Intended for healthcare professionals

Education And Debate Controversies in management

Should general practitioners perform diagnostic tests on patients before prescribing antibiotics?ForAgainst

BMJ 1999; 318 doi: (Published 20 March 1999) Cite this as: BMJ 1999;318:799

Should general practitioners perform diagnostic tests on patients before prescribing antibiotics?

As most antibiotics are prescribed by general practitioners, control of antibiotic resistance depends greatly on rational prescribing behaviour by these doctors. Here a Danish microbiologist argues for near patient testing before prescribing antibiotics, while a British general practitioner contends that there are considerable problems associated with testing and that time is often the best treatment for acute, self limiting infections.


  1. Hans J⊘rn Kolmos (hans.joern.kolmos{at}, consultant microbiologist
  1. Department of Clinical Microbiology, Hvidovre Hospital, University of Copenhagen, DK-2650 Hvidovre, Denmark
  2. Primary Medical Care, Southampton SO16 5ST

    As 80% of antibiotics are prescribed by general practitioners,1 control of antibiotic resistance depends greatly on rational prescribing behaviour in this group of doctors. Microbiological testing is an important tool in rational prescribing of antibiotics, not only in hospitals but also in general practice. However, the usefulness of laboratory based testing may be limited if it takes too long—particularly in patients who are considered to need immediate antibiotic treatment. These patients will therefore be treated empirically, and test results received afterwards from the laboratory may have little influence on the course of treatment.

    One way of overcoming these problems is to perform near patient microbiological testing. In Denmark, many general practitioners undertake selected microbiological tests using phase contrast microscopy and simple diagnostic kits. 2 3

    Advantages of near patient testing

    Practice based microbiological testing has several advantages. Most importantly, the test result can be available immediately, and this helps to decide whether the patient should be prescribed antibiotics. General practitioners are often under pressure from patients who believe they need antibiotics.4 With a negative test result at hand it is much easier for the doctor to refuse to prescribe an unnecessary antibiotic. Furthermore, if the patient needs treatment immediately, the test result may help in choosing the most appropriate agent.

    Near patient testing also reduces bureaucracy. General practitioners do not have to spend time filling in the forms that have to accompany specimens sent to the laboratory, communication problems with the laboratory are obviated, and money spent on envelopes and postage can be saved. This suggests that practice based testing will often be cheaper than laboratory based testing. Last, but not least, practice based testing may increase the satisfaction the doctor gains from working with his or her patients.

    Microbiological tests suitable for near patient testing in general practice

    View this table:

    Quality issues

    The standard of testing in general practice must be comparable to that in the microbiology laboratory. This imperative limits the types of tests that can be performed in general practice—only simple, rapid, and reliable tests are suitable. Furthermore, test results must be easy to read, and their clinical interpretation must be unequivocal. Tests that do not meet these requirements are unsuitable for the pressurised working conditions of general practice, where only a short time is set aside for each patient. In addition, a given test must be performed frequently, and the positive rate must be sufficiently high that a positive result is seen regularly. If these requirements are not met, the general practitioner will never develop enough experience, and the quality of testing will be unsatisfactory.

    The table shows the more important microbiological tests which, according to the above principles, are suitable for general practice. Some of these tests are performed with commercial diagnostic kits, but the most important diagnostic tool is a good microscope. It must have phase contrast—this allows microscopy to be carried out on unstained material, which simplifies the preparation of slides and saves time.

    Embedded Image

    Important near patient tests

    The most important microbiological tests carried out in general practice are those for vaginal discharge, dysuria, and sore throat. Vaginal discharge is one of the most common problems about which women consult their general practitioner. Comparative studies have shown that examination of a vaginal smear by phase contrast microscopy in the general practitioner's surgery gives a much more precise diagnosis than cultures performed in the microbiology laboratory.5 This techniques can also be used to validate the quality of cervical smears sent for cytology screening.

    “The most important diagnostic tool is a good microscope”

    Urine microscopy is excellent for detecting urinary tract infection. In experienced hands, it is as rapid as using urine test strips, and more informative. Furthermore, urine microscopy may help to identify patients in whom culture is indicated and those for whom other kinds of investigation—for example, chlamydia testing or examination for atrophic vaginitis—should be considered.

    Using diagnostic kits to detect group A streptococcal antigen is a much more precise way of identifying streptococcal tonsillitis than clinical criteria alone. Group A streptococci are the only well recognised cause of bacterial tonsillitis, and they are still uniformly susceptible to penicillin. Culture is therefore indicated only in patients with recurrent infection, in whom other organisms may occasionally play an aetiological part.

    Training and quality assurance

    Practice based testing calls for education and quality control. In particular, courses in microscopy are needed. In Denmark, general practitioners and clinical microbiologists have worked together for several years to develop courses in microscopy for beginners and more experienced operators. Quality assurance for tests to detect group A streptococcus antigen and other commercial kits is important because these can be sold in most countries without any official licence. Clinical performance studies of diagnostic kits for detecting group A streptococcal antigen marketed in Denmark show that the quality of these may vary appreciably.6 As decentralised microbiological testing becomes more common the need for formalised quality assurance programmes will grow. Microbiology laboratories must undertake this task, if necessary in collaboration with other laboratory specialities such as clinical biochemistry. Kolmos


    Competing interests: None declared.



    1. Paul Little (psl3{at}, MRC clinician scientist
    1. Department of Clinical Microbiology, Hvidovre Hospital, University of Copenhagen, DK-2650 Hvidovre, Denmark
    2. Primary Medical Care, Southampton SO16 5ST

      The use of tests in diagnosing medical problems is fundamental to medical training from the earliest stages. In acute infection, near patient testing to target antibiotics at those patients who are most likely to benefit seems a logical way of rationalising antibiotic prescribing. However, several aspects of diagnostic testing in general practice still need to be assessed critically before its value can be judged.

      Are tests feasible and valid?

      The commonest acute infections managed by general practitioners are those of the respiratory system (acute sore throat, acute cough or bronchitis, acute red ear or otitis media, acute sinusitis) and the urinary tract. 1 2 Near patient testing for microbiological diagnosis is currently not feasible in routine settings for sinusitis and acute otitis media without perforation, 3 4 and is not useful for acute cough.5 I will therefore focus on acute sore throat and acute urinary tract infection.

      The proposed standard diagnostic tests for acute sore throat are the throat swab or rapid test to detect group A β haemolytic streptococcus antigen.6 However, the throat swab does not differentiate between infection and carriage, and community studies have shown that its sensitivity, specificity, and positive predictive value are disappointing compared with the antistreptolysin-O test. 6 7

      Although examination of a midstream specimen of urine remains the standard investigation for acute urinary tract infection, debate continues about the cut off point that represents infection.8 In practice, most general practitioners do not have the time, resources, or expertise to perform their own microscopy and culture of midstream specimens of urine, and the diagnostic delay that occurs with laboratory testing limits the clinical usefulness. Urinary reagent test strips may be a feasible alternative, but the quality of studies assessing their use is poor. Few studies have been performed in primary care (RA Moore et al, personal communication), and preliminary evidence shows that test strips may not perform much better than clinical scoring methods.9 There is also evidence that the performance of reagent test strips is reduced in everyday conditions.10 This illustrates the general concern over the validity and effectiveness in daily practice of near patient testing.11

      Can tests hasten recovery?

      Even assuming that valid tests were available, does better diagnosis lead to an appreciable health gain? Both acute sore throat and acute urinary tract infection are short and self limiting conditions, and the need to diagnose the causal organism is therefore questionable. Although targeting treatment for sore throats at streptococcus may reduce the duration of symptoms by 20-30%, 12 13 estimates from three trials suggest that recurrence is also increased by 20%, which probably negates any benefit of more rapid symptom resolution.14

      With regard to acute urinary tract infection, the usefulness of a definitive diagnosis may also be questionable.15 Symptoms in 50% of patients with bacteriuria will probably resolve within three days without antibiotic treatment, and reasonable alternative treatments exist. 15 16

      Can tests prevent complications?

      It is usually assumed that bacterial infections are more likely to lead to complications and should therefore be targeted preferentially. However, viral infections are more common,6 and complications may also result from secondary bacterial complications of primary viral infections. Since there are few data about complications, we in primary care have little idea whether better microbiological diagnosis is the key or whether it is more important to identify clinical or sociodemographic subgroups who are at risk. Since complications are rare, large prospective cohorts will be needed to generate unbiased clinical and microbiological data. In the meantime, if we make the generous assumption that every complication is preventable, we would have to test hundreds of people to prevent even the most common non-life threatening complications such as quinsy.17

      Is there net patient benefit?

      Showing that a better diagnostic test is available is not enough—the test has to be seen to produce an appreciable health gain. Even using rapid tests, general practitioners changed their prescribing decisions in only 13% of cases, thus questioning the net health gain from better diagnostic tests.18 Rapid access to diagnostic tests for respiratory illness has not reduced antibiotic prescribing.19 Randomised trials of the use of diagnostic tests and other diagnostic strategies in self limiting conditions are needed before their widespread use is advocated.

      Are tests cost effective?

      Health service resources are scarce. A throat swab or midstream sample of urine costs £5-£7 per test, and using diagnostic tests will increase health service costs greatly, with little evidence of any health gain. The use of rapid tests or urinary test strips is not likely to reduce antibiotic prescribing costs for those general practitioners who are low prescribers of these drugs, but could reduce costs for high prescribers if other health service costs were not increased.9

      Could using diagnostic tests increase other costs? More testing will increase the number of patients with false positive results and will probably increase the psychological costs.20 Greater use of diagnostic tests will probably increase spending on secondary care.21 Furthermore, the doctor's behaviour during the consultation can increase the expectations of patients and their belief in the importance of seeing the doctor. Given the central importance of patients' expectations in determining the behaviour of doctors,22 clarification of the effect of diagnostic tests on the subsequent expectations of patients is crucial. With regard to sore throat, there is preliminary evidence from a large randomised controlled trial that the prescribing of antibiotics by doctors increases patients' expectations. By increasing belief in the importance of antibiotics and increasing reattendance by 40%, the prescribing of antibiotics “medicalises” a self limiting illness and increases health service costs. 14 17

      “Rapid access to diagnostic tests for respiratory illness has not reduced antibiotic prescribing”

      Although there is little evidence on the medicalising effect of performing diagnostic investigations, this probably creates the perception that it is important to see the doctor to have the test. Indeed, there are reports from countries where diagnostic testing for sore throat is widespread (for example, Denmark) that patients expect to have a diagnostic test (J Lous, personal communication). Thus, a major cost of diagnostic testing may be that it encourages rather than reduces the expectations which patients with self limiting illness have of their doctor. Diagnostic testing may encourage more patients to visit their general practitioner. If more people see their doctor, net antibiotic use would not be reduced, and the selective pressure favouring the development of antibiotic resistance would continue. 23 24 Before advocating a major change in use of resources towards increased diagnostic testing, we need to be clearer that the benefits exceed the costs and that this is the most cost effective use of funds.


      Diagnostic tests for acute infections should be used where there is clear evidence of validity, feasibility, and cost effective health gain—including the effects on patients' expectations and on net use of antibiotics. With current concerns about workload, patient expectations, and antibiotic resistance—and until better evidence is available for the use of diagnostics test and of clinical scoring algorithms—general practitioners should probably encourage patients who are not very ill to use symptomatic treatment for most common infections without relying on either diagnostic tests or antibiotics.


      I am grateful to Drs Michael Moore, Ian Williamson, Tom Fahey, Andrew Lowes, and Catherine Hawke for comments and suggestions.

      Competing interests: None declared.


      View Abstract