Preeclampsia prediction model using the dipstick test for proteinuria during early gestation

Abstract

Aim: This study aimed to develop a prediction model for preeclampsia (PE) using routinely examined items in early pregnancy, in particular, the dipstick test for proteinuria.

Methods: A total of 9,086 pregnant women recruited in the Tohoku Medical Megabank Project Birth and Three-Generation Cohort Study were included in the present study for analysis. Maternal basic characteristics were obtained by self-report, and blood pressure and dipstick test data were obtained from medical records. The assessed outcome was PE, including superimposed PE. We developed a prediction model without the dipstick test for proteinuria (Model 1) and a model with it (Model 2), and compared the two based on mean area under the receiver operating characteristic curve (mAUROC) using five-fold cross validation.

Results: The mAUROC of Model 1 was 0.769 (95% CI: 0.741 to 0.797) and that of Model 2 was 0.785 (95% CI: 0.758 to 0.812). The difference in mAUROC between the two models was 0.016 (95% CI: 0.004 to 0.028). In Model 2, detection rates were 40%, 49%, and 64% at false-positive detection rates of 5%, 10%, and 20%, respectively.

Conclusions: We improved a prediction model for PE using routine antenatal care items by including the dipstick test for proteinuria.

Introduction

Preeclampsia (PE) affects about 3.4% of all pregnant women¹⁾ and is a major cause of maternal and fetal morbidity.^2,3) Women with PE also have an increased risk of future cardiovascular disease.⁴⁾ Previous studies have suggested that early intervention in high-risk pregnant women is effective in preventing PE,^5,6,7) and several prevention methods for PE, such as daily aspirin, calcium, and L-arginine supplementation, have been reported. However, given the potential adverse effects and economic burden of PE, being able to predict which women are at high risk for the condition is critical for prompt and effective prevention.

A number of studies have developed prediction models for PE using maternal characteristics combined with several biomarkers,⁸⁾ such as uterine artery doppler,^9,10) placental growth factor (PLGF),¹⁰⁾ and maternal serum pregnancy-associated plasma protein-A (PAPP-A).^10,11) While these biomarkers improve the ability to predict PE, screening methods using these biomarkers are not common in clinical practice, partly due to cost-effectiveness and technical difficulties. In 2021, tests for soluble fms-like tyrosine kinase-1/PLGF ratio, which is also an effective biomarker for women at imminent risk,^12,13,14) became covered in Japan by public health insurance after 18 weeks’ gestation and are gradually becoming more widely adopted. Nonetheless, there is a need for PE prediction models which rely only on common clinical parameters, such as blood pressure and proteinuria in early pregnancy. This can be supplemented with more precise and invasive testing using blood samples.

Although the pathogenesis of PE is not well understood, the two-stage model is currently accepted as the standard theory. According to this theory, in the first stage, placental ischemia results from incomplete spiral artery remodeling and, in the second stage, the placenta releases antiangiogenic factors into maternal circulation.¹⁵⁾ These antiangiogenic factors cause systematic endothelial damage resulting in proteinuria, pulmonary edema, and eclampsia.¹⁵⁾

Proteinuria is one of the criteria for diagnosing PE, in addition to de novo hypertension present after 20 weeks’ gestation,^16,17) and may result from the dysfunction of multiple organs, including the kidneys.¹⁸⁾ Thus, proteinuria might be a good biomarker that reflects the clinical course of PE. Dipstick tests are more popular for proteinuria screening at prenatal care centers in Japan than 24 h urine collection, and the World Health Organization recommends that proteinuria should be measured at every antenatal care visit.¹⁹⁾ A prediction model that includes the dipstick test for proteinuria in early pregnancy as a predictor variable can be easily adopted in clinical practice, although no studies to date have used this test in prediction models.

The present study aimed to develop a prediction model for PE using items routinely examined in early pregnancy as predictor variables, including the dipstick test for proteinuria.

Materials and methods

Study population

In July 2013, the Tohoku Medical Megabank Project Birth and Three-Generation Cohort Study (the TMM BirThree Cohort Study)^20,21) began collecting in utero and subsequent exposure and outcome data with the aim of establishing personalized health care and medicine, with approximately 50 obstetric clinics and hospitals in Miyagi Prefecture participating in the recruiting process. Pregnant women and children, fathers, and grandparents were recruited between 2013 and 2017. A total of 22,493 pregnant women were included in the TMM BirThree Cohort Study.

In the present study, pregnant women from the TMM BirThree Cohort Study who were not singletons (n=328) and those who withdrew their consent (n=335) were excluded. A total of 12,744 pregnant women were missing the following data (with duplicates): diagnosis of PE (n=827), dipstick test for proteinuria at antenatal care (n=4,281), blood pressure at antenatal care (n=3,930), prepregnancy body mass index (BMI; n=751), age (n=1,663), history of systematic lupus erythematosus (SLE; n=8,761), diabetes mellitus (DM; n=8,761), maternal family history of PE (n=8,761), method of conception (n=858), parity (n=432), gestational age at previous delivery (n=2,544), and interval between present and previous delivery (n=429). The final study population consisted of 9,086 pregnant women.

The present study was conducted in accordance with the Declaration of Helsinki. Approval for the study was obtained from the Ethics Committee of the Tohoku Medical Megabank Organization (2013-1-103-1). All participants gave written informed consent prior to study inclusion.

Prediction model

Many studies have reported on prediction models for PE,⁸⁾ and these models differed from each other not only in variables but also in the statistical models used. Among these models, the competing risks model has demonstrated successful performance for predicting PE.^22,23,24) In the competing risks model, all pregnant women are assumed to eventually experience PE. However, in most cases, delivery occurs before the development of PE and a survival time analysis, wherein delivery without PE is considered a censored observation, is conducted. One advantage of this model is that the risk of delivery with PE can easily be calculated at any gestational age. It is a well-validated model in European countries^22,23,24) and was therefore also applied to our study population.

Predictor variables

We included variables that were reported to be risk factors for PE²⁵⁾ and common in clinical practice. We also referred to previous studies that used competing risks models to predict PE.^22,23)

Maternal basic characteristics, including age,²⁵⁾ SLE (present or absent),²⁵⁾ DM (present or absent),²⁵⁾ maternal family (mother or sisters), history of PE (present or absent),²⁵⁾ method of conception (in vitro fertilization (IVF) or not),²⁵⁾ parity (nulliparous, parous with previous PE, or parous without previous PE),²⁵⁾ gestational age at previous delivery,^22,23) and interval between present and previous delivery,^22,23) were self-reported.

It is standard practice in Japan for pregnant women to visit antenatal care clinics or hospitals once every 4 weeks until 23 weeks’ gestation, once every 2 weeks from 24–35 weeks’ gestation, and once a week after 36 weeks’ gestation. Therefore, data on blood pressure and the dipstick test for proteinuria (negative, ‘±’, or ≥‘1+’) at 10–13 weeks’ gestation were obtained from medical records, as this gestational age corresponded to the time point when most of our participants completed their first or second antenatal visit. In this process, only the first blood pressure measurement during the visit was used because some clinics and hospitals measured blood pressure only once. Mean arterial pressure (MAP)²⁵⁾ ([systolic blood pressure+(2×diastolic blood pressure)] / 3) and the log₁₀ transformed multiple of the median (log MoM) value of MAP were then calculated. Prepregnancy weight and height were obtained from medical records and used to calculate prepregnancy BMI.²⁵⁾

Outcome

Antenatal medical records were obtained to diagnose hypertensive disorders of pregnancy (HDP): chronic hypertension (CH), gestational hypertension (GH), and PE or PE superimposed on chronic hypertension (SP) based on previous guidelines of the American College of Obstetricians and Gynecologists (ACOG),²⁶⁾ which was standard at the time of participant recruitment.

GH was defined as a systolic blood pressure of 140 mmHg or more, or a diastolic blood pressure of 90 mmHg or more, on at least one visit after 20 weeks’ gestation in a woman with a previously normal blood pressure. PE was defined as GH with proteinuria (≥‘2+’ on dipstick test) on at least one visit after 20 weeks’ gestation. SP was diagnosed when women with CH developed proteinuria after 20 weeks’ gestation. These conditions were automatically diagnosed by computers according to an algorithm and validated by a doctor.

Statistical analysis

Blood pressure in early pregnancy and the dipstick test for proteinuria were compared between those with and without the onset of PE/SP using Welch’s t-test for continuous variables and the chi-square test or Fisher’s exact test for categorical variables.

Parametric survival time analysis that considered delivery without PE/SP as a censored observation was also conducted.²²⁾ Gaussian distribution was assumed for survival curves. Maternal basic characteristics, MAP, and the dipstick test for proteinuria were included in our model as predictor variables. Before developing the prediction model, we examined the relationship between each continuous variable and gestational age at delivery with PE/SP. Continuous variables were grouped, and the effect of each group on gestational age at delivery with PE/SP was then plotted. All continuous variables were centralized to their mean values before developing the model, and gestational age at previous delivery and the interval between present and previous delivery were considered only among parous women.

We compared two models: Model 1 did not include the dipstick test for proteinuria as a predictor variable, while Model 2 did. Delivery with PE/SP was considered the outcome. Five-fold cross-validation was applied, and the area under the receiver operating characteristic curve (AUROC) was calculated. The mean of five AUROCs (mAUROC) was used to determine model performance. The bootstrap method was used to obtain the distribution of each mAUROC and difference in mAUROCs between models, and to calculate 95% confidence intervals (CIs). Detection rates (DRs) at false positive rates (FPRs) of 5%, 10%, and 20% were also calculated.

To investigate the effectiveness of our model in other classifications, such as hypertension after 20 weeks’ gestation, we conducted a secondary analysis. Participants with CH or SP were excluded, and mAUROC and FPR were calculated considering only PE and GH/PE as the outcome. All statistical analyses were performed using R version 3.5.3 (https://www.r-project.org).

Results

In the present study, 336 participants delivered with PE/SP (Table 1). Compared to participants without PE/SP, those with PE/SP tended to be older and nulliparous and have larger BMIs, a medical history of CH and DM, a family history of PE, higher MAPs, and proteinuria (Table 1). The percentages of women with proteinuria at 10–13 weeks’ gestation who went on to develop PE and SP were 31/236 (13%) and 17/100 (17%), respectively.

Table 1. Participant characteristics

	PE/SP (n=336)	Not affected (n=8,750)	P value
Maternal age (years)	32.5 (5.2)	31.9 (4.8)	0.03
Body mass index (kg/m2)	22.9 (3.8)	21.3 (3)	<0.0001
Gestational age (weeks)	38.8 (2)	39.2 (1.6)	<0.0001
Chronic hypertension	100 (29.8)	238 (2.7)	<0.0001
Diabetes mellitus (type 1 or 2)	6 (1.8)	23 (0.3)	<0.0001
Systematic lupus erythematosus	1 (0.3)	8 (0.1)	0.3
Family history of PE	17 (5.1)	242 (2.8)	0.02
Parity
Nulliparous	220 (65.5)	4,662 (53.3)	<0.0001
Parous with previous PE	23 (6.8)	165 (1.9)
Parous with no previous PE	93 (27.7)	3,923 (44.8)
Interval (years)	4.6 (3.2)	3.8 (2.3)	0.01
Gestational age of previous delivery (weeks)	38.4 (2.4)	38.9 (1.8)	0.02
Conception by in vitro fertilization	30 (8.9)	477 (5.5)	0.009
Mean arterial pressure (mmHg)	89.7 (11.4)	79.9 (9.1)	<0.0001
Dipstick test for proteinuria
Negative	226 (67.3)	7,391 (84.5)	<0.0001
±	62 (18.5)	1,027 (11.7)
≥‘1+’	48 (14.3)	332 (3.8)

Data are expressed as mean (standard deviation) for continuous variables and n (%) for categorical variables.

PE, preeclampsia; SP, preeclampsia superimposed on chronic hypertension.

We fitted continuous variables to gestational age at delivery with PE (Figure 1). We assumed a linear relationship for prepregnancy BMI, the interval between present and previous delivery, and log MoM of MAP, and a broken-stick relationship for maternal age and gestational age at previous delivery. The mean gestational age at delivery with PE/SP among the reference population (age 35 years, BMI 21.4 kg/m², no medical history, no family history of PE, nulliparous, spontaneous conception, MAP 80 mmHg, negative for proteinuria) was 46.1 (95% CI: 45.6 to 46.7) weeks (Table 2). The standard deviation (SD) of the survival curve was estimated to be 3.10 weeks.

Figure 1.

Relationship between gestational age at delivery with preeclampsia and each continuous variable.

Effects on time to delivery with preeclampsia of log₁₀ transformed multiple of median of mean arterial pressure (logMoM MAP), maternal age, maternal body mass index (BMI), gestation weeks at previous delivery, and interval between previous and present delivery are shown with a fitted line. We assumed a linear relationship for log MoM of MAP, maternal BMI, and the interval between present and previous delivery, and a broken-stick relationship for maternal age and gestational age of previous delivery.

Table 2. Prediction model for delivery with preeclampsia including dipstick test for proteinuria as a predictor variable

	Coefficient	95% confidence interval
Maternal age - 35 if >35 (years)	−0.11	−0.2 to −0.01
Body mass index - 21.4 (kg/m2)	−0.054	−0.104 to −0.003
Chronic hypertension	−2.3	−2.9 to −1.8
Diabetes mellitus (type 1 or 2)	−1.2	−3.0 to 0.6
Systematic lupus erythematosus	−5.2	−8.2 to −2.3
Family history of PE	−1.1	−1.9 to −0.3
Parity
Nulliparous (reference)	0	—
Parous with previous PE	−3.2	−4.0 to −2.4
Parous with no previous PE	−0.7	−1.3 to −0.1
Interval - 1.8 (years)	−0.12	−0.22 to −0.02
Gestational age of previous delivery - 37 if >37 (weeks)	0.6	0.4 to 0.8
Conception by in vitro fertilization	0.1	−0.6 to 0.8
log MoM of MAP	−16.5	−20.4 to −12.6
Dipstick test for proteinuria
±	−0.8	−1.2 to −0.3
≥‘1+’	−2.2	−2.8 to −1.6
Intercepta	46.1	45.6 to 46.7

PE, preeclampsia; log MoM, log₁₀ transformed multiple of median; MAP, mean arterial pressure

Broken-stick relationship was assumed for maternal age and gestational age at previous delivery in weeks.

All continuous valuables were centralized to their means.

^a  The mean of gestational age at delivery with PE/SP among the reference population.

Figure 2 shows the ROC curve for predicting delivery with PE/SP, which differed from the results of bootstrap simulation due to random sampling. The mAUROCs of Model 1 and Model 2 were 0.774 and 0.789, respectively. Regarding the prediction of delivery with PE/SP by bootstrap simulation, the mAUROC of Model 1 was 0.769 (95% CI: 0.741 to 0.797) and that of Model 2 was 0.785 (95% CI: 0.758 to 0.812), for a difference of 0.016 (95% CI: 0.004 to 0.028) (Table 3). In Model 2, DRs at FPRs of 5%, 10%, and 20% were 40%, 49%, and 64%, respectively. Concerning the prediction of delivery with only PE, the mAUROCs of Model 1 and Model 2 were 0.717 (95% CI: 0.686 to 0.761) and 0.734 (95% CI: 0.705 to 0.776), respectively, and the difference in mAUROCs was significant (0.017 [95% CI: 0.004 to 0.034]) (Table 3). As for the prediction of delivery with GH/PE, the mAUROCs of Model 1 and Model 2 were 0.709 (95% CI: 0.692 to 0.737) and 0.707 (95% CI: 0.691 to 0.737), respectively, with no significant difference between the two models (−0.0016 [95% CI: −0.0023 to 0.0015]) (Table 3).

Figure 2.

Receiver operating characteristic curve for prediction of delivery with preeclampsia.

Receiver operating characteristic curve for prediction of delivery with preeclampsia in the two models. Predictor variables are mean arterial pressure and maternal basic characteristics (age, body mass index, systematic lupus erythematosus, diabetes mellitus, maternal family history of preeclampsia, method of conception, parity, gestational age at previous delivery, and interval between present and previous delivery). Model 1 did not include the dipstick test for proteinuria, while Model 2 included the dipstick test for proteinuria. Mean areas under the receiver operating characteristic curve (mAUROCs) of the two models are 0.774 and 0.789, respectively.

Table 3. Performance of two models for predicting delivery with preeclampsia

Outcome and model		mAUROC (95% CI)	DR at FPR of
			5%	10%	20%
PE/SP (n=336/9,086)
	Model 1	0.769 (0.741 to 0.797)	37	48	62
	Model 2	0.785 (0.758 to 0.812)	40	49	64
	Difference	0.016 (0.004 to 0.028)
Only PE (n=236/8,748)
	Model 1	0.717 (0.686 to 0.761)	21	36	50
	Model 2	0.734 (0.705 to 0.776)	23	35	55
	Difference	0.017 (0.004 to 0.034)
GH/PE (n=610/8,748)
	Model 1	0.709 (0.692 to 0.737)	17	32	49
	Model 2	0.707 (0.691 to 0.737)	18	31	49
	Difference	−0.0016 (−0.0023 to 0.0015)

Model 1 does not include the dipstick test for proteinuria. Model 2 includes the dipstick test for proteinuria.

mAUROC, mean of area under the receiver operating characteristic curve; CI, confidence interval; FPR, false positive rate; DR, detection rate; PE, preeclampsia; SP, preeclampsia superimposed on chronic hypertension; GH, gestational hypertension

Discussion

We developed a prediction model for PE/SP using biomarkers measured during routine antenatal care visits and found that the dipstick test for proteinuria could slightly improve the prediction model. We achieved a better mAUROC in the prediction of delivery with PE/SP than that with only PE or GH/PE, and including the dipstick test for proteinuria in the model did not improve the mAUROC. The competing risks model was successfully applied to an East Asian population, demonstrated by the fact that the mAUROC of our model was better than that reported in studies analyzed in a systematic review,⁸⁾ which reported a performance ranging from 0.61 to 0.88 with various biomarkers. Most of those studies used logistic regression for prediction models.

We excluded CH and SP when conducting the secondary analysis to investigate a prediction model for only PE and GH/PE. While direct comparison of the quality of the models is difficult, it appears that the model for PE/SP might have a higher fitness than models for only PE or GH/PE. Previous studies have also suggested that the pathophysiological background of SP differs from that of PE,^27,28) although this is somewhat controversial.²⁹⁾ Therefore, if a sufficient sample size can be secured, it may be possible to develop a precise prediction model by considering PE and SP as different disorders. On the other hand, the dipstick test for proteinuria improved the mAUROCs of the prediction models for PE/SP and only PE, while the mAUROC of the prediction model for GH/PE did not change after including the dipstick test for proteinuria. This may be explained by the differing criteria for each subtype. For example, the diagnosis of GH does not require proteinuria.

A previous study by Wright et al. used competing risks models to demonstrate the process of developing prediction models for PE.²²⁾ The predictor variables used in that study were similar to those used in the present study, thereby allowing for a comparison of the relationship between each predictor variable and gestational age at delivery with PE. Wright’s study assumed a broken-stick relationship between maternal age and gestational age at delivery with PE, which is consistent with our results. However, that study also assumed a polynomial relationship between the interval between present and previous delivery and gestational age at delivery with PE, whereas the present study assumed a linear relationship because it could fit the relationship well enough with simplicity and validity. We attributed differences in the relationship to differences in ethnicity and/or random noise, especially for the interval between present and previous delivery, which showed relatively high variance. The relationship with gestational age at delivery with PE was unclear.

In our prediction model, IVF was not a clear risk factor for PE. This result differs from that of a previous study³⁰⁾ which suggested that IVF is a risk factor for PE in the Japanese population. We assume that the number of pregnant women who conceived by IVF in our study was insufficient to assess it as a risk factor for PE. Nonetheless, to develop a prediction model with high external validity, we included variables that were well-known risk factors, regardless of their statistical significance.

Our findings have at least two implications. First, our model includes only variables measured at regular antenatal visits. Thus, the model can be easily applied in clinical practice. According to the regression model in Table 2, the mean gestational age of each pregnant woman at delivery with PE can be calculated from usual antenatal care variables. The risk of delivery with PE before a specific gestational age, set at the user’s discretion, can then be estimated from a normal cumulative distribution based on the calculated mean and common SD. The risk at various stages of delivery with PE can be identified at once, so that preventive interventions and the frequency of antenatal care can be optimized. If the risk of preterm birth with PE is high, pregnant women should receive intensive intervention. Previous studies^9,11) have shown better AUROCs using uncommon biomarkers, but since they are not measured routinely in clinical settings, we consider our model superior in terms of practicality. We recommend that every pregnant woman undergo a dipstick test in their early pregnancy, and intensive follow-up should be promoted for high-risk pregnant women throughout their pregnancy.

Second, the inclusion of the dipstick test for proteinuria in our model may help detect severe types of PE. Considering the heterogenous pathogenesis of PE, the prediction model for PE improved using the dipstick test for proteinuria as a predictor variable likely because it allows for the identification of pregnant women with PE who have proteinuria in early pregnancy. According to the two-stage theory, proteinuria results from kidney damage, during the process of systematic endothelial damage. Thus, PE with proteinuria in early pregnancy may indicate a severe type of PE,^31,32) although this view is somewhat controversial.³³⁾ If our model can detect a severe type of PE before its onset, even if only a subtle improvement in mAUROC (0.769 vs. 0.785) is observed, it will be useful in clinical practice.

This study selected predictor variables according to a previous study which used the competing risks model.²²⁾ There likely exist clinically common variables that can improve the prediction model (e.g., the dipstick test for proteinuria), especially for East Asian populations. Therefore, further studies that examine other clinically common variables are warranted.

The strength of the present study is that it was the first to develop a prediction model for PE that includes the dipstick test for proteinuria as a predictor variable. We also demonstrated the effectiveness of competing risks models for East Asian populations by showing a superior mAUROC relative to other previously reported models,⁸⁾ although the external validity of our model using a different East Asian cohort should be further investigated. Importantly, as mentioned above, our model can be easily applied in clinical practice in Japan, which is an important feature for prediction models.

The present study also has some limitations. First, the definition of PE adopted differed from the mainstream definition^16,17) because blood pressure was measured only once and diagnosed automatically by a computer algorithm. We assume that the improvement of mAUROC was overestimated when the dipstick test for proteinuria was included because proteinuria is essential for diagnosing PE in our algorithm according to a previous guideline. Second, the comparability of dipstick tests for proteinuria was not verified due to variability across obstetric clinics and hospitals in the brand of dipstick test used.³⁴⁾ Differences in the brand used across clinics and hospitals could have resulted in systematic error, since the prevalence of PE also differed among participating obstetric clinics and hospitals.

In conclusion, we developed a prediction model for PE using routine antenatal care variables, which improved upon inclusion of the dipstick test for proteinuria.

Acknowledgements

This work was supported by the Japan Agency for Medical Research and Development (AMED), Japan [Grant Nos. JP19gk0110039, JP17km0105001, JP21tm0124005].

HO prepared the original draft. HO, MI and TO carried out the statistical analysis. MI, TO, KM, TO, AN, FU, NI, MK, HM, JS and SK were involved in the acquisition and interpretation of data, as well as the review process. All authors approved the submitted version of the manuscript.

The authors thank all participants, as well as staff members of Tohoku Medical Megabank Organization, Tohoku University, Iwate Tohoku Medical Megabank Organization, and Iwate Medical University.

Conflict of interest

The authors report no conflict of interest.

References

• 1.   Ananth  CV,  Keyes  KM,  Wapner  RJ. Pre-eclampsia rates in the United States, 1980–2010: age-period-cohort analysis. BMJ. 2013; 347 (nov07 15): f6564–f72.
• 2.   Khan  KS,  Wojdyla  D,  Say  L,  Gülmezoglu  AM,  Van Look  PF. WHO analysis of causes of maternal death: a systematic review. Lancet. 2006; 367: 1066–1074.
• 3.   Duley  L. The global impact of pre-eclampsia and eclampsia. Semin Perinatol. 2009; 33: 130–137.
• 4.   Ahmed  R,  Dunford  J,  Mehran  R,  Robson  S,  Kunadian  V. Pre-eclampsia and future cardiovascular risk among women: a review. J Am Coll Cardiol. 2014; 63: 1815–1822.
• 5.   Atallah  A,  Lecarpentier  E,  Goffinet  F,  Doret-Dion  M,  Gaucherand  P,  Tsatsaris  V. Aspirin for Prevention of Preeclampsia. Drugs. 2017; 77: 1819–1831.
• 6.   Roberge  S,  Bujold  E,  Nicolaides  KH. Aspirin for the prevention of preterm and term preeclampsia: systematic review and metaanalysis. Am J Obstet Gynecol. 2018; 218: 287–293.e1.
• 7.   Rolnik  DL,  Wright  D,  Poon  LC, et al. Aspirin versus Placebo in Pregnancies at High Risk for Preterm Preeclampsia. N Engl J Med. 2017; 377: 613–622.
• 8.   De Kat  AC,  Hirst  J,  Woodward  M,  Kennedy  S,  Peters  SA. Prediction models for preeclampsia: A systematic review. Pregnancy Hypertens. 2019; 16: 48–66.
• 9.   Pedroso  MA,  Palmer  KR,  Hodges  RJ,  Costa  FDS,  Rolnik  DL. Uterine Artery Doppler in Screening for Preeclampsia and Fetal Growth Restriction. Rev Bras Ginecol Obstet. 2018; 40: 287–293.
• 10.   Rolnik  DL,  Wright  D,  Poon  LCY, et al. ASPRE trial: performance of screening for preterm pre-eclampsia. Ultrasound Obstet Gynecol. 2017; 50: 492–495.
• 11.   Yücel  B,  Gedikbasi  A,  Dündar  O, et al. The utility of first trimester uterine artery Doppler, placental volume and PAPP-A levels alone and in combination to predict preeclampsia. Pregnancy Hypertens. 2016; 6: 269–273.
• 12.   Ohkuchi  A,  Kondoh  E,  Yamamoto  T, et al. Expert consensus: Indication criteria and screening strategy for preeclampsia using the serum sFlt-1/PlGF ratio at 18–36 weeks of gestation in women at imminent/basal risk of preeclampsia under insurance coverage. Hypertens Res Pregnancy. 2020; 8: 51–56.
• 13.   Ohkuchi  A,  Saito  S,  Yamamoto  T, et al. Short-term prediction of preeclampsia using the sFlt-1/PlGF ratio: a subanalysis of pregnant Japanese women from the PROGNOSIS Asia study. Hypertens Res. 2021; 44: 813–821.
• 14.   Agrawal  S,  Cerdeira  AS,  Redman  C,  Vatish  M. Meta-Analysis and Systematic Review to Assess the Role of Soluble FMS-Like Tyrosine Kinase-1 and Placenta Growth Factor Ratio in Prediction of Preeclampsia: The SaPPPhirE Study. Hypertension. 2018; 71: 306–316.
• 15.   Phipps  E,  Prasanna  D,  Brima  W,  Jim  B. Preeclampsia: Updates in Pathogenesis, Definitions, and Guidelines. Clin J Am Soc Nephrol. 2016; 11: 1102–1113.
• 16.   Brown  MA,  Magee  LA,  Kenny  LC, et al.; International Society for the Study of Hypertension in Pregnancy (ISSHP). The hypertensive disorders of pregnancy: ISSHP classification, diagnosis & management recommendations for international practice. Pregnancy Hypertens. 2018; 13: 291–310.
• 17.  Gestational Hypertension and Preeclampsia: ACOG Practice Bulletin, Number 222. Obstet Gynecol. 2020; 135: e237–e260.
• 18.   Lindheimer  MD,  Kanter  D. Interpreting abnormal proteinuria in pregnancy: the need for a more pathophysiological approach. Obstet Gynecol. 2010; 115: 365–375.
• 19.  WHO. WHO recommendations on antenatal care for a positive pregnancy experience: WHO; 2016.
• 20.   Kuriyama  S,  Yaegashi  N,  Nagami  F, et al. The Tohoku Medical Megabank Project: Design and Mission. J Epidemiol. 2016; 26: 493–511.
• 21.   Kuriyama  S,  Metoki  H,  Kikuya  M, et al.; Tohoku Medical Megabank Project Study Group. Cohort Profile: Tohoku Medical Megabank Project Birth and Three-Generation Cohort Study (TMM BirThree Cohort Study): rationale, progress and perspective. Int J Epidemiol. 2020; 49: 18–19m.
• 22.   Wright  D,  Syngelaki  A,  Akolekar  R,  Poon  LC,  Nicolaides  KH. Competing risks model in screening for preeclampsia by maternal characteristics and medical history. Am J Obstet Gynecol. 2015; 213: 62.e1–e10.
• 23.   O’Gorman  N,  Wright  D,  Syngelaki  A, et al. Competing risks model in screening for preeclampsia by maternal factors and biomarkers at 11–13 weeks gestation. American Journal of Obstetrics and Gynecology. 2016; 214: 103 e1–e12.
• 24.   Wright  D,  Tan  MY,  O’Gorman  N, et al. Predictive performance of the competing risk model in screening for preeclampsia. American Journal of Obstetrics and Gynecology. 2019; 220: 199.e1–e13.
• 25.   Bartsch  E,  Medcalf  KE,  Park  AL,  Ray  JG; High Risk of Pre-eclampsia Identification Group. Clinical risk factors for pre-eclampsia determined in early pregnancy: systematic review and meta-analysis of large cohort studies. BMJ. 2016; 353: i1753.
• 26.  ACOG Committee on Obstetric Practice; American College of Obstetricians and Gynecologists. ACOG practice bulletin. Diagnosis and management of preeclampsia and eclampsia. Number 33, January 2002. Int J Gynaecol Obstet. 2002; 77: 67–75.
• 27.   Costa  RA,  Hoshida  MS,  Alves  EA,  Zugaib  M,  Francisco  RP. Preeclampsia and superimposed preeclampsia: The same disease? The role of angiogenic biomarkers. Hypertens Pregnancy. 2016; 35: 139–149.
• 28.   Tuuli  MG,  Rampersad  R,  Stamilio  D,  Macones  G,  Odibo  AO. Perinatal outcomes in women with preeclampsia and superimposed preeclampsia: do they differ? American Journal of Obstetrics and Gynecology. 2011; 204: 508.e1–e7.
• 29.   Piccoli  GB. What Is Superimposed Preeclampsia (and Does It Actually Exist)? Kidney Int Rep. 2019; 4: 759–762.
• 30.   Nagata  C,  Yang  L,  Yamamoto-Hanada  K, et al.; Japan Environment & Children’s Study Group. Complications and adverse outcomes in pregnancy and childbirth among women who conceived by assisted reproductive technologies: a nationwide birth cohort study of Japan environment and children’s study. BMC Pregnancy Childbirth. 2019; 19: 77.
• 31.   Li  B,  Lin  L,  Yang  H, et al. The value of the 24-h proteinuria in evaluating the severity of preeclampsia and predicting its adverse maternal outcomes. Hypertens Pregnancy. 2018; 37: 118–125.
• 32.   Dong  X,  Gou  W,  Li  C, et al. Proteinuria in preeclampsia: Not essential to diagnosis but related to disease severity and fetal outcomes. Pregnancy Hypertens. 2017; 8: 60–64.
• 33.   Özkara  A,  Kaya  AE,  Başbuğ  A, et al. Proteinuria in preeclampsia: is it important? Ginekol Pol. 2018; 89: 256–261.
• 34.   Vuljanić  D,  Dojder  A,  Špoljarić  V, et al. Analytical verification of 12 most commonly used urine dipsticks in Croatia: comparability, repeatability and accuracy. Biochem Med (Zagreb). 2019; 29: 010708.