Essential Medical Statistics

The method of generalized estimating equations (GEE) is often used to analyze longitudinal and other correlated response data, particularly if responses are binary. However, few descriptions of the method are accessible to epidemiologists. In this paper, the authors use small worked examples and one real data set, involving both binary and quantitative response data, to help end-users appreciate the essence of the method. The examples are simple enough to see the behind-the-scenes calculations and the essential role of weighted observations, and they allow nonstatisticians to imagine the calculations involved when the GEE method is applied to more complex multivariate data.

https://faculty.washington.edu/heagerty/Courses/VA-longitudinal/private/Hanley-2003AJE.pdf

Statistical Analysis of Correlated Data Using Generalized Estimating Equations: An Orientation

Background and Methods In preliminary trials, lamivudine, an oral nucleoside analogue, has shown promise for the treatment of chronic hepatitis B. We conducted a one-year, double-blind trial of lamivudine in 358 Chinese patients with chronic hepatitis B. The patients were randomly assigned to receive 25 mg of lamivudine (142 patients), 100 mg of lamivudine (143), or placebo (73) orally once daily. The patients underwent liver biopsies before entering the study and after completing the assigned treatment regimen. The primary end point was a reduction of at least two points in the Knodell necroinflammatory score. Results Hepatic necroinflammatory activity improved by two points or more in 56 percent of the patients receiving 100 mg of lamivudine, 49 percent of those receiving 25 mg of lamivudine, and 25 percent of those receiving placebo (P<0.001 and P=0.001, respectively, for the comparisons of lamivudine treatment with placebo). Necroinflammatory activity worsened in 7 percent of the patients receiving 10...

/pdf/a-one-year-trial-of-lamivudine-for-chronic-hepatitis-b-4xkh4vzkaf.pdf

A One-Year Trial of Lamivudine for Chronic Hepatitis B

Context.—Hospitals that treat a relatively high volume of patients for selected
surgical oncology procedures report lower surgical in-hospital mortality rates
than hospitals with a low volume of the procedures, but the reports do not
take into account length of stay or adjust for case mix.Objective.—To determine whether hospital volume was inversely associated with 30-day
operative mortality, after adjusting for case mix.Design and Setting.—Retrospective cohort study using the Surveillance, Epidemiology, and
End Results (SEER)–Medicare linked database in which the hypothesis
was prospectively specified. Surgeons determined in advance the surgical oncology
procedures for which the experience of treating a larger volume of patients
was most likely to lead to the knowledge or technical expertise that might
offset surgical fatalities.Patients.—All 5013 patients in the SEER registry aged 65 years or older at cancer
diagnosis who underwent pancreatectomy, esophagectomy, pneumonectomy, liver
resection, or pelvic exenteration, using incident cancers of the pancreas,
esophagus, lung, colon, and rectum, and various genitourinary cancers diagnosed
between 1984 and 1993.Main Outcome Measure.—Thirty-day mortality in relation to procedure volume, adjusted for comorbidity,
patient age, and cancer stage.Results.—Higher volume was linked with lower mortality for pancreatectomy (P=.004), esophagectomy (P<.001),
liver resection (P=.04), and pelvic exenteration
(P=.04), but not for pneumonectomy (P=.32). The most striking results were for esophagectomy, for which
the operative mortality rose to 17.3% in low-volume hospitals, compared with
3.4% in high-volume hospitals, and for pancreatectomy, for which the corresponding
rates were 12.9% vs 5.8%. Adjustments for case mix and other patient factors
did not change the finding that low volume was strongly associated with excess
mortality.Conclusions.—These data support the hypothesis that when complex surgical oncologic
procedures are provided by surgical teams in hospitals with specialty expertise,
mortality rates are lower.

https://jamanetwork.com/journals/jama/articlepdf/188202/JOC80665.pdf

Impact of hospital volume on operative mortality for major cancer surgery.

Aims: To develop and evaluate the International Consultation on Incontinence Questionnaire (ICIQ), a new questionnaire to assess urinary incontinence and its impact on quality of life (QoL). Methods: A developmental version of the questionnaire was produced following systematic literature review and views of an expert committee and patients. Several studies were undertaken to evaluate the psychometric properties of the questionnaire, including content, construct and convergent validity, reliability and sensitivity to change. Results: The ICIQ was easily completed, with low levels of missing data (mean 1.6%). It was able to discriminate among diierent groups of individuals, indicating good construct validity. Convergent validity was acceptable, with most items demonstrating ‘moderate’ to ‘strong’ agreement with other questionnaires. Reliability was good, with ‘moderate’ to ‘very good’ stability in test-retest analysis and a Cronbach’s alpha of 0.95. Items identi¢ed statistically signi¢cant reductions in symptoms from baseline following surgical and conservative treatment. Item reduction techniques were used to determine the ¢nal version and scoring scheme, which also demonstrated good psychometric properties. Conclusions: The ¢nal ICIQ comprises three scored items and an unscored self-diagnostic item. It allows the assessment of the prevalence, frequency, and perceived cause of urinary incontinence, and its impact on everyday life. The ICIQ is a brief and robust questionnaire that will be of use in outcomes and epidemiological research as well as routine clinical practice. Neurourol. Urodynam. 23:322 ^330, 2004. 2004 Wiley-Liss, Inc.

G. Berry

Papers

Statistical Methods in Medical Research.