The development of perinatal audit: 20 years' experience

In the early 1950s, maternal deaths were still in the order of 1 per 1000 births and, viewed as human tragedy, a more tangible problem than death of the offspring. The Confidential enquiries into maternal deaths in England and Wales set a standard for quality assessment in the field of obstetrics. Three-yearly reports were issued regularly from 1952 onwards, and from 1985 they were extended to cover scrutiny of every maternal death in Scotland and Northern Ireland too (1). There is no doubt that these Confidential enquiries were instrumental in assuring better care and treatment, by pointing out elements of substandard care and giving advice, which was heeded. During the five decades following the first report, maternal mortality fell by nine-tenths, and most of the mothers who die in the UK (and Scandinavia) today are victims of inevitable fate, with no one to blame. Up until the end of the 1960s cesarean section in the Scandinavian countries was performed in about 2% of all births, in most cases to save or relieve the mother, while fetuses in distress were taken care of with forceps, ventouse or dexterous manipulation. If the newborn was asphyxiated, the midwife or the obstetrician took resort to gentle back-slapping, body-tilting or cold water spray, and, if unsuccessful, called on the anesthesiologist for help. Extremely preterm babies were considered nonviable and often observed without further ado. In 1967, perinatal mortality in Norway was 21 per 1000 births, when all livebirths past 16 weeks' gestational age and stillbirths past 28 weeks were counted (2). This figure did not attract much attention at the time, although in retrospect there is good reason to believe that some of these deaths were due to substandard care. Avedis Donabedian gave an account of the principles for evaluating the quality of medical care, in 1966, citing studies from various fields of medicine in the USA from the 1950s and the early 1960s (3). He emphasized the importance of using well-defined outcomes as criteria of quality and pointed out that many factors besides medical care contribute to such outcomes as recovery, restoration and survival. He discussed at length clinical records and other sources of information (interviews, direct observation of clinical activities) and whether to use all available information or to restrict expert assessment to summaries and abstracts that can be collected by less skilled persons. Measurement standards must be developed. These can be empiric or normative. Empirical standards are based on actual practice and are generally used to compare medical care in one setting with that in another, with average performance as the yardstick. Normative standards are those which define the best medical practice, with textbooks and scientific publications, highly qualified practitioners or panels of physicians defining the norms. One should then agree on where to put the threshold, whether to strive for the best medical care that can be achieved, or to let a more modest level signify “acceptable” or “adequate” care. Twelve years later, in 1978, Donabedian returned to the theme and tabulated a classification system for use in quality assessment, divided into structure, process and outcome (4). He expanded on the discussion of how much information to use for assessment in the individual case. If every particular aspect of the case is scrutinized, then so-called implicit criteria are being used for assessment. Explicit criteria, on the other hand, represent a set of standards which are predefined and supposed to identify at least acceptable practice. In light of developments since the 1970s, explicit criteria describing the best obtainable practice would today be evidence-based, as a rule extracted from Cochrane database reviews. Donabedian also described a two-stage approach: first identifying cases that do not conform to explicit criteria, to be followed by detailed review of those cases by a panel of colleagues − “peer review” (4). The panel may be “internal,” that is, its members are employed at the hospital or institution which is being reviewed, or “external,” with members coming from outside. Quality assessment of perinatal care had previously been carried out routinely, by decree in East Germany and in Western Australia (5,6) and ad hoc in certain areas in the UK (7–9), but the model was not well established elsewhere. Those who wanted to follow the examples had to set their own rules. Before we go on to recount our initial and later experiences, however, it will be useful to take a closer look at definition and methodology, with special emphasis on perinatal audit. A working party set up by the European Association of Perinatal Medicine in 1993 came up with the following definition: Perinatal audit is the systematic, critical analysis of the quality of perinatal care, including the procedures used for diagnosis and treatment, the use of resources and the resultant outcome and quality of life for women and their babies (10). It is understood that perinatal care means clinical care as given by doctors, midwives, nurses and other health professionals. Although the word “perinatal” has increasingly come to signify the baby, perinatal audit will of course also consider care of the mother. The given definition covers a variety of quality assessment activities, with the implicit understanding that the audit feedback should lead to changes to improve services. In the present context, we restrict our discourse to adverse event monitoring, the adverse event being perinatal death. The objective of the critical analysis is to point out faults and weaknesses, which were originally called “avoidable factors.” However, the term “avoidable” was often misinterpreted and has therefore been replaced. We now label them “elements of substandard care.” The authors of the 1985–87 Confidential enquiries (1) gave a working definition of “substandard care”: it meant that the care the patient received, or the care that was made available to her, fell below the standard which the authors considered should have been offered to her during that triennium. The elements may be failures in the individual case, but also underlying factors in the system of care giving. This implies that criteria for good care should and will change with time, and that there is room for subjective judgment on the part of the audit panel members. Whereas the two concepts “avoidable factors” and “substandard care” are similar, there is a wide difference between avoidable factors and potentially avoidable outcome. The conceptual difference between substandard care and potentially avoidable poor outcome is not always recognized when grading of outcome is designed in audits. During audit it is a common verdict that substandard care is taken to be the reason for a potentially avoidable poor outcome, while the same type of substandard care is also found in cases without adverse outcome, and also identified in cases with unavoidable adverse outcome. Judging substandard care, one might also relate it to optimal or local conditions. A group of professionals of high standing should form an “audit panel” mandated to perform the analysis. There are no set rules for how a panel should be composed. The term “peer review” indicates that the care professionals should be judged by people of the same rank and position as themselves. When legislation in the USA opened up for establishment of state or local Professional Standards Review Organizations and responsibility was handed over to physicians' societies, there was widespread anxiety among other health professionals that panels would consist of physicians only (4). However, there is general agreement that a perinatal audit panel should be multidisciplinary. Considering that perinatal care is provided by physicians, midwives, specialist nurses in neonatology and anesthesiology and other health professionals, some of these should with good reason also be represented. A midwife, an obstetrician and a pediatrician (neonatologist) would be obvious choices. Others to consider would be a perinatal pathologist, a general practitioner and a statistician or an epidemiologist, all of whom may contribute with relevant expertise. Panels mandated to audit individual cases of perinatal death should assess each case history with all its ramifications, including laboratory tests before and after the event: ultrasound, biochemistry, cardiotocography (CTG) and autopsy results. The objective is, according to the definition, to critically analyze the quality of care. Standards of care will depend on the circumstances under which the care is given. Emergency situations will occur in the patient's own home, during transport to the intended place of delivery, in a small maternity home or in a hospital department, in each of which the resources available to the caregivers vary widely. An audit panel should therefore look into structure, process and outcome before giving its opinion. Structure is defined as the resources available for care, including the staff, equipment and facilities, and their organization. Process is the utilization of these resources in the provision of health care, and outcome the result of the health care process (11). In the present context, the outcome is a priori taken to be perinatal death, and the concern of the panel is whether the available resources were utilized adequately in the individual case. What constitutes good, as opposed to substandard, care? Audit panel members must agree on a set of criteria for good care, as a prerequisite for consensus. These criteria can be classified according to three main dimensions: they can be either implicit or explicit, they can be either general or disease-specific, and they can be either ideal or empiric (11). As mentioned above, an explicit criterion is set in advance, based on research, written rules or treatment protocols, whereas an implicit criterion is decided on the spot in response to the question: “What would I have done in a similar situation?” In the second dimension, disease-specific criteria presuppose one model for each disease or condition, incorporating appropriate diagnostic tests, monitoring systems and therapy (e.g. for labor and delivery: if, when and how often to perform CTG, when and how to react, if and when to do fetal blood scalp sampling for analysis, appropriate techniques for cesarean section and vaginal operative delivery, and so on), while general criteria would be exceptional events, such as eclampsia, rupture of the uterus or puerperal sepsis. The third dimension (ideal or empiric) runs along a scale, from what is known with absolute certainty to be the best possible care to accepted treatment, based on experience or personal preference; or in present day vernacular, from evidence (i.e. Cochrane) based to educated guesswork. In addition to this general framework, several other issues should be addressed before an audit panel starts its work, such as “Should panel members have access to the original case notes and records or only structured abstracts?”, “Should panel members be blinded as to time and place of the actual event?”, “Should a panel member be exempted from cases involving his or her own department?” and others, all of which may directly influence the panel's verdict. One should strive to avoid biased conclusions. The likelihood of bias is higher when implicit criteria form the bulk of the evidence. It should also be considered how the results will be disseminated. If the panel has been officially mandated, it should report to the body that assigned it the task. If the audit stems from funded ad hoc research, the funding body should have its feedback, in addition to publications in scientific journals. When audit is an ongoing process, results will range from annual reports to direct case feedback to health care givers. In the following, we give practical examples of how such problems have been addressed and resolved. A medical birth registry was established in Norway in 1967. Its first reports prompted an analysis of perinatal services in the country, indicating that were not up to in either or specialist and hospital care It was that while Norway up to the had had better perinatal mortality than its and the was in the was for a perinatal audit. The for a group to perform an audit of all perinatal deaths in five of Norway in about of the annual of births in the The audit panel of an obstetrician a a perinatal pathologist, a perinatal and a every case of perinatal death in five of Norway over a A pediatrician as with one local in each These were all for and every perinatal death on special The audit panel cases of perinatal death, of which before during birth and after the cases with information from the and the of that case was and were births in while were from or higher order The panel had one for each of case had been to each member in each the as a rule the original case each case, the gave a and information while case This was followed by a general discussion that on average per case. were and available for detailed of information and for use of the panel to An of cases in judgment of unavoidable avoidable death was classified into one of three avoidable given ideal and avoidable given present conditions. avoidable deaths were to those cases where from health care might have the course of for the the of the health care were not to label a death The given ideal was to cases of birth in local with no for on the newborn in cases of and similar circumstances local perinatal deaths out of were considered to be the were given present and were given ideal The of “avoidable factors” was as in the during labor and and in the In avoidable factors were identified in two or three of these This was the first medical audit of its in It was performed by an including one for a explicit criteria age less than weeks and some with criteria were along the was by blinded of cases and found to be The results were in the where the term was to were into A was which in a report with a of care during birth and the One of the was to perinatal mandated to audit perinatal deaths on a A the same panel to out an audit on case in a The was to time during a of attention to factors and a of of mothers to a hospital with for care. and were provided in to each panel The was carried out at a with the same as in the audit described However, the panel was blinded as to the of each cases were In the first of the cases were classified as in the second The was in judged to be due to the of This perinatal audit was useful in time and the by more In the there was a in to look for of perinatal death, in a quality report in to the report from the of every established a special mandated to on all perinatal deaths in order to identify factors that might lead to better perinatal outcome and the of potentially avoidable of the had perinatal in An report in that discussion of perinatal deaths had out to be It had the on perinatal and care. In the between the clinical had as well as the and between and health care in this The report emphasized that the perinatal and their audit had been a for and and of for perinatal care. The of a perinatal audit is to the quality of care. However, perinatal deaths are and of and of death widely. some deaths are potentially more avoidable than this one should strive for a system of perinatal death classification that the perinatal deaths in appropriate for quality including by and between It should on for into which should be with areas for health care In order to analyze the in perinatal mortality between and in a perinatal death classification was to the perinatal deaths into relevant for further audit. This classification was discussed and at a with and perinatal The classification system was the Perinatal Perinatal deaths with fetal were in a and the were by time of death during or after gestational and in An on Perinatal and was established by the for for the of in the USA with in Norway and were among the countries to this the first in one of the an audit of perinatal outcomes This was further by at the in the use of an panel for perinatal on the experience It was that such would be a of of perinatal outcome more The was under but several years later it a in the perinatal death a panel from and found that there were more deaths of in than in audit on case notes blinded by a panel of that there was more care and a higher of potentially avoidable deaths among the It was that a should be performed on and that more during should be in In order to the that factors in perinatal care services in perinatal deaths were more common among mothers from the of when with a perinatal audit was performed using a perinatal death classification it was found that the for and deaths for women were and blinded the most common factors for these women were in health care, mothers cesarean of of and When the perinatal death classification was with the the perinatal mortality in was by a of a in and a to in deaths of using a of the type of the higher of perinatal deaths was by a higher mortality from Perinatal deaths in the in were with those in in by using the classification for perinatal deaths A in perinatal mortality was found in the to by more deaths of deaths and deaths due to information not be the clinical general use of and of in the two were by or cesarean section for fetal distress is used in the which might the higher of deaths and that die in the One that was of the same in and the − with perinatal mortality − was deaths of fetuses with These often deaths to be of and level of care. This observation not the that an to is common among these in perinatal mortality have been in various European The was a by the European for the One objective was to whether factors were present in the cases of perinatal deaths collected in and their to the outcome. The was based on of perinatal deaths in the in European from the Norway Scotland and A audit in was by the panel at the of the We decided to those from the perinatal death classification in which care and treatment were most to have a on the fetal and 28 weeks or more of and deaths in after weeks or more of and the first weeks of An audit panel divided into the deaths by means of explicit audit criteria and by a of each case The audit panel midwives, and perinatal from the The elements of care were into and of and and each of a set of had to be in each case. The to the explicit criteria that were for the audit. These criteria were based on evidence from collected by the Cochrane and database in on evidence that was not established by and by criteria by among the panel members as elements of appropriate perinatal care. A of cases of perinatal deaths were more than of the was for of these cases of which deaths were considered potentially These were deaths where identified factors or were have the outcome. The and had the of potentially avoidable deaths and while the had the factors were identified in the care often to care delivery, with to as the most was also a among potentially avoidable Perinatal mortality are much higher in the less of the than in the more the for was perinatal deaths per 1000 births, to between and in the countries It to that perinatal audit a good in poor countries with high mortality however, may not be In a hospital with good it should be but a audit in areas with many home or with only to the information more We have examples of In a the classification was found to be for audit The of by time of death on the that in cases where we that death was potentially this would be by delivery, treatment with before death In this classification has a for deaths of which in the was found to a higher of potentially avoidable deaths than other of The use of gestational age might be by which is more defined and is for in In a at the with a perinatal mortality of per perinatal deaths were by the When with the and there was an of deaths and deaths weeks A audit for these cases was found useful for with several for quality in and and care. A of perinatal deaths was in in One and stillbirths and deaths were autopsy was performed to a in many cases with information from records and hospital of death were according to a system by the group The of deaths under the circumstances was for each case. deaths were most followed by and was the most common observed 21 and of the deaths were and There were deaths among the mothers who a perinatal death. The was that and treatment of and in the newborn should have high in along with care. This is a We do not that in the countries where perinatal audit was from to rules based on evidence during the more than years since In where the first were it was more a from rules by from local of to a set of in the Structure during the Perinatal did a in the We to the point that the and are misinterpreted and in often as followed by of It is, of to state with certainty how many babies might have been of but it would be than from the word of the given in the and in are high with to the of perinatal deaths in the countries and the audit those of deaths most to have a on the outcome. The between countries or are more than the actual The audit is not and is therefore a good point for quality in and in those countries not to However, adverse outcome audit only on cases and may without the on the adverse outcome audit should not but should be with other and quality assessment Perinatal audit as a of quality assessment is well although the as such has not been to scientific the use of perinatal audit to outcomes has to be One the of perinatal audit by areas using audit with areas not using a a an appropriate of this would evidence of the of perinatal audit.