Chapter 8. Cohort Studies

• Studies in which the investigator passively observes events without determining assignment of exposure are referred to as observational.
•  Cohort studies are observational studies in which the investigator determines the exposure status of subjects and then follows them for subsequent outcomes.
• A prospective cohort study is one in which the exposure and subsequent outcome status of each subject is determined after the onset of the investigation.
• A retrospective cohort study is one which utilizes historical information on exposure status and subsequent outcome.
• The risk ratio or relative risk is a measure of association between the exposure and disease and is defined as the risk among the exposed divided by the risk among the unexposed.
• A risk ratio > 1 implies that exposure increases the risk of disease, whereas a risk ratio < 1 implies that exposure decreases the risk of disease.
• The attributable risk percentage is a measure of the proportion of the total risk among exposed persons that is related to their exposure.

Vigorous efforts at resuscitation were initiated, including bag-and-mask ventilation with 100% oxygen and chest compressions, but the Apgar score at 1 minute of life was 1. The Apgar score (Table 8–1), an index of neonatal asphyxia, can range from 0 (very asphyxiated) to 10 (no asphyxia). Despite continuing resuscitation, the 5-minute Apgar score improved only to 2, with a heart rate of 110 beats/minute. The 10-minute Apgar score remained depressed at 3, and the neonate was transferred to the newborn intensive care unit. With aggressive medical management, the 3100-g neonate continued to improve without evidence of acute neurologic complications. He was discharged from the hospital on the twelfth day of life.

Perinatal asphyxia can be defined as fetal hypoxia during labor and delivery. Hypoxia in the perinatal period is believed to be a major cause of both perinatal deaths and impaired development and neurologic function among survivors. The causes of perinatal asphyxia are not completely understood, but a number of factors have been associated with hypoxia during labor and delivery, including preeclampsia or eclampsia, maternal hypotension, placental insufficiency, and prematurity. The passage of meconium and the presence of this material in the amniotic fluid indicate possible fetal distress.

The pathogenesis of perinatal asphyxia is presented in Figure 8–1. The development of severe metabolic acidosis in the fetus indicates a lack of oxygen. This is because tissues must resort to anaerobic glycolysis for energy production. The degree of hypoxia that a fetus can tolerate before cellular injury occurs is variable and depends on a variety of factors, including previous asphyxia during the pregnancy, metabolic needs versus metabolic reserves, and blood flow to vital organs.

In experimental studies on newborn primates, varying degrees of asphyxia have been induced. These studies have shown that for brain injury to occur, the fetus must be exposed to marked asphyxia for at least 25 minutes. These studies also indicate that this degree of asphyxia will probably lead to fetal death rather than survival with neurologic impairment. In general, research has shown that the immature nervous system is more resistant to hypoxic injury than the mature brain.

The parents of the newborn in the Patient Profile are understandably distraught by the unanticipated complications in their baby’s first hours of life. Naturally, they have questions about what the future will bring. Will their son develop normal mental capacity? Will he have physical disabilities? Questions such as these from concerned parents often can be answered from the pediatrician’s own clinical experience. Physicians who have seen only a few such patients, however, must consult the medical literature to respond to parental concerns.

In undertaking such a search, the pediatrician will find a variety of case reports concerning infants who have had severe perinatal asphyxia and have developed various acute and chronic medical problems. Some of these reports describe dismal outcomes, including death. However, there are also reports of instances of severe asphyxia during delivery in which the child later had normal neurologic development and excellent school performance. Thus the pediatrician may be uncertain about what to tell parents. Clearly, a broad spectrum of outcomes is possible. What the pediatrician in our example needs to find are studies that offer reliable evidence of the likelihood of each of the various possible outcomes.

The most definitive conclusions could be drawn from a clinical trial. This would be a study in which newborns are randomly assigned to groups with different levels of perinatal asphyxia and then followed with measurements of outcome, such as achievement of developmental milestones and school performance. This type of study has actually been performed on laboratory animals, but obviously such a study on human infants would be unethical. An investigator could not intentionally expose humans to potentially harmful conditions simply to learn about the effects on outcome.

Since intentional exposure of human newborns to asphyxia cannot be justified on ethical grounds, the investigator might resort to observing the outcomes of newborns who develop asphyxia under natural circumstances.

A large cohort study was conducted in the United States on the utility of Apgar scores as predictors of chronic neurologic disability. In this study, investigators evaluated 49,000 infants whose Apgar scores were recorded at 1 and 5 minutes of age. For those infants who did not achieve a score of 8 or higher at 5 minutes, Apgar scores were then recorded at 10, 15, and 20 minutes. All the children were then followed to the age of 7 years. The occurrence of seizures was determined through clinical observations in the newborn nursery; interval histories were recorded at 4, 8, 12, and 18 months of age and yearly thereafter. The presence of cerebral palsy was determined by physical examination at age 7 years. A psychological and developmental assessment was also performed at age 7. The design of this study is shown in Figure 8–3.

This study demonstrated that low Apgar scores are a risk factor for the development of cerebral palsy. However, 55% of the children with cerebral palsy at age 7 had Apgar scores of 7 or higher at 1 minute, and 73% scored 7 or higher at 5 minutes. Of the 99 children who survived and had Apgar scores of 0–3 at 10, 15, or 20 minutes, 12 were found to have cerebral palsy. Eleven of those 12 also had delayed mental development. Ten of those infants had seizures in the first 24 hours of life. Of the children who survived and had Apgar scores of 0–3 at 10 minutes or later, 80% were free of any major handicap at early school age.

This study of a large population of children provides the pediatrician in the Patient Profile with the kind of information needed to discuss the baby’s prognosis with his parents. The study represents a range of experience that no individual practitioner could compile, even in a lifetime of practice. The pediatrician can now advise the parents that although their baby does have an increased risk of cerebral palsy and developmental delay, such an outcome occurs in only about one of eight asphyxiated neonates. Because the baby did not have a seizure in the first 24 hours of life, the prognosis may be more favorable. It should be reassuring to the parents to learn that 80% of even the most severely asphyxiated newborns were free of major neurologic handicap at early school age.

Perhaps the contributions of cohort studies can be illustrated best by the Framingham Heart Study, one of the most widely recognized and most influential studies of this type. In that investigation, the status of the residents of Framingham, Massachusetts, was determined with respect to potential risk factors for cardiovascular disease. Beginning in 1950, a sample of 6500 individuals aged 30–59 years was chosen from a total population of approximately 10,000 people in that age group. Approximately 5100 subjects with no clinical evidence of atherosclerotic cardiovascular disease agreed to participate in the study. Each subject was examined at the beginning of the study and was reexamined every 2 years thereafter. For example, the investigators identified subjects who had elevated blood pressure, subjects who smoked, and subjects who had elevated serum cholesterol levels. The population was followed over 35 years to identify subjects who suffered a myocardial infarction, stroke, or other adverse cardiovascular event. This protocol allowed investigators to determine if an individual with hypertension, for example, was more likely to suffer a stroke than someone with normal blood pressure.

More than 250 research reports have been produced by the Framingham Heart Study investigators and their collaborators. The Framingham Heart Study is the source of much of our current knowledge about the risk factors for cardiovascular morbidity and mortality. This cohort has also been used to collect information regarding various other diseases.

Framingham was chosen as the site for the study for many reasons, but primarily because the community is stable and there is broad representation of occupations among the population. The Framingham study is limited, however, because its participants are mainly white, middle class individuals.

The advantage of the retrospective cohort design is that all the events under study have already occurred, and conclusions can therefore be drawn more rapidly. In addition, the cost of a retrospective cohort study might be substantially lower for the same reason. The retrospective approach may also be the only feasible way to study the effects of exposures that no longer occur, for example, discontinued medical treatments. On the other hand, a retrospective cohort study must usually rely on existing records or subject recall; both are usually less complete and accurate than data collected in a prospective study. Attributes of prospective and retrospective cohort studies are compared in Table 8–2.

It should now be clear that a prospective cohort study often takes a long time to complete. Furthermore, to ensure that there are enough subjects to reach a valid conclusion, a cohort study (either prospective or retrospective) usually requires a fairly large number of individuals with the potential to develop the outcome of interest. For example, in the study of perinatal asphyxia, data concerning 49,000 subjects were accumulated over a 7-year period. For this reason, such studies can be expensive to complete. In fact, if the outcome of interest under study is rare, the sample size required for a cohort study may be so large that it would be impractical to undertake such a study. On the other hand, if the exposure or risk factor is rare, a well-designed cohort study may offer superior statistical power to evaluate associations between exposure and subsequent development of disease, because this type of study allows selective inclusion of exposed persons. A cohort study of perinatal asphyxia, for example, could include all newborns with Apgar scores of 0–3, but only a sample of the larger pool of newborns with higher scores.

Following subjects over a long period of time can lead to various problems. Subjects may move away or leave the study for other reasons, including death from causes other than the disease under investigation. If the losses to follow-up are substantial, and the lost subjects differ in outcome from those who remain in the study, the validity of the results can be affected seriously. It is also possible for exposure status to change during the course of the study. Obviously, birth asphyxia occurs only once. In other circumstances, however, the exposure under study may be subject to variation over time. For example, a cigarette smoker may quit, or in an occupational cohort study, employees may change jobs, and therefore their level of exposure to an occupational hazard may change. Diagnostic methods used for the disease under study may also vary over time.

The advantages and disadvantages of cohort studies are presented in Table 8–3.

Selection of subjects for a cohort study is influenced by various factors, including (1) the type of exposure under investigation, (2) the frequency of the exposure in the population, and (3) the accessibility of subjects and the likelihood of their continuing participation. Both exposed and unexposed groups must be free of the outcome of interest at the start of the study, and they must be similarly eligible to develop the outcome of interest during the course of the study. If some subjects already have the outcome of interest at the onset of the study, the temporal relationship between exposure and outcome will be obscured.

The Exposed Group

The type of exposure under investigation is critical for selection of the exposed group. Some exposures during pregnancy are common, such as gestational diabetes or hypertension. For these exposures, a general population of pregnant women could be used to construct the cohort. Other exposures, such as in vitro fertilization, are not common. In a cohort study designed to evaluate whether in vitro fertilization is a risk factor for developmental disability in the offspring, it may be necessary to sample exposed subjects from an infertility clinic rather than from among pregnant women in the general population.

Feasibility issues are also important in selecting the exposed population. The investigator should identify an accessible population that is motivated to participate in the study and unlikely to discontinue participation. The availability of historical information such as medical records may also be a factor in selecting this group. Examples of groups that have been chosen for feasibility reasons include nurses, members of health maintenance organizations, residents of stable communities, and labor union members.

The degree of exposure may differ depending on the goals of the study. For some exposures, subjects are classified into one of two groups: exposed or unexposed. In vitro fertilization is an example of this type of dichotomization. Other studies involve a range of exposure levels, for example, Apgar scores as an indicator of perinatal asphyxia. The investigator may take a graded exposure variable, such as the Apgar score, and transform it into a categorical exposure by dividing subjects into those whose score exceeds a certain designated value (eg, Apgar score of 3) and those whose score falls below that value. The value chosen to separate groups is referred to as a cutoff point and can be selected in various ways. For example, the cutoff point might be selected on the basis of the underlying distribution of values, such as the point that separates the 10% of subjects with the lowest Apgar scores from the remainder of the population. Alternately, a standard cutoff point can be used that is believed to have pathophysiologic implications, regardless of the underlying distribution in the population.

Thus, Apgar scores can be classified in three different ways: dichotomous (0–3 versus > 3), multiple ordered (0–3, 4–6, 7–10), or continuous (by gradations). If the exposure can be categorized into multiple levels or gradations, the investigator can determine whether a relationship exists between the dose (of the exposure) and the response. In the present context, the study may seek to determine whether the risk of chronic neurologic disability rises as the Apgar score decreases. If such a trend is observed, the argument that perinatal asphyxia is a cause of chronic neurologic disability is strengthened.

The Unexposed Group

Feasibility issues for the unexposed group are similar to those for the exposed group. The unexposed group must be accessible for entry into the study and for follow-up. When the purpose of a cohort study is to investigate a community, such as in the Framingham Heart Study, that community is the source of the unexposed persons. Because there may be more unexposed people in the community than are needed for the investigation, a representative sample may be taken. In the Framingham Heart Study, several risk factors were of interest, all of which were relatively prevalent in the community. In this situation, a sample of the entire community was drawn and then subdivided into exposure groups, depending on the risk factor of interest in a particular analysis. In the study of perinatal asphyxia, the unexposed group was defined as the infants with the highest Apgar scores (7–10), indicating the lowest degree of perinatal asphyxia.

For cohort studies that involve the selection of a specific exposed population, selection of an appropriate comparison population may be less clear-cut. For example, in a study of in vitro fertilization as a risk factor for congenital malformations, the comparison group might be pregnant women who are followed in other obstetric practices. If, however, pregnant women in the comparison population are not followed with a comparable level of clinical scrutiny, or if the unexposed group of pregnant women differs from the exposed group of pregnant women in other ways that might be related to congenital malformations, the study may lead to a false conclusion. The investigator may relate an observed increased risk of congenital malformation to in vitro fertilization, when in fact the elevated risk is the result of other differences between the exposed and unexposed groups. This type of problem illustrates why a randomized controlled clinical trial may be less susceptible to error than a cohort study. With randomization, factors known to be related to the development of disease—as well as other factors not yet recognized as related to the disease—tend to be balanced between the groups. This increases confidence that an observed association is in fact the result of the exposure of interest rather than the result of some other characteristic.

The underlying principle in selecting the unexposed group is that it should yield a fair comparison with the exposed group. Occasionally, the frequency of outcome occurrence in the exposed population is compared with the occurrence in the general population. This is particularly useful when members of the general population are very unlikely to be exposed to the study factor. However, the general population may not be comparable with those in the exposed group. For example, follow-up for disease occurrence may be more (or less) complete than for the exposed study group. This may lead to erroneous conclusions. Furthermore, if the exposed and unexposed groups are chosen from different time periods (a nonconcurrent study), medical care or other factors may differ between the groups in a way that makes the comparison unfair and the results invalid. Suggestions for the selection of exposed and unexposed subjects are presented in Table 8–4.

The investigator must collect information on both the independent variable (exposure) and the dependent variable (response) during the course of a cohort study.

Exposure

It is essential to define the exposure clearly. Some exposures are acute, one-time episodes, never repeated in a subject’s lifetime (eg, asphyxia at birth). Other exposures are long term, such as cigarette smoking or the use of oral contraceptives. Exposures may also be intermittent, such as pregnancy-induced hypertension, which may occur during one pregnancy, disappear after delivery, and perhaps reappear during subsequent pregnancies. The types of exposure characteristics that should be considered are presented in Table 8–5.

A subject who originally satisfies the criteria for inclusion in a cohort study should not subsequently be excluded from the analysis because of a change in exposure status during follow-up. This type of exclusion may lead to a biased conclusion. Specifically, it is possible that a change in exposure status may indicate a change in outcome status. For example, in a study of the relationship between the use of an antinausea medication during pregnancy and subsequent risk of spontaneous abortion, the medication may be discontinued for a subject because of early signs of threatened abortion. Excluding this subject from the analysis, therefore, may result in an underestimate of the true link between use of the medication and risk of abortion. The potential for changes in exposure status has important implications for the frequency of follow-up. Frequent reassessment of exposure and outcome status may be required if exposure status changes over time.

The source of available information about exposure may constrain the ability of the investigator to define and measure exposure experience. If the information comes from medical records, as is sometimes necessary in a retrospective cohort study, the accuracy of exposure information may be poor. For example, there are inherent disadvantages in using Apgar scores from medical records. Sometimes the Apgar scoring system is recorded by the medical staff as part of the required paperwork after delivery, without careful timing of the observations and detailed assessment by multiple observers. The medical staff providing patient care may be distracted by other responsibilities. In the previously cited prospective cohort study of neonatal asphyxia, a specially trained independent observer who was not responsible for patient care recorded the score in a standard manner on a standardized study form at exactly 1, 5, and 10 minutes of life.

In general, objective measures of exposure or biological markers of exposure are preferred over subjective measures. For example, in a study of maternal use of illicit drugs and pregnancy outcome, one approach to exposure assessment would be to question pregnant women about their use of illicit drugs. Self-reports of illicit drug use, however, are likely to underrepresent actual exposure. Repeated measurements of drug metabolites in urine might provide a more accurate and reliable assessment of exposure.

Clinical Response

Before the start of the study, it is imperative to determine that subjects do not have the outcome (disease) under investigation. This may be particularly difficult if the outcome is a disease that develops slowly, has an insidious onset, and is asymptomatic until its late stages. One approach to this problem is to exclude cases in which the disease emerges early in the course of the investigation, under the assumption that the biological onset of disease preceded the beginning of the study.

The degree of surveillance for disease should be similar in the exposed and unexposed groups. The frequency of examination and the duration of follow-up depend on the type of exposure and the outcome under investigation. For some diseases, the time from exposure to development of disease is short. A cohort study of the relationship between exposure to perinatal asphyxia and death within the first week of life would have a short follow-up period. Other outcomes, such as chronic neurologic disability, may require years to assess. Because the investigators were interested in performance and cognitive ability at early school age, the study of the relationship of perinatal asphyxia to neurologic development required 7 years of follow-up.

Information on outcome status may come from various sources. Some cohort studies rely on information from the records of physicians and hospitals. This would be particularly pertinent for a cohort study focusing on a population with good access to health care and standardized recordkeeping practices, such as a prepaid health plan. Other cohort studies may combine physician records with periodic examinations by the investigators. The Framingham Heart Study is an example of this type of study. Another approach to collecting information on disease is to have the subjects report whether they develop the outcome of interest. The study may also involve reviews of medical records in a subset of subjects to confirm self-reports.

If the outcome under study is death from any cause, the investigator may use information from death certificates. Death certificates may have limited utility, however, if the study focuses on a specific disease, because cause-of-death information on death certificates may be inaccurate (see Chapter 4: Medical Surveillance). Obviously, in that circumstance the best information would come from autopsy reports. This approach may not be feasible, however, since most people who die are not autopsied.

If diagnostic evaluations are required by the investigator during the study, an appropriate diagnostic test for the disease must be available (see Chapter 6: Diagnostic Testing). This approach has limitations because diagnostic tests are not always available or feasible. To ensure a fair comparison between the exposed and unexposed groups, the accuracy and reliability of diagnosis must not differ between the groups. Thus it can be helpful if those who assess outcomes are unaware of the subjects’ exposure status. The study of perinatal asphyxia relied on standard neurologic examination and psychological tests. The examiners had no access to the medical records, and thus were blind to the Apgar scores of the children. This blinded approach should facilitate an assessment of neurologic outcomes that is comparable for exposed and unexposed children.

It is possible for exposure status to alter the surveillance for disease. An example of this problem could occur in a study of neonatal asphyxia and intellectual development. Physicians are more likely to administer psychological and developmental tests to an infant who had a difficult birth with low Apgar scores; that child is therefore more likely to be diagnosed as having subtle developmental problems than another child who has not been singled out for close surveillance. This could lead to an overestimate of the relationship between neonatal asphyxia and subsequent developmental disability. This problem, however, can be avoided, as in the cited study, by ensuring that a standard diagnostic protocol is followed, regardless of exposure status.

Several different approaches can be used to analyze the results of a cohort study, as described in the following sections.

Risk Ratio

The results of a cohort study can be summarized using the format shown in Table 8–6. In that table, the letters A–D represent numbers of subjects in the four possible combinations of exposure and outcome status (in this instance, death).

• A. Exposed persons who later die
• B. Unexposed persons who later die
• C. Exposed persons who do not die
• D. Unexposed persons who do not die

The total number of subjects in this study is the sum of A + B + C + D. The total number of exposed persons is A + C, and the total number of unexposed persons is B + D.

Among exposed persons, the risk (R) of death is defined as

As indicated in Chapter 2: Epidemiologic Measures, risk can vary between 0 (no exposed persons die) and 1 (all exposed persons die). As in all statements of risk, some time period for the development of the outcome must be specified. For example, the outcome might be the risk of death in the first year of life. Among unexposed persons, the risk of death is defined as

If the exposed and unexposed persons have the same risk of death, the RR is 1 (ie, the null value). That is, exposure is not related to the outcome.

The calculation of risk ratio can be illustrated from the study of perinatal asphyxia. The data in Table 8–7 relate to infants who weighed more than 2500 g at birth. Exposure is defined as an Apgar score of 0–3 at 10 minutes of life, and the comparison group of less exposed newborns had Apgar scores of 4–6 at 10 minutes. In the actual study, a third group with Apgar scores of 7–10 was included, but the data are not described here in detail.

The risk among exposed newborns is

That is, about one of three newborns weighing more than 2500 g and having very low Apgar scores at 10 minutes died during the first year of life. The risk among “less exposed” newborns is

In other words, one in eight neonates weighing over 2500 g and having intermediate Apgar scores at 10 minutes died during the first year of life.

Without any further calculations, it should be obvious that the neonates with very low 10-minute Apgar scores had a worse prognosis than those with intermediate 10-minute Apgar scores. Quantification of the magnitude of this effect is achieved by calculating the risk ratio

The RR of 2.8 means that newborns at this birth weight with very low 10-minute Apgar scores are almost three times more likely to die in the first year of life than similar-weighing newborns with intermediate 10-minute Apgar scores. The RR is a measure of the strength of association between exposure and outcome. The farther the RR is from the null value of 1, the stronger the association. The strength of association is an important criterion in evaluating whether an observed association is likely to represent a cause-and-effect relationship. The RR of 2.8 is consistent with a moderate-to-strong relationship between the exposure (10-minute Apgar score) and outcome (infant death).

As discussed in Chapter 7: Clinical Trials, a sense of the statistical precision of this estimated risk ratio can be obtained by calculating confidence intervals around the point estimate of 2.8. Using the approximation method described in Appendix B, the 95% confidence interval for the data presented in Table 8–7 is (1.9, 4.1). That is, at the 95% level of confidence, the range of RR values consistent with the observed data falls between 1.9 and 4.1. Thus the data are consistent with a risk of death in infants with very low Apgar scores between roughly a doubling and a fourfold increase (Figure 8–5). As demonstrated in Chapter 7: Clinical Trials, the point estimate does not lie in the middle of the RR confidence interval. The asymmetry of this interval derives from the skew of the range of values of the risk ratio toward the positive direction (ie, all beneficial effects are compressed into the range 0–1, whereas hazardous effects range from 1 to positive infinity).

Since the null value is excluded from this 95% confidence interval, it can be concluded that the findings are statistically significant. In other words, these data are not consistent with the null hypothesis of no association between Apgar scores and infant mortality (at the prespecified 95% level of confidence). An association as strong as that observed between Apgar scores and infant mortality, therefore, cannot be explained by chance alone.

As indicated earlier, the argument that the linkage between Apgar score and death in the first year of life is one of cause and effect is strengthened if a dose–response relationship can be demonstrated. A third group of newborns, with Apgar scores of 7–10, was therefore included in the study. Comparison of the risk of death in that group with the previous reference group, which had intermediate Apgar scores of 4–6, yields a risk ratio of 0.15, with an approximate 95% confidence interval of (0.11, 0.21). This result means that newborns with a 10-minute Apgar score of 7–10 have only about one-sixth the risk of death in the first year of life as newborns with Apgar scores of 4–6. This disparity is statistically significant, and the very narrow width of the confidence interval indicates a statistically precise estimate (because it is based on a large number of observations).

The dose–response relationship between Apgar score and the risk ratio of death in the first year of life for newborns weighing more than 2500 g is shown in Figure 8–6. The reference group against which others were compared in the preceding calculations was the group with Apgar scores of 4–6 (ie, the risk ratio for this group is defined as 1). A clear trend of decreasing risk ratio with increasing Apgar score is seen, and this trend is unlikely to have occurred by chance alone. Thus, there is strong evidence in these data for a dose–response relationship.

Attributable Risk Percent

The risk of a specified outcome can be compared with measures other than a ratio. For example, the risk for one group can be subtracted from the risk for another group. This measure is termed the risk difference, or excess risk. Some authors use the term “attributable risk” for this measure, but that expression is discouraged here because it may be confused with other expressions. The risk difference (RD) is defined as

Using the previously cited data relating 10-minute Apgar scores (0–3 versus 4–6) to the risk of death in the first year of life, we calculate the risk difference as

That is, the risk of death in the first year of life is increased by 0.219 for newborns who weigh more than 2500 g and have a 10-minute Apgar score of 0–3, compared with similar-weighing newborns with a 10-minute Apgar score of 4–6.

For the Apgar score–infant mortality data, the attributable risk percent is

In other words, almost two thirds of the total risk of infant mortality for newborns who weigh more than 2500 g and have 10-minute Apgar scores of 0–3 is related to an Apgar score below the 4–6 level. The attributable risk percent typically is used as an indicator of the public health impact of exposure. These data suggest that birth asphyxia is a major contributor to—but not the sole cause of—infant mortality among severely asphyxiated children.

Rate Ratio

The analyses presented thus far are based on comparisons of risk estimates across exposure groups. In a cohort study, the measured outcome may be an incidence (or mortality) rate rather than a risk. Rate data in a cohort study can be summarized using the format shown in Table 8–8. The rate ratio is derived as follows

The magnitude of the rate ratio is interpreted in the same manner as the risk ratio (< 1 = protective effect, 1 = no effect, > 1 = harmful effect of exposure). The farther away from the null value, the stronger the association between exposure and the rate of the outcome. The data collected in the study of perinatal asphyxia were not presented in a manner that allows calculation of rate ratios.

To illustrate this measure, data are drawn from the Chicago Heart Association Detection Project in Industry (Dyer et al, 1992). That investigation involved almost 40,000 men and women at 84 cooperating companies and institutions in the Chicago area. Subjects were enrolled between 1967 and 1973, screened for risk factors for cardiovascular disease, and then followed an average of 14–15 years. For white males aged 25–39 at entry, the relationship between baseline serum cholesterol level and subsequent rate of coronary heart disease (CHD) is shown in Table 8–9. The rate ratio is

In other words, the mortality rate from CHD among white males with borderline high cholesterol levels was about 3.5 times higher than that of white males with lower cholesterol levels. Adjustment for underlying age differences in the study groups reduced the observed rate ratio to 3.1. Comparison of mortality from CHD among white males aged 25–39 with high serum cholesterol levels (> 6.2 mmol/L [> 240 mg/dL]) with white males of the same age with normal serum cholesterol levels (< 5.1 mmol/L [< 197 mg/dL]) yielded an age-adjusted rate ratio of 5.1. Thus, a dose–response relationship was evident between baseline serum cholesterol level and subsequent CHD mortality.

In this chapter, the basic approach to the design and analysis of cohort studies is presented, with illustrations drawn primarily from the literature on birth asphyxia. A cohort study is a type of observational investigation in which subjects are classified on the basis of level of exposure to a risk factor and followed to determine subsequent disease outcome. Prospective cohort studies are conducted by making all observations on exposure and disease status after the onset of the investigation. Retrospective cohort studies involve observations on exposure and disease status prior to the onset of the study. The retrospective approach offers several pragmatic advantages, but may result in less accurate and complete information on exposure and disease status.

Cohort studies are statistically efficient for the study of rare exposures because the exposed individuals can be selectively included in the study. On the other hand, cohort studies are inefficient for the investigation of slowly developing or rare diseases. The evaluation of chronic diseases through the cohort approach requires a long follow-up period and increases the chances that subjects will be lost from the study. The evaluation of rare diseases with the cohort study approach requires a large sample size and therefore is expensive and labor intensive.

There are several basic strategies to analyze cohort studies. If data are collected on the risk of developing an outcome during a specified period of time, the summary measure of effect typically is the risk ratio, or the risk of the outcome among exposed individuals divided by the risk of the outcome among unexposed individuals. An alternative approach to contrasting risks is the risk difference, which is the risk among exposed persons minus the risk among unexposed persons. If the risk difference is divided by the risk among exposed persons, a measure termed the attributable risk percent is derived. The attributable risk percent is an indicator of the proportion of risk that may be attributable to the exposure per se. When data in a cohort study are based on the rate of disease outcome, the standard measure of effect is the rate ratio.

The prospective cohort study of perinatal asphyxia cited in this chapter indicates that Apgar scores can serve as a useful predictor of subsequent risk of death and neurologic disability. An inverse dose–response relationship occurs between Apgar score level and the risk of adverse neurologic outcome. In spite of the increased risk, however, most children with low Apgar scores survive and do not manifest neurologic or developmental disability. Through proper interpretation of the results of this cohort study, the pediatrician in the Patient Profile can inform the baby’s parents that although their child faces an increased risk of certain disabilities, there is about an 80% chance that no neurologic handicaps will develop.