Diagnostic value of C reactive protein in infections of the lower respiratory tract: systematic review
BMJ 2005; 331 doi: https://doi.org/10.1136/bmj.38483.478183.EB (Published 30 June 2005) Cite this as: BMJ 2005;331:26All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
A clinician can use C-reactive proteïne as a diagnostic tool in a
strategy to increase certainty on the presence or absence of pneumonia.
Victor van der Meer et.al. asked themselves if there is evidence in
literature that CRP is an accurate tool. In a systematic review and meta-
analysis way they try to deterimine accuracy cut-off points for CRP. They
conclude that methodological quality of the individual studies is generaly
poor. Combined with the heterogeneity between the studies (fig. 2) we must
be very careful to draw conclusions to answer the primary question on the
accuracy of CRP. The authors rightly conclude there is insufficient
evidence in literature to support this strategy. From here it can be
confusing to readers and apparently also the editor who writes "what this
study adds". It is important to realise that here is an essential and
subtle difference between what we call “no evidence of effect and the
evidence of no effect”. So the evidence is insufficient not jet the
strategy of using CRP. That means that we must not stop using (serial) CRP
in patients suspected to have pneumonia but we have to generate high
quality diagnostic research on this subject.
Competing interests:
None declared
Competing interests: No competing interests
Sir,
It was interesting to note that CRP was not very sensitive in prediction
or absence of lower respiratory tract infection.
There was an interesting study done in neonates infection which showed
Serial CRP levels are useful in the diagnostic evaluation of neonates with
suspected infection. Two CRP levels <1 mg/dL obtained 24 hours apart, 8
to 48 hours after presentation, indicate that bacterial infection is
unlikely.
I think CRP could be used as a method of exclusion, rather than a sole
predector of infection
Ref:
1.William E. Benitz, Michael Y. Han, Ashima Madan, and Pramela
Ramachandra:Serial Serum C-Reactive Protein Levels in the Diagnosis of
Neonatal Infection, PEDIATRICS Vol. 102 No. 4 October 1998, p. e41
Competing interests:
None declared
Competing interests: No competing interests
Metlay raises some methodological issues regarding our review (1, 2).
He states that we can’t conclude on heterogeneity of sensitivities and
specificities based on the ROC-curve, since this curve represents various
cut-off points of C reactive protein. However, as shown in figure 2, we
have also aggregated the data for every single cut-off point. This figure
shows that also for separate cut-off points the sensitivity and
specificity differ widely between the reported studies. In our discussion
paragraph we chose an example with a cut-off point from a study with a
favourable sensitivity and sensitivity . However, even this cut-off point
does not provide satisfying diagnostic information. Although presentation
of a high cut-off point would result in better positive predictive values
(at the cost of the negative predictive value), we can unfortunately not
yet determine the optimal cut-off point for use in clinical practice.
With regard to the reference standards, Metlay considers the absence
of information a major limitation. In our review we chose a chest
radiograph as a reference standard, because this is at present still
considered the reference standard in international guidelines. (3) The
limitations of microbiological work-up are evident. This reference
standard should, however, not be regarded as a gold standard. Our results
reflect the poor association of C reactive protein with the outcomes of
this other diagnostic tool.
Finally, the strenght of a systematic review is that is summarizes
all available evidence, taking into account the methodological quality of
the studies. On the basis of the currently available studies we still
conclude that the diagnostic value of C reactive protein in lower
respiratory tract infection is limited. This doesn’t mean that we should
wait for the perfect test, as remarked by professor Melbye (4). However,
we should also not feel obliged to use a test which additional value is
uncertain.
1. Metlay JP. A different look at the same data
http://bmj.bmjjournals.com/cgi/eletters/331/7507/26#111570
2. Van der Meer V, Knuistingh Neven A, Van den Broek PJ, Assendelft
WJJ. Diagnostic value of C reactive protein in infections of the lower
respiratory tract: systematic review. BMJ 2005;331:26-9.
3. BTS Standards of Care Committee. BTS guidelines for the management
of community acquired pneumonia in adults. Thorax 2001;56(Suppl
IV):iv1–64.
4. Melbye H, Gonzales R. Shall we wait for the perfect test?
http://bmj.bmjjournals.com/cgi/eletters/331/7507/26#111611
Competing interests:
None declared
Competing interests: No competing interests
Re: Victor van der Meer, et al. Diagnostic value of C reactive
protein in infections of the lower respiratory tract: systematic review.
BMJ 2005;331:26-29.
C reactive protein (CRP) is used by the practitioner as a marker for
infection and disease activity, and as a diagnostic adjunct. Patients who
present with the typical symptoms and signs of lower respiratory tract
infection are physically examined and a temptative diagnosis is
established on the basis of the clinical findings, fever, cough, chest
pain, and extrapulmonary symptoms (headache, malaise, myalgias, sore
throat, gastrointestinal distress). The age and environmental history are
also taken into consideration. Pneumonia can be diagnosed from clinical
symptoms and signs. A CXR is not needed in many instances. CRP will be
determined only when the clinical picture is not clear or when the doctor
feels insecure or is urged to undertake additional investigations for non-
medical reasons. Antibiotics will be prescribed or not according to the
sum of all the above, but never routinely.
Systematic reviews are in fashion. Searching the data-bases from different
providers according to a predetermined list of subject headings and text
words serves the investigator to dissect the articles which have been
retrieved. Articles and material which do not fit all criteria are
excluded from further analysis. Statistics usually are of a complex nature
and cannot be grasped by the practitioner in full detail. The problem is
compounded when researchers focus on the wrong questions. No sensible
clinician will ever use the CRP to detect signs of pulmonary consolidation
or infiltrates on CXR.
Ueli Bollag, paediatric and general practitioner
u.bollag@bluewin.ch
Competing interests:
None declared
Competing interests: No competing interests
Dear Sir,
We read the paper by van der Meer et al. with great interest (1).
They assessed the value of C reactive protein (CRP) in detecting
radiologically proven pneumonia and evaluated how well CRP can
differentiate between bacterial and viral infections of the lower
respiratory tract. They conclude that the current evidence does not
support a wide introduction of CRP as a rapid test to guide prescription
of antibiotics.
We, also, have attempted to seek a reliable, sensitive and specific
test to answer this question in children. We looked at the possibility
that procalcitonin may provide the answer (2). Procalcitonin is a 116-
aminoacid peptide and one of the precursors of calcitonin. Increases in
procalcitonin occur more rapidly than increases in CRP. This rapid and
specific induction of procalcitonin after an adequate stimulus and the
high and reliable production of procalcitonin in patients with severe
bacterial infections or sepsis suggest a pathophysiological function of
procalcitonin in the acute immune response. In bacterial infections,
procalcitonin levels increase from values in the picogram range to levels
ranging from 1 to 1000ng/mL.
In our review of the literature procalcitonin was found to be an excellent
marker for severe bacterial infections in general. However, it was
difficult to be clear about its value in lower respiratory tract
infections in children. We included 6 studies on the use of procalcitonin
as a marker of bacterial causes of lower respiratory infections. Three of
these studies concluded that procalcitonin differentiates between
bacterial and viral infections more effectively than CRP, white blood cell
count or interleukin –6 in emergency department settings ( 3,4,5). Three
other studies stated that measurement of procalcitonin is of little value
in differentiating between bacterial and viral pneumonia (6,7,8). Results
of studies on markers of infection depend on the accuracy of the
aetiological diagnosis of lower respiratory infection. Diagnostic work-up
varied between the studies, as was also noted by van der Meer et al (1).
Also the use of antibiotics before enrolment into the studies could be a
major confounding factor.
However, studies on respiratory tract infections in adults show the
value of procalcitonin measurements more clearly. A recent prospective,
cluster-randomised, controlled, single- blinded intervention trial by
Christ-Crain et al. found that procalcitonin guidance significantly
reduced the use of antibiotics in adults with lower respiratory tract
infections (9). In the procalcitonin group the proportion of patients with
lower respiratory tract infections who received antibiotics was reduced by
47% compared with the standard group.
It is clear from the study of van der Meer et al. (1) that CRP is not
a suitable test to guide antibiotic prescription in lower respiratory
infections. We suggest it may be worthwhile to consider the use of
procalcitonin in these situations.
References
1. van der Meer V, Neven AK, van den Broek J. Assendelft WJJ.
Diagnostic value of C reactive protein in infections of the lower
respiratory tract: systemic review. BMJ, doi:10.1136/bmj.38483.478183.EB
2. van Rossum AMC, Wulkan RW, Oudesluys-Murphy AM. Procalcitonin as
an early marker of infection in neonates and children. Lancet Inf Dis
2004;4:620-30
3. Prat C, Dominguez J, Rodrigo C, et al. Procalcitonin, C-reactive
protein and leukocyte count in children with lower respiratory tract
infection. Pediatr Infect Dis J 2003;22: 963-68
4. Hatzistilianou M, Hitoglou S, Gougoustamou D, et al. Serum
procalcitonin, adenosine deaminase and its isoenzymes in the aetiological
diagnosis of pneumonia in children. Int J Immunopathol Pharmacol 2002; 15:
119-27
5. Moulin F, Raymond J, Lorrot M, et al. Procalcitonin in children
admitted to hospital with community acquired pneumonia. Arch Dis Child
2001; 84: 332-36
6. Korppi M, Remes S, Heiskanen –Kosma T. Serum procalcitonin
concentrations in bacterial pneumonia in children; a negative result in
primary healthcare settings. Pediatr Pulmonol 2003; 35: 56-61
7. Korppi M, Remes S. Serum procalcitonin in pneumococcal pneumonia
in children. Eur Respir J. 2001;17: 623-27
8. Toikka P, Irjala K, Juven T et al. Serum procalcitonin, C-reactive
protein and interleukin-6 for distinguishing bacterial and viral pneumonia
in children. Pediatr Infect Dis 2000; 19: 598-602
9. Christ-Crain M, Jaccard-Stolz D, Bingisser R, et al. Effect of
procalcitonin-guided treatment on antibiotic use and outcome in lower
respiratory tract infections: cluster-randomised, single-blinded
intervention trial. Lancet 2004; 363: 600-7
Competing interests:
None declared
Competing interests: No competing interests
One may dream about, and wait for, the perfect test for pneumonia,
with a sensitivity and specificity close to 100%. The time to wait will
probably be long. Even PCR, which can detect infinitesimally small
numbers of microbes, does not distinguish between colonization and
infection. Thus, in the current era, pneumonia is an anatomical diagnosis
based on radiographic and clinical criteria. Pneumonia refers to
infections due to a wide assortment of microbiological aetiologies, and
the severity of pneumonia varies widely as a function of host and
virulence factors. Several studies have clearly demonstrated that the
clinical findings that doctors have relied on in their diagnosis of
pneumonia, like crackles and coloured sputum, are of limited value.1-2
Although the most severe pneumonias usually, but not always, present with
a classical clinical picture, the majority of pneumonias met in general
practice are not clear cut. Even the best clinical prediction rules have
positive posterior probabilities of only about 30 to 40 percent.3 New
diagnostic measures are sorely needed.
van der Meer and co-authors have clearly summarized our knowledge so far
about the diagnostic characteristics of C-reactive protein (CRP).4 What
surprises us is that, in spite of favourable test characteristics, which
are far better than those of chest signs and symptoms, the authors do not
support the use of the test. Of course, CRP is not the perfect pneumonia
test. The CRP test only reflects the acute phase response, and must be
interpreted within a clinical setting. The duration of illness has to be
taken into account,5-6 and symptoms and signs are still crucial, albeit
low specificity. A CRP value above 200 mg/l strongly predicts pneumonia
in coughing adult patients, and no risk is usually taken by withholding
antibiotics when the CRP value is normal (which is the case in the
majority of coughing patients who have been ill more than a week).
Although we still have a way to go to determine the optimal use and
interpretation of the test, CRP testing has been shown to be associated
with lower rates of antibiotic prescriptions,7 a feat not easily
accomplished with educational interventions.8 Prospective evaluations of
management algorithms that combine CRP with clinical criteria are needed
to assess the impact of CRP testing on other outcomes of interest—such as
duration of office visit, chest X-ray utilization and return office
visits.9 We should not rely on, nor can we afford to wait for, the
perfect test.
1. Diehr P, Wood RW, Bushyhead J, Krueger L, Wolcott B, Tompkins RK.
Prediction of pneumonia in outpatients with acute cough--a statistical
approach. J Chronic Dis 1984;37:215-25.
2. Hopstaken RM, Muris JW, Knottnerus JA, Rinkens PE, Dinant GJ.
Contributions of symptoms, signs, erythrocyte sedimentation rate, and C-
reactive protein to a diagnosis of pneumonia in acute lower respiratory
tract infection. Br J Gen Pract 2003;53:358-64.
3. Metlay JP, Fine MJ. Testing strategies in the initial management
of patients with community-acquired pneumonia. Ann Intern Med 2003;138:109
-18.
4. van der Meer V, Neven AK, van der Broek, Assendelft WJJ.
Diagnostic value of C reactice protein in infections of the lower
respiratory tract: a systematic review. BMJ 2005: 331 ……..
5. Melbye H, Hvidsten D, Holm A, Nordbo SA, Brox J. The course of C-
reactive protein response in untreated upper respiratory tract
infection.Br J Gen Pract 2004;54:653-8.
6. Melbye H, Straume B, Aasebo U, Dale K. Laboratory tests for
pneumonia in general practice: the diagnostic values depend on the
duration of illness. Scand J Prim Health Care 1992;10:234-40.
7. Andre M, Schwan A, Odenholt I, Swedish Study Group on Antibiotic
Use. The use of CRP tests in patients with respiratory tract infections in
primary care in Sweden can be questioned. Scand J Infect Dis 2004;36:192-
7.
8. Gonzales R, Sauaia A, Corbett KK, Maselli JH, Erbacher K, Leeman-
Castillo BA et al. Antibiotic treatment of acute respiratory tract
infections in the elderly: effect of a multidimensional educational
intervention. J Am Geriatr Soc 2004;52:39-45.
9. Flanders SA, Stein J, Shochat G, Sellers K, Holland M, Maselli J
et al. Performance of a bedside C-reactive protein test in the diagnosis
of community-acquired pneumonia in adults with acute cough. Am J Med
2004;116:529-35.
Competing interests:
None declared
Competing interests: No competing interests
In a recent systematic review of the diagnostic value of C reactive
protein in infections of the lower respiratory tract [1], the authors
concluded that the evidence does not support the introduction of C
reactive protein as a rapid test to guide antibiotic prescriptions. I
disagree with this conclusion for several reasons. First, the analysis of
C reactive protein as a stand-alone diagnostic test for community-acquired
pneumonia is misguided. It is known that individual signs and symptoms
have poor operating characteristics as diagnostic tests, but in
combination they are much more useful [2]. As an example, temperature has
poor operating characteristics as a stand-alone diagnostic test for
pneumonia but, still, clinicians include a consideration of temperature in
the overall assessment of the probability of pneumonia because it provides
incremental information. The more useful question is whether any new test
provides incremental diagnostic information. Recent studies by Flanders et
al and Hopstaken et al found that C-reactive protein was significantly
associated with radiographic pneumonia independent of clinical criteria
[3, 4].
Second, the data are presented in a somewhat misleading manner.
Presenting the range of sensitivities (from 10% to 98%) and range of
specificities (from 44% to 99%) in both the abstract and results suggests
that estimates of the diagnostic accuracy of the test are all over the
place and, by implication, worthless. But these wide ranges reflect the
results you would observe when you analyze a test over a range of cut-off
points. All tests will display a similar sensitivity-specificity trade-
off. This in no way is an indictment of the test. Indeed, it is clear
that at the lowest cut-off points for C reactive protein, sensitivity is
typically quite high and at the high cut-off points, specificity is very
high. So, if appropriately incorporated into diagnostic algorithms, the
information can be useful across the range of values. No one would expect
a single cut-off point to both rule in and rule out a diagnosis even
though the authors present a single example applying a cut-off value of 20
and demonstrate that (not surprisingly) the positive predictive value is
poor.
Third, the absence of information on the reference standards used to
define pneumonia or bacterial lower respiratory tract infections is a
major limitation. Chest radiography is an imperfect gold standard for
diagnosing community-acquired pneumonia [5] and, similarly, the
microbiological tests used have limited sensitivity and specificity.
Therefore, the concept of false positives and false negatives is
problematic when the gold standard is tarnished. From a practical
standpoint, the true gold standard in this setting should be whether C
reactive protein measurement contributes to the correct identification of
antibiotic responsive illnesses, and in that regard, the data are
encouraging.
A minor additional point, Figure 3 is mislabeled. An ROC curve plots
sensitivity against 1-specificity.
References
1. van der Meer V, Neven AK, van den Broek PJ, Assendelft WJ.
Diagnostic value of C reactive protein in infections of the lower
respiratory tract: systematic review. Bmj 2005; 331(7507): 26.
2. Metlay JP, Fine MJ. Testing strategies in the initial management of
patients with community-acquired pneumonia. Ann Intern Med 2003; 138(2):
109-18.
3. Flanders SA, Stein J, Shochat G, et al. Performance of a bedside C-
reactive protein test in the diagnosis of community-acquired pneumonia in
adults with acute cough. Am J Med 2004; 116(8): 529-35.
4. Hopstaken RM, Muris JW, Knottnerus JA, Kester AD, Rinkens PE, Dinant
GJ. Contributions of symptoms, signs, erythrocyte sedimentation rate, and
C-reactive protein to a diagnosis of pneumonia in acute lower respiratory
tract infection. Br J Gen Pract 2003; 53(490): 358-64.
5. Syrjala H, Broas M, Suramo I, Ojala A, Lahde S. High-resolution
computed tomography for the diagnosis of community-acquired pneumonia.
Clin Infect Dis 1998; 27(2): 358-63.
Competing interests:
None declared
Competing interests: No competing interests
Hopstaken et al. hold the view that C reactive protein (CRP) is very
helpful in excluding pneumonia. However our results, when applied to
primary care, do not support this opinion.
Exclusion of a diagnosis requires a sensitive test. In the discussion
of our paper we consider the data of Melbye et al.[1] (a point right on
the summary ROC curve). Although we find a high negative predictive value
of 98.6%, we must bear in mind that the pre-test probability of a negative
diagnosis is 95%, which means that in a low-prevalence situation the
additional value of CRP is negligible.
On the other hand, due to frequently reported low rates of
specificity (see figure 2 of our systematic review), the CRP test would
result in a considerably high rate of false positive tests, which in daily
practice will most likely lead to considerable incorrect prescriptions of
antibiotics. Whether the added value of CRP, being an imperfect test, will
outweigh the potential negative side-effect of extra prescription due to
false-positive tests can only be shown in clinical trials, evaluating the
use of a CRP test in addition to a clinical prediction rule. As yet these
trials are not available.
1. Melbye H, Straume B, Aasebo U, Brox J. Laboratory tests for
pneumonia in general practice: the diagnostic values depend on the
duration of illness. Scand J Prim Health Care 1992;19:234-40.
Competing interests:
None declared
Competing interests: No competing interests
According to the results - but not the conclusions - of the
systematic review of van der Meer et al. C reactive protein appears to be
very helpful in discriminating pneumonia from acute bronchitis in adults
with a lower respiratory tract infection presenting in general practice.
The relation of CRP with an infiltrate on chest radiography (reference
standard) showed an area under the curve of 0.80, and even 0.85 for the
studies that fulfilled the Lijmer criteria. We know that the accuracy of
symptoms and signs in discriminating pneumonia from acute bronchitis in a
patient with LRTI is very poor in general practice.[1 2] Classical
symptoms and signs of pneumonia, derived from hospital studies, are of
limited value in everyday general practice, because of the lower incidence
and smaller extent of disease found there. Reviewing the accuracy of CRP
as an isolated marker of LRTI would be comparable to evaluating only one
finding of physical examination (e.g. abnormality on auscultation). The
accuracy of a single CRP test is not 100%, but at least it appears to be
much stronger than all findings from history-taking and physical
examination. These results may have important implications. GPs may, in
particular, be able to rule out pneumonia more accurately and could thus
safely withhold unnecessary antibiotic prescriptions when CRP test results
are low (e.g., <20 mg/l). This important message is unfortunately not
reflected in the conclusions of the authors.
References
1. Melbye H, Straume B, Aasebo U, Dale K. Diagnosis of pneumonia in adults
in general practice. Relative importance of typical symptoms and abnormal
chest signs evaluated against a radiographic reference standard. Scand J
Prim Health Care 1992;10(3):226-33.
2. Hopstaken RM, Muris JWM, Knottnerus JA, Kester ADM, Rinkens PELM,
Dinant GJ. Contributions of symptoms, signs, erythrocyte sedimentation
rate and C-reactive protein to a diagnosis of pneumonia in acute lower
respiratory tract infection. Br J Gen Pract 2003;53:358-64.
Competing interests:
None declared
Competing interests: No competing interests
Ethnographic studies required
I find CRP levels have more uses in clinical decision-making than the
obviously binary ie. "CRP high, therefore bacterial, therefore I must
start antibiotics" vs. "CRP low, therefore viral, therefore I will
withhold antibiotics". Quantitative research studies unfortunately tend to
favour this sort of binary outcome, single goal-of-testing paradigm.
As an example, in young infants with a pyrexial illness, it is often
not possible to initially identify the focus of infection: pharyngitis,
pneumonia, meningitis and urinary tract infection may all present
similarly. When I see a markedly raised CRP without obvious focus, I will
request more invasive or specialised tests to look for the source of
infection.
I think there is a need for qualitative research or audit, looking at
how CRP is actually used in clinical practice, who it is requested in and
why, who it is not, did it affect management, etc. This will be required
particularly if we are considering introducing new tests such as serum
procalcitonin.
Competing interests:
CRP levels influence my clinical decision-making.
Competing interests: No competing interests