The question of which chronic disease most impacts upon quality of life can lighten
many a dreary hour. Anecdote piles upon anecdote, but the problem is that the plural
of anecdote is not data. What is needed is surveys using the same instrument to
measure quality of life, used in large enough samples of patients, with a similar
range of disease severity, and with a similar demographic base. Even then there may
be problems in interpretation, but it might provide some better insights into disease
impact on people. A study with about 15,000 patients from Holland gives us just this
[1]. But prospective readers should beware. This is definitely a two-brains paper,
but worth conquering.
Study
All research groups known to examine chronic diseases in the Netherlands were
contacted to see what data sets were available. Studies had to use a standardised
quality of life instrument, have full coverage of quality of life domains, include a
range of chronic diseases, be big (at least 200 patients), have medically confirmed
diagnoses, be obtained since 1992 and be geographically broad.
Eight data sets broadly fulfilling these categories were obtained, with information
on about 15,000 people. They all used SF-36 or SF-24. These were analysed by quality
of life dimension (physical functioning, physical role functioning, bodily pain,
general health, vitality, social functioning and mental health) according to:
- Disease clusters (grouping together similar diseases. For instance
musculoskeletal conditions of osteoarthritis, joint complaints, rheumatoid
arthritis and back impairments).
- Disease categories (ranking the individual diseases within the cluster).
- Patient characteristics (sociodemographic variables, like age, gender,
education).
The method used was the ranking of mean scores. Thus if three diseases scored (say)
5, 10, and 15 (with 5 the 'best' score), then they would be ranked 1, 2 and 3. This
was done for all quality of life domains, and the ranks for individual domains added
together. This summed rank produces low scores for the diseases or disease clusters
causing the least distress, and high scores for those causing the most problems.
Results
Disease clusters
The summed rank scores for chronic disease clusters are shown in the Figure.
Musculoskeletal conditions, renal disease, cerebrovascular/neurological conditions
and gastrointestinal conditions impacted most severely on quality of life.
Figure: Summed rank scores for disease clusters. Higher scores imply poorer
quality of life
|
Disease categories
Some examples are useful here. For instance, in musculoskeletal conditions,
ostoeoathritis had more adverse impact than back impairments, which scored higher
(worse) than rheumatoid arthritis. For neurological conditions, Parkinson's
disease or epilepsy, multiple sclerosis and stroke scored higher than migraine or
neuromuscular disease. For psychiatric disorders depression scored worse than
anxiety which in turn was worse than alcohol problems.
Patient characteristics
Patients who were older, female, had a low level of education, were not living
with a partner, and/or had at least one comorbid condition had the poorest
quality of life.
Comment
There will be limits to how far these sort of data can be subdivided and still
give us valid conclusions. So where there is the largest agglomeration of
information is where the strongest conclusions lie. For this analysis, this is
with the comparisons across disease categories.
Moreover, there are also issues within the quality of life measures that an
overall ranking will not highlight. This will be in the difference between mental
and physical functioning, for instance.
Many professionals will not be overly surprised at the ranking of disease
clusters, or the categories within each cluster, or the conclusions regarding
patient characteristics. But where there is limited information there will always
be room for argument. Though this ranking exercise should not be
over-interpreted, it does give us a firmer platform on which to base
decision-making, and on which to base research efforts.
The man on the stair
But hang on a minute, advised the man on the stair. Do you really believe that
musculoskeletal conditions cause a greater impact on quality of life than, say
visual impairment? Does this result have face validity? Could it simply be wrong?
What might be the causes of a wrong answer? Well the combining of data in the
meta-analysis may be incorrect, though the authors seem to have done a pretty
fine job, and discuss in detail possible problems and why they are unlikely to
occur. Then there are the original studies themselves used in the combining
process. Issues of validity of the original study was part of the defining
process of the analysis, which is why only eight of about 30 studies actually
made it into the review. The others were excluded because they were judged
inadequate.
Then there is the method used for measuring quality of life, in this case SF-36.
This measurement tool has been around for ages, is much used, and for which there
are manuals and methods written down in exquisite detail. Validity has been
measured in lots of different ways, hasn't it? And all that has been done so that
we can be confident of the results obtained by using it.
The only other conclusion left would be that SF-36 is fundamentally flawed. That
would have major repercussions, especially on all those discussions on things
like cost per QALY that policy makers use for judging whether medicines can be
bought by the NHS or other healthcare providers. Fundamental stuff, this, and why
this paper [1] is so important about thinking about thinking about healthcare
delivery.
References:
- MAG Sprangers et al. Which chronic conditions are associated with a better
or poorer quality of life? Journal of Clinical Epidemiology 2000 53:
895-907.
|
previous
story in this issue
