Know Your Source: Surveys Can Zero in on a Million Health Topics

Author(s)
Published on
December 14, 2012

If we think about a census as a butterfly net that captures a wide array of information about health and other topics, we might think about specialized surveys as magnifying glasses that let researchers study topics more closely.

There are four broad categories of surveys globally.

The first is a standardized, multi-country health interview survey. These focus on health outcomes, risk factors for poor health, and the type and level of access to health care systems. They often include exams by a medical professional. The best known is the Demographic and Health Surveys (DHS) program. When I moved into global health as a career, every time I heard someone say “DHS,” I assumed they were talking about the state Department of Health Services. Started in 1984, the “real” DHS includes more than 300 surveys in 74 low- and middle-income countries. The surveys are collected by ICF International, which receives the funding from USAID. The long-term plan for DHS is to encourage countries to start their own surveys, and, once they do, DHS focuses its expertise elsewhere.

The second is a standardized multi-country survey that includes some questions about health. A good example of this is the Living Standards Measurement Surveys (LSMS) from the World Bank. The program isn’t solely focused on health. Like the bank’s work in general, the idea is to help countries develop economically, but the surveys include questions that allow researchers to measure the use of health care resources and the amount of money spent on health care.

The third is a national health interview survey, which, like the DHS survey, can come with a physical exam. This can provide incredible data on blood pressure, cholesterol, blood sugar, eyesight, and other physical conditions. Much of what we know in the United States about these and other health factors comes from three national surveys. They’re so important that I’m going to give you a highlight or two form each one.

  • National Health Interview Survey (NHIS). This one has been collected since 1957 by the National Center for Health Statistics at the U.S. Centers for Disease Control and Prevention, which actually uses the U.S. Census Bureau to collect the data. Because of its long history, it is a valuable source of comparative data on health.
  • Behavioral Risk Factor Surveillance System (BRFSS). This is a telephone survey conducted by CDC. It is one of the largest surveys in the world, conducting more than 350,000 interviews every year (more than 10% of the population.) It aims to be representative in terms of race, income, education, and location. Survey takers collect data on health risk behaviors, preventive health practices, and health care access, with a strong focus on chronic disease and injury.
  • National Health and Nutrition Examination Survey (NHANES). This one is smaller than BRFSS, but it is also run by CDC. It covers about 15 counties every year and 5,000 people. A lot of what we know about people’s dietary habits and weight comes from NHANES.

And then there are disease- or condition-specific surveys. The World Mental Health Survey, for example, is used in 26 countries around the world to capture well-established and emerging mental health conditions.

Sounds like a lot of great information, right?

Surveys have limitations, too. They can be even less representative than a census, because they take into account far fewer people.

The results can be affected by the gender, race, or other characteristics of the survey taker. For example, a woman may be more likely to discuss specific reproductive health questions more openly with a female survey taker.

If the gaps between surveys are large, it can be difficult to measure trends over time. A dataset for one year is great, but if you can’t track which way a health problem is headed, it makes it difficult to find solutions or target appropriate policies.

And, when interviewing someone, it can be hard to determine which came first, the risk factor or disease. Some connections are obvious, like tobacco and lung cancer, but so many others are hard to tease apart.

This is why public health and global health has been relying more and more on highly sophisticated statistical tools that allow researchers to pull together vital statistics, censuses and surveys to create estimates for health trends over time and across populations. I’ll talk more about that in my next post.

Image by Playing Futures: Applied Nomadology via Flickr