11Multi-Item Scales
Multi-item PROs are typically applied to assess constructs that require the patient to
reconstruct, interpret, judge, compare, or otherwise evaluate complex or abstract
information. Under such circumstances multiple items may better capture the
essence of the phenomenon of interest [ 41 ]. Another important advantage of multi-
item scales is that more statistical information is provided so that scores are less
affected by measurement error. As a result multi-item questionnaires are usually
more reliable instruments. Although all multi-item PROs have in common that their
constituent items are combined to produce a total score, different measurement
models can be distinguished based on how individual items relate to the overarching
concept that the measure pertains to assess. That is, the items making up the scale
may either be hypothesized to be indicators of the measured trait, in which case the
individual items are referred to as effect indicators and the instrument is a scale or
the items together define the measured trait, in which case the items are referred to
as causal indicators and instrument an index [ 43 ]. For example, the Bath Ankylosing
Spondylitis Disease Activity Index (BASDAI) [ 44 ] is a patient-reported index of
disease severity that comprises 6 items pertaining to the five major symptoms of
AS: fatigue, spinal pain, joint pain/swelling, enthesitis, morning stiffness duration,
and morning stiffness severity. The total BASDAI score is obtained by averaging
the six individual items. The overarching, multidimensional concept of disease
activity is defined by, or emerges from, these individual constituting measures. If
any one of the five core symptoms of AS would be omitted from the BASDAI then
this would result in a less comprehensive assessment of disease activity. Disease
indices typically, but not exclusively, focus on PROs on the impairment level. By
contrast, the SF-36 physical functioning scale is a commonly used physical function
PRO that assesses disability across ten items that assess various aspects of physical
function [ 45 ]. Although a total score is similarly obtained by combining the indi-
vidual items, the individual items are considered more or less interchangeable
examples or realizations of the underlying construct of physical function.
Hypothetically, different specific indicators of physical function could have been
used, without changing the conceptual meaning of the trait that is assessed with the
scale. The reason that the same items have to be administered time and again relates
to the fact that total scores would not be comparable if patients would fill out differ-
ent items at different times. Items in a scale typically, but not exclusively, focus on
PROs at the disability or handicap level.
The measurement model that has been adopted has some implications for the
development or analysis of the respective PRO. Effect indicators are considered to
be more or less interchangeable realizations of the latent variable. Consequently
many statistical procedures that aim to evaluate the quality of individual items with
respect to measuring the desired outcome focus on the intercorrelation among items.
By contrast, causal indicators do not necessarily need to be correlated and when
developing an index, researchers tend to pay most attention to the ability of indi-
vidual indicators to discriminate between clinical relevant states [ 46 ]. Generally
speaking, items are considered to have high quality in a scale if they are substantially
1 PROMs and Quality of Care