As a practitioner, you need to be aware of important legal and professional indemnity issues when ordering and interpreting tests.

Is it the right test?
You can be held to be negligent for not ordering the correct test, or combination of tests, or failing to give enough information to the laboratory. The pathologist or radiologist needs to have a clear understanding of what information is sought e.g. a request for 'Rubella antibodies' is of little help. Although IgG and IgM antibodies can be measured by most laboratories, it is important to state whether you want to know a patient's immune status or whether you have a woman in early pregnancy concerned about recent exposure to rubella.

In pregnancy, it is critically important, if there are antibodies present, to know if they represent a recent infection or past infection or immunisation. In one such case which was settled with the plaintiff and hence never went to court, the pathologist had most of the liability, but the referring general practitioner's medical defence organisation contributed to the settlement because of the inadequate information on the request form.

Has the patient given informed consent?
A venerable obstetrician once said to me 'I've been ordering VDRLs as part of my routine antenatal screen for over 30 years, but if I now want to add a HIV antibody test I've got to go through an incredible rigmarole.' The reality is that for the last 30 years he has been acting improperly in not informing patients what tests he was ordering and why. The legislation enacted to impose, by statute, strict protocols for pre-HIV test counselling simply echoed an existing common law duty applicable to all medical procedures, including referral for investigations. You would not or should not prescribe penicillin without telling a patient what you plan to prescribe and why, without asking about past experience with penicillin and without mentioning possible serious and common adverse effects. Investigations are no different. Informed consent is essential.

Is it the right patient, the right test, the right label, the right request form ...?
Despite scrupulous attention to detailed protocols, there are still cases in which specimens are mislabelled or are otherwise wrongly identified. This may not matter in some cases, but it mattered very much to a middle-aged man who had a radical prostatectomy following a 'positive' biopsy for a cancer he did not have.

Has the test result been seen and acted upon?
The commonest error is failure to follow-up an abnormal test result. Classically, one member of a clinic orders a test and the result returns when he or she is not at the clinic. Someone puts the test aside to be drawn to the doctor's attention on their return and then someone else, noting that the test result has been 'left out', tidily puts it in the patient's file. If the patient does not telephone for the result, it will not emerge until the next visit, perhaps months later. If the patient was told (foolishly) ' If you don't hear from us, everything's OK', a tragedy can and has resulted - especially with reports of cervical smears.

The simplest - and surest - protocol is to require that no test be filed until it has been signed as having been seen by a doctor, preferably the one who ordered the test. That doctor should indicate on the result form that it is to be filed. If results are received electronically, and no hard copy is ever generated, you should ensure that your software leaves a 'footprint' to show when the result was formally transferred to the patient's computer record, and who opened the file to accept the result. Tests requiring patient follow-up should be subject to a protocol known by every member of the practice.

You have no duty to ensure that patients act on information you give them, but you do have a duty to make every reasonable attempt to get that information to them. Forget for the moment whether or not that is a legal duty. If you know a patient is at great risk, you clearly have an ethical duty to do what you can to protect them from harm by informing them of that risk. One or two attempts to telephone, at home and at work, and sending a letter by certified mail to the latest address they gave you, are examples of reasonable endeavours. These attempts should be recorded in the patient's notes.

It might be prudent to say to every patient having a cervical smear 'Please be sure to ring in two weeks to get your result'. However, the fact that they did not ring does not absolve you from a responsibility to attempt to contact them if the result is abnormal.

Is it a screening test or a diagnostic test?
Cervical smears and mammography are excellent screening tests for cervical and breast cancer. They are not 'diagnostic' tests. If reported as negative, they should not be relied upon alone in patients presenting with symptoms and/or signs. Almost every mammogram report concludes 'A negative mammogram should not preclude biopsy of clinically suspicious lesions'. A patient presenting with postcoital or intermenstrual bleeding requires full gynaecological assessment, not just a cervical smear.

The most accurate diagnostic test for cervical or breast cancer is the histopathology of the specimen removed at a hysterectomy or mastectomy! Anything less requires a substantial element of clinical judgement as to what more, if anything, should be done.

Does the result make sense?
It is axiomatic that if a test result is at odds with your clinical judgement, you should repeat the test. Have confidence in your judgement. All medical practitioners are entitled to rely on their own skills and those of their colleagues, but every medical practitioner has seen at least one instance when errors have occurred.