American Journal of Epidemiology Advance Access published online on June 7, 2007
American Journal of Epidemiology, doi:10.1093/aje/kwm139
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Practice of Epidemiology |
Validity of Preeclampsia-related Diagnoses Recorded in a National Hospital Registry and in a Postpartum Interview of the Women
1 Maternal Nutrition Group, Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark
2 Department of Obstetrics and Gynecology, H:S Hvidovre Hospital, University of Copenhagen, Hvidovre, Denmark
3 Ultrasound Clinic, Juliane Marie Center, H:S Rigshospitalet, Copenhagen, Denmark
Correspondence to Åse K. Klemmensen, Department of Obstetrics and Gynecology, H:S Hvidovre Hospital, University of Copenhagen, Kettegård Allé 30, DK-2650 Hvidovre, Denmark (e-mail: klem{at}dadlnet.dk).
Received for publication June 2, 2005. Accepted for publication June 8, 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In a population-based sample, the authors examined the validity of preeclampsia and related diagnoses recorded in a mandatory Danish national hospital discharge registry and in a standardized telephone interview of women who gave birth between 1998 and 2002. Using a "gold standard" for preeclampsia defined in accordance with the guidelines from the American College of Obstetricians and Gynecologists, the authors reviewed hospital charts of 3,039 women and found that 61of 88 preeclampsia cases (69.3%) and 24 of 55 cases of serious subtypes of preeclampsia (43.6%) were recorded as such by the registry. A total of 21 of 2,951 women without preeclampsia (0.71%) had a preeclampsia diagnosis in the registry. All registrations of serious subtypes of preeclampsia reflected true cases. The positive predictive value of a preeclampsia diagnosis in the registry was 74.4%. Including interview data reduced the sample size to 2,307 women. In this sample, of 62 women with preeclampsia, 45 (72.6%) reported in the interview to have had preeclampsia. Of 2,245 women with no preeclampsia, 31 (1.4%) reported to have had preeclampsia. The positive predictive value of the women's own report on preeclampsia was 59.2%. The authors conclude, for the purpose of etiologic studies, that the registry had acceptable validity, whereas the usefulness of self-reported information may be limited.
interviews; postpartum period; pre-eclampsia; registries; reproducibility of results; women
Abbreviations: ICD-10, International Classification of Diseases, Tenth Revision
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Preeclampsia remains a major problem in modern obstetric practice. It occurs in 2–6 percent of all pregnancies and is responsible for a large proportion of perinatal and maternal morbidity and mortality (1, 2). Preeclampsia has also been associated with diseases appearing later in the woman's or the offspring's life, such as cancer (3, 4), cardiovascular diseases (5–7), and the metabolic syndrome (8).
Identifying the causes and describing the consequences of preeclampsia require large, prospectively designed studies. When the number of study participants gets above a certain limit, however, it may become an insurmountable task to undertake detailed clinical chart reviews for all participants, and other methods for case ascertainment may be required. An option increasingly available in many countries is to make linkage to national or private administrative registries with information on diagnoses on individuals.
Very limited experience exists, however, regarding the validity of preeclampsia-related diagnoses collected in registries (9). In the present paper, we examine the validity of such diagnoses in a mandatory national hospital discharge registry, that is, the Danish National Patient Registry, which for many years has recorded virtually all discharge diagnoses for hospitalizations in Denmark (10). Additionally, we examined the validity of the women's own report on whether they had had preeclampsia or not. The latter information was obtained in a standardized, computer-assisted telephone interview undertaken 6 months after delivery as part of a large national survey (11). An evaluation of the diagnoses based on these two sources was performed against diagnoses based on a detailed review of all relevant clinical charts. The study included 3,084 pregnancies, making it the largest detailed validation study of registry-based and interview-based recordings of preeclampsia of which we are aware.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Sources used
The validation study comprised in principle all those women in the Danish National Birth Cohort who gave birth in the years 1998–2000 at three hospitals: Hvidovre University Hospital, Hilleroed Central Hospital, and the local hospital in Helsingoer. These three hospitals represented the three categories of hospitals in Denmark with obstetric service. If analyses were made separately for each of the three hospitals, they tended to give very similar estimates of sensitivity, specificity, positive predictive value, and kappa (data not shown).
The chart review.
All pregnant women in Denmark receive a "personal record" to carry with them at all visits to the midwife, general practitioner, or hospital. Blood pressure and urine dipstick (checking for proteinuria) measurements are noted at every contact; a copy of this personal record is made at the maternity ward and kept in the hospital archives together with documents related to the hospitalization. Registrations related to diagnoses of preeclampsia and hypertension were performed by one of the authors (Å. K. K.) and entered into a Microsoft Access database (Microsoft Corporation, Redmond, Washington). Each day that a woman had an abnormal blood pressure/urine dipstick measurement in either her chart or personal record, all data on blood pressure/urine test on that date were recorded, including the interval between recordings (figure 1). We refer to this source as the chart review or the gold standard.
|
The National Patient Registry.
The Danish National Patient Registry introduced the International Classification of Diseases, Tenth Revision (ICD-10), in 1994 (10). For validation, we explored the following ICD-10 codes: DO10–DO119 (covering hypertension before pregnancy); DO12–DO122 (pregnancy-induced edema or proteinuria without hypertension); DO13–DO139 (pregnancy-induced hypertension without proteinuria); and DO14–DO159 (the range of levels of preeclampsia) (Web table 1). (The information is described in the first of two supplementary tables; each is referred to as "Web table" in the text and is posted on the Journal's website (http://aje.oxfordjournals.org/).). The Danish National Patient Registry registrations are based on patient contacts, which means that every time a patient is admitted to a hospital as an inpatient or outpatient, registration of an ICD-10 diagnosis is made, although the same diagnosis may have been registered earlier in the pregnancy. We refer to this source as the registry.
The Danish National Birth Cohort.
The Danish National Birth Cohort started enrolment in 1996 and ended in October 2002, with 101,039 women included (11). The women were recruited at the first antenatal visit with their general practitioner and were interviewed four times by telephone: twice during and twice after pregnancy. Approximately 35 percent of all pregnant women in Denmark were included. A total of 70,296 women (70 percent) participated in the first postpartum interview, where information on elevation of blood pressure and preeclampsia was gathered. These questions were phrased as follows: "Did you have elevated blood pressure during pregnancy?"; "Did you have preeclampsia during pregnancy, including elevated blood pressure and urine protein loss?"; and "Did you take antihypertensive medication during pregnancy?" We refer to this source as the interview. Furthermore, information on parity, smoking, body mass index, and height was available from the two interviews during pregnancy.
Method of linkage between sources.
All citizens in Denmark receive a unique 10-digit personal identification number. The personal identification number is used as identification for the chart review, the registry, and the interview.
The study sample
A file with the relevant personal identification numbers was extracted from the Danish National Birth Cohort database on deliveries at the three hospitals for 3 consecutive years giving 4,911 deliveries, corresponding to 2.46 percent (4,911/199,479) of the total Danish deliveries during this 3-year period. At the university hospital, the patient files (n = 1,840) were kept as paper documents, whereas at the other two hospitals, the files (n = 3,071) were partly electronic (n = 1,244). The electronic obstetric chart was introduced in October 1999 at Frederiksborg County, where the two hospitals were located. We decided to use only the electronic charts, as they contained the same information as the paper documents. Our study group thus included 3,084 deliveries (figure 2).
|
Definition of gold standard
We compared the diagnostic criteria used in Denmark, in the United Kingdom, and in the United States in a previous paper (12). For the gold standard, we decided to use the recommended guidelines from the American College of Obstetricians and Gynecologists (figure 1), which are almost identical with the most recent Danish guidelines (13). These criteria are not absolute for eclampsia and were reported absent in 38 percent of the cases with eclampsia (14, 15). Elevation of the blood pressure should return to normal levels before 8 weeks postpartum (16). These criteria are a minimum, and many other symptoms or measurements can be added (17).
Statistical methods
We initially examined the marginal distribution of the diagnoses according to the registry, the interview, and the chart review, respectively. We proceeded with 2 x 2 cross-tables calculating specificity, sensitivity, and positive and negative predictive values for each of the registry and the interview, regarding the chart review as the gold standard. Moreover, Cohen's kappa (18) was estimated as a measure of agreement between the chart review and the diagnosis obtained from the registry/interview. Moreover, we examined whether the validity of preeclampsia diagnosis depended on known risk factors, by stratifying the analyses upon these. Finally, measures of association between preeclampsia diagnoses and known risk factors were assessed by means of logistic regression.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Among the 3,084 women, files were available and had information on the specific pregnancy in 3,039 (98.5 percent). In this group, 2,347 (77.2 percent) participated in the 6-month postpartum interview.
According to the gold standard, hypertensive disorders of pregnancy occurred in 5.86 percent, whereas the registry had less (3.27 percent) and the interview substantially more (12.01 percent) registrations (table 1). The gold standard incidence for preeclampsia (2.90 percent) was similar to that in the registry (2.72 percent), while the interview-based estimate was higher (3.36 percent). The specific subgroups of preeclampsia, not distinguishable in the interview, showed different values for the incidences in the gold standard and the registry, except for the most severe forms, such as the hemolysis elevated liver enzymes and low platelet count and eclampsia, which were nearly identical. The registry severely underestimated the incidence of gestational hypertension (0.55 percent vs. gold standard 2.96 percent), whereas the interview overestimated the incidence (8.64 percent).
|
When ascertainment was based on the registry, the sensitivity for all types of preeclampsia (69.32 percent) was moderate, whereas it was lower for serious preeclampsia (43.64 percent) and hypertensive disorders of pregnancy (48.88 percent) and much lower for gestational hypertension (10.00 percent) (table 2). The positive predictive value ranged from 56.25 to 100 percent and was highest for serious types of preeclampsia. Specificity was above 99.0 percent for all four diagnoses (100 percent for serious preeclampsia). For all diagnoses, kappa was moderate except for gestational hypertension (kappa = 0.16).
|
Implementation of data from the interview (table 3) reduced our sample by 23 percent as mentioned earlier. The sensitivities related to preeclampsia were similar in the registry and the interview (70.97 percent vs. 72.58 percent), but specificity was higher in the registry (table 3). Including a question about medication in the interview substantially increased specificity (from 98.62 to 99.69 percent) but simultaneously reduced sensitivity drastically (from 72.58 to 27.42 percent). Regarding gestational hypertension, the sensitivity was 58.46 percent in the interview compared with 12.31 percent in the registry; on the other hand, the specificity in the interview was poor (92.64 percent) compared with the registry (99.82 percent) (table 3).
|
For the preeclampsia diagnosis, we examined to what extent the validity depended on the mother's parity, height, body mass index, smoking, and previous diseases (table 4); each of these variables was dichotomized for the purpose of the analyses. Validity figures were quite similar across strata defined by these variables (table 4). We also examined to what extent the estimate of association between preeclampsia and these variables, expressed as odds ratios, depended on whether the preeclampsia diagnosis was based on the gold standard, the registry, the interview, or a combination (table 5). For all other variables than smoking, the estimates did not seem to depend much on the method of ascertaining preeclampsia.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Our study addressed a number of different validity issues pertaining to the use of information from a mandatory hospital discharge registry and from a postpartum interview to ascertain cases with preeclampsia for research purposes. The implications of our findings depend on the aim of the research.
If the aim is to determine the disease incidence in a population, some misclassification may be acceptable. This is particularly true if the number of cases wrongly classified to be noncases is similar to the number of noncases wrongly classified to be cases. We found that the incidences of preeclampsia estimated by the registry (2.72 percent) and the interview (3.36 percent) were both close to that based on our chart review (2.90 percent). Moreover, the registry-based incidences of the more serious diagnoses, the hemolysis elevated liver enzymes and low platelet count (0.26 percent) and eclampsia (0.03 percent), were close to the incidences based on the chart review. On the other hand, gestational hypertension was severely underascertained in the registry (0.55 percent vs. 2.96 percent according to the chart review), whereas the interview overestimated the incidence (8.64 percent).
If the aim is to identify as many as possible of the cases that occur in a population (e.g., in order to make a more detailed investigation of all potential preeclampsia cases), one may want to maximize sensitivity. The sensitivity of the registry was 70.97 percent for preeclampsia, whereas the interview had a sensitivity of 72.58 percent. It is noteworthy that, with a strategy based on both sources, requesting that the diagnosis should be present in at least one of the two, the sensitivity did not get above 86 percent, leaving 14 percent of preeclampsia cases unidentified (Web table 2).
We are aware that use of antihypertensive medication is different in North America compared with Europe. Our data on self-reported use of antihypertensive medication may therefore not be directly applicable to a US population.
If the aim is to investigate disease etiologies (be they the consequences of particular exposures for preeclampsia risk or the consequences of preeclampsia for diseases later in life in the mother (5, 6, 8, 19) or offspring (3, 20)), the situation gets more complex, as both sensitivity and specificity of the diagnosis may be critical. A low sensitivity will reduce the number of detected cases, which will lead to a reduced statistical power in the analysis. On the other hand, a low specificity will attenuate the observed odds ratio, compared with the underlying true odds ratio. In addition to the problem of a biased estimate, however, attenuation in the observed odds ratio will also in itself (everything else being equal) lead to a reduced statistical power to detect any true association, which may exist in our population. Numerical examples to illustrate these points are available from the authors.
An important validity concern in any etiologic analysis is whether the misclassification is differential (whether the degree of misclassification in the diagnosis depends on the exposure status). Whereas nondifferential misclassification can lead only to attenuation of the odds ratio, differential misclassification can potentially create biases in any direction (it can both exaggerate and attenuate the observed odds ratio). Differential misclassification is therefore generally less desirable. For parity, prepregnant weight, and previous diseases, no conspicuous differences were detected across strata in measures of positive predictive value, sensitivity, specificity, and kappa.
We also examined to what extent the observed associations between preeclampsia and known risk factors depended on the sources of the diagnosis. Again, for parity, prepregnant weight, and previous disease, it was reassuring that the odds ratios with preeclampsia were generally similar, and in several cases remarkably similar, irrespective of whether the diagnosis was based on the chart review, the registry, or the interview. Smoking behaved differently. Very unexpectedly, and contrary to expectations from earlier studies (21–23), the gold standard-based estimate tended to exhibit an elevated risk of preeclampsia among smokers. However, the registry- and interview-based estimates exhibited the expected (albeit insignificantly) reduced risk among smokers. We have no other explanation for this unexpected finding, than that it may be due to random fluctuations.
Our study had several strengths. To our knowledge, it is the largest of its kind concerning preeclampsia and related diagnoses. According to the clinical chart review, the incidence of preeclampsia was 2.90 percent in our study population, which is very similar to a previously published incidence (2.7 percent) for a Nordic population (24). This suggests that our study sample, when it comes to preeclampsia incidence, is representative for such populations. Our detailed chart review was, unlike some other studies (9), not restricted to only those pregnancies that had been identified as recognized cases according to the registry. This enabled us to estimate specificity and sensitivity, in addition to positive predictive value. Moreover, from the interviews, we had information on the woman's smoking, body mass index, and height, enabling us to consider such factors in our analyses.
Few earlier studies have evaluated the validity of registry-based preeclampsia diagnoses. Geller et al. (9) found a positive predictive value of 54 percent for preeclampsia, whereas it was 79 percent in our registry. A comparison of the Danish National Patient Registry and the Danish Medical Birth Registry (since 1994 an integrated part of the Danish National Patient Registry) was performed for 1,921 births (25). These were restricted to preterm births and therefore were not representative of the general population. Among these, 100 medical records were reviewed with respect to the validity of hypertensive disorders of pregnancy. A sensitivity of 75 percent and a kappa value of 0.7 were reported (25), which were higher values than in our data, possibly because their analysis was based on a preterm population.
The Danish National Patient Registry allows for registrations of diagnoses during multiple hospitalizations throughout pregnancy. This may be unusual for most comparable registries in other countries, which rely more heavily on the hospital admission related to delivery. When we restricted our analysis to take into account only the preeclampsia diagnosis recorded at the delivery hospitalization, we reduced the total number of registered cases by 13, from 82 to 69 cases. This led only to a slight decline in sensitivity and kappa and even to a slight increase in specificity and positive predictive value (sensitivity, specificity, positive predictive value, and kappa changed from 69.32 percent, 99.29 percent, 74.39 percent, and 0.71 (table 2) to 60.23 percent, 99.46 percent, 76.81 percent, and 0.67, respectively).
Despite our finding of a relatively high diagnostic validity of the Danish National Patient Registry, we believe that there is scope for substantial improvement in the quality of these registrations. Klemmensen et al. (12) recently reported a lack of consistency in guidelines regarding preeclampsia and difficulties in handling the criteria among the hospital personnel, who are responsible for the reporting to the registry. It is essential to have comprehensible national guidelines agreed upon by the obstetricians and their adherence to them. National guidelines also need to be similar to the international ones, to enable comparison of studies across borders.
We know of no other study that has evaluated the validity of women's own reports (shortly after pregnancy) on whether or not they had had preeclampsia or gestational hypertension. Not surprisingly, the interview overestimated incidences. Many of these women probably experienced one or more occasions with elevated blood pressure or other symptoms related to preeclampsia and, while they were under observation, got the impression of having preeclampsia and reported it as such. However, including a question about whether or not they received medication for the ailment increased specificity to acceptable levels (99.69 percent) but led also to a drastic reduction in sensitivity (27.42 percent). When preeclampsia is ascertained this way, the odds ratio is likely to be estimated with limited room for bias but with a drastically reduced power.
In conclusion, we found that the Danish National Patient Registry estimated incidences of preeclampsia correctly; this included other serious diseases, such as the hemolysis elevated liver enzymes and low platelet count and eclampsia. On the other hand, we found that gestational hypertension was severely underascertained by this source. For etiologic studies, we found that specificity for preeclampsia was very high in the Danish National Patient Registry.
As actual chart extraction is very time consuming, this might be restricted to those cases where women in the interview reported the disease to be present. This would dramatically reduce the number of charts to be assessed and, in our study, detect 70–85 percent of cases while maintaining a high specificity. Whether this would also be the case in other studies on self-reported diagnoses remains to be shown.
|
ACKNOWLEDGMENTS |
|---|
The study was supported by grants from the Faculty of Health Sciences, University of Copenhagen; the Danish Hospital Foundation for Medical Research in the regions of Copenhagen, the Faeroe Islands, and Greenland; the Copenhagen Medical Society; H:S Direktionens forskningspulje; and the Foundation of Desirée & Niels Yde. Financial support for the Danish National Birth Cohort was obtained from the March of Dimes Birth Defects Foundation, Health Foundation, European Union (QLK1-2000-00083), Danish Medical Research Foundation, Heart Foundation, Danish National Research Foundation, Pharmacy Foundation, Egmont Foundation, and Augustinus Foundation.
The managerial team of the Danish National Birth Cohort consisted of Jørn Olsen (Chair), Mads Melbye, Anne Marie Nybo Andersen, Sjurdur F. Olsen, Thorkild I. A. Sørensen, and Peter Aabye.
Conflict of interest: none declared.
|
NOTES |
|---|
Editor's note: An invited commentary on this article appears on page 000, and the authors' response is published on page 000.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
- Report of the National High Blood Pressure Education Program Working Group on High Blood Pressure in Pregnancy. Am J Obstet Gynecol (2000) 183(suppl):S1–22.[CrossRef][Web of Science][Medline]
- Roberts JM, Speer P. Antioxidant therapy to prevent preeclampsia. Semin Nephrol (2004) 24:557–64.[Web of Science][Medline]
- Ekbom A, Hsieh CC, Lipworth L, et al. Intrauterine environment and breast cancer risk in women: a population-based study. J Natl Cancer Inst (1997) 89:71–6.
[Abstract/Free Full Text] - Vatten LJ, Romundstad PR, Trichopoulos D, et al. Pre-eclampsia in pregnancy and subsequent risk for breast cancer. Br J Cancer (2002) 87:971–3.[CrossRef][Web of Science][Medline]
- Wilson BJ, Watson MS, Prescott GJ, et al. Hypertensive diseases of pregnancy and risk of hypertension and stroke in later life: results from cohort study. BMJ (2003) 326:845–51.
[Abstract/Free Full Text] - Wolf M, Hubel CA, Lam C, et al. Preeclampsia and future cardiovascular disease: potential role of altered angiogenesis and insulin resistance. J Clin Endocrinol Metab (2004) 89:6239–43.
[Abstract/Free Full Text] - Basso O, Wilcox AJ, Weinberg CR, et al. Height and risk of severe pre-eclampsia. A study within the Danish National Birth Cohort. Int J Epidemiol (2004) 33:858–63.
[Abstract/Free Full Text] - Pouta A, Hartikainen AL, Sovio U, et al. Manifestations of metabolic syndrome after hypertensive pregnancy. Hypertension (2004) 43:825–31.
[Abstract/Free Full Text] - Geller SE, Ahmed S, Brown ML, et al. International classification of diseases–9th revision coding for preeclampsia: how accurate is it? Am J Obstet Gynecol (2004) 190:1629–33.[CrossRef][Web of Science][Medline]
- Sorensen HT, Sabroe S, Olsen J. A framework for evaluation of secondary data sources for epidemiological research. Int J Epidemiol (1996) 25:435–42.
[Abstract/Free Full Text] - Olsen J, Melbye M, Olsen SF, et al. The Danish National Birth Cohort—its background, structure and aim. Scand J Public Health (2001) 29:300–7.[CrossRef][Web of Science][Medline]
- Klemmensen AK, Olsen SF, Wengel CM, et al. Diagnostic criteria and reporting procedures for pre-eclampsia: a national survey among obstetrical departments in Denmark. Eur J Obstet Gynecol Reprod Biol (2005) 123:41–5.[CrossRef][Web of Science][Medline]
- Johansen M, Kjaergaard N, Legarth J, et al. Hypertension in pregnancy. In: Sandbjerg guidelines (2001) (In Danish)Copenhagen, Denmark: Danish Society for Neuroscience.
- Douglas KA, Redman CW. Eclampsia in the United Kingdom. BMJ (1994) 309:1395–400.
[Abstract/Free Full Text] - Sibai B, Dekker G, Kupferminc M. Pre-eclampsia. Lancet (2005) 365:785–99.[Web of Science][Medline]
- ACOG practice bulletin. Diagnosis and management of preeclampsia and eclampsia. Int J Gynaecol Obstet (2002) 77:67–75.[CrossRef][Medline]
- Roberts JM, Pearson GD, Cutler JA, et al. Summary of the NHLBI Working Group on Research on Hypertension during Pregnancy. Hypertens Pregnancy (2003) 22:109–27.[CrossRef][Web of Science][Medline]
- Altman DG. Some common problems in medical research. In: Practical statistics for medical research (1991) London, United Kingdom: Chapman & Hall/CRC. 396–439.
- Irgens HU, Reisaeter L, Irgens LM, et al. Long term mortality of mothers and fathers after pre-eclampsia: population based cohort study. BMJ (2001) 323:1213–17.
[Abstract/Free Full Text] - Barker DJ, Gluckman PD, Godfrey KM, et al. Fetal nutrition and cardiovascular disease in adult life. Lancet (1993) 341:938–41.[CrossRef][Web of Science][Medline]
- Cnattingius S, Mills JL, Yuen J, et al. The paradoxical effect of smoking in preeclamptic pregnancies: smoking reduces the incidence but increases the rates of perinatal mortality, abruptio placentae, and intrauterine growth restriction. Am J Obstet Gynecol (1997) 177:156–61.[CrossRef][Web of Science][Medline]
- Newman MG, Lindsay MK, Graves W. Cigarette smoking and pre-eclampsia: their association and effects on clinical outcomes. J Matern Fetal Med (2001) 10:166–70.[Medline]
- Xiong X, Wang FL, Davidge ST, et al. Maternal smoking and preeclampsia. J Reprod Med (2000) 45:727–32.[Web of Science][Medline]
- Clausen T, Slott M, Solvoll K, et al. High intake of energy, sucrose, and polyunsaturated fatty acids is associated with increased risk of preeclampsia. Am J Obstet Gynecol (2001) 185:451–8.[CrossRef][Web of Science][Medline]
- Kristensen J, Langhoff-Roos J, Skovgaard LT, et al. Validation of the Danish birth registration. J Clin Epidemiol (1996) 49:893–7.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  V. Ehrenstein, K. J. Rothman, L. Pedersen, E. E. Hatch, and H. T. Sorensen Pregnancy-associated Hypertensive Disorders and Adult Cognitive Function Among Danish Conscripts Am. J. Epidemiol., October 15, 2009; 170(8): 1025 - 1031. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. R. Stein, D. A. Savitz, and M. Dougan Serum Levels of Perfluorooctanoic Acid and Perfluorooctane Sulfonate and Pregnancy Outcome Am. J. Epidemiol., October 1, 2009; 170(7): 837 - 846. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Catov, E. A. Nohr, L. M. Bodnar, V. K. Knudson, S. F. Olsen, and J. Olsen Association of Periconceptional Multivitamin Use With Reduced Risk of Preeclampsia Among Normal-Weight Women in the Danish National Birth Cohort Am. J. Epidemiol., June 1, 2009; 169(11): 1304 - 1311. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. L. Brantsaeter, M. Haugen, S. O. Samuelsen, H. Torjusen, L. Trogstad, J. Alexander, P. Magnus, and H. M. Meltzer A Dietary Pattern Characterized by High Intake of Vegetables, Fruits, and Vegetable Oils Is Associated with Reduced Risk of Preeclampsia in Nulliparous Pregnant Norwegian Women J. Nutr., June 1, 2009; 139(6): 1162 - 1168. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Lykke, J. Langhoff-Roos, B. M. Sibai, E. F. Funai, E. W. Triche, and M. J. Paidas Hypertensive Pregnancy Disorders and Subsequent Cardiovascular Morbidity and Type 2 Diabetes Mellitus in the Mother Hypertension, June 1, 2009; 53(6): 944 - 951. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. S. Wu, Y. Sun, M. Vestergaard, J. Christensen, R. B. Ness, C. L. Haggerty, and J. Olsen Preeclampsia and Risk for Epilepsy in Offspring Pediatrics, November 1, 2008; 122(5): 1072 - 1078. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Pinborg, O. Lidegaard, N. la Cour Freiesleben, and A. N. Andersen Vanishing twins: a predictor of small-for-gestational age in IVF singletons Hum. Reprod., October 1, 2007; 22(10): 2707 - 2714. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. M. Callaghan Invited Commentary: Identifying Women with Hypertension during Pregnancy--Is High Specificity Sufficient? Am. J. Epidemiol., July 15, 2007; 166(2): 125 - 127. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Osterdal, A. Klemmensen, A Tabor, and S. Olsen Osterdal et al. Respond to "Identifying Women with Hypertension during Pregnancy" Am. J. Epidemiol., July 15, 2007; 166(2): 128 - 129. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||