Editor's Choice

According to Webster's dictionary, the definition of risk is a factor, course or element involving uncertain danger. It is this definition that has caused so much trouble to obstetricians. A woman who sustains placental abruption, eclampsia or preterm labour is obviously ‘at risk’, but we tend not to classify her consciously as being at high risk, only of having an acute clinical problem, which requires immediate treatment. By ‘risk’, we mean the prediction of serious complications in childbirth, this prediction being derived from demographic and clinical factors in pregnancy. The concept of ‘risk’ has gripped modern medicine, and not just obstetrics. This concept is justified if the high risk groups identified on clinical and demographic grounds really are at high risk of a disease, and that men and women thought to be at low risk have substantially less chance of contracting the disease. If it is otherwise, the concept of risk is invalid, and it is unethical to concentrate resources on a group given the label ‘high risk’. Another difficulty with ‘risk’ is that if a large proportion of a population is thought to be at high risk of a disorder that occurs infrequently, many healthy persons will be subjected to unnecessary investigations, unnecessary interventions and unnecessary anxiety. Being a member of a high risk group may be dangerous to health, not because of the underlying disorder responsible for being included in a high risk group, but because of iatrogenic disease brought about by the ensuing investigations and interventions. These problems are explored by L. Høj et al. (pages 792–799), who tested the value of the risk card used in Guinea–Bissau to predict maternal mortality. In countries with limited facilities for health care, pregnant women who are thought to be at high risk of maternal mortality are identified; this identification is from demographic and clinical factors and is formulated in a risk score, often in the form of a risk card. The authors tested 4 of the 12 factors on the Guinea–Bissau risk card—age, parity, previous stillbirth and multiple pregnancy, in conjunction with 16 socio-economic and demographic variables. In a multiple regression analysis, only multiple pregnancy was associated with maternal mortality, while age, parity and previous stillbirth were not so associated. However, from the information in Table 6, multiple pregnancy is a poor predictor of maternal mortality, for its sensitivity is 9% and its positive predictive value is 2%. Although multiple pregnancy may be a useful predictor of perinatal mortality, it is only slightly useful in the prediction of maternal mortality. The multiple regression analysis showed that stillbirth in the present pregnancy or labour (often due to placental abruption or eclampsia) and distance to the maternity hospital were much stronger predictors of maternal mortality, but these are not clinical risk factors that can be identified or altered in pregnancy. The most important variable was living more than 25 km from a maternity hospital; two-thirds of all maternal deaths occurred in this group of women. The authors conclude that age, parity and previous stillbirth should not be criteria for delivery in a maternity hospital; that stillbirth in a present pregnancy is a sign of serious maternal disease and should indicate transfer to a maternity hospital; and above all, that maternal mortality will decline only with improved availability of obstetric care. The concept of risk and the formulation of clinical risk scores started about 40 years ago, and still we use risk to allocate women to different forms of care in pregnancy and labour. Thus, women are selected for domiciliary care on their being considered low risk. This has resulted in an unhealthy division between midwifery and obstetrics, midwives being responsible for the care of low risk women, obstetricians for the care of high risk women, with the construction of barriers between the two forms of care. But what if the concept of ‘risk’ is faulty? This was examined many years ago by Alexander and Keirse1, who found that formal risk scoring was too inefficient to be clinically useful, and that it was impossible to separate low risk women from women who were not at low risk. McCarthy et al.2 developed a risk score in one-quarter of a million women to predict perinatal mortality. This score was derived retrospectively and was not therefore used by obstetricians in their clinical care. The score was formulated 20 years ago, when fetal surveillance was much less rigorous and sophisticated than it is today. The sensitivity of the risk score was 38% and its positive predictive value 4%. This means that only one-third of perinatal deaths were predicted by a high risk score, two-thirds of perinatal deaths occurring in women judged to be at low risk. For every 100 instances where the risk score predicted perinatal death, 96 of these predictions were wrong, which would today result in unnecessary investigations, unnecessary interventions and unnecessary anxiety. The same outcome was found in all the clinical risk scores examined by Alexander and Keirse. Women who are thought to be low risk may be delivered in a ‘low risk’ area and undergo intermittent auscultation of the fetal heart in labour, a lesser form of fetal monitoring than cardiotocography. We do not know how efficient this separation is, based as it is on clinical risk factors derived antenatally, but it is likely to be inefficient, as in the study by McCarthy et al. It is likely that the majority of instances of birth asphyxia occur in low risk women, and that women given the label ‘high risk’ undergo unnecessary ultrasound scans, unnecessary inductions of labour and unnecessary caesarean sections. There is no intellectual justification for the separation on clinical and demographic grounds of women into high risk and low risk categories; and it is unethical to deny women a reliable and efficient test, cardiotocography, which can detect unpredicted birth asphyxia. As Høj et al. show, it is insufficient to determine high risk factors merely because of an increased odds ratio in a regression analysis. It is necessary to test the efficiency of these risk factors in the prediction of important clinical outcomes. If their sensitivity and positive predictive values are low, they cannot be regarded as efficient, and may be harmful if introduced into clinical practice. There is no evidence that identifying high risk groups in obstetrics on clinical grounds is useful, and plenty of evidence that it is harmful. We should abandon the concept of risk in pregnancy. A nightmare for obstetricians and paediatricians is impending preterm delivery at a very early gestational age at 24 or 25 weeks. We are uncertain of the benefits of corticosteroids, for the randomised trials included only a few pregnancies at such early gestational ages. Suppression of uterine contractions may be contraindicated if the pregnancy is complicated by antepartum haemorrhage or severe pre-eclampsia. We are not sure if caesarean section is preferred in breech presentation. If a caesarean section is performed, problems may arise because of poor formation of the lower segment of the uterus, requiring a classical caesarean section. The neonatal intensive care unit may have no available cots, resulting in transfer of the woman to another centre, compounding the difficulties in making decisions. Above all, there are uncertainties in advising the woman and her partner of the likely outcome for their infant if born at 24 or 25 weeks of gestation. This uncertainty has been largely resolved by Sidney Effer et al. (pages 740–745), who analysed the rates of survival of 860 newborn infants delivered in 13 tertiary centres in Canada between 1991 and 1996. This is the largest study of neonatal mortality at these early gestational ages. About half the infants born at 24 weeks and two-thirds born at 25 weeks survived. The study was large enough not only to calculate the survival according to completed weeks of gestation, but also by completed days of gestation. The study also suggests that prolonging pregnancy even by a few days is beneficial to the infant. Multivariate analyses showed that corticosteroids more than doubled the chances of survival, and that caesarean section for breech presentation increased survival by 76%. Effer et al. were unable to comment on neonatal morbidity in the long term owing to differences in follow up between the 13 centres. Nevertheless, the information on survival alone is useful in advising parents of the outcome that can be achieved in the short term with the best obstetric and neonatal care. The standard investigation in urinary incontinence in women is urodynamic testing by subtraction cystometry. Urodynamic investigations concentrate attention on the primary mechanism of continence, situated at the level of the neck of the bladder, and so far, there has been little investigation on the secondary mechanism of continence, situated at the level of the urogenital diaphragm. Roberto Pregazzi et al. (pages 821–827) used perineal ultrasound to measure the urethral angle in women with genuine stress incontinence, and found that it was significantly less than in a control group. The strength of this study is that it tested fully the intrinsic properties of this test, its reliability and its efficiency. Thus, two observers independently measured the urethral angle and found from the limits of agreement that its reliability was good. From receiver–operator characteristic curves, they established cutoff points for urethral angle and bladder neck mobility. They then used these cutoff points to measure the efficiency of the two tests in the detection of genuine stress incontinence. The likelihood ratio of an abnormal urethral angle was 12 and for a normal urethral angle it was 0.04. This is a good test. It is significantly better than bladder neck mobility. The authors conclude that urethral angulation is important in maintaining urinary continence in women, and suggest that measurement of the urethral angle may be helpful in investigating and treating women with stress incontinence. They could have gone further. For years, the standard surgical treatment for genuine stress incontinence has been colposuspension, an operation on the neck of the bladder, but this procedure now is complemented by the tension-free vaginal tape, an operation at the level of the urogenital diaphragm. Our concept of the mechanisms of continence is changing, and with it the site of the surgical procedure, for the tension-free vaginal tape is challenging the pre-eminence of colposuspension. The importance of the paper by Pregazzi et al. is that it gives us a pathophysiological explanation of the success of the tension-free vaginal tape, and suggests the possibility of selection of women for colposuspension or the tension-free vaginal tape, depending on the site of the main abnormality detected on perineal ultrasound, at the bladder neck or at the mid-urethra.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.