Guinan: I received my PhD from the University of Texas and MD from Johns Hopkins. I took an introductory course to statistics in graduate school. I later took statistics in an epidemiology training program here at [the Centers for Disease Control and Prevention] CDC called the Epidemic Intelligence Service.

Aczel: What special interest or focus brought you to the CDC?

Guinan: When I first graduated from college I worked in a university laboratory as a research technician, then I was employed as a flavor chemist for chewing gum by a large pharmaceutical company.

I came to the Centers for Disease Control and Prevention because I was interested in disease control and prevention from a population perspective rather than the individual perspective. In epidemiology, statistics are of paramount importance in tracking disease and injury in the community. Statistical data are used to determine health priorities in communities and to convince Congress, and state and local legislatures to invest in prevention and control of various health problems including AIDS.

Aczel: What type of research do you do at the CDC?

Guinan: I do epidemiologic research, which involves studying the risk factors and determinants of disease spread in the community for AIDS, which is caused by the human immunodeficiency virus.

Aczel: Could you talk about the current spread of AIDS, the population it is infecting, and the uncertainty of this epidemic.

Guinan: AIDS is a unique syndrome. HIV impairs the immune system and the manifestation of disease is very varied, from infections to tumors. AIDS is defined as one of these infections, tumors, or conditions, such as wasting syndrome. The time between HIV infection and disease manifestation averages 10 years, during which time the person may not know he/she is infected, but is infectious and can spread disease either sexually or by sharing injection drugs. So, a completely healthy appearing individual may be spreading infection. AIDS cases are indicators of long-term HIV infection. About 350,000 cases have been reported in the United States as of 1994. The proportion of cases in homosexual men has leveled off and the proportion of drug users is only rising slightly, but the proportion of cases transmitted during heterosexual intercourse is rapidly rising and of great concern. The heterosexual population is much greater than the populations of gay men or drug users, and the potential for spread in this population is well recognized by health authorities. However, political and religious concerns have kept public policy makers from confronting this issue.

Aczel: What caused the CDC to perceive that the AIDS virus was an epidemic?

Guinan: The CDC as part of its mission to prevent and control disease and injury often receives reports of unusual occurrences of disease. A group of physicians in San Francisco first contacted CDC when they noticed the occurrence of a rare pneumonia in previously healthy young gay men. The CDC has a weekly newsletter that it sends to public health workers, in which it reported these cases. Subsequent to that publication, CDC received hundreds of calls about similar cases occurring in New York, Florida, Georgia, Texas, and other states. An investigation was begun to determine the cause/causes of the unusual pattern of this pneumonia.

Aczel: So, now what are the issues you are working on?

Guinan: I am now focused on trying to emphasize prevention of transmission of HIV through encouraging healthy sexual behavior. Because we have no vaccine to prevent disease or no cure for HIV, we have to focus on behavior. This is very difficult and controversial. Many people want the message to be "just say no" to sex for adolescents and young people. I think that this is a good message, but not one that all young people will accept. CDC advocates the use of condoms in those persons who choose to have sex.

Aczel: How about the forecasting, how does the CDC gather the data for projections?

Guinan: CDC has a surveillance system for AIDS. States determine which diseases or conditions are reportable to health departments. In all states, AIDS is a reportable disease. We gather the data electronically from the state departments and analyze it for case numbers, geographic distribution, risk factors, and trends.

Aczel: Are there any unique statistical methods employed in tracking the AIDS virus? Or any data-gathering issues you had to confront with AIDS?

Guinan: What is unique about CDC’s statistics is that we employ strict surveillance case definitions. What is a case of AIDS is very difficult to determine, because it is not one disease. So we establish criteria that determine what we will count as a case. This means, for example, that many true cases may not be accepted in our counting system because the criteria have not been met. It is important to have a standard case definition because we must be counting the same thing in order to have accurate statistics. We have changed the case definition several times as we have learned more about HIV infection. This complicates the analysis of trends because we are counting different things. We have methods to determine trends using the old case definition and for analyzing how many of the increased annual cases are due to true incidence changes or to changes in the case definition.

Aczel: The press tends to report on "new AIDS cases"; is that reliable?

Guinan: The press is rarely reliable in reporting stories on AIDS.

Aczel: Back to the projections – with different spread rates for different populations, how do you forecast? Do you model each?

Guinan: We project the number of future cases of AIDS using a back calculation method which uses the slope of the curves of trends in annual cases for each transmission subpopulation. For example, the slope of the curve in gay men has flattened out while the slope of the heterosexually transmitted cases is rising acutely. The spread of the virus among various subpopulations has been estimated by doing surveys of samples of special populations. For example, injection drug users who come into treatment will be tested for HIV. The proportion of positives in various drug populations is different. For example, in New York up to 60 percent of some drug user populations are infected with HIV, but in California less than 10 percent of drug user populations tested are HIV positive. The spread of HIV into the heterosexual populations has occurred primarily through injection drug users. This has been determined by the surveillance data in which the risk factors for HIV are recorded. Having a sex partner who uses injection drugs is the most common reported risk factor for heterosexual cases.

The accuracy of projections varies with the different methodologies. What is the least well defined is the number of new cases of HIV infection occurring each year. We estimated in 1991 that approximately 1 million persons in the United States had been infected with HIV. This was estimated from a large group of HIV seroprevalence studies done on college campuses, in health clinics, in women delivering babies, and in various high risk populations, such as gay men and injection drug users.

Aczel: As an MD, what are your particular concerns or frustrations with forecasting for disease control?

Guinan: Most modelers of the AIDS epidemic do not have enough information on human behavior in order to make accurate projections. For example, we do not know what proportion of the U.S. population is engaging in unsafe sexual behavior. Any attempts to study sexual behavior are so controversial that they rarely get done properly. Public sentiment about sexual behavior studies is very polarized. Also, prevention efforts are very controversial such as promoting condom use among sexually active teens. Political and social issues have been the biggest barrier to trying to determine the extent of the HIV epidemic and the most effective measures to control it. Because of the discrimination against HIV-infected persons in housing, jobs, schools, and social settings, large groups of people at risk refuse to be tested. Unrealistic fears of HIV positive persons has complicated the approach to recognizing the infected, which is necessary in order to offer early treatment.

Aczel: Do you have any advice or suggestions to students, if this is their first, possibly only, statistics course?

Guinan: Students today should look at how the media use statistics, how politicians use statistics, and how we have become a nation "hooked on numbers." Statistics can be used to lure you into investing money, buying one particular item over another, or even where to go to college. Everyone should have a healthy skepticism about statistics if they are presented in order to convince you to buy something, or to vote a certain way. One should always ask, "What is it that they are not telling me?" For automobiles, for example, one particular model may be a best seller but its repair record may be poor. You can bet that the auto salesman doesn’t give you the statistics on repair rates. Mark Twain once said there are lies, great lies, and statistics. The general public usually doesn’t understand the use of statistics. Even one of our past presidents of the United States was shown lacking in this area. President Eisenhower was very surprised to find out that half of the American population had an IQ below average. Do you think he should have been surprised?