J ClinEpidemiolVol.43,No. 5,pp. 461466, Printed in Great Britain. All rights reserved

0895-4356/90 $3.00+ 0.00

1990

Copynght c 1990PergamonPressplc

HAEMATOLOGIC PARAMETERS PREGNANCY OUTCOME A PROSPECTIVE

COHORT

AND

STUDY IN THE THIRD

TRIMESTER

J. A. KNOTTNERUS,‘.* L. R. DELGADO,~ P. G. KNIPSCHILD,’ G. G. M. ESSED’ and F. SMITS~ Departments of ’General Practice, 2Epidemiology/Health Care Research and ‘Obstetrics and Gynaecology, University of Limburg, Maastricht, The Netherlands (Receined in renised form 9 September 1989)

Abstract-The relationship between haematologic parameters and pregnancy outcome was investigated in a prospective study, taking important covariates into consideration. A questionnaire asking information about smoking, alcohol consumption and iron intake was completed by a consecutive series of 796 ambulatory singleton pregnant women at 31-32 weeks of gestation. Subsequently, haematologic parameters were measured. After delivery, birth weight, gestation length and blood pressure values during pregnancy were abstracted from the obstetricians’ records. Low birth weight (< 2.500 g) and preterm birth (< 37 weeks), were significantly more frequently seen in women with high haemoglobin (> 8.0 mmol/l) or high haematocrit (2 38%) values, with Mantel-Haenszel odds ratios varying from 2.4 to 4.2. Also a high erythrocyte count (>4.5/pl) was correlated with low birth weight, while for mean corpuscular volume (MCV) no such relationships were found. Pregnancy induced hypertension was positively associated with haemoglobin, haematocrit and erythrocyte count, but could not fully explain the relationships between these parameters and adverse outcomes. Using multiple logistic regression with low birth weight and/or preterm birth as the dependent variable, the effect of haematocrit still was demonstrable after controlling for smoking, hypertension, parity, alcohol consumption and iron intake. These results are in agreement with the hypothesis that a higher blood viscosity is a risk factor for suboptimal placenta-perfusion. Haematologic Hypertension

parameters Parity

Pregnancy

INTRODUCTION It is generally haemoglobin

accepted

that very low maternal

(Hb) levels are associated with an unfavourable pregnancy outcome. More recently, increasing attention has been focused on the potential adverse effects of high Hb levels [l-6]

or insufficient

plasma

volume

expansion

*All correspondence should be addressed to: J. A. Knottnerus, Department of General Practice, University of Limburg, P.O. Box 616, 6200 MD Maastricht, The Netherlands. CE 4315-D

outcome

Cohort

study

Smoking

and haemodilution [7-lo] during pregnancy. However, most studies involved very small populations. Furthermore, few investigators adjusted their results for potential confounding factors such as smoking [l, 41 and high blood pressure [4]. In the important study of Garn [l] gestational age at the moment of the Hbdetermination was not taken into account. Since in cases of immature delivery the relatively high Hb levels of early pregnancy will have been included, this might have produced an artificially strong association between high 461

J. A. KNOTTNERUS et al.

462

Hb levels and an unfavourable pregnancy outcome. In this paper the authors report upon an investigation into the relationship between haematologic parameters, routinely measured in ambulatory pregnant women in the 31st or 32nd week of gestation, and pregnancy outcome, while taking into account the influences of smoking, alcohol consumption, high blood pressure, parity and iron administration. METHODS

Subjects Between 1 July 1985, and 1 May 1986, all pregnant women, paying a routine antenatal visit in the 31st or 32nd week of gestation to the obstetric outpatient department of the Maastricht University Hospital, were asked to participate in the study. Also invited were the pregnant women who paid the corresponding antenatal visit to one of two midwife practices in Maastricht, in the period between 1 October 1985, and 1 February 1986. Not included were pregnant women who had been referred during pregnancy from other (non-academic) hospitals; who had no appointment in the 31st or 32nd week of gestation; or who had been hospitalized in that period; and women having a twin pregnancy. 92% of the pregnant women visiting the obstetric outpatient department and 64% of those attending the midwife practices participated in the study. In total, complete questionnaires of 796 women were obtained (666 attending the outpatient clinic and 130 attending the midwife practices). They were 18-39 years of age (P50 = 27 years) and 48% were nulliparous. More detailed information on the distribution of age and parity in the study population is shown in Table 1. Data acquisition The women were asked to fill out a questionnaire concerning smoking (average number of cigarettes smoked daily in the preceding month) and alcohol consumption (average number of glasses of alcoholic drinks taken weekly in the preceding month). The questionnaires were returned before leaving the outpatient clinic or midwife office. After the completion of the questionnaire, blood was collected routinely in EDTA vacuum-tubes by venapuncture, and analyzed *1pl=10-6mm3;

lfl=lp-‘.

Table

1. Age and parity of the study lation

popu-

Parity Age

0

I

4S/pl) were clearly associated with an unfavourable outcome (Tables 2 and 3). Relationships between low values and pregnancy outcome could not be demonstrated, but extremely low values were infrequently seen. No associations were found between MCV and the outcome parameters. Using routine antenatal blood pressure measurements, we operationalized PIH as a diastolic blood pressure of at least 90 mmHg, or systolic pressure of at least 140 mmHg, or a rise

Outcome

463

in the former of at least 15 mmHg or in the latter of at least 30 mmHg, recorded in the second half of pregnancy [ 121. As can be seen in Tables 2 and 3, PIH was more frequent in the higher categories of the haematologic parameters, except for MCV. After stratification for PIH, as a possible confounding variable, the relationships between haematologic parameters (Hb, Ht and erythrocyte count) and outcome were still present, both in PIH and non-PIH cases: controlling for PIH, and dichotomizing at the highest cut-off points. The Mantel-Haenszel odds ratios for an adverse outcome were significantly different from 1.Oin most cases (Table 4). These results did not change after exclusion of women (n = 13) who were known with preexistent hypertension. 37% of the women still smoked in the month before entering the study. Among these women, we found a higher incidence of low birth weight (8%, as opposed to 4% in non- or ex-smokers, p = 0.03, Pearsons’ chi square test) and of preterm birth (6 vs 3%, p = 0.13). However, this could not explain the relationships between haematologic parameters and pregnancy outcome, because smoking was not correlated with haemoglobin level, haematocrit and erythrocyte count. Twenty-two percent of the women consumed weekly, on average, at least one glass of an alcoholic drink, in the preceding month. However, no relationships of alcohol consumption with pregnancy outcome, nor with haematologic parameters, could be demonstrated.

Table 2. Haemoglobin level and haematocrit at 31-32 weeks of gestation, relation to pregnancy outcome and high blood pressure Haemoglobin

(mmol/l):

Low birth weight g