Knight, HE; (2018) Using routinely collected data to evaluate the performance and quality of English NHS maternity services. PhD (research paper style) thesis, London School of Hygiene & Tropical Medicine. DOI: https://doi.org/10.17037/PUBS.04650762
Permanent Identifier
Use this Digital Object Identifier when citing or linking to this resource.
Abstract
This ‘publication style’ thesis comprises a collection of research papers, each of which seeks to address a different element of the overall aim: to determine the extent to which electronic data, captured routinely as part of clinical care and hospital administration, can be used to evaluate the performance and quality of English NHS maternity services. These routine data sources present opportunities for research groups to examine whether current practice and outcomes in NHS maternity services meet guidelines and standards, and to guide research and initiatives to improve the quality of maternity care at a regional and national level. However, the difficulty faced by clinicians, managers and service users in interpreting some of the currently available maternity statistics highlights the need to improve the usefulness of the information being produced to evaluate NHS maternity services. The first part of this thesis comprises a review of the advantages and limitations of existing routinely collected data sources for these purposes. The review identifies three key challenges relating to 1) the handling of missing or inconsistent information, 2) the definition of key exposure, outcome and confounding variables relevant to maternity care and 3) adjustment for confounding variables. In the second part, novel techniques are developed to address current weaknesses in the secondary analysis of these data. The findings show that these new methods can be used to derive accurate information on two key data items: 1) the method of delivery and 2) the parity status of women, although misclassification rates are higher for some subgroups of women. This section demonstrates that overall the quality of administrative data is sufficient to support the evaluation of maternity care but that some organisational-level statistics are sensitive to inconsistencies in the data. Consequently, it is recommended that publications of quality indicators should describe how data were prepared and analysed, in order for results to be replicable. In the third part, a series of retrospective cohort studies are described that illustrate how these new methodological techniques can be used to overcome the three challenges identified in the part 1. The first study calculated rates of attempted and successful vaginal birth after caesarean section, which had not previously been done using administrative data at national and provider-level basis (Chapter 6), and found that among women who attempted a trial of labour for their second birth, almost two-thirds successfully achieved a vaginal delivery. A second study evaluated a clinical intervention (induction of labour) designed to prevent rare outcomes such as perinatal mortality which are impractical to investigate by experimental methods (Chapter 7); it found that bringing forward the routine offer of induction of labour from the current recommendation of 41±42 weeks to 40 weeks of gestation in nulliparous women aged >=35 years might reduce overall rates of perinatal death. A third study examined an important health policy question about when staff should be present on the labour ward (Chapter 8) and involved the linkage of administrative, staffing and clinical datasets. The study found no difference in the rate of maternal and neonatal morbidity according to the presence of consultants on the labour ward. A final study examined whether administrative data provided a cost effective way of monitoring perinatal outcomes using a composite indicator of adverse outcomes. The study found that a measure developed in Australia could be adapted to English data, and had good concurrent and predictive ability (Chapter 9). The thesis concludes that hospital administrative datasets, linked with other sources of clinical data where necessary, are a valuable resource for population-based service evaluations. Taken together, the novel techniques developed, validated and applied as part of this programme of work, advance our understanding of the ways in which routinely collected maternity data can and cannot be used to support the evaluation of maternity services. Whilst these data are not perfect and there is certainly a need to improve their completeness and consistency, this research demonstrates that it is possible to develop techniques to identify and manage data errors, and methods to clearly define key exposure, outcome and confounding variables. Together, these allow answers to be found to many potential questions about maternity care.
Item Type | Thesis |
---|---|
Thesis Type | Doctoral |
Thesis Name | PhD (research paper style) |
Contributors | Cromwell, DA and Gurol-urganci, I |
Faculty and Department | Faculty of Public Health and Policy > Dept of Health Services Research and Policy |
Copyright Holders | Hannah Ellin Knight |
Download
Filename: 2018_HSRP_PhD_Knight_H.pdf
Licence: Creative Commons: Attribution-Noncommercial-No Derivative Works 3.0
Download