| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Clinical |
Department of Medicine & Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong, SAR China
Correspondence to: K.M. Chow, Department of Medicine & Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong, China. Chow_Kai_Ming{at}alumni.cuhk.net
 |
ABSTRACT |
|---|
 TOP ABSTRACT SUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Background: Relatively little is known of the
epidemiology and predictors of sudden death in peritoneal dialysis (PD)
populations. We aimed to identify the risk factors of sudden death among PD
subjects. Methods: To explore clinical correlates of sudden death
in PD patients, we conducted a population-based case-control study using data
from a single dialysis unit. Cases (n = 24) were defined as all PD
patients that met the criteria for sudden death during January 2003 through
December 2006. We also selected 48 control subjects that were selected from
the prevalent PD patient name list compiled in alphabetical order. Data on the
hemoglobin, potassium, and calcium levels, residual renal function, dialysis
adequacy, cardiovascular risks, comorbid conditions, concurrent use of
aspirin, beta-blockers, angiotensin-converting enzyme inhibitors, and
erythropoietin, electrocardiographic and echocardiographic findings were
extracted from case notes and computer records. Confounders were controlled by
logistic regression. Results: Over a period of 4 years, 24 PD patients (mean
age 61.4 ± 9.5 years, median duration of dialysis 3.1 years)
experienced sudden death. Univariate analyses showed that patients that died
suddenly were more likely to be male and to have diabetes mellitus, a history
of smoking, and a lower small solute clearance as measured by Kt/V. Cases of
sudden death were also more likely to have received blood transfusion within
the previous 1 year. There were no significant differences between patients
and controls for residual renal function, serum potassium levels, control of
blood pressure and mineral metabolism, or hemoglobin levels. Multivariate
regression analysis confirmed independent association between recent blood
transfusion and increased odds of sudden death [adjusted odds ratio (OR) 5.18,
95% confidence interval (CI) 1.44 – 18.6]. Two other factors
significantly associated with risk of sudden death were male gender (adjusted
OR 4.16, 95% CI 1.14 – 15.2) and diabetes mellitus (adjusted OR 5.33,
95% CI 1.53 – 18.6). Conclusion: This study shows that recent blood
transfusion is associated with an increased likelihood of sudden death in PD
patients. The mechanisms that underlie this observation are unclear.
KEY WORDS: Sudden death; mortality; cardiovascular disease; hemoglobin; transfusion.
Cardiovascular diseases are the leading cause of mortality in dialysis patients worldwide, with sudden death accounting for 10% – 40% of deaths from all causes (1,2). The high rate of sudden death in the dialysis population supports an urgent need to better characterize this common event and provide insights into potential methods of prevention. Yet to date, most studies of sudden death in dialysis populations have focused on hemodialysis subjects, with reference to the timing related to dialysis schedule, potassium level, and heart condition (1,3–5). While important, such studies do not provide information about factors that predict sudden death for peritoneal dialysis (PD) patients, who tend to have low rather than high serum potassium levels with daily dialysis (6). Indeed factors that predict sudden death may be intrinsically different from those for hemodialysis patients. Identifying these factors would not only fill a gap in the current literature but would also inform nephrologists of the best preventive strategies.
Accordingly, we investigated the clinical correlates of sudden death in a case-control study involving PD subjects. We sought to examine the potential factors associated with sudden death among PD patients.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patient data were abstracted from patient records and electronic databases. Patients that died during the study period were reviewed. Cause(s) of death for each patient were determined by reviewing the medical records, information provided by the attending physician, autopsy reports, and coroner's files. In case of death outside the hospital, family members were interviewed by telephone to ascertain the circumstances surrounding the death. Sudden death was defined as unexpected non-traumatic death occurring within 1 hour of the onset of symptoms and without any previous condition that would seem fatal (7). Deaths during sleep and un-witnessed deaths occurring at home were considered sudden deaths. For each PD subject that met the criteria of sudden death, 2 control subjects were extracted from the patient list matched for date of death and the alphabetical order of patients' unique family and given names. Each control was assigned an index date identical to the patient's date of sudden death during the study period.
All data were independently abstracted by two reviewers and disagreements were resolved by consensus. The medical data collected consisted of demographic details, comorbidity, and anthropometric data. Clinical and demographic data, including presence of diabetes mellitus, smoking status, and recent use of erythromycin or macrolides (within 90 days), regular medication including aspirin, beta-blocker, angiotensin-converting enzyme inhibitors, and recombinant human erythropoietin, were recorded. For analytical purposes, we dichotomized smoking status as "who have never smoked (<100 cigarettes in a lifetime)" versus "who ever smoked." Coronary artery disease was defined as a history of myocardial infarction, angina, positive stress test, or coronary artery bypass surgery. Peripheral artery disease was defined as a history of amputation, lower extremity vascular bypass surgery, or intermittent claudication. Cerebrovascular disease was defined as a history of stroke, transient ischemic attack, or carotid endarterectomy. Patients' most recent two-dimensional echocardiogram and standard 12-lead electrocardiography (ECG) were reviewed. We refer to the ECG performed when the patients were stable without active coronary ischemia. The QT interval was calculated as the mean QT interval obtained from 12 different leads on the ECG using Bazett's formula (QTc) (8). In addition, the hemoglobin, potassium, and adjusted calcium values were obtained from the three most recent consecutive clinic visits. Adequacy of PD was determined by measurement of Kt/V, which was performed at least yearly using a standard method. Protein nitrogen appearance rate was calculated by the modified Bergström formula and normalized by ideal body weight (9). Residual glomerular filtration rate, as inferred from the last dialysis adequacy assessment, which was performed at least yearly, was calculated as an average of 24-hour urinary urea and creatinine clearances by standard methods as described previously (10). Systolic and diastolic blood pressures were measured on every follow-up visit at 8-week intervals.
Statistical analysis was performed using SPSS software version 11.5 (SPSS Inc., Chicago, IL, USA). All data are expressed as mean ± SD for normally distributed data and median or range for skewed data. Data were compared by chi-square test, Fisher's exact test, Mann–Whitney test, or Student's t-test, as appropriate. We were particularly interested in current and past smoking history, gender, comorbid illnesses such as diabetes mellitus, QT interval prolongation, prescription of aspirin, and erythropoietin. Variables significant on univariate analysis were included in multivariate analysis using a conditional logistic regression. Multivariate models were constructed with the backward stepwise procedures for variables significant at the 10% level. We evaluated two separate regression models: regular use of aspirin was excluded from the first model because of a concern of bias by indication, but this variable was included in the second model. The estimations of both univariate and multivariate analysis resulted in odds ratios (OR) with their 95% confidence intervals (CI). All probabilities were two-sided and the level of significance was set at 0.05.
This study was initiated by the investigators; the pharmaceutical industry was not involved.
|
RESULTS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Table 1 shows the details for the 24 subjects with sudden death (cases) and 48 control subjects. The case and control subjects were similar in age, duration of dialysis, body size, residual renal function, pulse pressure, and serum potassium concentrations. We observed similar serum potassium levels in cases and controls (3.80 ± 0.59 vs 3.81 ± 0.46 mmol/L, p = 0.92). On the other hand, patients with sudden death were more likely than the control subjects to be male, to have a history of smoking, to have diabetes mellitus, and to receive regular aspirin. In particular, subjects with sudden death had a markedly increased prevalence of diabetes mellitus than did controls without sudden death (p = 0.005). Of the cases, 58% had received blood transfusion within 1 year, compared with only 31% of the control subjects (unadjusted OR 3.08, 95% CI 1.12 – 8.47). On average, patients with sudden death had been transfused 1.21 ± 0.72 units of red blood cells within 1 year, compared to 0.77 ± 1.04 units in those without sudden death (p = 0.068). Over 80% of the blood transfusions were prescribed for elective correction of anemia of end-stage renal disease rather than acute blood loss. In contrast, there was no significant difference in average hemoglobin concentration between patients and controls (9.2 ± 1.7 vs 9.0 ± 1.4 g/dL, p = 0.70). Solute clearance was lower in patients with sudden death than in those without sudden death (Kt/V 1.70 ± 0.34 vs 1.92 ± 0.41, p = 0.032) although the difference in their residual renal function did not reach statistical significance. Furthermore, no differences were noted in the frequency of echocardiographic findings of left ventricular hypertrophy and QT interval prolongation in patients with and without sudden death.
|
After adjusting for all other significant factors (Table 2), it was found that red blood cell transfusion within the previous 1 year remained significantly associated with increased odds of sudden death (adjusted OR 5.18, 95% CI 1.44 – 18.6). In the multivariate analysis, two other factors were significantly associated with risk of sudden death: male gender (adjusted OR 4.16, 95% CI 1.14 – 15.2) and diabetes mellitus (adjusted OR 5.33, 95% CI 1.53 – 18.6).
|
Further analyses (not prespecified) investigated the difference between transfused and non-transfused subjects. Frequencies of diabetes and duration of dialysis were similar between patients that were transfused within 1 year and those that were not. Patients that received red blood cell transfusion within 1 year were older (62.0 ± 12.0 vs 55.2 ± 10.6 years, p = 0.013) and had lower total Kt/V (1.70 ± 0.34 vs 1.94 ± 0.40, p = 0.013) and less residual renal function (0.91 ± 1.44 vs 2.07 ± 1.78 mL/min/1.73 m2, p = 0.006).
The association of aspirin prescription may have resulted in bias because patients on regular aspirin were more likely to be prescribed drugs for cardiovascular disease. Therefore, we initially excluded this variable from the multivariate model. However, even when we included it in the second model, it did not materially change the estimates for the relationship between most risk factors and sudden death (Table 2).
|
DISCUSSION |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The fivefold increased OR of diabetes mellitus for sudden death is consistent with data from previous studies among the general population (11) and the end-stage renal disease population (4). The observed association between diabetes and sudden death may result from a higher prevalence of autonomic neuropathy, coronary artery disease burden, and more pronounced atherosclerosis and arterial and aortic wall stiffness in diabetic subjects (12,13).
The finding of recent blood transfusion as an independent predictor of sudden death was relatively unexpected. The reasons for this interesting observation are unknown but possible explanations include confounding by the treatment-by-indication bias, a common source of selection bias (14). For example, "sicker" PD patients or those with less residual renal function are more likely to be given a blood transfusion. Conversely, dialysis patients that are placed on a renal transplant waiting list are more often offered recombinant human erythropoietin treatment instead of repeated blood transfusions. Given the important clinical concern with respect to adverse cardiovascular events among chronic kidney disease patients with higher target hemoglobin concentration, as shown in a recent meta-analysis (15), it is valid to pose the question of whether transfusion could affect mortality and possibly effect sudden death in our PD patients. It would seem, nevertheless, that association of blood transfusion with sudden death is not simply mediated by achievement of higher hematocrit or hemoglobin, and the comparable hemoglobin levels between the cases and controls in our study support this contention. That being said, the potential harmful effects of red blood cell transfusion on mortality merit further discussion. In fact, red blood cell transfusion has been previously shown to be harmful to patients with acute coronary artery syndrome and cardiac surgery (16–18). Testing the hypothesis that blood transfusion aggravates the burden of sudden death therefore requires a prospective study design with randomized intervention because observational data do not demonstrate causality.
Our univariate (but not our multivariate) analysis findings of the association between smoking history and increased risk of sudden death are consistent with previous reports that cigarette smoking and the number of cigarettes smoked per day are strongly related to the risk of sudden death in the general population (19,20). On the other hand, we did not find convincing evidence of an increased prevalence of hypokalemia among PD subjects with sudden death because there was no difference in average serum potassium levels between the cases and the controls. Noteworthy is the fact that potassium level was not obtained on the day of sudden death and we cannot exclude the possibility that sporadic hypokalemia triggered ventricular arrhythmia and sudden death.
Our case-control study has several limitations related both to observational studies in general and to this study in particular. First, the autopsy rates were too low in our sample to gain insights into mechanisms by which sudden death occurred. Second, we were unable to test the impact of ethnicity as data for the whole group were derived from Chinese subjects. A third limitation of our study was that our analyses were underpowered to address comprehensively the risk of medications, such as erythromycin, that can prolong cardiac re-polarization and thus cause torsades de pointes (21,22). The limitation secondary to the small number of sudden death events needs to be highlighted. It is therefore questionable whether the play of chance or real effect could explain our findings. Furthermore, we did not include information on a variety of behavioral risk factors, such as lack of physical activity and excessive consumption of saturated fats. Finally, we did not test whether therapy with beta-blockers and/or angiotensin-converting enzyme inhibitors was associated with improved survival after cardiac arrest, as previously shown in a large contemporary cohort of hemodialysis patients (23).
In summary, we demonstrated important relationships between variables associated with sudden death in PD patients. Although we emphasize above the causative factors that explain at least some of the findings, it should be borne in mind that this non-experimental and observational study merely indicates associations and not necessarily causal relationships. While acknowledging the limitations that result from the observational study design, the hypotheses that were generated from this study warrant more large-scale prospective trials. Until further information is available, this study suggests caution regarding routine blood transfusion in stable PD patients.
|
ACKNOWLEDGMENTS |
|---|
We thank Ms Shirley Sun Kiu Tsang for the clerical support.
Received 17 March 2008; accepted 13 May 2008.
|
REFERENCES |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. J. Bleyer CLUES AT THE SCENE OF THE CRIME: SUDDEN DEATH IN DIALYSIS PATIENTS Perit. Dial. Int., January 1, 2009; 29(1): 23 - 25. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
