- Email: [email protected]
Differences in perinatal mortality and suboptimal care between 10 European regions: results of an international audit
BJOG: an International Journal of Obstetrics and Gynaecology February 2003, Vol. 110, pp. 97 –105
Differences in perinatal mortality and suboptimal care between 10 European regions: results of an international auditB Jan H. Richardusa,*, Wilco C. Graafmansa, S. Pauline Verloove-Vanhorickb, Johan P. Mackenbacha, The EuroNatal International Audit Panel, The EuroNatal Working Group A list of the members of the EuroNatal International Audit Panel and the EuroNatal Working Group may be found on page 104 Objective A European concerted action (the EuroNatal study) investigated the background of differences in perinatal mortality between countries of Europe. The study aimed to determine the contribution of differences in quality of care, by looking at differences in the presence of suboptimal factors in individual cases of perinatal death. Design Retrospective audit study. Setting Regions of 10 European countries. Population 1619 cases of perinatal death. Methods Perinatal deaths between 1993 and 1998 in regions of 10 European countries were identified. Reviewed were singleton fetal deaths (28 or more weeks of gestational age), intrapartum deaths (28 or more weeks) and neonatal deaths (34 or more weeks). Deaths with (major) congenital anomalies were excluded. Cases were blinded for region and an international audit panel reviewed them using explicit audit criteria. Main outcome measures Presence of suboptimal factors. Results The audit covered 1619 cases of perinatal death, representing 90% of eligible cases in the regions. Consensus was reached on 1543 (95%) cases. In 715 (46%) of these cases, suboptimal factors, which possibly or probably had contributed to the fatal outcome, were identified. The percentage of cases with such suboptimal care factors was significantly lower in the Finnish and Swedish regions compared with the remaining regions of Spain, the Netherlands, Scotland, Belgium, Denmark, Norway, Greece and England. Failure to detect severe IUGR (10% of all cases) and smoking in combination with severe IUGR and/or placental abruption (12%) was the most frequent suboptimal factor. There was a positive association between the proportion of cases with suboptimal factors and the overall perinatal mortality rate in the regions. Conclusions The findings of this international audit suggest that differences exist between the regions of the 10 European countries in the quality of antenatal, intrapartum and neonatal care, and that these differences contribute to the explanation of differences in perinatal mortality between these countries. The background to these differences in quality of care needs further investigation.
INTRODUCTION Reviewing the circumstances of adverse health outcomes has a long tradition, and maternal and perinatal deaths, in particular, have received extensive interest1. Several countries have national programs for this purpose, such as the
a
Department of Public Health, Erasmus MC, Erasmus Medical Centre, Rotterdam, The Netherlands b TNO Prevention and Health, Leiden, The Netherlands * Correspondence: Dr J. H. Richardus, Department of Public Health, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands. B The EuroNatal study was funded by the BIOMED 2 program of the European Commission. D RCOG 2003 BJOG: an International Journal of Obstetrics and Gynaecology PII: S 1 4 7 0 - 0 3 2 8 ( 0 2 ) 0 2 0 5 3 - 0
National Fetal and Infant Mortality Review Program (NFIMR) in the US and the Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) in the UK. Audit methods have been developed, both at the national and international levels with the aim of improving perinatal care. In the early 1990s, the European Association of Perinatal Medicine attempted to achieve standardisation across Europe in the methodology used for perinatal audit by drawing up guidelines for a common approach to perinatal audit2. This report describes the results of the first large scale international audit of perinatal mortality. Previous international audit have compared only pairs of countries, such as Norway and Sweden3,4, and Denmark and Sweden5. This study offers a unique opportunity to compare the quality of care provided within different health care systems and its possible association with perinatal mortality rates. www.bjog-elsevier.com
98
J.H. RICHARDUS ET AL.
Differences in perinatal mortality rates between countries may possibly be associated with differences in the quality of antenatal and perinatal care, in addition to other factors6. This issue was addressed directly by investigating perinatal deaths in various European countries in relation to the quality of care provided. Results from the different audit regions are compared in order to look for variations in the presence of suboptimal care factors. This international audit was part of a European collaboration, the EuroNatal study, which investigated the background to differences in perinatal mortality in various European countries7.
METHODS Regions in 10 European countries were identified aiming to represent the country as a whole with respect to urban and rural characteristics, socio-economic characteristics and ethnicity, and level of perinatal care. Variations existed in the size of the area and in the period for data collection, but lay within the years 1993 –1998. Greece conducted a cross sectional National Perinatal Survey, based on all consecutive births and perinatal deaths for eight weeks of 1998. The cases selected for review by the audit panel were singleton fetal deaths at 28 or more weeks of gestational age, with or without growth restriction, all intrapartum deaths at 28 or more weeks of gestational age and all neonatal deaths at 34 or more weeks of gestational age. In each case, major congenital anomalies were excluded. This selection corresponds to the categories II (single growth retarded fetus, born at 28 or more completed weeks of gestational age), III (single fetus without growth restriction, born at 28 or more completed weeks of gestational age), VI (all intrapartum deaths after admission at 28 or more completed weeks of gestational age) and X and XI combined (neonatal deaths among live births at 34 or more weeks of gestational age regardless of Apgar score) of the Nordic –Baltic (N– B) Perinatal Death Classification5. The N – B classification is based on the assumption that subgroups of perinatal deaths may be more or less ‘potentially avoidable’. The essential parameters are primarily chosen because they are thought to be of relevance for quality improvement within antenatal care, obstetrics and early neonatal care. The structure presents the results in such a way that will help focus on possible avoidable cases to a greater extent than classifications used previously5. In the EuroNatal study, both antenatal categories II and III were included to give the flexibility to redefine IUGR in the future. Category VI includes intrapartum deaths after admission, with admission defined as the first patient – professional contact for labour. The definition of neonatal deaths included all deaths of babies with a gestational age of 34 completed weeks or more (N – B categories X and XI combined). Babies with major congenital anomalies are a separate category (I) in the N – B classification and were not included in the EuroNatal audit.
Because national growth charts were not available for all countries, the standard growth chart used by Denmark and Sweden was adopted in this study, presenting weight for gestational age for an unselected population including infants still in utero, with weights estimated on the basis of ultrasound measurement8. Two standard deviations below the average was defined as the upper limit for severe growth restriction. Explicit criteria based on international consensus and evidence of effectiveness were incorporated into a questionnaire. Additional information was provided on social or other relevant circumstances, and on the availability of records and documents pertaining to the cases. For each element of care in the selected categories of the N – B classification, explicit criteria for optimal clinical care were defined. These criteria, nearly 50 in all, fell into the following three groups: 1. Criteria based on clear evidence available from controlled trials. The source of information was The Cochrane Pregnancy and Childbirth database of 1995 (Issue 2). 2. Criteria for which the evidence is not as firmly established as the above criteria. These also originated from The Cochrane Pregnancy and Childbirth database of 1995 (Issue 2). 3. Criteria formulated by the EuroNatal International Audit Panel based on consensus about appropriate perinatal practice, and defined primarily for standardisation of the EuroNatal audit. The Cochrane Pregnancy and Childbirth database of 1995 was used because the criteria were applicable to the whole period of the audit (1993 – 1998). The international audit panel consisted of 12 members from all participating countries except Finland. Panel members were nominated by representatives from the various countries in the EuroNatal Working Group and were recognised experts in their field. Four subpanels with three members were formed. The panel consisted of obstetricians, neonatologists, perinatal epidemiologists and a midwife. To define suboptimal care, a grading system was adapted from the Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) in the UK9. Each case was blinded for country and distributed randomly to one of the four subpanels. Each subpanel member received a copy of the questionnaire, the narrative summary and a sheet that identified lack of compliance with the explicit audit criteria. After reviewing this, the panel member decided on whether there were instances of suboptimal care. Any such instances were listed on a standard form. The panel member marked the timing as antenatal, intrapartum or neonatal and whether it was ‘maternal/social’, ‘infrastructure/service organisation’ or ‘professional care delivery’. They also marked the contribution of each suboptimal D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
Table 1. Number and percentage [n (%)] of cases entered into the audit, with the total number of cases graded and the number of cases graded with 2 and 3, in the selected N – B categories in regions of 10 European countries. Audit region
0
1
207 (100)
19 (9.2)
0
188 (90.8)
46 (24.5)
46 (24.5)
269 (100)
6 (2.2)
3 (1.1)
260 (96.7)
54 (20.8)
221 (100)
5 (2.3)
1 (0.5)
215 (97.3)
175 (100)
10 (5.7)
2 (1.1)
110 (100)
4 (3.6)
165 (100) 142 89 106 135
(100) (100) (100) (100)
1619 (100)
2
n (%) with 95% CI of percentage 2 or 3 graded 2 or 3 3 96 (51.1)
43.9 – 58.2
73 (28.1) 111 (42.7) 22 (8.5)
133 (51.2)
45.1 – 57.2
37 (17.2)
63 (29.3)
91 (42.3) 24 (11.2)
115 (53.5)
46.8 – 60.1
163 (93.1)
74 (45.4)
37 (22.7)
41 (25.2) 11 (6.7)
52 (31.9)
25.1 – 39.4
1 (0.9)
105 (95.5)
20 (19.0)
31 (29.5)
51 (48.6)
54 (51.4)
41.9 – 60.9
5 (3.0)
3 (1.8)
157 (95.2)
40 (25.5)
41 (26.1)
54 (34.4) 22 (14.0)
76 (48.4)
40.7 – 56.2
2 4 4 6
(1.4) (4.5) (3.8) (4.4)
1 (0.7) 0 0 0
139 85 102 129
38 10 18 32
46 32 39 51
45 40 37 42
55 43 45 46
(39.6) (50.6) (44.1) (35.7)
31.7 – 47.9 40.0 – 61.1 34.7 – 53.9 27.8 – 44.2
65 (4.0)
11 (0.7)
715 (46.3)
43.9 – 48.8
(97.9) (95.5) (96.2) (95.6)
1543 (95.3)
(27.3) (11.8) (17.6) (24.8)
(33.1) (37.6) (38.2) (39.5)
72 (38.3) 24 (12.8)
3 (2.9)
(32.4) 10 (7.2) (47.1) 3 (3.5) (36.3) 8 (7.8) (32.6) 4 (3.1)
369 (23.9) 459 (29.7) 584 (37.8) 131 (8.5)
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE
Flanders (Belgium) Whole country (Denmark) Seven former NHS regions (England, UK) 13/22 Hospital districts (Finland) Whole country (Greece) Part South-Holland province (The Netherlands) Four counties (Norway) Grampian (Scotland, UK) Valencia province (Spain) Southern Health Care Region (Sweden) Total
Grade (% of total graded)
Total entered Not auditable Dissensus Total graded into audit (% of total audited) (% of total audited) (% of total audited)
99
100
J.H. RICHARDUS ET AL.
Fig. 1. Percentage of deaths graded 2 or 3 in regions of 10 European countries, with 95% confidence intervals. Dark grey bars: all categories together; light grey bars: category II (single growth restricted fetal deaths) excluded.
Consensus on a final subpanel grade was reached in an audit procedure that began with a sequence of assessments by correspondence, followed by plenary meetings of the subpanels. When consensus was not reached in the assessment rounds by correspondence, the case was referred for discussion in a plenary meeting. If more than one of the three subpanel members could not assign a grade to a case because of insufficient information, the case was considered ‘not-auditable’. The consensus grade was taken from the three grades determined by the subpanel. Grade 0 was when the grades were X00, X01, 000, 001; grade 1: X11,
factor to the fatal outcome as ‘unlikely’, ‘possibly’ or ‘likely’. Then they allocated a final grade from the following scale: 0: No suboptimal factors have been identified 1: Suboptimal factor(s) have been identified, but are unlikely to have contributed to the fatal outcome 2: Suboptimal factor(s) have been identified, and might have contributed to the fatal outcome 3: Suboptimal factor(s) have been identified, and are likely to have contributed to the fatal outcome
Table 2. Total number of cases in the selected categories with number and percentage of cases graded 2 or 3 [n (%)] in regions of 10 European countries. Audit region
Flanders (Belgium) Whole country (Denmark) Seven former NHS regions (England, UK) 13/22 Hospital districts (Finland) Whole country (Greece) Part South-Holland province (The Netherlands) Four counties (Norway) Grampian (Scotland, UK) Valencia province (Spain) Southern Health Care Region (Sweden) Total
Category II (single growth restricted; 28 weeks of gestational age)
Category III (single not growth restricted; 28 weeks of gestational age)
Category VI (all intrapartum deaths; 28 weeks of gestational age)
Category X/XI (neonatal deaths; 34 weeks of gestational age)
Overall graded
Total graded
Grade 2 or 3 (% of total)
Total graded
Grade 2 or 3 (% of total)
Total graded
Grade 2 or 3 (% of total)
Total graded
Grade 2 or 3 (% of total)
52
46 (88.5)
87
21 (24.1)
22
13 (59.1)
27
16 (59.3)
188
57
49 (86.0)
146
40 (27.4)
28
20 (71.4)
29
24 (82.8)
260
35
30 (85.7)
124
41 (33.1)
16
12 (75.0)
40
32 (80.0)
215
23
17 (70.8)
107
22 (20.6)
9
5 (55.6)
24
8 (33.3)
163
21 40
21 (100) 27 (67.5)
54 71
11 (20.4) 17 (23.9)
7 26
5 (71.4) 15 (57.7)
23 20
17 (73.9) 17 (85.0)
105 157
24 19 31 22
21 18 23 18
(87.5) (94.7) (74.2) (81.8)
95 52 52 87
24 15 10 17
270 (83.3)
875
324
(25.3) (28.8) (19.2) (19.5)
2 5 8 6
218 (24.9)
129
1 5 4 3
(50.0) (100) (50.0) (50.0)
18 9 11 14
83 (64.3)
215
9 5 8 8
(50.0) (55.6) (72.7) (57.1)
139 85 102 129
144 (67.0)
1543
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE
101
Table 3. Frequency of the most common suboptimal professional care factor (failure to detect severe IUGR) and maternal/social factor (smoking) graded with 2 and 3 in all categories in regions of 10 European countries. Values are given as n (%). Total graded
Audit region
Flanders (Belgium) Whole country (Denmark) Seven former NHS regions (England, UK) 13/22 Hospital districts (Finland) Whole country (Greece) Part South-Holland province (The Netherlands) Four counties (Norway) Grampian (Scotland, UK) Valencia province (Spain) Southern Health Care Region (Sweden) Total
Failure to detect severe IUGR
188 260 215 163 105 157 139 85 102 129 1543
011, 111; grade 2: X22, X23, 222, 223; grade 3: 233, 333. The combination X33 was not used. ‘X’ means that one subpanel member considered that there was no sufficient information to assign a grade. For each case with a final grade of 2 or 3, the suboptimal factors were identified which had ‘possibly’ or ‘likely’ contributed to the fatal outcome. Only factors identified as such by at least two out of three subpanel members were taken into consideration. As a result, one or more suboptimal factors were identified for each case and categorised as explained above. Mean proportions of cases graded 2 or 3 were calculated for each region and for all regions taken together, with 95% confidence intervals. Differences between countries in the
27 25 18 9 18 19 15 7 12 8 158
(14.4) (9.6) (8.4) (5.5) (17.1) (12.1) (10.8) (8.2) (11.8) (6.2) (10.2)
Smoking Smoking and IUGR
Smoking and abruption
13 19 12 4 7 11 10 5 5 6 92
4 16 12 5 2 3 14 8 2 3 69
(6.9) (7.3) (5.6) (2.5) (6.7) (7.0) (7.2) (5.9) (4.9) (4.7) (6.0)
(2.1) (6.2) (5.6) (3.1) (1.9) (1.9) (10.1) (9.4) (2.0) (2.3) (4.5)
Smoking, abruption and IUGR 1 7 2 1 0 0 1 5 1 1 19
18 42 26 10 9 14 25 18 8 10 180
(9.6) (16.2) (12.1) (6.1) (8.6) (8.9) (18.0) (21.2) (7.8) (7.6) (11.7)
proportion of cases graded 2 or 3 were calculated by simple logistic regression (parameterisation: deviation). The relationship between the percentage of deaths graded 2 or 3 in the selected categories in the various regions (Table 1, column 10), and the overall perinatal mortality rates per 1000 in those regions were analysed by linear regression using SPSS for Windows, release 9.0.0. This is shown in Fig. 2.
RESULTS A total of 1619 cases of perinatal death in the selected N –B categories from regions in 10 European countries
Fig. 2. Perinatal mortality rate and % deaths graded 2 or 3 in regions of 10 European countries.
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
(0.5) (2.7) (0.9) (0.6) (0) (0) (0.7) (5.9) (1.0) (0.8) (1.2)
Total smoking-related factors
102
J.H. RICHARDUS ET AL.
were entered into the audit. These represented 90% of all eligible cases in the selected categories. Of these 1619 cases, 65 (4.0%) had insufficient information for the audit panel to grade, and in 11 cases (0.7%) the audit panel could not reach consensus about the grading (Table 1). The remaining 1543 cases (95.3%) received a consensus grade. In total, 369 (23.9%) were graded as 0, 459 (29.7%) as 1, 584 (37.8%) as 2 and 131 (8.5%) as 3. Table 1 shows the distribution of grades across the regions that were audited, together with the number of cases that were unauditable or in which no consensus was reached. A total of 715 cases (46.3%) were graded with 2 or 3, indicating the presence of suboptimal factors that possibly or likely contributed to the fatal outcome. These will be referred to as ‘grade 2/3 cases’. Also shown in Table 1 is the percentage of grade 2 or 3 per region, with 95% confidence intervals. The percentage of cases with such suboptimal care factors was significantly lower in the Finnish and Swedish regions when compared with the remaining regions of Spain, the Netherlands, Scotland, Belgium, Denmark, Norway, Greece and England (see also Fig. 1). Figure 1 shows the percentage of deaths with grade 2/3 in all the selected categories together for each region (dark bars). Also shown as light bars is the percentage of deaths with grade 2/3 in the selected categories with the exception of category II (single growth restricted fetal deaths of 28 or more weeks of gestational age). A similar pattern is seen as with all categories together, with the percentage of cases with such suboptimal care factors being significantly lower in the Finnish and Swedish regions when compared with the other regions. The percentage of grade 2/3 antenatal deaths with growth restriction in all regions taken together is 83.3% (range: 67.5– 100%); 24.9% (range: 19.2– 28.8%) in antenatal deaths without growth restriction, 64.3% (range: 50.0 – 100%) in intrapartum deaths and 67.0% (range: 33.3– 85.0%) in neonatal deaths (Table 2). The number of cases in each category in each region is too low to indicate whether differences between regions are statistically significant. In general, it is observed that Finland and Sweden, in which the overall percentages of grade 2/3 cases are below the average of 46.3% for all European regions taken together, also show below average percentages in each of the four graded categories. In the professional care delivery category, failure to detect severe IUGR appeared as the most prominent factor (10,2%). Table 3 shows the differences between the regions in the percentage of this suboptimal factor (range: 5.5– 17.1%). The 95% confidence interval limit for Finland (2.6 –10.2%) is just below the overall average of 10.2%. Sweden also has a low percentage (6.2%), but this is not statistically significant. The other regions are within the range of the overall average percentage. The most prominent suboptimal maternal/social factor was related to maternal smoking (11.7%). Table 3 also shows the frequency of this suboptimal factor. Overall, the percentage
found in each region varied from 6.1% (95% CI: 3.0 – 11.0%) in the Finnish regions, which is statistically significant below the overall average, to 16.2% (95% CI: 11.7– 20.6%) in Denmark, 18.0% (95% CI: 11.6– 24.4%) in the Norwegian regions and 21.2% (95% CI: 13.1– 31.4%) in the Scottish region, which are all statistically significant above the overall average. The other regions are within the overall average percentage. The EuroNatal study included selected subgroups representing approximately 60% of all deaths. Figure 2 shows a statistically significant ( P ¼ 0.04) association between the percentage of cases in the selected categories graded 2/3 in a region, and the overall perinatal mortality rate in that region as recorded in this study. In particular, the regions in Finland and Sweden with the lowest percentages of grade 2/3 cases also have the lowest overall perinatal mortality rates.
DISCUSSION Clinical audits that identify suboptimal factors influencing death, and especially deficiencies in medical care, have become common in many fields of medicine. Audit is defined as an investigation into whether an activity meets explicit standards defined in an auditing document10. The primary objective of clinical audit is to establish whether the processes of care provided to a single patient or population of patients represent those therapies and procedures that are thought or known to be associated with the achievement of a good outcome. Thus, an audit does not establish what optimal care is, but investigates whether optimal care is actually provided. Other scientific methods, such as randomised controlled trials, can be used to determine the best possible therapy or procedure. In an audit, the effectiveness of certain therapies and procedures, based either on clinical evidence or on the experience of a group of professionals, is taken for granted. This leads to explicit standards by which the care process is audited. The definition of explicit criteria and the use of external panels are considered essential for comparison of results of an audit11,12. The EuroNatal audit is unique in that it is the first attempt at an international audit of perinatal deaths on a scale that includes regions from 10 different countries with an average of 162 cases per region. The EuroNatal study therefore offers a unique opportunity to compare aspects of the quality of care provided within different health care systems in cases where the outcome was death. To ensure valid comparison between regions, the audit had to satisfy several conditions. These have been discussed separately13. It was concluded that the inevitable differences per region in level of case identification and recruitment and quality of information were not a serious threat for the validity of the audit. A high level of case identification was achieved in Flanders, Denmark, England, D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE
Finland, Norway, Scotland and Sweden because of the possibility for linkage between medical and civil registrations and other routine case ascertainment procedures. In Greece, the Netherlands and Spain, special studies were necessary to identify all relevant cases of perinatal mortality, with less complete levels of case identification. Taking all regions together, case recruitment was over 90%, but the figures from the Netherlands, Spain and Greece are not as solidly established. Differences between regions in the way cases were reviewed could not be completely avoided. In England, for example, the cases had already been audited and narratives produced after multidisciplinary audit. It is unclear whether this different procedure in England caused a bias in the audit towards a higher proportion of cases where suboptimal care was identified when compared with the other regions. Although data collection on antenatal care proved to be difficult in a number of regions where the people responsible for the delivery did not provide antenatal care, no association was found between completeness of information in the antenatal care period and the grade allocated13. An acceptable level of objectivity and standardisation in preparing narrative summaries was achieved by having the narratives compiled by independent researchers and by using a common format. In 96% of cases, the use of a questionnaire and a narrative summary provided sufficient information for the audit panel to grade the extent of suboptimal factors. It was also demonstrated that the audit procedure itself was reliable and consistent13. Each panel member thought that the cases were well blinded and that it was possible to guess the country of origin only in a small number of cases. Within-subpanel variability for grading at a level of 0 or 1 versus 2 or 3 (no or unlikely contribution versus possible or likely contribution to the fatal outcome) was good (kappa value: 0.74). Reproducibility of grading between subpanels was reasonable (kappa value: 0.61). It was also established that differences between countries in assignment of grade 2 or 3 could not be explained by differences in the distribution of countries across the subpanels and by differences in the distribution of countries across time. Overall, the level of consensus reached through the EuroNatal international audit procedure was very high at 95%. The CESDI 1 in 10 analysis for England and Wales as a whole followed a similar audit procedure. In this, 50% of all cases received a grade 2 or 3 for the years 1996/1997. In the 10% sample from the seven regions in England, which were included in the CESDI 1 in 10 analysis and which also participated in the EuroNatal audit, 54% were graded 2 or 3. These percentages are very similar and provide an additional validation of the high reproducibility of the findings in at least one country14. The extent to which the regions represent the whole country is an important consideration when attempting to extrapolate the data from the regions to national level and make international comparisons. Data from Denmark and Greece covered the whole country and therefore represent D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
103
those countries completely. Flanders and Valencia represent sizeable regions of Belgium and Spain, but for various demographic and cultural reasons cannot be compared with the countries as a whole. The regions from Norway, Sweden and Finland cover such a large section of each country that they can be considered representative for those countries. In England, the 10% sample came from 7 of the 14 former NHS regions of the country. The extent to which these regions represent the whole of England, the Grampian region represents Scotland and the region in South Holland represents the Netherlands, remains to be established. Studies in Norway, Sweden, Denmark, The Netherlands, UK, New Zealand and the US have found suboptimal factors in 25% to 30% of all perinatal deaths3,5,15,16. Approximately 35% of the suboptimal factors could be categorised as maternal/social, 25% as antenatal, 20% as intrapartum and 20% as neonatal care. It is difficult to compare the results of these studies with each other and with the present study, however. This is due to differences in the categories of perinatal deaths audited and lack of standardisation of audit criteria and methodology. In a recently published study of stillbirths and neonatal deaths among babies weighing 2.5 kg or more in the West Midlands Region in England, a comparable audit methodology was followed. Overall, 62.7% of deaths were considered to have grade 2 or 3 factors. The majority (78.1%) of these factors involved clinical practice of care providers17. Differences between regions in the proportion of perinatal deaths where the quality of antenatal and perinatal care was suboptimal and the association with the overall perinatal mortality rate suggest that differences in perinatal mortality rates between countries might be explained in part by differences in the quality of antenatal and perinatal care. Differences in the quality of care might result from between-country differences in the presence and content of clinical practice guidelines for antenatal and perinatal care, and/or in barriers to adherence of these guidelines. A study of the determinants of good quality antenatal and perinatal care delivery is necessary to identify more precisely how the quality of care can be improved. The results from the EuroNatal audit study further suggest that a marked reduction in perinatal deaths in the antenatal period may be achieved by improving professional care delivery in the early detection of severe growth restriction and the management of growth retarded fetuses. Other studies have also identified inadequate detection and management of IUGR as important suboptimal factors18 – 20. A detailed discussion of definition and detection of fetal growth restriction in the EuroNatal study is given elsewhere21. As demonstrated in Fig. 1, differences between countries in the presence of suboptimal care do not change when omitting category II (single growth restricted fetal deaths of 28 or more weeks of gestational age) from the calculations. The other important factor in perinatal mortality is smoking. The review criterion applied by the audit panel was that if the mother smoked five or more cigarettes per
104
J.H. RICHARDUS ET AL.
day and severe IUGR and/or placental abruption occurred, a grade 2 (indicating a possible relationship with the fatal outcome) should be assigned within the ‘maternal/social’ category22 – 25. Mothers smoked in a relatively high percentage of cases of perinatal death in Denmark and the regions in Scotland and Norway. This could explain part of the differences in perinatal mortality between the regions. The findings of this international audit suggest that differences exist between the regions of the 10 European countries in the quality of antenatal, intrapartum and neonatal care, and that these differences contribute to the explanation of differences in perinatal mortality between these countries. The background to these differences in quality of care needs further investigation.
Acknowledgements The authors would like to thank Drs Erja Halmesma¨ki and Pertti Kirkinen for collecting Finnish data; Dr Karel Marsˇa´l for collecting Swedish data; Dr A. Bekaert for collecting Flemmish data; Drs Trond Markestad, Alf Meberg and Lauritz B. Dahl for collecting Norwegian data; Drs Judith Wolleswinkel and Corla Vredevoogt for collecting Dutch data; Pat Greig for collecting Scottish data; and Jean Sands, Mary Sidebotham, Irene Boller, Rachel Lewis, Ruth MacKenzie, Rosie Thompson, Majorie Renwick and Grace Edwards for extracting English data from CESDI records. The following members of the EuroNatal Working Group also contributed to the study: Godelieve MasuyStroobant, Louvain-la-Neuve (Belgium); Fiona Alderdice, Belfast (UK); Beatrice Blondel, Villejuif (France); Monique Kaminski, MSc, Villejuif (France); Johannes Leidinger, Mu¨nchen (Germany); Maria da Purificac¸a˜o Arau´jo, Lisbon (Portugal); and Luis F. Lopes de Oliveira, Coimbra (Portugal). The EuroNatal International Audit Panel Per Bergsjø (Chairman), Bergen (Norway); David J. Lloyd (Vice-Chairman), Aberdeen (UK); Leiv S. Bakketeig, Odense (Denmark); Elizabeth M. Bannon, Belfast (UK); Marion Borkent-Polet, Kamerik (The Netherlands); Leslie L. Davidson, London (UK); Paul Defoort, Ghent (Belgium); Ame´lia Esparteiro Leita˜o, Lisbon (Portugal); Jens Langhoff-Roos, Copenhagen (Denmark); Angel Moral Garcia, Barcelona (Spain); Niko E. Papantoniou, Athens (Greece); Margareta Wennergren, Goteborg (Sweden). The EuroNatal Working Group Willem Aelvoet, Brussels (Belgium); Birgit Bødker, Copenhagen (Denmark); Gunilla Lindmark, Uppsala (Sweden); Gillian Penney, Aberdeen (UK); Alison Macfarlane, London (UK); Paula Rantakalli, Oulu (Finland); AnnaLiisa Hartikainen, Oulu (Finland); Mika Gissler, Helsinki (Finland); Chryssa Bakoula, Athens (Greece); Vasso Lekea, Athens (Greece); Karin van der Pal, Leiden (The Netherlands); Marianne Amelink-Verburg, Leiden (The
Netherlands); Jan Holt, Bodø (Norway); Marisa Rebagliato, Alicante (Spain); Francisco Bolumar, Alicante (Spain).
References 1. Donabedian A. The study of preventability in adverse outcomes. Exploration in Quality Assessment and Monitoring. Ann Arbor, Michigan: Health Administration Press, 1985:284 – 307. 2. Dunn PM, McIlwaine G, editors. Perinatal Audit. New York: The Panthenon Publishing Group, 1996. 3. Eksmyr R, Larssen KE, Bakketeig LS, et al. Perinatal mortality in a Swedish county 1973 – 1978. Time trends revealed by perinatal audit. Acta Paediatr Scand 1986;75(1):17 – 23. 4. Bakketeig LS, Bergsjo P, Larssen KE. Perinatal audit experience in Norway: a model for international application. In: NCHS, editor. Proceedings of the International Collaborative Effort on Perinatal and Infant Mortality. Hyattsville, Maryland: U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control, National Center for Health Statistics, 1990:II3 – II10. 5. Langhoff-Roos J, Borch-Christensen H, Larsen S, Lindberg B, Wennergren M. Potentially avoidable perinatal deaths in Denmark and Sweden 1991. Acta Obstet Gynecol Scand 1996;75:820 – 825. 6. Richardus JH, Graafmans WC, Verloove-Vanhorick SP, Mackenbach JP. The perinatal mortality rate as an indicator of quality of care in international comparisons. Med Care 1998;36(1):54 – 66. 7. Richardus JH, Graafmans WC, van der Pal-de Bruin KM, AmelinkVerburg MP, Verloove-Vanhorick SP, Mackenbach JP. An European concerted action investigating the validity of perinatal mortality as an outcome indicator for the quality of antenatal and perinatal care. J Perinat Med 1997;25(4):313 – 324. 8. Marsal K, Persson PH, Larsen T, Lilja H, Selbing A, Sultan B. Intrauterine growth curves based on ultrasonically estimated foetal weights. Acta Paediatr 1996;85(7):843 – 848. 9. Anonymous. Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI). Annual Report for 1 January – 31 December 1993 (part 1). London: Maternal and Child Health Research Consortium, 1994. 10. Øvretveit J. Evaluating Health Interventions. Buckingham: Open University Press, 1998. 11. Shaw CD. Criterion based audit. BMJ 1990;300(6725):649 – 651. 12. Westerling R. Studies of avoidable factors influencing death: a call for explicit criteria. Qual Health Care 1996;5(3):159 – 165. 13. Anonymous. European comparisons of perinatal care. Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI): 8th Annual Report. London: Maternal and Child Health Research Consortium, 2001. 14. Anonymous. Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI). 6th Annual Report. Chapter 3: The 1 in 10 Enquiries 1996/7. London: Maternal and Child Health Research Consortium, 1998. 15. Eskes TKAB, Krakers RPM, Evers JLH. Vermijdbare aspecten van perinatale sterfte; consequenties voor scholing en nascholing. Ned T Geneesk 1980;124(43):1811 – 1814. 16. Eskes M, van Alten D, Treffers PE. The Wormerveer study; perinatal mortality and non-optimal management in a practice of independent midwives. Eur J Obstet Gynecol Reprod Biol 1993;51(2):91 – 95. 17. Tan KH, Wyldes MP, Settatree R, Mitchell T. Confidential regional enquiry into mature stillbirths and neonatal deaths—a multi-disciplinary peer panel perspective of the perinatal care of 238 deaths. Singapore Med J 1999;40(4):251 – 255. 18. Birdsall MA, Pattison NS. Lessons to be learnt from a perinatal audit. N Z Med J 1992;105(928):54 – 56. 19. Miranda JA, Herruzo AJ, Mozas J, et al. Influence of obstetric and perinatal care on perinatal mortality. Eur J Obstet Gynecol Reprod Biol 1996;67(2):103 – 107. 20. De Reu PA, Nijhuis JG, Oosterbaan HP, Eskes TK. Perinatal audit on
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE avoidable mortality in a Dutch rural region: a retrospective study. Eur J Obstet Gynecol Reprod Biol 2000;88(1):65 – 69. 21. Anonymous. Perinatal mortality in Europe: an outcome indicator for the quality of antenatal and perinatal care (Euronatal). Final Report for the European Union. Rotterdam: Department of Public Health, Erasmus University, 2000. 22. Meyer MB, Jonas BS, Tonascia JA. Perinatal events associated with maternal smoking during pregnancy. Am J Epidemiol 1976;103(5): 464 – 476. 23. Voigt LF, Hollenbach KA, Krohn MA, Daling JR, Hickok DE. The
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
105
relationship of abruptio placentae with maternal smoking and small for gestational age infants. Obstet Gynecol 1990;75(5):771 – 774. 24. Williams MA, Mittendorf R, Monson RR. Chronic hypertension, cigarette smoking, and abruptio placentae. Epidemiology 1991;2(6): 450 – 453. 25. Raymond EG, Cnattingius S, Kiely JL. Effects of maternal age, parity, and smoking on the risk of stillbirth. Br J Obstet Gynaecol 1994; 101(4):301 – 306. Accepted 17 October 2002
Differences in perinatal mortality and suboptimal care between 10 European regions: results of an international auditB Jan H. Richardusa,*, Wilco C. Graafmansa, S. Pauline Verloove-Vanhorickb, Johan P. Mackenbacha, The EuroNatal International Audit Panel, The EuroNatal Working Group A list of the members of the EuroNatal International Audit Panel and the EuroNatal Working Group may be found on page 104 Objective A European concerted action (the EuroNatal study) investigated the background of differences in perinatal mortality between countries of Europe. The study aimed to determine the contribution of differences in quality of care, by looking at differences in the presence of suboptimal factors in individual cases of perinatal death. Design Retrospective audit study. Setting Regions of 10 European countries. Population 1619 cases of perinatal death. Methods Perinatal deaths between 1993 and 1998 in regions of 10 European countries were identified. Reviewed were singleton fetal deaths (28 or more weeks of gestational age), intrapartum deaths (28 or more weeks) and neonatal deaths (34 or more weeks). Deaths with (major) congenital anomalies were excluded. Cases were blinded for region and an international audit panel reviewed them using explicit audit criteria. Main outcome measures Presence of suboptimal factors. Results The audit covered 1619 cases of perinatal death, representing 90% of eligible cases in the regions. Consensus was reached on 1543 (95%) cases. In 715 (46%) of these cases, suboptimal factors, which possibly or probably had contributed to the fatal outcome, were identified. The percentage of cases with such suboptimal care factors was significantly lower in the Finnish and Swedish regions compared with the remaining regions of Spain, the Netherlands, Scotland, Belgium, Denmark, Norway, Greece and England. Failure to detect severe IUGR (10% of all cases) and smoking in combination with severe IUGR and/or placental abruption (12%) was the most frequent suboptimal factor. There was a positive association between the proportion of cases with suboptimal factors and the overall perinatal mortality rate in the regions. Conclusions The findings of this international audit suggest that differences exist between the regions of the 10 European countries in the quality of antenatal, intrapartum and neonatal care, and that these differences contribute to the explanation of differences in perinatal mortality between these countries. The background to these differences in quality of care needs further investigation.
INTRODUCTION Reviewing the circumstances of adverse health outcomes has a long tradition, and maternal and perinatal deaths, in particular, have received extensive interest1. Several countries have national programs for this purpose, such as the
a
Department of Public Health, Erasmus MC, Erasmus Medical Centre, Rotterdam, The Netherlands b TNO Prevention and Health, Leiden, The Netherlands * Correspondence: Dr J. H. Richardus, Department of Public Health, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands. B The EuroNatal study was funded by the BIOMED 2 program of the European Commission. D RCOG 2003 BJOG: an International Journal of Obstetrics and Gynaecology PII: S 1 4 7 0 - 0 3 2 8 ( 0 2 ) 0 2 0 5 3 - 0
National Fetal and Infant Mortality Review Program (NFIMR) in the US and the Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) in the UK. Audit methods have been developed, both at the national and international levels with the aim of improving perinatal care. In the early 1990s, the European Association of Perinatal Medicine attempted to achieve standardisation across Europe in the methodology used for perinatal audit by drawing up guidelines for a common approach to perinatal audit2. This report describes the results of the first large scale international audit of perinatal mortality. Previous international audit have compared only pairs of countries, such as Norway and Sweden3,4, and Denmark and Sweden5. This study offers a unique opportunity to compare the quality of care provided within different health care systems and its possible association with perinatal mortality rates. www.bjog-elsevier.com
98
J.H. RICHARDUS ET AL.
Differences in perinatal mortality rates between countries may possibly be associated with differences in the quality of antenatal and perinatal care, in addition to other factors6. This issue was addressed directly by investigating perinatal deaths in various European countries in relation to the quality of care provided. Results from the different audit regions are compared in order to look for variations in the presence of suboptimal care factors. This international audit was part of a European collaboration, the EuroNatal study, which investigated the background to differences in perinatal mortality in various European countries7.
METHODS Regions in 10 European countries were identified aiming to represent the country as a whole with respect to urban and rural characteristics, socio-economic characteristics and ethnicity, and level of perinatal care. Variations existed in the size of the area and in the period for data collection, but lay within the years 1993 –1998. Greece conducted a cross sectional National Perinatal Survey, based on all consecutive births and perinatal deaths for eight weeks of 1998. The cases selected for review by the audit panel were singleton fetal deaths at 28 or more weeks of gestational age, with or without growth restriction, all intrapartum deaths at 28 or more weeks of gestational age and all neonatal deaths at 34 or more weeks of gestational age. In each case, major congenital anomalies were excluded. This selection corresponds to the categories II (single growth retarded fetus, born at 28 or more completed weeks of gestational age), III (single fetus without growth restriction, born at 28 or more completed weeks of gestational age), VI (all intrapartum deaths after admission at 28 or more completed weeks of gestational age) and X and XI combined (neonatal deaths among live births at 34 or more weeks of gestational age regardless of Apgar score) of the Nordic –Baltic (N– B) Perinatal Death Classification5. The N – B classification is based on the assumption that subgroups of perinatal deaths may be more or less ‘potentially avoidable’. The essential parameters are primarily chosen because they are thought to be of relevance for quality improvement within antenatal care, obstetrics and early neonatal care. The structure presents the results in such a way that will help focus on possible avoidable cases to a greater extent than classifications used previously5. In the EuroNatal study, both antenatal categories II and III were included to give the flexibility to redefine IUGR in the future. Category VI includes intrapartum deaths after admission, with admission defined as the first patient – professional contact for labour. The definition of neonatal deaths included all deaths of babies with a gestational age of 34 completed weeks or more (N – B categories X and XI combined). Babies with major congenital anomalies are a separate category (I) in the N – B classification and were not included in the EuroNatal audit.
Because national growth charts were not available for all countries, the standard growth chart used by Denmark and Sweden was adopted in this study, presenting weight for gestational age for an unselected population including infants still in utero, with weights estimated on the basis of ultrasound measurement8. Two standard deviations below the average was defined as the upper limit for severe growth restriction. Explicit criteria based on international consensus and evidence of effectiveness were incorporated into a questionnaire. Additional information was provided on social or other relevant circumstances, and on the availability of records and documents pertaining to the cases. For each element of care in the selected categories of the N – B classification, explicit criteria for optimal clinical care were defined. These criteria, nearly 50 in all, fell into the following three groups: 1. Criteria based on clear evidence available from controlled trials. The source of information was The Cochrane Pregnancy and Childbirth database of 1995 (Issue 2). 2. Criteria for which the evidence is not as firmly established as the above criteria. These also originated from The Cochrane Pregnancy and Childbirth database of 1995 (Issue 2). 3. Criteria formulated by the EuroNatal International Audit Panel based on consensus about appropriate perinatal practice, and defined primarily for standardisation of the EuroNatal audit. The Cochrane Pregnancy and Childbirth database of 1995 was used because the criteria were applicable to the whole period of the audit (1993 – 1998). The international audit panel consisted of 12 members from all participating countries except Finland. Panel members were nominated by representatives from the various countries in the EuroNatal Working Group and were recognised experts in their field. Four subpanels with three members were formed. The panel consisted of obstetricians, neonatologists, perinatal epidemiologists and a midwife. To define suboptimal care, a grading system was adapted from the Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) in the UK9. Each case was blinded for country and distributed randomly to one of the four subpanels. Each subpanel member received a copy of the questionnaire, the narrative summary and a sheet that identified lack of compliance with the explicit audit criteria. After reviewing this, the panel member decided on whether there were instances of suboptimal care. Any such instances were listed on a standard form. The panel member marked the timing as antenatal, intrapartum or neonatal and whether it was ‘maternal/social’, ‘infrastructure/service organisation’ or ‘professional care delivery’. They also marked the contribution of each suboptimal D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
Table 1. Number and percentage [n (%)] of cases entered into the audit, with the total number of cases graded and the number of cases graded with 2 and 3, in the selected N – B categories in regions of 10 European countries. Audit region
0
1
207 (100)
19 (9.2)
0
188 (90.8)
46 (24.5)
46 (24.5)
269 (100)
6 (2.2)
3 (1.1)
260 (96.7)
54 (20.8)
221 (100)
5 (2.3)
1 (0.5)
215 (97.3)
175 (100)
10 (5.7)
2 (1.1)
110 (100)
4 (3.6)
165 (100) 142 89 106 135
(100) (100) (100) (100)
1619 (100)
2
n (%) with 95% CI of percentage 2 or 3 graded 2 or 3 3 96 (51.1)
43.9 – 58.2
73 (28.1) 111 (42.7) 22 (8.5)
133 (51.2)
45.1 – 57.2
37 (17.2)
63 (29.3)
91 (42.3) 24 (11.2)
115 (53.5)
46.8 – 60.1
163 (93.1)
74 (45.4)
37 (22.7)
41 (25.2) 11 (6.7)
52 (31.9)
25.1 – 39.4
1 (0.9)
105 (95.5)
20 (19.0)
31 (29.5)
51 (48.6)
54 (51.4)
41.9 – 60.9
5 (3.0)
3 (1.8)
157 (95.2)
40 (25.5)
41 (26.1)
54 (34.4) 22 (14.0)
76 (48.4)
40.7 – 56.2
2 4 4 6
(1.4) (4.5) (3.8) (4.4)
1 (0.7) 0 0 0
139 85 102 129
38 10 18 32
46 32 39 51
45 40 37 42
55 43 45 46
(39.6) (50.6) (44.1) (35.7)
31.7 – 47.9 40.0 – 61.1 34.7 – 53.9 27.8 – 44.2
65 (4.0)
11 (0.7)
715 (46.3)
43.9 – 48.8
(97.9) (95.5) (96.2) (95.6)
1543 (95.3)
(27.3) (11.8) (17.6) (24.8)
(33.1) (37.6) (38.2) (39.5)
72 (38.3) 24 (12.8)
3 (2.9)
(32.4) 10 (7.2) (47.1) 3 (3.5) (36.3) 8 (7.8) (32.6) 4 (3.1)
369 (23.9) 459 (29.7) 584 (37.8) 131 (8.5)
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE
Flanders (Belgium) Whole country (Denmark) Seven former NHS regions (England, UK) 13/22 Hospital districts (Finland) Whole country (Greece) Part South-Holland province (The Netherlands) Four counties (Norway) Grampian (Scotland, UK) Valencia province (Spain) Southern Health Care Region (Sweden) Total
Grade (% of total graded)
Total entered Not auditable Dissensus Total graded into audit (% of total audited) (% of total audited) (% of total audited)
99
100
J.H. RICHARDUS ET AL.
Fig. 1. Percentage of deaths graded 2 or 3 in regions of 10 European countries, with 95% confidence intervals. Dark grey bars: all categories together; light grey bars: category II (single growth restricted fetal deaths) excluded.
Consensus on a final subpanel grade was reached in an audit procedure that began with a sequence of assessments by correspondence, followed by plenary meetings of the subpanels. When consensus was not reached in the assessment rounds by correspondence, the case was referred for discussion in a plenary meeting. If more than one of the three subpanel members could not assign a grade to a case because of insufficient information, the case was considered ‘not-auditable’. The consensus grade was taken from the three grades determined by the subpanel. Grade 0 was when the grades were X00, X01, 000, 001; grade 1: X11,
factor to the fatal outcome as ‘unlikely’, ‘possibly’ or ‘likely’. Then they allocated a final grade from the following scale: 0: No suboptimal factors have been identified 1: Suboptimal factor(s) have been identified, but are unlikely to have contributed to the fatal outcome 2: Suboptimal factor(s) have been identified, and might have contributed to the fatal outcome 3: Suboptimal factor(s) have been identified, and are likely to have contributed to the fatal outcome
Table 2. Total number of cases in the selected categories with number and percentage of cases graded 2 or 3 [n (%)] in regions of 10 European countries. Audit region
Flanders (Belgium) Whole country (Denmark) Seven former NHS regions (England, UK) 13/22 Hospital districts (Finland) Whole country (Greece) Part South-Holland province (The Netherlands) Four counties (Norway) Grampian (Scotland, UK) Valencia province (Spain) Southern Health Care Region (Sweden) Total
Category II (single growth restricted; 28 weeks of gestational age)
Category III (single not growth restricted; 28 weeks of gestational age)
Category VI (all intrapartum deaths; 28 weeks of gestational age)
Category X/XI (neonatal deaths; 34 weeks of gestational age)
Overall graded
Total graded
Grade 2 or 3 (% of total)
Total graded
Grade 2 or 3 (% of total)
Total graded
Grade 2 or 3 (% of total)
Total graded
Grade 2 or 3 (% of total)
52
46 (88.5)
87
21 (24.1)
22
13 (59.1)
27
16 (59.3)
188
57
49 (86.0)
146
40 (27.4)
28
20 (71.4)
29
24 (82.8)
260
35
30 (85.7)
124
41 (33.1)
16
12 (75.0)
40
32 (80.0)
215
23
17 (70.8)
107
22 (20.6)
9
5 (55.6)
24
8 (33.3)
163
21 40
21 (100) 27 (67.5)
54 71
11 (20.4) 17 (23.9)
7 26
5 (71.4) 15 (57.7)
23 20
17 (73.9) 17 (85.0)
105 157
24 19 31 22
21 18 23 18
(87.5) (94.7) (74.2) (81.8)
95 52 52 87
24 15 10 17
270 (83.3)
875
324
(25.3) (28.8) (19.2) (19.5)
2 5 8 6
218 (24.9)
129
1 5 4 3
(50.0) (100) (50.0) (50.0)
18 9 11 14
83 (64.3)
215
9 5 8 8
(50.0) (55.6) (72.7) (57.1)
139 85 102 129
144 (67.0)
1543
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE
101
Table 3. Frequency of the most common suboptimal professional care factor (failure to detect severe IUGR) and maternal/social factor (smoking) graded with 2 and 3 in all categories in regions of 10 European countries. Values are given as n (%). Total graded
Audit region
Flanders (Belgium) Whole country (Denmark) Seven former NHS regions (England, UK) 13/22 Hospital districts (Finland) Whole country (Greece) Part South-Holland province (The Netherlands) Four counties (Norway) Grampian (Scotland, UK) Valencia province (Spain) Southern Health Care Region (Sweden) Total
Failure to detect severe IUGR
188 260 215 163 105 157 139 85 102 129 1543
011, 111; grade 2: X22, X23, 222, 223; grade 3: 233, 333. The combination X33 was not used. ‘X’ means that one subpanel member considered that there was no sufficient information to assign a grade. For each case with a final grade of 2 or 3, the suboptimal factors were identified which had ‘possibly’ or ‘likely’ contributed to the fatal outcome. Only factors identified as such by at least two out of three subpanel members were taken into consideration. As a result, one or more suboptimal factors were identified for each case and categorised as explained above. Mean proportions of cases graded 2 or 3 were calculated for each region and for all regions taken together, with 95% confidence intervals. Differences between countries in the
27 25 18 9 18 19 15 7 12 8 158
(14.4) (9.6) (8.4) (5.5) (17.1) (12.1) (10.8) (8.2) (11.8) (6.2) (10.2)
Smoking Smoking and IUGR
Smoking and abruption
13 19 12 4 7 11 10 5 5 6 92
4 16 12 5 2 3 14 8 2 3 69
(6.9) (7.3) (5.6) (2.5) (6.7) (7.0) (7.2) (5.9) (4.9) (4.7) (6.0)
(2.1) (6.2) (5.6) (3.1) (1.9) (1.9) (10.1) (9.4) (2.0) (2.3) (4.5)
Smoking, abruption and IUGR 1 7 2 1 0 0 1 5 1 1 19
18 42 26 10 9 14 25 18 8 10 180
(9.6) (16.2) (12.1) (6.1) (8.6) (8.9) (18.0) (21.2) (7.8) (7.6) (11.7)
proportion of cases graded 2 or 3 were calculated by simple logistic regression (parameterisation: deviation). The relationship between the percentage of deaths graded 2 or 3 in the selected categories in the various regions (Table 1, column 10), and the overall perinatal mortality rates per 1000 in those regions were analysed by linear regression using SPSS for Windows, release 9.0.0. This is shown in Fig. 2.
RESULTS A total of 1619 cases of perinatal death in the selected N –B categories from regions in 10 European countries
Fig. 2. Perinatal mortality rate and % deaths graded 2 or 3 in regions of 10 European countries.
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
(0.5) (2.7) (0.9) (0.6) (0) (0) (0.7) (5.9) (1.0) (0.8) (1.2)
Total smoking-related factors
102
J.H. RICHARDUS ET AL.
were entered into the audit. These represented 90% of all eligible cases in the selected categories. Of these 1619 cases, 65 (4.0%) had insufficient information for the audit panel to grade, and in 11 cases (0.7%) the audit panel could not reach consensus about the grading (Table 1). The remaining 1543 cases (95.3%) received a consensus grade. In total, 369 (23.9%) were graded as 0, 459 (29.7%) as 1, 584 (37.8%) as 2 and 131 (8.5%) as 3. Table 1 shows the distribution of grades across the regions that were audited, together with the number of cases that were unauditable or in which no consensus was reached. A total of 715 cases (46.3%) were graded with 2 or 3, indicating the presence of suboptimal factors that possibly or likely contributed to the fatal outcome. These will be referred to as ‘grade 2/3 cases’. Also shown in Table 1 is the percentage of grade 2 or 3 per region, with 95% confidence intervals. The percentage of cases with such suboptimal care factors was significantly lower in the Finnish and Swedish regions when compared with the remaining regions of Spain, the Netherlands, Scotland, Belgium, Denmark, Norway, Greece and England (see also Fig. 1). Figure 1 shows the percentage of deaths with grade 2/3 in all the selected categories together for each region (dark bars). Also shown as light bars is the percentage of deaths with grade 2/3 in the selected categories with the exception of category II (single growth restricted fetal deaths of 28 or more weeks of gestational age). A similar pattern is seen as with all categories together, with the percentage of cases with such suboptimal care factors being significantly lower in the Finnish and Swedish regions when compared with the other regions. The percentage of grade 2/3 antenatal deaths with growth restriction in all regions taken together is 83.3% (range: 67.5– 100%); 24.9% (range: 19.2– 28.8%) in antenatal deaths without growth restriction, 64.3% (range: 50.0 – 100%) in intrapartum deaths and 67.0% (range: 33.3– 85.0%) in neonatal deaths (Table 2). The number of cases in each category in each region is too low to indicate whether differences between regions are statistically significant. In general, it is observed that Finland and Sweden, in which the overall percentages of grade 2/3 cases are below the average of 46.3% for all European regions taken together, also show below average percentages in each of the four graded categories. In the professional care delivery category, failure to detect severe IUGR appeared as the most prominent factor (10,2%). Table 3 shows the differences between the regions in the percentage of this suboptimal factor (range: 5.5– 17.1%). The 95% confidence interval limit for Finland (2.6 –10.2%) is just below the overall average of 10.2%. Sweden also has a low percentage (6.2%), but this is not statistically significant. The other regions are within the range of the overall average percentage. The most prominent suboptimal maternal/social factor was related to maternal smoking (11.7%). Table 3 also shows the frequency of this suboptimal factor. Overall, the percentage
found in each region varied from 6.1% (95% CI: 3.0 – 11.0%) in the Finnish regions, which is statistically significant below the overall average, to 16.2% (95% CI: 11.7– 20.6%) in Denmark, 18.0% (95% CI: 11.6– 24.4%) in the Norwegian regions and 21.2% (95% CI: 13.1– 31.4%) in the Scottish region, which are all statistically significant above the overall average. The other regions are within the overall average percentage. The EuroNatal study included selected subgroups representing approximately 60% of all deaths. Figure 2 shows a statistically significant ( P ¼ 0.04) association between the percentage of cases in the selected categories graded 2/3 in a region, and the overall perinatal mortality rate in that region as recorded in this study. In particular, the regions in Finland and Sweden with the lowest percentages of grade 2/3 cases also have the lowest overall perinatal mortality rates.
DISCUSSION Clinical audits that identify suboptimal factors influencing death, and especially deficiencies in medical care, have become common in many fields of medicine. Audit is defined as an investigation into whether an activity meets explicit standards defined in an auditing document10. The primary objective of clinical audit is to establish whether the processes of care provided to a single patient or population of patients represent those therapies and procedures that are thought or known to be associated with the achievement of a good outcome. Thus, an audit does not establish what optimal care is, but investigates whether optimal care is actually provided. Other scientific methods, such as randomised controlled trials, can be used to determine the best possible therapy or procedure. In an audit, the effectiveness of certain therapies and procedures, based either on clinical evidence or on the experience of a group of professionals, is taken for granted. This leads to explicit standards by which the care process is audited. The definition of explicit criteria and the use of external panels are considered essential for comparison of results of an audit11,12. The EuroNatal audit is unique in that it is the first attempt at an international audit of perinatal deaths on a scale that includes regions from 10 different countries with an average of 162 cases per region. The EuroNatal study therefore offers a unique opportunity to compare aspects of the quality of care provided within different health care systems in cases where the outcome was death. To ensure valid comparison between regions, the audit had to satisfy several conditions. These have been discussed separately13. It was concluded that the inevitable differences per region in level of case identification and recruitment and quality of information were not a serious threat for the validity of the audit. A high level of case identification was achieved in Flanders, Denmark, England, D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE
Finland, Norway, Scotland and Sweden because of the possibility for linkage between medical and civil registrations and other routine case ascertainment procedures. In Greece, the Netherlands and Spain, special studies were necessary to identify all relevant cases of perinatal mortality, with less complete levels of case identification. Taking all regions together, case recruitment was over 90%, but the figures from the Netherlands, Spain and Greece are not as solidly established. Differences between regions in the way cases were reviewed could not be completely avoided. In England, for example, the cases had already been audited and narratives produced after multidisciplinary audit. It is unclear whether this different procedure in England caused a bias in the audit towards a higher proportion of cases where suboptimal care was identified when compared with the other regions. Although data collection on antenatal care proved to be difficult in a number of regions where the people responsible for the delivery did not provide antenatal care, no association was found between completeness of information in the antenatal care period and the grade allocated13. An acceptable level of objectivity and standardisation in preparing narrative summaries was achieved by having the narratives compiled by independent researchers and by using a common format. In 96% of cases, the use of a questionnaire and a narrative summary provided sufficient information for the audit panel to grade the extent of suboptimal factors. It was also demonstrated that the audit procedure itself was reliable and consistent13. Each panel member thought that the cases were well blinded and that it was possible to guess the country of origin only in a small number of cases. Within-subpanel variability for grading at a level of 0 or 1 versus 2 or 3 (no or unlikely contribution versus possible or likely contribution to the fatal outcome) was good (kappa value: 0.74). Reproducibility of grading between subpanels was reasonable (kappa value: 0.61). It was also established that differences between countries in assignment of grade 2 or 3 could not be explained by differences in the distribution of countries across the subpanels and by differences in the distribution of countries across time. Overall, the level of consensus reached through the EuroNatal international audit procedure was very high at 95%. The CESDI 1 in 10 analysis for England and Wales as a whole followed a similar audit procedure. In this, 50% of all cases received a grade 2 or 3 for the years 1996/1997. In the 10% sample from the seven regions in England, which were included in the CESDI 1 in 10 analysis and which also participated in the EuroNatal audit, 54% were graded 2 or 3. These percentages are very similar and provide an additional validation of the high reproducibility of the findings in at least one country14. The extent to which the regions represent the whole country is an important consideration when attempting to extrapolate the data from the regions to national level and make international comparisons. Data from Denmark and Greece covered the whole country and therefore represent D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
103
those countries completely. Flanders and Valencia represent sizeable regions of Belgium and Spain, but for various demographic and cultural reasons cannot be compared with the countries as a whole. The regions from Norway, Sweden and Finland cover such a large section of each country that they can be considered representative for those countries. In England, the 10% sample came from 7 of the 14 former NHS regions of the country. The extent to which these regions represent the whole of England, the Grampian region represents Scotland and the region in South Holland represents the Netherlands, remains to be established. Studies in Norway, Sweden, Denmark, The Netherlands, UK, New Zealand and the US have found suboptimal factors in 25% to 30% of all perinatal deaths3,5,15,16. Approximately 35% of the suboptimal factors could be categorised as maternal/social, 25% as antenatal, 20% as intrapartum and 20% as neonatal care. It is difficult to compare the results of these studies with each other and with the present study, however. This is due to differences in the categories of perinatal deaths audited and lack of standardisation of audit criteria and methodology. In a recently published study of stillbirths and neonatal deaths among babies weighing 2.5 kg or more in the West Midlands Region in England, a comparable audit methodology was followed. Overall, 62.7% of deaths were considered to have grade 2 or 3 factors. The majority (78.1%) of these factors involved clinical practice of care providers17. Differences between regions in the proportion of perinatal deaths where the quality of antenatal and perinatal care was suboptimal and the association with the overall perinatal mortality rate suggest that differences in perinatal mortality rates between countries might be explained in part by differences in the quality of antenatal and perinatal care. Differences in the quality of care might result from between-country differences in the presence and content of clinical practice guidelines for antenatal and perinatal care, and/or in barriers to adherence of these guidelines. A study of the determinants of good quality antenatal and perinatal care delivery is necessary to identify more precisely how the quality of care can be improved. The results from the EuroNatal audit study further suggest that a marked reduction in perinatal deaths in the antenatal period may be achieved by improving professional care delivery in the early detection of severe growth restriction and the management of growth retarded fetuses. Other studies have also identified inadequate detection and management of IUGR as important suboptimal factors18 – 20. A detailed discussion of definition and detection of fetal growth restriction in the EuroNatal study is given elsewhere21. As demonstrated in Fig. 1, differences between countries in the presence of suboptimal care do not change when omitting category II (single growth restricted fetal deaths of 28 or more weeks of gestational age) from the calculations. The other important factor in perinatal mortality is smoking. The review criterion applied by the audit panel was that if the mother smoked five or more cigarettes per
104
J.H. RICHARDUS ET AL.
day and severe IUGR and/or placental abruption occurred, a grade 2 (indicating a possible relationship with the fatal outcome) should be assigned within the ‘maternal/social’ category22 – 25. Mothers smoked in a relatively high percentage of cases of perinatal death in Denmark and the regions in Scotland and Norway. This could explain part of the differences in perinatal mortality between the regions. The findings of this international audit suggest that differences exist between the regions of the 10 European countries in the quality of antenatal, intrapartum and neonatal care, and that these differences contribute to the explanation of differences in perinatal mortality between these countries. The background to these differences in quality of care needs further investigation.
Acknowledgements The authors would like to thank Drs Erja Halmesma¨ki and Pertti Kirkinen for collecting Finnish data; Dr Karel Marsˇa´l for collecting Swedish data; Dr A. Bekaert for collecting Flemmish data; Drs Trond Markestad, Alf Meberg and Lauritz B. Dahl for collecting Norwegian data; Drs Judith Wolleswinkel and Corla Vredevoogt for collecting Dutch data; Pat Greig for collecting Scottish data; and Jean Sands, Mary Sidebotham, Irene Boller, Rachel Lewis, Ruth MacKenzie, Rosie Thompson, Majorie Renwick and Grace Edwards for extracting English data from CESDI records. The following members of the EuroNatal Working Group also contributed to the study: Godelieve MasuyStroobant, Louvain-la-Neuve (Belgium); Fiona Alderdice, Belfast (UK); Beatrice Blondel, Villejuif (France); Monique Kaminski, MSc, Villejuif (France); Johannes Leidinger, Mu¨nchen (Germany); Maria da Purificac¸a˜o Arau´jo, Lisbon (Portugal); and Luis F. Lopes de Oliveira, Coimbra (Portugal). The EuroNatal International Audit Panel Per Bergsjø (Chairman), Bergen (Norway); David J. Lloyd (Vice-Chairman), Aberdeen (UK); Leiv S. Bakketeig, Odense (Denmark); Elizabeth M. Bannon, Belfast (UK); Marion Borkent-Polet, Kamerik (The Netherlands); Leslie L. Davidson, London (UK); Paul Defoort, Ghent (Belgium); Ame´lia Esparteiro Leita˜o, Lisbon (Portugal); Jens Langhoff-Roos, Copenhagen (Denmark); Angel Moral Garcia, Barcelona (Spain); Niko E. Papantoniou, Athens (Greece); Margareta Wennergren, Goteborg (Sweden). The EuroNatal Working Group Willem Aelvoet, Brussels (Belgium); Birgit Bødker, Copenhagen (Denmark); Gunilla Lindmark, Uppsala (Sweden); Gillian Penney, Aberdeen (UK); Alison Macfarlane, London (UK); Paula Rantakalli, Oulu (Finland); AnnaLiisa Hartikainen, Oulu (Finland); Mika Gissler, Helsinki (Finland); Chryssa Bakoula, Athens (Greece); Vasso Lekea, Athens (Greece); Karin van der Pal, Leiden (The Netherlands); Marianne Amelink-Verburg, Leiden (The
Netherlands); Jan Holt, Bodø (Norway); Marisa Rebagliato, Alicante (Spain); Francisco Bolumar, Alicante (Spain).
References 1. Donabedian A. The study of preventability in adverse outcomes. Exploration in Quality Assessment and Monitoring. Ann Arbor, Michigan: Health Administration Press, 1985:284 – 307. 2. Dunn PM, McIlwaine G, editors. Perinatal Audit. New York: The Panthenon Publishing Group, 1996. 3. Eksmyr R, Larssen KE, Bakketeig LS, et al. Perinatal mortality in a Swedish county 1973 – 1978. Time trends revealed by perinatal audit. Acta Paediatr Scand 1986;75(1):17 – 23. 4. Bakketeig LS, Bergsjo P, Larssen KE. Perinatal audit experience in Norway: a model for international application. In: NCHS, editor. Proceedings of the International Collaborative Effort on Perinatal and Infant Mortality. Hyattsville, Maryland: U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control, National Center for Health Statistics, 1990:II3 – II10. 5. Langhoff-Roos J, Borch-Christensen H, Larsen S, Lindberg B, Wennergren M. Potentially avoidable perinatal deaths in Denmark and Sweden 1991. Acta Obstet Gynecol Scand 1996;75:820 – 825. 6. Richardus JH, Graafmans WC, Verloove-Vanhorick SP, Mackenbach JP. The perinatal mortality rate as an indicator of quality of care in international comparisons. Med Care 1998;36(1):54 – 66. 7. Richardus JH, Graafmans WC, van der Pal-de Bruin KM, AmelinkVerburg MP, Verloove-Vanhorick SP, Mackenbach JP. An European concerted action investigating the validity of perinatal mortality as an outcome indicator for the quality of antenatal and perinatal care. J Perinat Med 1997;25(4):313 – 324. 8. Marsal K, Persson PH, Larsen T, Lilja H, Selbing A, Sultan B. Intrauterine growth curves based on ultrasonically estimated foetal weights. Acta Paediatr 1996;85(7):843 – 848. 9. Anonymous. Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI). Annual Report for 1 January – 31 December 1993 (part 1). London: Maternal and Child Health Research Consortium, 1994. 10. Øvretveit J. Evaluating Health Interventions. Buckingham: Open University Press, 1998. 11. Shaw CD. Criterion based audit. BMJ 1990;300(6725):649 – 651. 12. Westerling R. Studies of avoidable factors influencing death: a call for explicit criteria. Qual Health Care 1996;5(3):159 – 165. 13. Anonymous. European comparisons of perinatal care. Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI): 8th Annual Report. London: Maternal and Child Health Research Consortium, 2001. 14. Anonymous. Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI). 6th Annual Report. Chapter 3: The 1 in 10 Enquiries 1996/7. London: Maternal and Child Health Research Consortium, 1998. 15. Eskes TKAB, Krakers RPM, Evers JLH. Vermijdbare aspecten van perinatale sterfte; consequenties voor scholing en nascholing. Ned T Geneesk 1980;124(43):1811 – 1814. 16. Eskes M, van Alten D, Treffers PE. The Wormerveer study; perinatal mortality and non-optimal management in a practice of independent midwives. Eur J Obstet Gynecol Reprod Biol 1993;51(2):91 – 95. 17. Tan KH, Wyldes MP, Settatree R, Mitchell T. Confidential regional enquiry into mature stillbirths and neonatal deaths—a multi-disciplinary peer panel perspective of the perinatal care of 238 deaths. Singapore Med J 1999;40(4):251 – 255. 18. Birdsall MA, Pattison NS. Lessons to be learnt from a perinatal audit. N Z Med J 1992;105(928):54 – 56. 19. Miranda JA, Herruzo AJ, Mozas J, et al. Influence of obstetric and perinatal care on perinatal mortality. Eur J Obstet Gynecol Reprod Biol 1996;67(2):103 – 107. 20. De Reu PA, Nijhuis JG, Oosterbaan HP, Eskes TK. Perinatal audit on
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
PERINATAL MORTALITY AND SUBOPTIMAL CARE IN EUROPE avoidable mortality in a Dutch rural region: a retrospective study. Eur J Obstet Gynecol Reprod Biol 2000;88(1):65 – 69. 21. Anonymous. Perinatal mortality in Europe: an outcome indicator for the quality of antenatal and perinatal care (Euronatal). Final Report for the European Union. Rotterdam: Department of Public Health, Erasmus University, 2000. 22. Meyer MB, Jonas BS, Tonascia JA. Perinatal events associated with maternal smoking during pregnancy. Am J Epidemiol 1976;103(5): 464 – 476. 23. Voigt LF, Hollenbach KA, Krohn MA, Daling JR, Hickok DE. The
D RCOG 2003 Br J Obstet Gynaecol 110, pp. 97 – 105
105
relationship of abruptio placentae with maternal smoking and small for gestational age infants. Obstet Gynecol 1990;75(5):771 – 774. 24. Williams MA, Mittendorf R, Monson RR. Chronic hypertension, cigarette smoking, and abruptio placentae. Epidemiology 1991;2(6): 450 – 453. 25. Raymond EG, Cnattingius S, Kiely JL. Effects of maternal age, parity, and smoking on the risk of stillbirth. Br J Obstet Gynaecol 1994; 101(4):301 – 306. Accepted 17 October 2002