Are we missing the importance of missing values in HIV prevention randomized clinical trials? Review and recommendations

smdi: an R package to perform structural missing data investigations on partially observed confounders in real-world evidence studies

Objectives

Partially observed confounder data pose a major challenge in statistical analyses aimed to inform causal inference using electronic health records (EHRs). While analytic approaches such as imputation are available, assumptions on underlying missingness patterns and mechanisms must be verified. We aimed to develop a toolkit to streamline missing data diagnostics to guide choice of analytic approaches based on meeting necessary assumptions.

Materials and methods

We developed the smdi (structural missing data investigations) R package based on results of a previous simulation study which considered structural assumptions of common missing data mechanisms in EHR.

Results

smdi enables users to run principled missing data investigations on partially observed confounders and implement functions to visualize, describe, and infer potential missingness patterns and mechanisms based on observed data.

Conclusions

The smdi R package is freely available on CRAN and can provide valuable insights into underlying missingness patterns and mechanisms and thereby help improve the robustness of real-world evidence studies.

A study on the correlation between family dynamic factors and depression in adolescents

Objectives

To evaluate the relationship between systemic family dynamics and adolescent depression.

Methods

An offline survey was distributed to 4,109 students in grades 6-12, with the final analysis including 3,014 students (1,524 boys and 1,490 girls) aged 10-18 years. The questionnaire included the Self-Rating Scale of Systemic Family Dynamics (SSFD), the Self-Rating Depression Scale (SDS), and demographic characteristics.

Results

Family dynamics were negatively correlated with depressive symptoms, with better family dynamics (high scores) associated with lower levels of depression based on the SDS score. After adjusting for sociodemographic characteristics, an ordinal multiclass logistic regression analysis identified family atmosphere (OR = 0.952, 95% CI: 0.948-0.956, p < 0.001) as the most important protective family dynamic against depression, followed by individuality (OR = 0.964, 95% CI: 0.960-0.968, p < 0.001). Latent class analysis (LCA) created the low family dynamic and high family dynamic groups. There were significant differences in the mean SDS scores between the two groups (45.52 ± 10.57 vs. 53.78 ± 11.88; p < 0.001) that persisted after propensity matching. Family atmosphere and individuation had a favorable diagnostic value for depression, with AUCs of 0.778 (95% CI: 0.760-0.796) and 0.710 (95% CI: 0.690-0.730), respectively. The diagnostic models for depression performed well.

Conclusion

Poor family dynamics may be responsible for adolescent depression. A variety of early intervention strategies focused on the family may potentially avoid adolescent depression.

Debridement of diabetic foot ulcers: public health and clinical implications – a systematic review, meta-analysis, and meta-regression

Background

Diabetic foot ulceration (DFU) has devastating complications and a lifetime occurrence of 15%–34%. Debridement of DFU is regarded as an intervention that accelerates ulcer healing and may reduce complications including amputations, infections, and poor quality of life (QoL), which have serious public health and clinical implications. A systematic review (SR) of SRs and of randomized controlled trials (RCTs) with meta-analyses (MAs) on debridement of DFU that synthesizes all human experimental evidence is warranted.

Objectives

Are debridement methods in DFU beneficial over other forms and standard gauze dressings (control condition) in these outcomes?

Study eligibility criteria

All SRs/MAs/RCTs comparing debridement methods for DFU with alternative methods of debridement and with control.

Data sources

Cochrane Wounds Group Specialized Register, Cochrane Central Register of Controlled Trials (Cochrane Library), Ovid MEDLINE, PubMed, EMBASE, EBSCO, CINAHL, and Web of Science.

Participants and interventions

Adults with type 1/2 diabetes with DFU and any debridement method compared with alternative debridement methods or control.

Main Outcomes

Amputation rates, wound infections, QoL, proportion of ulcers healed, time to complete healing, ulcer recurrence, and treatment cost.

Study selection and analysis

Data extraction/synthesis by two independent reviewers pooled using a random-effects model with sensitivity analysis.

Results

10 SRs were retrieved and reported qualitatively. Six SRs included MAs. This SR included 30 studies, with 2654 participants, using 19 debridement combinations. The debridement methods were compared with findings pooled into MAs. Meta-regression (MR) did not identify significant predictors/moderators of outcomes.

Limitations

The studies may have been under-powered. The inclusion/exclusion criteria varied and the increased risk of bias contributed to low-quality evidence.

Discussion/Conclusion

Weak evidence exists that debridement methods are superior to other forms of debridement or control in DFU.

Implications

Researchers should follow standardized reporting guidelines (Consolidated Standards of Reporting Trials). Clinicians/investigators could use the findings from this SR/MA/MR in guiding patient-individualized decision making and designing future RCTs.

Randomized Community Trial Comparing Telephone versus Clinic-Based Behavioral Health Counseling for People Living with HIV in a Rural Setting

PURPOSE

To test the efficacy of a theory-based behavioral intervention delivered via telephone versus clinic-based counseling to improve HIV outcomes and reduce alcohol consumption for people at-risk for treatment failure in a rural setting.

METHODS

Patients receiving HIV care (N = 240) were randomized using a computer-generated scheme to one of three conditions: (a) telephone behavioral health counseling, (b) clinic-based behavioral health counseling, or (c) attention control nutrition education. Behavioral counseling was delivered by either a community nurse or a paraprofessional patient navigator, with differences examined. Participants were followed for 12 months to assess medication adherence using unannounced pill counts and alcohol use measured by electronic daily text message assessments, and 18 months for HIV viral load and retention in care extracted from medical records.

FINDINGS

There was evidence for telephone and office-based counseling demonstrating greater medication adherence than the control condition but only in the short term. Clinic-based behavioral counseling significantly reduced alcohol use to a greater degree than telephone counseling and the control condition. There were no other differences between conditions. There were also no discernable differences between counseling delivered by the community nurse and the patient navigator.

CONCLUSIONS

Telephone and clinic-based counseling demonstrated improved medication adherence in the short term, while clinic-based counseling demonstrated reductions in alcohol use. The modest outcomes suggest that intensive intervention strategies are needed for patients that clinicians identify as at-risk for treatment discontinuation and treatment failure.

Comparative effects of telephone versus in-office behavioral counseling to improve HIV treatment outcomes among people living with HIV in a rural setting

With the expansion of telehealth services, there is a need for evidence-based treatment adherence interventions that can be delivered remotely to people living with HIV. Evidence-based behavioral health counseling can be delivered via telephone, as well as in-office services. However, there is limited research on counseling delivery formats and their differential outcomes. The purpose of this study was to conduct a head-to-head comparison of behavioral self-regulation counseling delivered by telephone versus behavioral self-regulation counseling delivered by in-office sessions to improve HIV treatment outcomes. Patients (N = 251) deemed at risk for discontinuing care and treatment failure living in a rural area of the southeastern USA were referred by their care provider. The trial implemented a Wennberg Randomized Preferential Design to rigorously test: (a) patient preference and (b) comparative effects on patient retention in care and treatment adherence. There was a clear patient preference for telephone-delivered counseling (69%) over in-office-delivered counseling (31%) and participants who received telephone counseling completed a greater number of sessions. There were few differences between the two intervention delivery formats on clinical appointment attendance, antiretroviral adherence, and HIV viral load. Overall improvements in health outcomes were not observed across delivery formats. Telephone-delivered counseling did show somewhat greater benefit for improving depression symptoms, whereas in-office services demonstrated greater benefits for reducing alcohol use. These results encourage offering most patients the choice of telephone and in-office behavioral health counseling and suggest that more intensive interventions may be needed to improve clinical outcomes for people living with HIV who may be at risk for discontinuing care or experiencing HIV treatment failure.

Analytical methods used in estimating the prevalence of HIV/AIDS from demographic and cross-sectional surveys with missing data: a systematic review

BACKGROUND

Sero- prevalence studies often have a problem of missing data. Few studies report the proportion of missing data and even fewer describe the methods used to adjust the results for missing data. The objective of this review was to determine the analytical methods used for analysis in HIV surveys with missing data.

METHODS

We searched for population, demographic and cross-sectional surveys of HIV published from January 2000 to April 2018 in Pub Med/Medline, Web of Science core collection, Latin American and Caribbean Sciences Literature, Africa-Wide Information and Scopus, and by reviewing references of included articles. All potential abstracts were imported into Covidence and abstracts screened by two independent reviewers using pre-specified criteria. Disagreements were resolved through discussion. A piloted data extraction tool was used to extract data and assess the risk of bias of the eligible studies. Data were analysed through a quantitative approach; variables were presented and summarised using figures and tables.

RESULTS

A total of 3426 citations where identified, 194 duplicates removed, 3232 screened and 69 full articles were obtained. Twenty-four studies were included. The response rate for an HIV test of the included studies ranged from 32 to 96% with the major reason for the missing data being refusal to consent for an HIV test. Complete case analysis was the primary method of analysis used, multiple imputations 11(46%) was the most advanced method used, followed by the Heckman's selection model 9(38%). Single Imputation and Instrumental variables method were used in only two studies each, with 13(54%) other different methods used in several studies. Forty-two percent of the studies applied more than two methods in the analysis, with a maximum of 4 methods per study. Only 6(25%) studies conducted a sensitivity analysis, while 11(46%) studies had a significant change of estimates after adjusting for missing data.

CONCLUSION

Missing data in survey studies is still a problem in disease estimation. Our review outlined a number of methods that can be used to adjust for missing data on HIV studies; however, more information and awareness are needed to allow informed choices on which method to be applied for the estimates to be more reliable and representative.

Multiple Imputation for Incomplete Data in Environmental Epidemiology Research

PURPOSE OF REVIEW

Incomplete data are a common problem in statistical analysis of environmental epidemiological research. However, many researchers still ignore this complication. We evaluate the performance of two commonly used multiple imputation (MI) methods (fully conditional specification and multivariate normal) for handling missing data and compare them to complete case analysis (CCA) method. We further discuss issues that arise when these methods are being used.

RECENT FINDINGS

MI is a simulation-based approach to deal with incomplete data. In general, MI will perform better then ad hoc techniques such as CCA. MI is an approach which replaces the missing data with plausible values and allows for additional uncertainty due to the missing information caused by the incomplete data. To illustrate this, we use data of 944 women from the Collaborative Perinatal Project and compare estimates between these methods. The goal is to examine if each of two outcomes, birth-weight and spontaneous abortion, in the data set are associated with mothers' smoking status during pregnancy adjusting for baseline covariates in the model. Results indicate that MI is better suited for handling incomplete data and led to a significant improvement in parameter estimates compared to CCA. The two MI methods produced similar point estimates, but slightly different standard errors.

Accounting for non-response bias using participation incentives and survey design: An application using gift vouchers

Standard corrections for missing data rely on the strong and generally untestable assumption of missing at random. Heckman-type selection models relax this assumption, but have been criticized because they typically require a selection variable which predicts non-response but not the outcome of interest, and can impose bivariate normality. In this paper we illustrate an application using a copula methodology which does not rely on bivariate normality. We implement this approach in data on HIV testing at a demographic surveillance site in rural South Africa which are affected by non-response. Randomized incentives are the ideal selection variable, particularly when implemented ex ante to deal with potential missing data. However, elements of survey design may also provide a credible method of correcting for non-response bias ex post. For example, although not explicitly randomized, allocation of food gift vouchers during our survey was plausibly exogenous and substantially raised participation, as did effective survey interviewers. Based on models with receipt of a voucher and interviewer identity as selection variables, our results imply that 37% of women in the population under study are HIV positive, compared to imputation-based estimates of 28%. For men, confidence intervals are too wide to reject the absence of non-response bias. Consistent results obtained when comparing different selection variables and error structures strengthen these conclusions. Our application illustrates the feasibility of the selection model approach when combined with survey metadata.

The treatment of incomplete data: Reporting, analysis, reproducibility, and replicability

Proper analysis and reporting of incomplete data continues to be a challenging task for practitioners from various research areas. Recently Nguyen, Strazdins, Nicholson and Cooklin (NSNC; 2018) evaluated the impact of complete case analysis and multiple imputation in studies of parental employment and health. Their work joins interdisciplinary efforts to educate and motivate scientists across the research community to use principled statistical methods when analyzing incomplete data. Although we fully support and encourage work in parallel to NSNC's, we also think that further actions should be taken by the research community to improve current practices. In this commentary, we discuss some aspects and misconceptions related to analysis of incomplete data, in particular multiple imputation. In our view, the missing data problem is part of a larger problem of research reproducibility and replicability today. Thus, we believe that improving analysis and reporting of incomplete data will make reproducibility and replicability efforts easier. We also provide a brief checklist of recommendations which could be used by members of the scientific community, including practitioners, journal editors, and reviewers to set higher publication standards.

Principled Approaches to Missing Data in Epidemiologic Studies

Principled methods with which to appropriately analyze missing data have long existed; however, broad implementation of these methods remains challenging. In this and 2 companion papers (Am J Epidemiol. 2018;187(3):576-584 and Am J Epidemiol. 2018;187(3):585-591), we discuss issues pertaining to missing data in the epidemiologic literature. We provide details regarding missing-data mechanisms and nomenclature and encourage the conduct of principled analyses through a detailed comparison of multiple imputation and inverse probability weighting. Data from the Collaborative Perinatal Project, a multisite US study conducted from 1959 to 1974, are used to create a masked data-analytical challenge with missing data induced by known mechanisms. We illustrate the deleterious effects of missing data with naive methods and show how principled methods can sometimes mitigate such effects. For example, when data were missing at random, naive methods showed a spurious protective effect of smoking on the risk of spontaneous abortion (odds ratio (OR) = 0.43, 95% confidence interval (CI): 0.19, 0.93), while implementation of principled methods multiple imputation (OR = 1.30, 95% CI: 0.95, 1.77) or augmented inverse probability weighting (OR = 1.40, 95% CI: 1.00, 1.97) provided estimates closer to the "true" full-data effect (OR = 1.31, 95% CI: 1.05, 1.64). We call for greater acknowledgement of and attention to missing data and for the broad use of principled missing-data methods in epidemiologic research.

Investigating tangible and mental resources as predictors of perceived household food insecurity during pregnancy among women in a South African birth cohort study

RATIONALE

Food insecurity during pregnancy is concerning given the increased nutritional needs of the mother for proper fetal development. However, research is lacking within the South African context to investigate the association of economic and psychosocial factors and food insecurity among pregnant women, using comprehensive, conceptually driven models.

OBJECTIVE

This study applies the Network-Individual-Resource (NIR) Model to investigate individual, intimate dyadic, and family level predictors of perceived household food insecurity for pregnant women.

METHODS

826 pregnant women enrolled in the Drakenstein Child Health Study (DCHS), a birth cohort in two communities in a peri-urban area of South Africa. Hierarchical logistic regressions were used to investigate the impact of household/family, intimate dyads, and individual tangible and mental resources on perceived household food insecurity during the critical period of pregnancy. Perceived household food insecurity was assessed through an adapted version of the USDA Household Food Security Scale - Short Form.

RESULTS

Among 826 pregnant women in South Africa, individual-level tangible resources (e.g. income, social assistance, HIV status) and mental resources (e. g. depression, childhood trauma) predicted perceived household food insecurity and these predictors differed by community. Intimate dyadic and family level resources did not predict household food insecurity.

CONCLUSIONS

Our findings of the economic and psychosocial predictors of perceived household food insecurity among pregnant women in South Africa, mirror findings in general populations. This study provides support for the extension of the NIR model to perceived household food insecurity, particularly regarding individual-level mental and tangible resources, as well as the impact of community-level factors. Future research should investigate the extent to which resource sharing occurs within networks.

Diminishing Perceived Threat of AIDS and Increasing Sexual Risks of HIV Among Men Who Have Sex with Men, 1997-2015

Community-wide awareness that antiretroviral therapies (ART) provides protection against HIV has the potential to increase perceived safety and thereby increase condomless anal sex among men who have sex with men (MSM). Furthermore, reductions in condom use can increase exposure to sexually transmitted infections, which in turn can reduce the protective effects of ART on HIV transmission. The current study extends previous community-based behavioral surveillance research on beliefs regarding use of ART for HIV prevention and sexual practices among MSM. Anonymous cross-sectional community surveys were collected from 1831 men at the same gay pride event in Atlanta, GA four times over nearly two decades; 1997, 2005-2006 (the 2006 survey over-sampled African-Americans to diversify the study), and 2015. Results indicate clear and consistent trends of increasing beliefs that HIV treatments reduce HIV transmission risks, reflecting the dissemination of HIV prevention research findings. Changes in treatment beliefs coincide with increased rates of condomless anal intercourse. Increased beliefs that treatments prevent HIV and increased condomless anal sex were observed for both HIV positive men and men who had not tested HIV positive. Results illustrate the emergence of an era where ART is the focus of HIV prevention and community-held beliefs and behaviors regarding definitions of risk create a new and potentially problematic environment for HIV transmission.

Multiple Imputation in Three or More Stages

Missing values present challenges in the analysis of data across many areas of research. Handling incomplete data incorrectly can lead to bias, over-confident intervals, and inaccurate inferences. One principled method of handling incomplete data is multiple imputation. This article considers incomplete data in which values are missing for three or more qualitatively different reasons and applies a modified multiple imputation framework in the analysis of that data. Included are a proof of the methodology used for three-stage multiple imputation with its limiting distribution, an extension to more than three types of missing values, an extension to the ignorability assumption with proof, and simulations demonstrating that the estimator is unbiased and efficient under the ignorability assumption.

Latent Class Analysis of Incomplete Data via an Entropy-Based Criterion

Latent class analysis is used to group categorical data into classes via a probability model. Model selection criteria then judge how well the model fits the data. When addressing incomplete data, the current methodology restricts the imputation to a single, pre-specified number of classes. We seek to develop an entropy-based model selection criterion that does not restrict the imputation to one number of clusters. Simulations show the new criterion performing well against the current standards of AIC and BIC, while a family studies application demonstrates how the criterion provides more detailed and useful results than AIC and BIC.

An Audit of Protocol Deviations Submitted to an Institutional Ethics Committee of a Tertiary Care Hospital

Protocol deviations (PDs) may jeopardize safety, rights, and welfare of subjects and data integrity. There is scarce literature and no guidelines for Institutional Ethics Committees (IECs) to process PD reports. The PD reports submitted to IECs from Jan 2011 to August 2014 were analyzed retrospectively. Types of studies reporting PDs, category and type of PDs, PD rate per participant, time of reporting PD since its occurrence and corrective actions stated by principal investigator (PI) for major deviations were noted. Out of 447 PDs from 73/1387 total studies received during study period, 402 were from 126 pharma studies. Investigator initiated studies and dissertations reported negligible PDs. Median number of PDs was 4 per protocol. Out of 447 PDs, 304 were related to study procedure, 87, 47 and 9 were from safety, informed consent document (ICD) and eligibility category respectively. The most common reason for PDs was incomplete ICD (22/47). Maximum study procedure related PDs were due to patient visiting outside window period (126/304). Thirty five of 87 PDs were due to missed safety assessment. The overall PD reporting rate per participant was 0.08. In 90% of reports, date of occurrence of PD was not specified. The median delay for reporting PDs after occurrence was 94 days. PDs classified as Major were 73% (323/447). The most common corrective actions stated by PI were participant counseling (85/323) and caution in future (70/323). The study findings emphasize the need for GCP training at regular interval of study team members. IEC have to be vigilant and visit sites frequently, take initiative and formulate guidelines regarding PD reporting.

Viral suppression and antiretroviral medication adherence among alcohol using HIV-positive adults

BACKGROUND

Substance use is a known predictor of poor adherence to antiretroviral therapies (ART) in people living with human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome. Less studied is the association between substance use and treatment outcomes, namely, suppression of HIV replication.

METHODS

Adults living with HIV (N = 183) who reported alcohol use in the previous week and receiving ART were observed over a 12-month period. Participants completed computer interviews, monthly unannounced pill counts to monitor ART adherence, and daily cell-phone delivered interactive-text assessments for alcohol use. HIV viral load was collected at baseline and 12-month follow-up from medical records. Analyses compared participants who had undetectable HIV viral loads at baseline and follow-up (sustained viral suppression) to those with unsustained viral suppression. Analyses also compared participants who were adherent to their medications (>85 % pills taken) over the year of observation to those who were nonadherent.

RESULTS

Fifty-two percent of participants had unsustained viral suppression; 47 % were ART nonadherent. Overall results failed to demonstrate alcohol use as a correlate of sustained viral suppression or treatment adherence. However, alcohol use was associated with nonadherence among participants who did not have sustained viral suppression; nonadherence in unsustained viral suppression patients was related to drinking on fewer days of assessment, missing medications when drinking, and drinking socially.

CONCLUSIONS

Poor HIV treatment outcomes and nonadherence were prevalent among adults treated for HIV infection who drink alcohol. Drinking in relation to missed medications and drinking in social settings are targets for interventions among alcohol drinkers at greatest risk for poor treatment outcomes.

Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration

The TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) Statement includes a 22-item checklist, which aims to improve the reporting of studies developing, validating, or updating a prediction model, whether for diagnostic or prognostic purposes. The TRIPOD Statement aims to improve the transparency of the reporting of a prediction model study regardless of the study methods used. This explanation and elaboration document describes the rationale; clarifies the meaning of each item; and discusses why transparent reporting is important, with a view to assessing risk of bias and clinical usefulness of the prediction model. Each checklist item of the TRIPOD Statement is explained in detail and accompanied by published examples of good reporting. The document also provides a valuable reference of issues to consider when designing, conducting, and analyzing prediction model studies. To aid the editorial process and help peer reviewers and, ultimately, readers and systematic reviewers of prediction model studies, it is recommended that authors include a completed checklist in their submission. The TRIPOD checklist can also be downloaded from www.tripod-statement.org.

Factors associated with study attrition among HIV-infected risky drinkers in St. Petersburg, Russia

BACKGROUND

Participant attrition in HIV longitudinal studies may introduce bias and diminish research quality. The identification of participant characteristics that are predictive of attrition might inform retention strategies.

OBJECTIVE

The study aimed to identify factors associated with attrition among HIV-infected Russian risky drinkers from the secondary HIV prevention HERMITAGE trial. We examined whether current injection drug use (IDU), binge drinking, depressive symptoms, HIV status nondisclosure, stigma, and lifetime history of incarceration were predictors of study attrition. We also explored effect modification due to gender.

METHODS

Complete loss to follow-up (LTFU), defined as no follow-up visits after baseline, was the primary outcome, and time to first missed visit was the secondary outcome. We used multiple logistic regression models for the primary analysis, and Cox proportional hazards models for the secondary analysis.

RESULTS

Of 660 participants, 101 (15.3%) did not return after baseline. No significant associations between independent variables and complete LTFU were observed. Current IDU and HIV status nondisclosure were significantly associated with time to first missed visit (adjusted hazard ratio [AHR], 1.39; 95% CI, 1.03-1.87; AHR, 1.38; 95% CI, 1.03-1.86, respectively). Gender stratified analyses suggested a larger impact of binge drinking among men and history of incarceration among women with time to first missed visit.

CONCLUSIONS

Although no factors were significantly associated with complete LTFU, current IDU and HIV status nondisclosure were significantly associated with time to first missed visit in HIV-infected Russian risky drinkers. An understanding of these predictors may inform retention efforts in longitudinal studies.

Effectiveness of Circle of Life, an HIV-preventive intervention for American Indian middle school youths: a group randomized trial in a Northern Plains tribe

OBJECTIVES

We assessed the effectiveness of Circle of Life (COL), an HIV-preventive intervention developed specifically for American Indian and Alaska Native (AI/AN) middle school youths.

METHODS

By partnering with a tribal community, we conducted a longitudinal wait-listed group randomized trial with 635 seventh and eighth graders in 13 schools of a Northern Plains tribe. We surveyed participants at baseline, 3 months, and 12 months from 2006 to 2007.

RESULTS

COL was found to increase HIV knowledge in the short term, but had no effect on sexual activity compared with those who did not receive it. However, COL was found to be effective for delaying the onset of sexual activity, with the greatest reduction in risk occurring for those receiving COL at early ages.

CONCLUSIONS

Community partnership was key to successful project design, implementation, and analysis. The project confirmed the importance of the timing of interventions in early adolescence. COL may be a key resource for reducing sexual risk among AI/AN youths.

The use of web-based diaries in sexual risk behaviour research: a systematic review

BACKGROUND

An increasing number of studies have used the diary method, which provides quantitative event-level data about sexual encounters. Diaries are an attractive tool for sexual behaviour research, yet little is known about the range of uses, methodological issues and best practices associated with this technology.

OBJECTIVES

To conduct a systematic review of the literature regarding the use of web-based diaries in sexual risk behaviour studies.

DESIGN

Systematic review.

DATA SOURCES

Five bibliographical databases, supplemented by references from previous reviews.

METHODS

Eligible studies were published in English before August 2013, used the internet to transmit data from collection device to study staff, and measured behaviours affecting HIV or sexually transmitted infection transmission risk. The primary author conducted an initial screen to eliminate irrelevant articles. Both authors conducted full-text reviews to determine final articles. We abstracted data on diary methodology, validity and reactivity (behaviour change caused by diary completion).

RESULTS

Twenty-three articles representing 15 studies were identified. Most diaries were collected daily for 1 month via websites, and completion was generally high (>80%). Compensation varied by study and was not associated with completion. Studies comparing diary with retrospective survey data demonstrated evidence of over-reporting on retrospective tools, except for the least frequent behaviours. Most studies that assessed reactivity as a result of diary completion demonstrated some change in behaviour associated with frequent monitoring.

CONCLUSIONS

Web-based diaries are an effective means of studying sexual risk behaviour. More uniform reporting and further research on the extent of reactivity are needed.

Latent class regression: inference and estimation with two-stage multiple imputation

Latent class regression (LCR) is a popular method for analyzing multiple categorical outcomes. While nonresponse to the manifest items is a common complication, inferences of LCR can be evaluated using maximum likelihood, multiple imputation, and two-stage multiple imputation. Under similar missing data assumptions, the estimates and variances from all three procedures are quite close. However, multiple imputation and two-stage multiple imputation can provide additional information: estimates for the rates of missing information. The methodology is illustrated using an example from a study on racial and ethnic disparities in breast cancer severity.

Mi??ing data: should we c?re?

Across all fields of research, the issue of missing data arises. In most studies, surveys, or experiments, there are instances of nonresponse that need to be appropriately dealt with to conduct reasonable statistical analyses. There are several types of missing data as well as ways to deal with incomplete data sets.