Maternal recall of a history of early-onset preeclampsia, late-onset preeclampsia, or gestational hypertension: a validation study

ABSTRACT

Objective: This study assessed women’s ability to recall different types of hypertensive disorders of pregnancy because of its disproportionate cardiovascular risk later in life.

Methods: Participants were 5–10 years post-partum with a history of early-onset-, late-onset preeclampsia, or gestational hypertension. Recall was assessed by questionnaire and compared to medical records.

Results: Questionnaire-based recall of early-onset preeclampsia had high sensitivity and specificity, late-onset preeclampsia poor sensitivity and high specificity and gestational hypertension had very poor sensitivity and high specificity.

Conclusion: Early-onset preeclampsia can accurately be assessed using a simple questionnaire. Recall of late-onset preeclampsia and gestational hypertension was not accurate.

KEYWORDS:

• Cardiovascular risk
• hypertensive disorders of pregnancy
• prevention
• recall

Introduction

One-third of all women worldwide die from cardiovascular disease (CVD) (1). Despite the high prevalence of CVD among women, diagnosis is often delayed due to differences in clinical presentation and underlying pathophysiology compared to the male standard (2–5). The development of preeclampsia is believed to be one of the earliest female-specific factors indicating increased cardiovascular risk (6). It has become evident that women who have had hypertensive disorders of pregnancy (HDP), including preeclampsia, face an increased cardiovascular risk later in life (7–9). The risk is related to the severity and time of onset of the HDP (9–11). This is evident already shortly after pregnancy when the prevalence of hypertension differs between women with different severities of HDP in history; 45% of women who have had early-onset preeclampsia (delivery before 37 weeks of gestation), 40% who had gestational hypertension (irrespective of gestational age) and 25% of women who had late-onset preeclampsia (delivery from 37 weeks of gestation) (9). Later in life, it is demonstrated by the increased risk of cardiovascular death; twofold in women who have had gestational hypertension or preeclampsia and nine-fold in women who have had early-onset preeclampsia (10–12).

Early identification of women at risk creates an opportunity for early intervention and preventive strategies. Later in life, when a general practitioner, internist, or cardiologist want to estimate the cardiovascular risk of a female patient, it could be helpful to know if she has had an HDP earlier in life. However, obtaining medical records can be difficult years after pregnancy. Therefore, the physician often has to rely on women’s ability to accurately recall their obstetric history. Several studies have focused on maternal recall of HDP as a proxy for having an accurate medical history on record (13,14). Recall accuracy increased as the severity of HDP increased, but it was noted that the accuracy may be influenced by other markers of disease severity like birth weight and preterm delivery. In addition, no distinction has been made between early- and late-onset preeclampsia while this may be important in risk assessment.

Therefore, this study focused on the ability of a questionnaire to assess different severities of HDP. Women with early-onset preeclampsia, late-onset preeclampsia, and gestational hypertension were approached as three distinct groups because of their disproportionate risk of CVD later in life. Women with an uncomplicated term pregnancy in history and women with preterm birth in history participated as controls. The latter to study the influence of preterm delivery in women with early-onset preeclampsia in history.

Materials and methods

Study population

Three severities of HDP were investigated: early-onset preeclampsia, late-onset preeclampsia, and gestational hypertension. Women with early-onset preeclampsia delivered prior to 37 weeks of gestation and women with late-onset preeclampsia delivered from 37 weeks of gestation. Both groups were diagnosed with new-onset hypertension (≥140/90 mmHg) and proteinuria (≥300 mg/24 h or 1+ dipstick test) using the ISSHP 2001 criteria; the criteria used at the time of diagnosis of the study population (15). Women with gestational hypertension suffered from new-onset hypertension (≥140/90 mmHg) without proteinuria and no limitation on gestational age (15). There were two control groups of which the first consisted of women with uncomplicated term pregnancies (gestational age ≥37 weeks). The second control group consisted of women who delivered preterm (gestational age <37 weeks). The additional control group of preterm birth was included to test if recall of women with early-onset preeclampsia was influenced by preterm birth. Both the uncomplicated and preterm pregnancies had to be normotensive and in the absence of proteinuria. Exclusion criteria for all participants were: diagnosis of cardiovascular disease before index pregnancy, multiple pregnancies or congenital abnormalities in index pregnancy, and currently pregnant. All women delivered in the VU University Medical Center between 2006 and 2011, time post index pregnancy was 5 to 10 years.

Procedure

Electronic health records of women who delivered between 2006 and 2011 in the VU University Medical Center were screened consecutively (by AB, ABF, and LJZ), and all eligible women were sent a questionnaire to their home address. Not responding women were sent a second questionnaire and were called once to verify the home address, not to complete the questionnaire. An informed consent form was attached to the questionnaire and had to be completed for participation. Approval for the study was obtained from the medical ethics committee of the Amsterdam University Medical Center, location VU University Medical Center in Amsterdam (protocol 2015.502).

Survey

The questionnaire was based on a questionnaire used by Diehl et al. for a similar study on recall of preeclampsia 20 years post index pregnancy (Figure 1) (16). Questions were added to distinguish between early-onset preeclampsia, late-onset preeclampsia, and gestational hypertension.

Figure 1. Translated questionnaire

Recall of preeclampsia was confirmed when either the direct term “preeclampsia” and/or the descriptive terms “high blood pressure” and “proteinuria” were reported by the participant. Next to that, women with preeclampsia were asked to report their gestational age at delivery to distinguish between early-onset and late-onset disease. Gestational hypertension was verified when women could indicate the presence of high blood pressure during index pregnancy with the absence of preeclampsia, toxemia, eclampsia, or proteinuria.

Sample size

The sample size of the case groups, to assess sensitivity, was based on the 95% confidence interval found by Diehl et al. (16). A width of 16% with a power of 80% and a two-sided α of 0.05 resulted in 107 women in each case group. For specificity, they found a confidence interval of 8%. Conservatively, 10% was chosen with a power of 80% and a two-sided α of 0.05 which resulted in 94 women for each control group.

Statistical analyses

The variables maternal age at the questionnaire, time period after index pregnancy, parity, gestational age at delivery of index pregnancy, intensive care admission, and perinatal death were extracted from medical records. Other variables were derived from completed questionnaires. Education was categorized into three groups: low (no education or primary school), intermediate (secondary school or upper secondary education), high (higher vocational education or university). Respondents and non-respondents were compared by maternal age, time period after index pregnancy, parity, gestational age of index pregnancy, maternal intensive care admission, and perinatal death using unpaired Student’s t-test, Mann–Whitney U test, and Chi-square test when appropriate.

The three case groups were each separately compared to the control group of women with a history of uncomplicated pregnancy. Women with a history of early-onset preeclampsia were additionally compared to women with a history of preterm birth. Comparisons were made for maternal age, time period after index pregnancy, current hypertension, use of antihypertensive medication or a family history of hypertension, educational level, variables related to index pregnancy including parity, gestational age, admission for HDP, maternal intensive care admission, perinatal death, and HDP in a previous pregnancy using unpaired Student’s t-test, Mann–Whitney U test, and Chi-square test when appropriate. A Bonferroni correction was applied and p < 0.008 was considered statistically significant.

The primary outcome was the validity of the questionnaire to recall early-onset preeclampsia, late-onset preeclampsia, and gestational hypertension. Medical records on index pregnancy served as the “gold standard.” The sensitivity of each group expressed the percentage of women who recalled correctly the condition, among those who had the condition in the medical record. Specificity represented the percentage of women who indicated correctly that they did not have the condition, among the women in the four groups who did not have the condition in the medical record.

As a secondary outcome, we studied the influence of predictors including maternal age, time post index pregnancy, educational level, current hypertension, current use of antihypertensive medication, a family history of hypertension, parity at index pregnancy, gestational age at delivery, admission for HDP, and HDP in a previous pregnancy, on correct recall of different severities of HDP by using logistic regression analysis. In this analysis, all women with a history of HDP were tested as one group and were subdivided between correct and incorrect recall. All potential determinants were tested in univariate analysis and significant variables (p < 0.20) were tested in multivariate analysis.

In other analyses, a p < 0.05 was considered statistically significant. Data were analyzed using SPSS 22 software (IBM Corp, Armonk, NY).

Results

For this study, women with a history of early-onset preeclampsia, late-onset preeclampsia, gestational hypertension, uncomplicated pregnancy, or preterm birth were approached, resulting in 1175 potential participants. The overall response rate was 41% (Table 1). Respondents were older (41.4 vs. 40.4 years, p = 0.005) and the gestational age of the index pregnancy was shorter (37.6 vs. 38.9 weeks, p < 0.001) than non-respondents. Response rate was higher in women with previous early-onset- (58.8%) and late-onset-preeclampsia (53.7%, both p < 0.001) (Table 2). The calculated sample size was reached for all groups except for the group of late-onset preeclampsia. Women with previous early-onset preeclampsia and gestational hypertension had more often current hypertension (20.6% and 15.9%) compared to control groups (both 1%, p < 0.001 and p < 0.001). Women with both early-onset preeclampsia, late-onset preeclampsia, and gestational hypertension reported more often use of anti-hypertensive medication (15.9%, 12.1%, 16.8%, respectively) compared to control groups (1% and 0%, p < 0.001, p = 0.002, p < 0.001). Women with early-onset preeclampsia in history were more often primiparous in index pregnancy (80.4%) than women with an uncomplicated pregnancy in history (56.7%, p < 0.001). Perinatal death occurred more in women with previous early-onset preeclampsia (10.3%) than in women with uncomplicated pregnancy (0%, p = 0.001) but it was comparable to women with preterm birth in their index pregnancy (6.1%). As expected, women with a history of early-onset preeclampsia, late-onset preeclampsia, or gestational hypertension had a shorter gestational age at delivery (31.9, 38.4, and 39.1 weeks) compared to women with an uncomplicated pregnancy in history (40.3 weeks, p < 0.001, p < 0.001, p < 0.001). Gestational age at delivery was comparable between the groups of early-onset preeclampsia and preterm birth (31.0 weeks). All case groups were more often admitted for HDP (84.1%, 68.2%, and 50.5% vs. 1% in both control groups, all p < 0.001), and had more often HDP in a previous pregnancy (15.0%, 27.3%, and 24.3% vs. 0% in both control groups, all p < 0.001). Notable is that the majority of women in all groups were highly educated (61.6−75.8%).

Table 1. Characteristics of respondents compared to non-respondents

		Respondents n = 483		Non-respondents n = 692		p-value
Age, years	mean ± SD	41.4	± 5.2	40.4	± 5.7	0.005
Time post index pregnancy, years	median [IQR]	7.6	[6.4−9.3]	7.3	[6.2−9.5]	0.410
Primiparous	n (%)	316	(65.4)	425	(61.4)	0.177
Gestational age, weeks	median [IQR]	37.6	[32.1−39.7]	38.9	[35.4−40.3]	<0.001
Maternal IC admission	n (%)	2	(0.4)	5	(0.7)	0.707
Perinatal death	n (%)	17	(3.5)	25	(3.6)	1.000

IC ,intensive care.

Table 2. Baseline characteristics

		Case groups						Control groups
		Preeclampsia <37 weeks n = 107		Preeclampsia ≥37 weeks n = 66		Gestational hypertension n = 107		Uncomplicated pregnancy n = 104		Preterm birth n = 99
Response rate	%	58.8* **		53.7*		36.5		29.4		44.6
Age, years	mean ± SD	40.8	± 5.7	42.0	± 4.5	41.9	± 4.9	42.7	± 5.4	39.6	± 4.7
Time post index pregnancy, years	median [IQR]	7.9	[6.6−9.4]**	7.8	[6.7−9.5]	7.6	[6.4−9.2]	8.6	[6.0−9.6]	6.8	[6.3−9.3]
Current hypertension ^†	n (%)	22	(20.6)* **	6	(9.1)	17	(15.9)*	1	(1.0)	1	(1.0)
Use of antihypertensive medication ^†	n (%)	17	(15.9)* **	8	(12.1)*	18	(16.8)*	1	(1.0)	0	−
Family history of hypertension ^†	n (%)	60	(56.1)	37	(56.9)	60	(55.6)	49	(47.1)	44	(44.4)
Education ^†
Low	n (%)	12	(11.2)	6	(9.1)	6	(5.6)	6	(5.8)	8	(8.1)
Intermediate	n (%)	28	(26.2)	10	(15.2)	23	(21.5)	20	(19.2)	29	(29.3)
High	n (%)	67	(62.6)	50	(75.8)	77	(72.6)	78	(75.0)	61	(61.6)
Index pregnancy
Primiparous	n (%)	86	(80.4)*	44	(66.7)	63	(58.9)	59	(56.7)	64	(64.6)
Gestational age, weeks	median [IQR]	31.9	[29.1−35.0]*	38.4	[37.7−39.7]*	39.1	[37.9−40.0]*	40.3	[39.6−40.9]	31.0	[28.4−35.3]
Admission for HDP ^†	n (%)	90	(84.1)* **	45	(68.2)*	54	(50.5)*	1	(1.0)	1	(1.0)
Maternal IC admission	n (%)	1	(0.9)	0	−	0	−	1	(1.0)	0	−
Perinatal death	n (%)	11	(10.3)*	0	−	0	−	0	−	6	(6.1)
HDP in previous pregnancy ^†	n (%)	16	(15.0)* **	18	(27.3)*	26	(24.3)*	0	−	0	−

HDP, hypertensive disorders of pregnancy; IC, intensive care.

*After Bonferroni correction significantly different compared to women with a history of uncomplicated pregnancy

**After Bonferroni correction significantly different compared to women with a history of preterm birth.

^†Self-reported.

Table 3 shows the recall of early-onset preeclampsia, late-onset preeclampsia, and gestational hypertension of all participants subdivided by diagnosis by the medical record. The validity of the questionnaire to recall early-onset preeclampsia, late-onset preeclampsia, and gestational hypertension is presented in Table 4. The sensitivity of recalling early-onset preeclampsia was 98.1% and specificity 94.2%. Late-onset preeclampsia was recalled with a sensitivity of 68.2% and specificity of 91.1%. Recall of gestational hypertension had a sensitivity of 29.0% and specificity of 97.9%. When early- and late-onset preeclampsia were considered as one group, sensitivity was 94.2% and specificity was 85.2%. Among women with preeclampsia in index pregnancy, 156/173 (90%) responded positively to the direct term “preeclampsia” and 124/173 (72%) on the combination of descriptive terms “hypertension” and “proteinuria.” Of the 173 women with a history of preeclampsia, 39 only reported the direct term and 7 women only reported the combination of descriptive terms. Ten women did neither recall the direct nor the descriptive terms. In logistic regression analysis, higher education was of negative influence on correct recall. Primiparity, lower gestational age at delivery, and admission for HDP positively influenced correct maternal recall (Table 5). In multivariate analysis, only gestational age remained significant, and lower gestational age was associated with correct maternal recall.

Table 3. Recall of different severities of HDP of all participants subdivided by diagnosis by the medical record

	Medical record
Recall by questionnaire	Preeclampsia <37 weeks n = 107	Preeclampsia ≥37 weeks n = 66	Gestational hypertension n = 107	Uncomplicated pregnancy n = 104	Preterm birth n = 99
Preeclampsia <37 weeks	105	12	9	0	1
Preeclampsia ≥37 weeks	1	45	36	0	0
Gestational hypertension	1	6	31	0	1

Table 4. Performance of maternal recall questionnaire for different severities of HDP

		Sensitivity	95% CI	Specificity	95% CI
Preeclampsia <37 weeks	%	98.1	(93.4−99.8)	94.2	(91.3−96.3)
Preeclampsia ≥37 weeks	%	68.2	(55.6−79.1)	91.1	(88.0−93.7)
Gestational hypertension	%	29.0	(20.6−38.5)	97.9	(95.9−99.1)

CI, confidence interval.

Table 5. The influence of factors for the correct recall of different severities of HDP

	Univariate analysis		Multivariate analysis
	Crude OR (95%CI)		Adjusted OR (95%CI)
At the time of the questionnaire
Age	1.00	(0.95−1.05)
Time post index pregnancy	1.02	(0.86−1.22)
Education	0.61	(0.40−0.94)	0.64	(0.40−1.03)
Hypertension ^†	1.10	(0.56−2.16)
Using antihypertensive medication ^†	1.29	(0.64−2.61)
Family history of hypertension ^†	1.24	(0.76−2.03)
At the time of the index pregnancy
Primiparous	1.93	(1.15−3.23)	1.49	(0.82−2.69)
Gestational age	0.71	(0.64−0.79)	0.73	(0.66−0.82)
Admission for HDP ^†	2.62	(1.56−4.41)	1.78	(0.98−3.23)
HDP in previous pregnancy ^†	0.69	(0.37−1.28)

^†Self-reported data.

Discussion

Main findings

The ability of a questionnaire to assess different severities of HDP varied. The recall of women with a history of early-onset preeclampsia was highly accurate. Specificity for soliciting early-onset preeclampsia was also high in the control group of women with preterm birth in history. In women with late-onset preeclampsia, recall was poor and in women with gestational hypertension very poor.

Interpretation

The sensitivity and specificity of the maternal recall are important in research to assess associations between HDP and cardiovascular risk. The sensitivity and specificity of the questionnaire to solicit early-onset preeclampsia were highly accurate and even more accurate than research performed in women with severe preeclampsia (sensitivity 83.0−92.0%, specificity 100%) (17,18). Severe preeclampsia included hemolysis, elevated liver enzymes, and low platelet count syndrome, systolic pressure ≥180 mm Hg or diastolic pressure ≥110 mm Hg, early induction and admission for preeclampsia, irrespective of gestational age. This may suggest that preterm delivery in women with early-onset preeclampsia might have influenced the impact of the situation and led to higher recall accuracy. The accurate recall of preterm delivery in the control group of women with preterm birth in history (98%) supports this suggestion.

Sensitivity and specificity of soliciting late-onset preeclampsia were less accurate than previous studies querying overall preeclampsia (sensitivity 72.6−78.0%, specificity 96.0−98.6%) (14,16,18,19). In these studies, gestational age at delivery was left out of consideration. When we repeated the analysis for only the confirmation of preeclampsia, the sensitivity increased from 68.2% to 86.4%. Indicating that not preeclampsia was recalled worse compared to other studies, but that gestational age was recalled as if they had early-onset preeclampsia. Gestational hypertension was recalled most inaccurate what is in line with previous observations (14,19,20). Although, 42% of these participants recalled a preeclampsia and only 29% recalled a history without any HDP. The inaccurate recall is often ascribed to less severe disease, less sequelae, and a lack of patient-physician communication. However, women with gestational hypertension often recalled the term preeclampsia and not the descriptive term hypertension so it might also be caused by a lack of uniform terminology. Women with previous preeclampsia were also less familiar with descriptive terms. This corresponded to previous research on which questions should be asked to solicit HDP (14,20–22). The use of direct terms, like “preeclampsia,” leads to the highest recall accuracy, compared to the use of descriptive terms, like “proteinuria.” In clinical practice, it, therefore, seems beneficial to share with patients their final diagnosis and long-term possible consequences, in order to improve the long-term recollections.

Women with previous late-onset preeclampsia often inaccurately recalled early-onset preeclampsia and women with gestational hypertension often recalled preeclampsia. Both are “incorrect” and cause “over-diagnosis.” In research, this could lead to erroneous results. In clinical practice, however, this could be suitable. “Over-treatment” of these women would lead to counseling and cardiovascular risk assessment to prevent CVD. These women, together with those with early-onset preeclampsia in history, face an increased cardiovascular risk later in life. Moreover, women without HDP in history did not recall a history of HDP. Therefore, clinicians could make use of a history of HDP in estimating cardiovascular risk for early identification of women at risk.

We found that accurate recall was significantly influenced by gestational age at delivery: the shorter the gestational age, the better the recall. Women with shorter gestational age are more likely to have a complicated pregnancy including admission, administration of medication, cesarean section, premature child, and extensive aftercare. This finding supports both theories that events with higher impact leads to higher accuracy in recall and that accurate recall relies on repeated and extensive patient-physician communication. Three other studies assessed factors that possibly modify the accuracy of maternal recall (18,19,23). They found that multiparity and high educational level positively influenced recall and that chronic hypertension and preeclampsia in a previous pregnancy negatively influenced recall of HDP(18). We found unexpectedly contradictory results in univariate analyses for parity and education where primiparous women recalled more accurate than multiparous women and women with low/intermediate education recalled better than highly educated women. Women with multiple pregnancies might confuse recollections of different pregnancies. The effect of education cannot be explained other than the high percentage of highly educated women participating in the study. Both effects diminished after adjustment.

Strengths and limitations

Previous research in the field of maternal recall of HDP stated that accurate recall in women with more severe disease may be influenced by other markers of disease severity like birth weight and preterm delivery. Our study is the first that included women with preterm birth in history without HDP to assess whether recall of HDP was influenced by preterm delivery. The distinction between early- and late-onset preeclampsia is highly valuable for clinical practice and risk assessment studies since both conditions precede two different risk profiles for CVD later in life.

Despite the relatively low response rates we recruited enough patients for each group, except for the group of late-onset preeclampsia. In the set time frame between 2006 and 2011, there were not enough patients in the database that could be contacted to reach the calculated sample size. This can be explained by the tertiary character of our hospital where women with late-onset preeclampsia are more often treated in secondary hospitals. Predictive values could not be calculated since the prevalences in the study were not representative of the general population. By presenting the crude data in Table 3, predictive values could be calculated with all preferred prevalences.

Conclusions

Early-onset preeclampsia can accurately be detected by a simple questionnaire 5 to 10 years after pregnancy. The validity of detecting late-onset preeclampsia and gestational hypertension was too low for use in research. These last-mentioned women often overstated their risk by recalling a more severe hypertensive disorder. In clinical practice, this would be suitable to detect most women at risk. Future studies should focus on the implementation of using a history of HDP in cardiovascular risk assessment.

Disclosure statement

No potential conflict of interest was reported by the authors.