Safety, clinical effectiveness, predictive accuracy and cost effectiveness of blood based tests for women with suspected preeclampsia: a health technology assessment
Health technology assessment|
Blood test-based tests may possibly predict a pregnant woman's risk of developing preeclampsia. The tests are now method-evaluated by the Norwegian Institute of Public Health to see if they should be used in clinical practice, among other things to avoid unnecessary hospitalizations.
Two to eight percent of pregnant women are diagnosed with preeclampsia worldwide. Preeclampsia is a potentially life-threatening condition requiring hospital admission and close maternal and fetal monitoring in the second half of pregnancy. Tests based on biomarkers like placental growth factor (PlGF) and soluble fms-like tyrosine kinase-1 (sFlt-1) may predict preeclampsia and lead to better pregnancy outcomes. Such tests are also proposed as a means of identifying women at low risk so that a) unnecessary admissions can be avoided, and b) they can return home reassured. In this health technology assessment, we address safety, effectiveness, predictive accuracy, and cost-effectiveness of these tests among pregnant women with suspected preeclampsia from gestational week 20.
Clinical effectiveness (three included studies)
- PlGF tests may reduce the time to a preeclampsia diagnosis and may reduce the risk of severe maternal adverse outcomes e.g. cerebral haemorrhage.
- We are unable to conclude whether sFlt-1/PlGF or PlGF testing improve neonatal outcomes or not.
Health services utilisation (two included studies)
- sFlt-1/PlGF ratio tests may increase admission rates the first 24 hours after the test, but this result is uncertain
- sFlt-1/PlGF ratio tests were associated with little or no differences in hospital admissions within the first week after the test, and for the remainder of the pregnancy.
Predictive accuracy (four included systematic reviews)
- the predictive accuracy of sFlt-1/PlGF ratio tests is fairly good, and may be useful to “rule out” suspected preeclampsia.
Budget impact of adding blood-based test to clinical practice
We estimate that introducing sFlt-1/PlGF ratio or PlGF test for women with suspected preeclampsia will lead to a direct budget impact of approximately 12.4 million Norwegia kroner annually. Potential health care savings due to earlier diagnosis and reduction in maternal adverse outcomes remain to be explored. Therefore, the current evidence is insufficient to conclude about the comprehensive economic impact of introducing sFlt-1/PlGF or PlGF testing.
Benefit-, resource- and the severity criteria
The tests evaluated may aid to predict the risk of developing preeclampsia. We are, however, not sure how useful the tests are in clinical practice, and there are some serious limitations in the included studies. Using the tests might, or might not, reduce the use of health care resources. Preeclampsia is a very serious condition and early detection might prevent serious outcomes.
World-wide, two to eight percent of pregnant women are diagnosed with preeclampsia. Preeclampsia is defined as new onset hypertension arising after 20 weeks’ gestation accompanied by proteinuria, or other signs of maternal organ dysfunction. Symptoms of preeclampsia are strong headache, visual disturbance, epigastric pain, swelling of hands, face or feet and low output of urine. Preeclampsia may develop rapidly to seizure (eclampsia), HELLP syndrome (haemolysis, elevated liver enzymes and low platelets), stroke, disseminated intravascular coagulation (DIC) and organ failure. Therefore, preeclampsia is regarded as a life-threatening condition that requires hospital admission and close maternal and fetal monitoring.
At antenatal consultation, all pregnant women have their blood pressure assessed and urine tested by a midwife or a general practitioner. Women with high risk of preeclampsia (e.g. women with diabetes mellitus, BMI above 35 and multiple pregnancies) are referred to hospitals for closer follow up. In addition to standard clinical management, tests based on the biomarkers placental growth factor (PlGF) and fms-like tyrosine kinase-1 (sFlt-1) have been developed to help predict preeclampsia from second trimester. PlGF and ultrasound of arterial uterine blood flow are suggested for preeclampsia screening at 11-14 weeks, but that approach is not evaluated in this health technology assessment.
Such tests are proposed to identify pregnant women at increased risk of preeclampsia to reduce the risk of negative outcomes in the women and the neonates, and also as a means of identifying women at low risk so that a) unnecessary admissions can be avoided, and b) they can return home reassured. This should also result in cost-savings for the health services.
The objective of this health technology assessment is to address questions regarding:
1) safety, effectiveness and health services utilisation associated with the use of Elecsys immunoassay sFlt-1/PlGF ratio, Triage PlGF-test, DELFIA Xpress PlGF 1-2-3 test, BRAHMS sFlt-1 Kryptor/BRAHMS PlGF-pluss Kryptor PE ratio or other blood-based tests for predicting suspected preeclampsia in 2nd and 3rd trimester;
2) predictive accuracy of these tests;
3) cost-effectiveness and budgetary consequences of implementation;
and evaluate the intervention against the priority setting criteria applicable in Norway (benefit, resource use and severity).
The study selection criteria for question of safety and effectiveness were:
Population: Pregnant women with suspected preeclampsia in 2nd or 3rd trimester (week 20 to 36 (+6 days))
Intervention: Elecsys immunoassay sFlt-1/PlGF ratio; Triage PlGF-test; DELFIA Xpress PlGF 1-2-3 test; BRAHMS sFlt-1 Kryptor/BRAHMS PlGF-pluss Kryptor PE ratio or other relevant blood-based tests used as add-on to standard clinical assessment for predicting preeclampsia in 2nd and 3rd trimester
Control: Standard clinical assessment, e.g. measuring blood pressure, testing urine for proteinuria, blood tests for haemoglobin (Hb), liver enzymes, bilirubin, headache, oedema, visual disturbance, fetal growth restriction or ultrasound
Outcomes: Mortality, morbidity (maternal and perinatal), health services utilisation (hospital admission, number of days admitted to hospital, number of days admitted at neonatal intensive care unit (NICU), consultations), induction of labour, caesarean, eclampsia, HELLP, gestational age, prematurity
Study design: Randomised controlled trials, non-randomized controlled trials, and controlled before-and-after studies. We included primary studies published after 2002.
The study selection criteria for question of predictive accuracy were:
Population: Pregnant women with suspected preeclampsia in 2nd or 3rd trimester (week 20 to 36 (+6 days))
Index tests: Elecsys immunoassay sFlt-1/PlGF ratio, Triage PlGF-test, DELFIA Xpress PlGF 1-2-3 test, BRAHMS sFlt-1 Kryptor/BRAHMS PlGF-pluss Kryptor PE ratio or other relevant blood-based tests for predicting preeclampsia in 2nd and 3rd trimester. Index tests used in conjunction with standard clinical assessment, or in conjunction with standard clinical assessment excluding quantitative determination of proteinuria.
Comparison: Direct comparison between tests listed as index tests, e.g. diagnostic accuracy of Elecsys immunoassay sFlt-1/PlGF ratio compared to Triage PlGF
Reference: A clinical diagnosis of preeclampsia based on standard clinical assessment: measuring blood pressure, testing urine for proteinuria, blood tests for haemoglobin (Hb), liver enzymes, bilirubin, headache, oedema, visual disturbance, foetal growth restriction and ultrasound with foetal growth assessment
Outcomes: Sensitivity, specificity, predictive values, and likelihood ratios
Study design: Systematic reviews of observational studies published after 2016.
Exclusion criteria: We excluded studies where tests were used for screening purposes.
To answer questions about safety and effectiveness, an information specialist searched for randomised controlled trials and controlled trials in three databases and searched for on-going studies in three registries (October 2019). To address the question of predictive accuracy, we searched for systematic reviews in five databases and for ongoing systematic reviews (June 2019).
Two reviewers independently assessed titles and abstracts of all records from the searches. Potentially relevant records were retrieved and evaluated in full text. Articles meeting our inclusion criteria were included. Disagreement in selection of studies, were solved by consulting one of the other review authors. We assessed risk of bias in controlled studies using the Cochrane risk of bias tool and methodological quality of included systematic reviews using the AMSTAR and QUIPS checklists.
The included studies evaluating the clinical effectiveness of these tests were too different to conduct meta-analyses. We therefore present the findings narratively. As one of the included systematic reviews included meta-analyses on sensitivity and specificity of one of the biomarkers, we report these findings. We assessed the certainty of evidence using the GRADE-approach.
Results of safety, effectiveness and health services utilisation
We included three controlled studies.
Two studies including 1706 women with suspected preeclampsia showed that PlGF tests may reduce the time to preeclampsia diagnosis (median time to pre-eclampsia diagnosis, days time ratio 0.36 (95% CI 0.15-0.87)), and may reduce the risk of severe adverse maternal outcomes (adjusted OR 0.32 (95% CI 0.11-0.96)). We are uncertain whether sFlt-1/PlGF or PlGF testing improve neonatal outcomes.
Health services utilisation
One randomised controlled trial evaluated the effects of adding sFlt-1/PlGF test to standard clinical management among 374 women with suspected preeclampsia. The study showed that the sFlt-1/PlGF ratio was associated with little or no differences in the rate of hospital admission, both during the first week after the test, and for the remainder of the pregnancy. The results may indicate an increased rate of admission to hospital within the first 24 hours after the test: 63 more admissions per 1000 women (from 29 fewer to 183 more). The relative risk for admission within 24 hours was 1.24 (95% CI 0.89- 1.70). This result is imprecise and needs to be interpreted cautiously. One study (1023 women) found that adding PlGF testing to standard clinical management led to a reduction in outpatient visits.
We did not identify studies evaluating the effects of PlGF test on hospital admissions.
Results of predictive accuracy
One systematic review with meta-analyses estimated that sFlt-1/PlGF ratio had a sensitivity around 0.85 (95% CI 0.66-0.94) and a specificity around 0.87 (95% CI 0.76-0.93). These meta-analyses included seven studies with 943 women at high risk (17%) of developing preeclampsia. If we apply sFlt-1/PlGF tests on a group of 1000 women, 25 of the 170 women who will develop preeclampsia will be wrongly classified as negative, and 145 will have a true positive test. Among the 830 women who don’t have preeclampsia, 722 will be identified correctly whereas 108 will receive a false positive result. None of the included accuracy reviews included meta-analyses of PlGF tests. The evidence regarding PlGF tests is based on single studies with wide confidence intervals, so we are uncertain about the predictive accuracy of these tests.
Results of health economic evaluation
We estimated the direct cost of introducing preeclampsia tests as approximately 2 000 Norwegian kroner per tested woman, and the budget impact as approximately 12.4 million Norwegian kroner annually, given that 6000 pregnant women would be tested.
The sFlt-1/PlGF and PlGF tests showed favourable results, but we assessed the certainty of evidence as moderate to low. The results of predictive accuracy correspond with earlier published systematic reviews.
The included controlled trials are from high income countries, but the results cannot be easily transferred to a Norwegian setting. The alternative diagnostic strategies we compare in our analysis are complex and context-specific, depending on multiple factors. Maternity and perinatal care is good in Norway, and therefore, the impact of introducing new preeclampsia test into a routine practice remains uncertain.
Benefit, resource use and severity
Due to the methodological challenges as well as limitations in available evidence we were unable to perform a classic cost-utility analysis and thus quantify the benefit criterion as well as the severity criterion, i.e. calculate the expected QALY gain or the “absolute shortfall”, also measured in QALYs. However, we trust that the description of the condition severity and above findings together with approximate net budget impact will help inform decisions about implementing or not implementing the tests in routine practice.
The tests evaluated in this health technology assessment may aid to predict the risk of developing the condition. We are however not sure how useful the tests are in clinical practice, and there are limitations, as we have shown in the GRADE assessments, in the studies that have evaluated the test in clinical practice. Using the tests might, or might not, reduce the use of health care resources. Preeclampsia is a serious condition, and early diagnosis is important to reduce severe adverse outcomes.
PlGF tests may reduce the time to a preeclampsia diagnosis, and may reduce the risk of severe maternal adverse outcomes. The sFlt-1/PlGF ratio test may be useful to rule out suspected preeclampsia, but seems to be associated with little or no differences in the risk of short or long term admissions. We are uncertain whether the test can improve neonatal outcomes.
The direct cost of introducing preeclampsia tests is about 2 000 Norwegian kroner per tested woman, and the budget impact is approximately 12.4 million Norwegian kroner annually. It remains to be explored whether earlier and correct diagnosis translates into more favourable short- and long-term outcomes for mother and infant, Neither do we have evidence that identification of women at low risk of preeclampsia will lead to reduced use of specialist health care services and thus cost saving in Norwegian settings.