In recent years, multiple imputation has emerged as a convenient and flexible paradigm for analysing data with missing values. Essential features of multiple imputation are reviewed, with answers to frequently asked questions about using the method in practice.

Multiple imputation: a primer:

Objective: Despite growing interest in adult attention deficit hyperactivity disorder (ADHD), little is known about its prevalence or correlates. Method: A screen for adult ADHD was included in a probability subsample (N=3,199) of 18–44-year-old respondents in the National Comorbidity Survey Replication, a nationally representative household survey that used a lay-administered diagnostic interview to assess a wide range of DSM-IV disorders. Blinded clinical follow-up interviews of adult ADHD were carried out with 154 respondents, oversampling those with positive screen results. Multiple imputation was used to estimate prevalence and correlates of clinician-assessed adult ADHD. Results: The estimated prevalence of current adult ADHD was 4.4%. Significant correlates included being male, previously married, unemployed, and non-Hispanic white. Adult ADHD was highly comorbid with many other DSM-IV disorders assessed in the survey and was associated with substantial role impairment. The majority of cases were u...

/pdf/the-prevalence-and-correlates-of-adult-adhd-in-the-united-5fbxyszdr6.pdf

The prevalence and correlates of adult ADHD in the United States: results from the National Comorbidity Survey Replication.

Background Clinical trials have established the efficacy of ranibizumab for the treatment of neovascular age-related macular degeneration (AMD). In addition, bevacizumab is used off-label to treat AMD, despite the absence of similar supporting data. Methods In a multicenter, single-blind, noninferiority trial, we randomly assigned 1208 patients with neovascular AMD to receive intravitreal injections of ranibizumab or bevacizumab on either a monthly schedule or as needed with monthly evaluation. The primary outcome was the mean change in visual acuity at 1 year, with a noninferiority limit of 5 letters on the eye chart. Results Bevacizumab administered monthly was equivalent to ranibizumab administered monthly, with 8.0 and 8.5 letters gained, respectively. Bevacizumab administered as needed was equivalent to ranibizumab as needed, with 5.9 and 6.8 letters gained, respectively. Ranibizumab as needed was equivalent to monthly ranibizumab, although the comparison between bevacizumab as needed and monthly bevacizumab was inconclusive. The mean decrease in central retinal thickness was greater in the ranibizumab-monthly group (196 μm) than in the other groups (152 to 168 μm, P=0.03 by analysis of variance). Rates of death, myocardial infarction, and stroke were similar for patients receiving either bevacizumab or ranibizumab (P>0.20). The proportion of patients with serious systemic adverse events (primarily hospitalizations) was higher with bevacizumab than with ranibizumab (24.1% vs. 19.0%; risk ratio, 1.29; 95% confidence interval, 1.01 to 1.66), with excess events broadly distributed in disease categories not identified in previous studies as areas of concern. Conclusions At 1 year, bevacizumab and ranibizumab had equivalent effects on visual acuity when administered according to the same schedule. Ranibizumab given as needed with monthly evaluation had effects on vision that were equivalent to those of ranibizumab administered monthly. Differences in rates of serious adverse events require further study. (Funded by the National Eye Institute; ClinicalTrials.gov number, NCT00593450.).

/pdf/ranibizumab-and-bevacizumab-for-neovascular-age-related-7eskj9tklx.pdf

Ranibizumab and bevacizumab for neovascular age-related macular degeneration.

BACKGROUND Hospitals have difficulty justifying the expense of maintaining trauma centers without strong evidence of their effectiveness. To address this gap, we examined differences in mortality between level 1 trauma centers and hospitals without a trauma center (non-trauma centers). METHODS Mortality outcomes were compared among patients treated in 18 hospitals with a level 1 trauma center and 51 hospitals non-trauma centers located in 14 states. Patients 18 to 84 years old with a moderate-to-severe injury were eligible. Complete data were obtained for 1104 patients who died in the hospital and 4087 patients who were discharged alive. We used propensity-score weighting to adjust for observable differences between patients treated at trauma centers and those treated at non-trauma centers. RESULTS After adjustment for differences in the case mix, the in-hospital mortality rate was significantly lower at trauma centers than at non-trauma centers (7.6 percent vs. 9.5 percent; relative risk, 0.80; 95 percent confidence interval, 0.66 to 0.98), as was the one-year mortality rate (10.4 percent vs. 13.8 percent; relative risk, 0.75; 95 percent confidence interval, 0.60 to 0.95). The effects of treatment at a trauma center varied according to the severity of injury, with evidence to suggest that differences in mortality rates were primarily confined to patients with more severe injuries. CONCLUSIONS Our findings show that the risk of death is significantly lower when care is provided in a trauma center than in a non-trauma center and argue for continued efforts at regionalization.

https://www.nejm.org/doi/pdf/10.1056/NEJMsa052049

A National Evaluation of the Effect of Trauma-Center Care on Mortality

Correspondence analysis is an exploratory tool for the analysis of associations between categorical variables, the results of which may be displayed graphically. For longitudinal data, two types of analysis can be distinguished: the first focuses on transitions, whereas the second investigates trends. For transitional analysis with two time points, an analysis of the transition matrix (showing the relative frequencies for pairs of categories) provides insight into the structure of departures from independence in the transitions. Transitions between more than two time points can also be studied simultaneously. In trend analyses often the trajectories of different groups are compared. Examples for all these analyses are provided. Correspondence analysis is an exploratory tool for the analysis of association(s) between categorical variables. Usually, the results are displayed in a graphical way. There are many interpretations of correspondence analysis. Here, we make use of two of them. A first interpretation is that the observed categorical data are collected in a matrix, and correspondence analysis approximates this matrix by a matrix of lower rank[1]. This lower rank approximation of, say, rank M + 1 is then displayed graphically in an M-dimensional representation in which each row and each column of the matrix is displayed as a point. The difference in rank between the rank M + 1 matrix and the rank M representation is matrix of rank 1, and this matrix is the product of the marginal counts of the matrix, that is most often considered uninteresting. This brings us to the second interpretation, that is, that when the two-way matrix is a contingency table, correspondence analysis decomposes the departure from a matrix where the row and column variables are independent[2,3]. Thus, correspondence analysis is a tool for residual analysis. This interpretation holds because for a contingency table estimates under the independence model are obtained from the product of the margins of the table divided by the total sample size. Longitudinal data are data where observations (e.g., individuals) are measured at least twice using the same variables. We consider here only categorical (i.e., nominal or ordinal) variables, as only this kind of variables is analyzed in standard applications of correspondence analysis[4]. We first discuss correspondence analysis for the analysis of transitions. Thereafter, we consider analysis of trends with canonical correspondence analysis. 1 Leiden University, Leiden, The Netherlands 2 Utrecht University, Utrecht, The Netherlands Update based on original article by Peter G. M. Van Der Heijden, Wiley StatsRef: Statistics Reference Online, © 2014, John Wiley & Sons, Ltd Wiley StatsRef: Statistics Reference Online, © 2014–2015 John Wiley & Sons, Ltd. This article is © 2015 John Wiley & Sons, Ltd. DOI: 10.1002/9781118445112.stat05497.pub2 1 Correspondence Analysis of Longitudinal Data 1 Transitional Analysis

Analysis of Longitudinal Data.

25. Multiple Imputation for Nonresponse in Surveys. By D. B. Rubin. ISBN 0 471 08705 X. Wiley, Chichester, 1987. 258 pp. £30.25.

Roger A. Sugden

Papers

Multiple Imputation for Nonresponse in Surveys