OUP user menu

Haemoglobin-related mortality in patients undergoing percutaneous coronary interventions

Holger Reinecke, Torsten Trey, Jürgen Wellmann, Jan Heidrich, Manfred Fobker, Thomas Wichter, Michael Walter, Günter Breithardt, Roland M. Schaefer
DOI: http://dx.doi.org/10.1016/j.ehj.2003.09.008 2142-2150 First published online: 1 December 2003


Aims It has recently been proposed that anaemia is an independent risk factor for development of cardiovascular disease in the general population. The impact of anaemia on long-term survival of patients with manifest coronary heart disease (CHD) has not been assessed so far. In this study, we examined the influence of haemoglobin concentrations on the outcome after percutaneous coronary interventions (PCI).

Methods and results In a retrospective cohort study, we analysed in-hospital and long-term mortality in all male patients admitted to our institution for elective PCI from 1998 to 1999. In 689 cases, complete follow-up information could be obtained (98.4%). Depending on their baseline haemoglobin, patients were divided in quintiles. In all subgroups, angiographic success after PCI (90–94%) was comparably high and in-hospital mortality was low (0–0.7%). During follow-up (median 697 days), patients in the lowest haemoglobin quintile (≤12.9g/dl) were significantly more likely to suffer from all-cause death (22.2%) than those of the other quintiles (3.7–12.1%; estimated mortality rates from Kaplan–Meier models, P<0.0001, log rank test). In more detail, we found a U-shaped relationship between mortality and haemoglobin strata in steps of 1g/dl (P<0.0001, log rank test). After adjustment for potential co-variates, patients of the lowest haemoglobin quintile showed in Cox regression analysis a markedly higher risk for death (adjusted hazard rate ratio (HRR) 4.09, 95% confidence interval (CI) 1.52–11.05) compared to the quintile with a haemoglobin concentration of 14.6–15.2g/dl.

Conclusion These results indicate that anaemia is associated with markedly reduced survival in patients with CHD after elective PCI. Since PCI is a common intervention and anaemia is a frequent condition in the general population, strategies for the management of anaemic PCI patients and treatment of anaemic patients with CHD should be developed.

  • Anaemia
  • Angioplasty
  • Percutaneous coronaryintervention
  • Mortality

1 Introduction

Anaemia is a relatively frequent condition in the general population with many possible underlying causes such as iron, folate or B12 deficiency, malnutrition, cancer-induced blood losses, bone marrow depression, renal failure with decreased production of erythropoeitin and/or renal loss of erythropoeitin.1

A few recent studies provided evidence that chronic anaemia may also be an independent cardiovascular risk factor. In a prospective study in the general population with healthy subjects without coronary heart disease (CHD), anaemia was found to be associated with a significantly higher rate of cardiovascular events during the following 6 years.2It had previously been shown that anaemia is a risk factor in patients with kidney disease and in patients with heart failure, and that these patients may benefit from treatment with iron and erythropoeitin.3Finally, a recent study revealed the profound impact of anaemia on in-hospital mortality in CHD patients undergoing coronary bypass surgery.4

There are no data available about the long-term survival of patients with manifest CHD and anaemia. Therefore, we analysed the impact of pre-procedural haemoglobin concentrations on all-cause and cardiac mortality in patients undergoing percutaneous coronary interventions (PCI).

2 Methods

2.1 Patients

All patients patients who underwent PCI from 1 March 1998 to 31 December 1999 were identified from a computer database in which data of all patients are stored who undergo heart catheterization at our institution.5In this database, apart from information about the intervention, data about concomitant diseases were stored since 1 March 1998. During the period mentioned above, all patients admitted to our institution for elective PCI were included in the analysis if no previous treatment with thrombolytic substances or glycoprotein IIb/IIIa-inhibitors was performed which might have otherwise influenced baseline haemoglobin levels. Thus, a total of 901 patients could be identified (78% male, 22% female). Since haemoglobin levels in females are influenced by menstruation, the subsequent analyses were only performed for male patients. From 11 men, no baseline haemoglobin concentration could be retrieved, therefore these cases were excluded from analysis. Co-morbidities at presentation for PCI of the patients who died during follow-up were identified from the files.

2.2 Renal function

End-stage renal failure was assumed if the patient has ever required temporary or ongoing maintenance haemodialysis. Since end-stage renal failure is well known to have profound impact on haemoglobin concentrations, these patients were not included. Creatinine clearances for the other patients were calculated by the Cockroft–Gault formula as follows: Creatinine clearance [ml/min]=((140−age)×weight [kg])/(72×serumcreatinine [mg/dl]).

2.3 Cardiovascular risk factors

Cardiovascular risk factors were assessed at presentation for PCI and were defined as follows: history of smoking was defined if the patients has smoked within the last 10 years; hypertension if blood pressure >140/90mmHg has been documented; hyperlipidaemia if total cholesterol or triglycerides levels were higher than 200mg/dl, or levels of lipoprotein (a) higher than 20mg/dl; family history of cardiovascular disease if stroke, myocardial infarction (MI), or coronary intervention occurred in a first degree relative; diabetes was assumed if a patient was taking oral antidiabetic medication or insulin.

2.4 Haemoglobin

Haemoglobin values were retrieved from the computer database of the Institute for Clinical Chemistry. Only values which were determined 1 to 3 days before the procedure were considered for analyses.

2.5 Follow-up

A questionnaire asking for repeat interventions and adverse events was sent to all patients. If the patients did not return their questionnaire, a follow-up was performed by telephone calls with the patients, their relatives or referring physicians. This follow-up was performed from March to November 2001. From 50 patients, no information of repeat PCIs after index PCI could be retrieved. For patients who died during follow-up, the treating physicians were contacted to obtain information about causes of death.

2.6 Interventional procedures

Percutaneous coronary interventions were performed using arterial access from the femoral or brachial arteries. The coronary segment, the type of coronary stenosis dilated, and angiographic success were classified in accordance to the revised classification of the American Heart Association and the American College of Cardiology.6Thus, angiographic success after PCI was assumed if a residual stenosis in the vessel diameter of <50% could be achieved.6Left ventricular ejection fraction (EF) was assessed from the pre-PCI angiogram and determined from 30° right anterior oblique projections.

2.7 Statistics

Patients were divided in quintiles depending on their baseline haemoglobin before PCI. Since patients with identical haemoglobin level were assigned to the same group, the quintiles are slightly unbalanced. Differences in basic clinical characteristics between the quintiles were tested by ANOVA F-test for continuous variables, and overall chi-square test for dichotomous variables. The P-values for all of these tests are given in the tables. Correlation analyses were performed by two-sided Pearson test. Endpoint analyses were made by log rank tests or Cox regression analyses. Estimates of the mortality rates were taken from Kaplan–Meier models including 1200 days of follow-up since there were survival data and censored subjects in each subgroup for this time period. For a more detailed analysis of haemoglobin concentrations, patients were separated in strata of haemoglobin of ≤10.9g/dl, 11.0–11.9g/dl, 12.0–12.9g/dl, and so on in steps of 1g/dl until ≥18.0g/dl; 2-year all-cause mortality for these subgroups was compared by log rank test. Univariate predictors of mortality during follow-up were analysed by Cox regression, and calculation of hazard rate ratios (HRR) with 95% confidence intervals (95% confidence interval (CI)). Multivariate analysis was performed by Cox regression analyses for haemoglobin quintiles alone (crude HRR) and with potential covariates (adjusted HRR), taking quintile 4 as reference since it contains the largest proportion of patients within the reference/normal range of haemoglobin concentrations.7As covariates for adjustment, those parameters were chosen which were found to have a P-value lower than 0.1 in univariate analyses of death. Furthermore, family history of cardiovascular disease was added to the model, since it was significantly different between the quintiles. Alternatively, Cox regression models were calculated with both a linear term only, and a linear and quadratic term of haemoglobin together due to the U-shaped association of haemoglobin and mortality found in Fig. 1. These two models were compared by the likelihood ratio test.

Fig. 1

Cumulative 2-year mortality (in %) was plotted depending on the patient’s baseline haemoglobin concentration before PCI. At the top of each column, the number of patients in the group is noted. Differences in survival were compared by the log rank test (P<0.0001).

For all tests, P-values <0.05 were taken as significant. All statistical analyses were performed with SPSS 10.0 for Windows.

3 Results

Between 1 March 1998 and 31 December, 700 male patients without end-stage renal failure were scheduled for elective PCI. Due to missing haemoglobin values before PCI in 11 patients, 689 of them were included in the following analyses (98.4%). These patients were divided in quintiles depending on their baseline haemoglobin concentrations before PCI. Basic clinical parameters are given in Table 1. Significant differences were found between the quintiles regarding age, baseline haemoglobin concentrations, baseline serum creatinine and calculated creatinine clearance, and family history of cardiovascular disease.

View this table:
Table 1

Baseline clinical data and cardiovascular risk factors at presentation for PCI

Haemoglobin quintiles, g/dl
Patients in group, n (% of total)144 (20.9)133 (19.4)136 (19.7)134 (19.4)142 (20.6)P-value
Baseline haemoglobin, mean±SD, g/dl11.8±1.113.4±0.314.2±0.214.9±0.216.0±0.7
Age, mean±SD, years66±963±962±1060±1062±9<0.0001
Hypertension, n (%)54 (37.5)54 (40.6)56 (41.2)66 (49.3)71 (50.5)0.13
Hyperlipidaemia, n (%)64 (44.4)79 (59.4)76 (55.9)76 (56.7)79 (55.6)0.11
History of smoking, n (%)52 (36.1)36 (27.1)40 (29.4)51 (38.1)56 (39.4)0.12
Family history of cardiovascular disease, n (%)20 (13.9)26 (19.5)31 (22.8)37 (27.6)22 (15.5)0.029
Diabetes, n (%)27 (18.8)21 (15.8)13 (9.6)11 (8.2)21 (14.8)0.056
Previous myocardial infarction, n (%)37 (25.7)47 (35.3)49 (36.0)34 (25.4)41 (28.9)0.14
Previous PCI, n (%)28 (19.4)18(13.5)35 (25.7)21 (15.7)33 (23.2)0.065
Previous CABG, n (%)a22 (15.2)26 (19.5)29 (21.3)19 (14.1)23 (16.1)0.52
Serum creatinine level, mean±SD, mg/dl1.4±0.91.2±0.51.1±0.21.1±0.21.2±0.3<0.0001
Creatinine clearance, mean±SD, ml/min68±2580±2482±2385±2183±24<0.0001
  • a Missing data in 65 patients.CABG indicates coronary artery bypass grafting; PCI, percutaneous coronary intervention. The creatinine clearance was calculated by the Cockroft–Gault formula (not calculated in 98 patients due to missing body weight). Differences between the quintiles were tested by ANOVA for continuous variables, and the chi-square-test for dichotomous parameters.

3.1 Interventional data

Data of the coronary status and left ventricular function as well as procedural data of the PCI are displayed in Table 2. No statistically significant differences could be found between the quintiles regarding angiographic success rate after PCI and other parameters, especially the frequency of stenting.

View this table:
Table 2

Angiographic characteristics and PCI data

Haemoglobin quintiles, g/dl
Patients in group, n (% of total)144 (20.9)133 (19.4)136 (19.7)134 (19.4)142 (20.6)P-value
Patients with
1 vessel disease, %30.636.138.235.840.80.60
2 vessel disease, %38.231.632.439.632.4
3 vessel disease, %31.332.329.424.626.8
Left ventricular ejection fraction, mean±SD, %62±2068±1563±1765±1764±200.35
Ad hoc PCI (angioplasty at initial catheterization), n (%)76 (52.8)57 (42.9)59 (43.4)58 (43.3)57 (40.1)0.25
1 segment, %72.265.466.972.466.20.58
2 segments, %19.424.825.019.426.1
≥3 segments, %
AHA/ACC class of lesion6
low risk,%11.814.314.015.713.40.98
moderate risk, %66.064.762.559.765.5
high risk, %
Stent use, n (%)69 (47.9)64 (48.1)67 (49.3)65 (48.5)63 (44.4)0.94
Use of glycoprotein IIb/IIIa-inhibitors, n (%)11 (7.6)12 (9.0)9 (6.6)10 (7.5)11 (7.7)0.97
Angiographic success (residual stenosis <50%), %133 (92.4)122 (91.7)123 (90.4)124 (92.5)134 (94.4)0.81
  • a PCI indicates percutaneous coronary intervention. Differences between the quintiles were tested by ANOVA for continuous variables, and the chi-square-test for dichotomous parameters.

3.2 Results of follow-up

In-hospital mortality was low and was not significantly different between the various subgroups (Table 3). No non-fatal in-hospital myocardial infarction occurred.

View this table:
Table 3

Acute and long-term outcome, and major adverse coronary events after PCI

Haemoglobin quintiles, g/dl
Patients in group, n (% of total)689144 (20.9)133 (19.4)136 (19.7)134 (19.4)142 (20.6)P-value
In-hospital death, n (%)31 (0.6)01 (0.7)01 (0.7)0.079
Follow-up time, median [25/75 percentiles], days697 [488 864]674 [461 905]710 [490 880]692 [479 846]725 [500 858]692 [476 845]
Death during follow-up, n (estimated mortality, %)b6230 (22.2)7 (10.0)8 (6.2)5 (3.7)12 (12.1)<0.0001
Cardiac death, n (estimated mortality, %)b4419 (14.3)3 (6.1)8 (6.1)4 (3.0)10 (10.8)0.0016
Of these with documented fatal MI, n1262211a
Cancer, n532000a
Others, n641010a
Unknown, n741002a
Non-fatal myocardial infarction, n (%)264 (3.5)3 (2.4)3 (2.4)9 (7.0)7 (5.4)0.25
Repeat PCI, n (%)c21445 (36.9)50 (40.0)35 (27.3)42 (31.8)42 (31.8)0.24
CABG after index PCI, n (%)416 (4.2)8 (6.0)9 (6.6)9 (6.7)9 (6.3)0.89
PCI or CABG during follow-up, n (%)c23649 (34.0)53 (39.8)42 (30.9)47 (35.1)45 (31.7)0.56
  • a P-values not calculated due to small number of events.

  • b Estimated mortality rate including 1200 days of follow-up obtained from Kaplan–Meier models.

  • c From 50 patients no information could be retrieved about repeat PCIs. Differences between the quintiles concerning in-hospital mortality were tested by the chi-square-test. Event rates during follow-up were compared by log rank test. CABG indicates coronary artery bypass grafting; MI, myocardial infarction; PCI, percutaneous coronary intervention.

Both all-cause and cardiac mortality were significantly higher in the lowest quintile than in all others. Frequency of non-fatal MIs was not significantly different. Furthermore, frequencies of repeat PCIs and coronary artery bypass grafting during follow-up were also not significantly different between all subgroups.

A more detailed analysis of mortality depending on the baseline haemoglobin concentration was performed in steps of haemoglobin concentrations of 1g/dl, and yielded a U-shaped curve for 2-year all-cause mortality in these subgroups (Fig. 1, P<0.0001, log rank).

Information on cause of death could be retrieved in 55 of the 62 patients who had died (89%). Forty-four patients (71%) died from cardiac causes (MI, congestive heart failure or sudden cardiac death), 19 of whom (43%) were in the lowest haemoglobin quintile (P=0.0016, log rank). Of these cardiac deaths, fatal MI confirmed by electrocardiography and laboratory markers occurred in 12 patients. Five patients (8.1%) died of cancer (three in the lowest quintile), and six patients died of other causes (four in the lowest quintile). Deaths from other causes were caused by bleeding (n=2), ruptured aortic aneurysm (n=1), and sepsis (n=1) in the lowest haemoglobin quintile; by sarcoidosis (n=1) in the second lowest haemoglobin quintile; and by pneumonia (n=1) in the quintile with the second highest haemoglobin concentration.

The cumulative survival of all patients was analysed by Kaplan–Meier curves (Fig. 2). Patients with the lowest haemoglobin quintile were compared to the pooled other quintiles and showed a significantly reduced survival (P<0.0001, log rank).

Fig. 2

The cumulative survival (in %) was displayed for patients with a haemoglobin (Hb) level ≤12.9g/dl compared to those above 12.9g/dl by Kaplan–Meier curves. Numbers of patients still under observation in each group were displayed below the curve. Survival was found to be significantly different between both groups (P<0.0001, log rank). Nearly all deaths in the group with the lower haemoglobin concentration occurred within the first 600 days.

3.3 Co-morbidities of patients who died during follow-up

Relevant co-morbidities at presentation for PCI of patients who died during follow-up are given in Table 4, since these might have affected haemoglobin concentrations. No significant differences in the frequencies of the co-morbidities could be found between the quintiles apart from history of smoking, diabetes and renal failure. In more detail, endocrinological disorders included hyperthyroidism (n=2) and hypothyroidism (n=1). Chronic liver disease was found in two patients with active hepatitis C, one of them suffering from oesophageal varices. Chronic inflammatory disease was present by rheumatoid arthritis (n=2), sarcoidosis (n=1), Morbus Behcet (n=1) and HIV (n=1, stadium CDC C3). History of gastrointestinal bleedings included history of gastric ulcers in four patients some years ago with no actual signs of bleedings at presentation for PCI; one patient suffered fromongoing intermittent gastrointestinal blood losses due to jejunal angiodysplasia. Haematological disorders were present with thrombocytopenia without known cause (n=2), but no accompanying disturbances in white and red blood cell line.

View this table:
Table 4

Co-morbidities at presentation for PCI of patients who died during follow-upa

Haemoglobin quintiles, g/dl
Death during follow-up, n (% of all deaths)30 (48.4)7 (11.3)8 (12.9)5 (8.1)12 (19.3)P-value
Chronic obstructive lung disease, n (%)1 (3.3)01 (12.5)02 (16.7)0.42
History of smoking, n (%)20 (66.7)03 (37.5)3 (60.0)5 (41.7)0.023
Endocrinological disorders, n (%)2 (6.6)1 (14.2)0000.58
Chronic liver disease, n (%)2 (6.7)00000.70
Chronic inflammatory disease, n (%)4 (13.3)1 (14.2)0000.45
History of gastrointestinal bleedings, n (%)5 (16.7)00000.21
Hematological disorders, n (%)1 (3.3)1 (14.2)0000.47
Diabetes, n (%)4 (13.3)4 (57.1)01 (20.0)4 (33.3)0.044
Creatinine >1.3mg/dl, n (%)15 (50.0)6 (85.7)7 (87.5)3 (60.0%)11 (91.6)0.039
  • a Differences between the quintiles were tested by the chi-square-test.

3.4 Univariate and multivariate analyses of mortality

Univariate analyses of factors associated with mortality during follow-up are presented in Table 5. Thus, patients’ characteristics as age, multi-vessel disease, and smoking as well as creatinine and haemoglobin smoking were associated with a significantly higher mortality during death. Furthermore, patients who received implantation of coronary stents were significantly less likely to die during follow-up. In subgroup analyses, the benefit from stenting was comparable in the four lower quintiles. In the group with the highest haemoglobin concentrations, only one out of 63 patients who received a stent died (1.6%) versus 11 of the 79 patients without stent (13.9%, P=0.017, log rank).

View this table:
Table 5

Univariate predictors of death during follow-up analysed by Cox regression analysis

FactorCategoryNon-survivors(estimated mortality,%)aHRR (95% CI)P
Age (in years)1.05 (1.02–1.07)0.002
Multivessel diseaseYes49 (13.1)2.26 (1.22–4.16)0.009
No13 (7.7)
History of smokingYes31 (14.0)2.01 (1.22–3.31)0.006
No31 (9.6)
Diabetes, n (%)Yes13 (18.2)1.74 (0.94–3.20)0.077
No49 (10.0)
HypertensionYes22 (10.8)0.67 (0.41–1.15)0.15
No40 (11.1)
Family history of cardiovascular diseaseYes14 (13.8)1.19 (0.66–2.16)0.57
No48 (10.5)
Stent implantationYes20 (9.1)0.53 (0.31–0.91)0.021
No42 (13.2)
Serum creatinine level (in mg/dl)1.76 (1.40–2.21)<0.0001
Haemoglobin level (in g/dl)0.68 (0.59–0.78)<0.0001
  • a Estimated mortality rate including 1200 days of follow-up obtained from Kaplan–Meier models. Survival rates were analysed by unadjusted Cox regression models, with hazard rate ratios (HRR), 95% confidence intervals (CI) and P-values given. Age, creatinine and haemoglobin were entered as continuous variables, therefore there are no data in the first two columns.

Since haemoglobin concentration is well known to be correlated with renal function, mortality was also compared in the subgroup of patients with a normal creatinine level (≤1.3mg/dl). Also in this subgroup, all-cause mortality was higher in the patients of the lowest haemoglobin quintile (14.7%) compared to the other quintiles (5.1%, 5.7%, 2.5%, 8.6%, P=0.01, log rank). While haemoglobin concentrations were significantly correlated to serum creatinine in the whole cohort (r=−0.24, P<0.001, Pearson) and to the calculated creatinine clearance by the Cockroft–Gault formula (r=0.13, P=0.001, Pearson), they were not significantly correlated to creatinine (r=0.04, P=0.33, Pearson) or to creatinine clearance (r=0.06, P=0.14, Pearson) in the subgroup of patients with normal renal function.

A multivariate analysis was performed by Cox regression analysis. Thus, a crude HRR was calculated for the risk of death for each of the quintiles (Table 6). After adjustment for potential co-variates, a significant and markedly higher risk for death in patients of the lower quintile could be confirmed. In an alternative adjusted model, inclusion of a quadratic term of haemoglobin yielded a significant improvement in model fit (P=0.0059, likelihood ratio test) implicating a U-shaped (quadratic) relation between haemoglobin and mortality, as seen in Fig. 1.

View this table:
Table 6

Cox regression analysis with crude and adjusted hazard rate ratios (HRR) for death in the five hemoglobin quintiles

Patients,n (% of all)Deaths, n (estimated mortality,%)bCrude HRRAdjusted HRRa
Quintiles of haemoglobinHRR95% CIP-valueHRR95% CIP-value
1.Quintile (Hb ≤12.9g/dl)144(20.9)30(22.2)5.812.25-14.980.00002a4.091.52–11.050.008a
2.Quintile (Hb 13.0–13.8g/dl)133(19.4)7(10.0)1.390.44-4.371.290.40–4.16
3.Quintile (Hb 13.9–14.5g/dl)136(19.7)8(6.2)1.590.52-4.871.660.54–5.10
4.Quintile (Hb 14.6–15.2g/dl)134(19.4)5(3.7)11
5.Quintile (Hb ≥15.3g/dl)142(20.6)12(10.8)2.270.80-6.452.260.79–6.48
aCo-variates used for adjustment
Number of diseased vessels2.111.50–2.97
History of smoking2.761.64–4.66
Serum creatinine levels (in mg/dl)1.501.14–1.97
Stent implantation0.500.29–0.87
Age (in years)1.031.00–1.06
Family history of CVD1.680.89–3.16
  • a The given P-values refer to haemoglobin quintiles in the Cox regression model.

  • b Estimated mortality rate including 1200 days of follow-up obtained from a Kaplan–Meier model.CI indicates confidence interval; Hb, haemoglobin concentrations; HRR, hazard rate ratio.

4 Discussion

Chronic anaemia is an established risk factor for CVD outcomes in patients with kidney disease and in patients with heart failure.3,8,9Data of a recently performed prospective cohort study of 14 000 subjects (The Atherosclerosis Risk in Communities (ARIC) Study) suggest that anaemia may also be an independent cardiovascular risk factor in the general population.2This hypothesis is substantiated by previous results of the Framingham study demonstrating a U-shaped relation between haematocrit and mortality,10as similarly found in our analysis (Fig. 1). For the first time, we now analysed the impact of haemoglobin concentrations on long-term survival of patients with manifest CHD.

Although angiographic outcome and in-hospital mortality after PCI was comparable in all subgroups, we found a marked difference in survival with regard to both patients being divided in quintiles of haemoglobin (Table 6) as well as subgroups by haemoglobin strata in steps of 1g/d (Fig. 1). Remarkably, 48% of all deaths and 43% of all cardiac deaths occurred in the lowest haemoglobin quintile, which included only 21% of all PCI patients. This significant association between haemoglobin and death was confirmed by Cox regression models including potential co-variates, as age, cardiac status, classical coronary risk factors and creatinine.

Because of the retrospective data collection in our study, no firm conclusions on pathophysiologic background or therapeutical strategies can be drawn. We were interested in the fact, however, that the majority of patients in all haemoglobin quintiles died of cardiac reasons (44 of 62 patients) with a significantly increased cardiac mortality in the quintile with the lowest haemoglobin; and that only few patients (five of 62) died of cancer. Renal function may have influenced haemoglobin concentrations and outcome because creatinine levels were higher in the lowest haemoglobin quintiles. However, this inverse association of mortality and haemoglobin was also observed in patients with normal creatinine values, and in multivariate analysis afteradjustment for creatinine.

There are several potential mechanisms how chronic anaemia could increase cardiac mortality in patients with coronary heart disease. The presence of anaemia does result in ventricular remodelling and ventricular hypertrophy with higher oxygen consumption which is considerably unfavorable in CHD.9,11,12Moreover, chronic anaemia may pronounce myocardial ischemia. Thus, Wu et al. recently demonstrated that patients with acute MI and anaemia at admission had a significantly higher 30-day mortality than those with normal haemoglobin concentrations.13In contrast to this study by Wu et al. and one performed by Zindrou and coworkers about the higher in-hospital mortality of anaemic patients after bypass surgery,4we did not observe any impact of haemoglobin concentration on in-hospital mortality. This discrepancy might be due to the much lower acute haemodynamic stress and much shorter myocardial ischemia during PCI compared to bypass surgery and acute MI. On the other hand, almost all of the deaths in our study occurred within the first 600 days after the index PCI suggesting a more acute pathophysiological effect.

In summary, the most interesting and novel finding of this study is the markedly increased all-cause and cardiac mortality of anaemic patients within a relatively short time period after PCI. Since anaemia is a relatively frequent condition in the general population which remains often under-diagnosed and untreated as stated by others,14and in view of the high risk and large number of patients concerned, possible treatment options should be concerned and further evaluated.


The authors are indebted to Mr Ing. (grad.) Klaus Balkenhoff for establishing the PCI database. We thank all technicians and nurses in the catheterization laboratories for their ongoing support.


View Abstract