The United States lags behind other countries in the development of diagnostic and medical record systems that allow public health specialists to track the incidence and severity of health problems, and importantly, identify risk factors contributing to adverse health outcomes.
Birth and death certificates contain key public health data on all Americans that serves as foundational information in epidemiology. Gaps in the data collected in these certificates, and difficulty in linking the information in birth and death certificates to other sources of medical information, has slowed the advance of science in tracking health trends and sorting out what is driving them.
The information routinely recorded in a birth certificate has changed over the years. In years past, birth certificates contained relatively extensive information on birth defects, but medical professionals and thedecided a decade or so ago to reduce the number of reportable birth defects.
Other steps have been taken in an effort to more systematically record and track birth defects that are not currently reported on birth certificates, including those that are not apparent for months or even years after birth.
Perhaps even more important, new tools are needed to track and record longer-term impacts of pesticide exposures on neurological development and function,, or susceptibility to disease. A variety of government funded and private efforts have established registries for reporting birth defects and developmental abnormalities, but timeliness, completeness, accuracy, and linkages with other health data remain constraints for those carrying out .
Birth certificates are official documents that mark the arrival of a new citizen of the US. Additional, valuable information should ideally be collected and recorded when an infant’s birth certificate is created. Six additional data points should be recorded, the last three of which are especially critical in assessing the impacts of herbicide and other chemical exposures on birth outcomes:
- Any recognized health problems in the infant (e.g., respiratory distress) or mother (e.g. diabetic);
- Complications, if any, during delivery and actions taken;
- Days in a neonatal intensive care unit (Neonic), and reason for placement of a newborn in the Neonic unit;
- County of residence and/or zip code where the mother lived when the child was conceived, if different from the place where the infant was born;
- Estimated date of last menstrual cycle, in order to estimate length of gestation and identify pre-term deliveries; and
- Date of first ultrasound, and projected due date based on that ultrasound.
Information that is usually available in hospitals when a newborn dies should be added to the routine data recorded on death certificates. Death certificates tend to accurately record the immediate cause of death, but fail to add much information regarding why the infant succumbed to, for example, respiratory failure, an infection, or sudden death syndrome.
In most cases, the infant’s medical record will have extensive information on the underlying problems leading to death, including actions taken to address observed problems, the mother’s health status, the family’s history of disease, and related problems that did not trigger death (e.g., low birth weight), but which may have contributed to the infant’s decline.
Key additional information that should ideally be recorded on a child’s death certificate include:
- Factors thought to contribute to, or trigger the cause of death;
- Whether umbilical cord or child blood sample was tested, scope of testing, and what was found;
- Steps taken, and their outcome, in trying to help the infant overcome its health problem; and
- Indications during prenatal care that the baby’s development was not normal.
Enhancing the Statistical Power of Epidemiological Studies
While animal studies are valuable in identifying and quantifying possible health risks in the human population, studies focusing on the impacts of chemical exposures on people are even more crucial. Such epidemiological studies typically identify populations thought to be exposed to varying degrees, and compare health outcomes in highly exposed groups to those people thought to have lower exposures.
Three big challenges confront all teams carrying out epidemiological studies. First, they need access to accurate information on the incidence of a given, adverse health outcome that is the focus of the study, like individuals diagnosed with non-Hodgkin’s lymphoma, or women giving birth to an infant with a specific birth defect.
Second, information on what scientists call “confounding variables” must be collected, and taken into account in statistical models. Such factors include several indicators of general health status (age, weight, BMI, smoking, chronic disease), exposures to other chemicals, or other risk factors known to be associated with the adverse health outcome under study.
Third, across the people participating in the study, scientists need accurate measures of exposure to the chemical or chemicals that are of concern. Estimated exposure levels will be used to break the study population into three to five low-exposure to high-exposure groups, so that the incidence of the adverse health outcome can be compared in the high- vs. low exposure groups.
As noted above, the most accurate indicators of relative pesticide exposure come from biomonitoring, i.e. actual measures of the level of chemicals in the blood, urine, or tissues of people in the study.
In studies focused on adverse birth and developmental outcomes, such biomonitoring data are sometimes collected over the course of pre-natal care, at birth, and as the infant matures. But in most studies of pesticide-human health outcomes, biomonitoring data of sufficient quality and scope are not available.
Also, given that many chronic diseases develop over long time periods, scientists would ideally need biomonitoring data on individuals on an annual, or at least regular basis over a lifetime. Such data is rarely, if ever available.
As an alternative, scientists have developed several approaches to estimate relative pesticide exposure levels based on data on the amount of pesticides sprayed in the vicinity of where a person lives, or spends significant time (e.g., in schools, the workplace). The underlying assumption is that people living in cities or suburbs, many miles from intensively farmed areas, will be less frequently exposed to pesticides than people living in or near areas subject to frequent applications. Dozens of exposure studies have, in fact, supported this assumption
Published work also often supports the conclusion that across the U.S. population, a major source of exposure to many types of pesticides is the diet (food and beverages, including drinking water). Accordingly, when epidemiologists carry out a pesticide study that compares a health outcome in people living in rural areas versus a city, they have to deal with the complication that both urban and rural people ingest some level of pesticides on a daily basis from their diet. This reality makes it less likely that studies will detect any incremental impact of pesticide exposures among rural people.
Inaccurate estimates of pesticide exposure over time erode the statistical power of epidemiological research. Imprecise exposure estimates make it more difficult to identify whether pesticides are contributing to adverse health outcomes, and if they are, by how much. More accurate exposure data specific to individual active ingredients are also needed to determine which pesticides are most likely responsible for observed, adverse health outcomes.
New tools and analytical systems are needed to enhance the accuracy of herbicide exposure estimates based on proximity to intensively farmed fields and publicly available data on herbicide use. One promising approach is to combine GPS-based spatial modeling with detailed, county-level databases on herbicide use per acre. By calculating the pounds of herbicides applied within concentric one-mile or five-mile circles around the residence of a person participating in a study, scientists could produce more accurate indicators of relative exposure levels, especially if coupled with at least some biomonitoring data.
The need for better exposure-related information is especially great at times when patterns of pesticide use are changing. Concern is heightened when pesticides linked to human health risks are being used more heavily, or soon will be, as clearly the case in the rural Midwest.