Rheumatology 2000;39:990–997
Patient-initiated hospital follow-up for
rheumatoid arthritis
S. Hewlett, K. Mitchell, J. Haynes, T. Paine1, E. Korendowych
and J. R. Kirwan
Rheumatology Unit, University of Bristol Division of Medicine, Bristol Royal
Infirmary and 1The Family Practice, Western College, Bristol, UK
Abstract
Objectives. To evaluate the clinical efficacy, cost and acceptability of a shared care system
of patient- or general practitioner (GP)-initiated hospital review in rheumatoid arthritis (RA).
Methods. A 2-yr randomized controlled trial of routine rheumatologist-initiated review was
compared with a shared care system. Shared care patients had no routine follow-up but
patients or GPs initiated access to rapid review by the multidisciplinary team via a nurse-run
helpline. Control patients had a rheumatologist-initiated medical review at intervals of
3–6 months. Clinical and psychological status, resource use, and patient and GP satisfaction
and confidence were assessed. Three-monthly clinical data were assessed (blind) for safety
monitoring, with failure set at a 20% increase in pain, disability or disease activity.
Results. Two hundred and nine established RA patients participated, of whom 182 were
evaluable. Safety-net failures were not different between groups. Shared care patients had less
pain (24 months, 3.9 cm on a 10-cm visual analogue scale vs 4.8 cm for controls; P < 0.05), a
smaller increase in pain over 2 yr (+ 0.4 cm vs +1.6 cm for controls; P < 0.01), greater selfefficacy (6, 15, 18, 21 months, P < 0.05), used 33.5% less resources (£208 per patient per year
vs £313 for controls; P < 0.001) and were more confident in the system (6, 9, 12, 18, 21, 24
months, P < 0.01 to P < 0.001).
Conclusions. A patient-initiated system for hospital review over 2 yr offers some clinical
benefit compared with the traditional system, using fewer resources and attracting greater
patient confidence. Longer-term assessment of the system would be appropriate.
K : Rheumatoid arthritis, Hospital follow-up, Patient-initiated, Cost, Pain, Disability,
Self-efficacy, Satisfaction.
Rheumatoid arthritis (RA) is a chronic but variable
disease requiring continuing specialist and multidisciplinary care [1], often including lifetime routine reviews
by hospital rheumatologists. Patterns of hospital review
vary but most patients attend every 3 months initially,
extending to 6–12 months as the course of the disease
becomes established. Planned follow-up review of RA
patients forms up to 75% of the average rheumatologist’s
workload [2]. Consequently, out-patient clinics are heavily booked in advance and, as the system becomes
difficult to manipulate, both routine new patient waiting
time and the wait for urgent review during disease
exacerbations are unacceptably long, causing dissatisfaction for patient, GP and rheumatology staff. In addition,
at routine reviews the patient’s disease is often quiet
and little or no intervention is required—patients may
be seen when they do not require help but do not have
easy, rapid access to support at times of need.
Faulkner and Frankel [3] suggested a variety of ways
of optimizing out-patient services (e.g. increasing discharge rates, lengthening the interval between routine
follow-up visits), but when implemented locally these
had minimal impact on an unwieldy system.
Furthermore, they do not address the fundamental belief
underpinning the traditional system—that RA patients
require regular hospital-initiated review throughout their
lifelong illness.
General practitioners (GPs) do not wish to manage
their RA patients alone and lose the expertise of the
rheumatologist and specialist multidisciplinary team, but
the traditional system is not optimal [4]. A more flexible
system of review might reduce inappropriate follow-up
appointments, improve rapid access to specialist advice
and release resources to further improve service provision. If the patient is empowered to initiate specialist
review, then feelings of enhanced self-efficacy and
control may improve the clinical and psychological
outcome. Some rheumatologists, either by choice
or necessity, lengthen the interval between routine
Submitted 29 September 1999; revised version accepted 3 March
2000.
Correspondence to: S. Hewlett, Rheumatology Unit, Bristol Royal
Infirmary, Bristol BS2 8HW, UK.
990
© 2000 British Society for Rheumatology
Patient-initiated hospital follow-up in RA
follow-up visits such that patients are infrequently
reviewed. There is little hard evidence about whether it
is clinically safe to reduce routine specialist input; therefore, this study includes a ‘safety net’ to monitor the
clinical progress of patients receiving both traditional,
planned reviews and no routine reviews but rapid access
(shared care).
Controlled studies of shared care in other diseases
have been performed [5], but most did not incorporate
patient-initiated access to hospital review. Studies in RA
have concentrated on monitoring for adverse effects of
slow-acting anti-rheumatic drugs (SAARDs) [6 ]. Mowat
et al. [7] randomized 72 RA patients to GP, consultant
or specialist occupational therapist follow-up and found
that all three systems were equally effective, but they
did not use patient-initiated review.
This study used a randomized controlled trial to
compare traditional, routine rheumatologist-initiated
hospital care for patients with established RA with a
patient-initiated shared care system in which there were
no routine hospital reviews, but hospital review was
initiated by patients and GPs. The hypotheses addressed
are that the patient-initiated shared care system will
show an improvement in clinical and psychological
outcome, a reduction in the overall use of healthcare
resources, and greater satisfaction with care by patients
and GPs.
Method
Three hundred and two consecutive patients attending
the rheumatology out-patients department of the Bristol
Royal Infirmary and who had established RA according
to international criteria [8] were invited to take part,
the study having received approval from the United
Bristol Health Care Trust Research Ethics Committee.
All RA patients were invited, with no exclusion criteria.
After receiving information on both arms of the study,
patients who gave consent were randomized to either
shared care with the GP [no routine hospital review but
rapid access on request; the shared care group (SCG)]
or traditional hospital care [regular planned review; the
control group] for 2 yr. SCG patients were cared for
by their GPs, who were provided with management
guidelines based on previous publications [9, 10].
[Responsibility for monitoring SAARDs is held routinely by GPs in this locality.]
Access to rheumatologist, occupational therapist or
physiotherapist review
SCG patients or GPs requested review by any team
member through the nurse-run telephone helpline,
whereby general advice or assistance was also given.
Fortnightly ring-fenced rheumatology clinics gave a
maximum wait of 10 working days for review. Control
patients had a traditional medical review ordered routinely every 3–4 months or according to standard practice. Requests for medical review of control patients
ahead of schedule were dealt with according to normal
practice, GP requests being assessed by a rheumatologist
991
(usually not the study rheumatologist), who decided on
the timing of the review. In the rare event of an
emergency (e.g. septic arthritis), all patients would be
dealt with immediately. Control patients had access to
the occupational therapist (OT ) and physiotherapist
(PT ) via the traditional route of GP request, rheumatologist appointment and referral to OT or PT waiting
lists. At medical, OT or PT hospital reviews, all patients
were managed according to need and further follow-up
was given as clinically indicated, SCG patients eventually
returning to patient-initiated review and controls to
routine review.
Outcome measures
Clinical and psychological
To assess the overall state of RA, evaluations of plasma
viscosity, C-reactive protein, haemoglobin, hand X-rays
[11], grip strength, range of movement at the knee and
elbow, and articular index [12] were made at 0 and 24
months. Clinical and psychological status was measured
at 3-month intervals with a questionnaire that covered
pain and disease activity [10-cm visual analogue scale
( VAS )], disability (Health Assessment Questionnaire)
[13, 14], days lost from work, helplessness (Arthritis
Helplessness Index) [15], anxiety and depression
(Hospital Anxiety and Depression Scale) [16 ], selfefficacy [17] and changes in medication.
Satisfaction
Satisfaction with and confidence in the system of care
were recorded using a 10 cm VAS every 3 months. At
24 months, GPs were asked the same question and GPs
of SCG patients were also asked whether they wished
their patients to continue in the shared care system.
Resources
Visits to health professionals about arthritis (including
the use of hospital transport) were recorded in patients’
diary cards. The costs of all hospital visits were calculated using local NHS Trust figures while GP, district
and practice nurse visits and hospital transport journeys
were costed using published unit cost data [18].
Safety net monitoring
A safety net, using the 3-monthly questionnaires
(assessed blind), was set up to monitor all patients’
clinical status as half were no longer receiving reviews.
Based on clinical experience, an increase of 20% in
pain, disease activity or disability was deemed a safety
net failure. Shared care patients who failed the safety
net were telephoned to check on their well-being and
encouraged to see their GP, while those who failed on
two consecutive occasions were encouraged to come for
rheumatologist review. Hospital records of control
patients who failed were checked to ensure they were
not lost to follow-up. Case notes were reviewed at 24
months by an independent assessor, who documented
RA complications using a predetermined checklist.
992
S. Hewlett et al.
Sample size and statistics
In order to show a 12% change in pain using a 10 cm
VAS, at 95% power, the study required a sample size of
186 patients [19]. Analysis was performed on all those
who completed four or more of the nine data sets.
Changes in parameters between the two groups of
patients at intermediate time points and over 2 yr were
compared using Student’s t-test or the Mann–Whitney
U-test (non-parametric data).
Results
Of 302 subjects invited to participate, 209 (69.2%)
agreed. Patients who declined were older, had a higher
mean articular index and were more disabled ( Table 1).
No patient declined to take part after randomization,
although the GP of one SCG patient declined on
grounds of comorbidity (but did enter other patients).
Of the 208 entered, 26 (12.5%) withdrew or were withdrawn because of non-compliance with safety monitoring (SCG, 11; control group, 15); nine patients died
(SCG, 5; control group, 4) but their data are included
where available (n = 182; SCG, 93; control group, 89)
(Fig. 1).
At entry patients showed moderate inflammatory
activity, pain and disability and appeared representative
of a hospital-based RA population ( Table 1). There
were no significant differences between the two groups,
other than for grip strength (P < 0.05).
F. 1. Patient flow through the study.
T 1. Demographic data
Variable
Age (yr)*
Disease duration (yr)
Males:females
Disability (HAQ, 0–3)*
Articular index (0–534)**
Pain (0–10 cm)
Disease activity (0–10) cm
Early morning stiffness (min)
CRP (mg/l )
Plasma viscosity (mPa)
Hb (g/dl )
Range of movement (degrees)
Right elbow
Left elbow
Right knee
Left knee
Grip strength, left hand (kg)
Grip strength, right hand (kg)***
Anxiety (0–21)
Depression (0–21)
Helplessness (1–30)
Self-efficacy (pain) (0–100)
Self-efficacy (function) (0–100)
Self-efficacy (other) (0–100)
Control group
(n = 89)
Mean (..)
SCG
(n = 93)
Mean (..)
59 (13)
12 (8)
27:62
1.4 (0.8)
110 (112)
3.2 (2.2)
3.0 (2.3)
58 (156)
21 (22)
2 (0)
12 (1)
57 (13)
11 (9)
31:62
1.4 (0.8)
97 (93)
3.5 (2.4)
3.3 (2.2)
65 (155)
25 (26)
2 (0)
13 (1)
127 (21)
129 (23)
108 (21)
107 (31)
14 (8)
14 (10)
6.7 (3.8)
5.0 (3.2)
16.1 (4.6)
52.6 (19.5)
59.8 (24.9)
65.3 (19.4)
126 (24)
130 (18)
107 (18)
112 (19)
19 (17)
17 (11)
6.6 (3.7)
4.9 (3.2)
16.5 (4.4)
55.0 (18.3)
62.1 (24.9)
64.3 (19.9)
Declined
(n = 94)
Mean (..)
68 (10)
14 (7.8)
22:72
2.0 (0.8)
154 (119)
*P < 0.05, declined vs SCG and controls combined (t-test); **P < 0.05, declined vs SCG and controls combined (Mann–Whitney U-test);
***P < 0.05, controls vs SCG (t-test).
Patient-initiated hospital follow-up in RA
Clinical and psychological benefit
The majority of the 3-monthly questionnaires were
returned (SCG, 88.8%; control group, 88.9%). There
was a significant difference between the groups for pain
at 24 months (SCG, 3.9 cm, control group 4.8 cm;
P < 0.05) (Fig. 2), and change in pain over months
0–24 was significantly less in shared care patients (SCG,
+0.4; control group, +1.6 cm; P < 0.01). There was a
trend in favour of the SCG patients for changes in
patient opinion of disease activity over 24 months (SCG,
–8%; control group, +17%; not significant). Mean
disability fluctuated very little over the 2 yr (SCG
improved by 8%, control group deteriorated by 4%) and
there were no statistically significant differences for
change in mean disability over 24 months ( Fig. 3).
There were significant differences in self-efficacy for
function at 6, 15, 18 and 21 months (SCG stronger, all
993
P < 0.05; 24 months, P < 0.053) (Fig. 4). Although
again there were trends in favour of the shared care
group for changes in psychological status over 24
months (anxiety and depression increased by 3 and 4%,
respectively, in SCG and by 11 and 14% in the control
group), there were no significant differences between the
groups for changes in psychological status.
The frequency of safety-net failures was not significantly different between the groups (SCG, 26.5%; control group, 28.9%), although more control patients failed
the safety net on a second review 3 months after failing
a first review ( Table 2). Of the shared care patients who
failed, the majority had already sought advice or did
not require it.
The proportions of patients treated with non-steroidal
anti-inflammatory drugs (NSAIDs), SAARDs and
glucocorticoids were equal in the groups at entry and
F. 2. Pain scores: comparison of mean scores with 95% confidence interval.
F. 3. Disability scores: comparison of mean scores with 95% CI. (HAQ, Health Assessment Questionnaire.)
S. Hewlett et al.
994
F. 4. Self-efficacy (SE) scores for function: comparison of mean scores with 95% CI.
T 2. Safety-net data
T 3. Percentage of patients receiving medication
Shared care
(n = 93)
Control
(n = 89)
Months
Medication
Questionnaires returned
Failed (first occasion)
Failed (second occasion, 3 months
later)
Reasons for first failure
Pain
Disease activity (patient opinion)
Disability
Time lost from work
Upon contacting patient (first failure)
Already sought medical help
Subsequently requested help
Did not require assistance
Upon contacting patient (second
failure)
Already sought medical help
Subsequently requested help
Did not require assistance
661
181
26.5% (175) 28.9% (181)
3.2% (22)
6.1% (38)
68%
46.8%
21.7%
0%
66.8%
58.9%
13.1%
0%
53.6%
28%
18.3%
54.5%
27.3%
13.2%
remained relatively constant throughout the 2 yr
( Table 3). Assuming that NSAIDs might have been
changed at either a GP or a hospital consultant visit,
there was one NSAID prescription change for every
nine SCG doctor visits (i.e. an alteration ratio of 1:9, vs
1:18 for the control group) and, assuming only rheumatologists were likely to alter SAARDs, a SAARD
alteration ratio of 1:6 SCG consultant reviews (control
group, 1:10).
At the 24-month case note review, four instances of
RA complications were documented for SCG patients
(one instance each of respiratory and gastrointestinal
complications and two instances of nodules) while controls had 10 instances (three instances of vasculitis, three
of drug reactions, two of gastrointestinal complications,
one each of renal and nodules). At entry, Larsen scores
[12] for radiographs were 43.7 (.. 19.1) for SCG and
NSAIDs
Shared care
Control
Oral glucocorticoids
Shared care
Control
SAARDs
Shared care
Control
0
3
6
9
12
15
18
21
24
70
72
78
72
81
76
80
79
74
75
76
81
76
76
75
76
70
72
17
26
18
26
19
25
17
22
18
18
17
17
16
15
19
16
18
15
67
71
71
68
72
73
72
74
72
76
70
75
66 65 67
81* 84* 79
*P < 0.05.
48.1 (.. 26) for the control group; at exit they were
48.4 (.. 20.4) for SCG and 50.1 (.. 27.3) for the
control group (no significant difference).
Resource use
Shared care appointments were given within a mean of
4.9 working days (range 0–18 days). In the eight cases
where the delay was >10 working days, this coincided
with staff or bank holidays and a consequent clinic
postponement. There was a 43.8% difference in consultant reviews [SCG requested 262 appointments, with a
mean of 2.82; controls were given 466 reviews, with a
mean of 5.24; P < 0.001), but no significant differences
in visits to other professionals ( Table 4). The overall
cost for managing the 93 SCG patients was £38 635
(£208 per patient per year) vs £55 597 for the 89 control
patients (£313 per patient per year). This represents a
saving for the SCG of 30.5% overall, or, when the
difference in group sizes are taken into account, 33.5%
(P < 0.001). The helpline received a mean of 2.4 calls
per week, which in most units would be managed by
existing nurse specialists. However, even if 10% of a
Patient-initiated hospital follow-up in RA
995
T 4. Number of consultations and their costings
SCG
Costing
Hospital doctor
GP (surgery)
GP (home visit)
Practice nurse (surgery)
District nurse (home)
Occupational therapist
Physiotherapist
Orthotist
Podiatrist
Orthopaedic surgeon
Transport journeys
Total cost for group, 2 yr
Cost per patient per year
£70a
£10b
£30b
£6b
£12b
£18a
£18a
£18a
£18a
£70a
£33.59b
Visits
262
403
20
914
49
110
99
18
47
33
70
Control group
Costs
Visits
£18 340
£4030
£600
£5484
£588
£1980
£1782
£324
£846
£2310
£2351
£38 635
£208
466***
315
8
1005
18
975
92
26
65
53
119
Costs
£32 620***
£3150
£240
£6030
£216
£2340
£1656
£468
£1170
£3710
£3997
£55 597***
£313***
Difference between
SCG and
control group
£14 280***
− £880
£360
− £372
− £546
£360
− £126
£144
£1400
£324
£1646
£16 962***
£105***
*P < 0.05; **P < 0.01; ***P < 0.001.
aLocal NHS Trust costing.
bUnit costs (Netten and Dennet [18]).
nurse’s salary (£2000) was included to run the helpline,
there was still a 26.8% saving (£229 per SCG patient
per year; £313 per control patient per year).
To test the robustness of these costings, a sensitivity
analysis was undertaken, using various values for the
two main contributing factors: hospital consultant visits
and GP visits ( Fig. 5). Even at unrealistically low
costings for hospital consultant visits (£50) and inappropriately high costings for GP consultations (£55), i.e. a
situation in which consultations in primary care are
more expensive than consultations in secondary care,
the overall cost saving for the study was 14%.
Satisfaction
Satisfaction and confidence levels were high for both
systems and not significantly different at entry (satisfac-
tion: SCG, 8.35 cm; controls, 8.29 cm; confidence: SCG,
8.48 cm; controls, 8.17 cm). However, satisfaction was
significantly higher in the SCG when patients were asked
to look back over the 2 yr (SCG, 8.56 cm; controls,
7.89 cm; P < 0.05). Confidence in the system to care for
them was significantly higher in SCG, in six of the eight
cross-sectional assessments (Fig. 6). At 24 months, 82
SCG patients were invited to continue with shared care
(five died, four had other comorbidities, one was discharged, one was non-compliant with safety questionnaires, and one was diagnosed with systemic lupus
erythematosus) and all accepted. At 24 months there
was no significant difference between the groups for GP
satisfaction and confidence (satisfaction: SCG, 7.66 cm;
controls, 7.39 cm; confidence: SCG, 7.46 cm; controls,
7.65 cm). Of the 70% of SCG GPs who answered the
F. 5. Sensitivity analysis of various costings for consultant and GP visits.
996
S. Hewlett et al.
F. 6. Confidence in the system: comparison of mean scores with 95% confidence interval.
question, 80% wished their patients to continue in the
shared care system, 8% preferred them to revert to
traditional care and 12% expressed no preference.
Discussion
This randomized controlled trial compared a patientinitiated system of shared care including open access to
the hospital specialist team with the traditional system
of routine consultant review in RA. Results suggest no
clinical deterioration and some clinical benefit in the
shared care group, which was achieved at a 33.5%
resource saving and incorporated greater satisfaction
and confidence in the system.
The groups were similar in all respects at entry, and
it is therefore unlikely that differences in disease status
could have influenced overall findings. Patients who
declined to take part were significantly older and more
affected by their arthritis. It may be that the decliners
were anxious about making a major change in a wellestablished care system or were uncertain about initiating consultant review themselves, although no data are
available on this. Shared care might need to be targeted
at certain patient groups.
The increased ratio of prescription changes to doctor
visits implies that shared care hospital and GP consultations were more appropriate (i.e. that treatment needed
to be changed), but these results must be interpreted
with caution as specific data on the content of outpatient visits was not recorded and a visit may have
been appropriate without a change in medication. As
clinical data were collected on a calendar basis (unrelated
to clinic appointments), it is not possible to relate clinical
status to any specific hospital visit. A further prospective
study is under way to examine the appropriateness of
both routine and requested visits.
The number of GP visits was increased in the shared
care patients, although this was not statistically significant. However, although extra GP visits were inherent
in the study design, no GP declined to participate and
most wished patients to continue at the end.
Consultant availability had to be protected during
this research study as it was not known how many
patients would request appointments, nor how frequently. The reported 33.5% saving does not take
account of the ring-fenced shared care appointments
that were not utilized by patients, leaving some consultant appointments unused. Clearly the saving is only
achieved if fewer staff are employed or more patients
cared for. However, the data now allow us to estimate
that, by utilizing the unfilled consultant appointment
slots, approximately 50% more patients could be accommodated in this shared care system without extra medical
input. If the system were to be scaled up to incorporate
the majority of a consultant’s RA patients (perhaps
8-fold ), then more than one rapid access clinic would
be required each fortnight, which would increase the
number of clinics available to patients. The study implies
that involving patients in their management reduces the
overall need for medical consultations (GP plus hospital
consultant visits) without increasing either physiotherapy or occupational therapy input, although there was
additional nursing support via the nurse-led helpline.
The method of safety-net monitoring (20% deterioration in disease designated as failure) meant that patients
who deteriorated slowly (e.g. by 10% every 3 months)
or patients who commenced the study with a VAS pain
score already above 80% of the maximum did not trigger
the review system. However, no patient deteriorated
significantly over 2 yr. Furthermore, the failure rate was
similar in the two groups, indicating that these somewhat
arbitrary cut-off points were compatible with the dayto-day variation seen in patients under regular review
and, indeed, may even have been too sensitive.
Patient-initiated hospital follow-up in RA
Over half the shared care patients who failed the
safety net had already requested medical help and the
nurse telephone calls resulted in only a quarter seeking
help. This indicates that patients are able to recognize
difficulties and to access help appropriately when given
the responsibility and mechanism to do so.
The lead rheumatologist in the study could not be
blind to the group assignment of the patients, which
may be considered a weakness in the study design.
However, had different clinicians managed the two
groups, this would have introduced another variable
(personal approaches to patient management).
Employing a single approach to management enables
the effect of service delivery/access to be assessed.
The study raises two main questions about generalizability. First, the study was confined to a single teaching
hospital. The patients are likely to be representative of
the average rheumatologist’s RA caseload, as all consecutive patients in an RA follow-up clinic were invited,
with no exclusion criteria. Unlike district general hospitals (DGH ) this teaching hospital has a number of
specialist registrars but (as yet) no nurse practitioner or
nurse-led clinics. However, the work was carried out by
a single consultant and could therefore reasonably be
reproduced in a DGH. Secondly, the system has been
investigated for only 2 yr in a chronic disease. However,
the patient cohorts and systems have been maintained
beyond this and data collection is currently under way
for years 5 and 6, allowing long-term analysis.
This study raises further issues. For example, can
those patients who would benefit most from this system
be identified? Should patients be offered an education
programme on self-management as a prerequisite for
entering the new system? Could or should 3-monthly
postal reviews be replaced by (for example) a biennial
health professional or medical review? How appropriate
is the timing of the traditional consultant reviews compared with shared care reviews? Are there missed treatment opportunities (e.g. for OT intervention) when
patients are not reviewed regularly? These points will be
also be addressed in the extension of the study to 6 yr.
These results challenge the traditional view that chronically ill patients (in this instance those with RA) require
regular routine hospital review. It has provided evidence
that, in the short term (2 yr), a patient-initiated shared
care system can be clinically and economically effective
for established RA patients, is acceptable to GPs and
patients and is worthy of serious consideration.
Acknowledgements
The study was funded by an NHS Research and
Development National Programme Grant. The authors
would like to thank the advisory group for their assistance: Professor D. Sharpe ( University of Bristol
Division of Primary Health Care), Dr I. Hine (GP) and
Mrs A Hopwood (Rheumatology Clinical Manager,
Bristol Royal Infirmary). We would also like to thank
the clinical team for supporting the study—Mrs M.
Cottle (PT ), Ms G. Ludlum, Ms R. Squires and Ms L.
997
Blenkiron (OTs), Mrs S. Tippler (Nurse), Ms P. Kearsey
(NHS Secretary)—and clinic staff Mrs W. Harrison and
Mrs M. Bullock, and the patients who participated.
References
1. Scott DL, Shipley M, Dawson A, Edwards S, Symmons
DPM, Woolf AD. The clinical management of rheumatoid
arthritis and osteoarthritis: strategies for improving clinical
effectiveness. Br J Rheumatol 1998;34:546–54.
2. Kirwan JR, Snow S. Which patients see a rheumatologist?
Br J Rheumatol 1991;30:285–7.
3. Faulkner A, Frankel S, eds. Evaluation of delivery and
use of outpatient services. Bristol: Health Care Evaluation
Unit, University of Bristol, 1993:48–54.
4. Dickson DJ. What do general practitioners expect from
rheumatology clinics? Br J Rheumatol 1996;35:920.
5. Hampson JP, Roberts RI, Morgan DA. Shared care: a
review of the literature. Fam Pract 1996;13:264–79.
6. Pullar T, Hunter JA, Capell HA. Gold and penicillamine
therapy: is shared care with general practitioners effective
and safe? Rheumatol Rehab 1982;21:139–44.
7. Mowat AG, Nichols PJ, Hollings EM, Haworth JR,
Aitken LC. A comparison of follow-up regimes in rheumatoid arthritis. Ann Rheum Dis 1980;39:12–7.
8. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries
JF, Cooper NS et al. The American Rheumatism
Association 1987 revised criteria for the classification of
rheumatoid arthritis. Arthritis Rheum 1988;31:315–24.
9. Dieppe P, Kirwan JR, Cooper C, McGill N. Arthritis and
rheumatism in practice. London: Gower Medical
Publishing, 1990.
10. Dieppe P, Paine T. Referral guidelines for general practitioners—which patients with limb joint arthritis should be
sent to a rheumatologist? Reports on Rheumatic Diseases.
Chesterfield: Arthritis Research Campaign Publication,
1994.
11. Thompson PW, Silman AJ, Kirwan JR, Currey HLF.
Articular indices of joint inflammation in rheumatoid
arthritis: correlation with the acute phase response.
Arthritis Rheum 1987;30:618–23.
12. Larsen A, Dale K, Eek M. Radiographic evaluation of
rheumatoid arthritis and related conditions by standard
reference films. Acta Radiol Diagn 1977;18:48–91.
13. Fries FJ, Spitz P, Kraines RG, Holman HR. Measurement
of patient outcome in arthritis. Arthritis Rheum 1980;
23:137–45.
14. Kirwan JR, Reeback JS. Stanford Health Assessment
Questionnaire modified to assess disability in British
patients with rheumatoid arthritis. Br J Rheumatol
1986;25:206–9.
15. Stein MJ, Wallston KA, Nicassio PM. Factor structure of
the Arthritis Helplessness Index. J Rheumatol 1988;15:
427–32.
16. Zigmond AS, Snaith RP. The Hospital Anxiety and
Depression scale. Acta Psychiatr Scand 1983;67:361–70.
17. Lorig K, Chastain R, Ung E, Shoor S, Holman H.
Development and evaluation of a scale to measure perceived self-efficacy in people with arthritis. Arthritis
Rheum 1989;32:37–44.
18. Netten A, Dennett J. Unit costs of health and social care.
Personal Social Services Research Unit, University of
Kent, 1997.
19. Bellamy N. Standardised procedures for outcome measurement and parameters for calculating sample size for antirheumatic drug studies. In: Bellamy N, ed. Musculoskeletal clinical metrology. London: Kluwer Academic,
1993.