OUP user menu

Women have better long-term prognosis than men after cardiac resynchronization therapy

Stanislava Zabarovskaja, Fredrik Gadler, Frieder Braunschweig, Marcus Ståhlberg, Jonas Hörnsten, Cecilia Linde, Lars H. Lund
DOI: http://dx.doi.org/10.1093/europace/eus039 1148-1155 First published online: 7 March 2012


Aims Cardiac resynchronization therapy (CRT) improves prognosis in patients with moderate-to-severe heart failure, reduced left ventricular ejection fraction, and wide QRS complexes. However, CRT may be under-utilized in women and data on long-term follow-up are still scarce. The aim was to investigate long-term mortality and hospitalization and prognostic impact of gender after CRT.

Methods and results Data on 619 consecutive patients (19% women) that received CRT at a single centre between 1998 and 2008 were collected from electronic medical records and national death and hospitalization registries up to 2010. The primary endpoint was death of any cause and the secondary endpoint was combined death of any cause or heart failure hospitalization. Over a mean follow-up of 1320 ± 786 days, 215 (35%) patients reached the primary endpoint and 437 (71%) the secondary endpoint. Overall, 1-, 5-, and 10-year survivals were 91, 63, and 39%, respectively. Female gender was the only independent predictor of all-cause mortality; hazard ratio (HR) 0.44 [95% confidence interval (CI), 0.21–0.90; P= 0.025]. Women also had a trend towards lower risk for the secondary endpoint, HR 0.68 (95% CI, 0.45–1.04; P= 0.072).

Conclusion In this registry analysis, patients with CRT had similarly high short-term survival to those in controlled trials, and this favourable prognosis was sustained over the long term.Women had lower all-cause mortality than men.

  • Heart failure
  • Cardiac resynchronization therapy
  • Gender
  • Outcome
  • Long-term prognosis


Several large randomized controlled trials (RCTs) have shown that cardiac resynchronization therapy (CRT) decreases mortality1,2 and hospitalization,14 and improves functional capacity3,4 in New York Heart Association (NYHA) class III–IV heart failure (HF), impaired left ventricular ejection fraction (LVEF), and wide QRS complexes. More recently, CRT has appeared beneficial also in mild-to-moderate HF.58

However, data on long-term follow-up of CRT patients are scarce, and little is known about predictors of long-term outcomes post-CRT.

Heart failure is slightly less common in women than in men, and women have better prognosis in epidemiological surveys but less clearly in clinical drug trials, although the latter is clouded by female under-representation and limited power.9 Women are under-represented in RCTs of CRT, and the relative benefit for women compared with men in CRT studies and registries has been conflicting.24,6,7,1015 Cardiac resynchronization therapy appears to be under-utilized in women.16

Therefore, we analysed a large single institution cohort of CRT patients with regard to the role of gender and predictors of long-term outcome.



We retrospectively screened 627 consecutive patients who received CRT alone (CRT-P) or in combination with an implantable cardioverter defibrillator (ICD) (CRT-D) at Karolinska University Hospital, Stockholm, Sweden, between 1998 and 2008. Patients included in a trial on atrioventricular junction ablation and pacing for permanent atrial fibrillation, and children were excluded (total n= 8), leaving 619 for analysis. Electronic medical records were examined by the same individual and 95 variables were collected and entered into a database. The study was approved by the local ethics committee.


Right atrial leads were placed in the atrial appendage or roof. Right ventricular (RV) leads were placed in the apex or in a mid-septal position. Left ventricular (LV) leads were placed in a lateral or posterolateral vein branch, if possible. The aim was maximum electrical delay between the RV and LV leads. When upgrading patients from existing pacing systems, such measurements were generally unobtainable and placement in a posterolateral branch was considered optimal. Bipolar leads were used for atrial and RV leads. At the start of the registry period, only unipolar LV leads were available, but after availability of bipolar LV leads such were used.


Data on hospitalization and death and their causes were collected from medical records and from the Swedish death and hospitalization registries as of 2010. Implantable cardioverter defibrillator-10 codes for primary cause of death or hospitalization were used with definitions as listed in Table 2. Complete pre- and post-CRT data were not available in a large-enough number of patients to define or assess response. However, pair-wise changes in NYHA class, LVEF, and left ventricular end diastolic diameter (LVEDD) were assessed. Complications were defined as electrode dislocation, electrical dysfunction, local bleeding, infection, perforation, and pneumothorax.

Statistical analysis

Twenty-one variables were selected from the 95 as clinically relevant. Baseline characteristics in women and men were compared with un-paired t- and χ2 tests. The primary endpoint was defined as death of any cause and the secondary endpoint as combined death of any cause or HF hospitalization. Crude outcomes were analysed by Kaplan–Meier and causes of death and hospitalization were compared with χ2 tests. Predictors of outcomes were assessed with univariate and multivariate Cox regression. Twelve of 21 variables were significant univariate predictors of death of any cause and were entered into a multivariate analysis. For consistency, the same 12 variables were used for multivariate analysis also for the secondary endpoint. Overall and gender-specific changes in NYHA, LVEF, and LVEDD were analysed with paired t-tests and compared between the genders with un-paired t-tests. A P value of <0.05 was considered significant. Statistical analysis was performed in PASW Statistics 17.0.

The proportional hazard assumption was assessed by scaled Schoenfeld residuals. Some violation to the proportional hazard assumption was detected for aetiology and for haemoglobin for the secondary endpoint (but not the primary), but when stratified for aetiology or categorized by haemoglobin, no change in the significance of the variables entered into the multivariate analysis was detected. Proportional hazards assumptions were tested in R version 2.9.2 (2009 The R Foundation for Statistical Computing).


Baseline characteristics, at the time of implantation, are depicted in Table 1; 19% were women. There were no major or unexpected differences between the genders, but women had less frequent ischaemic aetiology. Mean follow-up was 4 years (1320 ± 786 days, range 2 days to 12 years). All patients alive at last follow-up had >1 year of follow-up; 56 patients had <1 year of follow-up and all these died within 1 year of implantation.

View this table:
Table 1

Baseline characteristics at the time of cardiac resynchronization therapy implantation

Total, n= 619Male, n= 500Female, n= 119P value
Age (years)68 ± 11 (17–90)68 ± 10 (28–90)67 ± 12 (17–89)0.142
Female gender119/19%
Late period of implantationa456/74%367/73%89/75%0.757
CRT including ICD211/34%180/36%31/26%0.040
Ischaemic aetiology331/61%287/65%44/42%<0.001
Pre-existing pacemaker175/28%135/27%40/34%0.150
HR72 ± 15 (30–150)71 ± 15 (30–150)74 ± 13 (47–110)0.166
MAP86 ± 13 (50–132)86 ± 14 (50–132)86 ± 12 (53–120)0.833
NYHA class0.666
 II49 (13%)41 (13%)8 (10%)
 III317 (81%)250 (80%)67 (85%)
 IV24 (6%)20 (7%)4 (5%)
Weight82 ± 17 (40180)85 ± 16 (50180)70 ± 16 (40120)<0.001
EFb24 ± 8.6 (7–55)24 ± 8.6 (7–55)24 ± 8.3 (10–50)0.791
LVEDD65 ± 9.6 (35100)66 ± 9.3 (35100)62 ± 9.9 (4085)0.001
Permanent atrial fibrillation or flutter223/43%192/44%31/34%0.072
Paroxysmal atrial fibrillation or flutter77/15%61/14%16/18%0.392
His ablation before CRTc26/4%18/4%8/7%0.127
His ablation after CRTc56/9%47/9%9/8%0.530
QRS (ms)d155 ± 32 (72–268)154 ± 32 (72–268)160 ± 27 (92–230)0.160
Aortic stenosis, moderate or severe14/3%12/2.4%2/2%0.635
Mitral insufficiency, moderate or severe157/25%128/26%29/24%0.782
Diabetes mellitus167/27%145/29%22/19%0.020
Hb (g/L)135 ± 16 (93179)136 ± 17 (93179)129 ± 14 (98164)<0.001
Na (mmol/L)139 ± 3.3 (123–150)139 ± 3.3 (123–150)138 ± 3.5 (124–144)0.239
K (mmol/L)4.2 ± 0.4 (2.65.9)4.3 ± 0.4 (2.65.9)4.1 ± 0.4 (2.95.3)0.002
GFRe62 ± 24 (11–242)62 ± 23 (11–139)65 ± 28 (18–243)0.246
  • Data are presented as mean ± standard deviation (SD) (range) or as n (%) when appropriate.

  • n and % may not add to 619 or 100%, respectively, because of missing data.

  • Hb, haemoglobin; Na, sodium; K, potassium; GFR, glomerular filtration rate (mL/min/1.73 m2). Bold numbers indicate a significant difference between the genders.

  • aLate period of implantation: 2003–2008 compared with early: 1998–2002.

  • bForty-two patients had EF > 35%.

  • cIn all patients, i.e. may have occurred in both permanent and paroxysmal atrial fibrillation or flutter.

  • dFifty-four patients had QRS < 120 ms.

  • eGlomerular filtration rate was calculated according to the Modification of Diet in Renal Disease (MDRD)17 [GFR in mL/min = 186 × (creatinine in mg/dL) −1.154 × age – 0.203 × 1.21 if black × 0.74 if female].

Primary and secondary endpoints

Two hundred and fifteen (35%) patients reached the primary endpoint. Overall survival was 91, 63, and 39% at 1, 5, and 10 years, respectively (Figure 1A). Women had lower hazard of death [multivariate hazard ratio (HR)] 0.439 [95% confidence interval (CI), 0.214–0.903; P= 0.025] (Figure 1B and Table 3). Four hundred and thirty-seven (71%) patients reached the secondary endpoint. Event-free survival was 57% at 1 year and 26% at 5 years (Figure 2A). Women had non-significantly lower hazard of reaching the secondary endpoint (HR 0.681, 95% CI, 0.449–1.035; P= 0.072) (Figure 2B).

Figure 1

(A) Kaplan–Meier curve for the whole population for the primary endpoint: all-cause mortality. (B) Graphic Cox regression for all-cause mortality shown by gender; adjusted for the 11 variables that were significant (other than gender) in univariate analysis (age, concomitant implantation of an implantable cardioverter defibrillator, aetiology, mean arterial pressure, New York Heart Association class, ejection fraction, aortic stenosis, mitral insufficiency, diabetes mellitus, haemoglobin, and glomerular filtration rate).

Figure 2

(A) Kaplan–Meier curve for the whole population for the secondary endpoint: combined all-cause mortality or heart failure hospitalization. (B) Graphic Cox regression for combined all-cause mortality and heart failure hospitalization, shown by gender; adjusted for the same 11 variables as in Figure 1B.

Causes of death and hospitalization

Causes of death are depicted in Table 2. Mortality was largely cardiovascular. Heart failure was the primary diagnosis for 15% of all deaths, whereas cardiovascular non-HF mortality was the cause of 56% of all deaths. There was no difference in cause of death between genders (P= 0.611). Heart failure was the primary or contributing cause of death in 68% of all deaths (68% overall: 67% for women and 68% for men, P= 0.880).

View this table:
Table 2

Cause of death

Mortality, total215/619 (35%)187/500 (37%)28/119 (24%)
Cause of death  P= 0.611a
P= 0.880b
 Unknown58/215 (27%)51/187 (27%)7/28 (25%)
 Known157/215 (73%)136/187 (73%)21/28 (75%)
 Heart failure mortality primarya24/157 (15%)20/136 (15%)4/21 (19%)
 Cardiovascular non-heart failure mortality88/157 (56%)75/136 (55%)13/21 (62%)
 Cancer mortality13/157 (8%)11/136 (8%)2/21 (10%)
 Other mortality32/157 (20%)30/136 (22%)2/21 (10%)
Heart failure mortality primary or contributingb107/157 (68%)93/136 (68%)14/21 (67%)
  • Date of death and the ICD-10 cause were obtained from the Swedish national death and hospitalization registry. P values are for difference between genders in cause of death. Causes of death were classified as follows: heart failure: I110, I420, and codes starting with I50; cardiovascular: I052, I080, I10, I119, I209, I219, I251, I255, I258, I259, I350, I422, I425, I429, I48, I713, and I709; cancer: C159, C179, C189, C259, C349, C61, C679, C80, C859, and C920; and other: A412, A419, B99, E117, E142, E149, G710, I639, I710, I713, I718, J449, K550, K729, N189, N19, N390, R990, R998, R999, W06, W19, and X590 (there were no additional causes of death).

  • aWith heart failure as the primary cause.

  • bWith heart failure as the primary or contributing cause.

Heart failure was the most common cause of hospitalization. Deceased patients had more hospitalizations per year than alive patients. Women had significantly less hospitalizations (1.0 vs. 1.7 per patient-year, P= 0.017), but HF hospitalizations were fewer only by trend (0.4 vs. 0.8 per patient-year, P= 0.148).

Predictors of outcomes

Twelve of the 21 clinically relevant variables were univariate predictors of survival (Table 3). These were entered into a multivariate model and only gender remained significant (HR 0.439, P= 0.025). Some but not all of the 12 variables that predicted death also predicted combined death/HF hospitalization. In multivariate analysis with the same 12 variables as in the model for death, female gender was still protective but this did not reach statistical significance (HR 0.681, P= 0.072).

View this table:
Table 3

Predictors of all-cause mortality

Hazard ratioPHazard ratioP value
Age1.038 (1.0231.053)<0.0011.027 (0.998–1.057)0.070
Female gender0.576 (0.3870.857)0.0070.439 (0.2140.903)0.025
Late period of implantation0.799 (0.591–1.080)0.145
CRT including ICD0.604 (0.4340.842)0.0030.792 (0.457–1.374)0.407
Ischaemic aetiology1.393 (1.0271.889)0.0331.194 (0.725–1.966)0.487
Pre-existing pacemaker0.886 (0.655–1.198)0.432
HR1.000 (0.988–1.011)0.940
MAP0.969 (0.9570.981)<0.0010.980 (0.960–1.001)0.059
NYHA class0.0010.751
 II vs. IV0.236 (0.1020.545)0.0010.714 (0.230–2.215)0.559
 III vs. IV0.421 (0.2410.736)0.0020.968 (0.407–2.304)0.941
Weight0.996 (0.985–1.006)0.431
EF0.966 (0.9480.984)<0.0010.985 (0.954–1.017)0.357
LVEDD1.005 (0.986–1.024)0.643
Permanent atrial fibrillation or flutter1.090 (0.815–1.459)0.561
QRS1.001 (0.996–1.007)0.615
Aortic stenosis, moderate or severe2.629 (1.3465.135)0.0051.647 (0.719–3.773)0.238
Mitral insufficiency, moderate or severe1.396 (1.0421.872)0.0251.061 (0.665–1.693)0.803
Diabetes mellitus1.381 (1.0321.847)0.0301.319 (0.832–2.091)0.239
Hb0.984 (0.9750.993)0.0010.986 (0.971–1.002)0.083
Na0.955 (0.911–1.001)0.055
K0.978 (0.673–1.421)0.907
GFR0.981 (0.974–0.988)<0.0010.998 (0.985–1.011)0.727
  • Cox regressions with the primary endpoint as outcome. Twelve of 21 variables were significant in the univariate analysis and were entered into the multivariate analysis.

  • Hazard ratios presented with 95% confidence interval in parentheses.

  • Hb, haemoglobin; Na, sodium; K, potassium; GFR, glomerular filtration rate (mL/min/1.73 m2).

  • Bold numbers indicate significance in univariate and/or multivariate analysis.

Permanent atrial fibrillation or flutter at baseline was common (34% of women vs. 44% of men, P= 0.072), but did not predict the primary (Table 3) or secondary endpoint, even in univariate analysis. Assessment of paroxysmal atrial fibrillation/flutter was less reliable, but was judged present in 18% of women and 14% of men (P= 0.392). Among all patients, His ablation was performed 1462 ± 456 days prior to CRT implantation in 7% of women and 4% of men (P= 0.127), and 349 ± 456 days post-CRT implantation in 8% of women and 9% of men (P= 0.530). Permanent or paroxysmal atrial fibrillation/flutter did not predict the primary or secondary outcome with or without His ablation.

In univariate analysis, combining CRT with an ICD predicted decreased risk and ischaemic aetiology predicted increased risk of the primary but not secondary endpoint, but these risks did not persist in multivariate analysis.


New York Heart Association class, LVEF, and LVEDD improved significantly in both women and men. The improvement was more pronounced in women for all of the above, although the difference in improvement between men and women was statistically significant only for ejection fraction (EF; P= 0.006).


Complication rates were higher among women (13%) than among men (8%), but only by trend (P= 0.079, Table 4).

View this table:
Table 4


MaleFemaleTotalP value
None462 (92%)104 (87%)566 (91%)0.079
Electrode dislocation8 (2%)5 (4%)13 (2%)0.075
Electrical dysfunction12 (2%)4 (3%)16 (3%)0.553
Local bleeding3 (1%)1 (1%)4 (1%)0.769
Infection/perforation3 (1%)2 (2%)5 (1%)0.237
Pneumothorax2 (0.4%)0 (0%)2 (0.3%)0.490
Other10 (2%)3 (3%)13 (2%)0.722

Off-label implantation

Of the 619 patients, 132 patients did not meet all classical indications for CRT (all patients were implanted prior to NYHA II becoming an indication), i.e. they had any of NYHA II, EF > 35% and/or QRS < 120 ms. Of these, 49 patients had NYHA II, 42 patients had EF > 35%, and 54 patients had QRS < 120 ms. Reasons for this included enrolment in the REVERSE8 and OPSITE18 studies, as well as mechanical rather than electrical dyssynchrony and atrio-ventricular block or His ablation, and conventional pacing indication. For the 487 patients who met all strict criteria, survival was similar to the overall 619 patients. Female gender but also higher mean arterial pressure (MAP) significantly predicted survival, and female gender was now a significant predictor also of event-free survival. For the 132 patients who did not meet all strict criteria, survival was higher than for the overall population, and female gender did not predict prognosis.


To our knowledge, this is the largest long-term follow-up registry of CRT patients to date. Our main findings are (i) female gender is a strong and independent predictor of survival after CRT; (ii) short-term outcomes in this large unselected single institution cohort are similar to those in RCTs;1,2,7 and (iii) this favourable prognosis is sustained over the long run.

Female gender predicts prognosis after CRT

In our study, women had a remarkably better prognosis, with an adjusted HR for all-cause mortality of 0.439 compared with men. Moreover, gender was the only significant predictor after adjustment for numerous variables that were significant predictors in univariate analysis and clinically relevant.

Female gender is believed to be associated with better prognosis in a general HF population.9 Also in CRT populations, there are suggestions that women fare better. In MADIT-CRT, the hazard ratio for death or HF for CRT-D vs. ICD alone was 0.31 for women vs. 0.72 for men (P for interaction <0.01).6,11 In the MIRACLE study, women had longer to first event than men.10 In contrast, in RAFT,7 COMPANION,19 CARE-HF,2 and REVERSE8 there was no difference in effect or gender–therapy interaction. In single-centre studies, female gender was either protective13,14 or not protective.15

In MADIT-CRT, the better prognosis among women was ascribed to baseline differences: women had more non-ischaemic aetiology, left bundle branch block, utilization of β-blockers, and less right bundle branch block, atrial fibrillation, and renal dysfunction. These factors may account for both better prognosis irrespective of CRT and better response to CRT. Also, among healthy subjects, women have smaller LV mass and on average 4–10 ms narrower QRS than men,20,21 and in MADIT-CRT, the authors speculated that female patients may have had greater dyssynchrony for a given QRS.11 In our study, women had less ischaemic aetiology and a trend towards less permanent atrial fibrillation, but after adjustment for 12 variables including these, female gender remained significant, suggesting other, unknown, factors are important. Women had a significantly better response with regard to EF and non-significantly better response with regard to NYHA class and LVEDD. Thus, the better prognosis for women may be due to more favourable reverse remodelling and better response to CRT. Future research should be directed at identifying reasons why women may have more favourable reverse remodelling when neurohormonal activation and cardiac remodelling are interrupted.9

CRT is under-utilized in women

The proportion of women in our study (19%) is representative. In Europe, 24% of CRT recipients are women,22 and the corresponding figure in the USA is 22–26%,16 in RCTs 17–32%;1,2,4,5,7,12 and in larger registries 17–24%.1315,23

Though incidence and prevalence of HF are slightly less in women than in men, the proportion of women in the overall HF population is actually >50%.9,24 Women with HF have more often than men preserved EF, and women with impaired EF tend to have better EF than men,16,25 but women tend to have more impaired functional capacity.25,26 Although women have been under-represented also in clinical drug trials, utilization of other evidence-based therapies is not nearly as skewed towards men. The utilization of β-blockers and angiotensin-converting enzyme inhibitors in men with systolic HF is ∼49 and 80%, respectively, and women with systolic HF ∼39 and 74%, respectively.27 Other studies also suggest that women receive evidence-based drug therapy to a similar or nearly similar extent as men do.28,29

The reasons for under-utilization of CRT in women are therefore not readily apparent. Studies of patient attitudes towards device therapy suggest lower enthusiasm for technology and higher concerns for complications among women.29 Women had a trend towards more complications in our study, which is consistent with data from the US National Cardiovascular Data Registry ICD Registry. Here, 27% of 161 470 ICD recipients were women, and 34% of men and 39% of women ICD recipients received concomitant CRT. Women had a significantly higher odds of any adverse event (odds ratio 1.32) and major adverse events (odds ratio 1.71). The authors speculated that smaller body size and vessel diameter may make women more susceptible to mechanical complications such as perforation, coronary venous dissection, lead dislodgment, and haemothorax and pneumothorax.30

Short-term outcomes with CRT

Randomized controlled trails are limited by strict inclusion and exclusion criteria and rigorous follow-up, which may impair generalizability. Specifically, patients with atrial fibrillation, moderately increased QRS width, higher age, and women were either excluded or under-represented.31,32 Indeed, based on RCT criteria, as few as 1% of patients with HF may qualify for CRT,33 although this figure may increase with indications expanding to milder disease. Estimated 1-year survival in RCTs was 85–95% and estimated 1-year event-free survival was 56–95%.1,2,6,7 In comparison, in our registry, at 1 year, overall survival was 91% and survival free from HF hospitalization was 57%. The latter figure was lower than that in RCTs, which may reflect a higher prevalence of atrial fibrillation, older age, and/or less rigorous follow-up in our registry, and/or MADIT-CRT,6 RAFT,7 and REVERSE8 including less symptomatic patients. These observations provide assurance that the benefits of CRT as demonstrated in large RCTs can also be achieved in un-selected populations with comorbidities and historical exclusion criteria.

Long-term outcomes with cardiac resynchronization therapy

Few long-term data are available from RCTs. In CARE-HF, LV reverse remodelling was most marked during the first 3–9 months and overall benefits were sustained for 29 months.34,35 Medium-term follow-up has been assessed in several cohorts. van Bommel et al.14 analysed 716 patients followed for a mean of 25 months and observed 1- and 2-year survival of 92 and 84%, respectively. Auricchio et al.13 analysed 1303 patients followed for a median of 34 months and observed 1- and 5-year survival of 92 and 56%. Gasparini et al.23 analysed 317 patients followed for a median of 36 months and observed 10% annual mortality. Kronborg et al.15 analysed 70 patients followed for 4 years and observed 85 and 47% 1- and 5-year survival. Only van Bommel et al.14 examined morbidity and demonstrated a survival free from cardiovascular hospitalization at 1 and 2 years of >80% and >70%. The majority of causes of death were cardiac and/or HF.1315,23 Our 619 patients were followed over a mean of 4 years (1320 ± 786 days) and had similar outcomes and causes of death. Previous studies13,15,23 have not analysed hospitalization, which may be difficult to assess outside the accurate documentation in RCTs. In contrast, linking our cohort to the nationwide hospitalization registry, we were able to retrieve complete and reliable information about hospitalizations and their causes. Our patients had survival free from HF hospitalization of 57% at 1 year and a remarkable 26% at 5 years.

Predictors of mortality and/or heart failure hospitalization

Numerous parameters known to predict prognosis in overall HF populations9 and in CRT populations,1315,23 such as higher age, higher NYHA class, lower LVEF and MAP, ischaemic aetiology, renal dysfunction, hyponatraemia, anaemia, and diabetes, also predicted all-cause mortality and/or combined mortality/HF hospitalization in univariate but not multivariate analysis in our study. Of interest is the high proportion of atrial fibrillation or flutter. Interestingly, and in contrast to previous studies,13,14 atrial fibrillation did not predispose to worse outcomes, even in univariate analysis and even in the absence of His ablation. Whether CRT-D improves prognosis beyond CRT-P is controversial; in our study it did so in univariate but not multivariate analysis. Again, the fact that female gender was the only independent predictor of prognosis and that it was stronger than other markers such as LVEF requires further study.

In summary, our observations suggest that women have better prognosis than men after CRT and that special efforts should be directed at reaching women for this life-saving and symptom-alleviating therapy. In this registry, we do not address cost effectiveness, but other recent data suggest that CRT is cost effective, even in mild–moderate HF.36 Future research should be directed at identifying reasons for these gender differences and at examining whether indications for CRT for women should be different and/or expanded.


Our study is limited by its retrospective registry character and lack of a suitable control group. We cannot show that women respond better to or have a greater mortality or morbidity benefit from CRT, but rather that among patients with CRT, women do better than men. However, the large sample size, the availability of a large number of clinically relevant variables for multivariate analysis, and large difference in hazard between men and women lend assurance to our conclusions. We do not have data on QRS configuration, such as left or right bundle branch block or on body surface area, precluding indexing. Owing to the retrospective nature, our registry did not have complete data on response, and we were not able to define or analyse predictors of response or whether the better prognosis in women was associated with favourable reverse remodelling and CRT response. The recent European CRT survey22 showed that CRT implantation outside of strict classical indications occurs widely, as it did in this registry. However, our findings persisted when patients that did not meet all classical criteria were excluded.


This work was supported by grants from the Stockholm County Council, Stockholm, Sweden, to L.H.L. and C.L. and the Swedish Heart-Lung Foundation, Stockholm, Sweden, to L.H.L. and C.L.


We thank Lina Benson, MSc, for statistical assistance.

Conflicts of interest: F.G. has received speakers' and consultant fees from Medtronic Inc., Kista, Sweden, and St. Jude Medical, Inc., Järfälla, Sweden. F.B. has received funding from Medtronic Inc., Kista, Sweden and speakers' fees from Medtronic Inc., Kista, Sweden, St. Jude Medical, Inc., Järfälla, Sweden, and Vitatron Inc., Kista, Sweden. J.H. has received speaker' fees from Medtronic Inc., Kista, Sweden, St. Jude Medical, Inc., Järfälla, Sweden and Vitatron Inc., Kista, Sweden. C.L. has received funding from Medtronic Inc., Kista, Sweden and speakers' fees from Medtronic Inc., Kista, Sweden, St. Jude Medical, Inc., Järfälla, Sweden, and Vitatron Inc., Kista, Sweden. L.H.L. has received consultant fees from St. Jude Medical, Inc., Järfälla, Sweden.


View Abstract