|Home | About | Journals | Submit | Contact Us | Français|
Sickle cell disease (SCD) is associated with early mortality. We sought to determine the incidence, cause, and risk factors for death in an adult population of patients with SCD. All patients aged ≥18 years seen at the Adult Sickle Cell Center at Duke University Medical Center between January 2000 and April 2005 were enrolled. Forty-three patients (21 males and 22 females) died during the study period. Median age of survival was 39 years for females (95% CI 34–56), 40 years for males (95% CI 34–48), and 40 years overall (95% CI 35–48). Cardiac causes of death accounted for 25.6% (11/43 patients); pulmonary, 14.0% (6 patients); other SCD related, 32.6% (14 patients); unknown, 14.0% (6 patients); and others, 14.0% (6 patients). Pulseless electrical activity arrest, pulmonary emboli, multi-organ failure, and stroke were the most frequent causes of death. Among the deceased patients, the most common pre-morbid conditions were cardiopulmonary: ACS/pneumonia (58.1%), pHTN (41.9%), systemic hypertension (HTN) (25.6%), congestive heart failure (CHF) (25.6%), myocardial infarction (20.9%), and arrhythmias (14.0%). Tricuspid regurgitant jet velocity (TRv) was significantly higher (3.1 m/s vs. 2.6 m/s, p<0.001) and hemoglobin significantly lower (8.3 g/dL vs. 9.2 g/dL, p<0.05) in deceased patients as compared to patients who lived, respectively. With improved preventive and therapeutic advances, including hydroxyurea therapy, acute complications such as infection are no longer the leading cause of death; instead causes of death and pre-morbid conditions are shifting to chronic cardiopulmonary complications. Further, arrhythmia leading to premature death is under-recognized in SCD and warrants further investigation.
Since sickle cell disease (SCD) was first discovered in the early twentieth century and then attributed to a genetic mutation causing abnormal hemoglobin chemistry in 1949–1950, mortality has decreased substantially. In 1973, Diggs estimated a median survival of 14.3 years, with one third of the deaths occurring before age 5, half between ages 5 and 30 years, and one sixth occurring after age 30(1–3). By 1989, Leikin reported that the probability of patients with SCD who were at least 2 months of age surviving to age 20 years was approximately 85%(4). Powars and colleagues also reported that childhood survival to age 20 years improved from 79% for patients born before 1975 to 89% for children born in or after 1975(5). This improvement in mortality among pediatric patients was attributed to early parental education and counseling, early antibiotic administration for febrile episodes, penicillin prophylaxis, and widespread adoption of newborn screening programs for SCD(4, 6–8).
In 1994, prior to the widespread use of hydroxyurea, Platt et al studied patients with SCD who ranged from birth to 66 years of age(9). They reported a median age at death of 42 years for males and 48 years for females with sickle cell anemia, and 60 years for males and 68 years for females with hemoglobin (Hgb) SC disease. Further, they found that in patients with sickle cell anemia, renal failure, seizures, ACS, a low fetal Hgb level, and a baseline white blood cell count greater than 15,000 cells per cubic millimeter were associated with decreased survival. Hydroxyurea has been shown to decrease the frequency of painful crises, episodes of ACS, and the need for transfusions(10), and has also been shown to improve survival among adult patients with SCD(11). Pulmonary hypertension (pHTN), thought to be due in part to chronic hemolysis and is resistant to hydroxyurea therapy, may at least be partially responsible for a significant proportion of sudden and unexplained deaths(3, 12–17); asthma(18) and diastolic dysfunction(19) were additional risk factors that portend early death. Thus, new mortality data, particularly in the context of widespread hydroxyurea administration among patients with SCD throughout the United States, are needed. Further, as systemic HTN has recently been found to be associated with pulmonary HTN and renal failure(20), identification of potentially reversible comorbidities is also vital in order to decrease mortality in patients with SCD.
Deaths among patients with SCD are often unexpected and sudden(12, 21–25). In one series of 306 autopsies of patients with SCD, death was noted to be sudden and unexpected in 40.8%(24). And while the most common causes of death in adult patients with SCD have been reported to be ACS, infection, stroke, pHTN, and sickle cell-related lung injury(9, 11, 13–15, 23–29), the cause of death is commonly unknown(24, 27, 28, 30).
Despite evidence that the risk of serious arrhythmias is significantly increased during vasoocclusive crises(31), the frequency of sudden cardiac death in patients with SCD has been scantily studied. Since sudden cardiac death has been prevented in other high-risk patient populations with implantable cardioverter defibrillator therapy(32, 33), the frequency of sudden cardiac deaths in patients with SCD should be determined, in order to assess the possible need for a similar approach. Therefore, we sought to describe the causes of death in our adult sickle cell population at Duke University Medical Center (DUMC) over a 5 year period. We report the median age of survival and the associated comorbidities of the patients who died, and also attempt to identify risk factors associated with premature death.
Of the 240 adult patients with SCD who were followed at DUMC from January 2000 through April 2005, 43 patients died. There were an equal number of men and women when comparing the group of subjects who died to the subjects who lived (Table I). The majority of patients in each group had Hgb SS disease, followed by Hgb SC disease. The median age at death in the deceased patients was 39 years (range 21 to 83 years) compared to a median age at last follow-up of 34 years in the group of patients who are still living (range 19 to 85 years); p<0.03 (Table I). When compared by gender, the median age differed between the 2 studied groups, but only reached statistical significance for the female subjects (43 vs. 36 years for deceased and living subjects, respectively; p=0.02). Three patients lived to be greater than 70 years of age (2 patients in the living group, one patient age 85 years with Hgb SC disease and one patient age 73 years with Hgb SS disease, and 1 patient in the deceased group with Hgb Sβ+-thalassemia disease lived to be age 83 years). Eight of the patients were less than thirty years old when they died. The percentage of patients who were treated with hydroxyurea was not significantly different between the 2 groups studied (p= 0.06) (Table II).
The median age of survival, calculated using Cox regression, was 40 years (95% CI 35–48). There was no significant difference between males (median 40 years, 95% CI 34–48) and females (median 39 years, 95% CI 34–56).
Transthoracic echocardiograms were performed in 74% of the deceased and 64% of the living patients (Table II). The median TRv was 3.1 m/s in the patients who died, as compared to 2.6 m/s in the patients who lived, and pHTN was more prevalent in the deceased than in the living patients (56% vs. 26%, respectively, p < 0.001; (Table II, Figure 1). Further, deceased patients were significantly more anemic than the living patients (Hgb level 8.3 g/dL vs. 9.2 g/dL, respectively, p = <0.05). Ejection fractions were not statistically different between patients who died and patients who were still living, and the degree of fractional shortening did not differ between the two groups. The left ventricular size, percentage of patients with prolonged QTc intervals, white blood cell (WBC) count, and oxygen saturation were also not significantly different between the patients who died and the patients who lived (Table II). Ferritin levels measured while at steady state in 26 deceased patients, ranged from 69 to 7440 mcg/L. Fourteen patients had ferritin levels ≥ 1000 mcg/L, and eight had ferritin levels measuring ≥ 2000 mcg/L. Average ferritin levels were higher in patients with cardiac complications as compared to those patients without cardiac complications, though the difference was not statistically significant (2319 mcg/L vs. 1664 mcg/L, respectively, p= 0.454). Ferritin levels were not measured in five patients that have a history of cardiac complications.
As shown in Table III, cardiopulmonary causes accounted for 17 of 43 deaths (39.5%). The most common cause of death was cardiac in origin, pulseless electrical activity arrest. The other causes of death attributable to SCD were stroke, multi-organ failure, liver failure, renal failure, anoxic brain injury, and “sickle cell anemia”, totaling 14 (32.6%). Other causes of death included narcotic overdose, assault, complications after fall, foreign body aspiration, and intestinal disorder, totaling 6 (14.0%). The cause of death for six of the patients is unknown (14.0%).
Medical complications often overlapped in both studied groups, and cardiopulmonary complications were frequent. The most common cardiopulmonary manifestations included ACS/pneumonia, pHTN, systemic hypertension (HTN), CHF, and stroke (Table IV). ACS and pneumonia did not necessarily predict pHTN and vice versa (data not shown). Deceased subjects were more likely to have a history of pHTN (p=0.001) and CHF (p=0.003), but not ACS/pneumonia and stroke. Some patients also had a history of cardiac arrhythmias, such as atrial fibrillation, supraventricular tachycardia, and nonfatal ventricular fibrillation. Atrial fibrillation was significantly more prevalent in deceased subjects (p=0.004). The most frequent sickle cell-related complications included cholethiasis, proteinuria, avascular necrosis, and renal insufficiency. Proteinuria, renal insufficiency, and avascular necrosis occurred significantly more frequently in deceased subjects (Table IV).
Platt et al reported in 1994 a median age of death of 42 years in adult patients with SCD(9). In contrast to Platt’s study, where age of death was determined from birth, patients in our study had to have survived to the age of 18 years to be included. Therefore, the patients in Platt et al’s work and our study represent two different populations and our median survival age cannot truly be compared. More recently based on an analysis from birth, Powars reported a median age of survival of 36.3 years for female patients and 38.7 years for male patients with SCD(5). The median survival ages from our study agree fairly well with the results by Platt et al and Powars, if 95% confidence intervals are considered. Further, Steiner el al have reported that in-hospital deaths and death rate in United States hospitals remained stable for adult patients with SCD from 1998–2004 as compared to 1994–1997(34). A likely explanation is that the causes of death are shifting from acute to chronic sickle-related complications.
Our results, representing an institution that specializes in the care of adult patients with SCD, show that a large proportion of the deaths were not from classic acute sickle-related complications. The most common known causes of death in our patient population while related to SCD were cardiopulmonary in nature, totaling 17 of 43 patients (39.5%) and the most prevalent was pulseless electrical activity arrest. This is likely in part secondary to hypoxia, myocardial infarction, CHF, and pulmonary disorders,. The other known causes of death included single or multi-organ failure (6 patients) and stroke (4 patients). Our findings are compatible with Steinberg et al(11), who showed that 21 of their 75 deaths were due to pulmonary diseases, with SCD crisis and stroke each accounting for an additional 9 and 6 patients. Steiner and Miller(34) also reported that of 113,098 hospital stays in the United States during which SCD was noted, cardiac and respiratory conditions accounted for 15% of deaths. Powars and colleagues reported that of 232 patients that died, the most common cause of death was chronic lung disease with pulmonary HTN and cor pulmonale in 47 patients (20%), and 12 of those patients had evidence of myocardial ischemia and fibrosis. An additional 7 patients (3%) died from cardiovascular disease(5). These studies collectively suggest that in the post-hydroxyurea era, acute non-sickle-related complications leading to death may become the most prevalent.
Three of the deceased patients in our study had a history of myocardial infarction, suggesting that a small proportion of SCD patients will develop this lethal disease; however, its clinical manifestations appear to be different than in patients without SCD. Exercise testing has a low positive predictive value of myocardial ischemia(35). Sickle vasculopathy has been shown to have many parallels with atherosclerosis, including endothelial activation and dysfunction, platelet activation, in situ thrombosis, and disordered apolipoproteins(36). Still, patients with SCD who are found to have evidence of myocardial infarction on autopsy generally have no signs of coronary atherosclerotic disease(35, 37–40). This lack of atheroma formation may be due to the significantly decreased total cholesterol and low-density lipoprotein levels that occur in patients with SCD(36).
The cause of death for six of the patients remains unknown. In addition, the cause of death listed on death certificates for three patients was “sickle cell anemia.” In past clinical series, the cause of death was stated as “unknown” or as “sickle cell disease” in 7–75% of cases(5, 24, 27, 28, 30). This high proportion of ill-defined causes of death is thought to be secondary to the absence of uniformity in data collection and analyses among medical centers, lack of morphological evidence of some fatal physiological event, and insufficient interest, time, and/or resources for a more in-depth evaluation of the specific causes of death(24). From our experience, even when the exact circumstances surrounding death were detailed, the exact cause of death was frequently difficult to decipher.
The pattern of pre-morbid conditions of SCD seen in our deceased patients (Table IV) were similar to previous reports, with ACS/pneumonia, cholethiasis, pHTN, and avascular necrosis at the top of the list. However, chronic conditions such as proteinuria, renal insufficiency, systemic HTN, and CHF were also very common. Further, many cardiopulmonary complications such as pHTN, CHF, myocardial infarction, and atrial fibrillation, and chronic conditions such as sickle cell nephropathy occurred significantly more commonly in deceased as compared to living patients. These observations suggest that while hydroxyurea may reduce acute sickle-related complications, end-organ damage continues to be very prevalent in this patient population.
In our study, seven patients had a history of significant cardiac arrhythmias, including five patients with atrial fibrillation, one with supraventricular tachycardia, and one with a history of ventricular fibrillation arrest. Inadequate blood supply to areas with fibromuscular dysplasia, microthrombosis, hypoxemia, and rheological abnormalities(41) may lead to the development of fatal cardiac arrhythmias. The demonstration of both recent and old areas of fibrosis and degeneration near the cardiac conduction system suggests a chronic process. The risk of serious arrhythmia is increased during vasoocclusive crisis, as Maisel et al. demonstrated significant arrhythmias in 24 of 30 patients with continuous electrocardiographic monitoring during a vasoocclusive episode(31). Further, they detected atrial and ventricular arrhythmias in 60 and 67% of patients, respectively, via 12-lead and 24-hour ambulatory electrocardiograms during vasoocclusive crisis. Nine of the 30 patients had “complex arrhythmias,” including two patients with ventricular tachycardia. Early detection of serious electrical disturbance may decrease the incidence of sudden cardiac death in patients with SCD.
Sudden cardiac death has been treated successfully in high-risk populations with implanted cardioverter defibrillators (ICD). Initial studies using the ICD focused on patients with ischemic cardiomyopathy and history of acute myocardial infarction(32, 33). More recently, the sudden cardiac death in heart failure trial (SCD HeFT) has extended the use of ICD therapy to patients without previous myocardial infarction(42). These patients had heart failure due to left ventricular dysfunction and no other risk for sudden cardiac arrest. Nevertheless, ICD therapy was effective in preventing all cause mortality and sudden cardiac arrest. The key element in the success of ICD therapy has been the identification of a patient population at risk for sudden cardiac arrest and subsequent placement of an ICD. In our SCD patient population, usual risk factors for sudden cardiac death, including decreased ejection fraction and increased left ventricular size, were not significantly different between patients who were living and deceased. Further study is indicated to identify cardiac risk factors associated with increased mortality such that a select group of patients with SCD can be followed more closely and be exposed to more aggressive and invasive clinical interventions as deemed necessary if we hope to be successful in preventing premature death.
Increased anemia was found to be a risk factor for early death in our study. Sebastiani et al. have shown that an elevated lactate dehydrogenase level, due to hemolysis, contributed to an increased risk of death in patients with SCD(43). Baseline WBC and oxygen saturations were not significantly different between deceased and living patients. The presence or absence of α-thalassemia was not available for our study groups. Platt and colleagues previously reported that α-thalassemia status was not associated with mortality in patients with SCD. Fourteen of the deceased patients had evidence of iron overload, with ferritin levels ≥ 1000 mcg/L, as a result of chronic transfusion therapy. Although our patients with history of cardiac complications had higher ferritin levels, this difference was not statistically significant. As ferritin levels do not correlate well with hepatic iron concentration and may not be a reliable marker of iron stores in patients with SCD(44, 45), our patients with cardiac complications may indeed have higher iron stores, contributing to the observed cardiac complications and premature mortality in our patients. In contrast to studies performed in patients with thalassemia(46, 47), severe iron overload has not been found to be associated with cardiac arrhythmias and CHF in patients with SCD. However, studies have shown that chronically transfused patients with SCD have increased mortality as compared to non-transfused patients with SCD(48, 49), though the contribution of iron overload to this increased mortality is difficult to decipher. Further, the ferritin level has been found to be a univariate predictor of death in patients with SCD(50).
In conclusion, despite advances in medical therapy, patients with SCD continue to die prematurely. Our patients died most commonly from cardiopulmonary disease and multi-organ failure, shifting from the previously reported infectious and sickle cell-related causes of death. As acute complications of SCD are ameliorated by hydroxyurea, end-organ damage continues to accumulate. Further studies are indicated to determine which patients should be aggressively treated to prevent early fatal cardiopulmonary complications.
All patients aged 18 years and older who were followed at the Adult Sickle Cell Center at DUMC between January 2000 and April 2005 were enrolled in the study. The Institutional Review Board issued a consent waiver to allow the review of their DUMC medical records.
Clinical, laboratory, and radiologic information, including type of SCD, age at death, cause of death, place of death, comorbidities, medications, complete blood count, ferritin, creatinine, urinalysis, oxygen saturation, electrocardiograms, and echocardiograms were reviewed. Variables, including patient age, baseline Hgb and WBC, left ventricular function, left ventricular size, QTc, and TRv, were compared between the patients who had died and those who were living. pHTN is defined as a TRv of at least 2.5 meters per second.
Cause of death was obtained from medical records at DUMC. If a patient died outside of DUMC, death certificate information was obtained from the medical examiner when available.
Statistical analyses, including evaluation of means, standard deviation, Fisher’s exact t-test and the chi-squared test, were performed. P-value <0.05 was considered statistically significant. The Cox regression model was used to analyze age of survival. The age at entry (January 2000) was used for left censoring and the age at last follow up for right censoring. Median survival ages and their confidence intervals were calculated for each group from the survival functions.
We would like to thank Matt M. Hsieh, M.D. and John F. Tisdale, M.D. for their help in preparing the manuscript. Also, we would like to thank Mary Abrams, MPH for her support on data collection and editorial review of this manuscript.
This research was supported by grants R01HL68959, R01HL79915 and U54HL070769, and the intramural research program of the National Institute of Diabetes, Digestive, and Kidney Diseases and the National Heart, Lung, and Blood Institute of the National Institutes of Health. Dr. Laura De Castro was also the Sickle Cell Scholar of the Duke-UNC Comprehensive Sickle Cell Center (U54HL070769).