Pregnancy outcomes and blood pressure visit-to-visit variability and level in three less-developed countries

In pregnancy in well-resourced settings, limited data suggest that higher blood pressure (BP) visit-to-visit variability may be associated with adverse pregnancy outcomes. Included were pregnant women in 22 intervention clusters of the CLIP (Community- Level Interventions for Preeclampsia) cluster randomized trials, who had received at least 2 prenatal contacts from a community health worker, including standardized BP measurement. Mixed-effects adjusted logistic regression assessed relationships between pregnancy outcomes and both BP level (median [interquartile range]) and visit-to-visit variability (SD and average real variability [ARV], adjusted for BP level), among all women and those who became hypertensive. The primary outcome was the CLIP composite of maternal and perinatal mortality and morbidity. Among 17 770 pregnancies, higher systolic and diastolic BP levels were associated with increased odds of the composite outcome per 5 mm Hg increase in BP (odds ratio [OR], 1.05 [95% CI, 1.03-1.07] and OR, 1.08 [1.06-1.11], respectively). Higher BP visit-to-visit variability was associated with increased odds, per a SD increase in BP variability measure, of (1) hypertension (systolic: OR, 2.09 [1.98-2.21] for SD and 1.52 [1.45-1.60] for ARV; diastolic: OR, 2.70 [2.54-2.87] for SD and 1.86 [1.76-1.96] for ARV); and (2) the composite outcome (systolic: OR, 1.10 [1.06-1.14] for SD and 1.06 [1.02-1.10] for ARV; diastolic: OR, 1.07 [1.03-1.11] for SD and 1.06 [1.02-1.09] for ARV). In 3 less-developed countries, higher BP level and visit-to-visit variability predicted adverse pregnancy outcomes, providing an opportunity for high-definition medicine. This was an unplanned secondary analysis of data from the 22 intervention clusters of the CLIP cluster randomized trials, aimed at externally validating findings from the CHIPS trial.12 Data can be accessed through the CLIP trials data access committee (Text S1 in the Data Supplement). The CLIP trials were conducted in 2014 to 2017 in India (N=6 intervention clusters), Pakistan (N=10), and Mozambique (N=6).13–16 The unit of randomization (cluster) was the local administrative unit. All pregnant women aged 15 to 49 years (12–49 years in Mozambique) were identified in their community by trained community health workers. All women provided written informed consent to participate. The trial was unmasked given the nature of the intervention, aimed at addressing the 3 delays in triage, transport, and treatment related to preeclampsia. First, community engagement addressed barriers and facilitators to accessing care. Second, existing cadres of community health workers were trained to task-share pregnancy hypertension-oriented care at CLIP contacts in women’s homes, using the CLIP PIERS (Preeclampsia Integrated Estimate of Risk Score) on-the-Move (POM) digital health application for risk stratification.17 Community health workers (1) responded to emergency conditions, if relevant; (2) took women’s BP and assessed dipstick proteinuria at the first and any hypertensive contact; (3) administered oral methyldopa 750 mg for BP of at least 160/110 mm Hg; (4) administered 10 g intramuscular magnesium sulfate for suspected severe preeclampsia; and (5) and referred to a comprehensive emergency obstetric care facility. Standardized BP measurement by trained community health workers used a semiautomated pregnancy- and preeclampsia-validated oscillometric device (Microlife 3AS1-2).18 Having rested for 5 minutes, women’s BP was measured at least twice; all readings were entered into the POM application, which averaged the first and second readings and if they differed by >10 mm Hg, a third reading was requested and the second and third readings averaged. The planned frequency of prenatal POM-guided CLIP contacts was every 4 weeks, at minimum. Trained surveillance teams conducted regular surveys of households (every 3–6 months), except in India where a prospective population-based surveillance system was established. The primary outcome was a composite of all-cause maternal and perinatal mortality and morbidity. Maternal death or morbidity occurred during or within 42 days of pregnancy; morbidity was one or more life-threatening complications of pregnancy, defined as a serious end-organ complication of preeclampsia (eg, eclampsia), another major cause of maternal mortality/morbidity (ie, obstetric sepsis or vaginal fistula), or receipt of a life-saving intervention. Perinatal death was stillbirth, early or late neonatal mortality, and morbidity a composite of problems that could be ascertained in community (eg, seizure or feeding difficulty). For detailed definitions, see Table S2 in the Data Supplement. Overall coordination and data management were by the Preeclampsia–Eclampsia Monitoring, Prevention and Treatment research group at the University of British Columbia, Canada. Ethical approvals were granted by the University of British Columbia (H12-03497) and relevant in-country research ethics boards (Aga Khan University, Pakistan, 2590-Obs-ERC-13; KLE University, India, MDC/IECHSR/2011-12/A-4, ICMR 5/7/859/12-RHN; and Centro de Investigação em Saúde de Manhiça (CIBS-CISM/038/14) and Mozambique National Bioethic Committee (219/CNBS/14). The trials are registered at URL: https://www.clinicaltrials.gov (Unique identifier: {“type”:”clinical-trial”,”attrs”:{“text”:”NCT01911494″,”term_id”:”NCT01911494″}}NCT01911494) and the related individual participant data meta-analysis on PROSPERO (CRD42018102564). We included CLIP participants in pregnancy, from enrollment until follow-up for the CLIP primary outcome, who had at least 2 antenatal contacts by community health workers (prerequisite for determining BP variability, see below). Mean systolic and diastolic BP levels were the mean of relevant values taken at all POM-guided CLIP contacts between enrollment and delivery. Within-participant visit-to-visit BP variability was assessed using all POM-guided CLIP contacts after enrollment until delivery. We evaluated 2 measures of BP visit-to-visit variability used outside pregnancy: (1) within-participant SD to reflect dispersion of measurements around mean BP and (2) average real variability (ARV) as the average of the absolute successive difference of all BP values, reflecting changes over short time intervals (so a decrease by 4 mm Hg and then an increase by 6 mm Hg would represent an ARV of 5). We adjusted for mean BP level, as higher levels are associated with more variability. Any correlation was explored between BP variability and number of measurements. In our primary analysis, relationships were explored between each major CLIP outcome and both BP level and visit-to-visit variability, using values before the outcomes: progression to hypertension (systolic BP of at least 140 mm Hg or a diastolic BP of at least 90 mm Hg, based on an average of 2 measurements), composite of maternal or perinatal mortality or morbidity (primary outcome), composite maternal outcome (mortality or morbidity) and composite perinatal outcome (stillbirth, early or late neonatal death, or neonatal morbidity) to evaluate whether the direction of effect on maternal outcomes was the same. In addition, we further examined the relationship between each major CLIP outcome and BP variability only among women who became hypertensive to see if BP variability could add further information to BP level. Data were summarized as median and interquartile range and counts (percentages) for continuous and categorical variables, respectively. The mean BP level-outcome relationship was explored by mixed-effects logistic regression. Adjustment was made for country and cluster (each as a random effect), maternal age at enrollment, maternal education, parity, and gestational age at enrollment. The odds ratio (OR) for each outcome was calculated per 5 mm Hg increase in mean BP from enrollment until delivery. The BP variability-outcome relationship was evaluated for all women, and specifically for women who developed pregnancy hypertension, by mixed-effects logistic regression, adjusted for average BP level (defined as the mean of the BP readings used to define visit-to-visit variability) and the variables described above for BP level analyses. The change in the scale of the OR was calculated per SD increase in both metrics of BP variability to compare the relative importance of one measure with another. Correlation between BP visit-to-visit variability and the number of measurements was assessed by Spearman correlation (r). In sensitivity analyses: (1) for all outcomes, we excluded BP values within 7, 14, 21, and 28 days before delivery to minimize the extent to which BP variability may be an artifact of the outcomes themselves (ie, reverse causality); (2) for all outcomes, we added further adjustment for the final antenatal BP measurement, to account for BP trajectory; (3) for all outcomes, we excluded repeat pregnancies for the same woman; and (4) for progression to hypertension, we incorporated diagnoses based only on trial surveillance data for women who became hypertensive after their last POM-guided visit. A P<0.05 was considered statistically significant, without adjustment for multiple comparisons. All statistical analyses were performed using and R 3.5.3 (R Development Core Team, Vienna, Austria). J. Bone had access to all data and takes responsibility for its integrity and the data analysis.