Does availability of point of care C-reactive protein measurement affect provision of antibiotics in a community respiratory service?

Resistance to antibiotics is an escalating concern, both worldwide and locally. Data from Public Health England's (PHE) 2020 English Surveillance Programme places the number of severe resistant infections at 65 162 in 2019 (equating to 178 daily new antibiotic-resistant infections a day), an increase of 7.2% from the previous year, which was a further 9% increase from the year preceding that. Across the world, mortality associated with antibiotic resistance is predicted to continue to rise (European Medicines Agency, 2009), with some estimates of costs in losses to productivity globally amounting to tens of trillions of pounds (PHE, 2015).

Between 2015-2019, antibiotic prescribing within general practice in England has declined by around 12%; however, antibiotic prescribing in community settings has increased by almost 29% over the same period when comparing defined daily doses per 1000 inhabitants per day (an estimate of the daily population proportion receiving antibiotics (World Health Organization, 2021). This is a large increase, although general practice prescribing currently accounts for 71% of all prescribing, compared to 4% within other community settings (PHE, 2020).

The management of infective exacerbations of both chronic obstructive pulmonary disease (COPD) and bronchiectasis primarily involves antibiotics. In the UK, the prevalence rate of COPD in the adult population stands at around 2% for those aged over 16 years and 4.5% for those aged over 40 years (Harries et al, 2014; National Institute for Health and Care Excellence (NICE), 2019a); around half of these experience an exacerbation at least once annually (Al-Ani et al, 2013). Bronchiectasis prevalence is estimated at approximately 0.3% of the UK population (Snell et al, 2019). NHS England (2014) data demonstrated 115 000 hospital admissions related to COPD per year, making it the second biggest cause of emergency admissions (NICE, 2015) and the largest cause of all readmissions (Wise, 2019).

Accurate diagnosis of COPD exacerbation should be based upon a clear definition that is universally agreed; however, such definitions are imprecise and lack objective measurements (Rodriguez-Roisin, 2000). One widely used definition of exacerbation is the Anthonisen classification (AC), which looks at the occurrence of one or more of three primary symptoms, including increased production of sputum, sputum purulence and increase in dyspnoea (Anthonisen et al, 1987). Although simple and clinically useful, this definition dismisses the use of objective measures in evaluating COPD exacerbation. NICE guidelines (2018) suggest looking at these symptoms alongside exacerbation and microbiology history and the indication for escalation to secondary care, dismissing current evidence surrounding the measurement of C-reactive protein (CRP) as not specific to COPD.

Bronchiectasis exacerbations are similarly defined, with the most recent British Thoracic Society guidelines (Hill et al, 2019) suggesting the presence of increased sputum production or viscosity change, increased purulence and worsening cough (with or without increasing wheeze, breathlessness, haemoptysis and systemic upset).

CRP is a blood plasma protein synthesised by the liver, the concentration of which rises in response to inflammation or trauma after a few hours (an acute phase reactant). Although CRP is not disease-specific, evidence suggests elevation in CRP levels are associated with exacerbations of COPD (Hurst et al, 2006; Chen et al, 2017), particularly bacterial infection (Gallego et al, 2016). Therefore, CRP is potentially a useful biomarker and an indicator of a beneficial response to antimicrobial treatment (Miravitlles et al, 2013) and readily available via point of care devices in primary care (Minnaard et al, 2015).

While NICE COPD guidelines (2018) have omitted CRP measurement for exacerbations, international guidelines suggest further clarification of CRP's role in exacerbation management of COPD is needed (Global Initiative for Chronic Obstructive Lung Disease, 2021), and guidance does suggest the use of CRP in pneumonia management, with indications for withholding, delaying or employing antibiotics based upon the result (NICE, 2019b). A randomised control trial within 86 general practices across the UK (Butler et al, 2019) found the inclusion of CRP measurement during a COPD exacerbation reduced patient-reported antibiotic use by 22 percentage points during an initial consultation, without a negative effect on reported quality of life, clinical or microbiological outcome measures. A similar pattern in primary care was found in a single centre study using the NICE pneumonia guidelines as a guide, with a shift from prescribing to providing ‘back-up’ antibiotics or none at all (Ward, 2018). A further study focusing on secondary care interventions (Prins et al, 2019) demonstrated a reduction of antibiotic use by almost 15 points within the hospital setting.

The local community respiratory service is responsible for providing respiratory care on behalf of GPs within the author's clinical commission group (CCG) locality for patients with a confirmed diagnosis of either COPD or bronchiectasis. The primary objective of the service is to manage acute exacerbations in the community and thus reduce unnecessary admissions. Patients on the service register have been educated on symptoms of exacerbations and are encouraged to contact the service when unwell. Antibiotic provision for infective exacerbation management, via prescription or patient group directive (PGD), is a key component of service provision. This places the service within the growing ‘other community setting’ category when looking at the prevalence of antibiotic prescription (as defined by PHE, 2020).

The service obtained two POC CRP measurement machines to support assessment of acute infective exacerbations. In this instance, the LumiratekC Analyzer reflectometry scanning device was acquired and used. A small sample (~5μl) of capillary blood is obtained for testing and mixed with a spoit (pipette) in a buffer tube before being placed onto a testing strip, which is then analysed, a process that returns a CRP measurement (range: 3-150 mg/L) within 3 minutes.

Staff had been initially trained on sample collection and device use, and guidance was issued to staff regarding consideration of antibiotics in patients with a CRP value of 50mg/L or greater (a level found to provide superior antibiotic response (Stolz et al, 2006; Daniels et al, 2009)) where diagnostic uncertainty arose, as a component of a comprehensive respiratory clinical assessment.

Aims

The primary outcome measured was antibiotic use for treatment of an acute exacerbation following a comprehensive respiratory clinical assessment, in both COPD and bronchiectasis, for those with a POC CRP measurement compared to those without one.

A secondary outcome looked at CRP elevation compared with Anthonisen criteria with respect to antibiotic use in the COPD cohort. The bronchiectasis cohort was not studied, as Anthonisen criteria were primarily formed in response to COPD management (Anthonisen et al, 1987), and suggested symptoms of an infective exacerbation involve one or more of a wider range of symptoms (NICE, 2021).

Method

A single-centre retrospective quantitative service review was established, examining all Egton Medical Information Systems (EMIS) electronic records for patients contacting the community rapid response respiratory service from 1-31 December 2019. Approval from the local department of clinical quality was obtained, following completion of the Medical Research Council/NHS Health Research authority ‘Is it research?’ tool to determine that the service review would not be considered research and ethics approval would, therefore, not be required.

An EMIS population search was performed, with the collection of the following data:

• Contact type (phonecall, face-to-face or visit)
• Age at contact
• Gender
• Smoking status
• Body mass index (BMI)
• Forced expiratory volume in 1 second (FEV1)
• Percentage of predicted FEV1 (%FEV1)
• Primary underlying condition (COPD/bronchiectasis)
• Measurement of CRP (Yes/No)
• Provision of antibiotics (Yes/No)
• Additional antibiotic treatment within 30 days.

Additionally, for the COPD cohort, the number of Anthonisen criteria symptoms documented was recorded, as was elevation in CRP of 50mg/L or greater.

Patients were excluded if they fell within the following categories:

• They required a referral to accident and emergency (AandE), as decisions to treat were then taken by secondary care clinicians
• They were already on antibiotic treatment, as this would not be a new event
• They were contacted to commence treatment following an existing sputum culture, as microbiology results are considered to be an important factor in the prescribing process (Musgrove et al, 2018)
• They had made contact with the service for a non-respiratory based query, as these contacts would not fit the criteria for community based antibiotic provision.
• Telephone contact-only episodes were also excluded, as the aim of the review was to compare CRP measurements, which would not be possible with non-contact visits, where other clinical assessment tools would also be unavailable.

If a patient required a repeat assessment during the reviewed period, only the initial contact was included, to avoid duplication of patients. However, follow-up contacts that required further antibiotic treatments were tallied for cohort and subgroup comparison.

This data was exported from the electronic record system in the form of a .xls file, and then entered into the IBM® SPSS® statistics package (version 26) for analysis.

Statistical tests were performed using SPSS®. Age, BMI, FEV1 and FEV1% as continuous measured data are represented with median and interquartile range (25th, 75th quartile) for robustness against outliers. With independent continuous data not normally distributed, Mann-Whitney U testing was selected. Gender, smoking status, antibiotic issue and CRP elevation data is categrical, presented as percentages, and the Fishers Exact Test was performed (Du Prel et al, 2010; Macridge and Rowe, 2017; Spriestersbach et al, 2019).

Results and discussion

A total of 695 service contacts were analysed, and exclusion criteria were applied (Figure 1) as described, resulting in a COPD (n=282) and a bronchiectasis cohort (n=46), with a ‘CRP’ and ‘No CRP’ subgroup for each. Within each subgroup, antibiotic provision was then recorded.

Demographic and clinical measurements for both cohorts and subgroups (Table 1) were collected, and statistical tests for significance were applied. Although the ‘CRP’ and ‘No CRP’ subgroups within each cohort were not equally populated (as this is a retrospective service review), difference in age, gender, current smoking status, BMI, FEV1, %FEV1 and further antibiotic use within 30 days did not appear to be statistically different between ‘CRP’ and ‘No CRP’ subgroups for both cohorts, with a significance level set at 0.05.

Primary outcome

Provision of antibiotics within the COPD cohort occurred with 70.1% of patients (n=141) who did not have a CRP measurement. In the ‘CRP’ measurement cohort, antibiotic provision occurred almost 25 percentage points less (24.4%), with 45.7% of patients (n=37), a statistically significant difference between the subgroups (p<0.001). A similar pattern is seen within the smaller bronchiectasis cohort, with antibiotic provision 58.7 percentage points lower when CRP is measured (81.8% vs 23.1%, or n=27 vs n=3), which is, once again, statistically significant (p<0.001). As current exacerbation guidelines suggest antibiotic treatment based upon symptom prevalence, it can be seen that antibiotic provision tended to follow symptom assessment, with CRP measurement more likely to be performed when there was only one Anthonisen criteria present, to support treatment decisions. Patients with two or more AC comprised 58% (n=47) of the COPD ‘CRP’ subgroup, compared with 72% (n=144) of the ‘No CRP’ subgroup (p=0.03).

Secondary outcome

Within the ‘CRP’ measured subgroup of the COPD cohort (n=81), Anthonisen criteria were also analysed for comparison with elevation in CRP measurement. Comparing those with two or more Anthonisen criteria (a group considered to respond more successfully to antibiotics (Anthonisen et al, 1987)), against those with less (Table 2), we again see no statistically significant difference between the groups when comparing age, gender, smoking status, FEV1, FEV1% or further antibiotics within 30 days. However, there is a statistically significant difference in elevation of CRP and antibiotics issued, as we would expect the more symptomatic patients to be assessed as more likely to require treatment, and to display an elevation in CRP levels in response to the inflammatory changes associated with an infective exacerbation.

Comparisons between these groups and the provision of antibiotics demonstrate some interesting differences (Figure 2), whereby provision of antibiotics occurs at a greater percentage for raised AC than an elevated CRP, and this phenomenon widens with an increased number of AC.

In total, for this group of 81 patients (COPD with CRP measured), CRP was 50mg/L or greater (elevated) in 24.7% (n=20) patients, and antibiotics were issued to 45.7% (n=37) patients. For patients with only one AC (42%, n=34), 8.8% (n=3) had an elevated CRP, while 20.6% (n=7) were provided with antibiotics. For patients with two AC (27.1%, n=22), 22.7% (n=5) had an elevated CRP, while 40.9% (n=9) received antibiotics; for patients with all three AC (30.9%, n=25), 48% (n=12) had an elevated CRP, whilst 84% (n=21) received antibiotics.

If we examine those with two or more AC (n=47), we can see that only 36.2% (n=17) of patients in this group had a CRP greater than 50mg/L; however, 63.8% (n=30) of patients went on to receive antibiotics, which we would expect based on existing symptom-focused exacerbation guidelines.

Limitations

Only two POC CRP measurement machines were available for the review period; however, more than two clinicians were on shift at certain times, restricting opportunities to measure CRP. Further variation in the measurement of CRP, which may have had an impact on the review, centres on the fact that CRP measurement was not integrated into the complete assessment as a required component at the time of this review. Instead, it was (and is) for the clinician to decide to utilise it based on an individual assessment, and no protocol was in place to determine this, other than to use as a guide during diagnostic uncertainty (for example, when only one AC is present). Thus, CRP will have been measured based on perceived clinical diagnostic need and tool availability.

Telephone contacts were excluded from this review as, during a telephone-only consultation, measurement of CRP was not an option. However, after other exclusions were applied, telephone contacts accounted for 41.3% of clinical contacts (n=258) and included advice on starting antibiotics. Thus, a large number of contacts and assessments were not studied, which may provide an area for further future review.

Documentation of the AC symptoms was found within the comment section of electronic notes and required manual tabulation, and depended on accurate clinical history-taking and then documentation, which is assumed for the purposes of this review. However, this may be an area of weakness to be addressed in any future study.

Further antibiotic use was recorded based on service contacts in the electronic record. However, this may not account for antibiotic provision from other care providers, potentially underreporting the incidence of antibiotic provision. Antibiotic treatment success was also not recorded, and may have played a part in further antibiotic provision.

BMI, FEV1 and %FEV1 data were taken from within 12 months of the assessment event, as patients under the care of the service received an annual review, where these were measured. While this reassessment timeframe applied to both cohorts and their respective subgroups equally, the time involved may potentially affect homogeneity of cohorts/subgroups, as the timespan before the presenting assessment was not calculated. Future work could address this.

A relatively large number of patients with COPD may also have an underlying diagnosis of bronchiectasis (Arram and Elrakhawy, 2010; Ni and Shi, 2016); however, within this study, each cohort was categorised as a distinct diagnosis, using the primary listed condition treated for each episode (as antibiotic treatment choices are slightly different for each, based on the local area prescribing committee formulary guidelines).

Finally, no data was recorded regarding any failed attempts to obtain a blood sample for CRP, or any device errors/failures for comparison.

Conclusion

POC CRP measurement was implemented in the service relatively recently, and this service review is the first attempt to analyse its impact. There are currently no specific national guidelines for POC CRP measurement in community respiratory services, although guidelines for use when treating community-acquired pneumonia have been published (NICE, 2019b). These were withdrawn and updated in response to the COVID-19 pandemic with guideline NG138 (NICE, 2019c), which does not include CRP guidance; however, there has been support for adopting the use of CRP measurement in this setting from clinical interest groups, such as the Primary Care Respiratory Society, which published a position statement (PCRS, 2021) advocating a review of POC CRP measurement within future national COPD guidelines.

The results in this study correlate with Butler et al's (2019) work, which showed a significant reduction in antibiotic prescribing following the introduction of CRP measurement. Butler et al (2019) found a 22 percentage point reduction in antibiotic prescribing, compared with a 24.4 percentage point reduction identified in this retrospective service review. Although the subgroup was smaller, this review also demonstrated a 58 percentage point reduction in antibiotic use in patients with a diagnosis of bronchiectasis. The cohort sizes represent all patients assessed by the service for a calendar month, and the initial number of contacts (n=695) presents a similar value to that of the Butler et al (2019) study (n=653), although the exclusions in this review reduced the studied number to approximately half of this initial amount.

When comparing CRP measurement with the number of Anthonisen criteria present during an exacerbation for COPD patients, an increase in antibiotic provision correlated with increased number of symptoms, and also for an elevated CRP rate; however, antibiotics were more likely to be provided with an increased number of Anthonisen criteria symptoms, rather than for an elevated CRP measurement, within this group. This further indicates a pattern of increased antibiotic provision based upon symptom history as opposed to purely CRP measurement. However, this difference may represent a clinician utilising a comprehensive approach during assessment of the presenting patient, rather than relying on a single diagnostic test or description of symptoms, and basing treatment on current guidelines, which emphasise symptom measurement. As seen with the wider cohort, antibiotic provision is still reduced when compared with the subgroup where no CRP had been measured (primary outcome).

A study by Trappenburg et al (2011) considered the use of a modified Anthonisen criteria, which also examines ‘minor’ symptoms, such as fever, as a way to potentially limit overestimation of prevalence of exacerbations. This has been shown to compare favourably with other outcomes, such as inflammatory markers (Bhowmik et al, 2000) and decline in lung function (Donaldson et al, 2002), and presents an area for further review and development within the service. A study by Miravitlles et al (2013) found that clinical failure (i.e., worsening of the exacerbation) was only significantly highlighted when sputum purulence was present with an associated elevated CRP (outlined in in this study as ≥40mg/L), which points to a further areas for study, both in symptom indicators and baseline level for CRP elevation.

The data collection for this study occurred before the COVID-19 pandemic and, while current local clinic practice with regards to sampling is unchanged (with the addition of recommended personal protective equipment), the importance of including COVID-19 testing where clinically indicated has been introduced into the service to support potential alternative diagnosis, as CRP elevation may also occur with acute COVID-19.

This study suggests the important role POC CRP measurement plays in supporting antibiotic use as an objective measure, alongside careful respiratory assessment (including symptom assessment, such as AC), when diagnostic uncertainty occurs. Scaled up with comprehensive local or national guidelines, POC CRP measurement has the potential to positively impact antimicrobial stewardship (Brett and Al-Hasan, 2019) and resistance by reducing unnecessary antibiotic provision, patient health (in terms of both morbidity and mortality resulting from antibiotic resistant infection burden) and costs in healthcare (such as those resulting from excessive provision and treating resistant infections) and beyond, through reducing loss of productivity (Frances et al, 2020).

KEY POINTS

• Effective antibiotic stewardship is important in the management of potential growth of resistant infections.
• Antibiotic provision increases with increased symptom count.
• Antibiotic provision is reduced when point of care C-reactive protein testing is performed.
• Point of care C-reactive protein measurement may aid exacerbation diagnosis when symptoms alone are not conclusive.

CPD reflective questions

• Consider the current assessment strategies you employ to assess exacerbations. Do they conform to existing strategies?
• Which symptoms might suggest the most likely group of patients to respond to antibiotic therapy?
• What threshold level of point of care C-reactive protein measurement would indicate suitability of antibiotic therapy?