and pain. The measures are available in multiple languages. Although the scales have not been extensively used in HIV research, there is some evidence for their validity in this context. HUI2 and HUI3 have been associated with disease severity, AIDS-related events and viral load. [4, 12] Construct validity has been established by correlating HUI scales with the MOS-HIV [4, 20]. The scales have also been found to be responsive to change in HIV disease states [12, 20] however, both the MOS-HIV and EQ-5D have been found to have better discriminatory capacity [12]. Based on emerging data, Clayson et al. (2006) considered the HUI to be a potentially useful adjuvant to an HIV specific measure [4].

The McGill quality of life questionnaire (M-QOL) [26]

The McGill Quality of Life Questionnaire comprises 16 items encompassing four domains: support, existential well-being, physical and psychological symptoms. Skevington et al. report that the face/content validity of the measure is improved by the inclusion of an existential dimension, which may be particularly relevant to people with HIV who have advanced disease (CD4 count <100) [20]. However, only scores for one subscale (physical symptoms) distinguished between people with HIV who had high and low CD4 counts, indicating poor discriminant validity of the other subscales [20]. There was a lack of information in the reviews on responsiveness or reliability. This measure has not been frequently used with people with HIV. There was therefore insufficient information available to determine the suitability of this instrument.

The SF-12 is one of three generic measures (along with the SF-20 and SF-36) from the Medical Outcomes Study (MOS) that met our inclusion criteria. It consists of 12 items in 8 domains: physical functioning, role-physical, role-emotional, bodily pain, general health, vitality, social functioning and mental health, allowing for the generation of physical and mental health summary scores. There were little validity and reliability data available in the reviews and mixed results were presented in terms of responsiveness to change in treatment [12]. Since the SF-12 shares items with other measures developed from the MOS, it was considered likely that the scale had similar floor and ceiling effects to other MOS scales [17]. Clayson et al. recommended the use of the SF-12 in place of the SF-36 where the length of the SF-36 is a problem [4].

The SF-36 comprises 36 items encompassing 9 domains: physical functioning, role-physical, bodily pain, general health, vitality, social functioning, role-emotional, mental health and reported health transition allowing for the generation of physical and mental health summary scores. The scale has been translated into several languages and takes approximately 10 min to complete. Scores on the SF-36 can be used in economic analyses by deriving utility weights [11]. The original SF-36 [28] was modified to improve the range and precision of some of the scales, improve the wording and format of the questionnaire, resulting in the SF-36 v2 [41], however the reviews largely failed to distinguish between the two versions. As a result, this review refers only to the SF-36 and does not specify whether the findings relate to Version 1 or Version 2. The reviews reported good to acceptable internal consistency and construct validity [13, 20]. People with HIV reported lower HRQoL on all dimensions compared to general population controls [13, 20]. Scale scores have been associated with treatment duration, less comorbidity and better social support [13]. The SF-36 has been found to be responsive to the initiation of ART, change in CD4 count, viral load and symptoms [4, 12, 20] however it may not be sensitive to change in ART medication in patients who are stable on ART [12]. Problems with floor and/or ceiling effects have been reported on some subscales [20]. The SF-36 was recommended for use in people with HIV in two of the reviews [4, 18], however a criticism of this and other MOS measures was that they were developed in US and translated into other languages without the input of patients to ensure conceptual and semantic equivalence, and therefore may not be relevant for use in cross-sectional research [20].

The WHOQOL-BREF was developed from the WHOQOL-100 instrument, which was produced within an international collaboration of 15 countries, using a spoke-wheel methodology to ensure conceptual and semantic equivalence across languages and cultures [20]. The instrument has been frequently used in people with HIV. It consists of 26 items over 4 domains: physical health, psychological health, social relationships and environment. It is available in 40 languages and takes less than 5 min to complete [20]. Good psychometric properties, including Cronbach’s alpha coefficients in the acceptable range, correlations with disease severity and CD4 count were reported [12, 20], however data on responsiveness were not available in the reviews.

Two additional instruments, the FLZM [24] and the SF-20 [28], were identified (Table 4), however the reviews did not assess their psychometric properties. There was insufficient information available to determine the suitability of these instruments.

Overview of selected HIV specific HRQoL measures

Properties of each of the HIV-specific HRQoL scales are summarised in Table 5.

Table 5 Qualities and psychometric properties of the selected HIV-specific measures extracted from the reviews

Full size table

The ACTG-21 consists of 21 items encompassing 7 HRQoL domains: physical functioning, energy/fatigue, social functioning, cognition, pain, health perception and emotional wellbeing. The reviews did not report reliability or validity data for the scale but floor and ceiling effects were anticipated given that the scale shares items with other MOS measures [17]. The measure has not been widely used in people with HIV. There was therefore insufficient information available to determine the suitability of this measure.

The HIV-QL-31 is a 31-item measure encompassing the following domains: sex life, pain, psychological aspects, relationships, limitations caused by HIV, symptoms and impact of treatment and care. The scale has high internal consistency [17, 20] and has been shown to discriminate between groups based on disease severity [20]. The measure was developed with patients and thereby covers a broad range of issues relevant to people with HIV [20]. However the HIV-QL-31 has not been widely used and the limited psychometric data available, including lack of information on responsiveness, make it difficult to establish the suitability of this measure.

The MOS-HIV is most widely used HIV-specific measure. Initially based on the SF-20, it consists of 35 items across 11 domains: physical functioning, pain, social functioning, role functioning, emotional well-being, energy/fatigue, cognitive function, health distress, health transition, general health and overall quality of life, allowing for the generation of physical and mental health summary scores. The instrument has been translated into at least 14 languages. The English version takes approximately 5–10 min to complete, but was reported to take twice as long for some translations, such as the Spanish version, where more words are required to express the concepts [20]. Scoring and interpretation has been described as complex [12, 20].

The reviews reported mixed findings on the psychometric properties of the MOS-HIV. Good internal consistency was generally reported [4, 12, 19, 20] however Carabin et al. reported acceptable internal consistency for some but not all domains [17]. Test-retest reliability was considered inadequate [17, 19]. The reviews contained mixed findings regarding construct validity, with some suggesting poor construct validity [17] and others suggesting good construct validity [20]. Acceptable convergent and discriminant validity was reported in several reviews [4, 12], however some studies did not find the expected relationship with CD4 count [12, 20]. Responsiveness has been established in a wide variety of contexts including adverse events, increased symptoms, opportunistic infections, AIDS defining events and initiation of ART [4, 10, 12, 18, 19], however negligible effects have been found in patients changing ART regimens [10]. Gakhar et al. found the MOS-HIV to be more responsive than the EQ-5D and HUI3 [12]. However, ceiling effects have been reported on several domains [19, 20].

Overall, the MOS-HIV was considered to have well-established psychometric properties. It was recommended as a suitable measure for assessing HRQoL in people with HIV [4, 10, 18], either individually or together with the EQ-5D [10]. However reservations expressed in the reviews included concerns about the continued relevance of the measure for people with HIV, given that it was one of the earliest HIV-specific scales to be developed [4], questions about whether there was a true advantage of using the MOS-HIV over the SF-36 [4] and scepticism about the ‘gold standard’ status that has been assigned to the MOS-HIV based on available evidence [20]. One criticism was that while patient interviews were conducted as part of the development process, they had not sampled extensively from people with HIV, limiting content validity [20]. Furthermore, like the SF-36, the MOS-HIV may lack conceptual relevance across different languages and cultures [20]. The need for more information about the performance of the scale in women, low income and disadvantaged groups was also identified [20].

Multidimensional quality of life questionnaire for persons with HIV/AIDS (MQOL-HIV) [37]

The MQOL-HIV consists of 40 items assessing 10 domains: physical health, physical functioning, mental health, social functioning, cognitive functioning, social support, financial status, sexual functioning, partner intimacy and access to care. These domains were developed through interviews with HIV positive patients and providers in different settings.

The MQOL-HIV was considered to be a useful instrument since attention had been paid to the concerns of people living with HIV in its development [20], however there was limited evidence to support the construct validity, reliability and responsiveness of this instrument [4, 12, 20] and floor and/or ceiling effects were reported in some domains [17].

The PROQOL-HIV consists of 43 items assessing 8 domains: physical health and symptoms, treatment impact, emotional distress, health concerns, body change, intimate relationships, social relationships and stigma. The items were developed through interviews conducted with people living with HIV in 9 countries. Psychometric data were not reported in the reviews.

Symptom quality of life adherence (HIV-SQUAD) [40]

The HIV-SQUAD consists of 24 items assessing HRQoL, a symptom checklist and a visual analogue scale to measure adherence. Preliminary psychometric data were reported in one review [12]. While Cronbach’s alphas indicted acceptable internal consistency for the physical component, there was unacceptable internal consistency for the psychological component. The instrument was able to discriminate between patients with different illness states (CD4 count, hepatitis infection) and was responsive to changes in HIV viral load [12]. However, there was no evidence of the measure’s use beyond the initial validation paper and limited psychometric data were available in the reviews.

WHOQOL-HIV BREF [39]

The WHOQOL-HIV BREF was developed in focus groups of people with HIV across 6 countries, ensuring that the items have conceptual and semantic relevance across cultures [20]. The instrument has 31 items, including several developed as a result of input from people with HIV (e.g. sexual activity, social inclusions and aspects of spirituality, such as forgiveness) and covers the six generic WHOQOL domains (physical functioning, psychological functioning, levels of independence, social relationships, environment and spirituality). No psychometric data on the scale were available in the reviews.