Effects In Multivariate Models Research Articles

Multivariate network meta-analysis incorporating class effects

BackgroundNetwork meta-analysis synthesises data from a number of clinical trials in order to assess the comparative efficacy of multiple healthcare interventions in similar patient populations. In situations where clinical trial data are heterogeneously reported i.e. data are missing for one or more outcomes of interest, synthesising such data can lead to disconnected networks of evidence, increased uncertainty, and potentially biased estimates which can have severe implications for decision-making. To overcome this issue, strength can be borrowed between outcomes of interest in multivariate network meta-analyses. Furthermore, in situations where there are relatively few trials informing each treatment comparison, there is a potential issue with the sparsity of data in the treatment networks, which can lead to substantial parameter uncertainty. A multivariate network meta-analysis approach can be further extended to borrow strength between interventions of the same class using hierarchical models.MethodsWe extend the trivariate network meta-analysis model to incorporate the exchangeability between treatment effects belonging to the same class of intervention to increase precision in treatment effect estimates. We further incorporate a missing data framework to estimate uncertainty in trials that did not report measures of variability in order to maximise the use of all available information for healthcare decision-making. The methods are applied to a motivating dataset in overactive bladder syndrome. The outcomes of interest were mean change from baseline in incontinence, voiding and urgency episodes. All models were fitted using Bayesian Markov Chain Monte Carlo (MCMC) methods in WinBUGS.ResultsAll models (univariate, multivariate, and multivariate models incorporating class effects) produced similar point estimates for all treatment effects. Incorporating class effects in multivariate models often increased precision in treatment effect estimates.ConclusionsMultivariate network meta-analysis incorporating class effects allowed for the comparison of all interventions across all outcome measures to ameliorate the potential impact of outcome reporting bias, and further borrowed strength between interventions belonging to the same class of treatment to increase the precision in treatment effect estimates for healthcare policy and decision-making.

Comparison of mixed effects models of antimicrobial resistance metrics of livestock and poultry Salmonella isolates from a national monitoring system

Antimicrobial use in agriculture is considered a pathway for the selection and dissemination of resistance determinants among animal and human populations. From 1997 through 2003 the U.S. National Antimicrobial Resistance Monitoring System (NARMS) tested clinical Salmonella isolates from multiple animal and environmental sources throughout the United States for resistance to panels of 16–19 antimicrobials. In this study we applied two mixed effects models, the generalized linear mixed model (GLMM) and accelerated failure time frailty (AFT-frailty) model, to susceptible/resistant and interval-censored minimum inhibitory concentration (MIC) metrics, respectively, from Salmonella enterica subspecies enterica serovar Typhimurium isolates from livestock and poultry. Objectives were to compare characteristics of the two models and to examine the effects of time, species, and multidrug resistance (MDR) on the resistance of isolates to individual antimicrobials, as revealed by the models. Fixed effects were year of sample collection, isolate source species and MDR indicators; laboratory study site was included as a random effect. MDR indicators were significant for every antimicrobial and were dominant effects in multivariable models. Temporal trends and source species influences varied by antimicrobial. In GLMMs, the intra-class correlation coefficient ranged up to 0.8, indicating that the proportion of variance accounted for by laboratory study site could be high. AFT models tended to be more sensitive, detecting more curvilinear temporal trends and species differences; however, high levels of left- or right-censoring made some models unstable and results uninterpretable. Results from GLMMs may be biased by cutoff criteria used to collapse MIC data into binary categories, and may miss signaling important trends or shifts if the series of antibiotic dilutions tested does not span a resistance threshold. Our findings demonstrate the challenges of measuring the AMR ecosystem and the complexity of interacting factors, and have implications for future monitoring. We include suggestions for future data collection and analyses, including alternative modeling approaches.

Neighborhood Design for Walking and Biking

Neighborhood designs often relate to physical activity and to BMI. Does neighborhood walkability/bikeability relate to BMI and obesity risk and does moderate-to-vigorous physical activity (MVPA) account for some of the relationship? Census 2000 provided walkability/bikeability measures-block group proportions of workers who walk or bike to work, housing age, and population density-and National Health and Nutrition Examination Study (NHANES 2003-2006) provided MVPA accelerometer measures. Regression analyses (2011-2012) adjusted for geographic clustering and multiple control variables. Greater density and older housing were associated with lower male BMI in bivariate analyses, but there were no density and housing age effects in multivariate models. For women, greater proportions of neighborhood workers who walk to work (M=0.02) and more MVPA was associated with lower BMI and lower obesity risk. For men, greater proportions of workers who bike to work (M=0.004) and more MVPA was associated with lower BMI and obesity risk. For both effects, MVPA partially mediated the relationships between walkability/bikeability and BMI. If such associations are causal, doubling walk and bike-to-work proportions (to 0.04 and 0.008) would have -0.3 and -0.33 effects on the average BMIs of adult women and men living in the neighborhood. This equates to 1.5 pounds for a 64-inch-tall woman and 2.3 pounds for a 69-inch-tall man. Although walking/biking to work is rare in the U.S., greater proportions of such workers in neighborhoods relate to lower weight and higher MVPA. Bikeability merits greater attention as a modifiable activity-friendliness factor, particularly for men.

Suicidal Behavior in Prodromal Huntington Disease

Background: Several studies have suggested a greater risk of suicide in Huntington disease (HD); however, unique risk factors for suicide in HD are not established. Objective: We sought to determine risk factors for suicidal behavior, defined as suicide or attempted suicide, in prodromal HD. Methods: From the prospective PREDICT-HD cohort, we identified 735 cases with HD gene expansion but no manifest symptoms of HD and 194 non-gene-expanded controls. In survival analysis, a number of potential risk factors for suicidal behavior were assessed, including symptoms of depression, hopelessness, substance abuse, marital status, gender, and psychiatric history. Results: During a mean of 3.7 years of prospective follow-up, 12 cases (1.6%) attempted suicide and 1 completed suicide (0.1%). No suicides were observed among controls. In univariate Cox proportional hazards regression models, a history of suicide attempts (HR 8.5, 95% CI 2.8–26.1, p < 0.0002) and a Beck Depression Inventory II score >13 (HR 7.2, 95% CI 2.3–22.0, p < 0.0006) were associated with suicidal behavior. These risk factors had independent effects in multivariate models. A history of incarceration in the past 2 years was also associated (HR 12.5, 95% CI 2.7–56.6, p < 0.002), though uncommon. No further risk factors were identified. Conclusion: A history of suicide attempts and the presence of depression are strongly predictive of suicidal behavior in prodromal HD. As these risk factors are among the most robust risk factors for suicide, established suicide risk factors appear applicable to those with prodromal HD.

Comparison of procedures to assess non‐linear and time‐varying effects in multivariable models for survival data

The focus of many medical applications is to model the impact of several factors on time to an event. A standard approach for such analyses is the Cox proportional hazards model. It assumes that the factors act linearly on the log hazard function (linearity assumption) and that their effects are constant over time (proportional hazards (PH) assumption). Variable selection is often required to specify a more parsimonious model aiming to include only variables with an influence on the outcome. As follow-up increases the effect of a variable often gets weaker, which means that it varies in time. However, spurious time-varying effects may also be introduced by mismodelling other parts of the multivariable model, such as omission of an important covariate or an incorrect functional form of a continuous covariate. These issues interact. To check whether the effect of a variable varies in time several tests for non-PH have been proposed. However, they are not sufficient to derive a model, as appropriate modelling of the shape of time-varying effects is required. In three examples we will compare five recently published strategies to assess whether and how the effects of covariates from a multivariable model vary in time. For practical use we will give some recommendations.

Incarceration and Risky Sexual Partnerships in a Southern US City

Incarceration is strongly associated with HIV infection and may contribute to viral transmission by disrupting stable partnerships and promoting high-risk partnerships. We investigated incarceration and STI/HIV-related partnerships among a community-based sample recruited for a sexual behavior interview while frequenting venues where people meet sexual partners in a North Carolina city (N = 373). Men reporting incarceration in the past 12 months were more likely than men without recent incarceration to experience multiple new sexual partnerships (unadjusted prevalence ratio [PR] 1.8, 95% confidence interval [CI]: 1.1-3.1) and transactional sex defined as trading sex for money, goods, or services (unadjusted PR: 4.0, 95% CI: 2.3-7.1) in the past 4 weeks. Likewise, women who were ever incarcerated were more likely than never-incarcerated women to experience recent multiple new partnerships (unadjusted PR: 3.1, 95% CI: 1.8-5.4) and transactional sex (unadjusted PR: 5.3, 95% CI: 2.6-10.9). Sexual partnership in the past 12 months with someone who had ever been incarcerated versus with partners with no known incarceration history was associated with recent multiple new partnerships (men: unadjusted PR 2.0, 95% CI 1.4-2.9, women: unadjusted PR 4.8, 95% CI 2.3-10.1) and transactional sex (men: unadjusted PR 3.3, 95% CI 1.7-6.6, women: unadjusted PR 6.1, 95% CI 2.4-15.4). Adjustment for demographic and socioeconomic variables had minimal effect on estimates. However, the strong overlap between incarceration, partner incarceration, and substance abuse had substantial effects in multivariable models. Correctional-facility and community-based HIV prevention, with substance abuse treatment, should reach currently and formerly incarcerated individuals and their sexual partners.

Detection of causal relationships between factors influencing adverse side-effects from anaesthesia and convalescence following surgery: a path analytical approach.

The anaesthesiologist's preoperative interview with the patient is important in preparing the patient for surgery. Its potential protective influence on adverse side-effects from anaesthesia and convalescence is rarely investigated within the context of other perioperative factors. Structural equation modelling allows detection and quantification of all causal relationships and mediator effects in multivariate models. Therefore, this method is presented as a tool and applied to discover the influence of the preoperative interview within socio-demographic variables and duration of surgery on complaints and recovery after anaesthesia. The influence of individual satisfaction with the anaesthesiologist's preoperative interview on postoperative events such as nausea/vomiting, difficulties in recovering from anaesthesia, experience of postoperative pain, physical discomfort and satisfaction with convalescence expressed by the patient was analysed by means of structural equation modelling. The variables gender, age and duration of surgery were also included as predictors in the analyses. The model in the total sample of 710 patients was then analysed for structural differences between groups treated either with propofol (n = 204) or with isoflurane + nitrous oxide (n = 267) for maintenance of anaesthesia. The model revealed that the anaesthesiologist's preoperative interview in combination with associated mediating side-effects explains 45% of the variance of 'feeling physical discomfort' and 18% of the variance of 'satisfaction with convalescence'. The same model could be fitted in the propofol and the isoflurane + nitrous oxide group. Moreover, the structure and the strength of causal relations between variables were identical in the two groups. The anaesthesiologist's efforts to improve the interview with the patient by more reassuring and proper information will result in less side-effects from anaesthesia and better recovery from surgery. It could be demonstrated that structural equation modelling is a powerful tool for detection of causal relationships and mediator effects in perioperative medicine.

Detection of causal relationships between factors influencing adverse side-effects from anaesthesia and convalescence following surgery: a path analytical approach

Background and objective: The anaesthesiologist's preoperative interview with the patient is important in preparing the patient for surgery. Its potential protective influence on adverse side-effects from anaesthesia and convalescence is rarely investigated within the context of other perioperative factors. Structural equation modelling allows detection and quantification of all causal relationships and mediator effects in multivariate models. Therefore, this method is presented as a tool and applied to discover the influence of the preoperative interview within socio-demographic variables and duration of surgery on complaints and recovery after anaesthesia. Methods: The influence of individual satisfaction with the anaesthesiologist's preoperative interview on postoperative events such as nausea/vomiting, difficulties in recovering from anaesthesia, experience of postoperative pain, physical discomfort and satisfaction with convalescence expressed by the patient was analysed by means of structural equation modelling. The variables gender, age and duration of surgery were also included as predictors in the analyses. The model in the total sample of 710 patients was then analysed for structural differences between groups treated either with propofol (n = 204) or with isoflurane + nitrous oxide (n = 267) for maintenance of anaesthesia. Results: The model revealed that the anaesthesiologist's preoperative interview in combination with associated mediating side-effects explains 45% of the variance of 'feeling physical discomfort' and 18% of the variance of 'satisfaction with convalescence'. The same model could be fitted in the propofol and the isoflurane + nitrous oxide group. Moreover, the structure and the strength of causal relations between variables were identical in the two groups. Conclusions: The anaesthesiologist's efforts to improve the interview with the patient by more reassuring and proper information will result in less side-effects from anaesthesia and better recovery from surgery. It could be demonstrated that structural equation modelling is a powerful tool for detection of causal relationships and mediator effects in perioperative medicine.