The value of obstetric surveillance systems within the National Health Service
Correspondence: Pippa Kyle. Email: email@example.com
- There are several obstetric surveillance systems currently used within the National Health Service (NHS).
- These systems employ a variety of techniques to monitor a range of important health conditions and medical services.
- Uses of surveillance data include monitoring disease, clinical audit, research and resource planning.
- A surveillance system must be designed appropriately if it is to achieve its desired purpose and remain a justifiable use of public funds.
- To be able to identify the different surveillance system types currently employed within obstetrics and their appropriate uses.
- To understand the advantages and limitations of surveillance system data in obstetrics.
- To appreciate the importance of appropriate surveillance system design in generating useful data.
- Surveillance systems contain large volumes of personal data. Relevant legal permissions are required to collect, store and permit access to such data that is often collected without explicit consent.
Effective monitoring of maternal and perinatal health is an essential component of obstetric care. By doing so, we can determine who is affected by disease; evaluate current healthcare and service provision; develop new services and technologies; and identify areas for future research.
Several surveillance systems are currently used to monitor maternal health and health care. Important examples include the Confidential Enquiry into Maternal Deaths (CEMD); the UK Obstetric Surveillance System; Hospital Episode Statistics maternity data; and regional congenital anomaly registers.
Methods of surveillance in obstetric practice within the National Health Service (NHS)
Surveillance is the ‘continuous and systematic process of collection, analysis, interpretation, and dissemination of descriptive information for monitoring health problems’. Methods of surveillance currently employed within NHS obstetrics include notifiable disease or event reporting; medical registries; information systems; and surveys.
Notifiable disease and death reporting
The surveillance of certain diseases and all deaths is governed by law. Clinicians who diagnose notifiable diseases such as rubella and acute infectious hepatitis during the antenatal period are personally responsible for reporting the case to the Health Protection Agency. All cases of maternal death are required by the Government to be submitted to the CEMD. This enquiry began in England in Wales in 1951 and became UK-wide in 1985. It examines and reports on all deaths of women while pregnant or within 42 days of the end of pregnancy regardless of the cause of death. The MBRRACE-UK (Mothers and Babies – Reducing Risk through Audits and Confidential Enquiries across the UK) collaboration has recently taken over the responsibility of enquiry data collection.
A medical registry is the ‘systematic collection of health and demographic data for patients with specific health characteristics, held in a central database for a predefined purpose’. Registries are an excellent method of continuous surveillance capable of storing large volumes of data. The development of a registry to the point where it produces meaningful data can take years and the administrative structures involved are expensive. Registry-based surveillance is therefore typically reserved for conditions such as epilepsy in pregnancy and congenital anomaly, which are likely to be of significant public health importance for many years to come. The UK Epilepsy & Pregnancy Register is a well-established registry that has provided data to facilitate counselling and advice regarding antiepileptic medications and risk of teratogenesis. Several regional and disease-specific registries are used to monitor trends in congenital anomalies within the UK.[6, 7] Examples include the Northern Congenital Abnormality Survey and the National Down Syndrome Cytogenetic Register. These registries contribute to larger surveillance networks including the British Isles Network of Congenital Anomaly Registers (BINOCAR); and the European network of congenital anomaly registers (EUROCAT).
Unlike disease-specific registries, databases such as Hospital Episode Statistics (HES), the Scottish Birth Record (SBR) and the Scottish Morbidity Records (SMR) typically collect data on more general patient populations. HES contains details of all admissions to NHS hospitals in England including those to maternity units. The SBR aims to collect data on every baby born in Scotland, whereas the Scottish Morbidity Records 02 (SMR02) database collects information on clinical and demographic characteristics and outcomes of all patients discharged from Scottish maternity units.[9, 10] National databases contain data that can be used to evaluate uncomplicated pregnancies and provide denominator data for more specialised surveillance systems. Because these databases are not designed according to any one investigator's specific surveillance objective, care must be taken to ensure that the data are appropriate for the analysis required. For example, although HES maternity data have been used as a quality indicator for effectiveness and performance across regions and hospitals within the UK there is minimal coverage of births occurring at home and in private maternity hospitals. The national birth rate would therefore be underestimated.
The UK Obstetric Surveillance System (UKOSS) is a prospective monthly case survey designed specifically for the study of rare pregnancy complications. With contributions from all consultant-led maternity units in the UK, nominated clinicians receive a report card from UKOSS on a monthly basis listing the conditions currently under surveillance. Cases reported via these cards are followed up by UKOSS who send the reporting clinician a second, more detailed form to complete from the patients' clinical notes. UKOSS studies are limited to conditions with an estimated incidence of fewer than one in 2000 births. Unlike many methods of surveillance, surveys can be used for short-term surveillance. In addition, it is possible to conduct case–control and cohort studies by asking participating centres to identify and supply data on women to be used in control groups. This surveillance system enables new studies to be commenced quickly, providing an excellent way to monitor emerging diseases, such as new strains of influenza, and their effect on fetal and maternal health. Nevertheless there are disadvantages to surveys such as UKOSS in that they rely heavily on the cooperation of reporting clinicians. Although investigators proactively survey for cases, unlike with the CEMD, there is no legal obligation to report to UKOSS. For some busy clinicians this surveillance may not be seen as a priority and so underreporting ensues. Although surveys are considered a relatively inexpensive and thus efficient surveillance technique, they do not reflect the time and thus hidden expense associated with clinicians reporting and submitting data.
The purpose of surveillance in obstetrics
Surveillance data have a variety of uses. Although systems are designed with specific purposes in mind, often the data collected can have additional research and clinical uses.
Monitoring disease trends
Surveillance data tell us when and where disease occurs and whom it affects. Over time surveillance allows us to identify changes in the nature and/or extent of disease and highlights factors that may be responsible for such changes. Changes in the rate and causes of maternal death are monitored by the CEMD. Recent observations made by the Enquiry include the high incidence of obesity among maternal deaths and the increasing rate of maternal death secondary to sepsis over the past decade.[3, 13] Such observations have allowed us to identify risk factors for maternal death and act as important stimuli for the improvement of care. Similar studies can be carried out by independent research groups using national information systems. For example, linked data from the Scottish Stillbirth and Infant Death Survey and the SMR02 were recently used to evaluate the causes and trends of intrapartum stillbirth and neonatal death at term in Scotland over the past two decades. Continuous, rather than ad hoc, surveillance such as that provided by regional congenital anomaly registers allows investigators to detect increases in disease incidence at the earliest possible opportunity. These registries have also been used evaluate public health interventions, such as folic acid supplementation, that aim to reduce the incidence of congenital disease.
Clinical audit and evaluation
The regular audit of performance is a key component of good medical practice. The CEMD is designed to collect data to audit the care provided to women who died during or shortly after pregnancy. Lessons learned from individual cases are used in the development of new protocols and guidelines such as the RCOG guidelines for thromboprophylaxis and new guidance on the management of women with obesity in pregnancy.[17, 18] The fall in incidence of maternal death secondary to thromboembolism following the publication of the former guideline is testament to the success of the contributing surveillance systems. Surveillance at a population rather than case-by-case level, via the National Down Syndrome Cytogenetic Register, has contributed to work done by the National Screening Committee Monitoring Programme in assessing the usefulness of NHS antenatal diagnosis strategies.
Surveillance as a research tool
Surveillance systems are powerful research tools even though they often contain only routinely collected clinical data. Increasing the use of routine data represents a potentially cost-efficient approach to expanding the capacity for obstetric research.
The development of evidence-based management techniques and strategies is an important component of obstetric research. Randomised controlled trials are often difficult to conduct in obstetrics and are typically impracticable for the study of rare diseases for which interventions may still be of benefit. In such cases surveillance data may be used to help generate a consensus on optimal management. UKOSS has conducted a number of prospective case–control and cohort studies which have improved the evidence base surrounding management of several rare pregnancy-related conditions including antenatal pulmonary embolism, tuberculosis in pregnancy, and amniotic fluid embolism.[21-23] The population-based nature of such data increases the likelihood that study findings can be generalised to the wider public. Although not frequently employed at present, registries can be used as a source of outcome data for randomised controlled trials; indeed the feasibility of evaluating an obstetric intervention using only routinely collected registry data has been demonstrated in Finland.
Data from congenital anomaly registries have improved our understanding of the natural history of congenital disease. The Northern Congenital Abnormality Survey recently reported a detailed analysis of 20-year survival rates for many different congenital disorders. Accurate information regarding long-term prognosis is invaluable to obstetricians when counselling patients following the diagnosis of a congenital anomaly.
National datasets can be used for population-based studies to answer a range of research questions. Examples include the recent characterisation of major risk factors for stillbirth and evaluation of the effects of maternal obesity on pregnancy outcomes.[26, 27]
Planning future clinical care, service provision and policies
Important outputs from information systems such as HES include annual national maternity statistics publications. These allow monitoring of year-on-year changes in service use, delivery parameters and perinatal outcomes at national, regional and local levels. By relating changes in patient demographics to changes in the use of various healthcare services, investigators can help healthcare providers to estimate future service requirements. Analysis of Scottish national data, for example, has suggested that more than one-third of the increase in unplanned caesarean rates seen among women in Scotland between 1980 and 2005 was directly attributable to women delaying the start of their families. Importantly this suggests that a significant component of the increasing caesarean rate cannot be accounted for by physician or patient preference and that such an increase in the caesarean rate is likely to remain so long as women continue to delay the onset of childbirth. Beyond clinical guideline development, surveillance data are used to inform government policies. England's National Service Framework for Maternity Services and its implementation strategy ‘Maternity Matters’ were both developed using outputs from the CEMD.[30, 31]
Surveillance system design and limitations
A successful surveillance system requires a clearly defined statement of purpose that should be reflected in the design of the system. All systems should collect some demographic data to allow investigators to generalise findings to other populations, but inclusion criteria and data elements should otherwise address specific surveillance needs. The limitations of surveillance must also be recognised. Surveillance systems are typically expensive to run; difficult to complete; and pose dilemmas with respect to confidentiality, consent and data protection.
Active versus passive ascertainment
A key feature of surveillance system design is the method by which cases are identified. Completeness, capturing all relevant cases and removing duplicates, is particularly important to accurately measure incidence and avoid selection bias. ‘Active’ case ascertainment requires the surveillance system operators to seek out cases of the disease(s). Multiple sources of data can be accessed to locate cases and to ensure accurate and complete ascertainment. A truly ‘active’ ascertainment process requires both the surveillance system and the contributing reporters to seek out cases. In the case of the CEMD there is a legal requirement to report cases. In regional congenital anomaly registers, however, although they survey a range of NHS maternity, neonatal and paediatric units, cytogenetic laboratories, and postmortem reports to identify cases within their catchment area, there can be no guarantee that the contributors report all cases.
Conversely, ‘passive’ case ascertainment relies on contributors voluntarily submitting data to the system as and when they come across a case. Whereas UKOSS is proactive in regularly providing clinicians with an opportunity to report relevant cases, it cannot be regarded as a truly active surveillance system as it depends entirely on the good will of reporting clinicians and their ability to identify all relevant cases at their unit. The disadvantages of passive ascertainment are highlighted by the former National Congenital Anomaly System (NCAS). Designed to provide early warnings of clusters of congenital anomalies following the thalidomide tragedy of 1958–61, NCAS relied on the voluntary reporting of congenital disease by NHS trusts to supplement data contributions from regional registries. Variable trust participation rates resulted in incomplete data collection. Such data were deemed to be of only limited value and consequently did not represent a justifiable use of public funds. NCAS ceased to collect data in 2009.
Financial cost is a major obstacle in the development and maintenance of surveillance systems. The financial commitments associated with surveillance are long term. In addition to costs incurred during the design and development of a surveillance system there are day-to-day running costs associated with data collection, processing, analysis, interpretation and dissemination. Systems that demand high levels of validity and completeness will incur costs associated with active ascertainment and quality control. These costs may be acceptable when the value of the data collected is high. SMR02, for example, is subject to regular quality assurance checks, but has been more than 99% complete for over three decades.
Consent, confidentiality and data protection
By definition, national registers and medical registries contain large volumes of personal data. Maintaining confidentiality is not only an ethical obligation; it is also vital if public and professional support of surveillance is to continue. The degree to which systems contain identifying data varies according to system design. Identifying data facilitate quality control and may allow linkage between surveillance systems. If patient contact data are stored then novel services can be offered and participants may be recruited for more detailed studies.
The complexities of issues surrounding the use of patient data for research when collected without informed consent are well documented. Statutory permission to collect such data must be granted before a surveillance system is put into operation and comes with a variety of restrictions. UKOSS is not permitted to hold any identifying data. Unfortunately the collection of medical data without explicit consent appears to be necessary to ensure completeness and avoid the unacceptable selection bias that has been reported to occur in registries that require informed consent for inclusion.
Several surveillance systems currently exist in the UK that are of relevance and value to obstetricians. These surveillance systems allow us to go far beyond descriptive epidemiology. Using surveillance data we are able to improve standards of care; better understand the natural history of disease; evaluate the safety and efficacy of diagnostic and therapeutic interventions; and plan for future service provision. The continued success of surveillance systems such as UKOSS and the CEMD will depend not only on adequate funding but also on clinicians' appreciation of these values and their continued participation both in reporting cases and suggesting future areas of research interest.
Disclosure of interests
None to declare.