Confounding Study

Learning Objectives

>

Introduction to Study

>

Student Role

>

Dr Stellman's Study

>

Dr Shapiro's Study

>

Discussion Questions

Step 3: Student Role - Your Plan of Action

You need to first familiarize yourself with these studies.

1. Listen to the introduction about the two studies
2. Read the following synopsis of each study

Questions in steps 4 and 5 require you to demonstrate critical thinking and knowledge of epidemiological concepts. Read carefully through the explanations of both correct and incorrect answers. Finally, answer the discussion questions in Step 6 found at the end of the exercise. Bring your answers to you seminar section and be prepared to discuss them in class.

Dr. Steven Stellman [ Listen ] [ Read Transcript ] [ Read Study Synopsis ]

Dr. Syd Shapiro [ Listen ] [ Read Transcript ] [ Read Study Synopsis ]

Synopsis 1: Artificial Sweetener and Bladder Cancer, S. Stellman et al.

Note: These synopses will be used as a background material for homeworks on Bias and Confounding.

Objectives

To assess whether use of artificial sweetener in daily diet increases the risk of bladder cancer.

Hypothesis

Artificial sweetener (AS) and diet beverage (DB) use is associated with bladder cancer.

Design

This is a matched case-control study.

Controls were matched to cases on age (in decades), sex, hospital, and hospital-room status (private, semiprivate, or ward). This was a 1:1 matching with matches found for all but 10 male cases and 14 female cases.

Synopsis 2: Recent and Past Use of Conjugated Estrogens in Relation to Adenocarcinoma of the Endometrium, Shapiro S., Kaufman D.W., et al.

Objectives

To determine whether prior use of estrogen is associated with endometrial cancer

Hypotheses

• Asymptomatic endometrial cancer is not caused by estrogen use.
• Bleeding caused by estrogen use does not cause asymptomatic endometrial cancer.

Design

Matched case-control study

Matching

1:4 (up to 4 controls were matched to each case according to decade of age and geographic areas).

Population at risk for disease

Postmenopausal women aged 50 to 69 years from Eastern Seaboard, Kansas, Arizona, California, and Canada.

Source Population

It is difficult to establish the precise source population for a hospital case-control study; cases might have come from far away to receive specialized treatment, while controls might have lived in the neighborhood surrounding the hospital.

Eligibility Criteria for Cases and Controls

• Cases: postmenopausal women aged 50-69, admitted to the hospitals located on the eastern seaboard, Kansas, Arizona, California, and Canada.
• Controls: women who were admitted for conditions not related to prior estrogen use from the same hospitals as cases and during the same time period.

Diagnoses of Matched Controls

Methods of Accrual of Cases and Controls

• Cases: all newly admitted patients with a diagnosis of endometrial cancer were identified and interviewed.
• Controls: female patients admitted to the medical, surgical, and orthopedic wards with diagnoses other than endometrial cancer were sampled in a systemic manner and interviewed.

Data Collection:

Measurement of Exposure: a questionnaire was used with questions pertaining to lifetime histories of regular use of noncontraceptive estrogens for any of the following indications: regulation of periods, menstrual problems, infertility, breast conditions, endometriosis, sexual difficulties, and menopausal symptoms.

Measurement of Outcome: diagnosis of adenocarcinoma of the endometrium recorded either in the discharge summary or the pathology report, within a year of the current admission.

Exclusion criteria:

• 5 cases and 11 controls who first used a noncontraceptive estrogen within two years of the date of diagnosis (for cases) or for whom the date of the first use was unknown were excluded (for controls)
• Use of noncontraceptive estrogens for a total duration of less than three months
• Use of unspecified female hormone only (20cases and 90 controls)

Data Analysis

Total number of cases: 149
Total number of controls: 453

The proportion of cases who used conjugated estrogens was greater among cases than controls (see Table 1).

Table 1: Relation of Use of Noncontraceptive Estrogens among 149 Cases and 402 Controls

Odds ratio estimates together with their 95% confidence limits were computed for various categories of estrogen use. Conjugated estrogen use was a statistically significant predictor of endometrial cancer (Table 2).

Table 2. Relation of Use of Noncontraceptive Estrogens to Risk of Endometrial Cancer among 149 Cases and 402 Controls

Conjugated estrogens use played a statistically significant role for all categories of time elapsed since latest use, except for the last time category, ≥ 5 yr (Table 3).

Table 3. Relation of Use of Conjugated Estrogens for Five Years or More to Risk of Endometrial Cancer, According to Time Elapsed since Latest Use

Results: The rate of endometrial cancer was higher in women who used conjugated estrogens, relative to those who did not. There was no evidence of an association for use lasting less than one year but the risk increased with duration of use.

Questions in steps 3, 4, and 5 require you to demonstrate critical thinking and knowledge of epidemiological concepts. Read carefully through the explanations of both correct and incorrect answers. Finally, answer the discussion questions in Step 6 found at the end of the exercise. Bring your answers to you seminar section and be prepared to discuss them in class. Please proceed to Step 4.