Logistic regression models are frequently used in cohort studies to determine the association between treatment and dichotomous outcomes in the presence of confounding variables. In a logistic regression model, the association between exposure and outcome is measured using the odds ratio (OR). The OR can be difficult to interpret and only approximates the relative risk (RR) in certain restrictive settings. Several authors have suggested that for dichotomous outcomes, RRs, RR reductions, absolute risk reductions, and the number needed to treat (NNT) are more clinically meaningful measures of treatment effect.
We describe a method for deriving clinically meaningful measures of treatment effect from a logistic regression model. This method involves determining the probability of the outcome if each subject in the cohort was treated and if each subject was untreated. These probabilities are then averaged across the study cohort to determine the average probability of the outcome in the population if all subjects were treated and if they were untreated.
Risk differences, RRs, and NNTs were derived using a logistic regression model.
Clinically meaningful measures of effect can be derived from a logistic regression model in a cohort study. These methods can also be used in randomized controlled trials when logistic regression is used to adjust for possible imbalance in prognostically important baseline covariates.
It has been well known that ignoring measurement error may result in substantially biased estimates in many contexts including linear and nonlinear regressions. For survival data with measurement error in covariates there has been extensive discussion in the literature with the focus being on the Cox proportional hazards models. However, the impact of measurement error on accelerated failure time (AFT) models has received little attention, though AFT models are very useful in survival data analysis. In this paper, we discuss AFT models with error-prone covariates and study the bias induced by the naive approach of ignoring measurement error in covariates. To adjust for such a bias, we describe a simulation and extrapolation method. This method is appealing because it is simple to implement and it does not require modelling the true but error-prone covariate process that is often not observable. Asymptotic normality for the resulting estimators is established. Simulation studies are carried out to evaluate the performance of the proposed method as well as the impact of ignoring measurement error in covariates. The proposed method is applied to analyse a data set arising from the Busselton Health study (Australian J. Public Health 1994; 18:129-135).
Access to health care services in Canada has been identified as an urgent priority, and chronic disease has been suggested as the most pressing health concern facing Canadians. Access to services for Canadians living with chronic disease, however, has received little emphasis in the research literature or in health policy reform documents. A systematic review of research into factors impeding or facilitating access to formal health services for people in Canada living with chronic illness is presented. The review includes 31 studies of Canadian populations published between 1990 and 2002; main results were analyzed for facilitators and barriers to access for people experiencing chronic disease. An underlying organizing construct of symmetry between consumers, providers, and the larger Canadian system is suggested as a relevant lens from which to view the findings. Finally, a discussion of the relationship between identified factors and the principles of primary health care is offered.
Epidemiological studies with two-stage designs typically gather information about some covariates from all study subjects in the first sampling stage, while additional data from only a subset of the subjects are collected in the second sampling stage. Appropriate analysis of two-stage studies maintains validity and can also improve precision. We describe an application of a weighted likelihood method, mean-score logistic regression, to accommodate data from a cross-sectional study of Helicobacter pylori infection in children, where the study sample was enriched with additional non-randomly sampled cases. The present work exemplifies how careful analysis of epidemiological data from complex sampling schemes can adjust for potential selection bias, improve precision and enable a more complete investigation of factors of interest. Our results highlight the importance of H. pylori infected mothers and siblings as risk factors for the infection in children in Sweden.
The biological fate of the fish feed additive, ethoxyquin (EQ) was examined in the muscle of Atlantic salmon during 12 weeks of feeding followed by a 2 weeks depuration period. Parent EQ (1,2-dihydro-6-ethoxy-2,2,4-trimethylquinoline), quinone imine (2,6-dihydro-2,2,4-trimethyl-6-quinolone), de-ethylated EQ (6-hydroxy-2,2,4-trimethyl-1,2-dihydroquinoline) and EQDM (EQ dimer or 1,8'-di(1,2-dihydro-6-ethoxy-2,2,4-trimethyl-quinoline) were found to be the ubiquitous metabolites of dietary EQ, with EQDM as a main metabolite. A rapid decrease in the level of EQ (2.4 days of half-life) was balanced by an increase in EQDM, giving an unchanged net sum following 2 weeks of depuration. The mandatory 14 days depuration period prior to slaughtering of farmed salmon in Norway was not sufficient for complete elimination of EQ-derived residuals. Post depuration, EQDM accounted for 99% of sum of the two compounds in all treatment groups; possible toxicological effects of EQDM are not known. The individual concentrations of EQ and EQDM and their sum are dependent on EQ level in the feed, consequently, their residual concentrations may be controlled. The theoretical amount of EQ and EQDM consumed in one meal of farmed salmon would be under the recommended ADI, provided that the fish were raised on feed with no more than 150 mg EQ/kg feed, which is the EU maximum limit for EQ in fish feed.
To determine whether Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) scores correlate with the clinician's impression of level of disease activity.
In total, 230 patients with SLE followed at the University of Toronto Lupus Clinic who had 5 visits 3 months apart in 1992-93 were studied. At each visit a standard protocol was completed. A clinician who did not know the patients or their SLEDAI scores evaluated each patient record and assigned a clinical activity level. "Flare" was defined by new or increased therapy for active disease, an expression of concern, or use of the term "flare" in the physician's notes. The SLEDAI score was calculated from the database.
SLEDAI scores described a range of clinical activity as recognized by the clinician. Median SLEDAI scores ranged from 2 (inactive disease) to 8 (persistently active or flare). When the clinician assessed the patient to be improved, the median SLEDAI score decreased by 2. When the clinician assessed that the patient was experiencing a flare, the SLEDAI score increased by a median of 4.
Based on our data we propose the following outcomes for patients with SLE: flare, an increase in SLEDAI > 3; improvement is a reduction in SLEDAI of > 3; persistently active disease is change in SLEDAI +/- 3; and remission a SLEDAI of 0. These outcomes will allow a more complete description of a patient's response to therapeutic intervention in a responder index.
GENESIS (General Ethnographic and Nursing Evaluation Studies In the State) is a tested and proven community analysis strategy that integrates ethnographic and epidemiologic data to arrive at a comprehensive, holistic description of the health of a community and its residents. Communities analyzed in most project GENESIS studies have been rural or semirural. ACTION (Assessing Communities Together in the Identification Of Needs) is an extension of the GENESIS community analysis model that was developed to meet the unique needs of community-level research and analysis in an urban, multicultural setting. Significant differences in the context in which the ACTION projects took place necessitated extensions in specific components of the GENESIS model. Application of the GENESIS model by the ACTION team is described. Based on the experiences with ACTION, recommendations are offered for future urban, multicultural community analysis projects.
Manitoba's hospital separations and physician medical files were linked for the fiscal years 1984-85 and 1985-86. The result was a study file consisting of records for 5,293 males and 3,143 females, who, during this period, suffered an Acute Myocardial Infarction (AMI), commonly called a heart attack. Merging the two types of files created a comprehensive data base for these AMI victims. The Manitoba age-sex standardized AMI rate was 38.0 per 10,000 population. Age-specific rates were higher for males than for females for all age groups. Hospitalized cases accounted for 7,201 individuals or 85.4% of AMI victims. Age-sex standardized rates of hospitalization per 10,000 population ranged from 27.1 in the Central region to 36.0 in the Westman region. The Manitoba age-specific rates of hospitalization for males in the 35-54 and 55-64 age groups were about three times the female rates for the same age groups. One quarter of AMI hospitalized victims died in hospital. The Manitoba age-specific death rates for males in the 35-54, 55-64 and 65-74 age groups were double the rates for females in the same age groups. Of the 8,436 AMI victims under study, 86.4% had at least one other concurrent medical condition such as angina, other forms of ischemic heart disease, diabetes, or hypertension. Of AMI victims, 93.8% underwent at least one of the following procedures: coronary artery bypass surgery, angiogram, electrocardiogram, cardiac catheterization, arteriography, or blood cholesterol testing. A higher percentage of procedures was performed on males than on females.
Recent adaptations of the spatial scan approach to detecting disease clusters have addressed the problem of finding clusters that occur in non-compact and non-circular shapes--such as along roads or river networks. Some of these approaches may have difficulty defining cluster boundaries precisely, and tend to over-fit data with very irregular (and implausible) clusters shapes.
We describe two simple adaptations to these approaches that can be used to improve the effectiveness of irregular disease cluster detection. The first adaptation penalizes very irregular cluster shapes based on a measure of connectivity (non-connectivity penalty). The second adaptation prevents searches from combining smaller clusters into large super-clusters (depth limit). We conduct experiments with simulated data in order to observe the performance of these adaptations on a number of synthetic cluster shapes.
Our results suggest that the combination of these two adaptations may increase the ability of a cluster detection method to find irregular shapes without affecting its ability to find more regular (i.e., compact) shapes. The depth limit in particular is effective when it is deemed important to distinguish nearby clusters from each other. We suggest that these adaptations of adjacency-constrained spatial scans are particularly well suited to chronic disease and injury surveillance.
Cites: Lancet. 1988 Feb 6;1(8580):272-32893085
Cites: Stat Med. 2006 Mar 15;25(5):723-4216453376
Cites: Stat Med. 1996 Apr 15-May 15;15(7-9):717-269132899
Cites: Am J Epidemiol. 1990 Jul;132(1 Suppl):S136-432356825