Prospective Structured Perinatal Audit and the Ten Group Classification System: Essential for understanding and improving childbirth

Childbirth is under scrutiny globally. In recent years in some countries there has been significant dissatisfaction with the quality and safety of care afforded to relatively low risk women with a single cephalic pregnancy.^{1, 2} Simultaneously there has been an increase in maternity enquiries investigating clinical practices with concerning findings.^3-5

Improving childbirth, in particular safety, is the responsibility of governments, professional specialist societies and individual health professionals. A cohesive strategy needs to be developed involving women and their families in all aspects of their care. The strategy needs to address areas where improvements are needed, considering the resources available, expectations and cultural contexts.

One of the challenges in modern maternity service delivery is the existence of different philosophies of care between mothers, between healthcare professionals, and between mothers and healthcare professionals. This is due to previous biases which are propagated by confusing evidence. Clear goals in childbirth need to be identified and agreed.

Evidence based information is often presented in a complicated scientific manner and currently relies either on randomised controlled trials or retrospective observational studies both of which are not always helpful to mothers or indeed clinicians.^{6, 7} Consequently, mothers find it difficult to decide on how to use their autonomy appropriately and make the best choices. Much of the information available is either not relevant or not available to them in a simple and understandable manner. Likewise, clinicians struggle to give advice when they themselves do not understand the information. This confusion then continues in relation to accountability and responsibility when mothers choose a certain type of care.

Attempts to improve care have been implemented with varying degrees of success and these need to continue. Most improvements have been related to changing processes and less effort has been directed toward improving the routine analysis of results (events and outcomes). In contrast most creditable organisations invest significantly in the collection of routine information for quality assurance.

The first measure of safety, quality and consistency in any birthing unit is knowing what your results are and this depends on routine data collection. The second measure of safety, quality and consistency is the ability to understand the results, how they interact with each other and how to use them to compare practice with other birthing units and within the same birthing unit over time. This is the purpose of classification, converting data to useful knowledge which can be used to improve quality of care.

It is therefore hard to understand why for childbirth it is at best difficult and at worst impossible to organise measurement of care on a routine basis. This is an issue for both clinicians and the organisations they work for, and which needs acknowledging and addressing. If better oversight on care had taken place in the past, then much dissatisfaction may have been prevented and multiple maternity enquiries not required.

To fully appreciate the purpose of this editorial it is important to understand that this is not about judging care in the first instance. It is about encouraging the use of a common language to analyse care. It is for women, individual birth units and organisations to make the judgement on their care and decide on further action. This will depend on both maternal and professional views and agreed goals. Nevertheless, the process of Prospective Structured Perinatal Audit (PSPA) can only start by clinicians acknowledging and appreciating the importance of knowing their own processes, events and outcomes in a manner which is simple and easily communicable to mothers. At the present time this does not seem to be forthcoming.

Routine data collection in childbirth is a global challenge. It is resource dependent and requires total commitment from everyone in the organisation. Most events and outcomes in childbirth are easy to define but some are not and pragmatic ways around this need to be found. The introduction of electronic health records has generally been disappointing and has not yet solved the problem of routine data collection.⁸ There are two main reasons for this. Firstly, clinicians cannot agree on the clinical definitions or data structure, and secondly electronic health records from a data analysis point of view are poorly structured, cumbersome to enter and retrieve data, and difficult and expensive to improve once they have been implemented.

Raw data are not adequate to assess safety and quality of care. The data need to be organised and classified to understand it. Interpretation of events and outcomes depend on their incidence and the incidence depends on the appropriate denominator being used, which in turn depends on which events or outcomes are being analysed. Only then can the incidences be interpreted logically and appropriately and be relatable to historical results in the same birth unit or indeed compared to other birth units.

When classifying events and outcomes it is important to use a classification system that can incorporate all relevant events and outcomes. Typically in childbirth, changes in one event or outcome might affect others and therefore having all events and outcomes altogether in one classification is helpful.

The classification system used needs to be prospective in structure. That means the different groups of pregnant women in which events and outcomes are being analysed need to be identifiable before starting the childbirth process. Knowing the incidence in clinically relevant prospectively determined groups of pregnant women will allow ‘the intention to treat’ principle to be used with the denominator remaining fixed. These groups of women should be identified early on in pregnancy, preferably at booking either according to the lead clinician involved, public or private, or indeed low- or high-risk pathways of care. The information collected and classified can be used as a tool to inform mothers, clinicians and the organisation before the process of childbirth begins. Importantly also the classification of all pregnant women must be both mutually exclusive and totally inclusive and all pregnant women over the chosen time period of analysis must be included for classification.

This philosophy is becoming known as PSPA. PSPA combines the important principles of a commitment to prospective audit and the structure within which this becomes possible and importantly the ‘the intention to treat’ concept used in randomised controlled trials. The results are then interpreted using the local guidelines for practice described by the individual units. Proponents of this philosophy believe that this will enable learning from each other to take place when the standardised objective reporting of the PSPA will encourage discussion on the more heterogeneous, clinical guidelines found between different birth units.

The Ten Group Classification System (TGCS) was designed to meet the requirements for measuring care and in particular implementing PSPA. Table 1 presents the prospective and retrospective perinatal and non-perinatal information that are important to understand in order to implement PSPA.

The TGCS structure derives from the prospective perinatal information related to pregnancy. The creation of ten groups as a starting point was an arbitrary decision trying to balance enough groups to enable some discernible, clinically relevant information but not too many to be confusing and difficult to remember. Importantly though, the TGCS is meant to act only as the initial, overall structure and represent the starting point for all perinatal audits. The groups can either be subdivided or indeed amalgamated depending on what events and outcomes are being analysed.

The premise for using the TGCS starts with the simple hypothesis that the practice of evidence-based medicine begins and ends with knowing your results. The TGCS is simple and presents a clarity of thought within an agreed structure. It needs discipline to implement and a fundamental belief that all events and outcomes are more clinically relevant after stratification using the ten groups and their subgroups. Also, no perinatal event or outcome should ever be considered in isolation from other events, outcomes and organisational issues.

Much has been written about the TGCS.^9-15 Initially popularised as a classification of caesarean sections, it was always designed to be a perinatal classification and structured so that all events and outcomes short term and long term could be incorporated. At the present time it is being used in over 110 countries with formal scientific publications from at least 90 countries. It has been endorsed by the World Health Organization (WHO), the International Federation of Gynaecology and Obstetrics (FIGO) and the European Board and College of Obstetrics and Gynaecology (EBCOG)^16-18 and many national societies. However, it is still poorly understood and presented, and not currently being used to its full potential for mothers or clinicians. Simple software programs, freely available and written using the TGCS structure are needed, especially in low-income countries or settings without electronic health records, which will make it easier to record clinical practice.¹⁹

Realising the benefits and need of data-driven dialogue, especially among frontline clinicians, the WHO now hosts the WHO Robson Classification Platform.²⁰ Routine data collection in childbirth and the commitment to using this data to improve clinical practice are global challenges, with healthcare providers finding limited tools and support available. This global interactive database is designed to enhance the understanding of caesarean section rates and other perinatal events and outcomes using the TGCS. It supports dialogue between countries and healthcare facilities with a shared interest in optimising the use of caesarean section and improving perinatal outcomes. The platform allows individual maternity units to share their perinatal data using the TGCS, create customised charts and graphs to visualise and compare their data with that of other maternity units, and contact other maternity units to engage in data-driven clinical practice discussions with providers worldwide.

It is critical that the standard TGCS table is used initially. It is important to appreciate how much information can be interpreted from the standard TGCS table which includes the sizes of the groups and the number of caesareans in each group. Interpretation of the TGCS table will direct where further detailed audit takes place. Conversely if there is a significant issue with any of the events and outcomes from an overall perspective, their incidence can be traced back to prospective groups of pregnant women.

Interpretation of TGCS data follows important principles. Before going into any detail about the incidences of events or outcomes in any of the groups, it is important without exception to analyse the group sizes and appreciate the pattern that they form. This is either a standard pattern or often a unique pattern that is associated with specific interpretation of the denominator data. Examples are when the organisation is a tertiary referral unit or in some countries when not all births are included in the overall denominator because some of these are occurring at primary healthcare facilities or at home. This latter reason may be because of a national policy of births outside the main hospitals or because of a lack of access to hospital facilities as in some low-income countries. The biggest contribution that governments could make is at the time of the registration of all births: include classification of the birth using the individual TGCS criteria (previous obstetric history, category of the pregnancy, pathway to birth and the gestational age).

When interpreting either the sizes of the groups or the incidences of events and outcomes, there are only three reasons why there should be variance within the same birth unit over time or between units. The first and most common reason currently, because of the difficulties described earlier in data collection, is data quality. This may be because of clinical definitions or poor data collection and classification. If quality of data collection and classification is confirmed, then the second reason is a difference in significant epidemiological variables. Only if these first two reasons have been validated should the third reason, differences in clinical practice, be considered. The WHO has also developed a manual to help interpret the TGCS table²¹ but all populations are unique and continuously changing.

Other classifications of clinical information can be added within the TGCS structure, such as indications for caesarean sections and inductions¹² all the way through to neonatal outcome and placental pathology. There is no limit to what can be analysed according to the TGCS both in the long or short term and they all can be related back to ‘the intention to treat’ in clinically relevant, identifiable groups of pregnant women. Complaints, medico-legal cases, and cerebral palsy are other examples of outcomes that, providing the raw data is available, can be analysed using the TGCS. Age, body mass index, case-mix and other known significant epidemiological factors can be used as prospective information and refine the TGCS analysis or can be analysed within the individual groups retrospectively.

Perinatal audit is not recognised as an entity, specialist area or even at all useful. Collection of routine quality data is resource dependent, requiring total organisational commitment. At the present time there are no universally accepted classifications, principles or training programs for perinatal audit. A universal language is needed to learn from each other, improve the care given to all women giving birth and bring everyone together.

But implementation of PSPA will not happen by accident. A strategic decision needs to be made by relevant authorities and a commitment made by all clinicians. Training courses are needed to explain the concept and to collect, analyse and interpret. Secondary classifications are needed within the TGCS¹² and professionals need to achieve consensus. Professional bodies and governments need to support the concept and reward implementation of PSPA to assess safety, quality and consistency. Importantly PSPA will complement other sources of evidence-based medicine, in particular randomised controlled trials, when changes of practice have been implemented.

The hope that the clinical care during childbirth could be standardised remains the aim of many clinicians and professional societies. For different reasons this will be difficult to achieve. The biggest contribution that senior clinicians could make at the present time is to embrace different types of care but agree and standardise the way that care is measured.