Nonresponse Weights Research Articles

Temporal changes in the risk of six-month post-covid symptoms: a national population-based cohort study.

It is unclear how the risk of post-covid symptoms evolved during the pandemic, especially before the spread of Severe Acute Respiratory Syndrome Coronavirus 2 variants and the availability of vaccines. We used modified Poisson regressions to compare the risk of six-month post-covid symptoms and their associated risk factors according to the period of first acute covid: during the French first (March-May 2020) or second (September-November 2020) wave. Non-response weights and multiple imputation were used to handle missing data. Among participants aged 15 or more in a national population-based cohort, the risk of post-covid symptoms was 14.6% (95% CI: 13.9%, 15.3%) in March-May 2020, versus 7.0% (95% CI: 6.3%, 7.7%) in September-November 2020 (adjusted RR: 1.36, 95% CI: 1.20, 1.55). For both periods, the risk was higher in the presence of baseline physical condition(s), and it increased with the number of acute symptoms. During the first wave, the risk was also higher for women, in the presence of baseline mental condition(s), and it varied with educational level. In France in 2020, the risk of six-month post-covid symptoms was higher during the first than the second wave. This difference was observed before the spread of variants and the availability of vaccines.

Trends in Opioid and Nonsteroidal Anti-Inflammatory Drug Use for Patients with Kidney Stones in United States Emergency Departments from 2015 to 2021.

Introduction: Renal colic is frequently treated with opioids; however, narcotic analgesic use can lead to dependence and abuse. We evaluated use trends of opioids and nonsteroidal anti-inflammatory drugs (NSAIDs) for pain management of kidney stones in United States emergency departments (EDs) from 2015 to 2021. Methods: Kidney stone encounters were identified using National Hospital Ambulatory Medical Care Survey data. We applied a multistage survey weighting procedure to account for selection probability, nonresponse, and population weights. Medication use trends were estimated through logistic regressions on the timing of the encounter, adjusted for selected demographic and clinical characteristics. Results: Between 2015 and 2021, there were an estimated 9,433,291 kidney stone encounters in United States EDs. Opioid use decreased significantly (annual odds ratio [OR]: 0.87, p = 0.003), and there was no significant trend in NSAID use. At discharge, male patients were more likely than females (OR: 1.93, p = 0.001) to receive opioids, and Black patients were less likely than White patients (OR: 0.34, p = 0.010) to receive opioids. Regional variation was also observed, with higher odds of discharge prescriptions in the West (OR: 3.15, p = 0.003) and Midwest (OR: 2.49, p = 0.010), compared with the Northeast. Thirty-five percent of patients received opioids that were stronger than morphine. Conclusion: These results suggest improved opioid stewardship from ED physicians in response to the national opioid epidemic. However, regional variation as well as disparities in discharge prescriptions for Black and female patients underscore opportunities for continued efforts.

Estimation of a Population Total Under Nonresponse Using Follow-up

Abstract In this article, we propose methods to minimize bias due to unit nonresponse. We consider a two-phase sampling design where the second phase is a probability subsample of nonrespondents from the first phase. In this context, we propose three weighting procedures to estimate the total when not all units in the subsample respond. The weighting is based on the response homogeneity group (RHG) model. Given the RHG model, theoretical results on bias and variance estimation are obtained for all estimators. In a simulation study, we evaluate the empirical properties of the three estimators as well as of estimators based on two commonly used procedures to handle unit nonresponse in single-phase sampling design. These two procedures include: (i) nonresponse calibration weighting, also known as the one-step approach, and (ii) nonresponse probability weighting followed by calibration, also known as the two-step approach. Our results indicate that when there is significant deviation from the assumed RHG model, the nonresponse follow-up estimators perform better in terms of bias and coverage.

Lifetime Suicide Attempts in Otherwise Psychiatrically Healthy Individuals

Not all people who die by suicide have a psychiatric diagnosis; yet, little is known about the percentage and demographics of individuals with lifetime suicide attempts who are apparently psychiatrically healthy. If such suicide attempts are common, there are implications for suicide risk screening, research, policy, and nosology. To estimate the percentage of people with lifetime suicide attempts whose first attempt occurred prior to onset of any psychiatric disorder. This cross-sectional study used data from the US National Epidemiologic Study of Addictions and Related Conditions III (NESARC-III), a cross-sectional face-to-face survey conducted with a nationally representative sample of the US civilian noninstitutionalized population, and included persons with lifetime suicide attempts who were aged 20 to 65 years at survey administration (April 2012 to June 2013). Data from the NESARC, Wave 2 survey from August 2004 to September 2005 were used for replication. Analyses were performed from April to August 2023. Lifetime suicide attempts. The main outcome was presence or absence of a psychiatric disorder before the first lifetime suicide attempt. Among persons with lifetime suicide attempts, the percentage and 95% CI of those whose first suicide attempt occurred before the onset of any apparent psychiatric disorders was calculated, weighted by NESARC sampling and nonresponse weights. Separate analyses were performed for males, females, and 3 age groups (20 to <35, 35-50, and >50 to 65 years). In the total sample of 36 309 respondents, 1948 persons had lifetime suicide attempts; 66.8% (95% CI, 64.1%-69.4%) were female, and 6.2% (95% CI, 4.9%-7.4%) had no apparent lifetime psychiatric diagnoses when surveyed. In addition, 13.4% (95% CI, 11.6%-15.2%) made their first suicide attempt prior to psychiatric disorder onset. Thus, an estimated 19.6% of respondents first attempted suicide without an antecedent psychiatric disorder. No significant age or sex differences were detected in the percentage of those with lifetime suicide attempts absent psychiatric disorders, although females were more likely than males to attempt suicide in the year of psychiatric disorder onset (14.9% [95% CI, 12.5%-17.3%] vs 8.6% [95% CI, 6.0%-11.2%]; P < .001), and attempts were less frequent among those older than 50 to 65 years (3.9% [95% CI, 3.5%-4.4%] vs 6.1% [95% CI, 5.4%-6.8%] for 35-50 years and 6.2% [95% CI, 5.6%-6.9%] for 20 to <35 years; P < .001). In this study, an estimated 19.6% of individuals who attempted suicide did so despite not meeting criteria for an antecedent psychiatric disorder. This finding challenges clinical notions of who is at risk for suicidal behavior and raises questions about the safety of limiting suicide risk screening to psychiatric populations.

Developing non-response weights to account for attrition-related bias in a longitudinal pregnancy cohort

BackgroundProspective cohorts may be vulnerable to bias due to attrition. Inverse probability weights have been proposed as a method to help mitigate this bias. The current study used the “All Our Families” longitudinal pregnancy cohort of 3351 maternal-infant pairs and aimed to develop inverse probability weights using logistic regression models to predict study continuation versus drop-out from baseline to the three-year data collection wave.MethodsTwo methods of variable selection took place. One method was a knowledge-based a priori variable selection approach, while the second used Least Absolute Shrinkage and Selection Operator (LASSO). The ability of each model to predict continuing participation through discrimination and calibration for both approaches were evaluated by examining area under the receiver operating curve (AUROC) and calibration plots, respectively. Stabilized inverse probability weights were generated using predicted probabilities. Weight performance was assessed using standardized differences of baseline characteristics for those who continue in study and those that do not, with and without weights (unadjusted estimates).ResultsThe a priori and LASSO variable selection method prediction models had good and fair discrimination with AUROC of 0.69 (95% Confidence Interval [CI]: 0.67–0.71) and 0.73 (95% CI: 0.71–0.75), respectively. Calibration plots and non-significant Hosmer-Lemeshow Goodness of Fit Tests indicated that both the a priori (p = 0.329) and LASSO model (p = 0.242) were well-calibrated. Unweighted results indicated large (> 10%) standardized differences in 15 demographic variables (range: 11 − 29%), when comparing those who continued in the study with those that did not. Weights derived from the a priori and LASSO models reduced standardized differences relative to unadjusted estimates, with the largest differences of 13% and 5%, respectively. Additionally, when applying the same LASSO variable selection method to develop weights in future data collection waves, standardized differences remained below 10% for each demographic variable.ConclusionThe LASSO variable selection approach produced robust weights that addressed non-response bias more than the knowledge-driven approach. These weights can be applied to analyses across multiple longitudinal waves of data collection to reduce bias.

Differences in the impact of the COVID-19 pandemic in the long-term trajectories of life satisfaction by sex in the UK: a prospective cohort study

BackgroundThe COVID-19 pandemic has disproportionately affected women's mental health, with most evidence focused on mental ill-health outcomes. Previous research suggests that differences in time-use might explain this disparity, as women generally spent more time doing psychologically taxing activities than men. We investigated (1) sex differences in the impact of the pandemic in the long-term trajectories of life satisfaction and (2) whether time-use differences partly explained that differential impact. MethodsIn this prospective cohort study, we used the 1970 British Cohort Study, a birth cohort representing people born in Great Britain in 1970. Since 1970, information has been collected for all cohort members every 4–10 years (ten sweeps). Additional data were collected at three timepoints during the pandemic (May 2020, September–October 2020, February–March 2021) as part of the COVID-19 Survey. Response rates within the target population ranged between 23·6% (May, 2020) and 32·5% (March, 2021) in the COVID-19 Survey. We used non-response weights to restore sample representativeness to the target population (cohort members alive and residing in the UK during the pandemic). Life satisfaction was prospectively assessed at ages 26 (1996), 30, 34, 42, 46, 50, 50·5, and 51 years (2021) using a single question with responses ranging from 0 (lowest life satisfaction) to 10 (highest life satisfaction). We used multiple-group piecewise latent growth curve models grouped by sex assigned at birth. Additional models were adjusted for self-reported number of hours spent working, volunteering, home-schooling children, taking care of children and other people, and doing housework during the pandemic. Participants provided verbal informed consent. Ethics approval was not sought as non-sensitive, open-access data with minimal risk of identification was used. FindingsWe analysed data from 6766 adults (3799 [56·2%] women and 2967 [43·8%] men). Compared with men, women had consistently higher levels of life satisfaction before the pandemic (Δintercept,unadjusted = 0·213, 95% CI 0·087 to 0·340, p=0·0009), and experienced a more accelerated decline with its onset (Δquad2,unadjusted = –0·018, –0·026 to –0·011; p<0·0001), even after accounting for time use (Δquad2,adjusted = –0·016, –0·031 to –0·001; p=0·0348). Results were robust to adjustment for concurrent financial and occupational situation (Δquad2,adjustedfull = –0·019, –0·035 to –0·004; p=0·0158). InterpretationOur study shows sex inequalities in the impact of the pandemic on the long-term life satisfaction trajectories of adults in their 50s. Self-reported time-use differences did not account for these inequalities. One of the limitations is that we relied on self-reported time-use and were not able to include factors like the so-called mental load (being responsible for organising or monitoring activities, regardless of being involved in their execution), which might partly explain these results. FundingEconomic and Social Research Council (ESRC) Centre for Society and Mental Health at King's College London.

Availability of Family Caregiver Programs in US Cancer Centers

Family caregivers provide the majority of health care to the 18 million patients with cancer in the US. Yet despite providing complex medical and nursing care, a large proportion of caregivers report no formal support or training. In recognition of this gap, many interventions to support cancer caregivers have been developed and tested over the past 2 decades. However, there are few system-level data on whether US cancer centers have adopted and implemented these interventions. To describe and characterize the availability of family caregiver support programs in US cancer centers. This cross-sectional national survey study was conducted between September 1, 2021, and April 30, 2023. Participants comprised clinical and administrative staff of Commission on Cancer-accredited US cancer centers. Data analysis was performed in May and June 2023. Survey questions about the availability of 11 types of family caregiver programs (eg, peer mentoring, education classes, and psychosocial programs) were developed after literature review, assessment of similar program evaluation surveys, and discussions among a 13-member national expert advisory committee. Family caregiver programs were defined as structured, planned, and coordinated groups of activities and procedures aimed at specifically supporting family caregivers as part of usual care. Survey responses were tabulated using standard descriptive statistics, including means, proportions, and frequencies. Of the surveys sent to potential respondents at 971 adult cancer centers, 238 were completed (response rate, 24.5%). After nonresponse weight adjustment, most cancer centers (75.4%) had at least 1 family caregiver program; 24.6% had none. The most common program type was information and referral services (53.6%). Cancer centers with no programs were more likely to have smaller annual outpatient volumes (χ2 = 11.10; P = .011). Few centers had caregiver programs on training in medical and/or nursing tasks (21.7%), caregiver self-care (20.2%), caregiver-specific distress screening (19.3%), peer mentoring (18.9%), and children caregiving for parents (8.3%). Very few programs were developed from published evidence in a journal (8.1%). The top reason why cancer centers selected their programs was community members requesting the program (26.3%); only 12.3% of centers selected their programs based on scientific evidence. Most programs were funded by the cancer center or hospital (58.6%) or by philanthropy (42.4%). In this survey study, most cancer centers had family caregiver programs; however, a quarter had none. Furthermore, the scope of programming was limited and rarely evidence based, with few centers offering caregiving education and training. These findings suggest that implementation strategies are critically needed to foster uptake of evidence-based caregiver interventions.

Graphical Causal Models for Survey Inference

Directed acyclic graphs (DAGs) are now a popular tool to inform causal inferences. We discuss how DAGs can also be used to encode theoretical assumptions about nonprobability samples and survey nonresponse and to determine whether population quantities including conditional distributions and regressions can be identified. We describe sources of bias and assumptions for eliminating it in various selection scenarios. We then introduce and analyze graphical representations of multiple selection stages in the data collection process, and highlight the strong assumptions implicit in using only design weights. Furthermore, we show that the common practice of selecting adjustment variables based on correlations with sample selection and outcome variables of interest is ill-justified and that nonresponse weighting when the interest is in causal inference may come at severe costs. Finally, we identify further areas for survey methodology research that can benefit from advances in causal graph theory.

Long-term psychological distress trajectories and the COVID-19 pandemic in three British birth cohorts: A multi-cohort study.

Growing evidence suggests that population mental health outcomes have worsened since the pandemic started. The extent that these changes have altered common age-related trends in psychological distress, where distress typically rises until midlife and then falls after midlife in both sexes, is unknown. We aimed to analyse whether long-term pre-pandemic psychological distress trajectories were disrupted during the pandemic, and whether these changes have been different across cohorts and by sex. We used data from three nationally representative birth cohorts comprising all people born in Great Britain in a single week of 1946 (National Survey of Health and Development, NSHD), 1958 (National Child Development Study, NCDS), or 1970 (British Cohort Study, BCS70). The follow-up data used spanned 39 years in NSHD (1982 to 2021), 40 years in NCDS (1981 to 2001), and 25 years in BCS70 (1996 to 2021). We used psychological distress factor scores, as measured by validated self-reported questionnaires (NSHD: Present State Examination, Psychiatric Symptoms Frequency, and 28- and 12-item versions of General Health Questionnaire; NCDS and BCS70: Malaise Inventory; all: 2-item versions of Generalized Anxiety Disorder scale and Patient Health Questionnaire). We used a multilevel growth curve modelling approach to model the trajectories of distress across cohorts and sexes and obtained estimates of the differences between the distress levels observed during the pandemic and those observed at the most recent pre-pandemic assessment and at the peak in the cohort-specific pre-pandemic distress trajectory, located at midlife. We further analysed whether pre-existing cohort and sex inequalities had changed with the pandemic onset using a difference-in-differences (DiD) approach. The analytic sample included 16,389 participants. By September/October 2020, distress levels had reached or exceeded the levels of the peak in the pre-pandemic life-course trajectories, with larger increases in younger cohorts (standardised mean differences [SMD] and 95% confidence intervals of SMDNSHD,pre-peak = -0.02 [-0.07, 0.04], SMDNCDS,pre-peak = 0.05 [0.02, 0.07], and SMDBCS70,pre-peak = 0.09 [0.07, 0.12] for the 1946, 1958, and 1970 birth cohorts, respectively). Increases in distress were larger among women than men, widening pre-existing sex inequalities (DiD and 95% confidence intervals of DiDNSHD,sex,pre-peak = 0.17 [0.06, 0.28], DiDNCDS,sex,pre-peak = 0.11 [0.07, 0.16], and DiDBCS70,sex,pre-peak = 0.11 [0.05, 0.16] when comparing sex inequalities in the pre-pandemic peak in midlife to those observed by September/October 2020). As expected in cohort designs, our study suffered from high proportions of attrition with respect to the original samples. Although we used non-response weights to restore sample representativeness to the target populations (those born in the United Kingdom in 1946, 1958, and 1970, alive and residing in the UK), results may not be generalisable to other sections within the UK population (e.g., migrants and ethnic minority groups) and countries different than the UK. Pre-existing long-term psychological distress trajectories of adults born between 1946 and 1970 were disrupted during the COVID-19 pandemic, particularly among women, who reached the highest levels ever recorded in up to 40 years of follow-up data. This may impact future trends of morbidity, disability, and mortality due to common mental health problems.

ADDRESSING SOCIAL ISOLATION IN NURSING HOMES DURING COVID-19: A NATIONAL SURVEY

Abstract COVID-19 related policies introduced extraordinary social disruption in nursing homes. In response to the unprecedented COVID-19 pandemic, congregated long term care living facilities attempted and/or implemented innovative intervention strategies to alleviate loneliness in residents. We surveyed Directors of Nursing/Administrators of 1,669 homes sampled in strata defined by size (number of beds 30–99, 100+) and quality ratings (1, 2–4, 5) between February-May 2022. The response rate was 30%. Almost 2/3rds of respondents completed it online and the rest via paper. Analyses included nonresponse survey weights to provide nationally representative results. Among a list of 17 situations that occurred, staff shortages was identified as extremely stressful by the majority. Staff were extremely stressed about doing more to meet resident needs and keeping up with rapidly changing regulations which often lacked clinical sense. One third of respondents were extremely concerned about their home’s ability to meet residents’ social needs before vaccines, dropping to 13% after vaccines. Nursing homes tried and perceived as most useful using technology (tablets, phones, emails), assigning staff as a family contact, and staff spending more time with residents. Nearly 60% were extremely concerned about staff burnout/mental health before vaccines and 40% remained extremely concerned after vaccines. Many nursing homes attempted to mitigate the harmful effects of social isolation during the pandemic, despite the stressful circumstances in which staff worked. The extent to which various approaches were implemented varied. While concerns about social isolation reduced after vaccines were available, administrators remain extremely concerned about staff burnout and mental health.

Iranian Multi-center Osteoporosis Study (IMOS), 2021–2022: the study protocol

BackgroundThis paper presents the protocol of the 4th round of Iranian Multi-center Osteoporosis Study (IMOS), a national survey with the primary objective of estimating the prevalence of osteoporosis and sarcopenia and their risk factors in a representative sample of urban and rural populations.MethodsThe target population of the survey is all individuals ≥ 50 years in Iran. A multi-stage random sampling method has been used in the study. We stratified the 31 provinces of the country into 5 strata based on the distribution of their potential risk factors for osteoporosis and randomly selected one or two provinces from each stratum. Then, we invited 2530 people aged ≥ 50 years recruited in the 8th National Survey of None Communicable Diseases (NCD) Risk Factors (STEPs-2021) in the selected provinces to participate in IMOS. Body composition measurements including bone mineral density, muscle mass, and fat mass are measured through Dual-energy X-ray Absorptiometry (DXA) method using HOLOGIC (Discovery and Horizon) devices; and Trabecular Bone Score (TBS) is measured on the DXA scans using iNsight software. Anthropometric measurement and physical examinations are made by a trained nurses and other required information are collected through face-to-face interviews made by trained nurses. Laboratory measurements are made in a central lab. The prevalence of osteoporosis and sarcopenia will be estimated after applying sampling design, non-response, and post-stratification weights to the data.DiscussionIMOS will provide valuable information on the prevalence and determinants of osteoporosis and sarcopenia at the national level, and the results can be used in evaluating health system interventions and policymaking in the field of musculoskeletal diseases.

Representativeness of survey participants in relation to mental disorders: a linkage between national registers and a population-representative survey.

Surveys and registers have provided important insights into the mental health of the community. However, both sources have strengths and limitations. While participation in surveys has been shown to be lower among those with mental disorders, misclassification and limited information on confounders are typical issues for registers. To examine if participants of the Central Denmark Region's 2017 How are you? survey were representative of the general population in terms of mental disorder diagnoses. By linking data from the Central Denmark Region's 2017 How are you? survey with the Danish national registers, we compared the frequency of mental disorder diagnoses among (a) participants in the survey (n = 32,417), before and after applying non-response weights, and (b) the entire population who were eligible to participate (n = 1,063,082; 16 years of age or older on 10th January 2017 and registered as living in the Central Denmark Region). Using logistic regression models, we estimated associations between being diagnosed with any mental disorder and nine general medical conditions to assess whether selection into the survey appeared to bias these associations. Based on register data, 10.4% (n = 110,492) of the eligible population had received a diagnosis of any mental disorder prior to the date of this survey. Among the unweighted survey sample, 8.2% (n = 2,648) had received a diagnosis; once non-response weights were applied, this corresponded to 9.5%. Representativeness varied by sex, age and type of mental disorder. For example, people with organic disorders or substance use disorders were generally underrepresented among survey participants of all ages; however, representativeness of common disorders such as mood or neurotic disorders was generally good. With respect to the association of any mental disorder and general medical conditions, we found that estimates were similar for survey samples (both weighted and unweighted) compared to the entire eligible population. People with a previous diagnosis of a mental disorder are slightly underrepresented in the survey. However, this selection bias was minimized when non-response weights were applied. Associations between mental disorders and general medical conditions did not appear to be affected by selection bias. With the application of non-response weights, the survey provided a sample representative of the general population in terms of mental disorder diagnoses.

Developing non-response weights to account for attrition-related bias in a longitudinal pregnancy cohort.

ObjectiveThe prospective cohort study design is ideal for examining diseases of public health importance. A main source of potential bias for longitudinal studies is attrition. In this study, we compare the performance of two models developed to predict sources of attrition and develop weights to adjust for potential bias. ApproachThis study used the All Our Families longitudinal pregnancy cohort of 3351 maternal-infant pairs. Logistic regression models were developed to predict study continuation versus drop-out from baseline to the three-year data collection wave. Two methods of variable selection took place. One method used previous knowledge and content expertise while the second used Least Absolute Shrinkage and Selection Operator (LASSO). Model performance for both methods were compared using area under the receiver operator curve values (AUROC) and calibration plots. Stabilized inverse probability weights were generated using predicted probabilities. Weight performance was assessed using standardized differences with and without weights (unadjusted estimates). ResultsLASSO and investigator prediction models had good and fair discrimination with AUROC of 0.73 (95% Confidence Interval [CI]: 0.71 – 0.75) and 0.69 ( 95% CI: 0.67 – 0.71), respectively. Calibration plots and non significant Hosmer-Lemeshow Goodness of Fit Tests indicated that both the LASSO model (p = 0.10) and investigator model (p = 0.50) were well-calibrated. Unweighted results indicated large (&gt;10%) standardized differences in 15 demographic data variables (range: 11% - 29%), when comparing those who continued in study with those that did not. Weights derived from the LASSO and investigator models reduced standardized differences relative to unadjusted estimates, with ranges of 0.1% - 5.3% and 0.3% - 12.7%, respectively. ConclusionThe data-driven approach produced robust weights that addressed non-response bias more than the knowledge-driven approach. The data driven approach, did, however still require content knowledge in how data were grouped, combined, or split. The weights can be applied to analyses across multiple waves of data collection to reduce bias.

Methods for conducting trends analysis: roadmap for comparing outcomes from three national HIV Population-based household surveys in Kenya (2007, 2012, and 2018)

BackgroundFor assessing the HIV epidemic in Kenya, a series of independent HIV indicator household-based surveys of similar design can be used to investigate the trends in key indicators relevant to HIV prevention and control and to describe geographic and sociodemographic disparities, assess the impact of interventions, and develop strategies. We developed methods and tools to facilitate a robust analysis of trends across three national household-based surveys conducted in Kenya in 2007, 2012, and 2018.MethodsWe used data from the 2007 and 2012 Kenya AIDS Indicator surveys (KAIS 2007 and KAIS 2012) and the 2018 Kenya Population-based HIV Impact Assessment (KENPHIA 2018). To assess the design and other variables of interest from each study, variables were recoded to ensure that they had equivalent meanings across the three surveys. After assessing weighting procedures for comparability, we used the KAIS 2012 nonresponse weighting procedure to revise normalized KENPHIA weights. Analyses were restricted to geographic areas covered by all three surveys. The revised analysis files were then merged into a single file for pooled analysis. We assessed distributions of age, sex, household wealth, and urban/rural status to identify unexpected changes between surveys.To demonstrate how a trend analysis can be carried out, we used continuous, binary, and time-to-event variables as examples. Specifically, temporal trends in age at first sex and having received an HIV test in the last 12 months were used to demonstrate the proposed analytical approach. These were assessed with respondent-specific variables (age, sex, level of education, and marital status) and household variables (place of residence and wealth index). All analyses were conducted in SAS 9.4, but analysis files were created in Stata and R format to support additional analyses.ResultsThis study demonstrates trends in selected indicators to illustrate the approach that can be used in similar settings. The incidence of early sexual debut decreased from 11.63 (95% CI: 10.95–12.34) per 1,000 person-years at risk in 2007 to 10.45 (95% CI: 9.75–11.2) per 1,000 person-years at risk in 2012 and to 9.58 (95% CI: 9.08–10.1) per 1,000 person-years at risk in 2018. HIV-testing rates increased from 12.6% (95% CI: 11.6%–13.6%) in 2007 to 56.1% (95% CI: 54.6%–57.6%) in 2012 but decreased slightly to 55.6% [95% CI: 54.6%–56.6%) in 2018. The decrease in incidence of early sexual debut could be convincingly demonstrated between 2007 and 2012 but not between 2012 and 2018. Similarly, there was virtually no difference between HIV Testing rates in 2012 and 2018.ConclusionsOur approach can be used to support trend comparisons for variables in HIV surveys in low-income settings. Independent national household surveys can be assessed for comparability, adjusted as appropriate, and used to estimate trends in key indicators. Analyzing trends over time can not only provide insights into Kenya’s progress toward HIV epidemic control but also identify gaps.

Disability and Comorbidity of Mood Disorders and Anxiety Disorders With Diabetes and Hypertension: Evidences From the China Mental Health Survey and Chronic Disease Surveillance in China.

BackgroundThe China Mental Health Survey was carried out using the same sampling frame with the China Chronic Diseases and Risk Factors Surveillance. This paper explores the relationship between the disability and the comorbidity of mood disorders and anxiety disorders with diabetes and hypertension.MethodsA large-scale nationally representative sample with both mental disorders and chronic diseases was collected from 157 Disease Surveillance Points in 31 provinces across China. Face-to-face interviews were conducted by trained lay interviewers to make diagnoses of mood disorders and anxiety disorders using the Composite International Diagnostic Interview. Diabetes and hypertension were diagnosed from self-report and blood examination or body measurement. Sampling design weights, non-response adjustment weights, and post-stratification adjustment weights were applied during the analyses of comorbidity and disability.ResultsTotally 15,000 respondents had information of mental disorders and physical diseases. In the patients with mood disorders or anxiety disorders, the weighted prevalence rates of diabetes or hypertension were not higher than those in persons without the above mental disorders, but the weighed disability rates increased when having the comorbidity of hypertension (P < 0.05). The severity of disability was higher among patients with comorbidity of diabetes and anxiety disorders, or hypertension and mood disorders, compared with that among patients without the physical comorbidity (P < 0.05). After adjusted by age, gender and education, patients with comorbidity of mental disorders and physical disorders had the highest disability, followed by the patients with mental disorders only, and physical diseases only.ConclusionsThe disability of mood disorders and anxiety disorders comorbid with diabetes and hypertension are more serious than that of any single disease. The relationship of mental and physical diseases is worth exploring in depth for comprehensive and integrated intervention to decrease the disability.

The impact of non-response weighting in health surveys for estimates on primary health care utilization.

BackgroundLow response rates in health surveys may affect the representativeness and generalizability of results if non-response is systematically related to the indicator of interest. To account for such potential bias, weighting procedures are widely used with an overall aim to obtain less biased estimates. The aim of this study was to assess the impact of applying calibrated weights on prevalence estimates of primary health care utilization among respondents compared to the entire sample of a representative Danish survey of adults aged ≥16 years.MethodsRegistry-based 1-year prevalence data on health care utilization of chiropractor/physiotherapist, dentist and psychologist in 2016 were linked to the entire sample (n = 312 349), including respondents (n = 183 372), from the Danish National Health Survey in 2017. Calibrated weights, which applied information on e.g. sex, age, ethnic background, education and overall health service use were used to assess their impact on prevalence estimates among respondents.ResultsAcross all included types of health care, weighting for non-response decreased prevalence estimates among respondents, which resulted in less biased estimates. For example, the overall 1-year prevalence of chiropractor/physiotherapist, dentist and psychologist utilization decreased from 19.1% to 16.9%, 68.4% to 62.5% and 1.9% to 1.8%, respectively. The corresponding prevalence in the entire sample was 16.5%, 59.4% and 1.7%.ConclusionsApplying calibrated weights to survey data to account for non-response reduces bias in primary health care utilization estimates. Future studies are needed to explore the possible impact of weighting on other health estimates.

A Bootstrap Variance Estimation Method for Multistage Sampling and Two-Phase Sampling When Poisson Sampling Is Used at the Second Phase

The bootstrap method is often used for variance estimation in sample surveys with a stratified multistage sampling design. It is typically implemented by producing a set of bootstrap weights that is made available to users and that accounts for the complexity of the sampling design. The Rao–Wu–Yue method is often used to produce the required bootstrap weights. It is valid under stratified with-replacement sampling at the first stage or fixed-size without-replacement sampling provided the first-stage sampling fractions are negligible. Some surveys use designs that do not satisfy these conditions. We propose a simple and unified bootstrap method that addresses this limitation of the Rao–Wu–Yue bootstrap weights. This method is applicable to any multistage sampling design as long as valid bootstrap weights can be produced for each distinct stage of sampling. Our method is also applicable to two-phase sampling designs provided that Poisson sampling is used at the second phase. We use this design to model survey nonresponse and derive bootstrap weights that account for nonresponse weighting. The properties of our bootstrap method are evaluated in three limited simulation studies.

Changes in the behavioural determinants of health during the COVID-19 pandemic: gender, socioeconomic and ethnic inequalities in five British cohort studies

BackgroundThe COVID-19 pandemic is expected to have far-reaching consequences on population health. We investigated whether these consequences included changes in health-impacting behaviours which are important drivers of health inequalities.MethodsUsing data...

Association between Social Integration and Face Mask Use Behavior during the SARS-CoV-2 Pandemic in Japan: Results from U-CORONA Study.

Face mask use is a critical behavior to prevent the spread of SARS-CoV-2. We aimed to evaluate the association between social integration and face mask use during the COVID-19 pandemic in a random sample of households in Utsunomiya City, Greater Tokyo, Japan. Data included 645 adults in the Utsunomiya COVID-19 seROprevalence Neighborhood Association (U-CORONA) study, which was conducted after the first wave of the pandemic, between 14 June 2020 and 5 July 2020, in Utsunomiya City. Social integration before the pandemic was assessed by counting the number of social roles, based on the Cohen’s social network index. Face mask use before and during the pandemic was assessed by questionnaire, and participants were categorized into consistent mask users, new users, and current non-users. Multinomial logistic regression analysis was used to examine the association between lower social integration score and face mask use. To account for possible differential non-response bias, non-response weights were used. Of the 645 participants, 172 (26.7%) were consistent mask users and 460 (71.3%) were new users, while 13 (2.0%) were current non-users. Lower social integration level was positively associated with non-users (RRR: 1.76, 95% CI: 1.10, 2.82). Social integration may be important to promote face mask use.

Evaluating access to health and care services during lockdown by the COVID-19 survey in five UK national longitudinal studies

ObjectiveAccess to health services and adequate care is influenced by sex, ethnicity, socioeconomic position (SEP) and the burden of comorbidities. Our study aimed to assess whether the COVID-19 pandemic further...