Missing data frequency and correlates in two randomized surgical trials for urinary incontinence in women

Introduction and hypothesis

Missing data is frequently observed in clinical trials; high rates of missing data may jeopardize trial outcome validity.

Purpose

We determined the rates of missing data over time, by type of data collected and compared demographic and clinical factors associated with missing data among women who participated in two large randomized clinical trials of surgery for stress urinary incontinence, the Stress Incontinence Surgical Treatment Efficacy Trial (SISTEr) and the Trial of Midurethral Sling (TOMUS).

Methods

The proportions of subjects who attended and missed each follow-up visit were calculated. The chi-squared test, Fisher’s exact test and t test were used to compare women with and without missing data, as well as the completeness of the data for each component of the composite primary outcome.

Results

Data completeness for the primary outcome computation in the TOMUS trial (62.3 %) was nearly double that in the SISTEr trial (35.7 %). The follow-up visit attendance rate decreased over time. A higher proportion of subjects attended all follow-up visits in the TOMUS trial and overall there were fewer missing data for the period that included the primary outcome assessment at 12 months. The highest levels of complete data for the composite outcome variables were for the symptoms questionnaire (SISTEr 100 %, TOMUS 99.8 %) and the urinary stress test (SISTEr 96.1 %, TOMUS 96.7 %). In both studies, the pad test was associated with the lowest levels of complete data (SISTEr 85.1 %, TOMUS 88.3 %) and approximately one in ten subjects had incomplete voiding diaries at the time of primary outcome assessment. Generally, in both studies, a higher proportion of younger subjects had missing data. This analysis lacked a patient perspective as to the reasons for missing data that could have provided additional information on subject burden, motivations for adherence and study design. In addition, we were unable to compare the effects of the different primary outcome assessment time-points in an identically designed trial.

Conclusions

Missing visits and data increased with time. Questionnaire data and physical outcome data (urinary stress test) that could be assessed during a visit were least prone to missing data, whereas data for variables that required subject effort while away from the research team (pad test, voiding diary) were more likely to be missing. Older subjects were more likely to provide complete data.

     

A retrospective audit of postoperative days alive and out of hospital,including before and after implementation of the WHO surgical safety checklist

We implemented the World Health Organization surgical safety checklist at Auckland City Hospital from November 2007. We hypothesised that the checklist would reduce postoperative mortality and increase days alive and out of hospital, both measured to 90 postoperative days. We compared outcomes for cohorts who had surgery during 18-month periods before vs. after checklist implementation. We also analysed outcomes during 9 years that included these periods (July 2004–December 2013). We analysed 9475 patients in the 18-month period before the checklist and 10,589 afterwards. We analysed 57,577 patients who had surgery from 2004 to 2013. Mean number of days alive and out of hospital (95%CI) in the cohort after checklist implementation was 1.0 (0.4–1.6) days longer than in the cohort preceding implementation, p < 0.001. Ninety-day mortality was 395/9475 (4%) and 362/10,589 (3%) in the cohorts before and after checklist implementation, multivariable odds ratio (95%CI) 0.93 (0.80–1.09), p = 0.4. The cohort changes in these outcomes were indistinguishable from longer-term trends in mortality and days alive and out of hospital observed during 9 years, as determined by Bayesian changepoint analysis. Postoperative mortality to 90 days was 228/5686 (4.0%) for Māori and 2047/51,921 (3.9%) for non-Māori, multivariable odds ratio (95%CI) 0.85 (0.73–0.99), p = 0.04. Māori spent on average (95%CI) 1.1 (0.5–1.7) fewer days alive and out of hospital than non-Māori, p < 0.001. In conclusion, our patients experienced improving postoperative outcomes from 2004 to 2013, including the periods before and after implementation of the surgical checklist. Māori patients had worse outcomes than non-Māori.     

A dynamic block-randomization algorithm for group-randomized clinical trials when the composition of blocking factors is not known in advance

We present an algorithm for randomizing units in blocks for controlled trials when the composition of blocking factors is not known in advance. For example, suppose the desired goal of an intervention study is to randomize units to one of two interventions while blocking on a dichotomous factor (e.g., gender), but the total number of units, and therefore number or composition, of males and females among those units assembled for randomization cannot be determined in advance. This situation arises in randomized trials when subjects are scheduled to be randomized as a group, but not all of the subjects show up for the visit. Since investigators do not know which of the scheduled subjects will or will not show up, a dynamic randomization scheme is required to accommodate the unknown composition of the blocking factor once a group of subjects (units) is assembled for randomization. These settings are further complicated when there is more than one blocking factor. In this paper, we present an algorithm that ensures the integrity of the randomization process in these settings.     

Trends in survival of leukemia among children,adolescents, and young adults: A population-based study in Osaka,Japan

This study focused on children as well as adolescents and young adults (AYAs) and aimed to examine trends in survival of leukemia over time using population-based cancer registry data from Osaka, Japan. The study subjects comprised 2254 children (0-14 years) and 2,905 AYAs (15-39 years) who were diagnosed with leukemia during 1975-2011. Leukemia was divided into four types: acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), chronic myeloid leukemia (CML), and other leukemias. We analyzed 5-year overall survival probability (5y-OS), using the Kaplan-Meier method and expressed time trends using the joinpoint regression model. For recently diagnosed (2006-2011) patients, a Cox proportional hazards model was applied to determine predictors of 5y-OS, using age group, gender, and treatment hospital as covariates. Over the 37-year period, 5y-OS greatly improved among both children and AYAs, for each leukemia type. Among AYAs, 5y-OS of ALL improved, especially after 2000 (65% in 2006-2011), when the pediatric regimen was introduced but was still lower than that among children (87% in 2006-2011, P < .001). Survival improvement was most remarkable in CML, and its 5y-OS was over 90% among both children and AYAs after the introduction of molecularly targeted therapy with tyrosine kinase inhibitors. Among patients with recently diagnosed AML, the risk of death was significantly higher for patients treated at nondesignated hospitals than those treated at designated cancer care hospitals. The changes in survival improvement coincided with the introduction of treatment regimens or molecularly targeted therapies. Patient centralization might be one option which would improve survival.