The life expectancy of patients with sickle cell disease has improved considerably since 1960, when Sir John Dacie described sickle cell disease as “essentially a disease of childhood.” “Indeed,” he wrote, “the mortality is high and relatively few patients reach adult life, even when the standard of medical care is high”1. In his 1973 review based on autopsies, Diggs estimated a median survival of 14.3 years, with 20 percent of the deaths occurring in the first 2 years of life, one third occurring before the fifth year of life, half between 5 and 30 years of age, and one sixth after the age of 302. In contrast, in the recent Cooperative Study of Sickle Cell Disease (CSSCD), approximately 85 percent of children and adolescents with sickle cell anemia (homozygous for sickle hemoglobin) and 95 percent of patients with sickle cell-hemoglobin C disease (heterozygous for hemoglobin S and C) survived to 20 years of age3. Among the patients less than 20 years of age in that series, mortality peaked between 1 and 3 years of age; the primary cause of death was infection — predominantly Streptococcus pneumoniae sepsis. A low level of fetal hemoglobin, a low level of total hemoglobin, and an elevated base-line white-cell count were associated with an increased risk of death.
In most studies of mortality among adult patients, the age at death has been determined either from autopsy data or from clinical records. The clinical events leading to death have been recorded only sporadically. In the Jamaican studies by Thomas et al., the acute chest syndrome was the principal cause of death after 10 years of age,4 with chronic organ damage such as renal or cardiac failure, cerebrovascular accident, and complications of pregnancy noted as concomitant causes of death. Powars and colleagues5 described similar findings in a cohort of patients followed in Los Angeles for 25 years, noting that survival estimates based on this group may have been influenced by changes in clinical practice over time. For example, they suggest that the high mortality they observed among women with sickle cell anemia reflected the high mortality among pregnant women before 1975. Accurate knowledge is needed about the contemporary mortality rate, the events that occur shortly before death, and the clinical conditions that are associated with an increased risk of death. Such information is of great importance in counseling patients, in anticipating dangerous clinical situations, and in designing targeted therapy and clinical trials. We undertook a prospective analysis of deaths among patients in the CSSCD, with emphasis on the risk factors for death in patients over 20 years of age.
The design of the CSSCD is described elsewhere6. The CSSCD is a prospective study of the clinical course of sickle cell disease in which more than 3764 patients were enrolled from birth to 66 years of age, at 23 clinical centers throughout the continental United States
7. Starting late in 1984, α-globin gene mapping was performed in patients with sickle cell anemia by Dr. Stephen H. Embury at the University of California, San Francisco8,9. Consent was obtained from all patients, their parents, or their legal guardians.). Patients were enrolled at these centers between September 1978 and 1988 and were seen at regular intervals for laboratory evaluation and physical examination. At enrollment, the genotype was confirmed by review of the hematologic data and by quantitative hemoglobin electrophoresis performed at the Centers for Disease Control and Prevention
Risk Factors for Early Death
Although males were at greater risk for early death than females, when sex was factored into the analysis of each of the other risk factors, no significant interaction was found (data not shown). Thus, there was no evidence that the risk factors for early death differed according to sex.
The patients with more symptomatic disease were at higher risk of early death. Mortality varied with the following acute and chronic conditions: renal insufficiency, acute painful episodes, acute anemic episodes, seizures, and the acute chest syndrome
The risk of early death was inversely associated with level of fetal hemoglobin
Figure 1). Differences within the group with values above the 75th percentile could not be determined because of the small number of deaths.). This association was dramatically illustrated by the enhanced survival of patients with sickle cell anemia who had fetal hemoglobin levels above the 75th percentile (
Two other laboratory values were found to be associated with mortality (Table 4). In each case, the increase in mortality was associated only with values at the end of the distribution; there was no smooth trend of mortality increasing with the laboratory value. Patients with sickle cell anemia who had hemoglobin values below the 10th percentile ( ≤ 7.1 g per deciliter) had a slightly higher risk of death than all other patients (2.8 vs. 1.1 deaths per 100 person-years). Similarly, patients with sickle cell anemia who had an elevated white-cell count (15,100 per cubic millimeter) had a slightly higher risk of death (2.2 vs. 1.2 per 100 person-years). In each case, if the 10 percent of patients in each of these groups was excluded from analysis, no significant associations were found between mortality and the hemoglobin level or the white-cell count.
There was no demonstrable relation between mortality and the presence or absence of α-thalassemia. A binary indicator for α-thalassemia did not make a statistically significant contribution to the proportional-hazards model. The delay in collecting data on α-thalassemia interfered with our ability to perform a more rigorous analysis of this variable.
Multiple Regression Analysis
Since the incidence of clinical conditions may vary with laboratory profile, and since two or more conditions can occur in the same patient, there may be redundancy in the results of the risk-factor modeling by single regression analysis. Therefore, the laboratory values and clinical events associated with significant variation in the risk of death were combined in a multivariate model, and backward elimination was used to reduce the model to the subset of factors that made statistically significant contributions to the variation in mortality (Table 5). This model showed that the fetal hemoglobin level, renal insufficiency, acute chest syndrome, seizures, and white-cell count were significant risk factors.
This report presents a snapshot of the life expectancy of patients with sickle cell disease in the United States during the 1980s. In contrast to the widely held assumption that patients with sickle cell anemia rarely survive to adulthood, the median age at death among such patients was 42 years for males and 48 years for females. This represents a decrease of roughly 25 to 30 years in life expectancy, as compared with that of the black American population in general. Among patients with sickle cell-hemoglobin C disease, the median age at death was 60 years for males and 68 years for females. The longer survival of females with both types of hemoglobinopathy is typical of the findings in black American and other normal populations.
The pattern of mortality varies with age. We have already reported that the peak incidence of death among children with sickle cell anemia occurred between 1 and 3 years of age and that deaths among patients less than 20 years of age were predominantly caused by pneumococcal sepsis3. We anticipate that the recent acceptance of universal screening of newborns and the early administration of prophylactic penicillin to infants with sickle cell anemia will mean that more than 85 percent of such children will survive to 20 years of age and that overall survival will be even greater than we report here. Unfortunately, the risk of death among those over 20 years of age is not easily assigned to a single preventable cause.
In this study, adults with sickle cell anemia had a high mortality rate, with few surviving into their 60s. The circumstances of their deaths were quite varied. Eighteen percent of the deaths occurred in chronically ill patients with clinically obvious organ-system failure (renal failure, congestive heart failure, or chronic debilitating cerebrovascular accident). Although a minority of these patients had other specific factors that contributed to organ failure (such as systemic lupus erythematosus, renal tuberculosis, and congenital valvular heart disease), sickling was undoubtedly an important accelerator of organ destruction. Thirty-three percent of deaths occurred in relatively healthy patients who did not have overt chronic organ failure but died during a classic sickle crisis. Seventy-eight percent of these patients died during an acute painful episode or an episode of acute chest syndrome. These patients had surely experienced and survived such episodes when they were younger, but they were not able to tolerate them in later years.
The precise causes of death under these circumstances remain elusive, probably involving bone marrow and fat embolization,13-16 excessive narcotization,17 the pain event itself,18 or a combination of these factors. Adults, even those who appear relatively fit, are susceptible to cardiovascular collapse and acute multiorgan-system failure and may die suddenly during episodes of acute pain, the chest syndrome, or both. This finding illustrates the need for more thorough study of acutely ill patients; perhaps some such unexpected deaths can be prevented. The other 22 percent of these “healthy” patients with sickle cell anemia died of acute stroke; all the strokes for which a type was specified were acute hemorrhages.
The clinically well patients were not all as well as they seemed. As Powars and colleagues have described, vascular damage caused by sickling insidiously accumulates in varying degrees5. Our ability to prevent early deaths hinges on the ability to identify patients at high risk so they can be treated before fatal, often silent, vascular lesions develop. We examined clinical and laboratory data to determine whether we could identify adults at high risk for early death. We studied steady-state laboratory values using only those obtained at the time of routine visits rather than during acute illness. We studied steady-state clinical status by excluding events that occurred during the last year of life; we hoped thus to avoid biasing the results with changes associated directly with the cause of death. Our previous reports on this population described a wide range of degrees of clinical severity19. In this analysis, we found that the adult patients who had more symptoms of the disease were at increased risk of early death.
The important laboratory risk factors were a high white-cell count and a low fetal hemoglobin level. Patients with a white-cell count over 15,000 per cubic millimeter were at increased risk. This interesting finding is also well documented in the normal population and is independent of factors, such as smoking, that are known to raise the base-line white-cell count20. In this study we may simply have “rediscovered” this normal enigmatic observation, or may have identified a feature peculiar to sickle cell anemia, such as the relation between inflammatory cytokines and the endothelium, adhesive interactions between white cells and endothelium, or even interactions between white cells and sickle red cells. It is tempting to speculate that lowering very high white-cell counts may be therapeutic in sickle cell disease and that the beneficial effects of hydroxyurea and other cytotoxic chemotherapeutic agents that raise the fetal hemoglobin level may relate to their tendency to lower the white-cell count.
The most straightforward laboratory risk factor was the fetal hemoglobin level. Patients with high levels had an improved life expectancy. Although this result is no surprise, its documentation is of utmost importance. Of all the risk factors we identified in this study, the fetal hemoglobin level is the one known to be relatively stable throughout life21. Adults who had low levels of fetal hemoglobin as children are likely to die earlier than those who had high levels. The fetal hemoglobin level should be considered an important factor in selecting children for high-risk interventions such as bone marrow transplantation and long-term treatment with agents that stimulate the production of fetal hemoglobin.