Presentation trends, characteristics and outcomes for women with early pregnancy bleeding in the emergency department: A 10‐year data linkage study

Women present to the emergency department (ED) with pregnancy complications including bleeding. They seek investigations, treatment and clear discharge and referral pathways.


INTRODUCTION
Per vaginal (PV) bleeding before 20 weeks' gestation is experienced in about 25% of pregnancies. 1 PV bleeding can be life threatening (eg ectopic pregnancy) and result in miscarriage and emotional distress and have a significant bearing on the mindset of women for current and future pregnancies. 1,2Early pregnancy complications are commonly managed in the emergency department (ED).Within the ED, expertise and access to resources can vary between hospitals and the timing of presentation.This can impact patients' access to appropriate care and health service outcomes, including health service costs.Appropriate care is one that is in line with consensus-based or evidence-based guidelines. 3idence supports the implementation of specialised early pregnancy services, including outpatient clinics, for example, Early Pregnancy Assessment Service (EPAS).5][6] Specialised early pregnancy services have shown to improve patient and health service outcomes, including reduced ED length of stay (EDLOS), representations and cost; however, equitable access is limited. 7ere is limited evidence describing ED presentation trends and outcomes across a health district with varied access to early pregnancy services and differing models of ED care and resource availability.A multicentre approach facilitates comparison and provides insight into how health service outcomes are impacted by ED presentation trends and patient characteristics.This study aimed to fill this knowledge gap and to identify presentation trends, characteristics and outcomes for women with early pregnancy bleeding across a regional health district in New South Wales, Australia.

Study design
This study was a retrospective data linkage cohort study of women of reproductive age, between 10 and 50 years, who presented to one of five EDs between 1 January 2011 and 31 December 2020 with early pregnancy complications, namely PV bleeding.

Setting
The health district is located south of Sydney, in New South Wales, Australia, and covers over 6000 km 2 . 8There were five EDs operating during the study period, including a level 6 (approximately 70 000 annual presentations), level 4 (approximately 40 000 annual presentations), level 3 (approximately 30 000 annual presentations) and two level 2 (one rural and one urban) (approximately 6 000 and 15 000 (rural) annual presentations) facilities. 8,9From 2012, the level 6 facility had a specialised EPAS in operation on weekdays. 10

Data collection
Data were extracted from the Illawarra Heath Information Platform, a databank of electronic non-identifiable patient records sourced from the health district. 11All female patients who presented to one of the five EDs between 1 January 2011 and 31 December 2020 and were of reproductive age (10-50 years) 12 with early pregnancy complications, namely PV bleeding, were included.Patients were excluded if they were male, their gender was not specified and they were not of reproductive age.Six data sets were extracted: emergency department, admitted patient, costs, non-admitted patient (NAP), pathology and radiology.ED Health district data on all ED presentations were extracted from the Australian Institute of Health and Welfare. 8The total population and the number of women of reproductive age in the health district were extracted from Census data for the years 2011, 2016 and 2021. 13

Data management and analysis
Data were processed using the SAS statistical software (version 9.4).Data were imported, checked and refined; exclusions were applied; and new variables were derived.Each data set was prepared for linking, and clinical phases of care were created, which constituted a 28-day window of clinical care, commencing from a woman's index ED presentation.Deterministic linking was used to map each woman's pathway starting with a presentation to ED.
The final linked data set was exported from SAS (version 9.4) and analysed using jamovi (version 2.3.12)software.
Descriptive statistics were used to analyse presentations, characteristics and outcomes.Means and SDs are reported for normally distributed data; for others, medians and interquartile ranges (IQR) are reported.Linear regression models were used to identify factors that influenced continuous outcomes, and binary logistic regression models were used to identify predictors for binary outcomes.Purposeful selection of covariates was undertaken.Univariate analysis was used to identify relationships with each predictor and outcome variable.The screening criterion for regression modelling included a significance level of 0.2. 14ltivariable regression modelling was then conducted with all variables identified as significant in screening.A P-value of 0.05 was statistically significant.Figure 1 shows a flowchart summary of the steps used in data collection, processing and analysis.

ETHICAL APPROVALS
Ethical approval was granted by the health district (ISLHD/ LNR/2021-14).A waiver of consent was obtained for the establishment of IHIP from the UOW and ISLHD HREC (2016/306).

Participant characteristics
The mean age of women presented to one of five EDs with early pregnancy bleeding was 29.1 years (SD 7.3).Women aged 35 years or more accounted for 22.5% (n = 3287).Over the study period, the average age for women at the time of presentation increased from 28.5 years (SD 7.5) to 29.3 years (SD 7.1).Of the women who presented to the ED during the study period, 7.7% (n = 761) identified as Aboriginal and/or Torres Strait Islander.
Women were born across 124 countries, with the majority born in Australia (86.1%, n = 8466).The dominant preferred language was English, with 1.9% (n = 187) requesting a translator upon presentation.

Presentations
There were 14 634 ED presentations with early pregnancy complications across the health district over the study period, representing 0.97% of the total ED presentations. 8The number of presentations increased by 19.6% between 2011 and 2020 (Fig. 2).
The distribution of presentations across the days of the week was consistent, and nearly two-thirds (60.7%, n = 8884) occurred during office hours (8am-6pm).Over 90% of women (n = 13 440) arrived by private car/transport.General practitioners (GP) referred only 5.5% of women to the ED (n = 803), with most presentations being self-referral (91.8%, n = 13 376).

Return visits
Of the 14 634 presentations, there were 2198 return visits within 28 days from initial presentation to the ED with early pregnancy bleeding.Almost 20% (n = 414) of these 2198 women attended a different hospital from their initial presentation, and less than half returned within 48 h following the initial visit (47.1%, n = 885).

Emergency department care
Most women (94.2%, n = 13 784) were assigned a triage category of 3 (treatment in 30 min) or 4 (treatment in 60 min). 15Less than 40% of presentations were seen by an ED medical officer (EDMO) within their allocated triage time; in comparison, 87% of all ED presentations were seen within their allocated triage time. 8The median wait time to see EDMO regardless of the triage category was 50 min (IQR 24-100).Almost 7% of presentations were seen by either an NP or a registered nurse (RN), rather than an EDMO, and the median wait time to see a nurse was 23 min (IQR 13-42 min) (Table S2).

Types of investigations
There were 6306 ultrasound tests (43.1% of presentations) and 24 602 pathology tests performed (mean 1.98 (SD 2.47) per presentation).One-third of the presentations received neither ultrasound nor pathology assessment during their presentation to the ED.

ED length of stay
The overall median EDLOS for early pregnancy PV bleeding was 3 h and 17 min (197 min, IQR 125-281).Compared to all ED presentations, the median EDLOS was 2 h and 24 min. 8The mean EDLOS was significantly reduced for women who engaged with NAP services (36 min; 95% CI 30, 43; P < 0.001).Hospital admission increased mean EDLOS by almost 2 h (117 min; 95% CI 108, 125; P < 0.001) for early pregnancy bleeding.Similarly, EDLOS for all presentations and those admitted was extended to 3 h and 23 min 8 (Table S3).Between the facilities of presentation, EDLOS was longest at level 4 ED, with a mean of 218 min (95% CI 203, 234), and the shortest EDLOS observed at urban level 2 ED, with a mean of 33 min (95% CI 80 −14) (Fig. S1).

Referral and departure pathways
Most women were seen, treated and discharged from the ED (n = 10 666, 72.9%), and over half (n = 7542, 51.5%) were referred to their GP.Almost 9% of women (n = 1279) left before being seen by an EDMO.Of the women who did not wait (DNW), 12.1% F I G U R E 1 Flowchart of methods used for data collection, management, linking and analysis in the study of presentation trends, characteristics and outcomes for women who present to the emergency department with early pregnancy bleeding: a 10-year data linkage study in regional New South Wales, Australia (ED, emergency department; EDLOS, ED length of stay; PV, per vaginal).

Data extracted from Illawarra Health Information Platform
Datasets n = 6

Data management and linking
Data imported, checked, and cleaned.Variables derived and reformatted.
Exclusions applied: data restricted to variables of interest.

Non-Admitted patient
• Clinics of interest identified (n = 88) using service unit full name and imported from raw dataset.

Radiology and Pathology
• Orders of interests (radiology n = 30, pathology n = 17) were imported from raw datasets.

Data analysis
Datasets prepared for linking.
Almost 17% (n = 2404) of women who presented were admitted to the hospital.The admission rate remained steady over the study period.An increase in maternal age (OR 1.03; 95% CI 1.03, 1.04; P < 0.001) and presenting on a weekday (OR 1.20; 95% CI 1.07, 1.36; P < 0.002) were positive predictors for admission (Table S5).

Costs
The total cost to the health district was over $16 million for the  S6).

DISCUSSION
There were almost 15 000 ED presentations for early pregnancy bleeding over the 10-year period.The annual presentation rate increased exponentially and was higher than the population growth for the health district. 13Despite limited investigations, health service costs increased over the 10 years.

F I G U R E 2
The number of emergency department presentations for early pregnancy bleeding, total emergency department presentations, total population and number of women of reproductive age in the health district (2011-2020) (ED, emergency department).

Women's characteristics
The mean age for women at the time of presentation increased over the study period.Increasing maternal age was associated with longer EDLOS and a greater likelihood of being admitted and receiving an ultrasound.In Australia, maternal age increased from 30.0 years in 2010 to 30.9 in 2020. 16Older maternal age will in turn increase demand on the ED and health service resources.
Women who identified as Aboriginal and/or Torres Strait Islander were over-represented among those presented with early pregnancy complications at 7.7%, despite comprising 3.5% of the district population. 13This over-representation of Aboriginal and/or Torres Strait Islander women is not new. 4Within the health district, there are NAP services that offer multidisciplinary support to Aboriginal families. 17However, the geographical location and hours of operation may impact access, leaving women to present to the ED.

Return visits
A total of 2198 women (15.02% of early pregnancy bleeding presentations) returned to the ED; this is more than three times the reported 4.9% of return visits within 3 days of a previous presentation for EDs in New South Wales. 18When restricted to the reproductive age, the return rate is estimated to range from 1.0 to 1.3% per annum.Although return visits are associated with lower acuity presentations (triage categories 4 and 5), the return rate among early pregnancy bleeding presentations is disproportionately high and needs further inquiry. 18ss than half returned within the first 48 h; this finding is congruent with other literature. 19Time frames to classify return visits to the ED vary within research, and current health policy stipulates 48 h.A time frame of 28 days was used for this study as suggested by Hutchinson et al. 19 Women who return are considered to be at risk, with inadequate initial care, higher rates of adverse events and increased costs. 20most 20% of return visits were to a different hospital within the health district.Surveillance is often conducted at one site, reducing the capture of return visits. 19By gathering data across a health district, greater accuracy results of and insights into return visits have been obtained.Women who returned to the ED were more likely to have an ultrasound.This finding implies that some return visits may have been planned for investigations.Guidelines suggest that an ultrasound can be delayed for clinically stable patients informed by clinical judgement and joint decision making. 21,22

ED investigations
Approximately half of the presentations did not receive a pathology assessment, and almost 60% did not receive an ultrasound.
Rates of compliance to appropriate care appear low in this study; however, the data do not specify whether investigations were clinically indicated, and there are no data on investigations performed before arrival.Ultrasounds performed within the ED may not have been captured, including bedside or point-of-care ultrasounds.To extract these data, patient notes would need to be reviewed, which was not feasible in this study.

Nurse practitioners
Almost 7% of presentations were seen and treated by an NP and/or RN.In response to increased service pressure, new roles within the ED have emerged.The role of the ED NP is expanding rapidly, with many seeing and treating women experiencing early pregnancy complications. 23Women who saw a nurse rather than an EDMO had a reduced wait time. 23Further inquiry into the role of the ED NP, and their care of women experiencing early pregnancy complications, is needed.

ED length of stay
Both smaller facilities (level 2) were found to have shorter mean EDLOS compared to larger higher-delineated EDs.The level 2 facilities do not have the resources to perform investigations, including ultrasound.Women would need to be transferred to a higher-delineated facility to access ultrasound.Following the initial assessment, women may choose to travel approximately 15-45 min by road to the higher-delineated facilities.The high proportion of return visits across the health district for this cohort may have captured this event.

NAP services
Access to NAP services significantly reduced the odds of DNW and shortened EDLOS.NAP services such as EPAS are the 'gold standard' for providing care to women experiencing early pregnancy bleeding. 24Access to NAP services was inequitable across the health district, with only one EPAS in operation.5][26] EPAS provides comprehensive and streamlined care with a 'women-centred' model of service delivery and should be the benchmark standard of care. 25This study is restricted by retrospective data and does not assess the impact of EPAS on patient satisfaction or timeliness and safety of care of women presented with early pregnancy bleeding.

Departure pathways
In almost 9% of ED presentations, women left before being seen, which is double the national proportion of 3.8%. 27DNW departure pathway is associated with adverse outcomes and increased return visits and admission. 28Research is needed to provide explanations for and insights into why women leave the ED without being assessed after triage.

Costs
Women who presented to the level 2 rural ED had significantly higher costs compared to those presented to the level 6 ED, despite lower odds of receiving investigations, being admitted and shorter EDLOS.ED funding models could explain the disparity in health service costs between rural and metropolitan hospitals.
Healthcare service inequities in rural Australia include lack of resources, isolation from specialist care and limited infrastructure, which affect the provision of services and in turn increase costs and reduce care delivery. 1,29,30th increasing maternal age and number of women presented to the ED with early pregnancy complications, the ED remains an important service provider.By using linked data across a health district, we have determined predictors of health service

ETHICS APPROVAL
Ethical approval was granted from the Regional Health District, approval number ISLHD/LNR/2021-14.
data set was the core data set used to apply inclusion/exclusion criteria and was the base data set for linking.The cost data set was limited to data from 2014, due to the non-availability of data for the previous years.The local health district costing unit provided cost data.Patient-level costs were calculated through PowerPerformance Manager, a clinical costing system, and recorded per the Australian Hospital Patient Costing Standards (version 8.0).Direct, indirect and corporate overhead costs for each clinical phase of care were extracted, and a sum amount was calculated.Direct costs refer to the provision of patient clinical services such as salary costs of nursing and medical staff, diagnostic tests, transport, allied health, procedures and ward supplies.Indirect costs are relevant to the delivery of a service but cannot be easily attached to a particular service; these include catering, cleaning, intrahospital transport and hospital management and administration.Corporate overhead costs are not directly related to patient care processes but are required to support service delivery such as the human resource department and executive management.NAP data set featured information related to NAP services, for example the type of outpatient service accessed. 11NAP services encompass patient care without formal hospital admission; instead, services are accessed through hospital outpatient or community-based clinics and in the home.Modality and access to care can differ in NAP services.
Reproductive age (10-50 years) on presentation to ED • Presenting problem relating to complication in early pregnancy including PV bleeding Data collection Datasets (n = 5) linked to core dataset (ED), using person code and arrival date.dataset (Emergency Department, Non-Admitted and Admitted patient, Costs, Pathology and Radiology datasets combined) •N = 14 634 •Unique persons = 9858 •Variables = 884 Clinical phases of care created (28day window of care starting from ED arrival date).

•
Presentation trends •Demographic characteristics •Emergency Department care •Pathways discharge and referrals Linear and logistic regression models: •Continuous -EDLOS and costs •Categorical -health service use (representation, ED investigations ordered) and outcomes (admission and departure pathways) (n = 155) returned to the ED.An increase in age (OR 0.98; 95% CI 0.98, 0.99; P = 0.001) and attending a NAP clinic (OR 0.78; 95% CI 0.67, 0.91; P < 0.001) reduced the odds of women who DNW (Table care of women who presented to ED with early pregnancy bleeding.The median cost per presentation increased by 330% from 2014 to 2020 (AUD$249.00to 1072.50).The total cost per clinical phase of care was significantly (P < 0.001) more for all triage categories higher than category 4. A category 1 mean cost was $2643 (95% CI AUD$1902, $3385), category 2 was $592 (95% CI AUD$365, $819) and category 3 was $163 (95% CI AUD$86, $240) more.Each pathology order increased the cost by $276 (95% CI AUD$260, $292; P < 0.001), and each ultrasound increased the cost by $437 (95% CI AUD$382, $492; P < 0.001) (Table outcomes such as EDLOS and have drawn comparisons between facilities with differing resource availability and variances in models of care.Findings from this study could guide the development of local resources and a strategy to improve current care pathways.Determination of the appropriateness of care and any unwarranted variation in care is needed to measure the quality of ED care.Limitations to this study involved the use of retrospective research data, which are inherently biased and rely on accurate documentation of patient data.An example of this limitation was the unavailability of data regarding care received before presentation to the ED.This study is illustrative of the population within one health district in Australia; therefore, results may not be generalisable to other geographical areas.Statistical significance may have been reached due to sample size and may not necessarily have clinical significance.Lastly, this study was able to describe only the outcomes captured in the electronic database for early pregnancy bleeding and could not be used to describe the experiences of the ED.ACKNO WLE DGE MENTS The authors acknowledge the Illawarra Health Information Platform research partnership established between the Illawarra Shoalhaven Local Health District and the University of Wollongong.The authors thank Brendan McAlister and Luise Lago for assistance with accessing, managing and linking data.The lead author acknowledges the financial support provided by the Skellern Family Foundation in the form of PhD candidature scholarship.Open access publishing facilitated by The University of Sydney, as part of the Wiley -The University of Sydney agreement via the Council of Australian University Librarians.