Systematic Reviews to Inform Practice, November/December 2022

In the last decade there has been an increasing focus on the social determinants of health and their role in health outcomes. Twenty years ago Geronimus proposed the weathering hypothesis, which posited that the early experience and repeated exposures of social and economic adversity and political marginalization experienced by Black people lead to a cumulative allostatic load resulting in a more rapid deterioration of their health.1 Since then many observational studies, summarized by a meta-analysis of 41 studies examining the role of weathering as an explanation for the racial disparities in health outcomes, have supported this hypothesis.2 Additionally, a few studies have focused on the biological mechanism by which this stress and marginalization cause weathering by measuring premature shortening of telomere length.2 Anyone can experience stress, especially if they have low income, live in unstable housing, and have limited employment options. The question is, does discrimination have a different or differential effect on allostatic load, physiologic functioning, and health outcomes? Race is a social, not a biological, construct. It is used to promote racism, which according to the Merriam-Webster.com Dictionary has 3 aspects: (1) “belief that race is a fundamental determinant of human traits and capacities and that racial differences produce an inherent superiority of a particular race”; (2) “systemic oppression of a racial group to the social, economic, and political advantage of another”; and (3) “political or social system founded on racism and designed to execute its principles.”3 To answer the question, accurate methods to measure experiences of racism or discrimination are essential. Many measures have been developed, mostly in the United States, to measure the experience of racism and effects of discrimination over a person's lifetime or everyday exposure.4 These instruments range from a single question used in many states in their Pregnancy Risk Assessment Monitoring System surveys to longer ones such as the Perceived Racism Scale (51 items) and the Telephone-Administered Perceived Racism Scale (61 items).4 Most scales have appropriately been tested on Black Americans who have been subjected to a long history of racism in the United States. Fewer scales, such as Krieger's experience of discrimination scale, were developed using a working class population of white, Black, and Latino persons.5 Many scales have good psychometric properties, although some are lacking in important criteria, especially adequate sample size, and many have not been replicated in different population samples.4 Several studies have documented the effects of racism and discrimination on poor general health, both physical and mental, unrelated to age, sex, or birthplace.6 The systematic review and meta-analysis by van Daalen and colleagues sought to determine the effects of racial discrimination on pregnancy outcomes.7 The search strategy was extensive, using 8 electronic databases without any time limit or language restriction. A study was included if it measured self-perceived racial discrimination and had an adverse neonatal or maternal outcome. The authors preidentified 6 neonatal and 8 maternal outcomes but were only able to analyze preterm birth (PTB), low birth weight (LBW) and small for gestational age (SGA) neonates, and hypertensive disorders of pregnancy. Study quality was assessed using the Newcastle-Ottawa Scale converted to good, fair, and poor using Agency for Healthcare Research and Quality criteria. GRADE (Grading of Recommendations Assessment, Development, and Evaluation) scores assessed the risk of bias. It should be noted that because all studies of this topic have to be observational, GRADE scores start at a low level and can increase one level with a large effect and accounting for all confounders. The search strategy ended up including 24 studies with all but 4 taking place in the United States. Over half were cohort studies (n = 14) followed by 6 cross-sectional and 4 case-control studies. All studies were published in the last 20 years with the number of participants ranging from 39 to nearly 9500. The table of studies and their characteristics was detailed, clearly identifying the populations studied and the discrimination measures used. Half the studies were judged as “good” quality, but 10 were identified as “poor” and 2 as “fair.” The overall GRADE score for neonatal outcomes was very low and, for hypertensive disorders of pregnancy, low. Racial discrimination was measured using at least 8 different scales, with Krieger's experience of discrimination scale or its adaptation used most frequently (n = 11); 3 studies used a single question to measure the discrimination experience. Half of the scales focused on lifetime experience with racism, and the other half dealt with daily or recent experiences. Thirteen studies contributed to the summary estimate for PTB with the 2 largest studies contributing about 2200 and 1450 participants each. The pooled estimate calculated that racial discrimination increased the odds ratio (OR) of PTB by 40% (OR, 1.40; 95% CI, 1.17-1.68). There was moderate heterogeneity among studies (I2 = 60%), and when influential studies were excluded, the ORs were still statistically significant at 1.33 and 1.48. When the analysis was limited to Black or African American persons the OR was 1.33 (95% CI, 1.13-1.57). In assessing for publication bias, the authors estimated 5 missing studies possibly attenuating the OR estimate to 1.20. Although a meta-analysis of LBW and very low birth weight (VLBW) was not possible, 5 studies showed a positive association between discrimination and LBW, and 5 trended toward a positive association. Compared with low or medium exposure to racism, those with high exposure in the past year had 2.5 times the odds of a preterm LBW neonate, and those with lifetime exposure had 1.5 times the odds. Similarly, those experiencing more discriminatory events had higher odds of a VLBW neonate. Three studies contributed to the pooled estimate for SGA. Among the 1588 participants with 290 SGA neonates, the OR was not statistically significant (OR, 1.23; 95% CI, 0.76-1.99). When the analysis was limited to African Americans, there also was no association (OR, 0.90; 95% CI, 0.82-1.10), but when limited to older African Americans, the OR was significant at 1.45. Only one study focused on perinatal outcomes, specifically hypertensive disorders of pregnancy. This single study of racism and hypertensive disorders of pregnancy did not find that racism explained the disparity in the prevalence of hypertension among a large multiethnic, geographically diverse cohort of nulliparous women.8 Authors of this systematic review did not include postpartum mental health disorder as a potential maternal outcome, although 3 recent studies have shown an association between racial discrimination and increased postpartum mood disorders.9-11 In summary, this analysis reinforces the findings of an integrative review done by Alhusen and colleagues12 that included 15 studies and found racial discrimination has a small increased effect on the rate of PTB and probably LBW. The overall quality of the evidence is low given difficulties in accounting for all the potential confounding variables. Whether timing of exposure to racism early in life compared with later makes more of a difference has not really been studied. The weathering hypothesis would posit it does, and early exposure to the stress of racism, even in utero, may be the mechanism that causes hypothalamic-pituitary axis dysfunction. It may be that recent experiences of discrimination during pregnancy exert a triggering effect on the hypothalamic-pituitary axis leading to the inflammatory cascade and vascular dysfunction that contribute to PTB.13 This conceptual model fits with the few studies in this review that showed recent experiences versus lifetime experiences exerted a stronger effect on PTB. This review also highlights that the effects of racism have been poorly studied in relation to maternal outcomes and deserve to be the focus of further research. However, given the weathering hypothesis, studying maternal effects may require meticulously designed cohort studies using the life course perspective, which includes a combination of epigenetic, environmental, and social variables and historical time effects.14,15 There are other limitations to this review, some of which the authors point out. Nearly all of the studies take place in the United States, which means they are more generalizable to the context of the experience of American Black pregnant people. Because racism is embedded in a historical context and established social institutions, the experience of racism in the United States may not translate to or have the same attributes in other global contexts. Additionally, the racism scales measure someone's self-identified race or ethnicity but not their socially perceived identity, which may be more important. A study examining the effects of being socially assigned as white, no matter what a person's self-assigned race or ethnicity category was, resulted in similar health outcomes as for those assigned as white and self-identified as white and much poorer outcomes for those assigned as nonwhite, regardless of self-identity.16 This phenomenon may partially explain the lesser disparities among Hispanic individuals living in the United States. Additionally, acculturation may be an important lens through which Hispanic persons view their experience of discrimination, with at least one study demonstrating that shortened telomere length in Latinx pregnant people was associated with acculturation stress but not discrimination.17 Midwives are concerned with providing quality care to their patients to help them achieve the best possible outcomes. This means it is the responsibility of all midwives to tackle the issue of racism in this country through striving to be anti-racist. A plethora of resources is available to grow and learn toward the goal of becoming anti-racist. A place to start is the Racial Equity Library available on the Journal of Midwifery & Women's Health website where relevant articles have been collated for easy searching. Highly recommended is the editorial by Michelle Drew, which includes an annotated bibliography of readings.18 Smith ER, Oakley E, Grandner GW, et al. Clinical risk factors of adverse outcomes among women with COVID-19 in the pregnancy and postpartum period: a sequential, prospective meta-analysis [published online August 23, 2022]. Am J Obstet Gynecol. doi: 10.1016/j.ajog.2022.08.038 It has long been known that because of physiologic changes in pregnancy, pregnant people are at increased risk for severe respiratory illness from viral infections, particularly from coronaviruses associated with influenza A, varicella, and severe acute respiratory syndrome.1 Infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been no exception, and many, including the Centers of Disease Control and Prevention, have highlighted that those infected who were also pregnant have increased severe disease, complications, and higher deaths rates than comparable nonpregnant women.2 With coronavirus disease 2019 (COVID-19) cases declining rapidly, data about its effect on pregnant people are now emerging. Several interim meta-analyses and series of cases have been published in 2020 and 2021.3-5 However, the current sequential, prospective meta-analysis using data from nearly 22,000 people with suspected or confirmed SARS-CoV-2 infection in pregnancy is the largest to date. To be included in the data, a pregnant person with COVID-19 had to be followed for 42 days after birth. The largest limitation of the study is the heterogenous methods for identifying patients with COVID-19; for example, some centers screened everyone coming to the hospital for labor or pregnancy issues, and others tested only those who were symptomatic. Therefore, the data on incidence of different complications are hard to interpret. Most of the studies were rated as having low to moderate risk of bias, with the issue of the sampling frame being the primary reason for lower scores on the study's bias assessment tool. The study examined 24 outcomes related to COVID-19 severity; maternal, fetal, and neonatal morbidities; and adverse birth outcomes. Because of the nature of the data, the present meta-analysis could only examine outcomes by different conditions or comorbidities, comparing those with the condition and those without, but could not account for presence or absence of another related comorbidity. Cases occurred between January 2020 and December 2021. They came from 33 countries with broad representation across the globe and by World Bank income group levels, but half of the data came from a single source, Mexico. The majority of cases were identified in the third trimester of pregnancy with the caveat that gestational timing for cases from Mexico was unknown. The mean age of the study population was 29.4 years but varied by country with the mean age being 3 years younger in Puerto Rico, India, and Kenya. The incidence of COVID-19 severity, measured as intensive care unit admissions, critical care need, and ventilation as well as maternal and fetal or neonatal outcomes, was visually depicted by World Bank income levels. However, interpretation of these data is difficult because of the very different methods by which cases were identified. Of note, no important outcome differences were noted by parity. Unsurprisingly, like the general population, pregnant people with COVID-19 and chronic conditions had increased risk for more severe disease and outcomes. Those with diabetes, hypertension, or cardiovascular disease (CVD) were significantly more likely to require critical care (2-3 times increased risk) and ventilation (5-6 times the risk). Most importantly, the large sample size enabled examining pregnancy-related death, which was significantly increased among those with COVID-19 who also had hypertension (relative risk [RR], 2.75; 95% CI, 1.76-4.28), diabetes (RR, 3.79; 95% CI, 2.61-5.50), or CVD (RR, 16.76; 95% CI, 4.42-63.63). Hypertensive disease of pregnancy was also elevated among those with these disorders, as was preterm labor and birth, but the increased risk was small. More notably, the risk of placental abruption with these 3 conditions was increased 7 to 10 times. This may partially explain the increased RR of stillbirth, around 6.5 and 9.0, for those with diabetes and CVD, respectively, and to a lesser extent for those with hypertension (RR, 3.4). Perinatal, early neonatal, and neonatal deaths were also markedly increased for those with these conditions, with the highest risk among those with CVD. A novel finding was the documentation of increased risk among persons with HIV infection. The co-occurrence of HIV increased RR across the maternal conditions in the range of 1.5 to 2.7 and markedly increased perinatal death (RR, 8.63; 95% CI, 1.40-53.31). Examination of COVID-19 cases with nutritional factors of obesity, underweight, and anemia provided some fresh insights. Although SARS-CoV-2–infected individuals with obesity had a small increased risk for critical care and ventilation (RR, ≤2) and some a small increased risk for perinatal outcomes, such as preeclampsia and cesarean birth, there was no increased risk for fetal or neonatal deaths. On the other hand, persons with COVID-19 who were underweight had 5.7 to 9.3 times increased risk of critical care and ventilation, respectively, and 14 times the risk of pregnancy-related death. Being underweight was also associated with a 14 times increased risk of a VLBW neonate and 7 times increased risk for moderate PTB (< 34 weeks). Anemia was also associated with a small increase in ventilation (RR, 1.78), pregnancy-related death (RR, 2.36; 95% CI, 1.15-4.81), and stillbirth (RR, 3.75; 95% CI, 1.00-14.11). In summary, beyond the morbidities of diabetes, hypertension, CVD, and obesity that have been highlighted in the many studies of the risk factors associated with severity of COVID-19 infection in the general population, this study highlights increased risk in pregnancy among those with HIV, underweight, and anemia. What this study cannot answer is the role of particular conditions controlling for other comorbidities. For example, what is the independent contribution of HIV in COVID-19 infection outcomes among those who are pregnant independent of anemia, the most common hematologic finding among those with HIV?6 An additional contribution of this study is documentation of the large increased risk of maternal mortality and placental abruption among pregnant persons with COVID-19 and diabetes, hypertension, CVD, or HIV. The increased consequences to pregnant people who acquire SARS-CoV-2, at least through to the delta variant, are real. But the same is true with influenza A, and we still have fewer than 50% of pregnant people receiving the vaccine in their second and third trimester during the influenza season.7 COVID-19 vaccine acceptance globally among pregnant people has also been below 50%, with the lowest rates among high-income countries.8, 9 It is understandable that the general population might be hesitant to accept a vaccine that was produced with record speed and used an unfamiliar process, messenger RNA (mRNA). In truth, long-term safety data of mRNA vaccines are lacking, with the longest data safety on this type of vaccines prior to 2021 coming from 3 years of data from an mRNA rabies vaccine phase I study of 101 participants.10 The data being collected in the general population to date have been promising on short-term safety of the current COVID-19 vaccines except for rare problem of myocarditis in younger people, particularly men.11, 12 The same can be said for data on the safety of COVID-19 vaccines in pregnancy from the United States and Canada, but both studies depended on self-reported data, which are much more likely to be biased.13, 14 Longer-term, well-designed cohort studies will be needed to provide the best, unbiased information. On the other hand, the influenza vaccine is a traditional vaccine has been administered to populations for 20 years. Yet during that time, only one high-quality randomized clinical trial of the influenza vaccine in pregnant people has been carried out.15, 16 Vaccine hesitancy is a complex behavior, and the 5C framework (confidence, complacency, convenience, risk calculation, and collective responsibility) is most frequently used to understand it.17, 18 The 5C framework can assist health care providers in engaging each patient individually to determine what their particular concerns are for a particular vaccine and to answer these concerns as honestly as possible, including acknowledging the limitations of our knowledge. Chen Q, Qiu X, Fu A, Han Y. Effect of prenatal perineal massage on postpartum perineal injury and postpartum complications: a meta-analysis. Comput Math Methods Med. 2022;2022:3315638. This meta-analysis on the effects of prenatal perineal massage on perineal outcomes at births has some significant flaws. However, it may be instructive for those reviewing meta-analyses and wanting to use them to change clinical practice. The following search criteria for study inclusion were identified: randomized clinical trial design, the intervention was done prenatally starting at 34 to 36 weeks’ gestation, there was a control group of no massage, and outcomes were related to perineal effects. No date or language restriction was imposed on the search. A similar meta-analysis was previously published in 2020, and it identified 11 eligible studies through August 20191 and a Cochrane systematic review from 2013.2 This current meta-analysis identified 16 studies, of which only one study, which was not a randomized trial, was published after 2019. This new review missed one study identified by both previously published meta-analyses.3 The discrepancies among these 3 meta-analyses highlight the importance of not only search terms and databases used but also of hand searching references of studies and possibly review papers and being clear about inclusion criteria. However, the most significant issues related to the studies included in this most recent meta-analysis is the inclusion of studies that did not meet the inclusion criteria. A large study with null results, which accounted for 13% of the weight of the final estimate for any perineal laceration and 28.6% for first- or second-degree lacerations, was an intervention performed in the second stage of labor and not prenatally.4 Additionally, a single study that had 2 publications for different outcomes was included twice in the meta-analyses, thereby increasing its weight in the final estimate from 13.8% to 26.4% for any laceration and from 29% to over 50% for first- or second-degree lacerations.5,6 Lastly, 3 studies included in this review were not randomized clinical trials—2 were observational and one was quasi-experimental—and therefore did not meet study criteria.7-9 The results of this most recent meta-analysis are similar to those of the 2 previously published, although the effect estimates favoring perineal massage are smaller, probably because of the previously mentioned issues. For example, this recent review found a 44% decreased risk for third- and fourth-degree lacerations (RR, 0.56; 95% CI, 0.47-0.67) with prenatal massage, which is a smaller effect than the 2020 review, which found a larger effect of 64% decreased risk (RR, 0.36; 95% CI, 0.14-0.89).1 The results for any perineal lacerations were more similar, about a 20% decrease with prenatal massage. Probably because of the decreased risk of lacerations, those practicing prenatal perineal massage also report significantly less postpartum pain at 3 days. This meta-analysis does not provide any new evidence in support of prenatal perineal massage, but it is a good reminder to midwives that it may be worthwhile to spend some time discussing this simple intervention and providing patients with a good handout such as the Share with Women handout on perineal massage.10