| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Department of Clinical Chemistry, Vejle County Central Hospital, DK-7100 Vejle, Denmark.
2
Department of General Practice, University of Aarhus,
DK-8000 Aarhus, Denmark.
a Address correspondence to this author at: Department of Clinical Chemistry, Vejle County Central Hospital, DK-7100 Vejle, Denmark. Fax 45 75 82 18 14.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: In a randomized crossover trial during intervention periods, general practitioners (GPs) were allowed to measure CRP within 3 min, using NycoCard® CRP. During control periods, they had to mail blood samples for CRP measurements to the hospital laboratory and received test results 24–48 h later. Twenty-nine general practice clinics participated (64 GPs), and 1853 patients were included in the study. Results were evaluated at both the level of participating GPs and the level of included patients.
Results: For participating GPs, the overall use of erythrocyte sedimentation rates (ESRs) decreased by 8% (95% confidence interval, 1–14%) during intervention periods, and the number of blood samples mailed to the hospital laboratory decreased by 6% (1–10%). No reduction in the prescription of antibiotics was seen. The proportion of study patients having a follow-up telephone consultation was reduced from 63% to 53% (P = 0.0001), and patients with CRP concentrations >50 mg/L had their antibiotic treatments started earlier when CRP was measured in general practices (P = 0.0161).
Conclusion: The implementation of the near-patient CRP test was cost-effective mainly on the basis of a reduction in the use of services from the hospital laboratory by GPs. If the implementation is followed by education and clinical guidelines, opportunities exist for additional reduction in the use of ESR and for a more appropriate use of antibiotics.© 1999 American Association for Clinical Chemistry
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In Denmark, general practitioners (GPs) request CRP by mailing blood samples to hospital laboratories, and the test is ordered for 34% of all patients having a blood sample mailed (10). Several near-patient tests for CRP measurements are available commercially. We have evaluated one of these tests (NycoCard® CRP) and found it reliable and robust for the use in general practice (11). In questionnaire surveys, it has been proposed that knowing the CRP value while having the patient in the office may give GPs better guidance for the prescription of antibiotics than ESR (12)(13).
Implementation of near-patient tests in general practice should be based on an outcomes assessment with documentation for the clinical outcomes (14). However, outcomes for the organization and economic outcomes are important features and should be included in an assessment (15). Therefore, results of a technology assessment should be available before the decision to introduce the test is made (16)(17).
The aim of this study was to assess the clinical, organizational, and economic consequences of introducing a near-patient test for CRP measurements in general practice, with special focus on the clinical use of the test, the additional use of ESR, prescription of antibiotics, and utilization of the hospital laboratory service.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
intervention
The clinics were randomized into two groups. One group was given
access to a near-patient test for CRP (NycoCard CRP Whole Blood;
Nycomed Pharma) in the office (intervention), whereas the other group
had to order CRP as usual, mailing a blood sample to the laboratory
(control). After a period of 3 months, the two groups interchanged
their status (crossover). The first period of intervention and control
was 3 months (April–June 1996), and the second period was 4 months
(July–October 1996). All clinics received an introductory visit with a
demonstration by the manufacturer before their intervention period. No
clinical guidelines for the use of CRP were distributed to the clinics.
inclusion of patients
The GPs filled out a registration card for each patient when a CRP
was measured in the office or requested at the laboratory. The date for
the consultation, the tentative diagnosis, and the CRP value were
registered together with the patient's personal registration number.
the danish health service system and the laboratory information
system
All inhabitants in Denmark have a personal registration number,
and it is therefore possible to collect information about any person
from independent registries (18). Of the population, 97%
are registered at their GPs (19). The GPs receive a basic
fee from the Danish Health Service System for every registered person
and an extra fee for several specified services such as consultations,
laboratory tests, and mailing of patient samples to laboratories. These
services, as well as prescription of drugs, are registered by the
Health Service System and can, on application, be used for scientific
purpose (18). Partnership general practices have the same
registration number, which allows no differentiation between individual
GPs in a clinic. In the county of Vejle, all test results from hospital
laboratories are stored in a Laboratory Information System for 5 years
for all 340 000 inhabitants.
outcome measurements at the clinic level
From the regional office of the Health Service System, we received
data on the following: (a) the number of patients registered
at each clinic; (b) the date of consultation and personal
registration number for all patients receiving one of the following
services: ESR, WBC, antigen test for group A streptococci, bacterial
culture, urine susceptibility test, and microscopy of urine and
bacteria; and (c) all antibiotics dispensed in 1996.
From the Laboratory Information System at Vejle Hospital, we received data for all blood tests requested in 1996.
outcome measurements at the patient level
Some outcome measurements were available only at the patient
level, including the number of follow-up consultations.
statistical analyses
Results were calculated at the patient level by use of
2 tests, unpaired t-tests, and
Mann–Whitney tests. At the clinic level crossover
t-test statistics were used (20). The inclusion
rates were assessed for each clinic in intervention and control
periods. If a clinic had an inclusion rate <75% in either period, all
patients from that clinic were excluded from the study. The clinic
itself was not excluded because all outcome measurements at the clinic
level were based on data from the Health Service System and the
Laboratory Information System, and therefore were independent of the
registration of included patients.
ethics
The study was approved by the Danish Data Protection Agency, and
the local scientific ethics committee was informed about the study.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
.
Results at the level of general practice clinics were suitable for
organizational and economic evaluation, and these results are presented
in Tables 5
and 6
.
|
|
|
|
|
|
results at the level of included patients
During the study period, 2915 patients were registered at their
GPs: 1560 (53.3%) during intervention periods and 1355 (46.5%)
during control periods. During the intervention periods, all 29
clinics had an inclusion rate near 100%; however, during the control
periods, 13 clinics had an inclusion rate <75% (range, 6–67%). All
1041 patients registered by these clinics were excluded. Of the
remaining 1874 patients, 21 were excluded because of incomplete
registration of personal registration numbers, leaving 919 patients
included in the intervention periods and 934 patients included in the
control periods.
Sex and age characteristics.
The mean age of the included
patients was 53.7 years [95% confidence interval (CI),
52.8–54.6%], and 60.2% (95% CI, 58.0–62.4%) of the patients were
women.
Reasons for measuring CRP and disease characteristics.
The
disease characteristics for the included patients are given in Table 1
. During intervention periods, the use of CRP was significantly
higher for "diagnosis of new diseases" and "infectious
diseases". For patients with sinusitis, the use of CRP during
intervention periods was threefold higher than during the control
periods, whereas changes for other specified infectious diseases are
insignificant. There was a significant decrease in follow-up testing
during intervention periods. The decrease in follow-up testing was seen
mainly in the group of patients without infections, with a decrease
from 18.1% during control periods to 12.8% during intervention
periods (P = 0.002). Follow-up testing for patients
with infectious diseases decreased from 8.1% to 5.9%
(P = 0.07). The use of CRP for chronic inflammatory
diseases and other diseases was equally distributed in the periods.
Additional tests, consultations, and laboratory services.
The
number of additional tests and consultations together with the use of
laboratory services is given in Table 2
. The use of ESR was significantly (11%) lower during
intervention periods, and the decrease was significant both for
patients with CRP values within the health-related reference
interval and for patients with high values. For patients with
CRP values 
10 mg/L, the decrease was 9.1% (P =
0.002); for patients with CRP values between 10 and 25 mg/L, the
decrease was 12% (P = 0.043); and for patients with
CRP values 
25 mg/L, the decrease was 12.6% (P =
0.014). The use of other supplementary tests remained unchanged. The
use of follow-up telephone consultations was also lower during
intervention periods, whereas the number of daytime and out-of-hours
consultations was unchanged. During control periods, 140 patients
(15.0%) had a blood sample forwarded for a CRP measurement only, but
the total number of blood samples forwarded to the laboratory declined
significantly, from 100% to 60.2%, during intervention periods,
which is a reduction of nearly 40%. This reflects a reduction in
the number of patient samples forwarded containing a request for other
analyses in addition to a CRP measurement. The mean number of
requests for additional laboratory tests for each forwarded blood
sample was 4.8 tests in intervention periods and 5.5 tests in control
periods (i.e., CRP exclusive). In both periods, creatinine, alanine
aminotransferase, alkaline phosphatase, WBC, and potassium were the
most frequently requested additional tests.
Prescription of antibiotics.
In Table 3
, the prescription of antibiotics is compared for different CRP
concentrations. Patients with infection as the tentative diagnosis and
with unspecific diagnoses such as fever, cough, or dyspnea are
included. Patients in a follow-up course and with appendicitis were
excluded. Antibiotics were prescribed for approximately one-third of
all patients with a sign of infection during both intervention and
control periods. During both periods, the frequency of prescription
increases with an increased magnitude of CRP values (test for trend;
P <0.0001). During both periods, 20% of the patients with
a health-related CRP value received an antibiotic prescription. These
patients had the same distribution of infectious diagnoses as described
in Table 1
.
There were no differences between types of antibiotics prescribed during intervention and control periods. In both periods, narrow spectrum penicillin was the most frequently used antibiotic (45%), followed by macrolides (22%), broad spectrum penicillins (20%), quinolones (8%), and sulfa drugs (3%).
Patient delays for collecting antibiotics.
By comparing the
date that CRP was measured in each patient with the date that the
prescribed antibiotic was dispensed by a pharmacy, a delay for
collecting antibiotics can be estimated for each patient (Table 4
). Patients with a CRP value >50 mg/L had a significantly lower
delay in collecting their antibiotics during intervention periods than
during control periods.
results at the level of general practice clinics
Intervention effects.
Quick access to CRP measurements led to
a reduction of the overall use of ESR by GPs of 8%, whereas their use
of other tests was unchanged. The number of CRP tests forwarded to the
laboratory was reduced by 65%, and the overall number of blood samples
forwarded by GPs was reduced by 6% (Table 5
).
Economic evaluation.
The cost-effectiveness for CRP as a
near-patient test was calculated (Table 6
) on the basis of intervention effects (Table 5
). Implementation
of CRP in general practice gave a rise in costs for the Health Service
System, but also considerable savings for the laboratories, yielding a
total savings of $111 160 per year for a Danish county with 340 000
inhabitants.
Selection bias.
The 29 clinics participating in the study were
compared with the 12 clinics in the catchment area that declined
participation. There were no significant differences for the number of
requests for CRP from the laboratory, the use of ESR, the prescription
of antibiotics, and the number and ages of the GPs affiliated with the
clinics. The same parameters were compared for the 16 clinics with an
inclusion rate >75% and the 13 clinics with a low inclusion rate, and
no significant differences were found.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
study validity
Some clinics had a low inclusion rate in the control period. This
probably reflects difficulties in remembering to include patients in a
routine situation. To ensure a high internal validity, we excluded all
patients from clinics with a low inclusion rate. In the test for
selection bias, participating clinics were found representative for all
clinics in the catchment area.
esr
Some results in this study were found significant at the level of
general practices as well as at the level of included patients. At the
level of included patients, the use of ESR decreases from 46.3% for
control patients to 35.3% for intervention patients (Table 2
). If we
look at the total use of ESR by GPs and not only at ESR used for the
patients included in this study, the decrease remains significant, at a
total of 8% (Table 5
). This decrease was achieved after only 3 to 4
months of intervention, and the decrease will probably continue over a
period of 4 to 5 years, as has been seen for clinical hospital
departments introduced to CRP (21).
blood samples mailed to the laboratory
The number of blood samples forwarded from general practices to
the laboratory decreased significantly both at the level of included
patients and at the level of GPs. In a previous study, we found that
CRP is requested in 34% of all blood samples mailed from general
practices (10). In that study, it was not possible to detect
whether CRP was requested as an additional test to other primarily
ordered tests or whether CRP was actually the blood test necessitating
the blood sample. The marked decrease in the number of forwarded blood
samples found in this study indicates that CRP actually is a blood test
that provokes a blood sample in general practice.
prescription of antibiotics
Surprisingly, we found no major changes in the prescription of
antibiotics by GPs (Table 3
). From previous published studies, it could
be hypothesized that the overall use of antibiotics would remain
unaffected on the basis of more appropriate use of antibiotics, which
means a decrease in prescriptions of antibiotics to patients with low
CRP values and an increase in the number of prescriptions to patients
with high CRP values. In this study, we found that exactly the same
number of patients were prescribed an antibiotic at a certain magnitude
of CRP values, regardless of whether the test results were given within
minutes or days (Table 3
). This observation together with a same-day
dispense rate of 70% for both periods (Table 4
) illustrates that the
effect of the clinical use of the test seems to be uniform for both
intervention and control periods. In control periods, antibiotics were
prescribed and dispensed before the GP knew the CRP value. The test
result, therefore, was not included in the diagnostic process, but
perhaps was used as a kind of quality assurance for the clinical
decision. It appeared that this pattern of application was copied
during the intervention periods, although the test result should have
been used as a part of the clinical decision-making process.
The group of patients with CRP values within the health-related
reference interval but having an antibiotic prescribed had exactly the
same distribution of infectious diseases as seen for the whole study
population (Table 1
). This means that tentative diagnoses such as
pneumonia, sore throat, and sinusitis were dominant and accounted for
>60% of suspected infectious diseases (Table 1
). In general practice,
the diagnostic value of CRP of these three diseases is well documented
(5)(6)(8). For suspected bacterial pneumonia, the negative
predictive value of a CRP value <50 mg/L is 95%, which means that
only 1 of 20 patients will have bacterial pneumonia and a CRP value
<50 mg/L simultaneously (5). The predictive value of a CRP
value <10 mg/L would be even higher; however, 20% of patients with
health-related CRP values had an antibiotic prescribed. The focus of
this study was to investigate changes in the behavior of GPs who have
CRP as a near-patient test, not to assess the ability of CRP to
distinguish between bacterial and viral infection; therefore, we did
not analyze culture data on every patient suspected for an infectious
disease. The available literature supports the ability of CRP, although
it is an nonspecific marker of infectious disease, to distinguish
between bacterial and viral diseases (5)(6)(7)(8)(9)(10)(11)(12)(13). If we accept
this ability to distinguish between diseases, we should have seen a
decrease in the prescription of antibiotics to patients with low CRP
values during the intervention periods and an increase in the
prescription of antibiotics to patients with high CRP values, but this
was not seen. Therefore, it seems reasonable to assume that
implementation of clinical guidelines for the use of CRP in general
practice could give a more appropriate use of antibiotics.
For a single subgroup of patients, a quick CRP result had an effect.
Patients tentatively diagnosed clinically as having an infectious
disease and with CRP values >50 mg/L collected their antibiotics
sooner from pharmacies when the CRP measurement was made in the clinics
(Table 4
). This suggests that the GPs encouraged their patients to
collect their antibiotics the same day, because a high CRP value
indicates not only a bacterial infection, but also the severity of the
infection (4).
telephone consultations
At the level of included patients, a significant decrease in the
number of telephone consultations during the follow-up periods during
intervention was seen, probably reflecting that patients had received
the test result at the primary consultation and that their case was
then closed. It was not possible for us to evaluate this result at the
level of GPs because we only had permission to receive data for
follow-up consultations for included patients.
reasons for measuring crp
Regarding the reasons for CRP measurements, only small differences
between the study periods were noticeable. For both periods, a major
part of the tests was used for diagnosing a new disease; however,
during intervention periods this use was significantly more frequent.
This result reflects that the GPs found having the test result of the
CRP test while the patient was still in the clinic more useful for
diagnostic purposes. A decrease in the frequency of using CRP for
follow-up purposes was revealed. CRP is known to be very useful for
monitoring the treatment course of infectious patients, by using the
patients as their own references (4). One could consider
that a decrease in follow-up testing could give rise to several
recurrences and, therefore, higher costs. We found that the decrease in
follow-up testing was seen mainly for the group of patients with
noninfectious disease, which typically means patients with chronic
inflammatory diseases.
cost effectiveness
Implementation of a near-patient test for CRP in general practice
in the county of Vejle would give the Health Service System a direct
cost of $69 400 per year. The laboratories in the county would save
$180 560 per year; because the counties are the owners of the
hospitals in Denmark, the overall cost reduction would be $111 160 for
the county of Vejle.
In conclusion, we have evaluated the clinical, organizational, and economic consequences of implementing a near-patient test for CRP in general practice. Although the clinical alterations were small but relevant, we still calculate a direct cost reduction of approximately $110 000 per year for a Danish county with 340 000 inhabitants. If implementation is followed by education and clinical guidelines, additional beneficial outcome effects seems possible, based on an additional reduction in the use of ESR and a more appropriate use of antibiotics.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  J. W L Cals, F. H F Chappin, R. M Hopstaken, M. E van Leeuwen, K. Hood, C. C Butler, and G.-J. Dinant C-reactive protein point-of-care testing for lower respiratory tract infections: a qualitative evaluation of experiences by GPs Fam. Pract., December 18, 2009; (2009) cmp088v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Cizman, T. Srovin, M. Pokorn, S. Cad Pecar, and S. Battelino Analysis of the causes and consequences of decreased antibiotic consumption over the last 5 years in Slovenia J. Antimicrob. Chemother., May 1, 2005; 55(5): 758 - 763. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. M Eldridge, D. Ashby, and G. S Feder Informed patient consent to participation in cluster randomized trials: an empirical exploration of trials in primary care Clinical Trials, April 1, 2005; 2(2): 91 - 98. [Abstract] [PDF]
|
|
|
|
 |
|
 C. P Price Regular review: Point of care testing BMJ, May 26, 2001; 322(7297): 1285 - 1288. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
