A systematic literature review of how and whether social media data can complement traditional survey data to study public opinion

In this article, we review existing research on the complementarity of social media data and survey data for the study of public opinion. We start by situating our review in the extensive literature (N = 187) about the uses, challenges, and frameworks related to the use of social media for studying public opinion. Based on 187 relevant articles (141 empirical and 46 theoretical) - we identify within the 141 empircal ones six main research approaches concerning the complementarity of both data sources. Results show that the biggest share of the research has focused on how social media can be used to confirm survey findings, especially for election predictions. The main contribution of our review is to detail and classify other growing complementarity approaches, such as comparing both data sources on a given phenomenon, using survey measures as a proxy in social media research, enriching surveys with SMD, recruiting individuals on social media to conduct a second survey phase, and generating new insight on "old" or "under-investigated" topics or theories using SMD. We discuss the advantages and disadvantages associated with each of these approaches in relation to four main research purposes, namely the improvement of validity, sustainability, reliability, and interpretability. We conclude by discussing some limitations of our study and highlighting future paths for research.

Improved estimates for individual and population-level alcohol use in the United States, 1984-2020

Aims

While nationally representative alcohol surveys are a mainstay of public health monitoring, they underestimate consumption at the population level. This paper demonstrates how to adjust individual-level survey data using aggregated alcohol per capita (APC) data for improved individual- and population-level consumption estimates.

Design and Methods

For the period 1984-2020, data on self-reported alcohol consumption in the past 30 days were taken from the Behavioral Risk Factor Surveillance System (BRFSS) involving participants (18+ years) in the United States (US). Monthly abstainers were reallocated into lifetime abstainers, former drinkers, and 12-month drinkers using the 2005 National Alcohol Survey data. To correct for under-coverage of alcohol use, we triangulated APC and survey data by upshifting quantity (average grams/day) and frequency (drinking days/week) of alcohol use based on national- and state-level APC data. Results were provided for the US as a whole and for selected states to represent different drinking patterns.

Findings

The corrections described above resulted in improved correspondence between survey and APC data. Following our procedure, national estimates of alcohol quantity increased from 45% to 77% of APC estimates. Both quantity and frequency of alcohol use were upshifted; by upshifting to 90% of APC, we were able to fit trends and distributions in APC patterns for individual states and the US.

Conclusions

An individual-level dataset which more accurately reflects the alcohol use of US citizens was achieved. This dataset will be invaluable as a research tool and for the planning and evaluation of alcohol control policies for the US. The methodology described can also be used to adjust individual-level alcohol survey data in other geographical settings.

Innovative methods to analyse the impact of gender norms on adolescent health using global health survey data

Background

Understanding how gender norms affect health is an important entry point into designing programs and policies to change norms and improve gender equality and health. However, it is rare for global health datasets to include questions on gender norms, especially questions that go beyond measuring gender-related attitudes, thus limiting gender analysis.

Methods

We developed five case studies using health survey data from six countries to demonstrate approaches to defining and operationalising proxy measures and analytic approaches to investigating how gender norms can affect health. Key findings, strengths and limitations of our norms proxies and methodological choices are summarised.

Findings

Case studies revealed links between gender norms and multiple adolescent health outcomes. Proxys for norms were derived from data on attitudes, beliefs, and behaviours, as well as differences between attitudes and behaviours. Data were cross-sectional, longitudinal, census- and social network-based. Analytic methods were diverse. We found that gender norms affect: 1) Intimate partner violence in Nigeria; 2) Unhealthy weight control behaviours in Brazil and South Africa; 3) HIV status in Zambia; 4) Health and social mobility in the US; and 5) Childbirth in Honduras.

Interpretation

Researchers can use existing global health survey data to examine pathways through which gender norms affect health by generating proxies for gender norms. While direct measures of gender norms can greatly improve the understanding of how gender affects health, proxy measures for norms can be designed for the specific health-related outcome and normative context, for instance by either aggregating behaviours or attitudes or quantifying the difference (dissonance) between them. These norm proxies enable evaluations of the influence of gender norms on health and insights into possible reference groups and sanctions for non-compliers, thus informing programmes and policies to shape norms and improve health.

•

This article presents effective methods to study gender norms in existing global health survey data.

•

We devised conceptual pathways linking gender norms to gender-based health disparities.

•

We identified gender norms proxies and reference groups enforcing the norm.

•

We tested hypotheses linking gender norms to health.

•

These methods can aid policy and programme design to advance gender equality and health.

Children with genetic conditions in the United States: Prevalence estimates from the 2016-2017 National Survey of Children's Health

PURPOSE

Estimating the overall prevalence of genetic conditions among children in the United States and the burden of these conditions on children and their families has been challenging. The redesigned National Survey of Children's Health provides an opportunity to examine the prevalence and burden.

METHODS

We used the combined 2016-2017 National Survey of Children's Health to estimate the prevalence of genetic conditions among children aged 0 to 17 years (N = 71,522). Bivariate analyses were used to assess differences in sociodemographic characteristics, health-related characteristics, and health care utilization between children with and without genetic conditions.

RESULTS

In 2016-2017, the prevalence of children aged 0 to 17 years with a reported genetic condition was approximately 0.039, roughly equating to 2.8 million children. A greater percentage of children with genetic conditions had a physical (50.9% vs 24.8%), mental (27.9% vs 5.8%), or behavioral/developmental/intellectual condition (55.6% vs 14.4%) than children without a genetic condition. Furthermore, they used more care and had more unmet health needs (7.6% vs 2.9%).

CONCLUSION

This study provides an estimate of the overall prevalence of children living with genetic conditions in the United States based on a nationally representative sample. It also highlights the physical, mental, and behavioral health needs among children with genetic conditions and their unmet health care needs.

How to Measure Retirement Age? A Comparison of Survey and Register Data

Due to an increasing heterogeneity in retirement transitions, the measurement of retirement age constitutes a major challenge for researchers and policymakers. In order to better understand the concept of retirement age, we compare a series of measures for retirement age assessed on the basis of survey and register data. We use data from Sweden, where flexible retirement schemes are implemented and register data are available. We link survey data from the Swedish Level of Living Survey with register data from the Swedish Longitudinal Integration Database for Health Insurance and Labour Market Studies. We create four measures of retirement age based on these datasets, applying approaches that have been used in previous literature. We analyse the means and distributions of these measures and evaluate the correlations between them. Finally, we regress common predictors of retirement age such as gender or education on the four measures of retirement age to examine potential differences in size, direction and statistical significance of the associations. We find that the survey measure of retirement age resembles the following two ways of defining retirement age in the register data: first, the age at which people receive more than half their income from old-age or disability pension and, second, the age at which they were not gainfully employed for at least 2 years. This insight gives us a better understanding of when in the retirement transition process, individuals identify with retirement. Moreover, it provides decision support for researchers working with register data to determine which measure to use.

Biases in Survey Estimates of Neonatal Mortality: Results From a Validation Study in Urban Areas of Guinea-Bissau

Neonatal deaths (occurring within 28 days of birth) account for close to one-half of all deaths among children under age 5 worldwide. In most low- and middle-income countries, data on neonatal deaths come primarily from household surveys. We conducted a validation study of survey data on neonatal mortality in Guinea-Bissau (West Africa). We used records from an urban health and demographic surveillance system (HDSS) that monitors child survival prospectively as our reference data set. We selected a stratified sample of 599 women aged 15-49 among residents of the HDSS and collected the birth histories of 422 participants. We cross-tabulated survey and HDSS data. We used a mathematical model to investigate biases in survey estimates of neonatal mortality. Reporting errors in survey data might lead to estimates of the neonatal mortality rate that are too high, which may limit our ability to track progress toward global health objectives.

Socio-Demographics, Gambling Participation, Gambling Settings, and Addictive Behaviors Associated with Gambling Modes: A Population-Based Study

Online and land-based gambling differ in terms of participation and harms. Multimode gambling has also been distinguished as a separate mode. The current study uses the Finnish Gambling 2019 population study sample of 18–74-year-old past-year gamblers (N = 3,077) to evaluate how these gambling modes differ in terms of socio-demographics, gambling participation, gambling settings, and addictive behaviors. We used land-based gambling as the reference group in a multinomial regression model. Male gender (OR 1.48), age between 18 and 54 (OR 1.88), and high income (OR 1.87) were associated with online gambling. The odds of online gambling were higher among those who gambled at least monthly (OR 1.34) and among those with the highest gambling spending (OR 3.62). Younger age (OR 2.31), high income (OR 1.51), gambling at least four game types (OR 2.96), spending the most money on gambling (OR 4.56), and gambling in at least three gambling settings were associated with multimode gambling. Socio-demographics and gambling participation were indicators of gambling modes. Online gambling was more intensive while multimode gambling was more frequent and versatile than land-based gambling. However, this was not reflected as increased addictive behaviors, probably due to the harmful nature of Finnish land-based gambling.

Serial cross-sectional data on the public's perception on the coronavirus during the first months of the pandemic in Germany

The coronavirus pandemic poses major challenges for governments and public health authorities. In order to implement appropriate measures, it is important to understand how the population is coping with the pandemic. This dataset contains serial cross-sectional survey data from the first months of the coronavirus pandemic in Germany. Data were collected between 24 March and 26 May 2020 in ten weekly surveys (ns ranging between 500 and 515, in total N = 5,076) as part of omnibus telephone interviews. Samples were drawn at random from landline and mobile telephone numbers. The main topics of the questionnaire were (a) the expected impact of the coronavirus on one's personal life, (b) perception of infection risk, (c) protective measures and (d) information behaviour. Data were weighted to ensure sociodemographic representativeness. To account for the rapidly changing situation of the coronavirus pandemic in Germany, the questionnaire underwent several adjustments during the data collection period.

Utility of linking survey and registry data to evaluate interventions and policies to address disparities in breast cancer survivorship among young women

PURPOSE

There is limited research linking data sources to evaluate the multifactorial impacts on the quality of treatment received and financial burden among young women with breast cancer. To address this gap and support future evaluation efforts, we examined the utility of combining patient survey and cancer registry data.

PATIENT AND METHODS

We administered a survey to women, aged 18-39 years, with breast cancer from four U.S. states. We conducted a systematic response-rate analysis and evaluated differences between racial groups. Survey responses were linked with cancer registry data to assess whether surveys could reliably supplement registry data.

RESULTS

A total of 830 women completed the survey for a response rate of 28.4 %. Blacks and Asian/Pacific Islanders were half as likely to respond as white women. Concordance between survey and registry data was high for demographic variables (Cohen's kappa [k]: 0.879 to 0.949), moderate to high for treatments received (k: 0.467 to 0.854), and low for hormone receptor status (k: 0.167 to 0.553). Survey items related to insurance status, employment, and symptoms revealed racial differences.

CONCLUSION

Cancer registry data, supplemented by patient surveys, can provide a broader understanding of the quality of care and financial impacts of breast cancer among young women.

Is South Africa closing the health gaps between districts? Monitoring progress towards universal health service coverage with routine facility data

BACKGROUND

South Africa is committed to advancing universal health coverage (UHC). The usefulness and potential of using routine health facility data for monitoring progress towards UHC, in the form of the 16-tracer WHO service coverage index (SCI), was assessed.

METHODS

Alternative approaches to calculating the WHO SCI from routine data, allowing for disaggregation to district level, were explored. Data extraction, coding, transformation and modelling processes were applied to generate time series for these alternatives. Equity was assessed using socio-economic quintiles by district.

RESULTS

The UHC SCI at a national level was 46.1 in 2007-2008 and 56.9 in 2016-2017. Only for the latter period, could the index be calculated for all indicators at a district level. Alternative indicators were formulated for 9 of 16 tracers in the index. Routine or repeated survey data could be used for 14 tracers. Apart from the NCD indicators, a gradient of poorer performance in the most deprived districts was evident in 2016-2017.

CONCLUSIONS

It is possible to construct the UHC SCI for South Africa from predominantly routine data sources. Overall, there is evidence from district level data of a trend towards reduced inequity in relation to specific categories (notably RMNCH). Progress towards UHC has the potential to overcome fragmentation and enable harmonisation and interoperability of information systems. Private sector reporting of data into routine information systems should be encouraged.

Patient Experiences With Telemedicine in a National Health Service Rheumatology Outpatient Department During Coronavirus Disease-19

The coronavirus disease-19 pandemic changed rheumatology practice with remote consultations being increasingly utilized where appropriate. We evaluated patient experiences with telephone consultations and report on patient attitudes toward current health care delivery and perspectives of telemedicine in a UK National Health Service rheumatology outpatient department. We analyzed 297 questionnaires from a postal survey conducted during the summer of 2020 after a telephone follow-up consultation. The mean age of respondents was 67 years and 68% were female. The 161 respondents (54%) reported it was their first telephone consultation and overall, 239 (84%) were satisfied with their health assessment. 60% would be happy to have future routine follow-up telephone consultations. Patients advised to shield shared similar satisfaction to the whole sample. However, with increasing age we identified a higher proportion were dissatisfied with telephone consultations and unlikely to have accessibility to video consultation or preferentially opt for this modality.

Estimating perceived parental substance use disorder: Using register data to adjust for non-participation in survey research

AIMS

To estimate the prevalence of parental substance use disorder (PSUD) in the general population based on young adults' reports adjusted for non-participation using register-based indicators of PSUD.

DESIGN

A national sample survey study combined with a retrospective register-based study. Setting Denmark. Participants 10,414 young people (aged 15-25 years) invited to two national sample surveys in 2014 and 2015 (5,755 participants and 4,659 non-participants).

MEASUREMENTS

A crude prevalence of PSUD was calculated based on participants' reports. Parental data from medical, mortality, prescription, and treatment registers (from the young adults' birth until the time of the surveys) were used to estimate a register-based prevalence of PSUD for both participants and non-participants. Differences between participants and non-participants were analysed using bivariate comparisons. Inverse probability weighting was used to adjust for bias due to non-participation. The crude prevalence of PSUD based on survey data was adjusted using the ratio of incidence proportion of the register-based PSUD compared with the survey-based PSUD.

FINDINGS

A total of 731 (12.7%) of the 5,755 survey participants reported PSUD. Register-based PSUD was more common among non-participants (856/4,659; 18.4%) compared with participants (738/5,755; 12.8%, OR = 1.53, 95% CI 1.38-1.70). The adjusted estimate of the survey-based PSUD increased by 2.5 percentage points, from 12.7% to 15.2%.

CONCLUSIONS

In the absence of register data, youth-reported PSUD is likely to underestimate the number of young people experiencing PSUD.

Adult death registration in Matlab, rural Bangladesh: completeness, correlates, and obstacles

Civil registration of vital events such as deaths and births is a key part of the process of securing rights and benefits for individuals worldwide. It also enables the production of vital statistics for local planning of social services. In many low- and lower-middle-income countries, however, civil registration and vital statistics (CRVS) systems do not adequately register significant numbers of births and, especially, deaths. In this study, we aim to estimate the completeness of adult death registration (for age 15 and older) in the Matlab health and demographic surveillance system (HDSS) area in Bangladesh and to identify reasons for (not) registering deaths in the national CRVS system. We conducted a sample survey of 2538 households and recorded 571 adult deaths that had occurred in the 3 years preceding the survey. Only 17% of these deaths were registered in the national CRVS system, with large gender differences in registration rates (male = 26% vs. female = 5%). Respondents who reported that a recent death in the household was registered indicated that the primary reasons for registration were to secure an inheritance and to access social services. The main reasons cited for not registering a death were lack of knowledge about CRVS and not perceiving the benefits of death registration. Information campaigns to raise awareness of death registration, as well as stronger incentives to register deaths, may be needed to improve the completeness of death registration in Bangladesh.

SUPPLEMENTARY INFORMATION

The online version contains supplementary material available at 10.1186/s41118-021-00125-7.

Changes in pregnancy desire after a pregnancy scare in a random sample of young adult women in a Michigan county

OBJECTIVES

We examined whether and how long young women became more or less likely to desire a pregnancy after experiencing a "pregnancy scare."

STUDY DESIGN

We used data from the Relationship Dynamics and Social Life (RDSL) study, based on a random, population-based sample of 992 young women from a county in Michigan. They were interviewed weekly for 2.5 years. We used fixed-effects logistic regression models to predict pregnancy desire after a pregnancy scare.

RESULTS

Of the 759 sexually experienced women we analyzed, 103 (14%) experienced 128 pregnancy scares. A woman's (adjusted) odds of desiring a pregnancy were 3.70 (95% CI 2.27-6.02) times higher during the week after, 3.04 (95% CI 2.30-4.10) times higher during the month after a pregnancy scare, and 2.31 (95% CI 1.71-3.11) times higher during all weeks after the pregnancy scare, compared to her other weeks during the study period. In a final model directly comparing each period to all weeks before the pregnancy scare, the odds of pregnancy desire were highest (aOR 5.08, 95% CI 3.06-8.42) during the first week, slightly smaller (aOR 3.01, 95% CI 2.11 - 4.30) during the subsequent three weeks, and remained elevated (aOR 1.58, 95% CI 1.19-2.09) throughout the remainder of the study period.

CONCLUSIONS

Our analyses suggest that the experience of a pregnancy "scare" does not scare young women away from wanting pregnancies. On the contrary, the state of possibly being pregnant actually made young women in our study more likely to want to be pregnant, on average.

IMPLICATIONS

Very few young women desire a pregnancy during the transition to adulthood; however, a salient life event like a pregnancy scare can abruptly generate a desire for pregnancy. Our study contributes to efforts to help women implement their pregnancy desires by furthering our understanding of those desires and the contexts in which they are formed.

Deferred Action for Childhood Arrivals (DACA) medical students - an examination of their journey and experiences as medical students in limbo

BACKGROUND

Although the value of DACA medical students has been hypothesized, no data are available on their contribution to US healthcare. While the exact number of DACA recipients in medical school is unknown, DACA medical students are projected to represent an increasing proportion of physicians in the future. The current literature on DACA students has not analyzed the experiences of these students.

METHODS

A mixed-methods study on the career intentions and experiences of DACA medical students was performed utilizing survey data and in-depth interviews. The academic performance of a convenience sample of DACA medical students was compared to that of matriculated medical students from corresponding medical schools, national averages, and first-year residents according to specialty.

RESULTS

Thirty-three DACA medical students completed the survey and five participated in a qualitative interview. The average undergraduate GPA (SD) of the DACA medical student sample was 3.7 (0.3), the same as the national GPA of 2017-2018 matriculated medical students. The most common intended residency programs were Internal Medicine (27.2%), Emergency Medicine (15.2%), and Family Medicine (9.1%). In interviews, DACA students discussed their motivation for pursuing medicine, barriers and facilitators that they faced in attending medical school, their experiences as medical students, and their future plans.

CONCLUSIONS

The intent of this sample to pursue medical specialties in which there is a growing need further exemplifies the unique value of these students. It is vital to protect the status of DACA recipients and realize the contributions that DACA physicians provide to US healthcare.

A note on multiply robust predictive mean matching imputation with complex survey data

Predictive mean matching is a commonly used imputation procedure for addressing the problem of item nonrespone in surveys. The customary approach relies upon the specification of a single outcome regression model. In this note, we propose a novel predictive mean matching procedure that allows the user to specify multiple outcome regression models. The resulting estimator is multiply robust in the sense that it remains consistent if one of the specified outcome regression models is correctly specified. The results from a simulation study suggest that the proposed method performs well in terms of bias and efficiency.

Short-term financial constraints and SMEs' investment decision: evidence from the working capital channel

This paper investigates the real effect of short-term financial constraints on small and medium-sized enterprises (SMEs). Under the working capital channel, cash credit constraints force entrepreneurs to forgo investment opportunities in order to finance their working capital needs. Building on survey data, I find that short-term credit constraints are as important as long-term ones in SMEs' investment decisions. Besides, the detrimental effect of cash credit constraints on corporate investment is stronger for SMEs with higher increase in working capital needs but lower for liquid ones that are able to adjust their accounts receivable and inventories. My results suggest that short-term finance is a major issue for SMEs. Plain English Summary Short-term credit constraints turn out to be as important as long-term ones in SMEs' investment decisions. Building on survey data, this paper investigates the real effects of short-term financial constraints on small businesses through an under-explored transmission mechanism. While the bulk of research on the effects of financial constraints focuses on long-term liabilities, I stress the importance of short-term finance. Owing to the competition between working and fixed capital in cash-flow uses, cash credit constraints force entrepreneurs to allocate additional cash-flow to finance the increase in their working capital needs to the detriment of long-term assets.The main implication of this work is that supports for short-term financing would provide a significant boost to long-term investment and economic growth.

Did state-owned enterprises do better during COVID-19? Evidence from a survey of company executives in China

In a survey of 1,182 company executives in China, state-owned enterprises (SOEs) reported less business reductions under COVID-19. This paper examines if SOEs' superior performance was resulted from government support rather than innate ability of coping with the pandemic. We construct a proxy for firm-level government support using firm's human resources (HR) action taken during the outbreak with firm's 2019 China revenue share as an instrument for the HR action variable. After controlling for the proxy for firm-level government support as well as other observed firm characteristics, we find SOEs in the sample performing significantly worse in the pandemic period.

A survey dataset to identify industry practices and challenges for mine rehabilitation completion criteria in Western Australia

The development of acceptable and achievable completion criteria is fundamental to the successful relinquishment of mined land to a post-mining use. Despite the central role of completion criteria, there is still a need to build capacity and understanding of how to set targets and develop measurable completion criteria that are accepted by all stakeholders involved. The work described in this paper aimed to elicit industry practice, barriers, and opportunities for the development of feasible and acceptable completion criteria. We developed a quantitative survey that was administered online. The target respondents consisted of mining companies, consulting businesses, and relevant regulators in Western Australia. The survey questionnaire, raw survey data, and summary statistics are provided in this paper to increase research transparency and facilitate reproducibility of the methods by researchers in other jurisdictions.

Using ranked survey data in education research: Methods and applications

Researchers and practitioners in education and school psychology regularly use ranked data to drive student- and systems-level decision-making. These types of data can be derived from assessments of individual preferences among researchers and practitioners, assessments of preferences among stakeholders including parents and children, and rankings of students on academic and social-emotional competency. However, the analysis of ranked data in education and psychology has typically been limited to simple approaches such as the examination of mean ranks assigned to items. This paper unifies a collection of classical methodologies, as well as proposes new techniques, for analyzing ranked data used across disciplines and applies the methods to data generated in school psychological research. The scope of the paper is to serve as a roadmap for researchers in education and school psychology who seek to more fully leverage information contained in ranked data. These methodologies include descriptive analyses, visualizations, tests of uniformity, cluster analyses, and predictive models. We demonstrate these techniques on the survey data of Fefer, DeMagistris, and Shuttleton (2016) and illustrate how using a broader set of tools can yield improved insights by researchers and practitioners.

Assessment of the COVID-19 pandemic situation: Data from two countries with different security measures taken by authorities (Belarus and Russia)

The article presents the data on the evaluation of COVID-19 situation by adult respondents (age from 18 to 76) coming from countries with the common past, language and similar mentality but differing in governmental security measures during pandemic spring outbreak (no precautionary measures in Belarus (N = 267); lockdown and financial support in Russia (N = 397)). The data was collected via online survey platform (Google forms) from 2020.04.11 to 2020.06.04 (during the period of lock-down in Russia). The data contains socio-demographical information (sex, age, country of citizenship), survey answers and the results of standardized psychological tests (to measure dangerous and threatening social world view and hardiness). The survey consists of four blocks: specific impact of the COVID-19 situation on various aspects of respondents’ life; estimation of different fears; estimation of various aspects of COVID-19 situation, and estimation of personal resources. All the items require participants to rate them on a 11-point Likert scale from 0 (totally disagree, absolutely no fear or no impact) to 10 (totally agree, the strongest fear or impact). Descriptive statistics as well as the comparison results are given. The data may be used to investigate the influence of lockdown, social distancing, and isolation on psychological well-being as well as the impact of personal resources in psychological well-being in stressful situations.

Multiple imputation validation study: addressing unmeasured survey data in a longitudinal design

Background

Questionnaires used in longitudinal studies may have questions added or removed over time for numerous reasons. Data missing completely at a follow-up survey is a unique issue for longitudinal studies. While such excluded questions lack information at one follow-up survey, they are collected at other follow-up surveys, and covariances observed at other follow-up surveys may allow for the recovery of the missing data. This study utilized data from a large longitudinal cohort study to assess the efficiency and feasibility of using multiple imputation (MI) to recover this type of information.

Methods

Millennium Cohort Study participants completed the 9-item Patient Health Questionnaire (PHQ) depression module at 2 time points (2004, 2007). The suicidal ideation item in the module was set to missing for the 2007 assessment. Several single-level MI models using different sets of predictors and forms of suicidal ideation were used to compare self-reported values and imputed values for this item in 2007. Additionally, associations with sleep duration and smoking status, which are related constructs, were compared between self-reported and imputed values of suicidal ideation.

Results

Among 63,028 participants eligible for imputation analysis, 4.05% reported suicidal ideation on the 2007 survey. The imputation models successfully identified suicidal ideation, with a sensitivity ranging between 34 and 66% and a positive predictive value between 36 and 42%. Specificity remained above 96% and negative predictive value above 97% for all imputed models. Similar associations were found for all imputation models on related constructs, though the dichotomous suicidal ideation imputed from the model using only PHQ depression items yielded estimates that were closest with the self-reported associations for all adjusted analyses.

Conclusions

Although sensitivity and positive predictive value were relatively low, applying MI techniques allowed for inclusion of an otherwise missing variable. Additionally, correlations with related constructs were estimated near self-reported values. Therefore, the other 8 depression items can be used to estimate suicidal ideation that was completely missing from a survey using MI. However, these imputed values should not be used to estimate population prevalence.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-020-01158-w.

Random survival forests using linked data to measure illness burden among individuals before or after a cancer diagnosis: Development and internal validation of the SEER-CAHPS illness burden index

PURPOSE

To develop and internally validate an illness burden index among Medicare beneficiaries before or after a cancer diagnosis.

METHODS

Data source: SEER-CAHPS, linking Surveillance, Epidemiology, and End Results (SEER) cancer registry, Medicare enrollment and claims, and Medicare Consumer Assessment of Healthcare Providers and Systems (Medicare CAHPS) survey data providing self-reported sociodemographic, health, and functional status information. To generate a score for everyone in the dataset, we tabulated 4 groups within each annual subsample (2007-2013): 1) Medicare Advantage (MA) beneficiaries or 2) Medicare fee-for-service (FFS) beneficiaries, surveyed before cancer diagnosis; 3) MA beneficiaries or 4) Medicare FFS beneficiaries surveyed after diagnosis. Random survival forests (RSFs) predicted 12-month all-cause mortality and drew predictor variables (mean per subsample = 44) from 8 domains: sociodemographic, cancer-specific, health status, chronic conditions, healthcare utilization, activity limitations, proxy, and location-based factors. Roughly two-thirds of the sample was held out for algorithm training. Error rates based on the validation ("out-of-bag," OOB) samples reflected the correctly classified percentage. Illness burden scores represented predicted cumulative mortality hazard.

RESULTS

The sample included 116,735 Medicare beneficiaries with cancer, of whom 73 % were surveyed after their cancer diagnosis; overall mean mortality rate in the 12 months after survey response was 6%. SEER-CAHPS Illness Burden Index (SCIBI) scores were positively skewed (median range: 0.29 [MA, pre-diagnosis] to 2.85 [FFS, post-diagnosis]; mean range: 2.08 [MA, pre-diagnosis] to 4.88 [MA, post-diagnosis]). The highest decile of the distribution had a 51 % mortality rate (range: 29-71 %); the bottom decile had a 1% mortality rate (range: 0-2 %). The error rate was 20 % overall (range: 9% [among FFS enrollees surveyed after diagnosis] to 36 % [MA enrollees surveyed before diagnosis]).

CONCLUSIONS

This new morbidity measure for Medicare beneficiaries with cancer may be useful to future SEER-CAHPS users who wish to adjust for comorbidity.

A survey dataset to evaluate the changes in mobility and transportation due to COVID-19 travel restrictions in Australia, Brazil, China, Ghana, India, Iran, Italy, Norway, South Africa, United States

COVID-19 pandemic has heavily impacted the global community. To curb the viral transmission, travel restrictions have been enforced across the world. The dataset documents the mobility disruptions and the modal shifts that have occurred as a consequence of the restrictive measures implemented in ten countries: Australia, Brazil, China, Ghana, India, Iran, Italy, Norway, South Africa and the United States. An online questionnaire was distributed during the period from the 11st to the 31st of May 2020, with a total of 9 394 respondents. The first part of the survey has characterized the frequency of use of all transport modes before and during the enforcement of the restrictions, while the second part of the survey has dealt with perceived risks of contracting COVID-19 from different transport modes and perceived effectiveness of travel mitigation measures. Overall, the dataset (stored in a repository publicly available) can be conveniently used to quantify and understand the modal shifts and people's cognitive behavior towards travel due to COVID-19. The collected responses can be further analysed by considering other demographic and socioeconomic covariates.

Data on antimicrobial use in livestock: Lessons from Uganda

In 2016, the General Assembly of the United Nations recognised inappropriate Antimicrobial Use (AMU) in livestock as one of the leading causes of increasing Antimicrobial Resistance (AMR). This is happening at a time when livestock production is expected to increase dramatically particularly in Africa, in response to the large rise in aggregate demand due to population growth, urbanisation and increasing income levels. Therefore, understanding the characteristics and appropriateness of AMU in livestock in this region is of utmost importance, yet data is seldom available. We propose to collect information on AMU in livestock by including related questions in nationally representative agricultural surveys that are carried out regularly (annually or every 2-3 years) by National Statistical Offices. This approach, with its limitation though, is a viable and cost-effective way to gather essential information on AMU in livestock farming. The Uganda Bureau of Statistics (UBOS) in collaboration with the Food and Agriculture Organization of the United Nations (FAO) piloted the introduction of key AMU questions in the Annual Agricultural Survey (AAS), gathering data from 6 thousand agricultural households nation-wide. Results show that AMU is considerable among livestock keeping households (35%), who use antibiotics not only for curative treatment (~58%) but also for disease prevention (~44%) and growth promotion (~5%). Data from the AAS also allows users to explore linkages between antibiotics use, livestock production practices (e.g. herd composition and size, feeding, breeding techniques, etc.) and other household / farm characteristics (e.g. location, education, household size, etc.), thereby effectively informing policy decisions.