Randomised clinical trial on the effect of the Dutch obstetric peer review system.

The project "Obstetric Peer Review Interventions" (Verloskundige Onderlinge Kwaliteitsspiegeling Interventies, VOKSINT) was set-up in The Netherlands in 1994. It provided annual comparison data (quality ranking, league tables) for secondary care obstetric departments adjusted for population differences, based on the data registered in the Perinatal Database of The Netherlands (Landelijke Verloskunde Registratie, LVR). The aim of the so-called VOKS reports was to influence obstetricians' interventions in such a way that they led to a more homogeneous policy. To assess this influence, a trial was set-up, with departments randomly assigned to be or not to be informed about the VOKS results. Obstetric intervention rates and the morbidity of newborns including neonatal neurological examinations (NNEs) were assessed. Obstetric intervention rates were similar in the report group and the control group. Practice in the report group became more homogeneous (adjusted for population differences) than in the control departments, but this was only statistically significant for term caesarean section.

The reliability of perinatal and neonatal mortality rates: differential under-reporting in linked professional registers vs. Dutch civil registers.

Official Dutch perinatal mortality rates are based on birth and death certificates. These civil registration data are not detailed enough for international comparisons or extensive epidemiological research. In this study, we linked and extrapolated three national, incomplete, professional registers from midwives, obstetricians and paediatricians, containing detailed perinatal information. This linkage and extrapolation resulted in one detailed professional database which is representative of all Dutch births and from which gestational age-specific perinatal mortality rates could be calculated. The reliability of these calculated mortality rates was established by comparing them with the rates derived from the national civil registers. The professional database reported more perinatal deaths and fewer late neonatal deaths than the civil registers. The under-reporting in the civil registers amounted to 1.2 fewer perinatal deaths per 1000 births and was most apparent in immature newborns. We concluded that under-reporting of perinatal and neonatal deaths depends on the data source used. Mortality rates for the purpose of national and international comparison should, therefore, be defined with caution. This study also demonstrated that combining different incomplete professional registers can result in a more reliable database containing detailed perinatal information. Such databases can be used as the basis for extensive perinatal epidemiological research.

Large differences in obstetrical intervention rates among Dutch hospitals, even after adjustment for population differences.

Within the framework of the 'Obstetric Peer Review' project (Verloskundige Onderlinge Kwaliteitsspiegeling, VOKS) statistical models have been developed to predict department specific intervention rates, based on the distribution of risk factors in each department. Subsequently the difference between the expected number of interventions (labour inductions, caesarean sections and vaginal operative deliveries) and the actual numbers were calculated for each year and subpopulation defined by the level of prematurity. Data used were available from the Perinatal Database of the Netherlands (Landelijke Verloskunde Registratie, LVR) concerning the years 1988-1992. Even after adjusting for many clinical risk factors the interdepartmental differences of intervention rates are both clinically relevant and statistically significant. It was also obvious, especially for the term population, that departments showed a consistent pattern with respect to the difference between their expected and observed intervention rates over these 5 years.

The Dutch obstetric intervention study--variations in practice patterns.

OBJECTIVE: To compare obstetric intervention rates between Dutch hospitals. METHODS: A total of 28,934 hospital births under secondary care (specialist care for medium-/high-risk pregnancies) in 1990 were analyzed in a stratified, random sample of Dutch hospitals based on the records of the Dutch Netherlands perinatal database. Comparisons were made of the intervention rates between hospitals. RESULTS: The intervention rates of the various hospitals differed widely. The most striking difference was in the cesarean section (CS) rate for non-vertex first twins, with a range of 0-100% and a mean rate of 47.6%. On average a CS for a term breech was performed in 30.8% of cases and sedation or analgesics were administered in 16.2% of cases. The mean rate of episiotomy for a term breech was 71.5%, the lowest rate being 18.8%. CONCLUSION: Our results show relatively low intervention rates with considerable interhospital variation.

Trends in caesarean section rates among high- and medium-risk pregnancies in The Netherlands 1983-1992.

During the last two decades, the rates of operative deliveries have been rising constantly in all industrialized countries including the Netherlands. Within the framework of the project 'Obstetric Peer Review' (Verloskundige Onderlinge Kwaliteitsspiegeling), the trends in the caesarean section rates were investigated, using the data of the Perinatal Database of the Netherlands (LVR), but only among the high- and medium-risk pregnancies. To that end homogeneous high-risk subgroups, with respect to pregnancy- or delivery-related complications, were defined in various ways and caesarean section rates were calculated for these groups. Irrespective of the definition of such a group (e.g. multiple pregnancy or breech presentation), an increase of the caesarean section rates evidently emerges. The rates of planned caesarean section appear to increase more than the rates of emergency caesarean section. The increase for the multiparae is greater than for the primiparae. It is striking that the largest increase was found within the medium-risk group (namely; singletons, at term, vertex presentation, normal birthweight, mothers aged 20-35 years and with a normal diastolic blood pressure). Although no evident pathology can be found in the data, these women do not belong to the low-risk group, because they were referred to an obstetrician. For the multiparae in this group, the risk of a planned caesarean section yearly increased by a factor 1.09 (i.e. 9%).

How to compare and report department specific mortality rates for peer review using the Perinatal Database of The Netherlands.

Since characteristics of patient populations of obstetric departments vary substantially with respect to the pathology presented, (perinatal) mortality rates as such can not be used for a fair peer review without adjustment for those differences. Using the Perinatal Database of The Netherlands (LVR), data on approximately 80,000 newborns annually from 1985 to 1991 inclusive were used in statistical models to predict the perinatal mortality risks of four subpopulations of different gestational age in about 125 obstetric departments. As predictors for perinatal mortality we used only those risk factors which were judged to reflect the 'pathology' of the patient; risk factors associated with or resulting from hospital care and/or policy are (by definition) excluded. The statistical technique of logistic regression was used. The expected and observed mortality within each of these four subpopulations (for 1985-1991) was sent anonymously to each department as a specially designed graphical overview. This method of (anonymous) peer review was met favourably. In peer review the difference between observed and expected mortality should be used, rather than mortality per se, to adjust for differences in basic risk in the intake populations on which the departmental policy has no influence.

Comparing mortality and morbidity in hospitals: theory and practice of quality assessment in peer review.

The incidence of mortality in a specific hospital depends on many risk factors. These risk factors may be divided roughly into two categories. The intake category, consists of those risk factors for which the hospital has hardly any influence upon their incidence; and the care category being those for which the incidence depends partly or completely on the treatment policy of the hospital. A hospital with a high incidence of risk factors in the intake category will have a higher mortality rate than a hospital with a low incidence, even if their care is exactly the same (i.e., if they treat their infants equally well). Therefore, a fair comparison between one hospital and a reference cohort, or among several hospitals (using a national registry) should adjust e. g. correct for those risk factors belonging to the intake category. A practical method is proposed, based on logistic regression, to effectuate such a "fair" judgment. The regression technique enables to compare "observed" and "expected" rates in a specific hospital and to test whether a difference between these rates is statistically significant. Both clinical and statistical aspects of the method are discussed, as well as the actual implementation of an automated annual reporting system. The method has been implemented in the Netherlands as an annual peer review and quality assessment system in obstetric care.

Differences in obstetrical intervention rates between Dutch hospitals.

Within the framework of the project 'Obstetric Peer Review' (Verloskundige Onderlinge Kwaliteitsspiegeling, VOKS) differences between Dutch hospitals concerning various obstetrical interventions were investigated. Using data of the Perinatal Database of the Netherlands from hospitals with at least 2000 newborns in the 5-year period 1987-1991, remarkable differences in frequencies of labour induction, caesarean section and vaginal operative deliveries can be shown, even when these interventions were considered within homogeneous subgroups with respect to pregnancy- or delivery-related complications. The incidence of caesarean section (and labour induction and vaginal operative delivery) appeared to depend more on the specific hospital policy than might be explained by populations differences alone. These intervention differences between hospitals clearly demonstrated the need to carry out a multivariate analysis to compare hospitals in a relevant and unbiased way, while adjusting for these population differences. This will estimate the intervention risks in a way which allows a fair comparison of hospital intervention policies.

Obstetric characteristics profiles as quality assessment of obstetric care.

In The Netherlands only about 50% of all pregnancies are defined as high risk pregnancies and consequently come into the domain of care of an obstetrician. In order to qualify as a high risk pregnancy, the pregnancy must satisfy certain criteria contained in an officially approved list of indications. Due to varying perceptions of these selection criteria by the selectors, larger differences in the treated population, obstetric interventions and results, respectively, were expected than in other countries. Within the framework of the project 'Obstetric Peer Review' (Verloskundige Onderlinge Kwaliteitsspiegeling, VOKS) the type and the magnitude of the differences of the values of the obstetric characteristics of population, interventions and results, respectively have been investigated. For the high risk pregnancies the data contained in the Perinatal Database of The Netherlands were used for the years 1983-1987, both inclusive. Because the various obstetric departments in The Netherlands had been unaware of these differences, consequently they had to become conscious about these differences and be informed about their own position relative to the others. For an instant comparison a single sheet graphical method has been designed, affording the assessment of one's own obstetric characteristics relative to those of other departments for the above mentioned 5 years for each of the 25 items considered. All departments, which participated in the Perinatal Database of The Netherlands during the years 1983-1987, received their own profiles and were invited to reply. Large differences were observed for the various characteristics considered.(ABSTRACT TRUNCATED AT 250 WORDS)

Estimate of perinatal mortality risk.

Perinatal mortality is usually calculated according to the World Health Organisation as stillbirth and first week mortality at a specified week of gestation divided by all births at that same gestational week. This is not a meaningful indicator of the risk of future perinatal death for a living fetus. We have developed an approach to estimate the prospective risk of perinatal mortality. Data were derived from the Perinatal Database of the Netherlands. We calculated the prospective risk of perinatal mortality by dividing all future perinatal deaths from a certain week of gestation by all fetuses that remained undelivered. Using this statistic there is a decline in risk from 16 to 39 gestational weeks and an increase from 39 weeks onwards.

[Underreporting of first-week mortality in premature childbirth in the National Obstetrics Register]. / Onderrapportage van de eersteweeksterfte bij vroeggeboorte in de Landelijke Verloskunde Registratie.

The reliability of the perinatal mortality as recorded in the Perinatal Database of the Netherlands (LVR), was studied over 1983. For this year is was possible to make a comparison between preterm infants in the LVR and the same infants recorded by paediatricians in the database of the Project on Preterm and Small for Gestational Age Infants in the Netherlands 1983 (POPS). The comparison between the recorded mortality of the same infants in these two anonymous databases was realised by a simple matching procedure. For the premature infants a 30% lower first week mortality was found in the LVR than in the POPS-group, which means that 10% of the total perinatal mortality (stillbirths and first-week deaths together) has not been registered in the LVR for this group, due to lack of adaptation of data after later deaths. The mortality in the LVR should be interpreted with caution while it is not yet possible to match the obstetric data to the data of the National Neonatology Registration.

Perinatal mortality in breech presentations as compared to vertex presentations in singleton pregnancies: an analysis based upon 57819 computer-registered pregnancies in The Netherlands.

In 1982, nationwide registration of obstetric data was instituted in The Netherlands with about 70% of all Dutch hospitals participating. The resultant data from 57819 singleton pregnancies in vertex or breech presentation at delivery was studied. The vertex and breech groups were compared. The proportion of breech presentations relative to vertex presentations was greater in low gestational age infants and those of low birthweight. After correction for gestational age and birthweight, the perinatal mortality was higher in the breech groups than in the vertex groups. Congenital malformations occurred more frequently in the breech group but, even after exclusion of infants with congenital malformations, perinatal mortality remained higher in the breech group at any gestational age. Caesarean section was more frequently performed in the breech group than in the vertex group but did not appear to improve the outcome of breech presentation. It is possible that breech presentation is not coincidental but is a consequence of poor fetal quality, in which case medical intervention is unlikely to reduce the perinatal mortality associated with breech presentation to the level associated with vertex presentation.