Rheumatology 2000;39:369–376
Does waiting matter? A randomized controlled
trial of new non-urgent rheumatology out-patient
referrals
N. P. Hurst, C. M. Lambert, J. Forbes1, A. Lochhead, K. Major2
and P. Lock2
Rheumatology Unit, Western General Hospital NHS Trust, Crewe Road,
Edinburgh EH4 2XU, 1Department of Public Health Sciences, Teviot Place,
University of Edinburgh and 2Ayrshire & Arran Health Board, Boswell House,
10 Arthur Street, Ayr KA7 1QJ, UK
Abstract
Objective. To examine the effect of waiting times on the health status of patients referred
for a non-urgent rheumatology opinion.
Methods. The study was a randomized controlled clinical study evaluating a ‘fast track’
appointment with a 6-week target waiting time against an ‘ordinary’ appointment in the main
city out-patient clinic of the rheumatology service for the Lothian and Borders region
(population ~1 million). Health status was measured using the SF12 physical and mental
summary component T-scores and pain was measured with a 100 mm visual analogue pain
scale. Secondary outcomes were health utility and perceived health both measured with the
EuroQol instrument, mental health measured with the Hospital Anxiety and Depression scale,
disability with the modified Health Assessment Questionnaire and economic costs measured
from a societal perspective.
Results. Mean waiting times were 43 days (s = ±16) and 105 days (s = ±51) for ‘fast
track’ and ‘ordinary’ appointments, respectively. Both groups showed significant
improvements in mean [95% confidence interval (CI )] scores for pain: 11 (7, 16) (P < 0.001);
physical health status: 4 (2, 5) (P < 0.001); mental health status: 2 (0.1, 4) (P < 0.02); and
health utility: 0.11 (0.07, 0.16) (P < 0.001) by the end of the 15-month period of the study,
but there was no significant difference between either arm of the study.
Conclusions. Rationing by delay was not detrimental to either mental or physical health
and patients in both arms of the study showed significant and similar improvement in health
by 15 months. Expenditure of resources on waiting times without regard to clinical outcomes
is likely to be wasteful and additional resources should be directed at achieving the greatest
clinical benefit. More research into effective methods of controlling demand and better
identification of those who would benefit from access to specialist care is needed.
K : Rationing, Waiting list, Arthritis, Health status, Health outcome, Economic cost.
The search for effective and equitable policies for managing the rising costs of health care has taken different
forms in different countries. Examples include the use
of health maintenance organizations in the USA, various
types of health insurance and fee for item of service
schemes in Europe and, in New Zealand, explicit nationally agreed rationing criteria [1].
In the UK, rationing methods include the constraints
imposed by national and health authority cash allocations, the exclusion or severe limitations by health
boards of certain treatments such as in vitro fertilization
(IVF ), and the control of demand through the manage-
ment of waiting lists (rationing by delay) [2, 3]. To the
extent that such rationing decisions are based on arbitrary and implicit preferences rather than on explicit
criteria, they could be criticized as being unethical and
inequitable [2, 4, 5]. Furthermore, the absence of any
clear national policies and the devolution of rationing
decisions to the local level, have led to wide variations
in health care provision [4] and to increasing public
concern and confusion about the process of rationing
and priority setting.
The political focus on waiting lists, either in the form
of national waiting time guarantees or the allocation of
resources for waiting list initiatives, has further distorted
priorities and expectations. These manipulate just one
aspect of a dynamic system, and set artificial targets
Submitted 23 July 1999; revised version accepted 19 October 1999.
Correspondence to: N. P. Hurst.
369
© 2000 British Society for Rheumatology
370
N. P. Hurst et al.
which bear no relation to health outcomes or quality of
service, and fail to address the equally important problem of demand. Consequently they produce either little
or no effect on waiting lists, distort clinical priorities
and raise unrealistic public expectations of the health
service. Investment in rheumatology services is clearly
desirable, but such investment should be used to produce
optimal clinical benefit and should satisfy unmet clinical
needs. To test whether investment in waiting time reduction would be beneficial in terms of improved health
outcomes, we have undertaken a randomized controlled trial.
This trial has examined the health impact of a reduction in ‘rationing by delay’ for non-urgent referrals to a
rheumatology service [3]. Non-urgent new patients were
randomized either to an ‘ordinary’ out-patient appointment with a mean waiting time of about 15 weeks or to
an intervention arm in which they received a ‘fast track’
appointment reducing their waiting time to approximately 6 weeks. The health status of each group and
the resources incurred were measured at baseline, the
time of their appointment and again 1 month postappointment and 15 months post-referral. The hypothesis tested was that current waiting times do not
adversely affect the health status of patients with nonurgent conditions. The secondary aim was to examine
the effect of waiting time on economic costs from a
societal perspective.
Methods
Setting
The trial was conducted between 1997 and 1999 in the
main rheumatology out-patient clinic for the city of
Edinburgh. The rheumatology service covers the
Lothian and Borders region and has a catchment population of approximately 1 million.
Patient population
All general practitioner (GP) referral letters are graded
‘urgent’, ‘soon’ or ‘routine’, by a consultant or experienced registrar, on the basis of the clinical information
provided and using criteria which are locally agreed and
made known to all local GPs [5]. These criteria are by
necessity simple but practical, and their application is
very dependent on the quality of the information provided. The criteria are ‘urgent’—active/acute recent
onset inflammatory disease; ‘soon’—patients with subacute problems, e.g. poorly controlled rheumatic disorder
needing management review, or non-urgent diagnostic
problems; ‘routine’—chronic non-acute problems, e.g.
cervical spondylosis, soft tissue or regional disorders.
Patients given an ‘urgent’ priority are currently seen
within 6 weeks and were excluded. Those eligible for
entry to the trial were any new patients referred to either
of two consultant rheumatologists or to ‘any doctor’,
who had been allocated to a non-urgent (‘soon’ or
‘routine’) appointment and were over 18 yr of age.
Eligible patients were randomized off-site by telephone,
using consecutive sealed envelopes, either to an ‘ordin-
ary’ out-patient appointment or to a ‘fast track’ appointment. To accommodate the ‘fast track’ patients,
additional clinic appointments were made available
each week.
Recruitment and informed consent
Because waiting time was the main intervention under
investigation, consent to gather health status and other
data was obtained from patients after they had been
randomized into either a ‘fast track’ or ‘ordinary’
appointment; patients remained blind to the randomization procedure and intervention. Recruitment continued until 250 patients had been randomized. Patients
who declined to participate did not have their allocated
appointment changed. The protocol received institutional ethical approval.
Health status assessment
Patients were mailed their appointment, consent form
and a baseline set of health status questionnaires to be
completed and returned as soon as possible if they
consented. Further questionnaires were completed on
the day of the booked clinic appointment, 1 month postappointment and 15 months post-referral. Patients were
asked to complete health status questionnaires without
assistance. Non-respondents were sent a second letter
within 14 days and also telephoned. If still no response
was obtained, details of the patients’ age and sex were
recorded.
Pain and physical function were the primary health
outcome indicators and were assessed using a visual
analogue pain scale ( VA-pain) [6 ] and the SF12 physical
component summary score (SF12-PCS) [7]. The
VA-pain scale measures joint pain on a 100 mm visual
analogue scale where a score of 100 indicates worst
possible pain and 0 no pain. The SF12 scores are
expressed as T-scores with standardized normal population mean of 50 and standard deviation (s) of 10. Low
scores indicate worse health.
Secondary outcome measures were collected, including the SF12 mental component summary score
(SF12-MCS) [7], the Hospital Anxiety and Depression
scale (HAD) [8], EuroQol ( EQ-5D) [9] and the modified
Health Assessment Questionnaire (MHAQ) [6 ]. The
SF12-MCS and the HAD both measure psychological
distress, the SF12-MCS is expressed as a T-score, whilst
the HAD ranges from 0 to 42, 0 indicating no symptoms
of anxiety or depression. The EQ-5D is a generic measure of health-related quality of life [9]. Part I of EQ-5D
is a questionnaire from which a value for health utility
can be obtained by applying preference weightings [10]
and part II is a visual analogue scale. The utility scores
( EQ-5Dutility) value health on a scale from 0 to 1,
where 1 represents perfect health and 0 represents death.
The visual analogue scores ( EQ-5D VAS ) indicate perceived health and range from 0 (worst imaginable health)
to 100 (best imaginable health). The MHAQ is scored
from 0 to 3, high scores represent worse disability.
Clinicians remained blind to all assessments. If one
or two items were missing from any questionnaire, the
Does waiting matter?
371
F. 1. Consort flow diagram. Assessments at each time point—1health status, 2employment, 3community health resources,
4clinic resources, 5personal expenditure.
patient was asked if they would complete them. If more
than two items were missing no such attempt was made
and the questionnaire was rejected as invalid. Published
SF12 scoring rules were followed [7].
Age, sex, post code, provisional diagnosis of inflammatory or non-inflammatory disease based on GP referral letter, and the final diagnosis after consultation were
recorded.
Economic data collection
At each time point employment status was categorized
as employed, self-employed, looking after family, sick
and either intending or not intending to seek work,
retired or performing alternative unpaid work. Those in
employment were asked how many hours/week they
worked. Health resource use in the community was
recorded as the number of visits patients made to
National Health Service or other health workers (GP,
nurse, physiotherapist, occupational therapist, osteopath, chiropracter, etc). This information was obtained
by asking patients, at the clinic and follow-up assessments, to recall the number of visits made to health
professionals during the preceding month. Health
resource use arising as a direct result of the consultation,
i.e. laboratory investigations (haematology, biochemistry, immunology), radiology, procedures, prescribed
drugs and orthotics, referral to other specialists or
paramedicals, was also recorded.
Patients were asked about expenditure on over the
counter (OTC ) medication, medical aids, home adaptations and alternative therapies for a 1-month period at
baseline and at the end of the study.
372
N. P. Hurst et al.
Statistical methods
To test the hypothesis that waiting time did not adversely
affect the health status of patients awaiting their appointment, the largest acceptable clinical differences between
groups were chosen as 15 mm on the 100 mm VA-pain
scale or a five-point difference on the SF12-PCS. Seventy
subjects are required per group to detect these differences
with a power of 90% for the VA-pain scale and 80% for
the SF12-PCS scale (alpha = 0.05, two-tailed t-test) [11,
12]. A total of 250 patients were randomized which,
allowing for a 30% non-response or drop-out rate,
would provide approximately 80 patients in the two
arms of the trial.
Between-group comparison of change in health status
over time was performed using analysis of covariance
(ANCOVA). For categorical or ordinal data, nonparametric methods were used. Analyses were performed
on the basis of intention to treat. Data were analysed
using SPSS for Windows version 7.5.2.
Results
Demographics and waiting times
Two hundred and fifty patients were randomized to ‘fast
track’ or ‘ordinary’ appointments, of whom 180 gave
consent ( Fig. 1., Table 1). The mean age [48.2 yr, 95%
confidence interval (CI ) 44, 52] and the proportion of
females (63%) of the 70 patients (29 ‘fast track’ and 41
‘ordinary’) who did not consent did not differ significantly from those who consented. The final diagnoses
recorded after the clinic consultation are shown in
Table 1; by way of comparison, 90 ‘urgent’ patients were
audited in parallel to the trial. Of these, 70% had a
clinical diagnosis of inflammatory rheumatic disease and
26% rheumatoid arthritis.
The distribution of waiting times during the trial
compared with Scotland or England and Wales during
the same period is shown in Table 2. Overall, the distribution of waiting times for new patients seen in Lothian
is typical of the rest of the UK, and if anything waiting
times are slightly longer. However, because urgent referrals who have shorter waiting times were excluded from
the trial, the distribution for the non-urgent trial patients
is different. Only 47% of ‘ordinary’ patients were seen
within 13 weeks with a median wait of 97 days, while
98% of ‘fast track’ patients were seen within 13 weeks
with a median waiting time of 40 days.
Health status at baseline
Baseline health status questionnaires were returned
promptly (mean 17 days; 95% CI 15, 20). At baseline
there was no significant difference between the ‘fast
track’ and ‘ordinary’ appointment categories either in
mean age (t-test, P = 0.36), sex ratio (P = 0.41), proportion with a provisional diagnosis of inflammatory vs
non-inflammatory disease (P = 0.9) or mean health
status scores ( Table 3).
The profile and prevalence of health problems
reported by patients in part I of EQ-5D are shown in
T 1. Comparison by treatment group of baseline patient characteristics. Numbers (%) for categories, mean (95% CI ) for continuous
variables
Fast track
appointment
(n = 96)
Ordinary
appointment
(n = 84)
Age (mean, 95% CI )
47.1 (44.4, 49.9) 48.6 (45.9, 51.5)
Females
68 (71)
52 (62)
Employmenta
Self-employed or employed
55 (57)
43 (51)
Unemployed
2 (2)
3 (4)
Retired or home duties
47 (49)
46 (54)
Sick (intending or not
11 (11)
11 (13)
intending to seek work)
Alternative unpaid work
5 (5)
7 (8)
Hours paid work
0 h/week
38 (40)
40 (48)
<20 h/week
7 (7)
7 (8)
20–35 h/week
18 (19)
12 (14)
>35 h/week
33 (34)
25 (30)
Patients reporting problems on EQ-5D with:
Mobility
58 (61)
52 (63)
Self-care
32 (34)
29 (35)
Usual activities
80 (84)
66 (79)
Pain
87 (91)
80 (95)
Anxiety/depression
50 (52)
54 (64)
Clinical diagnoses
Non-inflammatory
60 (62)
58 (61)
Degenerative joint disease
37
37
and non-specific arthralgia
Regional syndromesb
15
9
Psychiatric diagnosesc
5
3
Miscellaneous
3
2
Inflammatory
36 (38)
31 (37)
Rheumatoid arthritis
8
6
Spondylarthritis
8
9
Crystal arthritis
3
3
Connective tissue disease
2
3
Other inflammatory arthritisd 15
10
Missing
0
2
aEmployment total exceeds 100% as some patients in more than
one category.
bRotator cuff disease, capsulitis, tennis elbow, carpal tunnel
syndrome.
cDepression, anxiety, chronic fatigue, anorexia.
dMonoarthritis, palindromic rheumatism, polymyalgia rheumatica,
unspecified polyarthritis.
T 2. Waiting times for rheumatology out-patient appointments
Total
England & Wales
204 429
1997/1998a
Scotland 1997/1998a
13 224
Lothian clinics
2311
All patients randomized
180
in trial
‘Ordinary’ appointments
84
in trial
‘Fast track’ appointments
96
in trial
<13 13–25 >26 Median
weeks weeks weeks
days
73%
23%
4%
61
71%
68%
74%
23%
24%
22%
6%
8%
4%
55
63
53
47%
44%
8%
97
98%
2%
0%
40
aWaiting time data from the National Health Service executive and
Information and Statistics Division (Scotland).
Does waiting matter?
aNo significant difference between mean change scores for fast track and ordinary group (ANCOVA all P values > 0.05).
VA-pain: no pain = 0; maximal pain = 100. SF12-PCS and SF12-MCS: expressed as T-score, normal population mean = 50, s = 10. HAD: no symptoms of anxiety or depression = 0;
maximum symptom score = 42. EQ-5D (utility): perfect health = 1; death = 0. EQ-5D ( VA scale): best imaginable health = 100; worst = 0. MHAQ: least disabled = 0; most disabled = 3.
−11, 5
−2, 5
−2, 4
−2, 1
−0.10, 0.06
−7, 4
−0.25, 0.02
15 (8, 21)
10 (3, 16)
5 (2, 7)
3 (1, 6)
2 (0, 5)
2 (−2, 5)
1 (−1, 2)
1 (−1, 2)
0.11 (0.03, 0.18) 0.12 (0.05, 0.20)
0 (−5, 6)
4 (0, 9)
0 (−0.08, 0.09)
0.01 (0.01, 0.23)
−3, 9
−1, 2
−3.5, 0.9
−1, 1
−0.12, 0
−4, 5
−0.08, 0.06
5 (−1, 11)
0 (−2, 2)
2 (0, 5)
0 (−1, 1)
0.06 (−0.002, 0.12)
1 (−4, 5)
0.02 (−0.04, 0.07)
2 (−1, 5)
1 (0, 2)
1 (0, 2)
0 (−1, 1)
0.00 (−0.03, 0.03)
1 (−1, 4)
0.01 (−0.03, 0.05)
52 (26)
37 (11)
42 (12)
15 (7)
0.47 (0.33)
61 (21)
0.71 (0.58)
52 (27)
35 (10)
44 (13)
14 (8)
0.48 (0.35)
63 (23)
0.78 (0.64)
Ordinary
Mean (95% CI )
Fast
Ordinary
Fast
Health status
VA-pain
SF12–PCS (physical health)
SF12-MCS (mental health)
HAD (mental health)
EQ-5D (utility)
EQ-5D (global health)
MHAQ (disability)
Ordinary
Fast
Mean (95% CI )
95% CI for
difference between
the fast and
ordinary change
scoresa
Mean (s)
Change score baseline to first appointment
95% CI for
difference between
the fast and
ordinary change
scoresa
Baseline
Employment status and use of health resources
The employment status of subjects at baseline is summarized in Table 1. There was no significant change in
patient’s employment status or hours of work between
either arm of the study over the 15 months (x2 test;
P > 0.1). There was no significant difference in community resource use between fast and ordinary track
groups—these groups made a total of 171 and 157 visits,
respectively, to GPs, and 136 and 83 visits, respectively,
to other health professionals during the three sampling
periods (P > 0.1). Use of community resources by both
groups was significantly less during the final sampling
period at 15 months than during the two earlier sampling
periods ( Friedman test P = 0.000).
At the consultation, fewer patients in the ‘fast track’
group were prescribed analgesics or underwent routine
laboratory tests. Thus, in the ‘fast track’ group 35% of
patients were prescribed analgesics compared with 55%
of patients in the ‘ordinary’ group (P < 0.01), and 29%
T 3. Health status at baseline, and subsequent change in health status
Effect of waiting time on health status
Two patients did not attend the first consultation, and
three of the 178 who attended failed in error to complete
health questionnaires. The mean waiting times were 43
days (s = ±16) for ‘fast track’ appointments and 105
days (s = ±51) for ‘ordinary’ appointments. One
month after the consultation 154 of 178 patients (86%)
completed further questionnaires; a final assessment was
undertaken 15 months post-referral by 138 (77%)
patients.
There was no significant difference between either arm
of the study in change in health status over time
measured at the clinic appointment, 1 month postconsultation (not shown) or at the final 15-month review
(ANCOVA) ( Table 3). There was no improvement in
health status between baseline and the 1 month postconsultation assessment, but by 15 months, there were
significant improvements in mean (95% CI ) pain [11
(7, 16) P < 0.001], physical health status [4 (2, 5)
P < 0.001], mental health status [2 (0.1, 4) P < 0.02]
and health utility [0.11 (0.07, 0.16) P < 0.001] ( Table 3).
Overall health change was also analysed according to
final clinical diagnosis. Patients with rheumatoid arthritis showed large and significant improvements over 15
months in both of the primary outcome measures and
each secondary outcome except mental health ( Table 4).
Those with spondylarthritis also improved in terms of
physical health. Patients with non-specific arthralgia
improved in mental health and health utility but in
contrast, patients with osteoarthritis showed no significant improvement in any of the outcomes.
Change score baseline to 15 months post-referral
Table 1. The health of patients at baseline in terms of
physical function, mental health and pain are reflected
in the health status scores ( Table 3). The HAD score is
shown as a combined anxiety and depression score
( Table 3); when the depression and anxiety subscale
scores are considered separately, 29% and 50%, respectively, had scores >8 (i.e. indicative of a clinically
significant mood problem) on the two scales.
373
N. P. Hurst et al.
374
T 4. Change in health status over 15 months in four diagnostic groups [mean change (95% CI )]a
Rheumatoid arthritis
(n = 13)
VA-pain
23 (11, 35)**
SF12-PCS (physical
7 (1, 13)*
health)
SF12-MCS (mental
4 (−2, 11) ns
health)
HAD (mental health)
2 (−2, 6) ns
EQ-5D (utility)
0.21 (0.05, 0.36)*
EQ-5D (global health) 15 (6, 25)**
MHAQ (disability)
0.34 (0.04, 0.64)*
Spondylarthritis
(n = 13)
15 (−3, 32) ns
9 (3, 15)**
6 (−1, 13) ns
4 (0.1, 7)*
0.22 (0.05, 0.39)*
11 (1, 21)*
0.08 (−0.15, 0.32) ns
Non-specific arthralgia
(n = 27)
Osteoarthritis
(n = 22)
9 (−2, 19) ns
2 (1, 4) ns
4 (−4, 12) ns
−1 (−4, 3) ns
5 (2, 9)**
−2 (−6, 2) ns
1 (−1, 3) ns
0.11 (0.02, 0.21)*
2 (−7, 10) ns
0.01 (−0.14, 0.16) ns
−3 (−1, −5)*
0.01 (−0.1, 0.12) ns
−3 (−13, 6) ns
−0.02 (−0.16, 0.12) ns
aPaired t-test: baseline vs 15 months.
ns=not significant; *=P<0.05; **=P<0.01.
underwent routine laboratory tests compared with 53%
of patients in the ‘ordinary’ group (P < 0.001). In both
groups, those who required tests underwent a mean of
2.5 tests per patient. Prescription of other drugs, use of
radiology and referral to paramedical services did not
differ significantly between the two groups. The median
number (range) of visits to out-patients during the study
period did not differ between the fast track [1 (1–5)]
and ordinary arms [1 (1–9)] (Mann–Whitney P > 0.1).
Patient costs
The median [interquartile range (IQR)] expenditure on
home adaptations, medical aids, OTC medication and
alternative therapies was £2.30 (11.60) per patient in
the fast track and £1.55 (11.75) per patient in the
ordinary appointment arm (P > 0.1). Expenditure on
OTC items was lower at 15 months than at baseline
( Wilcoxon P = 0.012), but for other items was the same
during the two sample periods.
Discussion
Main findings
The principal finding was that rationing by delay, within
the limits of this study, was not detrimental to either
mental or physical health, and that shortening waiting
times to 6 weeks did not produce any additional health
benefit. Similarly, the resource consequences measured
in terms of employment status, use of community
resources or personal expenditure on non-prescribed
items, were no different for patients who received fast
track appointments from those receiving an ordinary
appointment. There were, however, some minor differences in use of hospital resources. At the consultation,
patients randomized to ‘ordinary’ appointments were
more often prescribed analgesics and were more likely
to undergo routine laboratory tests. One possible
explanation for this difference is that the laboratory test
results supplied in the GP’s letter were more likely to
be considered out of date in those with longer waiting
times and, therefore, more likely to be repeated by the
clinic doctor. The higher rate of analgesic prescription
remains unexplained.
Study design
The patients in each arm of this study were well matched
for all important demographic and health characteristics. The overall case mix and distribution of waiting
times was typical of other UK rheumatology units and
the follow-up period was sufficiently long to capture
health change. The criteria used to prioritize patients
were locally agreed and may differ from criteria used by
other units. The method is also clearly very dependent
on the quality of information provided by the GP letter
and the judgement of the doctor assessing this information—both processes are subject to error. The case mix
of non-urgent appointments was dominated by noninflammatory disease and only 14 patients (8%) had
rheumatoid arthritis. In contrast, an audit of ‘urgent’
appointments showed that the majority (70%) had
inflammatory disease and 26% had rheumatoid arthritis.
Thus, although there was clearly a failure rate in our
selection process, it was not unduly high.
Using a target of 6 weeks the mean waiting time was
reduced in the ‘fast track’ group to 43 days compared
with 105 days in the ‘ordinary’ group. While this was
perhaps not a large absolute reduction, it was associated
with a major shift in the distribution of waiting times.
The result was that virtually all ‘fast track’ patients were
seen within 13 weeks compared with less than 50% in
the ‘ordinary’ appointment group.
A sampling method was used to estimate community
costs because it was not considered practical to ask
patients to maintain diaries on expenditure and use of
community services for the whole 15 months. Although
our data suggest it is unlikely, we cannot exclude the
possibility that we have failed to capture additional
demands placed on GPs and community health services
while patients were waiting for their appointment.
Although no change in employment status was found,
we did not document days lost from work due to
sickness. There are methodological difficulties in identifying the impact of sickness on productivity [13, 14].
For example, it may be difficult to determine whether
days off sick are directly attributable to the illness, social
factors or even misguided advice from other health
professionals. Conversely, some patients may avoid
Does waiting matter?
taking time off, but work at a much lower capacity.
Quantifying this loss of productivity is therefore difficult
and for these reasons we chose the cruder but less
ambiguous end-point of actual change in work status.
Health change
With respect to health benefits, it could be argued that
those patients who benefited from treatment in the fast
track arm received that benefit some 6 weeks earlier.
This issue would be of most importance to the small
number of rheumatoid arthritis patients, who showed
the biggest clinical improvement. However, this earlier
benefit can only be assumed since our trial design did
not permit a detailed study of the time course of health
change. Also the numbers in the diagnostic subgroups
were small and comparisons of outcome between these
groups should therefore be interpreted with caution.
None the less these data emphasize the need to identify,
at the time of referral, patients who may have rheumatoid arthritis and who require more urgent specialist
attention.
Waiting lists and referral thresholds
Musculoskeletal disease is a common and important
public health problem [15, 16 ], and represents one area
where long waiting times have been particularly evident.
Demand for specialist services is high and in many units,
including our own, is controlled using a triage system
giving different priority to different clinical problems,
resulting in waiting times which greatly exceed the
9-week target set in the National Health Service outpatient’s charter. It appears that where demand for
more rheumatology resources has been met by increased
provision of consultants, the effect has been to reduce
both waiting times and the threshold for referring
patients with, arguably, less serious complaints [17, 18].
However, there is no evidence that shortening waiting
times has a beneficial effect on health or reduces disability in this group of patients, and the direct and indirect
costs and benefits are unknown. In patients with less
severe forms of osteoarthritis, for example, conservative
interventions have only limited benefit [19] and it should
be possible to deliver these at primary care level. It may
therefore reasonably be asked to what extent rising
demand for musculoskeletal services should be dealt
with by increasing resource allocation rather than by
continuing control through waiting lists and clinical
prioritization?
Implications for health policy
We believe that this is the only randomized controlled
trial to have explicitly tested the impact of waiting on
health and our results may have significant consequences
for health policy. A long waiting time may not affect
health status, but is, at the very least, inconvenient to
the individual patient. In the 1997/1998 year, the median
waiting time in Lothian for new referrals to consultantled out-patient clinics was 63 days with 32% of patients
waiting longer than 13 weeks and 8% waiting longer
than 26 weeks. These times are comparable to other
375
UK units. Waiting lists of this magnitude are difficult
to reduce and despite the government introducing a
number of high profile waiting list initiatives during
recent years the effects appear to be only temporary.
Referral thresholds are dynamic [17, 18] and evidence
from other centres [20], and anecdotally from our own
is that reduction of waiting times is short lived both
because this encourages additional referrals and because
there is a rising background demand for rheumatology
out-patient appointments [21]. The policy implication
is that waiting lists will only be eliminated by a substantial increase in capacity so that any patient can be seen
at any referral threshold. Given that this is unrealistic,
and the evidence that waiting is not detrimental to the
health of non-urgent patients, a delay may be efficient
in that it ensures that urgent cases are seen promptly
and that hospital services are used to capacity. For less
serious rheumatic complaints, further work exploring
patient preferences and attitudes is clearly needed to
inform policy on access to health care. Patients may
prefer additional resources to be spent on improving
aspects of care other than waiting lists. The only conclusion that can be drawn from these data is that longer
waiting times, within the limits studied, do not compromise the health status of non-urgent referrals and do
not increase economic costs. Thus, investment aimed
solely at reduction in waiting lists is an inefficient use
of resources. Investment in waiting time reduction would
be better directed at improving services for specific
conditions such as rheumatoid arthritis, for which there
is evidence that early intervention is beneficial [22, 23].
Conclusions
Given the dynamic relationship between length of waiting lists and public demands and expectations, simply
spending resources to try to shorten waiting times or
numbers on waiting lists, without regard to clinical
outcome is likely to be wasteful. Clinicians are committed to the development of evidence-based medicine and
there is clearly a need for government to make a similar
commitment to evidence-based policy. More research
into effective methods of controlling demand, better
identification of those with early rheumatoid arthritis
who need and benefit from prompt access to specialist
care [22, 23], and better diagnosis and management of
non-progressive conditions at primary care level are
required. Such measures are more likely to achieve a
sustained reduction in waiting times and improvements
in specialist care for those who really require it. Where
additional resources are made available, these should be
aimed at achieving the greatest clinical and health
benefit.
Acknowledgements
This work was supported by a grant from the Chief
Scientist’s Office of the Scottish Home & Health
Department. Dr R. Elton kindly provided advice on the
statistical analysis of the results. The authors are also
376
N. P. Hurst et al.
very grateful to all the patients who willingly gave their
time to complete the various assessments.
12.
References
1. Dixon J, New B. Setting priorities New Zealand style. Br
Med J 1997;314:86–7.
2. Light DW. The real ethics of rationing. Br Med
J 1997;315:112–5.
3. Klein R. Dimensions of rationing: who should do what?
Br Med J 1993;307:309–11.
4. Maynard A. Rationing health care. Br Med
J 1996;313:1499.
5. Hurst NP, McRorie ER. The short term health outcome
of out-patient rheumatology consultations in relation to
rationing: a pilot study. Br J Rheumatol 1997;37:509–13.
6. Felson DT, Anderson JJ, Boers M et al. The American
College of Rheumatology preliminary core set of disease
activity measures for rheumatoid arthritis clinical trials.
Arthritis Rheum 1993;6:729–40.
7. Ware JE, Kosinski M, Keller SD. SF12: How to score the
SF12 Physical and Mental Health Summary scales, 2nd
ed. Boston: The Health Institute, New England Medical
Center, 1995.
8. Zigmond AS, Snaith RP. The Hospital Anxiety and
Depression Scale. Acta Psychiatr Scand 1983;67:361–70.
9. The EuroQol group. EuroQol—a new facility for the
measurement of health related quality of life. Health Policy
1990;16:199–208.
10. Dolan P, Gudex C, Kind P, Williams A. Social tariff for
EuroQol: results from a UK general population survey.
Discussion paper 138. Centre for Health Economics, York
University.
11. Hurst NP, Ruta D, Kind P. Comparison of the MOS.
Short Form-12 (SF12) health status questionnaire with
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
the SF36 in patients with rheumatoid arthritis. Br
J Rheumatol 1998;37:862–9.
Hurst NP, Kind P, Ruta D et al. Measuring health-related
quality of life in rheumatoid arthritis: validity, responsiveness and reliability of EuroQol (EQ-5D). Br
J Rheumatol 1997;36:551–9.
Weinstein MC, Siegel JE, Garber A et al. Productivity
costs, time costs and health related quality of life: a
response to the Erasmus group. Health Economics
1997;6:505–10.
Brouwer WBF, Koopmanschap MA, Rutten FFH.
Productivity costs in cost effectiveness analysis: numerator
or denominator: a further discussion. Health Economics
1997;6:511–4.
Cohen G, Forbes J, Garraway M. Interpreting selfreported limiting long term illness. Br Med
J 1995;311:722–4.
Hurst NP. Rheumatoid arthritis in Scotland—A review of
its impact on public health, the economy and health
services. Health Bulletin 1994;52:198–206.
Kirwan JR, Dieppe P, Snow S. Consultant appointments
and their effect on out-patient appointments. Br
J Rheumatol 1989;28:453–4.
Bamji A, Dieppe P, Haslock I, Shipley ME. What do
rheumatologists do? A pilot audit study. Br J Rheumatol
1987;26(Suppl.):56–7.
Dieppe P. Osteoarthritis: time to shift the paradigm. Br
Med J 1999;318:1299–1300.
Newton JN, Henderson J, Goldacre MJ. Waiting list
dynamics and the impact of earmarked funding. Br Med
J 1995;311:783–5.
Kirwan JR. Rheumatology out-patient workload increases
inexorably. Br J Rheumatol 1997;36:481–6.
American College of Rheumatology ad hoc Committee
on Clinical Guidelines. Guidelines for the management of
rheumatoid arthritis. Arthritis Rheum 1996;39:713–22.
Emery P, Salmon M. Early rheumatoid arthritis: time to
aim for remission? Ann Rheum Dis 1995;54:944–7.