2. the measure contained more than one domain; and
3. the measure could be used on patients with all cancer types.
Measures that have been used in cancer care but were specific to
a particular patient group, for example, leukaemia patients
(Cancer Leukaemia Group B Studies - CALGB), or measures
which concentrated on only one life domain, for example,
physical symptoms (McGill Pain Questionnaire), were
excluded from the review (see Bowling3). It was also important
to identify only those measures that had been used for patients
receiving palliative care or proposed for use measuring
outcomes at this stage of the disease trajectory. For example,
measures specifically designed to assess the outcome of nonpalliative
cancer chemotherapy, such as the Breast Cancer
Chemotherapy Questionnaire (BCCQ), were not included.3
Assessment of identified measures
Measures were then assessed following the criteria outlined in
Table 1 ? Content validity was further assessed by whether the
measure covered the particular domains reported to be relevant to
palliative care (physical, psychological and spiritual dimensions),
and how many items were contained in each domain.
Results
In total, 41 measures were identified (see Tables 2 and 3).
Twelve of these satisfied the inclusion criteria. These measures
contained between five and 56 items and covered the physical,
psychological and spiritual domains of life to differing extents
(see Table 3).
To summarize the 12 measures: three are completed by a
professional;6'8'9 seven by the patient himself or herself;10"16 two
contain both patient and professional completion elements;17'18
eight assess items relating only to the patient,8'10"15'17 whereas
four may also consider the family or carer unit;619'1618 seven
have been validated in just one setting;6'10"12114'1518 five contain
30 or more items;10'12'14'16-17 two were designed for the
assessment of clinical trial interventions.12-14 This paper will
now describe each of these measures in more detail.
An initial assessment of suffering10
This measure was developed on 259 advanced cancer patients
in acute hospitals. A five-point Likert Scale with scores ranging
from five for 'good' to one for 'bad' was used to record the
answers to the 43 questions either by the patient unaided or by a
trained nurse interviewer. The questions have been refined to
give a shorter 20-item questionnaire suitable for use during the
initial assessment by a member of any profession in the hospice
or palliative care team.
Edmonton Symptom Assessment Schedule (ESAS)11
The ESAS was developed for quick assessment of outcomes in
routine practice. This tool consists of nine Visual Analogue
Scales (VASs). The patient draws a mark along a 100 mm line
corresponding to how they feel, with the far left end of the line
corresponding to the least degree of symptoms, and the far right
'worst' symptoms. The ESAS is completed on admission to
hospital and twice daily thereafter by the patient, or with the
assistance of a nurse. Patients who are unable to respond owing
to cognitive failure are assessed by their nurse or a specially
trained family member. The score for each item is recorded on a
bar graph, allowing staff to visualize patterns of symptom
control over time. Further testing of this measure's validity and
reliability are required, particularly with reference to the
potential bias introduced by a change in the person recording
the answers on the VAS as care continues.
European Organisation for Research on Cancer
Treatment (EORTC QLQ-C30)12
Developed with lung cancer patients to evaluate the quality of
life of those patients participating in international clinical trials,
this self-reporting questionnaire is both a reliable and valid
measure of the quality of life of cancer patients in research
settings. Questions cover the past week and responses are
mainly in the format of a straightforward four-point Likert
Scale, ranging from one for 'not at all' to four for 'very much'.
It contains a generic core with cancer-specific modules and
work is being carried out to extend the questionnaire for
Table 1 Criteria used to assess outcome measures7
Validity - the instrument measures what it intends to measure
Content validity - does the measure cover those domains considered important?
Criterion validity - does the measure correlate with superior measures or predict future outcome?
Construct validity - does the measure conform with the results using other established scales (or
discriminate between groups of patients)?
Reliability -the instrument produces the same results when repeated on an unchanged population
Inter-rater reliability - does the measure produce similar results when used by different observers?
Test-retest reliability - does the measure produce similar results when used at different points in time?
Internal consistency - do individual items within the instrument correlate with each other?
Responsiveness to change - the instrument is able to detect clinically s