Editor's Choice


Medicine as ritual

The philosophy of the journal is the publication of the highest quality original scientific research in obstetrics, gynaecology and women's health; but we are also keen to publish clinical opinion. The Letters to the Editor column is inadequate for this purpose, owing to constraints of space; instead we prefer clinical opinion to appear as Commentaries. Thus a Commentary is a personal view by an author, designed to stimulate the imaginations of readers, and sometimes to make them angry. Because it is a personal view it need not be supported by an exhaustive list of references; in fact, the author may prefer to have his Commentary completely unencumbered by references.

In this issue we publish three Commentaries and two main articles on the contentious subject of clinical practice guidelines. Savas Menticoglou and Philip Hall (pages 485–491) dissect the evidence supporting the guidelines governing the practice of routine induction of labour at 41 weeks gestation. These are the guidelines of the Society of Obstetricians and Gynaecologists of Canada, but they are nearly identical to those published by the Royal College of Obstetricians and Gynaecologists. The guidelines recommend that induction of labour should be offered at 41 weeks gestation, but too often this is interpreted as a recommendation that induction of labour should beperformed at 41 weeks gestation. Few obstetricians realise that the risk of perinatal mortality between 41 and 42 weeks gestation is one in 1000 pregnancies, between 42 and 43 weeks gestation it is one in 500 pregnancies, and even after 43 weeks gestation it is one in 370 pregnancies – all very small risks. Yet about 20% of all women will reach 41 weeks gestation, suggesting that many hundreds of women will undergo induction of labour routinely at 41 weeks gestation in the hope of preventing one perinatal death. Few obstetricians realise that evidence supporting the apparent reduction in perinatal mortality with a policy of induction of labour is based on only eight perinatal deaths. This evidence is surely too tenuous to promote a policy of such an invasive intervention in so many women. Few obstetricians realise the increased risk of caesarean section caused by induction of labour. An analysis of randomised trials by intention-to-treat suggests a modest reduction in the rate of caesarean section with induction of labour; however an analysis by treatment actually given shows that caesarean section occurs nearly twice as frequently with induction of labour. Menticoglou and Hall give convincing reasons why this increased risk of caesarean section is due to the process of induction of labour, and not to underlying placental pathology resulting in fetal hypoxia. The authors recommend that these clinical practice guidelines be withdrawn.

A few years ago the Royal College of Obstetricians and Gynaecologists published a clinical practice guideline which recommended that the vacuum extractor should be the instrument of first choice for operative vaginal delivery. This guideline resulted in a progressive increase in the number of women delivered by vacuum extraction and a decrease in the number delivered by forceps. The use of Kielland's forceps declined rapidly, and in a few years Kielland's forceps may become extinct, a situation deplored by Karl Oláh (pages 492–494). The author describes the evidence from randomised trials which apparently show that vacuum extraction results in fewer maternal injuries than forceps. It is this finding alone which has been used to justify the primacy of vacuum extraction over forceps. Yet vacuum extraction fails to deliver the infant in one-quarter of instances, leaving the obstetrician with the formidable problem of “failed vacuum”. Furthermore the impression has been gained that the vacuum extractor requires less skill than forceps, and so can be performed by persons with less training. Does this account for the increasing number of case reports and case series of that dreaded complication in the newborn, subaponeurotic haemorrhage? Karl Oláh recommends that a trained obstetrician should have facility with both instruments, the vacuum extractor and Kielland's forceps, and use either instrument as the occasion demands.

In his Commentary Nick Johnson (pages 495–497) identifies some of the absurdities in guidelines. It is no wonder that obstetricians are confused, when on the one hand they are told from the evidence of randomised trials that electronic fetal heart rate monitoring should not be used in low-risk pregnancies, and on the other hand they are told from the reports of the Confidential Enquiries into Stillbirths and Deaths in Infancy that electronic fetal heart rate monitoring can prevent death and handicap, even in low-risk pregnancies. It is no wonder that gynaecologists are confused, when their surgical treatment of a woman with cancer of the ovary is judged to be substandard if the cervix is not removed, although the benefits of such radical surgery have not been proved. It is no wonder that perinatologists are confused, when guidelines state that in order to maintain competence an arbitrary minimum number of chorion villus samples is required, when the perinatologist's results may be good, although he performs fewer than the minimum number of chorion villus samples.

Nick Johnson is right to refer to the political influence of guidelines. The Royal College of Obstetricians and Gynaecologists has produced guidelines on induction of labour for post-term pregnancy and electronic fetal monitoring which purport to be evidence-based, and which have the imprimatur of an organisation of the UK Government, the National Institute of Clinical Excellence. This is a dangerous road for a professional body, for its alliance with a government organisation must surely undermine its independence. Furthermore, examination of the guidelines shows that they are not evidence-based. Thus the Clinical Outcomes Group of the NHS Executive has three grades of recommendation: Grade A, based on evidence of at least one randomised trial; Grade B, based on well-conducted observational studies; and Grade C, based on expert committee opinion. There were 32 recommendations in the guidelines for induction of labour; 12 (38%) were Grade A, 2 (6%) Grade B and 18 (56%) Grade C. Not all the Grade C recommendations were trivial. There were 31 recommendations concerning electronic fetal monitoring: 6 (19%) were Grade A, 10 (32%) were Grade B and 15 (48%) were Grade C. Not all the Grade C recommendations were trivial. The Confidential Enquiries into Stillbirths and Deaths in Infancy recommend that the interval between the decision to perform an emergency caesarean section and delivery of the infant should not exceed thirty minutes, a Grade C recommendation, and that maternity hospitals should be required to perform audits of the decision-to-delivery interval. Ian MacKenzie and Inez Cooke (pages 498–504) performed a prospective observational study of more than four hundred emergency caesarean sections, in which fewer than 40% were achieved in less than thirty minutes. In order to improve their performance W.H. Helmy and colleagues (pages 505–508) performed four audit cycles and found that they could achieve delivery in less than thirty minutes in only 71% of caesarean sections. If the decision-to-delivery interval in emergency caesarean section is important, which is doubtful, then a realistic decision-to-delivery must be based on evidence and not on an arbitrary standard set by a professional body.

But evidence-based medicine is predicated on randomised trials. Evidence-based medicine has its origin in obstetrics, for it was an obstetrician who was one of the principle founders of the evidence-based medicine movement. What a paradox therefore that it is in obstetrics that scientific evidence is so hard to come by. Yet we should not be surprised at this. It is in general medicine that evidence-based medicine comes into its own, for in general medicine the evidence is firmly entrenched in randomised trials. Treatments in general medicine involve drugs of uniform potency, diseases where the diagnosis is unequivocal, and outcomes which are incontrovertible. Thus the ISIS 2 trial investigated the use of streptokinase and aspirin in myocardial infarction. Each vial of streptokinase and each tablet of aspirin were of uniform potency; the diagnosis of myocardial infarction was unequivocal, and the outcome, mortality at 35 days, incontrovertible. The combination of streptokinase and aspirin reduced mortality by 40%. These results are convincing, such that when I have my first myocardial infarction I hope to be treated with streptokinase and aspirin.

The situation is different in obstetrics, for very few of our interventions involve drugs. Randomised trials of induction of labour in post-term pregnancy are confounded by variations in the definition of post-term pregnancy, in the state of the cervix at induction of labour, in the methods of induction of labour, and in the methods of fetal surveillance in women treated conservatively. Randomised trials of operative vaginal delivery are confounded by variation in the definition of significant maternal injury, poor training of the operators, failure to separate easy outlet deliveries from difficult mid-cavity deliveries, and ascertainment bias. The concept of a comparison between vacuum extraction and forceps is itself flawed, for the two instruments should complement each other, not compete with each other. Randomised trials of electronic fetal monitoring are confounded by being too small, by variation in the populations studied, by poor training of the persons interpreting the fetal heart rate traces, and by focusing on surrogate outcomes. But the concept of a randomised trial of electronic fetal monitoring is itself flawed, for electronic fetal monitoring is a screening test, and screening tests are tested by other means, formal measurement of reliability by an observer agreement study, and formal measurement of efficiency, by estimation of the sensitivity, specificity and positive and negative predictive values in the prediction of an important outcome, hypoxic-ischaemic encephalopathy. It is foolish to combine these trials into meta-analyses, and it is foolish to write guidelines on induction of labour for post-term pregnancy, operative vaginal delivery and electronic fetal monitoring, guidelines which purport to be evidence-based, but which on examination are not.

The culture of guidelines is not a scientific phenomenon, it is a social and political phenomenon of the latter part of the twentieth century and the first part of the twenty-first century. Social and political systems occur in cycles of birth, vigour and death, and so does the culture of guidelines, which was born, is now flourishing and will decline. Guidelines are a manifestation of the present desire for regulation and control which now affects the practice of medicine, as it does organisations such as the police and the teaching profession. From guidelines emerge targets, audit, league tables and disciplinary action, a sequence which has already affected hospitals in England.

The culture of guidelines is the antithesis to the ethos of medicine. The practice of medicine is founded upon divine principles of care, so that the physician complements the priest; the physician cures diseases of the body, the priest cures diseases of the soul. The physician undergoes a rigorous training, in physics, chemistry and biology; anatomy, physiology and biochemistry; pathology, microbiology and pharmacology; medicine and surgery; obstetrics, gynaecology and paediatrics; and much else. Is all this rigorous training to be denied? We are breeding a race of medical robots who are incapable of thinking problems from first principles, but who reach for the nearest guideline for advice. In 43 BC Marcus Tullius Cicero in his book On the Nature of God wrote, The philosophers of the Academy have been wise in withholding their consent from any proposition that has not been proved. There is nothing worse than a hasty judgment, and nothing could be more unworthy of the dignity and integrity of a philosopher than uncritically to adopt a false opinion or to maintain as certain some theory that has not been properly explored or understood. We can substitute physician for philosopher. The physicians of our medical schools will have been wise in withholding their consent from any clinical practice whose worth has not been proved. There is nothing worse than a hasty judgment, and nothing could be more unworthy of the dignity and integrity of a physician than uncritically to adopt a false opinion or to maintain as certain some clinical practice that has not been properly explored or understood.

What can the physician do to maintain the integrity of his clinical practice? If he is sent a clinical practice guideline he should obtain the original scientific research on which the evidence supporting the guideline is based. Many physicians are intimidated by guidelines, for they contain the mathematics of meta-analysis, odds ratios and relative risks, with which they may not be familiar. But mathematics is a disadvantage to the physician, for it distracts from his appraisal of the clinical aspects of the randomised trial, and whether these clinical aspects are relevant to his own practice. If from this analysis the physician cannot agree with the guideline, he should not protest by himself, for he will be isolated, and if legal consequences arise from his practice, his position will be difficult to defend. Instead he should discuss the guidelines with his colleagues, and if they agree, then as a group of physicians they can formulate their treatments according to the needs of their own clinical practice. This is a stronger position to defend. Then can the physician use all of his training and experience. He can reassure women that the risk to a pregnancy lasting longer than 41 weeks is tiny; he can interpret electronic fetal heart rate traces in the knowledge that this is a reliable and efficient test; and if he is called in the dead of night to treat delay in the second stage of labour due to a persistent occipito-posterior position, he can use his knowledge and training to perform Kielland's forceps safely, to deliver a healthy infant. He will then be satisfied with his practice, and in the early morning may find some time to rest, when the roofs and chimney pots are dark against the red sky, and songbirds salute the birth of a new day.