Plotting regression coefficients and other estimates

Graphical display of regression results has become increasingly popular in presentations and in scientific literature because graphs are often much easier to read than tables. Such plots can be produced in Stata by the marginsplot command (see [R] marginsplot). However, while marginsplot is versatile and flexible, it has two major limitations: it can only process results left behind by margins (see [R] margins), and it can handle only one set of results at a time. In this article, I introduce a new command called coefplot that overcomes these limitations. It plots results from any estimation command and combines results from several models into one graph. The default behavior of coefplot is to plot markers for coefficients and horizontal spikes for confidence intervals. However, coefplot can also produce other types of graphs. I illustrate the capabilities of coefplot by using a series of examples.

The Blinder-Oaxaca decomposition for linear regression models

The counterfactual decomposition technique popularized by Blinder (1973, Journal of Human Resources, 436–455) and Oaxaca (1973, International Economic Review, 693–709) is widely used to study mean outcome differences between groups. For example, the technique is often used to analyze wage gaps by sex or race. This article summarizes the technique and addresses several complications, such as the identification of effects of categorical predictors in the detailed decomposition or the estimation of standard errors. A new command called oaxaca is introduced, and examples illustrating its usage are given.

Making regression tables simplified

estout, introduced by Jann (Stata Journal 5: 288–308), is a useful tool for producing regression tables from stored estimates. However, its syntax is relatively complex and commands may turn out long even for simple tables. Furthermore, having to store the estimates beforehand can be cumbersome. To facilitate the production of regression tables, I therefore present here two new commands called eststo and esttab. eststo is a wrapper for offcial Stata’s estimates store and simplifies the storing of estimation results for tabulation. esttab, on the other hand, is a wrapper for estout and simplifies compiling nice-looking tables from the stored estimates without much typing. I also provide updates to estout and estadd.

Making regression tables from stored estimates

Organizing and archiving statistical results and processing a subset of those results for publication are important and often underestimated issues in conducting statistical analyses. Because automation of these tasks is often poor, processing results produced by statistical packages is quite laborious and vulnerable to error. I will therefore present a new package called estout that facilitates and automates some of these tasks. This new command can be used to produce regression tables for use with spreadsheets, LaTeX, HTML, or word processors. For example, the results for multiple models can be organized in spreadsheets and can thus be archived in an orderly manner. Alternatively, the results can be directly saved as a publication-ready table for inclusion in, for example, a LaTeX document. estout is implemented as a wrapper for estimates table but has many additional features, such as support for mfx. However, despite its flexibility, estout is—I believe—still very straightforward and easy to use. Furthermore, estout can be customized via so-called defaults files. A tool to make available supplementary statistics called estadd is also provided.

3D Intervertebral Disc Localization and Segmentation from MR Images by Data-Driven Regression and Classification

In this paper we propose a new fully-automatic method for localizing and segmenting 3D intervertebral discs from MR images, where the two problems are solved in a unified data-driven regression and classification framework. We estimate the output (image displacements for localization, or fg/bg labels for segmentation) of image points by exploiting both training data and geometric constraints simultaneously. The problem is formulated in a unified objective function which is then solved globally and efficiently. We validate our method on MR images of 25 patients. Taking manually labeled data as the ground truth, our method achieves a mean localization error of 1.3 mm, a mean Dice metric of 87%, and a mean surface distance of 1.3 mm. Our method can be applied to other localization and segmentation tasks.

Fully automatic segmentation of AP pelvis X-rays via random forest regression with efficient feature selection and hierarchical sparse shape composition

In clinical practice, traditional X-ray radiography is widely used, and knowledge of landmarks and contours in anteroposterior (AP) pelvis X-rays is invaluable for computer aided diagnosis, hip surgery planning and image-guided interventions. This paper presents a fully automatic approach for landmark detection and shape segmentation of both pelvis and femur in conventional AP X-ray images. Our approach is based on the framework of landmark detection via Random Forest (RF) regression and shape regularization via hierarchical sparse shape composition. We propose a visual feature FL-HoG (Flexible- Level Histogram of Oriented Gradients) and a feature selection algorithm based on trace radio optimization to improve the robustness and the efficacy of RF-based landmark detection. The landmark detection result is then used in a hierarchical sparse shape composition framework for shape regularization. Finally, the extracted shape contour is fine-tuned by a post-processing step based on low level image features. The experimental results demonstrate that our feature selection algorithm reduces the feature dimension in a factor of 40 and improves both training and test efficiency. Further experiments conducted on 436 clinical AP pelvis X-rays show that our approach achieves an average point-to-curve error around 1.2 mm for femur and 1.9 mm for pelvis.

ROBREG: Stata module providing robust regression estimators

robreg provides a number of robust estimators for linear regression models. Among them are the high breakdown-point and high efficiency MM-estimator, the Huber and bisquare M-estimator, and the S-estimator, each supporting classic or robust standard errors. Furthermore, basic versions of the LMS/LQS (least median of squares) and LTS (least trimmed squares) estimators are provided. Note that the moremata package, also available from SSC, is required.

ERS task force statement: diagnosis and treatment of primary spontaneous pneumothorax.

Primary spontaneous pneumothorax (PSP) affects young healthy people with a significant recurrence rate. Recent advances in treatment have been variably implemented in clinical practice. This statement reviews the latest developments and concepts to improve clinical management and stimulate further research.The European Respiratory Society's Scientific Committee established a multidisciplinary team of pulmonologists and surgeons to produce a comprehensive review of available scientific evidence.Smoking remains the main risk factor of PSP. Routine smoking cessation is advised. More prospective data are required to better define the PSP population and incidence of recurrence. In first episodes of PSP, treatment approach is driven by symptoms rather than PSP size. The role of bullae rupture as the cause of air leakage remains unclear, implying that any treatment of PSP recurrence includes pleurodesis. Talc poudrage pleurodesis by thoracoscopy is safe, provided calibrated talc is available. Video-assisted thoracic surgery is preferred to thoracotomy as a surgical approach.In first episodes of PSP, aspiration is required only in symptomatic patients. After a persistent or recurrent PSP, definitive treatment including pleurodesis is undertaken. Future randomised controlled trials comparing different strategies are required.

Dynamic changes of intramural hematoma in patients with acute spontaneous internal carotid artery dissection.

BACKGROUND We prospectively investigated temporal and spatial evolution of intramural hematomas in patients with acute spontaneous internal carotid artery dissection using repeated magnetic resonance imaging over six-months. AIM The aim of the present study was to assess dynamic changes of intramural hematoma in patients with acute spontaneous internal carotid artery dissection at multiple follow-up time-points with T1w, PD/T2w, and magnetic resonance angiography. METHODS We performed serial multiparametric magnetic resonance imaging in 10 patients with spontaneous internal carotid artery dissection on admission, at days 1, 3, 7-14 and at months 1·5, 3, and 6. We calculated the volume and extension of the hyperintense intramural hematoma using T1w and PD/T2w fat suppressed sequences and assessed the degree of stenosis due to the hematoma using magnetic resonance angiography. RESULTS Mean interval from symptom onset to first magnetic resonance imaging was two-days (SD 2·7). Two patients presented with ischemic stroke, three with transient ischemic attacks, and five with pain and local symptoms only. Nine patients had a transient increase of the intramural hematoma volume, mainly up to day 10 after symptom onset. Fifty percent had a transient increase in the degree of the internal carotid artery stenosis on MRA, one resulting in a temporary occlusion. Lesions older than one-week were predominantly characterized by a shift from iso- to hyperintese signal on T2w images. At three-month follow-up, intramural hematoma was no longer detectable in 80% of patients and had completely resolved in all patients after six-months. CONCLUSIONS Spatial and temporal dynamics of intramural hematomas after spontaneous internal carotid artery dissection showed an early volume increase with concomitant progression of the internal carotid artery stenosis in 5 of 10 patients. Although spontaneous internal carotid artery dissection overall carries a good prognosis with spontaneous hematoma resorption in all our patients, early follow-up imaging may be considered, especially in case of new clinical symptoms.

Management of spontaneous intracranial hypotension - Transorbital ultrasound as discriminator.

OBJECTIVE Spontaneous intracranial hypotension (SIH) is most commonly caused by cerebrospinal fluid (CSF) leakage. Therefore, we hypothesised that patients with orthostatic headache (OH) would show decreased optic nerve sheath diameter (ONSD) during changes from supine to upright position. METHODS Transorbital B-mode ultrasound was performed employing a high-frequency transducer for ONSD measurements in the supine and upright positions. Absolute values and changes of ONSD from supine to upright were assessed. Ultrasound was performed in 39 SIH patients, 18 with OH and 21 without OH, and in 39 age-matched control subjects. The control group comprised 20 patients admitted for back surgery without headache or any orthostatic symptoms, and 19 healthy controls. RESULTS In supine position, mean ONSD (±SD) was similar in patients with (5.38±0.91 mm) or without OH (5.48±0.89 mm; p=0.921). However, in upright position, mean ONSD was different between patients with (4.84±0.99 mm) and without OH (5.53±0.99 mm; p=0.044). Furthermore, the change in ONSD from supine to upright position was significantly greater in SIH patients with OH (-0.53±0.34 mm) than in SIH patients without OH (0.05±0.41 mm; p≤0.001) or in control subjects (0.01±0.38 mm; p≤0.001; area under the curve: 0.874 in receiver operating characteristics analysis). CONCLUSIONS Symptomatic patients with SIH showed a significant decrease of ONSD, as assessed by ultrasound, when changing from the supine to the upright position. Ultrasound assessment of the ONSD in two positions may be a novel, non-invasive tool for the diagnosis and follow-up of SIH and for elucidating the pathophysiology of SIH.

Modeling absolute differences in life expectancy with a censored skew-normal regression approach.

Parameter estimates from commonly used multivariable parametric survival regression models do not directly quantify differences in years of life expectancy. Gaussian linear regression models give results in terms of absolute mean differences, but are not appropriate in modeling life expectancy, because in many situations time to death has a negative skewed distribution. A regression approach using a skew-normal distribution would be an alternative to parametric survival models in the modeling of life expectancy, because parameter estimates can be interpreted in terms of survival time differences while allowing for skewness of the distribution. In this paper we show how to use the skew-normal regression so that censored and left-truncated observations are accounted for. With this we model differences in life expectancy using data from the Swiss National Cohort Study and from official life expectancy estimates and compare the results with those derived from commonly used survival regression models. We conclude that a censored skew-normal survival regression approach for left-truncated observations can be used to model differences in life expectancy across covariates of interest.