Now We Are Sick: Labeling and Hypertension

One of the unchallenged dogmas of medicine is that prevention is better than cure. It follows from this that early screening for disease to detect it before it has caused irreversible damage is desirable. In the case of hypertension, there is currently great interest in prehypertension, and the possibility that early treatment may alter the course of the disease, as discussed earlier in this series.1 Corporations have been encouraged to adopt screening programs so that individuals with previously undiagnosed hypertension can be diagnosed and treated, to the benefit of both the patient and the employer. Things do not always turn out the way that they are planned, however, and a well recognized phenomenon is the “law of unintended consequences.” The consequences of screening for hypertension have been the subject of many studies, some of which have suggested that screening a population for hypertension may cause unintended harm. The landmark study of this phenomenon was performed by Haynes et al.,2 who screened Canadian steelworkers and followed 208 subjects for 1 year who were found to be hypertensive. In comparison with the previous year, absenteeism increased by about 80% in the year after the screening program, mostly from self-reported illness. In the year before the screening, the subjects who knew that they were hypertensive had twice as much absenteeism as those who were unaware of their hypertension, but almost all of the increase in absenteeism in the year after screening occurred in the newly diagnosed patients. These changes could not be explained either by changes in prescribed medication or blood pressure (BP), but were interpreted as being the direct consequence of individuals being labeled as hypertensive, and adopting a sick role. Another analysis of the steelworkers found that those who had recently been labeled as hypertensive showed a deterioration of marital and home life in the ensuing year3 which was independent of treatment and absenteeism. Somewhat similar results were obtained in a screening program conducted for the Massachusetts Mutual Life Insurance Company,4 where the rate of absenteeism was also lower in hypertensives who were unaware of their condition before the screening than in those already diagnosed, but again showed a greater increase in the year after diagnosis. Two cross-sectional studies had similar results. The first was a screening of 52,948 people in the Kaiser Permanente health care system which measured subjects' BP and conducted psychological testing, and also asked whether they had been diagnosed with hypertension.5 The subjects were divided into four groups: labeled hypertensives, unaware hypertensives, mislabeled normotensives (those who thought they were hypertensive, but with normal BP), and normotensives. Measures of psychological distress were higher in the labeled hypertensives and the mislabeled normotensives than in the other two groups. The second study was part of the National Health and Nutrition Examination Survey (HANES I).6 Well-being was greatest in the normotensive and unaware hypertensives, and lowest in the labeled and treated hypertensives. Another study, which did not include unaware hypertensives,7 found that both newly diagnosed (within 6 months of the interview) and previously diagnosed hypertensive patients complained of more worry about their health, more symptoms, and a lower perception of their general health than normotensives. These symptoms were independent of their BP or its treatment. An interesting study was reported by Bloom and Monterossa,8 who during a community screening program identified 71 subjects who had previously been told they were hypertensive, but who had normal Bps without ever having been treated. In comparison with the other normotensive individuals, they had a worse perception of their health, and more symptoms of depression. Some other reports did not confirm these findings, however. An example is a study by Rudd et al.,9 who screened 5888 electronics workers in a three-stage process. Subjects who were found to be hypertensive at their first visit were randomized into two groups, one of which received a “traditional” debriefing which told them of the risks related to hypertension and the need for treatment. The other got a “reassurance” debriefing, which emphasized the variability of BP, and made an attempt to reduce the subjects' fears. The rates of absenteeism in the following year were no different, however. They also analyzed the data according to whether the hypertensive individuals were previously diagnosed or not, and found no difference in the absenteeism rates. The British Medical Research Council trial10 of treatment of moderate hypertension found no adverse effects of labeling 1 year after the diagnosis had been made, which may be explained by the fact that the intervention was run in special clinics run by nurses. Both of these studies incorporated more intensive interactions between the patients and their health care providers than is often the case in everyday practice. The “traditional” debriefing in the Rudd study lasted 20 minutes, and was conducted by health educators. Thus, the adverse effects of labeling are avoidable. One of the most important studies in this field was conducted as a population survey in northern Spain11 in which 466 randomly selected subjects were recruited. Researchers performed a physical exam and administered a set of questionnaires, which included the SF-36, one of the most commonly used instruments for evaluating the quality of life. Those who had a history of diagnosed hypertension, or who were found to be hypertensive at the exam, wore a 24-hour monitor. There were 69 patients (15% of the total) with previously diagnosed hypertension, and they reported significantly worse scores than the normotensives on four of the eight SF-36 subscales: physical functioning, general health, vitality, and mental health. Another 70 subjects had elevated BP (>140/90 mm Hg) on the physical exam, of whom half were diagnosed as white coat hypertension after the 24-hour recording. The subjects with previously unknown hypertension did not show any worse scores on the SF-36 than the normotensive subjects. Absenteeism is not the only negative aspect of the labeling phenomenon. A survey of households in Harlem, NY, found that symptoms of depression were associated with self-report of having been diagnosed with hypertension, but not with the actual elevation of the BP.12 And a study of 200 consecutive cases of severe hypertension published in 1953,13 before effective antihypertensive treatment was available, found that the best predictor of the occurrence of headaches was not the severity of the hypertension, but whether or not the patient had already been diagnosed as hypertensive. For many years, there has been debate as to whether there is a “hypertensive personality,” but this issue remains unsettled. The concept originated with Alexander,14 who proposed that the hypertensive individual experiences repressed hostility, or “anger-in,” which is channeled into the autonomic nervous system, resulting in an increased BP. Although a large number of studies have reported correlations between personality and BP,15 no clear picture has emerged. A potential problem with such studies is that characteristics such as anger and anxiety, which are frequently associated with hypertension, may be a consequence of receiving the diagnosis, rather than being etiologic factors—and there is always the problem of knowing what constitutes an appropriate control group. A good example is a study by Irvine et al.,16 who gave personality questionnaires to 162 subjects who were found to have persistently elevated Bps over a period of 5 months. Half of the subjects knew they had hypertension, while the others did not. The aware hypertensives scored significantly higher than both the unaware hypertensives and normotensives on neuroticism, anxiety, and self-reported “type A” behavior. The authors concluded that these differences were attributable to the consequences of labeling. There are, in principle, three explanations for the findings of impaired quality of life in hypertensive patients: 1) a direct effect of the BP itself; 2) side effects of treatment; or 3) the consequences of labeling. The effects of treatment can be discounted, because several studies have shown that treatment with antihypertensive drugs can actually improve the quality of life. This was shown in the Treatment of Mild Hypertension Study (TOMHS),17 which compared the effects of five different drugs and placebo, and found improvement in all the groups, but the greatest changes were in the β blocker- and diuretic-treated groups. Another good example of this is the Hypertension Optimal Treatment (HOT) trial,18 where patients were randomized to three treatment groups with different diastolic BP goals (85–90, 80–85, and <80 mm Hg). The main finding was that the lower the BP, the greater the improvement in the quality-of-life scores. Not surprisingly, there were some symptoms that did get worse, such as the sex life score, cough, and swollen ankles, all of which could be explained as side effects from the more intensive drug treatment. It should be pointed out that studies such as these recruited only patients with previously diagnosed hypertension, so they cannot address the issue of whether the process of diagnosing a patient with hypertension affects quality of life. The fact that intensifying treatment improves quality of life, however, suggests that it is unlikely to be responsible for the adverse effects of being diagnosed with hypertension. As discussed above, the height of the BP does not appear to be an important factor in this regard. One of the best examples of the impact of labeling comes from a series of papers by Rostrup and his colleagues,19–21 who studied 19-year-old men who had been evaluated for the military draft in Norway. This procedure included two measurements of resting BP taken with an automated device. None of the men were told at the time what their BP was, or that they had any health problem. The experimenters identified 32 out of 3861 recruits whose Bps fell above the 98th percentile.19 One year later, these subjects were sent a letter asking them to return for an additional study. The subjects were divided into two groups. Subjects in the first group were sent a neutral letter telling them that they should return in 2 weeks for further evaluation, and subjects in the second group were told that they had high BP and also that they should come back in 2 weeks. At their initial evaluation, the two groups had similar Bps (156/96 and 159/95 mm Hg), heart rates, and body mass indices. At the second exam, BP was measured repeatedly over a 45-minute period of testing using the same automatic device. Although both groups showed somewhat lower BP during the second testing than at the initial screening, there was a much greater decrease in the uninformed group. Thus the informed group had a persistently higher systolic and diastolic BP throughout the session, and the heart rate was also higher than in the uninformed group. The informed group also showed an exaggerated BP response to a cold pressor test. Plasma norepinephrine and epinephrine were similar at rest in the two groups, but plasma epinephrine increased more in the informed group during the cold pressor test. In a subsequent study conducted on another set of recruits,20 the design was very similar, except that BP was measured intra-arterially on the return visit. In this instance, the resting BP was not significantly different between the groups, although the diastolic BP was initially higher in the informed group. The informed group showed a bigger anticipatory increase of both BP and heart rate when they were told that they would undergo a cold pressor test, and the same thing was seen before a mental arithmetic task. Both plasma and platelet catecholamines were higher in the informed group while at rest, and there was a greater increase of epinephrine during the challenge in the informed group. Thus, these studies demonstrated very clearly that the act of labeling someone as hypertensive may cause a subsequent increase in BP when they next enter a medical setting, and that this is most probably mediated by increased sympathetic activity. We do not know from these studies whether the BP in the informed group was affected in other settings, for example, when they were at home, nor how long the phenomenon would last. We suspect that the effects may be most marked in the medical setting, because in a study of the white coat effect, we have found that the white coat effect is strongly related to anxiety at the time of the BP assessment, but not to general measures of anxiety.22 This discussion may be dismissed by some as being of only academic interest, on the grounds that the critical task for practicing physicians is to focus on controlling the major risk factors such as hypertension and smoking, and if we have to scare our patients to get them to comply with our treatments, so be it. In actual fact, there is extensive evidence that self-rated ill health can have a major influence on patients' prognoses. Reviews of 46 community health studies concluded that 40 of them found a positive association between self-rated health and mortality.23,24 This type of evaluation is very simple: patients are asked whether they view their health in general as good or bad. A case-control study of stroke patients in Sweden found that self-rated ill health contributed to 20% of strokes, and had an impact similar to that of hypertension and smoking,25 particularly in men. Although the exact mechanisms by which these factors operate remain unclear, there are important implications. When we diagnose and treat patients with hypertension, we should emphasize the positive aspects of care. Our message to the public should not be that hypertension is the “silent killer,” but that it is a risk factor that can be readily controlled by an ongoing partnership between patients and their health care providers.