On Graphically Checking Goodness-of-fit of Binary Logistic Regression Models

OBJECTIVES: This paper is concerned with checking goodness-of-fit of binary logistic regression models. For the practitioners of data analysis, the broad classes of procedures for checking goodness-of-fit available in the literature are described. The challenges of model checking in the context of binary logistic regression are reviewed. As a viable solution, a simple graphical procedure for checking goodness-of-fit is proposed. METHODS: The graphical procedure proposed relies on pieces of information available from any logistic analysis; the focus is on combining and presenting these in an informative way. RESULTS: The information gained using this approach is presented with three examples. In the discussion, the proposed method is put into context and compared with other graphical procedures for checking goodness-of-fit of binary logistic models available in the literature. CONCLUSION: A simple graphical method can significantly improve the understanding of any logistic regression analysis and help to prevent faulty conclusions.

LOGITCPRPLOT: Stata module to graph component-plus-residual plot for logistic regression

logitcprplot can be used after logistic regression for graphing a component-plus-residual plot (a.k.a. partial residual plot) for a given predictor, including a lowess, local polynomial, restricted cubic spline, fractional polynomial, penalized spline, regression spline, running line, or adaptive variable span running line smooth

RRLOGIT: Stata module to estimate logistic regression for randomized response data

rrlogit fits a maximum-likelihood logistic regression for randomized response data.

Reporting of research quality characteristics of studies published in 6 major clinical dental specialty journals

The objective of this article was to record reporting characteristics related to study quality of research published in major specialty dental journals with the highest impact factor (Journal of Endodontics, Journal of Oral and Maxillofacial Surgery, American Journal of Orthodontics and Dentofacial Orthopedics; Pediatric Dentistry, Journal of Clinical Periodontology, and International Journal of Prosthetic Dentistry). The included articles were classified into the following 3 broad subject categories: (1) cross-sectional (snap-shot), (2) observational, and (3) interventional. Multinomial logistic regression was conducted for effect estimation using the journal as the response and randomization, sample calculation, confounding discussed, multivariate analysis, effect measurement, and confidence intervals as the explanatory variables. The results showed that cross-sectional studies were the dominant design (55%), whereas observational investigations accounted for 13%, and interventions/clinical trials for 32%. Reporting on quality characteristics was low for all variables: random allocation (15%), sample size calculation (7%), confounding issues/possible confounders (38%), effect measurements (16%), and multivariate analysis (21%). Eighty-four percent of the published articles reported a statistically significant main finding and only 13% presented confidence intervals. The Journal of Clinical Periodontology showed the highest probability of including quality characteristics in reporting results among all dental journals.

Publication and non-publication of clinical trials: longitudinal study of applications submitted to a research ethics committee

BACKGROUND: Not all clinical trials are published, which may distort the evidence that is available in the literature. We studied the publication rate of a cohort of clinical trials and identified factors associated with publication and nonpublication of results. METHODS: We analysed the protocols of randomized clinical trials of drug interventions submitted to the research ethics committee of University Hospital (Inselspital) Bern, Switzerland from 1988 to 1998. We identified full articles published up to 2006 by searching the Cochrane CENTRAL database (issue 02/2006) and by contacting investigators. We analyzed factors associated with the publication of trials using descriptive statistics and logistic regression models. RESULTS: 451 study protocols and 375 corresponding articles were analyzed. 233 protocols resulted in at least one publication, a publication rate of 52%. A total of 366 (81%) trials were commercially funded, 47 (10%) had non-commercial funding. 346 trials (77%) were multi-centre studies and 272 of these (79%) were international collaborations. In the adjusted logistic regression model non-commercial funding (Odds Ratio [OR] 2.42, 95% CI 1.14-5.17), multi-centre status (OR 2.09, 95% CI 1.03-4.24), international collaboration (OR 1.87, 95% CI 0.99-3.55) and a sample size above the median of 236 participants (OR 2.04, 95% CI 1.23-3.39) were associated with full publication. CONCLUSIONS: In this cohort of applications to an ethics committee in Switzerland, only about half of clinical drug trials were published. Large multi-centre trials with non-commercial funding were more likely to be published than other trials, but most trials were funded by industry.

Obesity and overweight associated with lower rates of colorectal cancer screening in Switzerland

Screening for colorectal cancer (CRC) is associated with reduced CRC mortality, but low screening rates have been reported in several settings. The aim of the study was to assess predictors of low CRC screening in Switzerland. A retrospective cohort of a random sample of 940 patients aged 50-80 years followed for 2 years from four Swiss University primary care settings was used. Patients with illegal residency status and a history of CRC or colorectal polyps were excluded. We abstracted sociodemographic data of patients and physicians, patient health status, and indicators derived from RAND's Quality Assessment Tools from medical charts. We defined CRC screening as colonoscopy in the last 10 years, flexible sigmoidoscopy in the last 5 years, or fecal occult blood testing in the last 2 years. We used bivariate and multivariate logistic regression analyses. Of 940 patients (mean age 63.9 years, 42.7% women), 316 (33.6%) had undergone CRC screening. In multivariate analysis, birthplace in a country outside of Western Europe and North America [odds ratio (OR) 0.65, 95% confidence interval (CI) 0.45-0.97], male sex of the physician in charge (OR 0.67, 95% CI 0.50-0.91), BMI 25.0-29.9 kg/m2 (OR 0.66, CI 0.46-0.96) and at least 30.0 kg/m2 (OR 0.61, CI 0.40-0.90) were associated with lower CRC screening rates. Obesity, overweight, birthplace outside of Western Europe and North America, and male sex of the physician in charge were associated with lower CRC screening rates in Swiss University primary care settings. Physician perception of obesity and its impact on their recommendation for CRC screening might be a target for further research.

Risk Factors and Outcomes for Late Presentation for HIV-Positive Persons in Europe: Results from the Collaboration of Observational HIV Epidemiological Research Europe Study (COHERE)

Background Few studies have monitored late presentation (LP) of HIV infection over the European continent, including Eastern Europe. Study objectives were to explore the impact of LP on AIDS and mortality. Methods and Findings LP was defined in Collaboration of Observational HIV Epidemiological Research Europe (COHERE) as HIV diagnosis with a CD4 count <350/mm3 or an AIDS diagnosis within 6 months of HIV diagnosis among persons presenting for care between 1 January 2000 and 30 June 2011. Logistic regression was used to identify factors associated with LP and Poisson regression to explore the impact on AIDS/death. 84,524 individuals from 23 cohorts in 35 countries contributed data; 45,488 were LP (53.8%). LP was highest in heterosexual males (66.1%), Southern European countries (57.0%), and persons originating from Africa (65.1%). LP decreased from 57.3% in 2000 to 51.7% in 2010/2011 (adjusted odds ratio [aOR] 0.96; 95% CI 0.95–0.97). LP decreased over time in both Central and Northern Europe among homosexual men, and male and female heterosexuals, but increased over time for female heterosexuals and male intravenous drug users (IDUs) from Southern Europe and in male and female IDUs from Eastern Europe. 8,187 AIDS/deaths occurred during 327,003 person-years of follow-up. In the first year after HIV diagnosis, LP was associated with over a 13-fold increased incidence of AIDS/death in Southern Europe (adjusted incidence rate ratio [aIRR] 13.02; 95% CI 8.19–20.70) and over a 6-fold increased rate in Eastern Europe (aIRR 6.64; 95% CI 3.55–12.43). Conclusions LP has decreased over time across Europe, but remains a significant issue in the region in all HIV exposure groups. LP increased in male IDUs and female heterosexuals from Southern Europe and IDUs in Eastern Europe. LP was associated with an increased rate of AIDS/deaths, particularly in the first year after HIV diagnosis, with significant variation across Europe. Earlier and more widespread testing, timely referrals after testing positive, and improved retention in care strategies are required to further reduce the incidence of LP.

First results from a Swiss level I trauma centre participating in the UK Trauma Audit and Research Network (TARN): Prospective cohort study

QUESTIONS UNDER STUDY: Patient characteristics and risk factors for death of Swiss trauma patients in the Trauma Audit and Research Network (TARN). METHODS: Descriptive analysis of trauma patients (≥16 years) admitted to a level I trauma centre in Switzerland (September 1, 2009 to August 31, 2010) and entered into TARN. Multivariable logistic regression analysis was used to identify predictors of 30-day mortality. RESULTS: Of 458 patients 71% were male. The median age was 50.5 years (inter-quartile range [IQR] 32.2-67.7), median Injury Severity Score (ISS) was 14 (IQR 9-20) and median Glasgow Coma Score (GCS) was 15 (IQR 14-15). The ISS was >15 for 47%, and 14% had an ISS >25. A total of 17 patients (3.7%) died within 30 days of trauma. All deaths were in patients with ISS >15. Most injuries were due to falls <2 m (35%) or road traffic accidents (29%). Injuries to the head (39%) were followed by injuries to the lower limbs (33%), spine (28%) and chest (27%). The time of admission peaked between 12:00 and 22:00, with a second peak between 00:00 and 02:00. A total of 64% of patients were admitted directly to our trauma centre. The median time to CT was 30 min (IQR 18-54 min). Using multivariable regression analysis, the predictors of mortality were older age, higher ISS and lower GCS. CONCLUSIONS: Characteristics of Swiss trauma patients derived from TARN were described for the first time, providing a detailed overview of the institutional trauma population. Based on these results, patient management and hospital resources (e.g. triage of patients, time to CT, staffing during night shifts) could be evaluated as a further step.

Detection of Malnutrition in Patients Undergoing Maintenance Haemodialysis: A Quantitative Data Analysis on 12 Parameters

Background Protein-energy-malnutrition (PEM) is common in people with end stage kidney disease (ESKD) undergoing maintenance haemodialysis (MHD) and correlates strongly with mortality. To this day, there is no gold standard for detecting PEM in patients on MHD. Aim of Study The aim of this study was to evaluate if Nutritional Risk Screening 2002 (NRS-2002), handgrip strength measurement, mid-upper arm muscle area (MUAMA), triceps skin fold measurement (TSF), serum albumin, normalised protein catabolic rate (nPCR), Kt/V and eKt/V, dry body weight, body mass index (BMI), age and time since start on MHD are relevant for assessing PEM in patients on MHD. Methods The predictive value of the selected parameters on mortality and mortality or weight loss of more than 5% was assessed. Quantitative data analysis of the 12 parameters in the same patients on MHD in autumn 2009 (n = 64) and spring 2011 (n = 40) with paired statistical analysis and multivariate logistic regression analysis was performed. Results Paired data analysis showed significant reduction of dry body weight, BMI and nPCR. Kt/Vtot did not change, eKt/v and hand grip strength measurements were significantly higher in spring 2011. No changes were detected in TSF, serum albumin, NRS-2002 and MUAMA. Serum albumin was shown to be the only predictor of death and of the combined endpoint “death or weight loss of more than 5%”. Conclusion We now screen patients biannually for serum albumin, nPCR, Kt/V, handgrip measurement of the shunt-free arm, dry body weight, age and time since initiation of MHD.

Publication of statistically significant research findings in prosthodontics & implant dentistry in the context of other dental specialties

OBJECTIVES To assess the hypothesis that there is excessive reporting of statistically significant studies published in prosthodontic and implantology journals, which could indicate selective publication. METHODS The last 30 issues of 9 journals in prosthodontics and implant dentistry were hand-searched for articles with statistical analyses. The percentages of significant and non-significant results were tabulated by parameter of interest. Univariable/multivariable logistic regression analyses were applied to identify possible predictors of reporting statistically significance findings. The results of this study were compared with similar studies in dentistry with random-effects meta-analyses. RESULTS From the 2323 included studies 71% of them reported statistically significant results, with the significant results ranging from 47% to 86%. Multivariable modeling identified that geographical area and involvement of statistician were predictors of statistically significant results. Compared to interventional studies, the odds that in vitro and observational studies would report statistically significant results was increased by 1.20 times (OR: 2.20, 95% CI: 1.66-2.92) and 0.35 times (OR: 1.35, 95% CI: 1.05-1.73), respectively. The probability of statistically significant results from randomized controlled trials was significantly lower compared to various study designs (difference: 30%, 95% CI: 11-49%). Likewise the probability of statistically significant results in prosthodontics and implant dentistry was lower compared to other dental specialties, but this result did not reach statistical significant (P>0.05). CONCLUSIONS The majority of studies identified in the fields of prosthodontics and implant dentistry presented statistically significant results. The same trend existed in publications of other specialties in dentistry.

Late presentation for HIV care across Europe: update from the Collaboration of Observational HIV Epidemiological Research Europe (COHERE) study, 2010 to 2013.

Late presentation (LP) for HIV care across Europe remains a significant issue. We provide a cross-European update from 34 countries on the prevalence and risk factors of LP for 2010-2013. People aged ≥ 16 presenting for HIV care (earliest of HIV-diagnosis, first clinic visit or cohort enrollment) after 1 January 2010 with available CD4 count within six months of presentation were included. LP was defined as presentation with a CD4 count < 350/mm(3) or an AIDS defining event (at any CD4), in the six months following HIV diagnosis. Logistic regression investigated changes in LP over time. A total of 30,454 people were included. The median CD4 count at presentation was 368/mm(3) (interquartile range (IQR) 193-555/mm(3)), with no change over time (p = 0.70). In 2010, 4,775/10,766 (47.5%) were LP whereas in 2013, 1,642/3,375 (48.7%) were LP (p = 0.63). LP was most common in central Europe (4,791/9,625, 49.8%), followed by northern (5,704/11,692; 48.8%), southern (3,550/7,760; 45.8%) and eastern Europe (541/1,377; 38.3%; p < 0.0001). There was a significant increase in LP in male and female people who inject drugs (PWID) (adjusted odds ratio (aOR)/year later 1.16; 95% confidence interval (CI): 1.02-1.32), and a significant decline in LP in northern Europe (aOR/year later 0.89; 95% CI: 0.85-0.94). Further improvements in effective HIV testing strategies, with a focus on vulnerable groups, are required across the European continent.

Progression of coronary artery disease during long-term follow-up of the Swiss Interventional Study on Silent Ischemia Type II (SWISSI II)

Background: This study evaluates cardiovascular risk factors associated with progression of coronary artery disease (CAD) in patientswith silent ischemia followingmyocardial infarction. Hypothesis: Coronary artery disease only progresses slowly with comprehensive risk factor intervention. Methods: A total of 104 of 201 patients (51.7%) of the Swiss Interventional Study on Silent Ischemia Type II (SWISSI II) with baseline and follow-up coronary angiography were included. All patients received comprehensive cardiovascular risk factor intervention according to study protocol. Logistic regression was used to evaluate associationsbetween baseline cardiovascular risk factors and CAD progression. Results: The mean duration of follow-upwas 10.3 ± 2.4 years. At baseline, 77.9% of patients were smokers, 45.2% had hypertension, 73.1% had dyslipidemia, 7.7% had diabetes, and 48.1% had a family history of CAD. At last follow-up, only 27 patients of the initial 81 smokers still smoked, only 2.1% of the patients had uncontrolled hypertension, 10.6%of the patientshad uncontrolled dyslipidemia, and 2.1%of the patientshad uncontrolled diabetes. Coronary artery disease progression was found in up to 81 (77.9%) patients. Baseline diabetes and younger age were associatedwith increased odds of CAD progression.The time intervalbetween baseline and follow-up angiography was also associatedwith CAD progression. Conclusion: Coronary artery disease progressionwas highly prevalent in these patients despite comprehensive risk factor intervention. Further research is needed to optimize treatment of known risk factors and to identify other unknown and potentiallymodifiable risk factors.

Assessment of publication bias in dental specialty journals

OBJECTIVE: The purpose of this study was to investigate the presence of publication bias (acceptance of articles indicating statistically significant results). METHODS: The journals possessing the highest impact factor (2008 data) in each dental specialty were included in the study. The content of the 6 most recent issues of each journal was hand searched and research articles were classified into 4 type categories: cross-sectional, case-control, cohort, and interventional (nonrandomized clinical trials and randomized controlled trials). In total, 396 articles were included in the analysis. Descriptive statistics and univariate and multivariate logistic regression was used to examine the association between article-reported statistical significance (dependent variable) and journal impact factor and article study type subject area (independent variables). RESULTS: A statistically significant acceptance rate of positive result was found, ranging from 75% to 90%, whereas the value of impact factor was not related to publication bias among leading dental journals. Compared with other research designs, clinical intervention studies (randomized or nonrandomized) presented the highest percentage of nonsignificant findings (20%); RCTs represented 6% of the examined investigations. CONCLUSIONS: Compared with the Journal of Clinical Periodontology, all other subspecialty journals, except the Journal of Oral and Maxillofacial Surgery, showed significantly decreased odds of publishing an RCT, which ranged from 60% to 93% (P < .05).

Appropriateness of reporting statistical results in orthodontics: the dominance of P values over confidence intervals

The purpose of this study was to search the orthodontic literature and determine the frequency of reporting of confidence intervals (CIs) in orthodontic journals with an impact factor. The six latest issues of the American Journal of Orthodontics and Dentofacial Orthopedics, the European Journal of Orthodontics, and the Angle Orthodontist were hand searched and the reporting of CIs, P values, and implementation of univariate or multivariate statistical analyses were recorded. Additionally, studies were classified according to the type/design as cross-sectional, case-control, cohort, and clinical trials, and according to the subject of the study as growth/genetics, behaviour/psychology, diagnosis/treatment, and biomaterials/biomechanics. The data were analyzed using descriptive statistics followed by univariate examination of statistical associations, logistic regression, and multivariate modelling. CI reporting was very limited and was recorded in only 6 per cent of the included published studies. CI reporting was independent of journal, study area, and design. Studies that used multivariate statistical analyses had a higher probability of reporting CIs compared with those using univariate statistical analyses. Misunderstanding of the use of P values and CIs may have important implications in implementation of research findings in clinical practice.