Cost Comparison of Repeat Placental Growth Factor-Based Testing in Women with Suspected Preterm Pre-eclampsia, Compared to Usual Care: The PARROT-2 Trial

Abstract

Objectives

The aim was to evaluate health resource use and cost, and conduct a cost comparison of repeat placental growth factor (PlGF)-based testing for suspected pre-eclampsia, compared with usual care.

Methods

This was a health economic evaluation in women participating in the PARROT-2 trial of repeat revealed PlGF-based testing, compared to usual care, for suspected preterm pre-eclampsia in 22 maternity units in England, Scotland, and Wales, (ISRCTN85912420, 25/11/2019). We conducted a cost comparison analysis, describing health resource use and associated cost per woman, infant, and mother–infant dyad according to randomised allocation to repeat revealed PlGF-based testing or usual care with repeat concealed testing. Additional analysis was stratified according to the initial PlGF-based test result (normal, abnormal, or very abnormal). A post-hoc, within-trial cost-effectiveness analysis was also conducted.

Results

Between December 17, 2019, and September 30, 2022, 1253 participants were randomised. Costs per woman, infant, or mother–infant dyad were similar between the repeat revealed testing group and the group receiving usual care with repeat concealed testing. In the revealed group compared to the usual care group, there were significantly greater costs associated with neonatal admissions for special care (mean difference £1169 95% confidence interval 398–1939), but not neonatal intensive care or high-dependency care. Costs correlated with the initial PlGF-based test result, with the highest costs per mother–infant dyad in the group with a very abnormal initial test result (£38,284 [standard deviation {SD} 35,997]) and lowest costs in the group with a normal initial test result (£13,156 [SD 9568]). Repeat testing was unlikely to be cost-effective, and the cost-effectiveness acceptability curve demonstrated low probability for a range of decision thresholds.

Conclusion

There is no evidence of an overall significant difference in cost associated with a policy of universal, routine repeat testing, compared to usual care, in women receiving an initial PlGF-based test for suspected pre-eclampsia.

Supplementary Information

The online version contains supplementary material available at 10.1007/s41669-025-00629-4.

Key Points for Decision Makers

These novel findings demonstrate there is no evidence of cost-effectiveness for a policy of universal, routine repeat placental growth factor (PlGF)-based testing for all women receiving an initial test for suspected preterm pre-eclampsia.

Repeat testing is cost neutral overall. Repeat testing was associated with an increase in preterm birth, and infant special care costs are significantly increased, but total infant costs are not significantly increased.

By preventing unnecessary repeat testing, this analysis lowers the barriers to single PlGF-based testing for suspected pre-eclampsia, as there is robust evidence for this, and it is recommended in national and international guidance.

This research is applicable to other high-income settings with a similar prevalence of adverse outcomes. Further research is needed in low- and middle-income settings, where the burden of pre-eclampsia mortality and morbidity lies.

Introduction

Pre-eclampsia has been reported to affect 2.8% of pregnancies in the United Kingdom (UK) [1]. Suspected pre-eclampsia affects approximately 10% of pregnant women, accounting for a substantial proportion of emergency presentations to maternity services [2]. As well as the burden of maternal and perinatal adverse outcomes, pre-eclampsia represents substantial economic impact. Unsurprisingly, studies have demonstrated that pregnancies affected by pre-eclampsia cost significantly more than uncomplicated pregnancies. This is largely driven by the burden of infant costs relating to preterm birth, as well as adverse maternal and perinatal outcomes [3, 4]. In 2012, the estimated cost of pre-eclampsia in the first 12 months after delivery was US$2.18 billion in the United States. More recent data estimated medical care costs between $25,437 and $41,790 per mother–infant pair [3, 5].

Diagnosing pre-eclampsia can be challenging, particularly in a background of medical co-morbidities, such as chronic hypertension or chronic kidney disease. Abnormal angiogenic imbalance, with abnormally low maternal concentrations of placental growth factor (PlGF) or high concentrations of soluble fms-like tyrosine kinase 1 (sFlt-1)/PlGF, can be detected prior to the clinical onset of pre-eclampsia, and accurately predicts severe maternal adverse outcomes [6, 7]. Diagnostic test accuracy studies of concealed PlGF-based testing have confirmed PlGF outperforms all other investigations commonly used for suspected pre-eclampsia [8, 9]. Based on these studies, economic modelling predicted that PlGF-based testing would lead to a cost-saving of between £330 and £1032 per woman tested [10–12]. The PARROT-1 trial, a pragmatic, real-world trial of revealed PlGF testing (alongside a clinical management algorithm) compared to usual care with concealed testing, demonstrated that PlGF testing reduced time to diagnosis by 2 days (geometric mean 1·9 vs 4·1 days; time ratio 0·36, 95% confidence interval [CI] 0·15–0·87) and reduced severe maternal adverse outcomes from 5.4 to 3.8% (adjusted odds ratio 0·32, 95% CI 0·11–0·96). [13] A cost-effectiveness analysis of the PARROT trial demonstrated a cost-saving of £149 per woman tested (when including the cost of the test), representing a possible cost-saving of £2,891,196 each year in the NHS in England; this is largely due to reduced outpatient attendances in women with a normal, reassuring PlGF-based test result [10].

The PARROT-2 trial evaluated whether repeat PlGF-based testing reduced perinatal adverse outcomes in women presenting with suspected preterm pre-eclampsia and receiving an initial revealed PlGF-based test (initial PlGF-based testing is recommended in UK National Guidance) [14]. Women were randomised to repeat revealed PlGF-based testing (either QuidelOrtho PlGF testing or Roche sFlt-1/PlGF testing, alongside a clinical management algorithm) or usual care with repeat concealed testing. The results demonstrated no significant difference in the primary composite outcome of stillbirth, early neonatal death, or neonatal unit admission between the revealed testing group (195 [31.2%] of 625 women) and the concealed testing group (174 [27.8%] of 626 women; relative risk 1.21 [95% CI 0.95–1.33]; p = 0.18) [15]. Repeat testing revealed testing reduced time to diagnosis of pre-eclampsia by 4 days (arithmetic mean 19.1 [standard deviation {SD} 20.4] vs 22.5 [SD 22.0]; mean difference − 3.79 days [95% CI − 7.10 to − 0.47], p = 0.0025); however, repeat testing reduced gestational age at delivery (36.7 vs 37.1 weeks’ gestation; − 0.40 [95% CI − 0.68 to − 0.12], p = 0.005) and increased preterm birth before 34 weeks’ gestation (14.4% vs 8.8%; 1.63 [95% CI 1.19–2.24], p = 0.002), as well as increasing Caesarean birth (68.3% vs 59.9%; adjusted risk ratio 1.14 [95% CI 1.05–1.23]; p = 0.002). [15] The stratified analysis of the PARROT-2 trial demonstrated that there is no evidence of clinical benefit of repeating PlGF-based testing if the initial result is abnormal. However, clinically indicated repeat testing can be considered in women with an initial normal test, particularly those who re-present with new symptoms or signs of pre-eclampsia. [15, 16]

The aim of this study was to conduct a within-trial health economic analysis to describe the health resource use and cost associated with implementing repeat PlGF-based testing in maternity services in the NHS in England, Scotland, and Wales compared with usual care.

Methods

This was a planned health economic evaluation of the PARROT-2 trial, from an NHS perspective [14]. The protocol [14], main trial results [15], and a stratified analysis of the main trial have been published. [16]

Trial Processes

The PARROT-2 trial was a randomised controlled trial of repeat revealed PlGF-based testing, compared to repeat concealed PlGF-based testing with usual care, in women with suspected pre-eclampsia between 22 weeks’ gestation and 35 weeks and 6 days’ gestation (ISRCTN 85912420). Ethical approval was from the Cambridge East Research Ethics Committee (ref. 19/EE/0322). Women and birthing people (subsequently referred to as women) were recruited from 22 consultant-led maternity units in England, Scotland, and Wales. Women aged 18 years or over with a singleton live foetus and suspected (but not confirmed) pre-eclampsia were eligible to participate. Suspected pre-eclampsia was defined as at least one of the following: new or worsening hypertension, proteinuria, severe headache or neurological symptoms, right upper quadrant or epigastric pain, suspicion of foetal growth restriction, or abnormal blood test results (hepatic or renal dysfunction, thrombocytopenia, or haemolysis). Women with a documented diagnosis of confirmed pre-eclampsia were not eligible. Women provided written consent and were individually randomised to repeat revealed PlGF-based testing or repeat concealed testing with usual care, with minimisation for maternity unit, indication for testing (hypertension or other), and gestational age at randomisation (22⁺⁰ to 27⁺⁶, 28 to 31⁺⁶, >32⁺⁰). Clinical management was based on the National Institute for Health and Care Excellence guidance on management of hypertension in pregnancy [17]. The trial was pragmatic, to replicate how a strategy of repeat PlGF-based testing for women with suspected preterm pre-eclampsia may be implemented in a real-world setting.

All women had an initial revealed PlGF-based test to guide surveillance, in accordance with national guidance, as part of routine clinical care [17]. Maternity units using either the QuidelOrtho PlGF test or the Roche sFlt-1/PlGF test for an initial PlGF-based test were eligible to be participating sites, as these were the commercial assays recommended in national guidance at the time of trial inception and have comparable test performance [2]. As far as possible, women provided blood samples for tests at routine clinical appointments, to a maximum of four times during their pregnancy. Women randomised to the repeat revealed testing group and their clinicians received revealed PlGF-based test results. A management algorithm was provided, based on national guidance for interpretation of repeat PlGF-based testing results to guide surveillance strategies (Figure S1; see the electronic supplementary material). Test results were classified into three groups based on nationally recommended thresholds: normal (PlGF ≥ 100 pg/ml or sFlt-1/PlGF ≤ 38), abnormal (PlGF 12–99 pg/ml or sFlt-1/PlGF > 38 or < 85), or very abnormal (PlGF < 12 pg/ml or sFlt-1/PlGF ≥ 85) [18]. Women with abnormal or very abnormal repeat PlGF-based test results were recommended to receive heightened surveillance, with additional monitoring, consideration of admission, and ultrasound scan surveillance. Full details are provided in the published protocol. [14]

Outcomes

Participants were followed up until primary postnatal discharge of the mother–infant dyad (or the end of the trial, for two infants with prolonged inpatient stays who were still inpatients at the end of the trial). The primary outcome for the main clinical study was a perinatal composite outcome of stillbirth, early neonatal death, and neonatal unit admission. Secondary outcomes have been published previously and include maternal and perinatal adverse outcomes [15]. Lay advisors with lived experience of pre-eclampsia were involved in selecting trial outcomes. Our trial co-investigator group and the trial steering committee included members with lived experience of pre-eclampsia. The composite perinatal outcome includes adverse consequences that matter to women and clinicians: neonatal unit admission (involving separation of baby from mother, and strongly endorsed by lay contributors) and perinatal death, considered important to recognise and include, even if infrequent in high-income settings.

Resource Use and Costs

Prespecified resource use data were collected. Maternal resource use included outpatient appointments, ultrasound scans, antenatal admission, and hospital admission associated with delivery. Maternal admission days were separated into days on antenatal ward, labour ward, high-dependency unit, intensive therapy unit, and postnatal ward. Infant resource use included neonatal admission days, separated into intensive care, high dependency, special care, transitional care, and normal care days. Neonatal unit admissions involved separation of the mother and infant. Transitional care involved a higher level of care support than normal newborn care, but with the mother as the resident primary care provider. Total maternal and infant costs were calculated separately, and total cost per dyad, both excluding and including the PlGF-based test cost.

Unit costs were obtained from NHS National Schedule of NHS costs 2020/21 [19]. Total cost per woman, per infant, and per mother–infant dyad were then calculated by multiplying unit costs by resource use data. Maternal delivery costs were calculated based on induction/no-induction, mode of delivery (vaginal with or without intervention; planned or unplanned caesarean), and weighting for the number of additional diagnoses, with higher costs if a woman had more than two diagnoses recorded. The cost of antenatal and postnatal inpatient admissions was based on diagnosis and long (> 5 days antenatal and > 6 days postnatal) versus short stay, with an additional cost per bed day after 5 days for long stays. Neonatal stays were costed based on ward type, with a cost of £1809 per bed day for neonatal intensive care, £1242 per bed day for high-dependency care, £892 per bed day for special care (weighting for with or without a carer present), £736 per bed day for normal care, and £773 per bed for transitional care (weighting for all types of transitional care). Women and infants with no record of using a particular healthcare resource were costed as zero.

The cost of PlGF-based tests is taken from diagnostic guidance on PlGF-based testing from the National Institute of Health and Care Excellence [20], with the QuidelOrtho PlGF test and the Roche sFlt-1/PlGF test having different unit costs. It was assumed that all women in the trial had at least one PlGF test. Additional tests were recorded on the trial database in line with the protocol. All costs are in 2020/21 British pounds, with no discounting, as the duration of follow-up was less than 1 year. NHS England costs were used for all sites, although there was one site in Scotland and one site in Wales.

Sample Size

The sample size for the main PARROT-2 trial analysis was 1208 participants. All participants fulfilling eligibility criteria, with outcome data, were included for this analysis.

Statistical Analysis

Healthcare resource use was analysed using linear regression adjusting for gestational age at randomisation and site, with bootstrapped and bias corrected 95% CIs to account for the skewed data. Mean cost and SD were calculated, with effect size and 95% CIs by bootstrapping with bias correction with acceleration, adjusting for gestational age at randomisation and site. Due to the relatively complete data (only one woman lost to follow-up), no assumptions were made regarding missing data. The database was designed so that if a resource use was not recorded, this was because it was not used.

Cost or cost savings associated with repeat PlGF-based testing may vary depending on initial PlGF-based test result (normal, abnormal, or very abnormal) or by type of PlGF test (QuidelOrtho PlGF test or Roche sFlt-1/PlGF ratio), as demonstrated in previous cost-effectiveness studies [10, 21]. Subgroup analysis according to initial PlGF-based test result or type of PlGF-based test used was performed.

For this analysis, it was pre-specified in the health economics analysis plan (published with the protocol and included in the electronic supplementary material) [14] that a mean incremental cost per reduction in the perinatal composite outcome of stillbirth, early neonatal death, and neonatal unit admission would only be calculated if there was a significant reduction in the intervention group. A post-hoc cost-effectiveness analysis was conducted, with a cost-effectiveness analysis curve for the primary outcome reporting the probability for which repeat testing might be cost-effective for a relative 1% decrease in the primary outcome. The primary composite outcome of stillbirth, early neonatal death, or neonatal unit admission is a binary outcome. This model would not run in bootstrap due to the small number of events (neonatal admission was removed as it is already in the cost outcome), so a Poisson model was run. Mean incremental costs per dyad were calculated as the marginal increase of costs using a general linear model with a family gamma and log link, given the significant skew in the cost data. Two-stage bootstrapping was used to calculate the net monetary benefit (NMB) of the marginal decrease in outcome for a range of decision thresholds. NMB was calculated as the marginal percentage increase multiplied by − 100 to change this to a point decrease in stillbirth, early neonatal death, or neonatal admission multiplied by a decision threshold.

Data analyses were done with Stata version 17 (StataCorp, College Station, Texas, USA).

Results

Between December 17, 2019, and September 30, 2022, 1253 women were randomised in the PARROT-2 trial; one woman was randomised in error, not meeting eligibility criteria with a twin pregnancy, and one woman was lost to follow-up. A total of 1251 women were included in this analysis, 625 in the repeat revealed PlGF-based testing group and 626 in the repeat concealed PlGF-based testing group.

For the subgroup analysis with results stratified by initial PlGF-based test result, 716 women (57.2%) had a normal initial result, 335 women (26.8%) had an abnormal initial result, and 200 women (16.0%) had a very abnormal initial test result. For the subgroup analysis with results stratified by test type, 788 women (63.0%) had QuidelOrtho PlGF testing (392 in the revealed and 396 in the concealed groups) and 463 women (37.0%) had Roche sFlt-1/PlGF testing (233 in the revealed and 230 in the concealed groups).

Participant Demographics

Full participant demographics are detailed in Table S1 (see the electronic supplementary material). The groups were broadly balanced for demographics, including age, body mass index, ethnicity, deprivation quintile, and pregnancy history.

Healthcare Resource Use

Descriptive statistics for health resource use are reported in Tables 1 and 2. There was no evidence of a significant difference in maternal health resource use between groups. In the revealed testing group compared to the concealed testing group, there was a significantly higher number of special care unit days (12.3 [SD 11.5] vs 9.2 [SD 11.1] days; mean difference 1.31 days [95% CI 0.47–2.15]). There was no significant difference in days in neonatal intensive care, high-dependency care, or transitional care.

Table 1.

Health resource use for repeat revealed PlGF-based testing, compared to usual care with repeat concealed PlGF-based testing

	Revealed (intervention) N = 625	Concealed (usual care) N = 626	Effect size
	Mean (SD)	Mean (SD)	Difference (95% CI)a
Maternal health resource use
Outpatient attendances	6·70 (5·00)	6·87 (5·18)	− 0·22 (− 0·70 to 0·27)
Ultrasound scans performed	2·64 (2·40)	2·59 (2·17)	0·03 (− 0·19 to 0·25)
Inpatient days	7·63 (6·51)	7·58 (7·41)	0·03 (− 0·73 to 0·79)
Maternal length of stay in following location (number of days, mean [SD])
Antenatal ward	3·32 (5·24)	3·32 (5·54)	− 0·03 (− 0·61 to 0·54)
Labour ward and delivery	0·99 (1·03)	0·99 (0·85)	0·00 (− 0·09 to 0·09)
Obstetric high-dependency unit	0·37 (1·12)	0·33 (1·14)	0·03 (− 0·09 to 0·15)
Non-obstetric high-dependency unit	0·01 (0·17)	0·10 (1·87)	− 0·09 (− 0·26 to 0·08)
Intensive therapy unit	0·02 (0·36)	0·04 (0·45)	− 0·02 (− 0·07 to 0·03)
Postnatal ward	2·92 (2·29)	2·81 (2·21)	0·12 (− 0·14 to 0·37)
Infant health resource use
Any neonatal unit admission	193 (30·88%)	171 (27·32%)
Neonatal inpatient days (for those admitted, mean [SD])
Intensive care	2·81 (7·70)	3·51 (9·08)	− 0·10 (− 0·59 to 0·39)
High dependency	5·45 (12·13)	5·41 (14·70)	0·22 (− 0·58 to 1·02)
Special care	12·30 (11·49)	9·19 (11·06)	1·31 (0·47 to 2·15)*
Total neonatal unit inpatient days	20·55 (23·06)	18·11 (24·12)	1·14 (− 0·14 to 3·00)
Postnatal ward stay
Transitional	0·42 (1·48)	0·79 (7·75)	− 0·36 (− 1·01 to 0·28)
Normal	1·75 (1·97)	1·79 (2·94)	− 0·05 (− 0·31 to 0·22)
Neonatal outcomes
Discharged home	623 (99·68%)	624 (99·68%)
Transferred to another hospital	2 (0·32%)	2 (0·32%)
Died before discharge	0 (0%)	0 0(%)

BCa bias correction with acceleration, CI confidence interval, PlGF placental growth factor, SD standard deviation

^a95% CIs by bootstrapping with BCa adjusting for gestational age at randomisation and site

Costs

Average costs by trial intervention group are reported in Table 2. The cost of the PlGF-based tests was calculated at £60.47 per participant in the concealed group and £176.95 per participant in the revealed group; all trial participants received an initial revealed test, and the mean number of repeat PlGF-based tests was two per participant in the revealed arm (Table S2; see the electronic supplementary material).

Table 2.

Healthcare cost (£) for repeat revealed PlGF-based testing, compared to usual care with repeat concealed PlGF-based testing

	Revealed (intervention) N = 625	Concealed (non-intervention) N = 626	Effect size
	Mean (SD)	Mean (SD)	Difference (95% CI)a
Cost of PlGF-based test	176.95 (95.24)	60.47 (14.31)	115.73 (109.41 to 122.05)
Maternal healthcare cost
Outpatient attendances	1246.05 (930.40)	1273.48 (963.51)	− 40.36 (− 129.34 to 48.61)
Ultrasound scans	449.07 (407.86)	441.02 (369.52)	4.40 (− 31.41 to 40.22)
Antenatal ward	2119.11 (3279.26)	2076.37 (3210.17)	28.39 (− 328.86 to 385.65)
Labour ward and delivery	4903.61 (1210.88)	4790.46 (1245.99)	107.68 (− 12.34 to 238.70)
High-dependency units	676.87 (2145.06)	769.89 (4513.22)	− 97.74 (− 513.83 to 318.35)
Intensive therapy Unit	53.31 (1206.69)	122.42 (1496.98)	− 68.66 (− 217.93 to 80.62)
Postnatal ward	1553.40 (1045.97)	1498.20 (1002.84)	54.47 (− 55.61 to 164.55)
Total maternal cost	11,001.43 (5441.93)	10,971.84 (7132.33)	− 11.81 (− 674.75 to 651.14)
Infant healthcare cost
Intensive care	1568.76 (8072.57)	1733.87 (9020.05)	− 179.51 (− 1059.35 to 700.33)
High dependency	2130.28 (8943.39)	1835.22 (9983.60)	271.09 (− 734.65 to 1276.82)
Special care baby	3438.96 (7642.25)	2250.59 (6294.46)	1168.70 (398.18 to 1939.22)*
Transitional	321.57 (1146.14)	608.77 (5989.93)	− 280.00 (− 710.76 to 150.76)
Normal	1285.94 (1447.04)	1316.81 (2167.43)	− 33.10 (− 219.63 to 153.43)
Total infant cost	8745.51 (19,250.74)	7745.25 (21,262.71)	947.17 (− 1278.52 to 3172.86)
Total maternal and infant cost	19,746.94 (21,425.98)	18,717.10 (23,739.34)	935.36 (− 1474.99 to 3345.72)
Total maternal, infant, and PlGF test cost	19,923.89 (21,421.26)	18,777.57 (23,739.55)	1051.10 (− 1199.53 to 3301.73)
Number of repeat QuidelOrtho PlGF tests per participant	2.18 (1.14)	2.07 (1.18)
Number of repeat Roche sFlt-1/PlGF tests per participant	2.26 (1.16)	2.42 (1.22)

Women and infants with no record of using a resource are costed as 0

BCa bias correction with acceleration, CI confidence interval, sFlt-1 soluble fms-like tyrosine kinase 1, PlGF placental growth factor, SD standard deviation

^a95% CIs by bootstrapping with BCa adjusting for gestational age at randomisation and site

There was no evidence of significant difference in maternal costs between groups; repeat PlGF-based testing was cost neutral. Although there were no significant differences in total infant costs between groups, in the revealed testing group compared to the concealed testing group, there was a greater cost associated with neonatal admissions for special care, with mean cost of £3438.96 (SD 7642.25) in the revealed group versus £2250.59 (SD 6294.46), and a mean difference of £1168.70 (95% CI 398.18–1939.22). There was no significant difference in total infant cost or total mother–infant dyad cost (either excluding or including cost of repeat PlGF-based testing).

Average costs by subgroup stratified by initial PlGF-based test result (normal initial test, abnormal initial test, or very abnormal initial test) are reported in Table 3, and detailed costs for the stratified analysis are presented in Table S3 (see the electronic supplementary material). There were no significant differences in total maternal costs, infant costs, or mother–infant dyad costs between the revealed and concealed groups, when stratifying according to initial PlGF-based test result. Costs were lowest in participants with a normal initial PlGF-based test (£13,156.33 [SD 9568.11] in the revealed group, compared to £13,804.94 [SD 17,677.93] in the concealed group; mean difference − £694.84 [95% CI − 2799.07 to 1409.39]), then participants with an abnormal initial PlGF-based test (£21,055.41 [SD 18,495.32] in the revealed group, compared to £19,005.97 [SD 16,888.93] in the concealed group; mean difference £328.18 [95% CI − 2970.62 to 3626.97]), and highest in participants with a very abnormal initial PlGF-based test (£38,284.47 [SD 35,996.65] in the revealed group, compared to £38,807.61 [SD 41,080.57] in the concealed group; mean difference − £1857.08 [95% CI − 11,022.93 to 7308.78]) (Table 3). Average costs by type of PlGF-based test (QuidelOrtho or Roche) are reported in Table 4. There were no significant differences in total maternal costs, infant costs, or mother–infant dyad costs between the revealed and concealed groups, when stratified according to test type.

Table 3.

Healthcare cost (£) for subgroup analysis by initial PlGF-based test result

	Revealed (intervention) N = 625		Concealed (non-intervention) N = 626		Effect size
	n	Mean (SD)	n	Mean (SD)	Difference (95% CI) a
Maternal costs
Normal first test	350	9560.86 (4855.03)	366	9754.26 (5415.58)	− 121.85 (− 861.18 to 617.47)
Abnormal first test	162	12,040.70 (5731.35)	173	12,233.93 (9801.72)	− 727.18 (− 2454.68 to 1000.33)
Very abnormal first test	113	13,973.44 (5231.37)	87	13,584.44 (6054.28)	178.85 (− 1364.43 to 1722.14)
Infant costs
Normal first test	350	3595.47 (7199.69)	366	4050.68 (15,665.12)	− 572.98 (− 2386.63 to 1240.66)
Abnormal first test	162	9014.71 (15,579.60)	173	6772.04 (12,316.82)	1055.36 (− 1582.39 to 3693.10)
Very abnormal first test	113	24,311.03 (35,026.34)	87	25,225.17 (39,644.54)	− 2035.93 (− 11,051.94 to 7040.45)
Maternal and infant costs
Normal first test	350	13,156.33 (9568.11)	366	13,804.94 (17,677.93)	− 694.84 (− 2799.07 to 1409.39)
Abnormal first test	162	21,055.41 (18,495.32)	171	19,005.97 (16,888.93)	328.18 (− 2970.62 to 3626.97)
Very abnormal first test	113	38,284.47 (35,996.65)	87	38,807.61 (41,080.57)	− 1857.08 (− 11,022.93 to 7308.78)

Normal = PlGF ≥100 pg/ml or sFlt-1/PlGF ≤38; abnormal = PlGF 12–99 pg/ml or sFlt-1/PlGF > 38 or < 85; very abnormal = PlGF < 12 pg/ml or sFlt-1/PlGF ≥ 85

CI confidence interval, sFlt-1 soluble fms-like tyrosine kinase 1, PlGF placental growth factor, SD standard deviation

^aAdjusting for gestational age at randomisation and site

Table 4.

Healthcare cost (£) for subgroup analysis by PlGF test type: QuidelOrtho PlGF compared to Roche sFlt-1/PlGF

	Revealed (intervention) N = 625		Concealed (non-intervention) N = 626		Effect size
	n	Mean (SD)	n	Mean (SD)	Difference (95% CI) a
Cost of test
QuidelOrtho PlGF test	392	138.62 (65.75)	396	49.58 (0)
Roche sFlt-1/PlGF test	233	241.43 (102.26)	230	79.23 (0)
Maternal costs
QuidelOrtho PlGF test	392	11,453.47 (5841.95)	396	11,213.01 (7924.98)	96.86 (− 870.19 to 1063.9)
Roche sFlt-1/PlGF test	233	10,240.93 (4605.32)	230	10,556.62 (5498.35)	− 100.32 (− 947.60 to 746.96)
Infant costs
QuidelOrtho PlGF test	392	8914.01 (20,609.42)	396	7236.84 (21,292.94)	1257.84 (− 1577.61 to 4093.29)
Roche sFlt-1/PlGF test	233	8462.03 (16,756.66)	230	8620.60 (212,228.31)	704.76 (− 2500.26 to 3909.77)
Maternal and infant costs including cost of test
QuidelOrtho PlGF test	392	20,506.1 (22,762.78)	396	18,499.43 (23,980.22)	1442.77 (− 1551.56 to 4437.10)
Roche sFlt-1/PlGF test	233	18,944.36 (18,958.25)	230	19,256.45 (23,363.51)	768.24 (− 2496.73 to 4032.21)

CI confidence interval, sFlt-1 soluble fms-like tyrosine kinase 1, PlGF placental growth factor, SD standard deviation

^aAdjusting for gestational age at randomisation and site

In women with a normal initial result, maternal costs and infant costs were not significantly different in the repeat revealed PlGF-based testing group (£9563.62 [SD 4848.36] for maternal costs and £3591.73 [SD 7189.74] for infant costs) compared to women randomised to concealed testing (£9763.58 [SD 5420.08] for maternal costs and £4057.75 [SD 15,686.04] for infant costs; mean difference for total mother–infant dyad costs £− 710.23 [95% CI − 2810.40 to 1389.84]).

Cost-Effectiveness

Clinical care with repeat revealed PlGF-based testing, implemented with a management algorithm, did not result in fewer maternal or perinatal adverse outcomes. In a post-hoc cost-effectiveness analysis, repeat testing was unlikely to be cost-effective and the cost-effectiveness acceptability curve demonstrated low probability for a range of decision thresholds (Figure S2; see the electronic supplementary material).

Discussion

Main Findings

In this health economic analysis of the PARROT-2 trial, repeat PlGF-based testing for suspected preterm pre-eclampsia was cost-neutral, and there were no significant differences in total cost for the woman, infant, or mother–infant dyad. We demonstrated that for the infant, special care admission days and associated costs were significantly increased. This is an important finding, as studies have demonstrated that the main drivers of economic burden associated with pre-eclampsia are infant healthcare costs [3]. However, given that the main trial results demonstrated a reduction in gestational age at delivery and increased preterm birth before 34 weeks’ gestation in the revealed testing arm of the trial [15], it is reassuring that there was no significant increase in total infant costs, neonatal intensive care days, or high-dependency care days. The main trial results demonstrated an increase in Caesarean birth in the group with repeat revealed testing, but this did not confer significantly increased costs relating to delivery in the repeat revealed testing group, compared to the repeat concealed testing group.

The analysis stratified by initial PlGF-based test result showed that total costs correlate with the initial test result, with the lowest costs in the normal initial test result group and the highest costs in the very abnormal initial test result group. This correlation reflects the distribution of maternal and perinatal adverse events, with higher prevalence in women with a very abnormal initial test result. These results align with other published data, which have shown that women with more severe hypertensive disorders of pregnancy have higher hospitalisation costs than those with milder disease, or uncomplicated pregnancies [3–5]. To our knowledge, this is the first time that correlation between initial PlGF-based test result and total cost per dyad has been reported.

The secondary analysis of the PARROT-2 trial demonstrated that there was no benefit in repeating PlGF-based testing if the initial test was abnormal or very abnormal [22]. However, the secondary analysis suggested there may be benefit in considering repeat PlGF-based testing in women with an initial normal test result presenting again with symptoms and signs of pre-eclampsia, at least 2 weeks after the initial test, to detect changing PlGF-based test category and enabling earlier diagnosis of pre-eclampsia (median time to diagnosis 7 days earlier). Therefore, it is reassuring that total maternal costs and infant costs were not different in women with an initial normal result randomised to repeat revealed PlGF-based testing compared to women randomised to concealed testing. However, no formal significance testing has been undertaken on this subgroup analysis as CIs were wide.

Strengths and Limitations

The PARROT-2 trial was a robustly conducted, individualised, randomised, controlled trial, according to a published protocol; there was a pre-specified, rigorous health economic analysis plan for this within-trial analysis [14]. It was a pragmatic clinical trial, with a low attrition rate. The large multi-centre nature of the trial, across varied maternity units in England, Scotland, and Wales, ensured diverse characteristics of participants, both in terms of demography and disease severity, and this enhances the generalisability of study findings across the NHS. It is a strength of the study that lay advisors with lived experience of pre-eclampsia were involved in the study design, conduct, and dissemination of results.

There was a chance imbalance between groups in the proportion with a very abnormal initial PlGF test in the main trial, which may have impacted the overall health economic analysis. The increase in infant costs for special care in the revealed group may be due to the increase in preterm birth before 34 weeks’ gestation (as described above), but special care costs and preterm birth before 34 weeks’ may have been impacted by the higher prevalence of very abnormal initial PlGF results in the revealed group. Despite achieving the target sample size according to the power calculation for the primary trial outcome, for some of the subgroup analyses, CIs were wide, and the analysis may have been underpowered. Furthermore, we did not collect data on health-related quality of life (mother–infant quality-adjusted life years) and wellbeing scores, impact on loss of earnings, or broader costs, such as those borne by family members. Additionally, we did not collect data on maternal or neonatal costs following primary hospital discharge, as complete ascertainment of these can be complex due to postnatal women not always staying at their primary residence immediately after discharge.

Interpretation

Despite absence of evidence of clinical effectiveness to date, repeat PlGF-based testing is recommended in some local, national, and international consensus guidelines [23]. The results of the PARROT-2 trial and this cost-effectiveness analysis suggest that there is no evidence of clinical or cost-effectiveness for a policy of universal, routine repeat PlGF-based testing. By preventing unnecessary repeat testing, this analysis thus supports cost-effective single testing (as demonstrated in the PARROT-1 trial [13]), reducing barriers to adoption and implementation. The promising emerging advent of point-of-care PlGF-based testing means that we are on the tipping point of a diagnostic adjunct becoming much more readily available and scalable beyond high-income settings.

Conclusion

Our results do not support a policy of universal, routine repeat PlGF-based testing in all women who have had an initial PlGF-based test as part of clinical management for suspected preterm pre-eclampsia. Repeat PlGF-based testing is cost-neutral overall, and there was no evidence of clinical benefit and an increase in preterm birth; costs for special care admission days for the infant are significantly increased, but there was no significant difference in costs for neonatal intensive care or high-dependency care days. From previous studies, there is robust evidence of clinical and cost-effectiveness for a single PlGF-based test for suspected pre-eclampsia.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

The trial was funded by the Jon Moulton Charitable Foundation (80) and Tommy’s (80). Funding for PlGF-based tests was also received from the Guy’s and St Thomas’ NIHR Biomedical Research Centre (BRC) and Roche.

Declarations

Conflict of interest

AHS has received funds from Perkin Elmer (Revvity) and QuidelOrtho for expenses for meetings and has received money from Roche as a consultant on strategy. All remaining authors declare no competing interests.

Data sharing

The dataset will be available to appropriate academic parties on request from the Chief Investigator (LCC) in accordance with the data sharing policies of King’s College London, with input from the co-investigator group where applicable.

Ethics approval

Ethical approval was from the Cambridge East Research Ethics Committee (ref. 19/EE/0322). The study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and later amendments.

Consent to participate

Women provided written, informed consent to participate in the trial.

Author contributions

AHS, LCC, and RMH conceived the study. LW, KED, JM, CB, PTS, KC, MG, AHS, and LCC were involved in securing funding for the study. AH, JS, LW, ZV and LCC coordinated the study conduct and data collection. RMH, AH, and PTS did the study analyses, supervised by LCC. RMH did the health economic analysis. RMH, AH, LW, and LCC wrote the article, with assistance from JS, KED, PTS, JM, CB, KC, ZV, and AHS. All authors approved the final version of the manuscript.

Consent for publication

All participants gave written informed consent for publication of trial results.

Code availability

The code will be available to appropriate academic parties on request from the Chief Investigator (LCC) in accordance with the data sharing policies of King’s College London.