The role of appropriate diagnostic testing in acute respiratory tract infections: An antibiotic stewardship strategy to minimize diagnostic uncertainty in primary care
Antibiotic resistance crisis
Antibiotic resistance (ABR) is a critical threat to public health globally which, if unchecked, could result in 10 million deaths per year at a cumulative cost of U$ 100 trillion by 2050. 1 Overuse and misuse of antibiotics has resulted in the emergence of multidrug resistant (MDR), extensively-drug resistant (XDR) and pan-drug resistant (PDR) bacteria, which are increasingly common in South Africa. 2 To alleviate the crisis, antibiotic stewardship (AS) is a concept that has emerged that embodies the appropriate use of antibiotics with the goal of optimising patient outcomes whilst reducing the emergence of resistant bacteria. 3 The majority (75-80%) of antibiotics for systemic use in adults and children are prescribed in the community, with acute respiratory-tract infections (ARTI), such as bronchitis, pharyngitis and sinusitis the most common indications. 4,5,6,7. Although most ARTIs are of viral origin, are self-limiting, and as such non-antibiotic treatment options should be preferred, they still account for most of the antibiotics prescribed in primary health care. It is estimated that almost 60% of patients with ARTI receive an unnecessary antibiotic, which is unlikely to be of benefit. 5,6,8,9,10
The reasons for the excessive prescription of antibiotics in the community are complex and include the lack of a precise diagnosis as to the cause of the presenting respiratory illness, perceived patient and parental preference for the receipt of an antibiotic, and a lack of appreciation of the negative impact of unnecessary prescriptions, particularly with regard to the development of resistance in the community. Fear of litigation, both consciously and unconsciously, might also be a factor. Perhaps equally important are constraints on the time available to perform a full clinical assessment and the considerable diagnostic uncertainty that arises as a consequence of overlapping clinical features between bacterial and viral infections and even non-infectious respiratory illnesses.10,11.
This problem is exacerbated in community practice by the lack of availability of sensitive, specific and cost-effective tests to distinguish viral from bacterial infections. Even the presence of bacteria at the site of a suspected infection does not necessarily identify whether this is bacterial or not, as without evidence of inflammation this probably represents colonisation or contamination.12,13 This implies that a good clinical history and examination alone may not be sufficient and the judicious use of biomarkers, such as point-of-care testing (POCT) for C-reactive protein (CRP), may aid in diagnosis, and so reduce antibiotic consumption.11
The purpose of this review is to provide a simple framework for clinical decision-making regarding antibiotic use, and where applicable the appropriate use of laboratory tests for common ARTI syndromes in primary practice. This ‘diagnostic stewardship’ should be a key component of AS, providing assistance to clinicians in everyday practice to differentiate bacterial from viral infections or non-infectious conditions, and in so doing boosting confidence in decision making. Conceptually this review should be understood as one strategy within a multi-modal primary care AS program with the ultimate aim of reducing redundant antibiotic use as depicted in Figure 1.14
Patients presenting with acute cough
Acute cough, defined as lasting less than three weeks (and without more worrying features such as haemoptysis, weight loss or other symptoms of tuberculosis, chest pain or a history of aspiration) is estimated to represent the most common cause for consultation with a general practitioner and one of the commonest reasons for prescription of antibiotics in community practice; in one study, 52.7% of adult patients presenting with acute cough received an antibiotic.15 This is despite Whaley et al having recently demonstrated that the most common infective causes of acute cough were viral; the common cold, non-specific URTIs and acute bronchitis, for which antibiotic therapy is not indicated.16. Non-infectious conditions causing cough include post-nasal drip from allergic rhinitis, gastro-oesophageal reflux, smoking, angiotensin converting enzyme inhibitors and undiagnosed asthma. The symptoms of an acute cough can take up to 3 weeks to settle, and it is this prolonged duration that often precipitates the initial antibiotic prescription (and possibly even a second one), out of desperation either on the part of the patient or the doctor. 17
The common cold
It should be understood that the clear nasal secretions characteristic of the common cold frequently become purulent without signifying superimposed bacterial infection (Figure 2). This is a common misperception and hence, even with coughing, which is a normal accompaniment of a post-nasal drip, antibiotics are not required. 8 Patients should be advised that the average duration of the illness (before and after seeing the doctor) is in the region of 10-11 days. 17. Obtaining nasal swabs for culture is strongly discouraged even if the cough is persistent, and is particularly wasteful in resource-challenged, low- and middle-income countries (LMICs).
Only symptomatic treatment should be provided, which may include analgesics, such as paracetamol (for relief of fever, headache or sinus discomfort). Aspirin should be avoided in children less than 18 years of age because of the risk of Reyes disease. In those ≥12 years of age, oral or topical decongestants may relieve rhinorrhoea and nasal congestion; however these should not be used for more than 5 days as longer durations may lead to rebound congestion known as rhinitis medicamentosa. Early use of a nasal steroid may reduce the duration of the post nasal drip-induced cough.
Clinically it can be difficult to differentiate acute bronchitis from other conditions that may present with cough.18 The major differential diagnosis is community acquired pneumonia (CAP) which does require antibiotics as depicted in Figure 3.
However, the presentation is very similar to that of a patient with the common cold. 19 The latter is usually associated with nasal congestion and rhinorrhea and typically lasts only 7-10 days, but it is important to recognise that bronchitis also seldom occurs in isolation as it is frequently results from a “sinobronchitis” occurring as a consequence of a virally-induced post-nasal drip. The presence of discoloured sputum should also not be used as an indicator of bacterial, as opposed to viral, infection as both of these and also non-infectious conditions, such as asthma, frequently present with purulent or purulent looking sputum.18
In most cases, history and physical examination are sufficient to identify more serious conditions requiring chest radiography (CXR). Only in select circumstances may additional diagnostic testing be required. If it is suspected that an outbreak of bronchitis might be due to Mycoplasma pneumoniae or Chlamydophila pneumoniae, especially when presenting in young adults, a specimen for molecular confirmation may be warranted. A swab for Bordetella pertussis polymerase chain reaction (PCR) is warranted in unvaccinated patients with a paroxysmal or ‘whooping’ cough. 18
Data is limited regarding the management of acute bronchitis in children. Most would regard it as an extension of a non-specific viral URTI and no laboratory testing would be advised. Microscopy, culture and sensitivity (MC&S) on sputum specimens is strongly discouraged, as antibiotics may inappropriately be prescribed for growth of colonisers or normal flora. Perhaps the most important feature of a more severe condition requiring a CXR is tachypnoea.
Antibiotics do not alter the natural course of acute bronchitis and should not be prescribed unless co-morbid risk factors are present,18 as depicted in Figure 3. Despite most cases being viral in origin, physicians in the United States continue to prescribe antibiotics to adults with this diagnosis despite clear-cut guideline recommendations to the contrary.16 In fact, despite significant interventions, including guidelines, quality measures, and more than 15 years of educational effort by the Centres for Disease Control and Prevention (CDC), the antibiotic prescribing rates for acute bronchitis, which should be zero, have remained at 71% and has even increased between the periods 2006 to 2010.20.
Management should include advice to the patient as to the benign nature and course of the condition, with the potential for symptom resolution to take up to 3 weeks, together with a nasal steroid and with instructions for the patient to return should symptoms fail to resolve or if a pyrexia, pleuritic chest pain or haemoptysis ensue. A macrolide antibiotic should be prescribed for confirmed Bordetella pertussis infections and although it has limited symptomatic effect, it does reduce the risk of transmission.
Respiratory viruses are the most common causes of acute pharyngitis/tonsillitis whilst group A β-haemolytic streptococci (GABHS) (S. pyogenes), is the most important bacterial pathogen, accounting for 15 and 40% of adult and paediatric cases, respectively. 21,22 It is important to note that asymptomatic carriage of GABHS occurs in up to 20% in children and in up to 5% of adolescents and young adults. 23
Certain clinical features may assist in distinguishing viruses from GABHS (Figure 4) and several clinical prediction rules to diagnose GABHS have been published. 24 However, reliance on these rules alone may still lead to antibiotic over-prescription. 25 Currently, rapid antigen testing (RADT) methods have not been validated in South Africa and are therefore not routinely available. In a recent meta-analysis, the accuracy of RADT in children with pharyngitis selected as likely to have GABHS by means of clinical prediction rules, ranged from 24% (95% CI, 21 to 27%) to 86% (95% CI, 84 to 89%).25 According to the authors, none of 16 RADTs tested had good correlation with the diagnosis of GABHS by culture particularly since the disease spectrum, the size of the bacterial inoculum and the skill of the operator all affect test performance. 25, 26 Molecular methods including polymerase chain reaction (PCR), offer the highest accuracy for GABHS detection (approaching 95-100%) but widespread use is limited by limited resources. 25
CRP point-of care testing does not distinguish between those that do or do not require antibiotic therapy. 27 In addition, CRP values of 10-60 mg/l measured in the laboratory may not be able to distinguish between viral and bacterial pathogens in patients with symptoms of less than 7 days duration. 28 Whereas measurement of the absolute neutrophil count may increase diagnostic accuracy for GABHS, when rapid antigen testing is not available, routine use is also not recommended, 29 nor are anti-streptococcal antibody titres as they reflect past and not current infections. 30 Taking all these factors into account, the diagnostic and antibiotic treatment criteria are summarized in Figure 4. Correctly performed throat cultures are still considered the gold standard for the diagnosis of acute GABHS pharyngotonsillitis with a high sensitivity of 90-95%. 30 To achieve this, it is recommended that the tonsils and posterior pharyngeal wall be vigorously swabbed. 21
It is important to note that a delay in antibiotic prescription pending availability of culture results does not reduce efficacy in the prevention of acute rheumatic fever and as such a delayed antibiotic prescription strategy is advised for all patients where a throat swab has been sent for MC&S. However, throat swabs for confirmation of GABHS may not be feasible in many SA settings owing to increased direct and indirect financial costs and the need for additional healthcare visits. In such cases, empiric antibiotic therapy is advised for all patients aged 3-21 years that fit the clinical prediction rules. Generally speaking, the mainstay of the management of acute pharyngitis is symptomatic which includes adequate analgesia and antipyretics.
Acute otitis media (AOM)
Correct clinical diagnosis of AOM is the key to reduce overall antibiotic prescribing for this very common condition. Approximately 75% of children have had at least one episode by 3 years of age.8 Time and clinical acumen is required to visualise the eardrum. Symptom presentation varies with age and because typical symptoms overlap with other conditions, a clinical history alone is not sufficient to predict whether AOM is present or not. To confirm the diagnosis, inflammation of the eardrum and a middle ear effusion i.e. fullness, bulging, cloudiness and redness of the tympanic membrane (TM), must be identified (Figure 5).
Although redness of the TM is an early sign of otitis media, alone it is not diagnostic as there are numerous other causes which include crying, otitis externa, myringitis and barotrauma.8 As a consequence, AOM is frequently mis- or over-diagnosed. Although clear visualization of the TM is difficult at times, and because AOM is typically self-limiting, a high degree of diagnostic certainty is essential to minimize antibiotic overuse. Previous studies have shown that with correct diagnosis, unnecessary antibiotic use can be reduced by up to 66%. 31 Pneumotoscopy and tympanometry are very useful in determining the presence of a middle ear effusion. Routine tympanocentesis on children with uncomplicated otitis is not feasible because of ethical and resource related considerations. Swabs and/or aspirations of otorrhoea (clear or purulent secretions) from a fresh tympanic perforation in patients without prior exposure to antibiotics, that is neither recurring nor chronic, are also not routinely advised. However, with cases of AOM that closely follows swimming in children with perforations or Grommets, a MC&S is very useful in excluding Pseudomonas aeruginosa which might be critical. Of note, CRP is unhelpful in determining whether antibacterial therapy can be withheld.
As AOM is often viral in aetiology (50% of cases), and with most bacterial cases (commonly Streptococcus pneumoniae, Haemophilus influenzae) resolving spontaneously, treatment of pain is the most critical aspect of the management of AOM. Antibiotics may be deferred for 48 hours in children ≥2 years of age while symptomatic therapy is administered (particularly where good follow-up is possible), if there is unilateral involvement and if the disease is not severe.8,32 (Figure 5).
Antibiotic prophylaxis in recurrent cases is of no value.
Acute bacterial rhinosinusitis (ABRS)
ABRS is usually preceded by a viral URTI (Figure 6). Allergy, trauma, dental infection or other factors that cause inflammation of the nose and paranasal sinuses may also predispose individuals to ABRS.8 A raised CRP has been suggested as a potential discriminator between coryza and sinusitis but absolute values have not been validated. 11
Attempts to establish an aetiological diagnosis of sinusitis are typically reserved for patients who have not responded to conventional medical treatment within 48-72 hours, in immuno-compromised patients and in the presence of complications. Sinus puncture and aspiration are the reference standard for a bacteriologic diagnosis of ABRS. As most studies have shown poor correlation between nose and throat cultures with maxillary sinus aspirates, nasal or nasopharyngeal swabs are not recommended.
A recent meta-analysis of treatment outcomes for ABRS has shown marginal benefit of antibiotics over placebo. 33 Overall, the number needed to treat for one adult to benefit is 13 (95% CI 9 - 22). Neither topical nor oral decongestants and/or antihistamines are recommended as adjunctive treatment whereas both intranasal irrigation with either physiologic or hypertonic saline and intranasal corticosteroids are recommended, the latter primarily in patients with allergic rhinitis.
Influenza epidemics occur every year in South Africa during winter, typically from April to September, and should be considered in the differential diagnosis during this period. The clinical presentation varies from asymptomatic to severe and life-threatening infections. Typical and atypical presentations are depicted in Figure 7. 34 The clinical diagnosis of influenza is often difficult and unreliable and, where indicated, laboratory testing should be performed. 35 Situations where this is necessary include where there is a cluster of cases within institutions such as nursing homes, or in severe illness, where neuraminidase inhibitors (such as oseltamivir) are indicated. The latter appears to provide only marginal benefit in the non-hospitalised (in contrast to the hospitalised) patient in the community and is not recommended outside of highly immune-compromised patients or those with severe chronic respiratory, cardiac or neurological conditions. 36 The laboratory test of choice is a respiratory tract PCR on a nasopharyngeal swab. Rapid tests are not recommended as their sensitivity is often low and negative results do not exclude the diagnosis. 37
Acute exacerbations of chronic obstructive pulmonary disease (AECOPD)
With regard to chronic obstructive pulmonary disease (COPD), there really are two situations in which antibiotics may be used. The first is for the treatment of acute exacerbations of COPD (AECOPD), which is the focus of this section. The second is the long-term antibiotic use for COPD patients who continue to have recurrent exacerbations despite optimal COPD treatment; description of this is beyond the scope of this document.
With regard to AECOPD, antibiotics are not always required and there is still uncertainty as to which patients would most benefit from antibiotic use. Up to 60% of AECOPD is said to be due to airway infection, but not all of these are bacterial, and viral infections do play a significant role. 38 In general terms however, antibiotic use appears to be of more value in patients with more severe COPD and in those with more severe exacerbations of COPD. Severity of AECOPD is usually classified using the Anthonisen criteria.39 Type 1 exacerbations, which are considered to be severe, have all the three cardinal symptoms of increasing dyspnoea, increasing sputum volume and sputum purulence, while type 2 exacerbations (moderate exacerbations) have two of those three symptoms and type 3 exacerbations have one of the symptoms, together with one other symptom such as URTI, fever, wheeze, cough, increased respiratory or heart rate. 39 The recommendations for routine antibiotic use in non-hospitalized patients and CRP testing are summarized in Figure 8. 40, 41, 42
Community-acquired pneumonia (CAP)
Most definitions of CAP include symptoms of a LRTI (productive cough, pleuritic pain and dyspnoea), focal chest signs (dullness to percussion, bronchial breathing and crackles), as well as systemic features such as tachypnoea, tachycardia and a temperature of ≥38°C. 43 These features are however non-specific and can be present in other infections, such as acute bronchitis, acute sinusitis and non-infectious conditions although tachypnoea and tachycardia are uncommon. Symptoms and signs are also often less pronounced in elderly patients.44 If there is doubt as to the diagnosis,45 POCT of CRP may be helpful and is discussed below and depicted in Figure 3.
Routine chest radiographs (CXR) are not always available in primary care. If readily available X-ray is recommended if the signs and symptoms of CAP as described above are present or the POCT CRP is > 100mg/L, and if the diagnosis is uncertain. The former is useful to exclude conditions other than pneumonia such as empyema, lung abscess and bronchiectasis. Due to low sensitivity, routine microbiological tests such as Gram-stain, sputum or blood cultures are also not indicated for the management of outpatients with low- severity CAP.43, 46, 47 In children, urinary pneumococcal antigen testing is not recommended for the diagnosis of pneumococcal pneumonia due to a high rate of false-positives.46
Outpatient versus inpatient management: Severity assessment
In adults, once a clinical diagnosis of CAP has been made the primary care physician should establish the optimal site for therapy (out-patient or in-patient) using clinical judgment together with a severity of illness score such as the CURB-65 or CRB-65. 17, 48 According to Bradley et al,46 severity assessment for children 3-6 months and older should also be performed by means of presence or absence of major and minor risk factors but this is complex and might require referral.49
The role of biomarkers in differentiating types of acute respiratory tract infection
There is increasing evidence for the role of biomarkers, such as CRP and procalcitonin (PCT), in the management of patients presenting with ARTI.45 In fact, a recent Cochrane review concluded that POCT for biomarkers (e.g. CRP) to guide antibiotic treatment of ARTIs in primary care can reduce antibiotic use and used as an adjunct to a doctor’s clinical examination, this reduction in antibiotic use did not affect patient-reported outcomes, including recovery from and duration of illness. 45,50
PCT is produced primarily by the C-cells of the thyroid gland. Markedly raised levels are seen with bacterial infections and only minimal increases reported with viral infections, which through the stimulation of interferon-gamma, inhibits PCT release. Conditions such as asthma or allergic rhinitis also do not themselves, result in PCT release. However, PCT lacks the accuracy in less severely ill patients with ARTI seen in primary care, 51 and does not perform any better than a POCT CRP measurement in this setting.45 Because the negative predictive value for exclusion of bacterial infection is enhanced only by serial measurements, and by concomitant negative cultures, and because rapid PCT testing on uncentrifuged venous blood is not readily available, this test is currently unlikely to influence management of RTI in primary care.51, 52
In contrast, where there is diagnostic uncertainty measurement of CRP levels can assist the physician in the management of patients presenting with features of ARTI. Recommendations for the use of CRP are based on reports from two large research groups; the IMPAC3T programme and the GRACE consortium study. These studies, incorporated into the 2014 NICE guidelines, 53 are very useful in providing evidence to support the clinician who is uncertain as to whether to include antibiotics in the management of LRTI or not. 45 Based on POCT CRP level measurement the following is recommended (Figure 3):
- CRP <20mg/L makes a diagnosis of CAP unlikely and antibiotic therapy is not routinely indicated
- CRP between 20 -100mg/L: watchful waiting approach is recommended, where antibiotic prescription is provided only if the clinical condition deteriorates and if co-morbid risk factors are present. These would include age < 3 months or > 75 years: in children, cardiovascular and pulmonary conditions (except asthma); in adults: heart failure, severe COPD, diabetes mellitus (especially with use of insulin), neurological conditions, severe renal insufficiency or immune-compromise.
- CRP >100 mg/L makes CAP more likely and antibiotic therapy would be appropriate. 53, 54,55 These cut-offs should be used as a guide and CRP values should not be utilized in isolation without a clinical examination. Of note, it is also possible that viral infections, including uncomplicated influenza, may present with CRP values > 100 mg/L.56, 57
It is also useful to utilise CRP to identify patients presenting with AECOPD most likely to benefit from antibiotics. In one study of AECOPD comparing amoxicillin/clavulanate with placebo the best CRP cut-off for predicting clinical failure in the placebo arm was 40 mg/L, with an area under the curve of 0.732 (95% confidence interval, 0.614-0.851).42 (Figure 8).
The consequences of ABR in terms of patient outcomes and economic impact are not a distant threat, but are being played out ‘here and now’. In South Africa, increasing levels of MDR, XDR and PDR organisms have been spawned by rampant over- and incorrect use of antibiotics. 1,58. As such, antibiotics should be seen as a precious resource and the prescribing physician, particularly the GP, the guardians of this resource.
Antibiotic stewardship requires a collaborative effort on the part of policymakers, healthcare providers, health-care insurance companies and patients. Besides the urgent need for ABS governance in primary care including optimal process and outcome measures, a drastic behavioural change is required by both doctors and patients with more emphasis on non-antibiotic treatment options, particularly in the outpatient setting. To achieve this, primary care physicians need to assess their patient’s clinical presentation carefully, attempt to follow simple frameworks for clinical decision-making regarding antibiotic use and where applicable, the appropriate use of POCT testing of biomarkers and cultures in order to minimise diagnostic uncertainty. 59 Understanding their patient’s expectations along with improved communication of antibiotic resistance principles, education, and empowerment are key to reducing antibiotic use in the outpatient setting.
Without techniques to change behaviour, such as goal setting for GPs to reduce antibiotic prescriptions, provision of feedback to enable self-monitoring and action planning based upon guidelines such as these to reduce diagnostic uncertainty, antibiotic prescribing rates will remain stagnant. 60 Furthermore, according to the so called ‘Pareto principle’ or ‘Law of the Vital Few’ (derived from economic theory and implying that 80% of outcomes result from only 20% of potential causes), by focusing on a few vital interventions, one can have a significant impact on outcome with less effort. 61 From a community stewardship point of view, it implies that even if the focus is placed on the appropriate diagnosis of ARTIs alone, this would represent a major target with effect sizes of sufficient magnitude that it would have the potential to reduce the incidence of ABR bacteria in the community.11
Disclaimer: This review clinical practice guideline is not intended as a sole source of guidance in the management of patients with conditions described in this paper. Rather, it is intended to assist clinicians in decision-making. It is not intended to replace clinical judgment or establish a protocol for the care of all patients with the conditions described. These recommendations may not provide the only appropriate approach to the management of patients with such conditions.
Source: Brink AJ, van Wyk J, Moodley M, Corcoran C, Ekermans P, Nutt L, Boyles T, Mendelson M, Perovic O, Feldman C, Richards GA.