Variance Estimation Research Articles

The Pre-Operative Evaluation Clinic: An Underutilized Service in Optimizing Analgesic Outcomes in Patients on Buprenorphine, Methadone, and Naltrexone for Substance Use Disorder or Chronic Pain.

Patients receiving methadone, buprenorphine, and naltrexone for either chronic pain or substance use disorder (SUD) pose perioperative challenges. Due to their complex pharmacology, perioperative recommendations continue to evolve. Deviations from these recommendations may result in worse perioperative outcomes. A formal preoperative evaluation (POE) and optimization of patients on these medications are recommended to address these concerns. A single-center retrospective electronic health record review was performed with adult patients on methadone, buprenorphine, and naltrexone undergoing elective surgery between January 1, 2010 and December 31, 2020. The primary outcome of interest was the percentage of patients referred to the POE clinic for evaluation prior to the scheduled elective surgery. In addition, we assessed differences in variables (perioperative opioid, hospital length of stay, perioperative multimodal analgesics, perioperative complications, inpatient pain service consult, readmission within 30 days, cancellation of surgery, addiction medicine consult) based on POE clinic evaluation. This analysis was performed separately for patients prescribed these medications for SUD versus chronic pain. Continuous outcomes were analyzed using linear regression with generalized estimating equations (GEE) and robust variance estimates. A total of 714 patients were included in the final analysis, of which 572 (80%) took buprenorphine, methadone, or naltrexone for chronic pain and 142 (20%) took these medications for SUD. Within the chronic pain and SUD subpopulations, 193 (34%) and 35 (25%) patients had formal POE clinic assessments, respectively. Among those taking these medications for chronic pain, POE clinic evaluation was associated with a higher likelihood for receiving non-opioid multimodal analgesics perioperatively (p = 0.016). Formal preoperative evaluations are currently underutilized in patients who take buprenorphine, methadone, or naltrexone for chronic pain or SUD. These patients may benefit from POE clinic assessment to optimize perioperative outcomes.

Functional difficulties in children and youth with autism spectrum disorder: analysis of the 2019 Canadian Health Survey on Children and Youth.

This study examined the prevalence of functional difficulties and associated factors in Canadian children/youth aged 5 to 17 years diagnosed with autism spectrum disorder (ASD). We analyzed data from the 2019 Canadian Health Survey on Children and Youth (CHSCY), a nationally representative survey of Canadian children/youth that used the Washington Group Short Set on Functioning (WG-SS) to evaluate functioning in six daily tasks. For each functional domain, binary outcomes were derived (no/some difficulty, a lot of difficulty/no ability). We used logistic regression to identify associations between demographic characteristics, educational experiences, and perceived mental and general health and the most common functional difficulties, namely those related to remembering/concentrating, communication and self-care. All estimates were weighted to be representative of the target population. The bootstrap method was used to calculate variance estimates. Analysis of the records of 660 children/youth with ASD revealed that the most common functional difficulties were remembering/concentrating (22%; 95% CI: 18-27), communicating (19%; 95% CI: 15-23) and self-care (13%; 95% CI: 10-17). Lower perceived mental health was associated with increased functional difficulties with remembering/concentrating. ASD diagnosis at a lower age and lower perceived general health were associated with increased functional difficulty with communication. Parental expectations for postsecondary education were associated with decreased functional difficulty for self-care. One or more functional difficulties from the WG-SS was present in 39% of Canadian children/youth aged 5 to 17 years with ASD. Functional difficulties with remembering/concentrating, communication and self-care were most common.

Association between the home-to-healthcare center distance and hearing aid abandonment among older adults.

Access to audiology services for older adults residing in sparsely populated regions is often limited compared to those in central urban areas. The geographic accessibility to follow-up care, particularly the influence of distance, may contribute to an increased risk of hearing aid abandonment. To assess the association between the home-to-healthcare-calibration-center distance and hearing aid abandonment among older adults fitted in the Chilean public health system. 455 patients who received hearing aids from two public hospitals in two regions were considered. Univariate and multivariate Poisson regression models with robust variance estimation were used to analyze the association between the geographical distance and hearing aid abandonment, accounting for confounding effects. Approximately 18% of the sample abandoned the hearing aid, and around 50% reported using the hearing aid every day. A twofold increase in distance between home and the hearing center yielded a 35% (RR = 1.35; 95% CI: 1.04-1.74; p = 0.022) increased risk of hearing aid abandonment. Also, those in the second quintile had a 2.17 times the risk of abandoning the hearing aid compared to the first quintile (up to 2.3 km). Under the assumption that patients reside within the first quintile of distance, a potential reduction of 45% in the incidence of hearing aid abandonment would be observed. The observed risk remained consistent across different statistical models to assess sensitivity. A higher distance between the residence and the healthcare center increases hearing aid abandonment risk. The association may be explained by barriers in purchasing supplies required to maintain the device (batteries, cleaning elements, potential repairs, or maintenance).

On variance estimation of random forests with Infinite-order U-statistics

On variance estimation of random forests with Infinite-order U-statistics

Deep Variation Prior: Joint Image Denoising and Noise Variance Estimation without Clean Data.

With recent deep learning based approaches showing promising results in removing noise from images, the best denoising performance has been reported in a supervised learning setup that requires a large set of paired noisy images and ground truth data for training. The strong data requirement can be mitigated by unsupervised learning techniques, however, accurate modelling of images or noise variances is still crucial for high-quality solutions. The learning problem is ill-posed for unknown noise distributions. This paper investigates the tasks of image denoising and noise variance estimation in a single, joint learning framework. To address the ill-posedness of the problem, we present deep variation prior (DVP), which states that the variation of a properly learnt denoiser with respect to the change of noise satisfies some smoothness properties, as a key criterion for good denoisers. Building upon DVP and under the assumption that the noise is zero mean and pixel-wise independent conditioned on the image, an unsupervised deep learning framework, that simultaneously learns a denoiser and estimates noise variances, is developed. Our method does not require any clean training images or an external step of noise estimation, and instead, approximates the minimum mean squared error denoisers using only a set of noisy images. With the two underlying tasks being considered in a single framework, we allow them to be optimised for each other. The experimental results show a denoising quality comparable to that of supervised learning and accurate noise variance estimates.

Factors influencing how informal caregivers of people with multiple sclerosis access and use a curated intervention website: Analysis from an RCT.

Healthcare consumers and providers are increasingly turning to digital solutions, such as curated websites. Knowing who accesses/benefits from these may improve design and development. This study investigated website usage of informal caregivers of people with multiple sclerosis and shifts in outcome plausibly associated with usage. Secondary analysis of data from a randomized clinical trial of 148 caregivers compared effectiveness of a website + tele-coaching to a website only intervention for caregivers. Groupwise differences in means/proportions were tested using t-tests and chi-square. Modified Poisson regression with a robust variance estimator and ordinal logistic regression tested the relationship between group and likelihood of website log-in. Ordinal logistic regression models examined whether caregiver characteristics were associated with website use. Generalized estimating equations (GEE) with an autoregressive correlation structure modeled the relationship between website usage and outcomes. Females were more likely to access the website than males (60% vs. 43%; p = 0.05). Though not statistically significant, a possible association (POR: .85, 95% CI: .69, 1.03) between caregiver burden and website access emerged; caregivers experiencing highest levels of burden appeared less likely to engage. Usage patterns differed by treatment arm: the website-only group accessed the Caring for yourself topic significantly more (61.67% vs. 38.33%: p = .04) with similar, but insignificant, trends for other topics. Clinicians can be confident referring females with moderate levels of burden to website-based interventions. By contrast, male caregivers and those experiencing high levels of burden may be less likely to access these resources, pointing to the need for alternative interventions. Clinicaltrials.gov, registration number: NCT0466208.

Propensity Score Regression for Causal Inference With Treatment Heterogeneity

Understanding how treatment effects vary on individual characteristics is critical in the contexts of personalized medicine, personalized advertising and policy design. When the characteristics are of practical interest are only a subset of full covariate, non-parametric estimation is often desirable; but few methods are available due to the computational difficult. Existing non-parametric methods such as the inverse probability weighting methods have limitations that hinder their use in many practical settings where the values of propensity scores are close to 0 or 1. We propose the propensity score regression (PSR) that allows the non-parametric estimation of the heterogeneous treatment effects in a wide context. PSR includes two non-parametric regressions in turn, where it first regresses on the propensity scores together with the characteristics of interest, to obtain an intermediate estimate; and then, regress the intermediate estimates on the characteristics of interest only. By including propensity scores as regressors in the non-parametric manner, PSR is capable of substantially easing the computational difficulty while remain (locally) insensitive to any value of propensity scores. We present several appealing properties of PSR, including the consistency and asymptotical normality, and in particular the existence of an explicit variance estimator, from which the analytical behaviour of PSR and its precision can be assessed. Simulation studies indicate that PSR outperform existing methods in varying settings with extreme values of propensity scores. We apply our method to the national 2009 flu survey (NHFS) data to investigate the effects of seasonal influenza vaccination and having paid sick leave across different age groups.

Multi-Scale Spatio-Temporal Memory Network for Lightweight Video Denoising.

Deep learning-based video denoising methods have achieved great performance improvements in recent years. However, the expensive computational cost arising from sophisticated network design has severely limited their applications in real-world scenarios. To address this practical weakness, we propose a multiscale spatio-temporal memory network for fast video denoising, named MSTMN, aiming at striking an improved trade-off between cost and performance. To develop an efficient and effective algorithm for video denoising, we exploit a multiscale representation based on the Gaussian-Laplacian pyramid decomposition so that the reference frame can be restored in a coarse-to-fine manner. Guided by a model-based optimization approach, we design an effective variance estimation module, an alignment error estimation module and an adaptive fusion module for each scale of the pyramid representation. For the fusion module, we employ a reconstruction recurrence strategy to incorporate local temporal information. Moreover, we propose a memory enhancement module to exploit the global spatio-temporal information. Meanwhile, the similarity computation of the spatio-temporal memory network enables the proposed network to adaptively search the valuable information at the patch level, which avoids computationally expensive motion estimation and compensation operations. Experimental results on real-world raw video datasets have demonstrated that the proposed lightweight network outperforms current state-of-the-art fast video denoising algorithms such as FastDVDnet, EMVD, and ReMoNet with fewer computational costs.

The Neutrophil-to-Lymphocyte Ratio as a Predictor of Acute Exacerbations Among Patients With COPD in Uganda.

The neutrophil-to-lymphocyte ratio (NLR) is an inexpensive biomarker that potentially predicts acute exacerbations of chronic obstructive pulmonary disease (AECOPDs). We evaluated the association of baseline NLR and respiratory hospitalization risk within one year among chronic obstructive pulmonary disease (COPD) patients in Uganda, a low- and middle-income country. A total of 312 COPD patients were followed for one year. Clinical characteristics and exacerbation rates were collected. Poisson regression with robust variance estimators was used to measure the association between NLR and hospital admissions due to COPD exacerbations. Receiver-operator characteristic (ROC) curves and the area under the curve were used to assess the ability of NLR to predict AECOPDs. The median (Q 1, Q 3) age was 64 years (53, 71). Females comprised 50.96% (n=159) of the cohort, and 71.2% (n=222) of participants had moderate or severe COPD. A total of 9.9% (n=31) of participants experienced a COPD exacerbation during the period of follow-up. At baseline, the median (Q 1, Q 3) NLR ratio among participants who experienced an exacerbation was 1.46 (0.92, 2.33) compared to 1.03 (0.72,1.42) among those who did not experience one during the follow-up period (p=0.002). Using Youden and Liu's methods, the optimal NLR cutoff for predicting COPD exacerbation was 1.17. This cutoff resulted in a ROC curve area of 0.64 (95% confidence interval: 0.56, 0.73). The NLR could be used as a risk predictor, in low- and middle-income countries, for hospital admissions due to COPD exacerbations. A cutoff of 1.17 was an independent predictor of hospitalization due to acute exacerbations of COPD within one year.

Intelligent Analysis Method of E-Commerce Data Based on Multiple Machine Learning Algorithms

Machine learning algorithm is one of the most popular research fields in the 21st century, and it is widely used in various industries. This paper first discusses the development status, characteristics and intelligent classification technology of e-commerce based on various methods. Then, through statistical analysis of multiple samples, this paper establishes a model to predict the quantity of goods and other time-related changes in the future period of time, and proposes an improved variance estimate to achieve the best effect under different data scenarios. Finally, the experiment proves that the response time of the linear regression algorithm fluctuates between 20ms and 25ms, and the overall fluctuation is small. For different samples, the response time of the linear regression algorithm is maintained in a relatively stable range, showing the stability of different input data. The response time of the decision tree algorithm fluctuates between 28ms and 33ms, showing a certain fluctuation. For different samples, the response time of decision tree algorithm fluctuates greatly, which may be affected by the characteristics of input data and show different processing efficiency. The response time of support vector machine algorithm fluctuates between 14ms and 18ms, and the change is relatively stable. For different samples, the response time of SVM algorithm is kept at a relatively stable level, showing the consistency of input data processing ability.

Comment on “The Treatment of Uncertainty in Hydrometric Observations: A Probabilistic Description of Streamflow Records” by de Oliveira and Vrugt

AbstractQuantifying uncertainties and errors in hydrometric observations is critical to improve our ability to predict streamflow. de Oliveira and Vrugt (2022, https://doi.org/10.1029/2022wr032263) expand on an earlier publication (Vrugt et al., 2005, https://doi.org/10.1029/2004wr003059) to describe a difference‐based variance estimation method that is, used to estimate the aleatory observation uncertainty in streamflow records directly from a time series of such data. The method is then applied to hourly data from 500+ catchments across the contiguous United States. This comment outlines our concerns that the assumptions needed to effectively use difference‐based variance estimation methods are not always met by hourly streamflow records. We illustrate our concerns through the use of a specific test case. We argue that more work is needed to quantify the individual sources of errors and uncertainties in hydrometric observations.

Temporal trends in medical device implant procedures in Australia 2008-22: evidence from the Australian Institute of Health and Welfare National Hospital Morbidity database.

ObjectiveExamine the temporal trends in medical device implant procedures in the Australian population.MethodsWe used data from the Australian Institute of Health and Welfare from the financial years 2007-08 to 2021-22 and chose the most frequently performed medical device implant procedures. We estimated the annual change in volume of procedures and age-standardised rates by calculating the compound annual growth rate (CAGR), and used regression with the Newey-West robust variance estimator to examine whether there was a linear trend in the age-standardised rates for each procedure.ResultsFor procedures including cardiac pacemakers, heart valves, hip and knee arthroplasties, and intraocular lenses, the crude CAGR was over 3%. For the age-standardised rates, the CAGR was largest for cardiac pacemaker, followed by heart valve replacement and hip arthroplasty procedures. For some procedures, the growth was more than in the Australian population, including cardiac pacemakers (β=1.00; 95% CI: 0.14-1.86), heart valve replacements (β=0.41; 95% CI: 0.28-0.54), hip arthroplasty (β=3.50; 95% CI: 1.61-5.38), and knee arthroplasty (β=4.31; 95% CI: 0.54-8.09) procedures. The trend of standardised rates of procedures, including incisional hernia with mesh, breast implants, coronary stents, and cardiac defibrillators, grew at the same rate as the population, whereas the rate for gastric banding procedures decreased (β=-3.14; 95% CI: -4.92 to -1.34).ConclusionThe findings from the current study, showing a large increase in medical device implant procedures, will assist in future healthcare planning and efforts in post-market surveillance of safety of medical devices.

Is it really a neuromyth? A meta-analysis of the learning styles matching hypothesis.

Learning styles have been a contentious topic in education for years. The purpose of this study was to conduct a meta-analysis of the effects of matching instruction to modality learning styles compared to unmatched instruction on learning outcomes. A systematic search of the research findings yielded 21 eligible studies with 101 effect sizes and 1,712 participants for the meta-analysis. Based on robust variance estimation, there was an overall benefit of matching instruction to learning styles, g = 0.31, SE = 0.12, 95% CI = [0.05, 0.57], p = 0.02. However, only 26% of learning outcome measures indicated matched instruction benefits for at least two styles, indicating a crossover interaction supportive of the matching hypothesis. In total, 12 studies without sufficient statistical details for the meta-analysis were also examined for an indication of a crossover effect; 25% of these studies had findings indicative of a crossover interaction. Given the time and financial expenses of implementation coupled with low study quality, the benefits of matching instruction to learning styles are interpreted as too small and too infrequent to warrant widespread adoption.

Correcting for measurement error under meta-analysis of z-transformed correlations.

This study mainly concerns correction for measurement error using the meta-analysis of Fisher's z-transformed correlations. The disattenuation formula of Spearman (American Journal of Psychology, 15, 1904, 72) is used to correct for individual raw correlations in primary studies. The corrected raw correlations are then used to obtain the corrected z-transformed correlations. What remains little studied, however, is how to best correct for within-study sampling error variances of corrected z-transformed correlations. We focused on three within-study sampling error variance estimators corrected for measurement error that were proposed in earlier studies and is proposed in the current study: (1) the formula given by Hedges (Test validity, Lawrence Erlbaum, 1988) assuming a linear relationship between corrected and uncorrected z-transformed correlations (linear correction), (2) one derived by the first-order delta method based on the average of corrected z-transformed correlations (stabilized first-order correction), and (3) one derived by the second-order delta method based on the average of corrected z-transformed correlations (stabilized second-order correction). Via a simulation study, we compared performance of these estimators and the sampling error variance estimator uncorrected for measurement error in terms of estimation and inference accuracy of the mean correlation as well as the homogeneity test of effect sizes. In obtaining the corrected z-transformed correlations and within-study sampling error variances, coefficient alpha was used as a common reliability coefficient estimate. The results showed that in terms of the estimated mean correlation, sampling error variances with linear correction, the stabilized first-order and second-order corrections, and no correction performed similarly in general. Furthermore, in terms of the homogeneity test, given a relatively large average sample size and normal true scores, the stabilized first-order and second-order corrections had type I error rates that were generally controlled as well as or better than the other estimators. Overall, stabilized first-order and second-order corrections are recommended when true scores are normal, reliabilities are acceptable, the number of items per psychological scale is relatively large, and the average sample size is relatively large.

On the bias in the AUC variance estimate

The area under the Receiver Operating Characteristic (ROC) curve (AUC) is a standard metric for quantifying and comparing binary classifiers. A popular approach to estimating the AUCs and the associated variabilities – the variance of the AUC or the full covariance matrix of multiple correlated AUCs – is the one proposed by DeLong et al. (1988), which is based on the Mann Whitney two-sample U-statistics. The bias of a variance estimator is an important factor in applications such as hypothesis testing and construction of confidence intervals – a negatively biased variance estimator may lead to incorrect conclusions, and a positive bias is conservative hence preferable. In this work, we show that the (co-)variance estimate in DeLong’s approach is always positively biased. More specifically, the difference matrix between the expectation of the estimated covariance and the true covariance is a positive semi-definite matrix. This bias is non-negligible when the sample size is small, and quickly diminishes as the sample size increases. Our method relies on constructing, from the AUC kernel, a random variable whose (co-)variance matrix coincides with the bias, thereby establishing the claim. We also discuss alternative approaches to AUC variance estimation that may potentially reduce the bias.

Genetic variation and trait association of fine rice genotypes

To explore high yielding genotype and significant yield contributing trait, field experiment was conducted using alpha lattice design replicated three times at the National Wheat Research Program (NWRP), Bhairahawa, Nepal in 2022. Eighteen fine rice genotypes were evaluated based on nine quantitative traits (Heading days, maturity days, plant height, panicle length, effective tillers, filled grains per panicle, unfilled grains per panicle, thousand grain weight and grain yield per hectare). Variance analysis, correlation and estimation of genetic parameters were conducted to comprehend genetic variations and interrelationships among traits within genotypes. The results showed most of traits except unfilled grains had statistical significance differences. Tarahara-107 (3940 kg/ha) had the highest grain yield followed by Tarahara-2 (3700 kg/ha) and NR 2381-RGA1-RGA2 (3665 kg/ha). Similarly, grain yield showed positive and significant association with filled grains at both phenotypic and genotypic level (rp = 0.7074 ** and rg= 0.9482 **). Elevated values of PCV and GCV were recorded for filled grains (22.05 and 17.57). The high value GAM coupled with high heritability was recorded for filled grains (28.83%), plant height (23.10%), number of effective tillers (21.57%) and thousand grain weight (19.37%). Thus, identified genotypes NR 2380-RGA1-RGA2 and NR 2381-RGA1-RGA2 exhibited early and potential high yielding genotypes while Tarhara-107 and Tarhara-2 identified as high yield, medium maturity and good filled grain number per panicle. In addition, filled grains trait displayed tightly linked with grain yield and possess high GCV, PCV, GAM with moderate heritability indicated direct selection could be employed for trait enhancement in breeding program.

Improved estimators of population variance using robust measures in case of missing data

In this paper, an attempt has been made in proposing some improved estimators of finite population variance when random non-response is present in the study variable, on the estimators given by Sharma and Singh (2018). Four improved regression type estimators have been proposed using information on two auxiliary variables, one of which has missing values in the data, using two-phase sampling scheme. Different robust regression methods have been applied on the proposed techniques so as to improve the efficiency of the estimators in case of outliers. It has been observed that robust estimators are more efficient than the proposed ones under some conditions. All the proposed estimators are compared with existing estimators for different values of probability of non-response and the results are verified by using empirical and simulation studies.

Estimates of Phenotypic and Genotypic Variance and Heritability in Eighty-Nine Bambara Groundnut [Vigna subterranea (L.) Verdcourt] Accessions Collected from Six Regions of Niger

The bambara groundnut [Vigna subterranea (L.) Verdc. (Fabaceae)] is a legume mainly cultivated by women, for the nutritional quality of these seeds. It is a so-called minor culture and the improvement of the plant and its popularization remains to be promoted. The objective of this present work is to evaluate the genetic variability of bambara groundnut accessions. The experiment was carried out according to a completely randomized block device with four replications. Twenty-two (22) characters including four (4) phenological, four (4) morphological and fourteen (14) related to yield were evaluated for accessions characterization. Descriptive analysis showed significant differences. The coefficients of variation ranged from 3.51% (maturity date) to 38.87% (shell weight). Significantly high values (CV˃20%) for 8 of the metric parameters were observed. Pod weight per plant and seed weight per plant (r=0.943), yield in kg/ha (r=0.943); seed weight per plant and yield in kg/ha (r=0.999) showed the strongest correlations. The phenotypic and genotypic coefficients of variation were high for dry biomass weight (PCV=42.23%; GCV=28.40%), shell weight (PCV=63.46%; GCV=22.46%) and 100-seed weight (PCV=25.57%; GCV=25.25%). Maturity date (95.77%) and 100-seed weight (99.84%) had high heritabilities. Broad-sense heritability and genetic gain are high for 100-seed weight (H2=99.84%; GA=52.58%). The Ascending hierarchical classification produced four groups of which group 4 is the most efficient in yield with short (20.58cm) early accessions (80.32 days). Groups 1 and 2 include, late accessions (~22cm) with respectively maturity dates (DM=85.67 days and DM=86.53 days.

Intimate partner violence against women during covid-19: A population-based study in Vitória, state of Espírito Santo, Brazil.

Violence against women has a negative impact on multiple dimensions of women's health. During the Covid-19 pandemic, intimate partner violence against women has continued, and in some contexts has intensified. The aim of this study was to identify the prevalence of intimate partner violence against women during covid-19 pandemic and its association with socioeconomic, behavioral, and life-experience factors. Cross-sectional, population-based study conducted in the municipality of Vitória, state of Espírito Santo, from January to May 2022, where 1,086 women aged 18 years and over were interviewed. The World Health Organization (WHO) instrument on violence against women was used to screen outcomes. The prevalence of violence during the pandemic (psychological, physical, and sexual) and bivariate analysis with sociodemographic, behavioral, family, and life history characteristics of women were estimated. The multivariate analysis was carried out for each type of violence, the Poisson regression model was performed with an estimate of robust variance, inserting the variables of interest with (p<0.20). Those with p<0.05 remained in the adjusted model. The prevalence of violence psychological against women perpetrated by an intimate partner during the pandemic was the most frequent (20.2%), followed by physical (9.0%) and sexual violence (6.5%). Women with less schooling and who were single had a higher prevalence of physical and psychological violence, as did those with a history of sexual abuse in childhood and whose mothers had been beaten by their intimate partners. Sexual violence was more prevalent among non-white, with up to eight years of schooling, whose mothers had a history of intimate partner violence, and who consumed alcohol during four days or more (p<0.01). Psychological, physical, and sexual violence perpetrated by the intimate partner during the pandemic presented high magnitude among women living in Vitória. Sociodemographic, behavioral factors, and personal and maternal experiences of violence were associated with the phenomenon.

Disparities in Hypertension Prevalence, Awareness, Treatment, and Control Among Women Living With and Without HIV in the US South.

Hypertension-related diseases are major causes of morbidity among women living with HIV. We evaluated cross-sectional associations of race/ethnicity and HIV infection with hypertension prevalence, awareness, treatment, and control. Among women recruited into Southern sites of the Women's Interagency HIV Study (2013-2015), hypertension was defined as (1) systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg according to clinical guidelines when data were collected, (2) self-report of hypertension, or (3) use of antihypertensive medication. Awareness was defined as self-report of hypertension, and treatment was self-report of any antihypertensive medication use. Blood pressure control was defined as <140/90 mm Hg at baseline. Prevalence ratios for each hypertension outcome were estimated through Poisson regression models with robust variance estimators adjusted for sociodemographic, behavioral, and clinical risk factors. Among 712 women, 56% had hypertension and 83% were aware of their diagnosis. Of those aware, 83% were using antihypertensive medication, and 63% of those treated had controlled hypertension. In adjusted analyses, non-Hispanic White and Hispanic women had 31% and 48% lower prevalence of hypertension than non-Hispanic Black women, respectively. Women living with HIV who had hypertension were 19% (P = .04) more likely to be taking antihypertension medication when compared with women living without HIV. In this study population of women living with and without HIV in the US South, the prevalence of hypertension was lowest among Hispanic women and highest among non-Hispanic Black women. Despite similar hypertension prevalence, women living with HIV were more likely to be taking antihypertensive medication when compared with women living without HIV.