Two Emerging Issues in Epidemiologic Research Access to populations for epidemiologic study Side effects of specialization.

Two Emerging Issues in Epidemiologic Research

• Access to populations for epidemiologic study

• Side effects of specialization

Issue #1: Access to Populationsfor Epidemiologic Study

• Often use pre-existing data

• But need primary data collection when:– No surveillance system exists

– Need exposure data not already available

• Common situation: population-based case-control study

• Present in ~95% of U.S. households

• Used for sampling, data collection, or both

• Random digit dialing (RDD)– Yields probability sample

– Captures unlisted phone numbers

Number of MedLine Citations with “Random digit” and “Epidemiologic-methods” by Year

0102030405060708090

100

1975-9 1980-4 1985-9 1990-4 1995-9 2000-2

Calendar year

Now-Standard RDD Refinements

• Distribution of calls among weekday, evening, weekend

• Later recontacting and conversion of “soft refusers”

• High-yield sampling methods:– Mitofsky-Waksberg two-stage sampling

– List-assisted sampling

Are RDD response rates declining?

Reported Participation among Controls in 69 Case-Control Studies, by Year of Publication

(Olson, 2001)

Source: Am J Epidemiol 2001; 154:574-81

Trend in Response Rates in CDC Behavioral Risk Factor Surveillance System, 1994-2000

0

10

20

30

40

50

60

70

80

1994 1995 1996 1997 1998 1999 2000

Calendar year

Source: BRFSS 1998-9 Quality Control Reports

Trend in Calling Effort in Behavioral Risk Factor Surveillance System, 1995-1999

0

1

2

3

4

5

6

7

1995 1996 1997 1998 1999

Calendar year

Phone numbers called per

eligible respondent

Source: BRFSS 1999 Quality Control Report

“And to think if I hadn’t been home having dinner I might have missed this wonderful

investment opportunity.”

Impact of Telemarketing in the U.S.

• Employed ~6 million people in 2001

• Generated $257 billion in direct sales to consumers in 2000

• ~19 billion telemarketing dialings each month

• “SUGging” - Selling Under the Guise of a survey

• “FRUGging” - Fund Raising Under the Guise of a survey

Other Contributing Factors

• More single-person households

• More households in which all adults work outside the home

• Answering machines

• Caller ID

“I don’t know why - I just suddenly felt like calling.”

“As of July 2002 CMS will no longer release the Medicare Names and Addresses file to researchers. CMS has imposed a moratorium on the release of this file as means for notifying beneficiaries about research studies.

“We are reevaluating our procedures to ensure that we continue to have in place the best privacy protections for Medicare beneficiaries.”

Source: www.cms.gov/data/requests/moratorium.asp, accessed 6/5/03

Center for Medicare and Medicaid Services Moratorium

Some Unfortunate Possible Responses

• Use only pre-existing data

• Use only populations of convenience– Self-identified volunteers

– Existing samples from prior studies

– Familiar, easy-access settings: workplaces, etc.

• Shift work abroad

Some General Measures We Can Take

• Work toward more standardized reporting of study participation

• Promote a positive public image of epidemiologic research– Feed back interesting information to participants

– Be mindful of respondent burden

– Minimize discomfort

• Scrupulously protect privacy and confidentiality

Specific Coping Strategies:1. Worry More Selectively?

• Bias due to non-response depends on:

(a) Proportion who participate

(b) Extent of difference between participants

and non-participants

• Sometimes (b) is small

• But can’t always predict when

Specific Coping Strategies:2. More Incremental Improvements to RDD

For example...

• Use answering-machine messages to distinguish research from telemarketing

• Increase number of call-backs

• Optimize selection and training of interviewers

• Offer incentives for participation

The National “Do Not Call” Registry

Percentage of U.S. Households with Internet Access by Annual Household Income, 2001

0 20 40 60 80 100

<$5K

$5-10K

$10-15K

$15-20K

$20-25K

$25-35K

$35-50K

$50-75K

$75K+

Source: Bureau of the Census Current Population Survey

Percentage of Households

Possible Future Portals of Access to Populations: 1. Internet

• Useful data-collection method once study participation arranged

• Not (yet?) a promising successor to RDD for sampling

Possible Future Portals of Access to Populations: 2. Cellular Phones

• Linked more to individuals, not households

• Population coverage may become very high

• Fewer calling plans charge for incoming calls

• Text messaging or other features may permit

initial contact without annoying interruptions

Some Currently Available Alternatives

• Area sampling

• Lists

– Commercial mailing lists

– Drivers license records

– Voter registration lists

– Other: town lists, dwelling units, etc.

Issue #2: Side Effects of Specialization

How epidemiologists vary:

• Chronic disease / Infectious disease

• Clinical training / Non-clinical background

• Academia / Government / Industry / Other

• Basic research / Field epidemiology

• By disease or exposure area

Side Effects of Specialization

• Fragmentation and

compartmentalization

• Risk of intellectual isolation from

each other

Divisions and Interdependencies

• Chronic disease / Infectious disease

• Clinical training / Non-clinical background

• Academia / Government / Industry / Other

• Basic research / Field epidemiology

• By disease area or exposure area

Thank you