Health promotion and disease prevention may be news as the nation plans a massive reform of its health care system, but nurse-midwives have long recognized these concepts as the basis of good health. For many of us, it is the focus on health and prevention that drew us to midwifery rather than to other more pathology-oriented disciplines. Even as we are necessarily drawn into dealing with the diagnosis and management of certain high-risk conditions, our focus rings loud and clear. Are these conditions preventable? If not, can we prevent adverse outcomes by early detection? If not, how can we best promote the health of mother, baby, and family within the context of their disease? In public health, these are the basics of prevention. Primary prevention of disease or outcome calls for altering susceptibility or reducing known risk factors; secondary prevention involves early detection and treatment; tertiary prevention is focused on the alleviation of disability and promotion of health within the disease state.
Much of primary care practice is aimed at primary and secondary prevention. To this end, clinicians have adopted an assortment of health risk assessment methods for the purposes of screening for disease, educating patients, or stimulating behavior change. Risk factors are duly noted and summarized into an impression about a particular patient's potential for preterm birth, human immunodeficiency virus infection, or some other health problem. Once an individual is labeled “at risk,” a series of interventions, from education to advice to diagnostic tests to assorted therapies, is set in motion in the hope of preventing or ameliorating the expected adverse outcome. In a risk-benefit equation, such assessment is presumed to be beneficial, in the best interests of the patient, and without ill effects.
There are several important things to note about the risk factors that comprise assessment tools. First, risk factors are statistically associated with adverse outcomes, but this does not necessarily mean they cause that outcome. In some cases, they are presumed to be causal: the role cigarette smoking plays in lung cancer is an example. Many risk factors, however, are not causal; they are surrogate markers for the real but as yet unidentified causes of disease. Lack of a high school education is often associated with higher risk. Does anyone believe that knowledge of algebra or English literature improves one's health? Clearly there are other causes of disease, and the association of disease with poor education is because the true causes are also found more frequently among people who do not finish high school. Many established risk factors fall into this category.
Second, risk factors cannot always be eliminated in the hope of preventing adverse outcome. Some may be inherited, such as genetic susceptibility to sickle-cell disease. Many socio-demographic risk factors, such as age, race, and poverty, fall into this category. We may have the laudable goal of eliminating poverty and racism in society, but for the individual woman whose risk we are assessing, the end results are already part of her risk profile.
Finally, most risk factors, even if they are strongly associated with outcomes in populations, do not predict adverse outcomes very well for individuals. Suppose for example that a certain fact in the health histories of pregnant women is associated with a fourfold increase in the development of gestational diabetes. This makes for an impressive risk factor, but it does not mean that all women with the risk factor will develop gestational diabetes. It may mean that, on average, 12 of 100 women with this factor and only three of 100 women without it will develop the condition: a fourfold increase. But most women, even if they have the risk factor, will never develop diabetes. The combination of poor specificity (a test with many false-positives) and the low prevalence of most adverse outcomes creates a low predictive value for positive test results (1). In other words, even the presence of many risk factors is no guarantee that a bad outcome will occur. The predictive value of preterm birth risk assessment tools, for example, is relatively low: only 10% to 25% of women designated as high risk go on to deliver preterm (2–5).
When risk factors (and their limitations) are incorporated into formal risk assessment screening and intervention programs, another problem emerges. Preventive health care strategies assume that, given certain risk factors or high risk scores, interventions are necessary to prevent adverse outcome. But because some risk assessment tools have low predictive ability, these interventions may be applied to many women who are not really at risk for the outcome (those with false-positive risk assessments). The result: Our advice and other interventions appear to work quite well in preventing adverse outcomes. The reason: Many of the women who were recipients of the interventions were never destined to have the outcome in the first place. In preterm birth prevention programs, the interventions seemed to work because most of the women labeled at risk delivered healthy babies at term. But if 75% to 90% of women were erroneously labeled as at risk for preterm delivery, the interventions may have had little or no role in preventing any adverse outcome. The appearance of benefit, however, reinforces the perceived value of risk screening and timely intervention, and makes it difficult to question whether assessment and intervention programs are beneficial or not.
Are there any risks associated with risk assessment? Risk assessment assigns risk labels, which result in interventions, which then produce more interventions. How many women labeled “at risk for prematurity” fell victim to such a cascade of intervention in preterm birth prevention programs? There is ample evidence that those risk assessment strategies did not identify women destined for preterm birth with accuracy. Nonetheless, overassessed women were diagnosed, monitored, hospitalized, tocolyzed, given cerclages and/or weekly pelvic exams, and advised any number of other interventions, all without any discernible effect on preventing preterm birth (6–9).
Unnecessary medical interventions are problematic enough, but there is more to consider. What happens psychologically or emotionally when women are labeled “at risk”? Do they perceive themselves as “problems waiting to happen”? What does this do to self-image? Are there adverse outcomes that could result from feelings of anxiety or helplessness? What happens when someone at risk for prematurity cannot follow advice to rest in bed and then delivers preterm? Does she blame herself for her premature baby (although it is possible that nothing we currently know could have prevented the outcome)? What if she follows all of the advice and still delivers a preterm infant? Even intervention strategies that might produce benefits in large populations may fail to prevent an adverse outcome in an individual woman. Will she feel she has “failed”? Some risk factors describe who the woman is; if she is poor or uneducated, will she feel inadequate? If she does not give up a preventable risk factor, like smoking, is she guilty? Clinicians would never deliberately pronounce such judgments, but is it possible in our enthusiasm for prevention to communicate them, however inadvertently? When risk screening assigns labels, or if interventions fail, do our patients judge themselves?
In public health policy there is an important consideration when determining whether to recommend a presumably beneficial screening and intervention program for the population at large: Are there health risks associated with the screening or intervention? The majority of people targeted are not at risk for the outcome one is trying to prevent. Thus, it is important to know that there is no threat to the health of a large number of “no-risk” individuals that could undercut the benefit of preventing a bad outcome in a small group of those at true risk. For example, before recommending routine alpha-fetoprotein testing for all pregnant women, advisory committees consider whether the benefits of detecting a few babies with neural tube defects could be offset by any possible increased pregnancy loss due to amniocentesis or false-positive diagnoses in women not at risk. Risk assessment as a routine screening and intervention program, although widespread in health care, is rarely considered in the same light. We assume its benefit, and only its benefit.
There are questions that should be asked, however, when initiating formal risk assessment programs. How accurate are the screening tools? What is the true value of the resulting interventions (not just the “apparent” value when applied to many not really at risk)? Are there any risks for those with “false-positive” risk profiles? What about the additional diagnoses, tests, procedures, and other interventions that will automatically be set in motion once that label of risk is assigned? What if these are completely unnecessary? What if they create health risks of their own? What if the screening questions are perceived as an invasion of personal privacy or dignity? Are there psychological ill effects when labels are assigned? If these questions cannot be answered, then we must consider this. What are the ethics of assessing all women to identify hypothetical risk factors (that may not predict disease with accuracy) in order to prescribe interventions (which may be of dubious value and possible harm) in the hopes of preventing an outcome (that will never happen to most of those subjected to this process)?
Both clinical practitioners and society in general view the benefits of prenatal and other health care as the result of prevention activities and early detection/treatment of disease. Risk assessment thus defines quality care; not to assess risk might certainly be considered negligent. But enthusiasm for risk assessment needs to be tempered with reality. For every mother or baby whose health is apparently assured by perinatal risk identification and intervention, there may be others who undergo unnecessary hospitalizations, cesarean sections, false diagnoses of disease, unmeasured psychological impact, etc., that result from inaccurate screening. Despite good intentions, the truth is that we still do not know how to predict or prevent many adverse outcomes. Appreciating the limitations of risk assessment is not a license for “laissez-faire” health care. However, as health professionals we need to understand the limitations and work toward minimizing their impact. Enthusiasm for prevention should not be allowed to create new risks of unnecessary intervention, unrealistic expectation, or undeserved blame.