Flexible parametric modelling of the cause-specific cumulative incidence function.

Competing risks arise with time-to-event data when individuals are at risk of more than one type of event and the occurrence of one event precludes the occurrence of all other events. A useful measure with competing risks is the cause-specific cumulative incidence function (CIF), which gives the probability of experiencing a particular event as a function of follow-up time, accounting for the fact that some individuals may have a competing event. When modelling the cause-specific CIF, the most common model is a semi-parametric proportional subhazards model. In this paper, we propose the use of flexible parametric survival models to directly model the cause-specific CIF where the effect of follow-up time is modelled using restricted cubic splines. The models provide smooth estimates of the cause-specific CIF with the important advantage that the approach is easily extended to model time-dependent effects. The models can be fitted using standard survival analysis tools by a combination of data expansion and introducing time-dependent weights. Various link functions are available that allow modelling on different scales and have proportional subhazards, proportional odds and relative absolute risks as particular cases. We conduct a simulation study to evaluate how well the spline functions approximate subhazard functions with complex shapes. The methods are illustrated using data from the European Blood and Marrow Transplantation Registry showing excellent agreement between parametric estimates of the cause-specific CIF and those obtained from a semi-parametric model. We also fit models relaxing the proportional subhazards assumption using alternative link functions and/or including time-dependent effects. Copyright © 2016 John Wiley & Sons, Ltd.

Is speed of healing a good predictor of eventual healing of pyoderma gangrenosum?

BACKGROUND: Pyoderma gangrenosum is a rare inflammatory skin condition. Two prospective studies have evaluated treatments for pyoderma gangrenosum using a primary outcome of healing speed at 6 weeks. OBJECTIVE: Using data from both studies we assessed the predictive value of 3 early predictors for healing at 6 months: speed of healing, Investigator Global Assessment (IGA), and resolution of inflammation, recorded at 2 and 6 weeks. METHODS: Logistic regression models were applied and the effectiveness of the 3 measures was assessed through estimating the positive and negative predictive values and the area under the receiver operating characteristic curve. RESULTS: The positive and negative predictive value at 6 weeks were, respectively, 63.5% (95% confidence interval [CI] 52.4%-73.7%) and 74.6% (95% CI 62.5%-84.5%) for speed of healing; 80% (95% CI 68.7%-88.6%) and 74.2% (95% CI 64.1%-82.7%) for IGA; and 72.1% (95% CI 59.9%-82.3%) and 68.1% (95% CI 57.7%-77.3%) for resolution of inflammation. IGA had the best combined positive predictive value, negative predictive value, and area under the receiver operating characteristic curve at 2 and 6 weeks. LIMITATIONS: We were limited by data available from existing datasets. CONCLUSION: Speed of healing, IGA, and resolution of inflammation were all shown to be good predictors of eventual healing of pyoderma gangrenosum.

Strategies used for measuring long-term control in atopic dermatitis trials: A systematic review.

BACKGROUND: Atopic dermatitis (AD) is a chronic inflammatory skin disease. There are no standardized methods for capturing long-term control of AD. OBJECTIVE: We sought to identify how long-term control has been captured in published randomized controlled trials (RCTs). Results will initiate consensus discussions on how best to measure long-term control in the core outcome set for AD. METHODS: We conducted a systematic review of RCTs of AD treatments published between 2000 and 2013, with a follow-up period of 3 months or longer, at least 1 outcome measure recorded at 3 or more time points, full article available, and published in English. RESULTS: In all, 101 of 353 RCTs were eligible. Methods to capture long-term control included: repeated measurement of AD outcomes (92 RCTs; 91%), use of AD medication (29 RCTs; 28.7%), and AD flares/remissions (26 RCTs; 25.7%). Repeated measurements of AD outcomes were typically collected 3 to 5 times during a trial, but analysis methods often failed to make best use of the data. Time to first flare was most commonly used for trials including flare data (21/52). Medication use was recorded based on quantity, potency, and frequency of application. LIMITATIONS: We included RCT data only. CONCLUSION: This review illustrates the difficulties in measuring long-term control, and points to the need for improved harmonization of outcomes.

How Clinically Relevant Are Treatment Comparisons of Topical Calcineurin Inhibitor Trials for Atopic Eczema?

We sought to explore the architecture of trials of calcineurin inhibitors for atopic eczema to document the extent to which comparisons with active treatments such as topical corticosteroids might have been included or avoided. We identified all eligible randomized controlled trials using the Global Resource for EczemA Trials (GREAT) database. Network plots were produced where the nodes represented a treatment type and the lines between the nodes represented the number of trials or participants involved in the various treatment comparisons. A total of 174 randomized controlled trials for atopic eczema treatments were identified in which pimecrolimus, tacrolimus, or topical corticosteroids were compared with another intervention or a vehicle/emollient. Of 39 trials involving pimecrolimus and 41 trials involving tacrolimus, 8 (20.5%) and 13 (31.7%), respectively, made comparisons with topical corticosteroids, and 25 (64.1%) and 15 (36.6%), respectively, were vehicle-controlled studies. The high rate of comparisons with vehicle controls in randomized controlled trials that assessed the efficacy of pimecrolimus or tacrolimus long after efficacy had been established is a matter of concern. Active comparators (mild topical corticosteroids for pimecrolimus and moderate to potent topical corticosteroids for tacrolimus) are best placed to determine how topical calcineurin inhibitors compare with established clinical practice.

How Often Do Comparative Randomised Controlled Trials in the Field of Eczema Fail to Directly Compare the Treatments Being Tested?

The objective of the study was to identify all parallel design randomised controlled trials (RCTs) comparing treatments for eczema in recent dermatology literature that have failed to report a between-group analysis. The GREAT database (www.greatdatabase.org.uk) was searched to identify parallel group RCTs comparing two or more interventions published in the English language in the last decade, 2004 to 2013. The primary outcome was the number of studies that had not reported a between-group analysis for any of the outcomes. Where possible we re-analysed the data to determine whether a between-group analysis would have given a different conclusion to that reported. Out of a total of 304 RCTs in the study period, 173 (56.9%) met the inclusion criteria. Of the 173 eligible studies, 12 (6.9%) had not conducted a between-group analysis for any of the reported outcomes. There was no clear improvement over time. Five of the eight studies that were re-analysed yielded non-significant between-group differences yet reported significant within-group comparisons. All but one of the 12 studies implied that the experimental intervention was successful despite not undertaking any between-group comparisons. Although the proportion of all RCTs that fail to report an appropriate between-group analysis is small, the fact that any scientist who purports to compare one treatment against another then chooses to omit the key comparison statistic is worrying.

National trends in acute kidney injury requiring dialysis in England between 1998 and 2013.

Acute kidney injury (AKI) severe enough to require dialysis is increasing and associated with high mortality, yet robust information about temporal epidemiology of AKI requiring dialysis in England is lacking. In this retrospective observational study of the Hospital Episode Statistics (HES) data set covering the entire English National Health Service, we identified all patients with a diagnosis of AKI requiring dialysis between 1998 and 2013. This incidence increased from 774 cases (15.9 per million people) in 1998-1999 to 11,164 cases (208.7 per million people) in 2012-2013. The unadjusted in-hospital case-fatality was 30.3% in 1998-2003 and 30.2% in 2003-2008, but significantly increased to 41.1% in 2008-2013. Compared with 2003-2008, the multivariable adjusted odds ratio for death was higher in 1998-2003 at 1.20 (95% CI: 1.10-1.30) and in 2008-2013 at 1.13 (1.07-1.18). Charlson comorbidity scores of more than five (odds ratio 2.35; 95% CI: 2.20-2.51) and emergency admissions (2.46 (2.32-2.61) had higher odds for death. The odds for death decreased in patients over 85 years from 4.83 (3.04-7.67) in 1998-2003 to 2.19 (1.99-2.41) in 2008-2013. AKI in secondary diagnosis and in other diagnoses codes had higher odds for death compared with AKI in primary diagnosis code in all three periods. Thus, the incidence of AKI requiring dialysis has increased progressively over 15 years in England. Improvement in case-fatality in 2003-2008 has not been sustained in the last 5 years.

Risk and Cause of Death in Patients Diagnosed With Myeloproliferative Neoplasms in Sweden Between 1973 and 2005: A Population-Based Study.

PURPOSE: Myeloproliferative neoplasms (MPNs) are associated with a shortened life expectancy. We assessed causes of death in patients with MPN and matched controls using both relative risks and absolute probabilities in the presence of competing risks. PATIENTS AND METHODS: From Swedish registries, we identified 9,285 patients with MPN and 35,769 matched controls. A flexible parametric model was used to estimate cause-specific hazard ratios (HRs) of death and cumulative incidence functions, each with 95% CIs. RESULTS: In patients with MPN, the HRs of death from hematologic malignancies and infections were 92.8 (95% CI, 70.0 to 123.1) and 2.7 (95% CI, 2.4 to 3.1), respectively. In patients age 70 to 79 years at diagnosis (the largest patient group), the HRs of death from cardiovascular and cerebrovascular disease were 1.5 (95% CI, 1.4 to 1.7) and 1.5 (95% CI, 1.3 to 1.8), respectively; all were statistically significantly elevated compared with those of controls. In the same age group, no difference was observed in the 10-year probability of death resulting from cardiovascular disease in patients with MPN versus controls (16.8% v 15.2%) or cerebrovascular disease (5.6% v 5.2%). In patients age 50 to 59 years at diagnosis, the 10-year probability of death resulting from cardiovascular and cerebrovascular disease was elevated, 4.2% versus 2.1% and 1.9% versus 0.4%, respectively. Survival in patients with MPN increased over time, mainly because of decreased probabilities of dying as a result of hematologic malignancies, infections, and, in young patients, cardiovascular disease. CONCLUSION: Patients with MPN had an overall higher mortality rate than that of matched controls, primarily because of hematologic malignancy, infections, and vascular events in younger patients. Evidently, there is still a need for effective disease-modifying agents to improve patient outcomes.

Physiotherapy treatment for atraumatic recurrent shoulder instability: early results of a specific exercise protocol using pathology-specific outcome measures.

BACKGROUND: Recurrent shoulder instability is usually caused by a traumatic event resulting in structural pathology, although a small subgroup of patients experience symptomatic recurrent shoulder instability without trauma. These patients are usually treated non-operatively but limited evidence exists regarding effective conservative management. In particular, there is a lack of reproducible exercise regimes and none that have been tested with condition-specific outcome measures. METHODS: A service evaluation was conducted over a 15-month period to assess our current treatment protocol used in the management of patients with atraumatic recurrent shoulder instability. The regime is reproducible with target-led progression milestones. Oxford Instability Shoulder Scores (OISS) and Western Ontario Shoulder Index (WOSI) scores were compared between baseline and final follow-up. RESULTS: Eighteen consecutive patients were included with mean follow-up of 4.5 months (range 1.35 months to 11.77 months). A statistically significant improvement was seen in both outcome measures. Mean OISS improved by 16.67 points (confidence interval: 12.34 to 20.99; p < 0.001). Mean WOSI improved by 36.76% (confidence interval: 28.46 to 45.06; p < 0.001). CONCLUSIONS: For this small group of patients with recurrent atraumatic shoulder instability, the Derby Shoulder Instability Programme produced significant improvements over the short term, with a high level of patient compliance. This is the first study to include pathology-specific patient-reported outcome measures to assess outcomes from a specific and reproducible exercise regime in this group of patients. The findings support further research to evaluate the exercise protocol in a larger group of patients over the longer term.