Evaluating a point-of-care C-reactive protein test to support antibiotic prescribing decisions in a general practice

Introduction

Respiratory tract infections (RTIs) are commonly encountered in primary care and healthcare professionals often have to decide whether the patient should be prescribed an antibiotic. The unnecessary prescribing of antibiotics is a well debated subject and it has been reported that RTIs account for around 60% of antibiotics issued within primary care^[1]
. However, it has been suggested that acute RTIs are often viral, self-limiting and do not require an antibiotic^[1],[2],[3]
. Prescribing an unnecessary antibiotic will potentially expose the patient to needless adverse effects without aiding recovery. The concern about increasing antibiotic resistance and the growing problem of Clostridium difficile are important arguments for the prudent use of antibiotics to treat RTIs. The recent prescription of an antibiotic in primary care is recognised as a clear risk factor for individual patients developing an infection with an antibiotic-resistant organism^[4]
. Reports indicate that around 5–12% of patients presenting in primary care with symptoms of a lower RTI are diagnosed with community-acquired pnemonia^[5],[6]
and 22–42% of these patients are admitted to hospital^[7]
. The main cause of pneumonia is bacterial infection and, thus, treatment with antibiotics will be required^[7]
. However, identifying those patients presenting with a RTI who are at risk of developing complications and require antibiotics can be a challenge for healthcare professionals^[1]
.

The National Institute for Health and Care Excellence (NICE), England and Wales’s health technology assessment body, recommends that point-of-care (POC) C-reactive protein (CRP) testing should be considered in patients presenting with symptoms of lower RTI in primary care where clinical assessment is inconclusive and it is unclear whether antibiotics should be prescribed^[7]
. CRP is a non-specific marker for infection and inflammation, and is a positive acute phase protein produced by the liver on stimulation with inflammatory cytokines^[8]
. CRP combines with bacterial polysaccharides or phospholipids released from damaged tissue to become an activator of the complement pathway. A rapid increase and changes in CRP are detectable as early as six hours after the inflammatory insult, with a half-life of 19 hours. Serum CRP is widely measured, with raised concentrations occurring in both bacterial infections and inflammation. Only minor elevations in CRP levels are observed in viral infections^[9]
. Geographical location and transport barriers affect the turnaround time of results from the hospital laboratory, therefore, CRP testing has previously been of limited value within primary care to support the timely diagnosis of acute RTIs. However, POC CRP analysers can now provide a rapid CRP result within the general practice setting.

A Cochrane systematic review concluded that POC CRP testing reduced antibiotic use in patients with acute respiratory infections, with no difference in the overall clinical recovery of patients. In one study evaluated there was an increase in hospital admission in the CRP group; however, this was based on a small number of events and may be coincidental^[10]
. Studies aimed at evaluating the cost-effectiveness of POC CRP testing in managing RTIs have also highlighted a potential cost saving^{[11],[12],[13],[14]}
. The NICE adult pneumonia clinical guideline recommends that antibiotic therapy should not be routinely offered if the CRP is <20mg/L, a delayed antibiotic prescription should be considered if CRP is 20–100mg/L and antibiotic therapy should be offered if the CRP is >100mg/L^[7]
. POC CRP testing is still not routinely used within primary care in the UK, despite widespread use in many European countries^[15]
. This paper describes the pilot of a POC CRP test to support antibiotic prescribing decisions in adult patients presenting with symptoms of lower RTIs within a single GP surgery in Wales.

Method

The pilot was undertaken from November 2015 to January 2016 in a GP surgery in Anglesey, North Wales. The practice has a list size of 10,200 patients. This practice was selected because of its high antibiotic prescribing rate and the POC CRP analyser was introduced as part of a wider initiative to reduce antibiotic prescribing rates. Funding for purchasing the POC CRP analyser was obtained via a bid to the Welsh government for healthcare equipment. The Welsh government provided the funding for the CRP machine and cartridges. No funding was provided for the researchers; the pharmacists, biochemists and GPs were involved in this study as part of their work.

Approval by ethics committee was not required as the work undertaken was considered to be part of a service evaluation. The evaluation was undertaken in four key phases, as described below.

Phase 1: Device selection, procurement, protocol development and verification

The objective of phase 1 was to ensure that an appropriate POC CRP analyser was selected, procured and verified according to the local POC testing procedure^[16]
. This phase was led by a principal clinical biochemist and POC medical laboratory assistant from the blood sciences department at the local general hospital (Ysbyty Gwynedd). The health board POC committee was also consulted to approve recommendations on device selection and procurement. This was to ensure that all the clinical governance issues relating to the POC CRP analyser were considered (i.e. safety, reliability, quality, suitability, efficiency, accuracy, result interpretation and cost-effectiveness).

Laboratory evaluation was undertaken to establish the accuracy and precision of the selected POC CRP analyser compared with the Beckman AU5800 laboratory analyser (Beckman Coulter, Inc.) at the hospital (Ysbyty Gwynedd). A comparison of the CRP results from the laboratory analyser, compared with the POC analyser, was undertaken for 40 serum samples taken from patients in both primary and secondary care. The antibiotic outcome decision that would have been recommended as per the NICE clinical guideline on pneumonia in adults was also compared for the laboratory analyser and POC CRP analyser result for each of the 40 samples. The laboratory staff also provided feedback on the usability of the selected analyser in practice, including the test error rates.

A study protocol for the use of the POC CRP analyser within the GP surgery was developed based on the NICE clinical guideline on pneumonia in adults. A principal clinical biochemist and senior pharmacist developed the documents in consultation with the local POC committee, consultant microbiologist, company representative and practice manager. Standard operating procedures (SOP) for the CRP analyser, including internal quality control (IQC) procedures, monitoring, basic troubleshooting and competency documents, were prepared by the POC medical laboratory assistant in consultation with the principal biochemist. Issues relating to clinical governance, risk management, user competence training, IQC and external quality control (EQA) as defined by ISO 22870 standards for POC analysers were considered when developing these documents^[17]
. During the evaluation period, the laboratory team led on auditing error rates, quality control/assurance, troubleshooting support and organising delivery of analyser cartridges. Laboratory support also included site visits to the practice to monitor competencies and compliance with IQC monitoring.

Phase 2: Data evaluation

The objective of phase 2 was to evaluate the use of the POC CRP machine in practice. The data evaluation was split into three categories: patient outcomes; impact on overall antibiotic prescribing rates; and POC analyser outcomes.

Patient outcomes

The evaluation included all patients who had received a POC CRP test as part of their consultation within the surgery during the pilot period (November 2015 to January 2016). The practice uses the EMIS web computer system for documenting and storing patient clinical records. Each patient registered with the surgery has a unique EMIS practice code. The POC CRP analyser has a facility to enter a patient identification number and users were requested to enter this unique patient code every time a test was undertaken. Patients were identified by extracting the patient codes from the POC CRP analyser. The patient record was reviewed by a clinical pharmacist and the following information was recorded: patient age; sex; presenting complaint/diagnosis; POC CRP result; antibiotic decision; re-attendance within four weeks for same complaint; and the outcome of any re-attendance consultations. Data were extracted and analysed using Microsoft Excel 2007. The online NHS Wales Shared Services ‘primary care prescribing catalogue’ was reviewed retrospectively to check whether any of delayed antibiotic prescriptions issued were dispensed within four weeks of the initial consultation.

Impact on overall antibiotic prescribing rates

The POC CRP analyser was introduced as part of the overall initiative with the practice to reduce antibiotic prescribing rates. The overall antibiotic prescribing rates for the practice were extracted from the online NHS Wales Shared Services ‘primary care prescribing catalogue’. This database provides data for all primary care prescriptions dispensed in Wales. The overall prescribing rate between November 2015 and January 2016 was compared with the same period the previous year (November 2014 to January 2015) as an indication of the overall percentage change in antibiotic prescribing rate for the practice. The overall difference in primary care antibiotic prescribing observed across all practices within the health board (Betsi Cadwaladr University Health Board) for the same time period was also calculated as a comparison.

POC analyser outcomes

IQC compliance was monitored by the POC laboratory team to ensure that IQC samples were within the desirable specifications for imprecision of CRP reporting^[18]
and within manufacturer recommendations. Cartridge error rates were analysed by the laboratory team.

Phase 3: User feedback

A 14-item survey was distributed to all practice staff members (n=11) who had an involvement with the POC CRP analyser. The survey consisted of closed and open-ended questions to evaluate: ease of machine use, perceived impact on antibiotic prescribing, impact on patient consultation time or flow of appointments, general satisfaction with the analyser and any open-ended comments. The written survey was distributed and returned via the practice manager at practice meetings.

The written survey was distributed and returned via the practice manager at practice meetings.

Phase 4: Patient feedback

A patient focus group was undertaken to evaluate the views of patients on the POC CRP machine. A letter was sent to each patient (n=94) who had received a POC CRP test during the pilot period inviting them to participate in a focus group. Six patients responded that they were willing to participate in the focus group. These individuals were then provided with additional information about the purpose of the focus group and what to expect. In the end, five patients participated (three males and two females) as one patient did not turn up on the day. All participants provided written informed consent. Assurances were given regarding confidentiality and all participants agreed to be audio-recorded. The focus group was conducted by two primary care pharmacists, using a semi-structured guide with questions centred on patient understanding of why the POC CRP test was undertaken, experience of having the test and patient acceptability. The focus group was undertaken on the GP premises. The discussion was audio-recorded and transcribed verbatim. Transcripts were analysed manually by the two primary care pharmacists who conducted the focus group and a consensus was reached on interpretation of the contents.

Results

Phase 1: Device selection, procurement, protocol development and verification

The Alere Afinion AS100 analyser (Alere; MA, United States) was selected in agreement with the local POC committee. The cost of purchasing the analyser was £3,500. The operating cost was £3.50 per test owing to consumables (a new disposable cartridge required with each CRP test undertaken). The manufacturer of the analyser estimated that IQC would be £136 per year and EQA registration cost to be around £200–240 per year.

The comparison with the laboratory analyser was undertaken with 40 serum CRP samples across the range of 5–198mg/L. Linear regression demonstrated that the Alere POC CRP analyser correlated well with the Beckman AU5800 analyser for serum samples (R² =0.995). Bland–Altman difference plots established a negative bias of the Alere POC CRP analyser compared with the Beckman AU5800 laboratory analyser with mean difference 6mg/L/10%. There were no differences in the advised antibiotic prescribing outcome according to the NICE clinical guideline on pneumonia in adults on comparing the Alere Afinion POC analyser and Beckman laboratory analyser CRP result. Laboratory staff fed back that the analyser was straightforward to use and no problems were encountered during the initial laboratory evaluation. No errors or test failures were recorded during the laboratory evaluation of the POC test analyser.

As a result of the protocol development, several documents were produced and provided to the GP surgery to support the use of the POC CRP analyser in practice:

• A guide on the use and interpretation of the POC CRP analyser to support antibiotic prescribing in adults presenting with symptoms of a lower RTI (see ‘Figure 1: Standard operating procedure’);
• An SOP for ‘Routine analysis of samples using the Alere Afinion AS100 CRP analyser’ (outlining installation requirements, how to operate the analyser, interpretation guide, management of error codes, maintenance and IQC requirements);
• Risk assessment and control of substances hazardous to health (COSHH) documents;
• Operator competency assessment documentation.

The protocols were presented to the practice staff during a practice meeting. A representative from the manufacturer of the analyser undertook the initial operator training and the POC laboratory team from the hospital blood sciences department undertook the subsequent competency assessments with the practice staff. Competency assessments were completed by six users (one healthcare assistant [HCA], four nurses and one GP). Each trained user was given a unique user code that should be entered into the analyser with each test undertaken, as per the SOP.

Phase 2: Data evaluation

Patient outcomes

During the pilot period, a total of 94 patients received a POC CRP test. Of these, 62.8% were female and 37.2% male. Patients ranged in age from 18 years to 84 years (mean age: 49.4 years). The indications for undertaking the POC CRP test are presented in Table 1.

A total of 71 patients presented with symptoms of an acute RTI and the results are summarised in ‘Figure 2: Patient outcomes for patients presenting with symptoms of an acute respiratory tract infection’. Of these, 77.5% (n=55) had a CRP of less than 20mg/L and 22.5% (n=16) had a CRP of 20–100mg/L. There were no patients in this group with a CRP of greater than 100mg/L.

In the sub-group with a CRP of <20mg/L (n=55), 51 (92.7%) of patients were treated according to the guideline and initially were not given antibiotics. Seven of these patients re-attended with the same complaint within four weeks, two of whom had a repeat CRP and were not issued with any antibiotics. The other five were prescribed antibiotics without re-checking the CRP. Antibiotics were given in the initial consultation for two of the remaining patients with a CRP of <20mg/L and a delayed antibiotic prescription to the other two patients. Both delayed antibiotic prescriptions were not dispensed. One patient in this group was lost to follow-up as they were admitted to hospital for an unrelated condition.

None of the patients in the sub-group with a CRP of 20–100mg/L (n=16) were treated according to the guideline. An antibiotic was issued during the consultation for 62.5% (n=10) of the patients in this group, rather than a delayed antibiotic as recommended. The remaining 37.5% (n=6) patients did not receive an antibiotic. One of these patients re-attended and was subsequently given an antibiotic without repeating the CRP, while the other five patients did not re-attend. Overall, of the 71 patients in the pilot presenting with symptoms of an acute RTI, 74.6% (n=53) were not given an antibiotic. The remaining 23 patients who received the POC CRP test were for indications that were not within the agreed respiratory protocol from NICE and a summary of patient outcomes for this group is presented in Figure 3: ‘Patient outcomes for other indications’.

When extracting the CRP data from the patient computer record and comparing these with the CRP result recorded in the analyser, a minor transcription error was identified in two patient results (i.e. the user had transcribed the incorrect number from the analyser into the patient record). The transcription errors in both cases were minor and did not have an impact on clinical decision making or data analysis.

Overall antibiotic prescribing rates

The overall change in antibiotic prescribing rates for the practice is presented in table 2. A mean reduction in items of 21.39% was observed in the pilot period of November 2015 to January 2016, compared with the same period the previous year. The overall difference in primary care antibiotic prescribing observed across all practices within the health board during the same period was a 10.6% reduction (this was calculated using comparative analysis system for prescribing audit [CASPA] data, which are available through Prescribing Services Wales).

Statistical analysis

To test the null hypothesis that the rate of change in other practices was 10.6%, the research rate of reduction needs to be >10.6% to be significant. A statistical T-test was performed using Excel statistical data analysis. The p value was <0.05 and, therefore, statistically significant (see table 3).

Analyser outcomes

The laboratory POC test team reported that all IQC samples were within acceptable limits and within the desirable specifications for imprecision reporting^[18]
:

• IQC control 1, mean CRP 16.5mg/L (n=10, %CV 6.1, manufacturer states <9% for CRP <20mg/L) and;
• IQC control 2, mean CRP 54.8mg/L (n=10, %CV 4.6, manufacturer states <5% for CRP >20mg/L).

Analysis of the error failure rates revealed 26 CRP test failures during the pilot period: error codes demonstrated that 61.6% resulted from cartridge failure, 30.8% resulted from insufficient sample volume (e.g. empty capillary, air bubble in capillary or capillary incompletely filled), 3.8% were owed to the wrong sample material and 3.8% resulted from condensation detected on the cartridge or cartridge not equilibrated to room temperature.

Phase 3: User feedback

Of the 11 user surveys given out, 63.3% (n=7; four GPs, two nurses and one practice manager) responded. Of those who used the machine in practice (i.e. carried out a test on a patient), all reported that the analyser was either ‘very easy’ or ‘easy’ to use (n=3). ‘Table 4: Direct quotations from the open-ended questions in the user survey’ outlines the comments received to the open-ended survey questions.

All respondents (100%, n=7) answered ‘yes’ to the question on whether they would welcome the routine use of the POC CRP analyser within the practice. In terms of general satisfaction with the analyser, 28.6% (n=2) responded that they were ‘satisfied’ and 71.4% (n=5) were ‘very satisfied’.

Phase 4: Patient feedback 

Three concepts relating to the POC CRP test were identified from the focus group transcript analysis: understanding of test purpose, acceptability of the test and perceptions of antibiotic prescribing (see ‘Box 1: Direct quotations from patients relating to the point-of-care (POC) C-reactive protein (CRP) test’). 

Discussion

Outcomes versus NICE guidelines

There was a good uptake of the POC CRP analyser during the pilot period to support clinical decision making in patients presenting with symptoms of an acute RTI. Overall, 25.4% of the patients in this pilot presenting with symptoms of an acute RTI received an antibiotic to treat their condition. This compares favourably with the UK reported figure of a 54% antibiotic prescribing rate for RTI consultations within primary care^[19]
.

However, the patients were not always managed according to the guidance provided by NICE. In patients with a CRP of less than 20mg/L, the majority of patients (92.7%) were treated according to the protocol, with the remaining patients given either an antibiotic or a delayed antibiotic prescription. It is interesting to note that both delayed antibiotic prescriptions were not actually dispensed, suggesting it was an effective strategy to promote prudent antibiotic prescribing in these cases. 

In patients with a CRP of 20–100mg/L, NICE suggests that a delayed antibiotic prescription should be considered. A delayed prescription is where a prescription is written but instructions are given to the patient that the antibiotic should only be dispensed if symptoms worsen. It is usually kept in the GP’s reception area for 48 hours and the patient can collect it if their symptoms worsen. Patients are more likely to get this dispensed if they have a prescription in their possession. Public Health Wales produce a leaflet, ‘Get well without antibiotics’, for the patient.

However, none of the patients in this pilot with a CRP within this range were managed according to the guidance. No antibiotics were initially prescribed in 37.5% of the patients and only one of these patients re-attended and were subsequently given antibiotics. It could be argued that issuing a delayed antibiotic prescription could have avoided the need for another face-to-face consultation with the patient in this case, thus freeing GP appointments. The remaining 62.5% were given an antibiotic in the initial consultation and not a delayed antibiotic prescription as suggested by NICE. Healthcare professionals within the practice were educated on the POC CRP testing protocol at a practice meeting at the beginning of the pilot. However, there appears to be a reluctance to issue delayed antibiotic prescriptions for patients with this CRP range. There were 16 patients in the respiratory group who had a CRP of 20–100mg/L and 10 of these patients we given an antibiotic. The antibiotics were given if the level was nearer to 100mg/L than 20mg/L, and patients presenting on a Friday afternoon were more likely to have an antibiotic prescribed than those presenting on a Monday. The CRP result is an aid to the overall clinical presentation and can inform a GP’s decision on whether an antibiotic is needed. Therefore, further exploration is needed with the healthcare professionals within this particular GP surgery to identify the barriers. This is particularly important as the delayed antibiotic prescription strategy has frequently been cited as an effective method of reducing antibiotic prescribing rates in RTIs^{[20],[21],[22],[23]}
.

Although the POC CRP analyser was introduced for managing acute RTIs only, there were 23 patients who received a test for other indications. This suggests that healthcare professionals within primary care may find rapid access to a CRP result useful for managing other clinical conditions. CRP requests are also sent to the hospital for various indications, however, having a CRP machine in the practice that was able to provide an immediate result was a major advantage in the GPs’ viewpoint. For this reason, the results were divided into respiratory and non-respiratory. Even after the discussion with the GPs after the study period regarding the non-respiratory patients, they found a CRP machine very useful in the practice for determining the seriousness of some inflammatory conditions. One patient had a CRP of >200mg/L resulting from a leg infection and the POC test meant that the decision to admit the patient to hospital was quicker as there was no delay in waiting for a blood CRP result. However, further research is needed to evaluate the validity, safety and cost-effectiveness of POC CRP testing outside of acute RTIs before routine use can be recommended. The cost of undertaking each test is £3.50. Further discussion with this GP surgery to clarify which clinical indications the POC CRP analyser can be used for if it were to be introduced routinely beyond the pilot period is one way in which additional data could be collected. Furthermore, if the POC CRP analyser is rolled out to other practices within the health board, a robust process for auditing and monitoring its use would also be needed.

For the non-respiratory patients, the CRP was tested as a POC test; if the surgery did not have the machine, they would probably have requested a serum CRP. Even when the CRP was less than 20mg/L, antibiotics were given on three occasions, suggesting that the cut-off points from the respiratory NICE guideline were not relevant in this group. These results have been presented in a similar way to the RTI patients as there are no specific cut off’s for all the various indications of these patients — gout, infrapatellar bursitis, cellulitis, wound infection, urinary tract infection, diarrhoea with rigors and diverticulitis. Therefore, this report is a true reflection of the experience of a CRP machine in real life, instead of a controlled study, and reflects clinical practice.

During the pilot period there was a 21.39% statistically significant (P =0.04) reduction observed in the overall prescribing of antibiotics within the practice compared with the same period the previous year. This was greater than the 10.6% overall reduction in primary care antibiotic prescribing that was observed across all practices within the health board during the same period. The POC CRP analyser cannot account for the entire reduction as it was introduced as part of an overall initiative to reduce antibiotic prescribing. The other initiative was education from the prescribing team showing All Wales Medicines Strategy Group prescribing indicators and a visit from the antibiotics pharmacist to discuss resistance and C. difficile. However, it proved to be a useful focus from which other prudent prescribing initiatives could be developed.

Use and patient feedback

In the user survey, healthcare professionals responded positively to the analyser and all respondents reported that they would welcome the routine use of the POC CRP analyser. Healthcare professionals found the analyser useful not only to aid clinical decision making, but also to reassure patients and support their advice that an antibiotic was not needed. The survey responses indicated that users felt the CRP test did increase consultation time, but this appeared generally to be acceptable. The process by which the test was carried out varied depending on the GP. Some GPs undertook the test themselves as part of the consultation, whereas others directed the patient to a nurse or HCA and then consulted with the patient afterwards. The actual impact of this unscheduled test on the patient appointment flow is unclear. However, the survey responses indicate that this was more problematic for the nurse or HCA appointment scheduling. The impact on patient appointment flow has been cited as a potential barrier for the routine adoption of POC CRP testing and community pharmacies are a possible alternative setting for undertaking the test^[15]
.

The focus group indicated that patient acceptability of the POC CRP test was high; patients were given an explanation by the healthcare professional of why the test was undertaken and this was generally well understood. However, patients indicated that handheld information (e.g. a leaflet explaining why antibiotics were not prescribed) would have been useful. Other positive aspects highlighted by patients include the rapidness of the results, convenience, less discomfort compared with a venous sample, and usefulness to support healthcare professional antibiotic decision making. The patients had no views on the impact of the test on appointment consultation or waiting time.

Error rate

No errors were encountered during the initial laboratory evaluation, however, a high cartridge error rate was identified by the laboratory staff (as reported in the user survey). The observed high cartridge error rate highlights the ongoing support required by non-laboratory trained staff to run diagnostic tests and this is a well-recognised issue within POC testing^[24]
.

The practice ran out of the cartridges at one point during the pilot period and this was indicated as a drawback by one responder of the user survey. The cartridges have to be stored in the fridge and during the pilot period the blood sciences took the responsibility for storing any excess boxes. This was mainly owing to the limited storage space available for the boxes of cartridges within the practice. However, the expensive cost of each cartridge also meant that the laboratory staff did not want excess stock stored within the practice in case issues, such as a fridge failure, affected the stability of the stock. The laboratory staff would then deliver any required additional stock during site visits to the practice. If the POC CRP was rolled out to more GP surgeries, then consideration must be undertaken of the practical aspect for the routine storage and supply of cartridges.

Several other error logs were also identified on the analyser by laboratory staff; these were a result of a software update that had not been undertaken. This was because a prompt for the software upgrade had not been received from the analyser manufacturer. Once identified, the software update was a straightforward procedure involving an USB memory stick download. However, this demonstrates the need for a robust process for communicating and ensuring that any necessary software updates are undertaken.

Laboratory staff also noted that several CRP tests had been undertaken where the unique user number had not been entered. Upon investigation, it was highlighted that a GP who had not completed the competency assessment process had been using the analyser. The high error rate could, therefore, be attributed to individuals using the analyser who had not been trained sufficiently. This highlights the importance of a robust competence assessment process when introducing a POC CRP analyser into practice. The ‘user lock out’ facility on the analyser was not utilised during the pilot. This is where trained users are provided with a barcode (e.g. placed on their identification badge) that must be scanned before the analyser can be used, therefore, only trained users are able to activate and use the analyser. It is recommended that the lock out facility is routinely used in any further roll-out of the POC CRP analyser. A robust competency assessment process, which is routinely audited and supported by the laboratory POC team, is also a necessity.

A transcription error was also noted in two patients where the CRP result recorded in the analyser was different to the result that had been manually entered into the patient computer record. In both cases the transcription error was minor and did not impact on clinical outcomes, however, it does highlight a potential risk and governance issue. The Medicines and Healthcare products Regulatory Agency highlights the importance of clinical governance in the management and use of POC test devices^[25]
. The facility to either scan or directly upload results into clinical record and laboratory information management system should be included when considering any future wide-scale procurement of POC CRP analysers.

Sufficient funding to ensure blood sciences and medicines management staff are available to provide support would need to be considered as part of any business case for the widespread roll-out of POC CRP analysers. This would be essential to support clinical governance requirements, ensure best practice, minimise the cost of error test failures and to audit clinical outcomes. The laboratory staff have an essential role to support the clinical governance requirements of POC testing because users may lack appreciation of the potential for error^[17]
. This laboratory role includes not only initial selection and evaluation of POC analysers in comparison with laboratory analysers, but also supporting protocol development, ongoing competence assessment, audit, IQC and EQA monitoring, supporting the supply of consumables/quality control reagents and basic day-to-day troubleshooting. It was important for the GPs to have faith that the CRP machine gave the same result as a hospital serum CRP test. The health board POC committee also approved its use following the quality control analysis from biochemistry. It is essential that any device used has the assurance that it is clinically accurate from a clinical governance perspective. Audit support, antibiotic prescription monitoring and prescribing advice capacity from medicines management teams must also be considered to ensure that the POC CRP analysers are used appropriately in day-to-day practice and to further support antibiotic stewardship within primary care. A full health economic analysis will need to be carried out as a further study.

Conclusion

In conclusion, a multidisciplinary team from pharmacy, blood sciences and primary care successfully piloted the use of POC CRP analyser to support antibiotic decision making in patients presenting with an acute RTI within a GP surgery. With the rolling out of initiatives supporting the implementation of wider clinical roles for pharmacists across the UK, including the introduction of pharmacists in GP practices, their skills could be utilised in the development and implementation of similar pilots and initiatives. In this pilot in North Wales, the analyser was well accepted within the practice and there was a positive impact on antibiotic decision-making. However, this report only describes the pilot of a POC CRP analyser in one GP surgery. A comparison with the antibiotic prescribing rates in patients presenting with symptoms of an acute RTI who did not receive a POC CRP test was not carried out and this evaluation would need to be undertaken if the analyser is rolled out to other practices within the health board.