Intended for healthcare professionals

Practice Diagnosis in General Practice

Test of time

BMJ 2009; 338 doi: (Published 15 June 2009) Cite this as: BMJ 2009;338:b1878
  1. Susanna C Almond, academic clinical fellow in general practice,
  2. Nick Summerton, clinical consultant
  1. 1Department of Primary Health Care, University of Oxford, Oxford OX3 7LF
  1. Correspondence to: S C Almond susanna.almond{at}

    The test of time is not appropriate when “red flag” symptoms warrant immediate referral, but it can be helpful in patients presenting with common problems, such as diarrhoea (doi: 10.1136/bmj.b1877), that have a known clinical course

    What is it?

    Patients often present to general practitioners with symptoms or signs that are ill defined and evolving, making it difficult to slot them neatly into a diagnostic category. In a cohort study of 500 patients presenting with common symptoms, 70% had improved at two weeks’ follow-up, and 60% of the remainder had improved at three months.1 Making a precise diagnosis of all presenting problems is arguably not only unnecessary but also potentially damaging both to the patient and to the overall healthcare budget.2

    General practice presents a good opportunity for using the test of time for diagnosis since arranging reassessment is relatively easy. Gray says that the general practice consultation should be seen as a total of 47 minutes spread over the course of a year, rather than a discrete 10 minute entity, allowing repeated evaluation in situations of diagnostic uncertainty.3

    Diagnosis by test of time involves a careful initial assessment of the patient’s presenting problem(s), followed by one or more reassessment(s) after predefined period(s) of time. At the reassessment(s) the symptom or sign might have become more clearly defined, might have resolved, might be unchanged, or might have worsened, or additional signs or symptoms may have appeared. Diagnosis by time therefore relies on a good understanding of the course of condition(s) to enable appropriate reassessment(s). This article considers the test of time as a diagnostic method that is often used for defining the final diagnosis (figure).4


    Stages and strategies in arriving at a diagnosis

    When can it be used?

    In applying this diagnostic strategy it is important to distinguish those patients who require rapid investigation, treatment, or referral from those who would better be served by observation over a period. Test of time should be used only in situations where the benefits of delay (lower risks of false positive diagnoses or avoiding unnecessary investigations) outweigh the harms of delayed or missed diagnoses. The test of time strategy can help to protect the patient from the harm of unnecessary investigations and the “cascade” effect of unexpected abnormal results generating further tests.5

    The test of time is not appropriate when red flag symptoms warrant immediate referral, but it can be helpful in patients presenting with common problems, such as diarrhoea, where the initial assessment has not identified any red flags and the clinical course is reasonably well established for the commoner infectious causes. Some conditions in which the test of time is well accepted have limited published data on clinical course: in the management of an isolated enlarged lymph node, for example, current practice recommends waiting 4-6 weeks before arranging further specific investigations such as biopsy, providing there are no red flag symptoms or active infection.6

    How does it go wrong?

    Problems in the initial assessment

    Inadequate assessment

    The “safety net” of a follow-up appointment could tempt a clinician to undertake an inadequate initial assessment. This may result in poor definition of the presenting problem or any associated symptoms and a failure to identify red flags. Cancer presenting in the emergency department has been cited as a marker of inadequate initial assessment in primary care.7 Primary care clinicians should be aware of the challenge of assessment in the course of an illness, before more typical features have emerged. Another potential pitfall is a reluctance to measure vital signs that might help to distinguish, for example, between serious bacterial infection and viral illness.

    Inappropriate discriminators

    If we are going to observe patients over time then the baseline assessment must be based on discriminant information applicable to early symptoms or signs (and, ideally, also relevant to the setting and the patient spectrum). For example, the classic clinical features of meningococcal disease appear late in the illness; within the first four to six hours of the onset of meningococcal disease children have non-specific features such as fever, poor feeding, and irritability.8

    Another example of an inappropriate discriminator is the “thunderclap” headache for subarachnoid haemorrhage. However, case series show that up to two thirds of patients with subarachnoid haemorrhage do not have an abrupt onset within seconds, but that the “worst headache of their lives” can evolve over several minutes.9 Consequently, many doctors incorrectly exclude the possibility of subarachnoid haemorrhage by considering that the absence of an abrupt “thunderclap” onset within seconds is a key discriminator.10

    Many colorectal cancer guidelines focus on the symptom of rectal bleeding, but the most common initial presentation for this condition is a change in bowel habits.11 The difficulty with using this symptom is that it has poor specificity and reliability.12

    Problems in reassessment

    Failed reassessment

    The patient may not return, especially if the responsibility is placed solely on them. For example, in relation to cancer, Nylenna and Hjortdahl identified an important gap between doctors’ and patients’ understanding and interpretation of some key symptoms and signs: the doctors tended to recommend a quicker response rate for most of the symptoms and signs than did the patients.13

    Biased reassessment

    Especially when reassessing a subjective symptom or a physical sign, doctors will be influenced by earlier findings,14 15 and such bias could undermine the safety of the test of time diagnostic strategy. Clinicians should be aware of this and critically re-evaluate the working diagnosis if the condition does not follow the expected course. Patients’ recollections of their symptoms are often inconsistent,16 and an individual’s physical and psychological state (which may have changed since the initial assessment) can affect their assessment of the severity of a current symptom.17

    Inability to determine important change

    It can be difficult to establish whether a clinically important change has taken place if the course of the sign or symptom or the extent of any biological variability is uncertain—for example, diagnosis of hypertension may require repeated measurement of blood pressure over time.18 Problems can occur if the usual trajectory of the symptom or sign is not linear or if it has poor reliability,16 as is the case with change in bowel habit.12

    Incorrect time frame

    It is important to identify the clinical course of the problem diagnosed or suspected, since the time frame for reassessment is specific to the condition. For example, the time course for deterioration in meningitis (a matter of hours) is different from that for viral gastroenteritis, where dehydration can develop over days.

    Narrow focus

    In general practice, it is important to appreciate that diagnosis by time is not simply about the exclusion of serious physical illness but also the simultaneous consideration of medically unexplained symptoms and psychiatric disorders. Depression and anxiety often present with somatic symptoms that may resolve with prompt and effective treatment.19

    How can we improve?

    Careful initial assessment

    Careful initial assessment should be a prerequisite for any test of time and must include information that can discriminate between possible diagnoses. The presenting problem, any associated symptoms or signs (especially red flags), comorbidities, and the patient’s current physical and emotional state must be clarified.17

    Developing an explicit and comprehensive re-assessment plan

    To minimise misunderstandings about reattendance, the follow-up plan must be explicit about reassessments, and this must be agreed with the patient. It should be based, as far as possible, on the known course of the presenting problem and the clinically important difference that needs to be detected.

    From the patient’s perspective, a good test of time focuses on specific features or changes in the condition that suggest a need for reassessment, and a time frame during which the patient should observe these features before seeking reassessment (see box for examples).

    Time frames of common symptoms

    • Acute cough in children—50% recover within 5-10 days and 90% within 20-25 days23 24

    • Children with pertussis positive serology—85% had cough two months after onset of symptoms, compared with 49% of those with negative serology25

    Sore throat
    • Sore throat (including Streptococcus positive) resolves in 40% by day 3 and in 85% by day 726

    • In serologically diagnosed Epstein-Barr virus infection, 74% have sore throat at diagnosis, 16% at 1 month, and 11% at 2 and 6 months27

    • By day 5, symptoms have resolved in 65% and by day 7, symptoms have resolved in 83% of patients28 29

    • Infectious diarrhoea in adults and children—at 3 weeks, 76% no longer had symptoms 30

    • Diarrhoea lasted three or more days in 72% of children with Campylobacter infection and 22% with norovirus 31

    Back pain
    • Acute lower back pain (without sciatica)— at 6 weeks, 40% were completely pain free; at 12 weeks, 58% were pain free; at 1 year, 73% were pain free32

    • Sciatica—at 2 years, 55% still had symptoms, and of those who recovered from sciatica, 61% still had lower back pain33

    Knee complaint
    • New episodes of pain and stiffness (excluding suspected fracture)—at 3 months symptoms persisted in 75% of patients and at 12 months symptoms persisted in 56%34

    Monitor progress with objective, reproducible measures

    Wherever possible, monitoring by the general practitioner or practice nurse should use features that can be objectively assessed and that have good reproducibility. Examples include using dermatoscopy to photograph and measure skin lesions that do not warrant immediate removal, using a ruler or callipers to record the sizes of lymph nodes, using a tape measure for serial measurements of head circumference in children, and monitoring diarrhoea by assessing the frequency of stools.

    When the patient is asked about changes in subjective symptoms, the options need to be balanced. For example, in dyspepsia, such options would be that the condition has improved, is unchanged, or has worsened.20

    Broaden the focus if necessary

    After a test of time, another diagnostic strategy, such as the test of treatment, may be needed—for example, bronchodilators for cough that does not resolve.21 Frequent attendance with the same symptom over a period, combined with an inability to make sense of the presenting symptom, should alert us to avoid thinking solely about organic disease when attempting to make a diagnosis.22


    Cite this as: BMJ 2009;338:b1878


    • This series aims to set out a diagnostic strategy and illustrate its application with a case. The series advisers are Kevin Barraclough, general practitioner, Painswick, and research fellow in community based medicine, University of Bristol; Paul Glasziou, professor of evidence based medicine, Department of Primary Health Care, University of Oxford; and Peter Rose, university lecturer, Department of Primary Health Care, University of Oxford.

    • We thank Paul Glasziou and Kevin Barraclough for their helpful comments.

    • Contributors: SA and NS drafted the initial text of the article together. SA searched for and summarised the evidence presented in the box. Both authors approved the final version. NS is guarantor.

    • Competing interests: None declared.

    • Provenance and peer review: Not commissioned; externally peer reviewed.