Choice of an instrument: Why AQoL?

There is no demonstrated ‘best’, ‘all purpose’ instrument for measuring quality of life.

Numerous non-utility and multiple multi attribute utility (MAU) instruments exist. Choice unavoidably requires judgement by the researcher. The wrong choice can invalidate measurement.

Choice should reflect three considerations

(i)    Dominating all others, the instrument must be capable of capturing all facets of the health state relevant to the research question;

(ii)    There should be evidence that measured magnitudes correspond with preferences;

(iii)    All else equal, brief instruments should be preferred.

The last criterion is commonly given excessive weight. None of the MAU instruments take more than 5-6 minutes to complete. If QoL is an important part of a study it is unwise to jeopardise the integrity of its measurement for a minimal time saving. We suspect this occurs due to the wrong belief that any ‘validated’ instrument satisfactorily measures QoL.

Why AQoL?

The AQoL instruments were designed because of the perceived limitations in the other instruments, at least in some contexts. Each of the AQol instruments increased instrument content and length (although AQoL8D, the longest, only takes an average of 5.4 minutes to complete). The increased content may sometimes - but not always - be necessary for the context and overall instrument validity. Researchers must select the instrument which best suits them from the options below.

The different instruments, their history, construction, use, strengths and weaknesses are summarised in Richardson, McKie and Bariola (Working Paper 2011).

Other options

MAU instruments in use include: EQ-5D, SF-12, HUI3, 15D, QWB. See chart comparing items in areas covered by AQoL and other Multi-Attribute Utility (MAU) instruments.

Validity

Instruments should be ‘valid’; that is they should have the ability to correctly measure what the researcher wants to measure. This is not always simple to judge and the claim that an instrument ‘has been validated’ is often misleading (see Validation). Instruments are ‘more or less’ validated and validity is context specific depending, inter alia, upon the purpose of the instrument.

Purpose of an instrument

Disease progression/recovery: If a study seeks only to monitor changes in the progression of a disease (after treatment) one of the many disease specific instruments may be suitable. See, for example, Bowling (2005).

Subjective wellbeing: If the study purpose is to monitor or compare subjective wellbeing (SWB) and satisfaction with (aspects of) life a number of SWB instruments exist. (For a review see Andrews and Robinson (1991). In Australia the Personal Wellbeing Index (PWI) has been used extensively in this capacity (see (International Wellbeing Group 2006). The Satisfaction With Life Scale (SWLS) has been widely used internationally.

Economic evaluation: If the study purpose is an economic evaluation based upon Quality Adjusted Life Years (QALYs) then an MAU instrument must be used.

Choice of MAU instrument

No MAU instrument has been validated for all contexts.

MAU instruments vary in their conceptualisation, dimensions, coverage, item detail, weighting technique and combination model. Correlations between scores from the same population are comparatively low. The large 5 instrument comparative study of 976 individuals by Hawthorne et al (2001) obtained correlations between 0.64 and 0.8. (R2 of 0.41, and 0.64 respectively). Incremental change varied by up t 100 percent. In sum, the validity of instruments varies in different contexts and little work has been conducted to compare results. For a summary see Khan and McKie (2010).

The chief consideration in selecting an instrument should be its capacity to measure what the research seeks to measure. This is the responsibility of the researcher. Note that sensitivity to a condition does not necessarily imply that the condition must be named. For example, without mentioning the condition the most important QoL effects of hospital dialysis in addition to its indirect effects upon lifestyle, may be captured by items relating to restricted activity, depression, vitality, etc. In contrast, a condition such as blindness may require naming if sight per se is important. The comparison chart provides a summary of the items in the major MAU instruments including the AQoL instruments.

Omission of items and dimensions is probability the greatest threat to validity. Importantly, a validation study demonstrating a correlation between an instrument and criterion measure (a quasi gold standard) does not guarantee that an instrument is valid in another context (see Validation).

Must one instrument be used to achieve comparability?

This argument is beguiling but false. An instrument which is sensitive in one context may be insensitive in another. The use of only this instrument will favour interventions in the context where it is sensitive. To take an extreme example (one aspect of) physical wellbeing may be very accurately measured by the question ‘how quickly can you run/walk 100 metres?’ The instrument would fail to measure the benefits of psychiatric procedures and cost utility analysis using this instrument would systematically discriminate against psychiatric care. The problem, of course, is generally more subtle than this!