Roll-over test in primigravidae attending a public primary care service

Objective:To determine the roll-over test (ROT) performance in predicting pregnancy-induced hypertension (PIH) in primigravidae aged 15-29 years in a public primary care service. Method: Prospective cohort study enrolling 369 consecutive and initially normotensive primigravidae. The ROT was applied within 28-32 weeks of pregnancy. PIH was defined as diastolic blood pressure (DBP) 90 mm Hg or systolic blood pressure (SBP) 140 mm Hg, or a rise in DBP 15 mm Hg or a rise in SBP 30 mm Hg. The ROT prognostic properties were calculated, and a receiver operating characteristic (ROG) curve was constructed. Results: For the 20 mm Hg cutoff point, sensitivity was 20% and specificity was 93%. Positive and negative predictive values were, respectively, 23% and 92%, for a PIH cumulative incidence of 9.5%. With other cutoff points, the ROG curve showed a poor discriminatory value of the test. ConcliJsion: The ROT was not useful for predicting PIH in a primary prenatal care setting.


INTROOUCTION H
ypertension which complicates pregnancy is a major cause of prematurity and perinatal mortaIity.I Any Iarge unseIected group of pregnant women with raised bIood pressure will contain individuaIs whose hypertension antedates the pregnancy, those whose bIood pressure began to rise after the end of the second trimester (pregnancy-induced hypertension -PIH), and a few with pre-existing hypertension exacerbated by the pregnancy.It is recognized that primigravidae are at highest risk ofPIH. 2 In most parts of the world, prenatal care services, when accessible, have the most profound effect on the

Adress for correspondence:
Luiz Francisco Marcopito Rua Pedro de Toledo, 675 São Paulo/SP -Brasil -CEP 04039-032 consequences of hypertension in pregnancy, but not in its prevention.If the roll-over test (ROT), published by Gant et aI. 3 in 1974 and still taught in medicaI schools, is in fact accurate, it could be useful for identifying higher risk groups of primigravidae, who would deserve special attention.This prospective study was therefore aimed at determining the vaIue of the ROT in predicting PIH development in primigravidae aged 15-29 years Iiving in the eastem region of Greater Sao Paulo, Brazil, a 10w-income urban area, who attended a primary prenatal care public service (Leonor Mendes de Barros Maternity Hospital).

PARTICIPANTS ANO METHOOS
The incIusion cri teria were as follows: a) spontaneous demand for routine prenatal assistance in the service where the study was based; b) reporting being primigravida at the time of registration; c) 15-29 years old; and d) normotensive (blood pressure lower than 140/90 mm Hg before the 24th week of pregnallcy) without the use of anti-hypertensive drugs.
The exclusion criteria (and the respective numbers excluded by each criterion) were as folIows: a) in the first contact, expectation of moving to an unknown "address (7); b) continued inability to give a precise home address at the second contact (10); c) miscarriage before the 24th week of pregnancy (5); d) determination, through further contact, that the woman was either not a primigravida (4) or not pregnant (3); e) suffered congenital vascular disease (1); or f) made regular use of aspirin or drugs known as vasoactives (zero).Out of 422 primigravidae considered eligible, 394 were submitted to ROT.
Loss during folIow-up was minimized by special efforts aimed at those (68%) who missed at least one appointment: 46 telephone calIs were made, 731letters were sent (maximum ofthree per woman)4, and 151 home visits were performed (in the case of non-response to the third letter).These efforts resulted in folIow-up losses of only 6.3% (25 participants).AIso, 34 hospitaIs had to be visited in order to get the medicaI records of those (more than 50%) who gave birth in hospitaIs other than that where the study was based.AlI the remaining 369 were folIowed up from 24 weeks of pregnancy until one month after delivery.
The ROT consisted of the same procedure performed by Gant et aI. 3 , except that the participants' responses to angiotensin II after the test were not evaluated.At the .time of study, each participant was placed in the left lateral recumbent position, and blood pressure measurements in the right upper arm were repeated at five minutes intervals, until a constant baseline diastolic blood pressure had been established.When the diastolic blood pressure had become constant, the subject was turned to the supine position and the blood pressure was measured immediately and again five minutes later while the participant was stilI supine.
The ROT was conducted by the same observer every time, who was trained to measure blood pressure with the London School of Hygiene and Tropical Medicine sound tapes.AlI the participants were invited to urinate before the test.The Korotkoff first phase was taken as the systolic blood pressure.The diastolic was recorded as the Korotkoff fifth phase -or the fourth, if the fifth phase was not detected before the value reached zero.
An increase in diastolic blood pressure of at least 20 mm Hg, from the left lateral to the supine position, was taken as a positive test, but other cutoff points were considered in the analysis.Sensitivity and specificity were calculated with such different cutoff points, in order to generate a receiver operating characteristic (ROC) curve 6 • PIH was defined as 7 a diastolic blood pressure (DBP) of at least 90 mm Hg or a systolic blood pressure (SBP) of at least 140 mm Hg, or a rise in DBP of at least 15 mm Hg or a rise in SBP of at least 30 mm Hg.The diagnosis of PIH was confirmed when such blood pressure readings were obtained on at least two occasions, six hours or more apart.
Regarding the review of medicaI.records from around the time of delivery, the same blood pressure criteria were applied, and the medicaI record review was done by the same observer, blind to the previous blood pressure leveIs of the participants.

RESULTS
Some characteristics of the 369 primigravidae analyzed are shown in Table 1.In these women, the overalI risk ofPIH development was 9.5% (35/369), with the 95% confidence interval between 6.5% and 12.5%.
Table 3 shows the ROT sensitivities and specificities for different ~DBP cutoff points.Those numbers were used to construct the ROC curve exhibited in the Figure, which shows the poor ROT prognostic value for PIH in the studied primigravidae.* Pregnancy-induced hypertension pump) after the ROT, and observed a high correlation with the ROT results.This procedure, however, was not tested in the present study, as it would not have been feasibIe in a public prenataI care service as crowded as the one where this study was based.This time-consuming (aImost thirty minutes) test wouId certainIy be considered for preventive purposes, if it presented an attractive performance in defining a high risk group of primigravidae for deveIoping PIH.However, as has been shown, that was not the case.

DISCUSSION
Many investigators have studied the ROT performance in severaI parts of the worId: some 8 -12 have recommended its use, and others l3 -18 have noL MethodoIogicaI discrepancies may account for such a division of opinions: in two of these studies lO ,17, the cutoff point of a positive ROT was different to i1DBP ~20 mm Hg; in two I2 ,16, diagnosis of PIH incIuded criteria other than bIood pressure only; in at Ieast three 8,14,18 it was not stated whether the medicaI record reviewer for PIH diagnosis was bIind to the ROT resuIts.The remaining 13 ,15 still showed conflicting resuIts on the ROT performance", possibIy because of differing PIH diagnosis criteria, in spite of being based on bIood pressure onIy.None of these studies was carried out in a primary care setting.

CONCLUSION
The poor performance of the ROT in predicting further PIH does not recommend its u"se in primigravidae in a primary prenataI care setting.
Difference from supine to left lateral diastolic blood pressure, in mm Hg. *