The evidence base for the diagnostic test accuracy of CRP testing in primary care is characterised by a high level of heterogeneity in patient populations, diagnostic criteria (including use of CRP levels in isolation or as part of a clinical algorithm), CRP cut-points, how the performance of the test is reported
and the absence of a universal reference standard for the diagnosis of RTIs requiring antibiotic treat- ment. Meta-analysis of the data was therefore not appropriate and a narrative review was presented. Planned subgroup analysis (children, older adults [≥65 years of age] patients attending out-of-hours (OOH) services and those in long-term care (LTC) facilities) were not possible due to limited data. Only one study included children (aged 3 months to 15 years) and no study specifically reported on CRP testing in older adults. Therefore no conclusions can be drawn on the diagnostic test accuracy of CRP testing in these subgroups and instead, results should be interpreted in terms of the general population. The results of this systematic review do provide important insights into the performance of CRP as a test to help identify patients who will benefit from antibiotic treatment and to aid decision- making for a number of conditions.
The evidence in relation to the diagnostic test accuracy of CRP in URTI is inconclusive. Limited evi- dence was found for the use of CRP testing in patients presenting with symptoms of acute sinusitis. Given this, and that current clinical guidelines do not generally recommend the use of antibiotics in acute sinusitis, it is unclear what the aim of CRP testing, on its own or as part of a clinical prediction rule, would be even if a suitable threshold could be established.
In pharyngitis/tonsillitis, treatment with antibiotics is generally only recommended in those with group A streptococcal (GAS) infection (5% to 30% of those presenting with sore throat). A cut-point of 35 mg/L CRP may be useful in discriminating bacterial from non-bacterial pharyngitis. One study sug- gests that at this threshold CRP may be useful as part of a clinical prediction rule in patients present- ing with sore throats, whose diagnosis remains inconclusive after clinical examination; however this score required further validation. In contrast, at a threshold of 6 mg/L the use of CRP in combination with a clinical prediction rule to rule out GAS could lead to unnecessary prescribing of antibiotic. This study also reported that CRP POCT may not perform better than other available tests (i.e. RADT) in the detection of GAS infection. (D1002)
Current clinical guidelines recommend antibiotic treatment for pneumonia irrespective of the aetiology. With the exception of patients at higher risk of complications, antibiotics are generally not recom- mended for other LRTIs as these are generally considered to be self-limiting with limited clinical bene- fit from antibiotic treatment.
The high level of heterogeneity across studies evaluating the diagnostic test accuracy of CRP testing in pneumonia patients, mainly concerning the type of intervention and how performance of the test was reported, made comparison across studies difficult. Patients with pneumonia may present with low levels of CRP, therefore use of CRP levels in isolation may lead to cases of pneumonia being missed. At a CRP cut-point of 20 mg/L, three out of four studies found the sensitivity to be <0.75, and considered it too low to use as a rule-out threshold for pneumonia, while most studies found a CRP cut-point of 50 or 100 mg/L as sufficiently specific to use as a rule-in threshold for the diagnosis of pneumonia. Many of the DTA studies used CRP testing in combination with a clinical prediction rule, making it difficult to determine the effect CRP testing had on its own. While the value of CRP testing in addition to clinical signs and symptoms for the diagnosis of pneumonia in primary care is unclear, it appears most useful in patients where primary care physicians have diagnostic uncertainty. (D1005)
rates of misdiagnosis across RTI types. Taken at face value, based on the diagnostic test accuracy, CRP POCT is not a very good test for distinguishing between viral and bacterial RTIs. However, that finding is contradicted by the significant impact on antibiotic prescribing observed in the clinical effec- tiveness trials. It may therefore be that the accuracy of the test is of lesser importance, and what is more critical is that it facilitates a discussion between the clinician and the patient and perhaps a more conservative treatment approach to managing acute RTIs.
Only four studies related to CRP POCT in primary care; all other studies used CRP tests carried out by laboratory staff either using a CRP POCT device or standard laboratory equipment. Therefore, It is not certain if the findings are applicable to CRP POCT when performed at the point-of-care by the intended user. It is noted that one study reported that differences in CRP measurements obtained with POCT devices compared to standard laboratory measurement did not translate into clinically relevant differences in the diagnosis of radiographic pneumonia in adults presenting with acute cough in primary care.
The review was limited to patients presenting to primary care with symptoms of acute RTI. This crite- rion was strictly applied, so studies that included patients presenting to other treatment settings such as hospital emergency departments, urgent care centres and outpatient clinics were excluded unless the data specific to primary care could be extracted. The applicability of data from these settings to primary care was considered limited due to differences in staffing, access to diagnostic services and the spectrum of presenting patients. Data in relation to a number of CRP POC devices were therefore excluded from this systematic review. However, it is noted that the restriction may not be relevant to all countries, where certain outpatient clinics and urgent care centres may be considered part of the primary care system.