Power and Type I error rates of goodness-of-fit statistics for binomial generalized estimating equations (GEE) models

Biological control of the stink bug Nezara viridula (Heteroptera: Pentatomidae) by two parasitoids and their interaction in non-crop habitats: a simulation model

Non-cultivated areas are resting, overwintering, feeding, and/or reproducing habitats for insects, and also places from where crop areas are colonized; thus, they are essential for understanding the biological control programs in agroecosystems. We developed a simulation model for a non-cultivated area of Buenos Aires province (Argentina), and we analyzed the control of Nezara viridula achieved by the action of two parasitoids: the oophagous Trissolcus basalis and the tachinid Trichopoda giacomellii, which attack older nymphs and adults. The model is a discrete time, deterministic, phenomenological, spatially homogeneous with a 1-week time interval simulation model, based on the age-structure and/or stage-structure of N. viridula and its two parasitoids. The host-parasitoid interactions were combined with a degree-day model affecting development times of T. giacomellii pupae and T. basalis pre-imaginal stages. The simultaneous attack of both parasitoid species enables the persistence of the system at low host densities, mediated by the functional response of the parasitoids, identified as population regulation factors. However, if only one parasitoid exists (i.e., only T. basalis or only T. giacomellii) the interaction N. viridula-parasitoid persisted but at higher density of N. viridula. These results explain the successful biological control of N. viridula after the introduction of T. basalis in the 1980s, when T. giacomellii was the only parasitoid present, unable to control N. viridula. Our model shows an indirect competition when both parasitoids are present: the attack of one of them diminished the potential number of hosts available to the other parasitoid species. In the field this interaction is obscured by the hibernation period which acted as a reset mechanism affecting the density and age/stage structure of all three populations. Our model was supported by field observations, and never exhibited the extinction of any of the parasitoids from the interaction.

Mixed-Method Quasi-Experimental Study of Outcomes of a Large-Scale Multilevel Economic and Food Security Intervention on HIV Vulnerability in Rural Malawi

The objective of the Savings, Agriculture, Governance, and Empowerment for Health (SAGE4Health) study was to evaluate the impact of a large-scale multi-level economic and food security intervention on health outcomes and HIV vulnerability in rural Malawi. The study employed a quasi-experimental non-equivalent control group design to compare intervention participants (n = 598) with people participating in unrelated programs in distinct but similar geographical areas (control, n = 301). We conducted participant interviews at baseline, 18–, and 36–months on HIV vulnerability and related health outcomes, food security, and economic vulnerability. Randomly selected households (n = 1002) were interviewed in the intervention and control areas at baseline and 36 months. Compared to the control group, the intervention led to increased HIV testing (OR 1.90; 95 % CI 1.29–2.78) and HIV case finding (OR = 2.13; 95 % CI 1.07–4.22); decreased food insecurity (OR = 0.74; 95 % CI 0.63–0.87), increased nutritional diversity, and improved economic resilience to shocks. Most effects were sustained over a 3-year period. Further, no significant differences in change were found over the 3-year study period on surveys of randomly selected households in the intervention and control areas. Although there were general trends toward improvement in the study area, only intervention participants’ outcomes were significantly better. Results indicate the intervention can improve economic and food security and HIV vulnerability through increased testing and case finding. Leveraging the resources of economic development NGOs to deliver locally-developed programs with scientific funding to conduct controlled evaluations has the potential to accelerate the scientific evidence base for the effects of economic development programs on health.

Methods and protocol of a mixed method quasi-experiment to evaluate the effects of a structural economic and food security intervention on HIV vulnerability in rural Malawi: The SAGE4Health Study

Background

Poverty and lack of a predictable, stable source of food are two fundamental determinants of ill health, including HIV/AIDS. Conversely, episodes of poor health and death from HIV can disrupt the ability to maintain economic stability in affected households, especially those that rely on subsistence farming. However, little empirical research has examined if, and how, improvements in people’s economic status and food security translate into changes in HIV vulnerability.

Methods

In this paper, we describe in detail the methods and protocol of an academic-NGO collaboration on a quasi-experimental, longitudinal study of the mechanisms and magnitude of the impact of a multilevel economic and food security program (Support to Able-Bodied Vulnerable Groups to Achieve Food Security; SAFE), as implemented by CARE. Primary outcomes include HIV vulnerability (i.e., HIV risk behaviors, HIV infection), economic status (i.e., income, household assets) and food security (including anthropometric measures). We recruited participants from two types of areas of rural central Malawi: traditional authorities (TA) selected by CARE to receive the SAFE program (intervention group) and TAs receiving other unrelated CARE programming (controls). In the intervention TAs, we recruited 598 program participants (398 women, 200 men) and interviewed them at baseline and 18- and 36-month follow-ups; we interviewed 301 control households. In addition, we conducted random surveys (n = 1002) in the intervention and control areas with a 36-month assessment interval, prior to and after implementation of SAFE. Thus, we are examining intervention outcomes both in direct SAFE program participants and their larger communities. We are using multilevel modeling to examine mediators and moderators of the effects of SAFE on HIV outcomes at the individual and community levels and determine the ways in which changes in HIV outcomes feed back into economic outcomes and food security at later interviews. Finally, we are conducting a qualitative end-of-program evaluation consisting of in-depth interviews with 90 SAFE participants.

Discussion

In addition to examining pathways linking structural factors to HIV vulnerability, this research will yield important information for understanding the impact of a multilevel environmental/structural intervention on HIV, with the potential for other sustainable long-term public health benefits.

Genotype calling for the Affymetrix platform

The analysis of high-throughput genotyping data in genome-wide association (GWA) studies has become a standard approach in genetic epidemiology. Data of high quality are crucial for the success of these studies. The first step in the statistical analysis is the generation of genotypes from signal intensities, and several approaches have been proposed for obtaining as accurate genotypes as possible. For the Affymetrix Genome-Wide Human SNP Array 6.0, the genotype calling algorithms Birdseed and CRLMM are commonly used in applications. After a brief description of the statistical methods for both algorithms, their usage is described in detail. Links are provided to the software and to sample code for the installation and execution of the algorithms. Additionally, a suggestion for processing the result files is made.

A nonparametric smoothing method for assessing GEE models with longitudinal binary data

Studies involving longitudinal binary responses are widely applied in the health and biomedical sciences research and frequently analyzed by generalized estimating equations (GEE) method. This article proposes an alternative goodness-of-fit test based on the nonparametric smoothing approach for assessing the adequacy of GEE fitted models, which can be regarded as an extension of the goodness-of-fit test of le Cessie and van Houwelingen (Biometrics 1991; 47:1267-1282). The expectation and approximate variance of the proposed test statistic are derived. The asymptotic distribution of the proposed test statistic in terms of a scaled chi-squared distribution and the power performance of the proposed test are discussed by simulation studies. The testing procedure is demonstrated by two real data.