Epidemiology for Neonatologists

Introduction

Epidemiology is the study of health and disease and the associated causes. The role of epidemiology in neonatal-perinatal medicine is to define the prevalence and causes of illness in women and children by exploring risk factors and their associations. Specifically, maternal and infant mortality rates are used to assess the levels of health, access to needed health care, and quality of health care provided by a region or country.

Common Epidemiologic Concepts

In epidemiology, risk describes the probability of disease occurrence; however, it may represent a wide variety of statistical measures that include incidence, prevalence, rate, or odds. A risk factor often implies an increase in the outcome of interest with exposure, although risk measures may also represent protective effects. Risks are often compared using relative measures such as the risk ratio, odds ratio, and rate ratio ( Table 2.1 ). Relative measures are helpful in the identification of risk factors but can be misleading if not accompanied by absolute measures such as the risk difference. For example, a relative risk of 2.0 indicates that the exposed group is twice as likely to develop the disease as the unexposed group. However, twice as likely may represent an absolute change in risk from 0.2 (20%) to 0.1 (10%; risk difference of 0.1 or 10%) or may represent an absolute change in risk from 0.002 (0.2%) to 0.001 (0.1%; risk difference of 0.001 or 0.1%). A small change in absolute risk, therefore, may have little to no clinical impact.

In neonatology, the impact of single risk factors on outcomes is generally small, and caution should be used when inferring causality. Most diseases such as intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC), and bronchopulmonary dysplasia (BPD) have many component causes, some of which remain either unknown or unmeasurable. These component causes can occur at different times, and the sequence of their occurrence may be important in the development of the disease.

Health Statistics and Data Sources

A clear understanding of the definitions used to calculate the epidemiologic data for health statistics is essential. In the United States, population-level health data is primarily derived from birth and death certificates submitted to the National Vital Statistics System (NVSS) by individual states and territories. The standardized terminology and definitions allow direct comparison of important population-level health markers such as birth, death, and outcome rates ( Table 2.2 ). A standard set of reporting measures are reported to the NVSS; however, individual states may choose to collect additional data of importance to their specific population. In the United States, completion of a birth certificate form is required for all births regardless of length of gestation or weight and uses uniform definitions ( Table 2.3 ). Fetal death reporting requirements, however, vary by state and may be based on gestational age or birth weight criteria.

The ability to make inferences based on health statistics relies on the availability of large datasets such as those found through the NVSS. However, many of the important questions regarding survival and neurodevelopmental outcomes of the smallest and sickest infants in neonatology represent only a small total number nationwide. Single centers often do not have large enough populations of neonates to have the statistical power to determine clinically important differences in outcomes. To meet this need, several networks of neonatal and maternal centers have been established.

The National Institute of Child Health and Human Development (NICHD) Neonatal Research Network (NRN), established in 1986 to improve the medical care for newborns and their families, is a consortium of US neonatal intensive care units (NICUs) that conducts multicenter clinical trials and observational trials on infants. Among the ongoing observational trials, the Generic Database Study (GDB) has been collecting information on very low birth rate (VLBW) infants since 1987, including data on hospital admission as well as follow-up data that includes standardized neurodevelopmental assessments. In addition, the NRN designs and executes high-quality clinical trials that provide epidemiologic data on subset populations and specific neonatal diseases.

The Vermont Oxford Network (VON) is an international collaborative of more than 1000 NICU and medical centers established in 1988. The mission of VON is “to improve the quality and safety of medical care for newborn infants and their families through a coordinated program of research, education, and quality improvement projects.” VON maintains a database of inpatient hospital information on VLBW infants using standardized measures. In total, more than 2.2 million infants are represented in this database, providing a rich resource for information on the care and hospital outcomes of high-risk infants.

National and international quality collaboratives are filling an important role in neonatology by providing a platform to gather information on maternal child health as well as improving the care delivered to newborns. While the scope and direction of the quality initiatives vary, several perinatal state collaboratives, such as California Perinatal Quality Care Collaborative, the Ohio Perinatal Quality Collaborative, the Perinatal Quality Collaborative of North Carolina, and Tennessee Initiative for Perinatal Quality Care, have contributed valuable data on rates of late onset sepsis, nonindicated cesarean section, antenatal corticosteroid use, and narcotic abstinence syndrome.