Hypertensive disorders of pregnancy are a major cause of morbidity and mortality worldwide. Reliable, published individual patient data from units and countries are lacking. Without these data, clinicians are unable to benchmark their incidence, treatments and outcomes, and patient safety is unable to be routinely assessed. Available data suggest that a notable proportion of the adverse events that occur with hypertensive disease of pregnancy may be preventable. Theory and practice indicate several methods that can offer the possibility of averting these preventable adverse events. These methods include benchmarking outcomes, standardisation of care processes, simulation, and enhancement of patient knowledge. However, data on optimal methods to enhance patient safety and quality of care of pregnant women with hypertensive disease remain limited, and further research is required.
Benchmarking healthcare: the concept
The use of benchmarking techniques has increased in the healthcare sector since the 1990s. The use of these techniques in health care has largely been pioneered in the USA through the publication of annual reports such as America’s Best Hospitals annual survey. Performance measurement systems that measure financial performance in non-healthcare sectors have been modified to consider multiple factors, not just financial benefit (i.e. consumer satisfaction, length of stay and morbidities). The theory behind the transference of these techniques from manufacturing to health care is that the latter is also driven by improved performance outcomes in which monetary gain is replaced with less morbidity and better consumer health. It would be naïve to assume that financial performance was not also at play in the healthcare sector, as better health care (less morbidity, shorter hospital stay) may also equate to less cost for the healthcare provider. This is a consideration not to be dismissed considering that the healthcare ‘industry’ comprises 14% of the gross domestic product of the USA and 8% in Australia.
There are many examples of how benchmarking has been used in health care, which highlight the techniques involved and their effect. On an international scale, the World Health Organization instituted a practical tool to monitor and improve the performance of hospitals. The Performance Assessment Tool for Quality Improvements in Hospitals (PATH) was designed to measure clinical effectiveness, efficiency, staff orientation, responsive governance, safety and patient centredness. Fifty-one hospitals from six regions completed the initial data collection. Data were compared between countries, but of greater relevance between regions of similar economic development. Other benchmarking initiatives, such as the Acute Physiology and Chronic Health Evaluation (APACHE) model developed in 1993, have targeted specific service providers, such as intensive care units. Now in version IV, APACHE as a statistical model benchmarks expected or predicted outcomes and compares them with actual outcomes. This type of predictive testing is what fuels the case-mix modelling, and has been incorporated into the Australian Diagnostic Related Groups framework. This type of data collection allows national comparisons to be made on predicted factors based on patient and hospital profiles.
Benchmarking is also used in disease-specific areas on a regular basis. These initiatives are aimed at measuring the performance of individuals, units, services, or both. Many programmes call themselves benchmarking without a true understanding of the cyclical process involved. Data on stroke rehabilitation, pain management and diabetes outcomes for example, can all be reliably collected and comparisons made; however, the cyclical process of change implementation and re-evaluation need to be undertaken to complete the process and allow the full benefit of its power to influence change in clinical practice.
Benchmarking obstetric morbidity worldwide
United Kingdom
The UK Obstetric Surveillance System was established by the National Perinatal Epidemiology Unit and the Royal College of Obstetricians and Gynaecologists in 2005 to monitor rare obstetric conditions (one case per 2000 births). These include amniotic fluid embolism, extreme obesity, myocardial infarction, pulmonary vascular disease, non-renal solid organ transplant, renal transplant, stroke in pregnancy, failed intubation, malaria, uterine rupture and treatments for postpartum haemorrhage. Reporting to this system is voluntary, and return rates are 87–97% per annum since initiation.
United States of America
The US Center for Disease Control and Prevention reports maternal mortality, rates of hypertension (pregnancy associated) and hypertension (chronic) via the National Vital Statistics Report. This mandatory report on all births in the USA does not include other markers of maternal morbidity.
European Union
In the European Union, perinatal information is collected through a wide variety of methods. Data collection is compulsory via a specialised birth registry in the following countries: Belgium, the Czech Republic, Denmark, Estonia, Finland, Germany, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Norway, the Slovak Republic, Slovenia and Sweden. In Austria, Cyprus, Finland, France, Hungary, Poland, and Spain, birth data are collected from hospital coding systems. France, Italy and Spain regularly conduct surveys to collect perinatal information from medical records and from interviews with mothers. The definitions of morbidity vary between each data collection system used, and these methods are best used to monitor mortality rather than morbidity.
Australia
The National Perinatal Data Collection are constructed from data collected under the Midwives Data Collection (MDC). In New South Wales, Australia, data are compiled into the NSW Mothers and Babies Report. This mandatorily records every birth 20 weeks’ gestation or over or greater than 400 g. A limited dataset of events occur during the initial birth hospitalisation. Morbidity data are non-specific in that severity of complications arising from diagnoses are not recorded. Two validation studies have been undertaken of the MDC: the first in 1990 and the second in 1998. Both studies concluded that, although the MDC was accurate in most data fields (i.e. birth date, birth weight and gender), reporting of maternal medical conditions during and before pregnancy needed to improve. An independent validation study comparing maternal diagnoses as hand collected and those reported in the MDC conducted in 2003, concluded that the MDC miscoded the type of hypertension women experienced in 56% of cases reviewed and under-reported cases of hypertension by 12%.
These type of data include, for example, statistics of the number of women experiencing a complication such as the hypertensive disorders of pregnancy (HDP). The current version of the MDC has the potential to capture the hypertensive diagnoses of chronic hypertension and pregnancy-induced hypertension (proteinuric or non-proteinuric) terminology, which is not in line with the current consensus or previous consensus statements. Information is not available on hypertensive medication prescription, rates of intra-uterine growth-restricted neonates born to hypertensive mothers, maternal complications (i.e. acute renal failure, acute pulmonary oedema), intra-cerebral events (i.e. stroke), eclampsia, and length of stay variations.
The common theme throughout all of these databases is the recording of the incidence of an event. Databases such as these, for example, can provide information on the number of cases of amniotic embolism but are not necessarily designed for the purpose of benchmarking per se . Some collections, such as the New South Wales Mothers and Babies Report do compare outcomes between health sectors or health services within a geographically defined area, but most report on the state or area in which they were recorded. The voluntary reporting nature of many of these datasets also limits the validity of the data captured. The reliance upon clinicians to report events to mandatory bodies is a further limiting factor in the reliance upon existing datasets to establish standards of clinical care that reflect optimal outcomes. With a paucity of reliable international or nation-wide information collected on pregnancy outcomes, how are we to assess the efficacy of the service we provide for the most common complication of modern pregnancy.
Benchmarking obstetric morbidity worldwide
United Kingdom
The UK Obstetric Surveillance System was established by the National Perinatal Epidemiology Unit and the Royal College of Obstetricians and Gynaecologists in 2005 to monitor rare obstetric conditions (one case per 2000 births). These include amniotic fluid embolism, extreme obesity, myocardial infarction, pulmonary vascular disease, non-renal solid organ transplant, renal transplant, stroke in pregnancy, failed intubation, malaria, uterine rupture and treatments for postpartum haemorrhage. Reporting to this system is voluntary, and return rates are 87–97% per annum since initiation.
United States of America
The US Center for Disease Control and Prevention reports maternal mortality, rates of hypertension (pregnancy associated) and hypertension (chronic) via the National Vital Statistics Report. This mandatory report on all births in the USA does not include other markers of maternal morbidity.
European Union
In the European Union, perinatal information is collected through a wide variety of methods. Data collection is compulsory via a specialised birth registry in the following countries: Belgium, the Czech Republic, Denmark, Estonia, Finland, Germany, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Norway, the Slovak Republic, Slovenia and Sweden. In Austria, Cyprus, Finland, France, Hungary, Poland, and Spain, birth data are collected from hospital coding systems. France, Italy and Spain regularly conduct surveys to collect perinatal information from medical records and from interviews with mothers. The definitions of morbidity vary between each data collection system used, and these methods are best used to monitor mortality rather than morbidity.
Australia
The National Perinatal Data Collection are constructed from data collected under the Midwives Data Collection (MDC). In New South Wales, Australia, data are compiled into the NSW Mothers and Babies Report. This mandatorily records every birth 20 weeks’ gestation or over or greater than 400 g. A limited dataset of events occur during the initial birth hospitalisation. Morbidity data are non-specific in that severity of complications arising from diagnoses are not recorded. Two validation studies have been undertaken of the MDC: the first in 1990 and the second in 1998. Both studies concluded that, although the MDC was accurate in most data fields (i.e. birth date, birth weight and gender), reporting of maternal medical conditions during and before pregnancy needed to improve. An independent validation study comparing maternal diagnoses as hand collected and those reported in the MDC conducted in 2003, concluded that the MDC miscoded the type of hypertension women experienced in 56% of cases reviewed and under-reported cases of hypertension by 12%.
These type of data include, for example, statistics of the number of women experiencing a complication such as the hypertensive disorders of pregnancy (HDP). The current version of the MDC has the potential to capture the hypertensive diagnoses of chronic hypertension and pregnancy-induced hypertension (proteinuric or non-proteinuric) terminology, which is not in line with the current consensus or previous consensus statements. Information is not available on hypertensive medication prescription, rates of intra-uterine growth-restricted neonates born to hypertensive mothers, maternal complications (i.e. acute renal failure, acute pulmonary oedema), intra-cerebral events (i.e. stroke), eclampsia, and length of stay variations.
The common theme throughout all of these databases is the recording of the incidence of an event. Databases such as these, for example, can provide information on the number of cases of amniotic embolism but are not necessarily designed for the purpose of benchmarking per se . Some collections, such as the New South Wales Mothers and Babies Report do compare outcomes between health sectors or health services within a geographically defined area, but most report on the state or area in which they were recorded. The voluntary reporting nature of many of these datasets also limits the validity of the data captured. The reliance upon clinicians to report events to mandatory bodies is a further limiting factor in the reliance upon existing datasets to establish standards of clinical care that reflect optimal outcomes. With a paucity of reliable international or nation-wide information collected on pregnancy outcomes, how are we to assess the efficacy of the service we provide for the most common complication of modern pregnancy.
Worldwide epidemiology of the hypertensive disorders of pregnancy
For such a commonly occurring pregnancy complication, clinical outcomes and incidence of HDP have not been studied extensively. The oldest printed reference (1935) to an outcome evaluation of women following a hypertensive disorder of pregnancy was by Herrick and Tillman of 930 women over a 15-year period. This study concluded that toxaemia affected women’s long-term vascular health and shortened their life expectancy. An epidemiological study of pre-eclampsia conducted in 1970 concluded that this disease affected women in all nations, although maternal mortality differed greatly. A study conducted in the USA over a 12-year period examining 0.7% of births concluded that women with pre-eclampsia had a 3–35 fold increased risk of severe complications such as placental abruption, thrombocytopaenia, disseminated intravascular coagulation, pulmonary oedema and aspiration pneumonia. Outcomes for women with proteinuric compared with non-proteinuric disease presentations have also been undertaken without inclusion of other variations in disease presentation and with varying results. A study conducted in 1997 using hand-recorded prospective pregnancy details for 1183 women diagnosed with the Australasian Society for the Study of Hypertension in Pregnancy criteria, reported that women with superimposed pre-eclampsia had more signs and symptoms than women with other forms of hypertensive diseases of pregnancy and a greater number of fetal complications. This finding has been replicated in other studies.
The outcomes for 4589 women and their babies enrolled in the Calcium for Pre-eclampsia Prevention Study showed that women with hypertension had increased maternal morbidity (i.e. operative delivery, placental abruption and renal dysfunction) and increased neonatal morbidity.
A database study of all women who delivered their babies in New South Wales over a 2-year period was undertaken. Relying upon the MDC, this study concluded that 9.8% of pregnant women were affected by a hypertensive disorder, and that these women suffered greater morbidity and mortality than the normotensive cohort. A South Australian study was conducted between 1998 and 2001, which examined the outcomes for 5356 hypertensive women over a 4- year period relying upon data collected via the South Australian perinatal data collection form. This study compared birth type, delivery gestation and neonatal characteristics such as birthweight, Apgar scores and special care nursery admissions between women with pregnancy-related hypertension and those with pre-existing hypertension. Women with pregnancy-related hypertension suffered more operative delivery and worse perinatal outcomes. Studies such as these, however do not tell us is that variation in clinical practices may have resulted in outcome variation.
The common theme among all of these studies is that inter-unit, inter-state or international comparisons were not undertaken, and although each study adds significantly to our knowledge of hypertensive diseases of pregnancy at a given time point, none address the issue of outcome improvement across time or units.
Can we benchmark outcomes between units and countries in a cohesive meaningful manner?
In an attempt to address these deficits, a statewide and international benchmarking comparison was undertaken over a 3-year period between five tertiary referral obstetric units in New South Wales, Australia, and one unit in Vancouver, Canada.
Individual patient data were collected by hand review of medical records of women who experienced a viable pregnancy and who were recorded as being hypertensive during that pregnancy by standardised coding systems (International Classification of Disease 10 th edition). A series of clinical indicators upon which to base comparison were agreed upon for the exercise and applied to this dataset. These include the following: maternal mortality, acute pulmonary oedema, acute renal failure, eclampsia, perinatal mortality, admission of term neonates to neonatal intensive care units, birth weights (adjusted for gender and gestation) below the 10 th and 3 rd centile and breast feeding at discharge.
The aim of these indicators was to include a cross-section of maternal, fetal and neonatal measures. The process for indicator establishment was based upon the SMART mnemonic, originally attributed to the management consultant Drucker.
Data were collected in a disorder-specific database, and were validated for inter-rater reliability (K = 0.95).
Disorder incidence
The incidence and type of hypertension in the units studied are presented in Table 1 . The published rate of pre-eclampsia is frequently quoted as affecting 5% of all pregnancies ; however, at the 2010 International Society for the Study of Hypertension in Pregnancy, data presented indicated that the incidence of pre-eclampsia is now lower than reported in all developed countries, other than the USA, where the rate is actually increasing. An incidence of 3% of all pregnancies is a more accurate figure in most nations rather than the 5% previously calculated.
| Unit number | Total births, 2005 | Total hypertensive disorders of pregnancy a | Pre-eclampsia a | Gestational hypertension a | Chronic hypertension a | Pre-eclampsia superimposed on chronic hypertension a | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| N | % | N | % | N | % | N | % | N | % | ||
| 1 | 4725 | 472 | 10 | 190 | 4.0 | 208 | 4.5 | 45 | 1.0 | 29 | 0.5 |
| 2 | 2902 | 174 | 6 | 60 | 2.0 | 77 | 2.5 | 26 | 1.0 | 11 | 0.5 |
| 3 | 3409 | 374 | 11 | 134 | 4.0 | 188 | 5.5 | 32 | 1.0 | 20 | 0.5 |
| 4 | 2498 | 169 | 7 | 69 | 2.5 | 71 | 3.0 | 23 | 1.0 | 6 | 0.5 |
| 5 | 1969 | 128 | 7 | 50 | 3.0 | 57 | 2.5 | 12 | 1.0 | 9 | 0.5 |
| 6 | 7355 | 409 | 6 | 179 | 2.5 | 153 | 2.0 | 49 | 1.0 | 28 | 0.5 |
| All units | 22,858 | 1726 | 7.5 | 682 | 3.0 | 754 | 3.0 | 187 | 1.0 | 103 | 0.5 |
Diagnostic features
With a plethora of consensus statements available to classify these disorders, the statement of the Society of Obstetric Medicine of Australia and New Zealand (SOMANZ) was applied to the dataset. The SOMANZ consensus statement is more inclusive in parameters and reflects the multi-system nature of these disorders. The diagnostic features collated using this classification system are presented in Table 2 .
| Diagnostic feature | Pre-eclampsia (% of women) | Pre-eclampsia superimposed on chronic hypertension (% of women) | P value a |
|---|---|---|---|
| Hypertension: systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, or both, on two separate occasions 4 h apart, after 20 weeks’ gestation | 100 | 100 | NS |
| Proteinuria: ≥ 300 mg/24 h | 77 | 82 | NS |
| Renal insufficiency: serum/plasma creatinine 0.09 mmol/) | 10 | 19 | NS |
| Liver disease: abnormal liver function tests or pain | 39 | 18 | < 0.001 |
| Neurological problems: eclampsia, hyperreflexia with clonus, severe headache with hyperreflexia, persistent visual disturbance | 33 | 25 | NS |
| Haematological disturbance: thrombocytopaenia, disseminated intravascular coagulation, haemolysis | 28 | 10 | 0.002 |
| Fetal growth restriction | 10 | 8 | NS |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
