Effectiveness by gender and age of renin-angiotensin system blockade in heart failure-A national register-based cohort study.

PURPOSE: Investigate effectiveness by gender and age and equity implications of treatment with renin-angiotensin system blockade (RASb) in heart failure (HF) patients.
METHODS: In this population-based register study, we used inpatient data from 2006 to 2010 for patients age 20 years or older with no HF hospitalisation for minimum of 1 year before an index hospitalisation. A wash-out period for RASb of 6 months preceding admission was used. Hospital data were linked with drug dispensation data and cause of death data. The associations between time-dependent RASb exposure and all-cause death and HF death, respectively, were examined by Cox regression models. Interactions by gender and age were also investigated on the multiplicative and additive scales.
RESULTS: Thirty thousand seven hundred twenty-one patients were analysed. Fifty-one percent were women. Median age was 83. Fifty-three percent of women and 64% of men received RASb after the index hospitalisation. Younger patients were more likely to receive RASb than older ones. One-year mortality was 28%. RASb was associated with an overall hazard ratio (HR) for all-cause death of 0.72 (95% confidence interval 0.69-0.75), and an HR of 0.85 (0.77-0.93) for HF death. Interaction analyses showed HRs for all-cause death associated with RASb between 0.12 (0.10-0.13) in the youngest, and 0.80 (0.76-0.84) in the oldest patients.
CONCLUSIONS: RASb appeared effective for women and men and for patients of all ages in this hospitalised HF cohort. No gender difference in effectiveness was found. RASb exposure was low overall, indicating a need for improved adherence to treatment guidelines. Treatment with RASb may be inequitable for women and older patients.

Despite evidence of efficacy of renin‐angiotensin system blockade (RASb) in heart failure (HF), RASb is underused, particularly in female and older patients.

RASb effectiveness in unselected HF patients is incompletely understood—some data suggest less effectiveness among women.

This study investigated effectiveness of RASb by gender and age among HF patients.

RASb appeared effective in women and men. No gender difference in effectiveness was found. RASb seemed effective in all age groups, although less among older patients.

HF treatment may be inequitable for women and older patients, but more detailed clinical data are needed to determine RASb eligibility.

INTRODUCTION

Efficacy of renin‐angiotensin system (RAS) blocking drugs in heart failure (HF) with reduced ejection fraction (HFrEF) is established in clinical trials,1, 2, 3 where mortality reductions are seen across gender, age, and ethnicity groups.1, 4, 5, 6 Hence, treatment guidelines recommend RAS blockade (RASb) for the vast majority of HF patients.7 Effectiveness8 studies (ie, observational studies of RASb use in HF patients) also show mortality reductions overall and that RASb is effective among elderly and female patients,9, 10, 11, 12, 13 although women may benefit less than men.11, 14

Although RASb reduces mortality by 16% to 40% in patients with HFrEF,1, 2, 3 under‐prescription as well as gender and age inequity in RASb treatment have been demonstrated.9, 15, 16, 17

Equitable health care requires efficacy studies in different patient groups, as effects and side effects may vary.18, 19 Women and older patients are typically underrepresented in HF research.20, 21 Moreover, all patients regardless of age, gender, or sociodemographics should have equal chances of receiving effective evidence‐based treatment according to need. Consequently, observational studies of effectiveness of RASb in different groups of HF patients are relevant in the context of equity in health care.

To our knowledge, no previous study has analysed total population data for RASb effectiveness according to age and gender in HF.

Aim

Building on our previously reported data, revealing less RASb treatment among female and elderly HF patients,15 we aimed to study the effectiveness of RASb in the same cohort.

The research questions were as follows:

Is RASb exposure associated with mortality reduction in a hospitalised Swedish HF population?

Does the association between RASb exposure and mortality differ by gender or age?

METHODS

Material

This population‐based cohort study used register data from the Swedish National Patient Register,22, 23, 24 the Swedish Prescribed Drug Register,25, 26 and the Swedish Cause of Death Register.27 The registers are held by the Swedish National Board of Health and Welfare. The coverage of the patient register is over 99% for inpatients. The Prescribed Drug Register records dispensed drugs in Sweden with 95% to 100% completeness.25 The Cause of Death register comprises all deaths among persons registered in Sweden at the time of death, with a coverage of 99%.27

Study population

We included inpatients aged 20 years or older from the patient register discharged from hospital 1 July 2006 to 30 December 2010, who survived hospitalisation with a primary diagnosis of HF (The International Classification of Diseases [ICD‐10] codes I11.0, I13.0, I13.2, I42.0, I42.3‐I42.9, I50.0, I50.1, and I50.9) and had not been hospitalised for HF 1 July 2005 to 30 June 2006 nor had a RASb dispensation in the 6 months preceding hospitalisation (Figure 1). The discharge date for the first HF hospitalisation during the study period was defined as the index date. We included 30 721 patients.

Figure 1

Flow chart of patient selection and inclusion. HF, heart failure; RASb, renin‐angiotensin system blockade

Outcomes

Two outcomes were analysed separately: HF death and all‐cause death. Date of death and underlying cause of death were retrieved from the cause of death register. Time was from index date until event, censoring, or end of study, ie, 31 December 2010. Subjects were followed up to 4.5 years.

Exposures

Renin‐angiotensin system blockade

At least one dispensation of either an angiotensin‐converting enzyme inhibitor (ACEI) or an angiotensin receptor blocker (ARB) at any time during follow‐up was regarded as RASb exposure. Doses were not recorded. Using the date of the first dispensation after the index date, a time‐dependent exposure variable for RASb was computed, such that follow‐up time before the first dispensation was defined as unexposed time, and follow‐up time after the first dispensation (until death, censoring, or end of study) was defined as exposed time, to avoid survival bias. Only the first dispensation was considered, ie, exposure information was not updated. No specific assumptions regarding the length of prescription were made. Stock‐piling and nonadherence were not addressed. RASb exposure was defined as protective.

Gender

Gender was categorised as woman or man. Female gender was considered protective; thus man was the reference.

Age

Age was categorised as 20 to 64 years, 65 to 74 years, 75 to 84 years, and 85 years or older. The highest age category was as the reference.

Covariates

Comorbidity, defined as additional ICD‐10 diagnoses at any inpatient hospitalisation 1.5 years previous to and including the index hospitalisation, comprised hypertension, angina pectoris, myocardial infarction, atrial fibrillation/flutter, pacemaker, coronary artery bypass grafting (CABG), stroke, peripheral vascular disease, lung disease, renal dysfunction, diabetes mellitus, anaemia, dementia, cancer, liver disease, and rheumatic disease.

As entry time into the study varied for the subjects, hospitalisation‐free time in the study period before the index hospitalisation was computed.

Other HF therapies, ie, beta blocker (BB) and mineralocorticoid receptor antagonist (MRA) dispensed within 1 year following index, were recorded.

Statistical analysis

Descriptive statistics are reported as medians with quartiles for continuous variables or proportions and counts for categorical variables.

Cox proportional hazards regression models were fitted to estimate associations (hazard ratios [HRs]) between exposures and outcomes for all‐cause death and HF death, respectively. We computed 95% confidence intervals (CI) for the estimates.

Univariate models were fitted to estimate crude HRs for each exposure. Model 2 was adjusted for age. Model 3 was adjusted for age and gender. Model 4 further included comorbidities, hospitalisation‐free time, and BB or MRA medication.

To study whether the adjusted associations between RASb and the outcomes differed by gender, model 5 further included the interaction term between RASb and gender. HRs were then calculated for the combined exposures RASb and gender with the common reference category man without RASb. Similarly, in model 6, we added the interaction term between RASb and age to model 4, and HRs for the combined exposures RASb and age were calculated with the common reference category 85 years or older without RASb. For HF death, interaction was estimated for gender and RASb. Event rates were too low in some age groups for age interaction analysis to be reliable. To determine interaction on the additive scale, the Relative Excess Risk due to Interaction (RERI) was calculated. RERI≠0 denotes interaction, where RERI>0 designates positive interaction, and RERI<0 designates negative interaction.28 All exposures were coded as protective to facilitate interpretation of interaction effects.

To examine whether misclassification of comorbidities would bias estimates, we first assumed that the misclassification was due to underreporting, ie, all patients recorded as having a diagnosis have it, but some patients coded as not having a diagnosis actually also have it. Although a simplification, it is reasonable to assume this to be true and that misclassification is one directional. Thus, the association between the measured confounder and the true confounder is always positive.

A directed acyclic graph (Figure S1) was used to draw our assumptions about this one‐directional association and about associations of comorbidity with RASb and death, respectively. Due to these assumptions and because misclassification should not depend on RASb exposure, we could follow a similar reasoning as Vanderweele29 for unmeasured confounders. For comorbidity positively associated with both exposure and outcome, the magnitude of the exposure‐outcome association was underestimated and for comorbidity negatively associated with the exposure and positively associated with the outcome, the association was overestimated. The bias thus depends on comorbidity diagnosis.

All statistical analyses were performed in SPSS (v22, v23, v25). The significance level was 0.05, and tests were two‐sided.

Ethics

The study conforms to the declaration of Helsinki and was approved by the Swedish Central Ethical Review Board (reg. no. Ö 29‐2011).

RESULTS

Patient characteristics

Fifty‐one percent were women (Table 1). Median age was 83 years overall, and women were older than men. Patients receiving RASb were younger than those without RASb. Median hospitalisation‐free time before index was 3.6 years. Median follow‐up time was 450 days.

Table 1

Patient characteristics by exposure to renin‐angiotensin system blockade, age, and gender

	Number (%)	Distribution (%)
	Total	RASb	No RASb	Age 20‐64	Age 65‐74	Age 75‐84	Age ≥ 85	Women	Men
Number (%)	30 732	17 836 (58.0)	12 896 (42.0)	3876 (12.6)	4216 (13.7)	9779 (31.8)	12 861 (41.8)	15 667 (51.0)	15 065 (49.0)
Age (y)
Median	83	79	86					85	80
25th percentile	74	69	80					78	70
75th percentile	88	85	90					89	86
Age 20‐64	3876 (12.6)	3237 (18.1)	639 (5.0)					1240 (7.9)	2636 (17.5)
Age 65‐74	4216 (13.7)	3197 (17.9)	1019 (7.9)					1706 (10.9)	2510 (16.7)
Age 75‐84	9779 (31.8)	6101 (34.2)	3678 (28.5)					4828 (30.8)	4951 (32.9)
Age ≥ 85	12 861 (41.8)	5301 (29.7)	7560 (58.6)					7893 (50.4)	4968 (33.0)
Gender
Women	15 667 (51.0)	8233 (46.2)	7434 (57.6)	1240 (32.0)	1706 (40.5)	4828 (49.4)	7893 (61.4)	8233 (52.5)
Men	15 065 (49.0)	9603 (53.8)	5462 (42.4)	2636 (68.0)	2510 (59.5)	4951 (50.6)	4968 (38.6)	9603 (63,7)
Follow‐up time
Total person years	47 068
Range	1‐1644	1‐1644	1‐1643	1‐1644	1‐1643	1‐1644	1‐1642	1‐1644	1‐1644
Median	450	606	263	736	618	491	317	440	460
25th percentile	153	265	71	322	247	172	99	144	161
75th percentile	897	1032	640	1179	1079	936	689	884	912
Comorbidity
Hypertension	10 376 (33.8)	6091 (34.2)	4285 (33.2)	1099 (28.4)	1433 (34.0)	3336 (34.1)	4508 (35.1)	5730 (36.6)	4646 (30.8)
Diabetes Mellitus	4746 (15.4)	2745 (15.4)	2001 (15.5)	585 (15.1)	909 (21.6)	1743 (17.8)	1509 (11.7)	2250 (14.4)	2496 (16.6)
Angina Pectoris	3662 (11.9)	1887 (10.6)	1775 (13.8)	213 (5.5)	407 (9.7)	1185 (12.1)	1857 (14.4)	1890 (12.1)	1772 (11.8)
Myocardial infarction	5503 (17.9)	2990 (16.8)	2513 (19.5)	424 (10.9)	723 (17.1)	1881 (19.2)	2475 (19.2)	2517 (16.1)	2986 (19.8)
Atrial fibrillation/flutter	12 791 (41.6)	7281 (40.8)	5510 (42.7)	958 (24.7)	1568 (37.2)	4393 (44.9)	5872 (45.7)	6528 (41.7)	6263 (41.6)
Pacemaker	2026 (6.6)	1102 (6.2)	924 (7.2)	115 (3.0)	197 (4.7)	678 (6.9)	1036 (8.1)	939 (6.0)	1087 (7.2)
Stroke	2484 (8.1)	1145 (6.4)	1339 (10.4)	120 (3.1)	263 (6.2)	934 (9.6)	1167 (9.1)	1244 (7.9)	1240 (8.2)
Renal dysfunction	2609 (8.5)	922 (5.2)	1687 (13.1)	208 (5.4)	335 (7.9)	939 (9.6)	1127 (8.8)	1043 (6.7)	1566 (10.4)
Vascular disease	1045 (3.4)	522 (2.9)	523 (4.1)	65 (1.7)	153 (3.6)	421(4.3)	406 (3.2)	459 (2.9)	586 (3.9)
Rheumatic disease	1067 (3.5)	506 (2.8)	561 (4.4)	65 (1.7)	138 (3.3)	416 (4.3)	448 (3.5)	750 (4.8)	317 (2.1)
Lung disease	5456 (17.8)	2848 (16.0)	2608 (20.2)	582 (15.0)	947 (22.5)	2059 (21.1)	1868 (14.5)	2805 (17.9)	2651 (17.6)
Liver disease	287 (0.9)	146 (0.8)	141 (1.1)	92 (2.4)	75 (1.8)	87 (0.9)	33 (0.3)	106 (0.7)	181 (1.2)
CABG	1720 (5.6)	1097 (6.2)	623 (4.8)	193 (5.0)	343 (8.1)	746 (7.6)	438 (3.4)	482 (3.1)	1238 (8.2)
Anaemia	3192 (10.4)	1385 (7.8)	1807 (14.0)	176 (4.5)	358 (8.5)	1021 (10.4)	1637 (12.7)	1776 (11.3)	1416 (9.4)
Cancer	2033 (6.6)	914 (5.1)	1119 (8.7)	118 (3.0)	304 (7.2)	771 (7.9)	840 (6.5)	694 (4.4)	1339 (8.9)
Dementia	1401 (4.6)	452 (2.5)	949 (7.4)	9 (0.2)	58 (1.4)	520 (5.3)	814 (6.3)	854 (5.5)	547 (3.6)
RASb	17 836 (58.0)			3237 (83.5)	3197 (75.8)	6101 (62.4)	5301 (41.2)	8233 (52.5)	9603 (63.7)
Beta blocker ≤1 ya	19 918 (64.8)	13 627 (76.4)	6291 (48.8)	3196 (82.5)	3145 (74.6)	6466 (66.1)	7111 (23.1)	9713 (62.0)	10 205 (67.7)
Mineralocorticoid receptor antagonist ≤1 ya	10 847 (35.3)	6678 (37.4)	4169 (32.3)	1604 (41.4)	1711 (40.6)	3410 (34.9)	4122 (32.1)	5533 (35.3)	5314 (35.3)

Abbreviations: CABG, Coronary Artery Bypass Grafting; RASb, renin‐angiotensin system blockade.

From index date.

Atrial fibrillation/flutter was the commonest comorbidity, present among 42%. Myocardial infarction, CABG, cancer, and renal dysfunction were more prevalent among men, while women had more hypertension, rheumatic disease, anaemia, and dementia. Most comorbidities were more prevalent with higher age, although diabetes mellitus, lung disease, CABG, and cancer decreased in the oldest.

Treatment

RASb was dispensed to 58% of patients, and more commonly among men and younger, compared with women and older patients. All comorbidities except hypertension and CABG were more frequent in the group not receiving RASb, with the largest difference in renal dysfunction (5% vs 13%) between RASb‐exposed and nonexposed patients. Among those receiving RASb, BB was more prevalent compared with the non‐RASb group (Table 1).

Outcome

Overall, 13 398 (44%) patients died during follow‐up. One‐year mortality was 28%, and 2‐year mortality was 37%. Death from HF occurred in 1909 patients (6%). There were 5665 (31.8%) deaths, including 796 (4.5%) HF deaths in those exposed to RASb, compared with 7733 (60.0%) and 1113 (8.6%) in those unexposed.

Table 2 shows that female gender was associated with a crude HR of more than one, which changed to 0.91 (95% CI, 0.88‐0.95) after adjustment for age, hospitalisation‐free time, RASb, comorbidity, and BB/MRA (model 4). Lower age was associated with a lower HR for all‐cause death, the HR for 20 to 64 year old being 0.19 (0.17‐0.21) (model 4). The unadjusted HR for all‐cause death associated with RASb was 0.46 (0.45‐0.48), and the adjusted overall HR was 0.72 (0.69‐0.75) (Table 2; model 4).

Table 2

Hazard ratio (HR) for death from all causes

N = 30 721 13 398 events (43.6%)	Model 1 Crude HR	Model 2 HR Adjusted for Age	Model 3 HR Adjusted for Age and Gender	Model 4 HR Adjusted for Age, Gender, Hospitalisation‐Free Timea, Comorbidityb, and Other Medicationc	Model 5 Model 4 Plus Interaction Gender * RASb	Model 6 Model 4 Plus Interaction Age Class * RASb	RERI
RASb	0.46 (0.45‐0.48)	0.62 (0.59‐0.64)	0.61 (0.59‐0.63)	0.72 (0.69‐0.75)	0.70 (0.67‐0.74)	0.80 (0.76‐0.84)
Gender (female)	1.11 (1.07‐1.15)		0.85 (0.82‐0.88)	0.91 (0.88‐0.95)	0.90 (0.86‐0.94)	0.91 (0.88‐0.94)
Age
20‐64		0.18 (0.16‐0.19)	0.17 (0.15‐0.18)	0.19 (0.17‐0.21)	0.19 (0.17‐0.21)	0.29 (0.25‐0.34)
65‐74		0.36 (0.33‐0.38)	0.35 (0.32‐0.37)	0.35 (0.33‐0.37)	0.35 (0.33‐0.38)	0.42 (0.39‐0.47)
75‐84		0.64 (0.62‐0.67)	0.63 (0.60‐0.65)	0.61 (0.59‐0.64)	0.61 (0.59‐0.64)	0.65 (0.62‐0.68)
≥85		Ref	Ref	Ref		Ref
Interaction female gender * RASb					1.04 (0.97‐1.11)
Interaction age * RASb
20‐64						0.50 (0.42‐0.60)
65‐74						0.69 (0.61‐0.79)
75‐84						0.86 (0.80‐0.93)
≥85						Ref
RERI female gender+RASb							0.05 (−0.00‐0.11)
RERI age 20‐64 + RASb							0.03 (−0.03‐0.09
RERI age 65‐74 + RASb							0.01 (−0.05‐0.07)
RERI age 75‐84 + RASb							0.00 (−0.05‐0.05)

Abbreviations: RASb, renin‐angiotensin system blockade; RERI, Relative Excess Risk due to Interaction.

Time in the study before the first hospitalization.

Hypertension, diabetes mellitus, angina pectoris, myocardial infarction, atrial fibrillation/flutter, pacemaker, stroke, renal dysfunction, vascular disease, rheumatic disease, lung disease, liver disease, Coronary Artery Bypass Grafting, anaemia, cancer, dementia.

Beta blocker and mineralocorticoid receptor antagonist.

Table 3 demonstrates the HRs for HF death, showing similar relations as for all‐cause death. The crude HR for HF death associated with RASb was 0.47 (0.42‐0.51) and the adjusted HR 0.85 (0.77‐0.93).

Table 3

Hazard ratio (HR) for death from HF

N = 30 721 1909 events (6.2%)	Model 1 Crude HR	Model 2 HR Adjusted for Age	Model 3 HR Adjusted for Age and Gender	Model 4 HR Adjusted for age, Gender, Hospitalisation‐free Timea, Comorbidityb, and Other Medicationc	Model 5 Model 4 + interaction Term for female Gender * RASb	RERI
RASb	0.47 (0.42‐0.51)	0.74 (0.67‐0.81)	0.73 (0.67‐0.81)	0.85 (0.77‐0.93)	0.82 (0.71‐0.95)
Gender (female)	1.31 (1.20‐1.43)		0.91 (0.83‐1.00)	0.93 (0.85‐1.03)	0.91 (0.81‐1.03)
Age
20‐64	0.06 (0.04‐0.08)	0.07 (0.05‐0.09)	0.07 (0.05‐0.09)	0.07 (0.05‐0.09)	0.07 (0.05‐0.09)
65‐74	0.09 (0.07‐0.12)	0.10 (0.07‐0.13)	0.10 (0.07‐0.13)	0.10 (0.07‐0.13)	0.10 (0.07‐0.13)
75‐84	0.35 (0.31‐0.39)	0.37 (0.33‐0.41)	0.37 (0.33‐0.41)	0.37 (0.33‐0.41)	0.37 (0.33‐0.41)
≥85	Ref	Ref	Ref	Ref	Ref
Female gender * RASb					1.05 (0.87‐1.26)
RERI female gender +RASb						0.05 (−0.11‐0.21)

Abbreviations: HF, heart failure; RASb, renin‐angiotensin system blockade; RERI, Relative Excess Risk due to Interaction.

Time in the study before the first hospitalization.

Hypertension, diabetes mellitus, angina pectoris, myocardial infarction, atrial fibrillation/flutter, pacemaker, stroke, renal dysfunction, vascular disease, rheumatic disease, lung disease, liver disease, Coronary Artery Bypass Grafting, anaemia, cancer, dementia.

Beta blocker and mineralocorticoid receptor antagonist.

To assess the bias of misclassification in our data, we ran analyses stratified by comorbidity diagnosis, showing small differences in HRs.

Interaction analyses

The interaction term for female gender and RASb was nonsignificant for all‐cause death. There was significant interaction between lower age class and RASb for all‐cause death. (Table 2; model 6). Interaction effects are also depicted as HRs for all‐cause death for each exposure combination compared with a common reference category in Figure 2, showing HRs between 0.12 (0.10‐0.13) in the youngest and 0.80 (0.76‐0.84) in the oldest patients.

Figure 2

Multiplicative interaction effects presented as adjusted hazard ratios with 95% confidence intervals (95% CI) for all‐cause death for different combinations of gender and renin‐angiotensin system blockade (RASb), and age and RASb, with a common reference. Adjusted for hospitalisation‐free time (time in the study before the first hospitalization); comorbidities (hypertension, diabetes mellitus, angina pectoris, myocardial infarction, atrial fibrillation/flutter, pacemaker, stroke, renal dysfunction, vascular disease, rheumatic disease, lung disease, liver disease, Coronary Artery Bypass Grafting, anaemia, cancer, dementia); and other medication (beta blocker and mineralocorticoid receptor antagonist)

The RERI for all‐cause death was not significant for gender and RASb nor age and RASb (Table 2). For HF death, the interaction term for female gender and RASb was nonsignificant (Table 3). HRs for each exposure category are presented in Figure 3. The RERI was not significant (Table 3).

Figure 3

Multiplicative interaction effects presented as adjusted hazard ratios with 95% confidence intervals (95% CI) for heart failure death for different combinations of gender and renin‐angiotensin system blockade (RASb) with the common reference man without RASb. Adjusted for hospitalisation‐free time (time in the study before the first hospitalization); comorbidities (hypertension, diabetes mellitus, angina pectoris, myocardial infarction, atrial fibrillation/flutter, pacemaker, stroke, renal dysfunction, vascular disease, rheumatic disease, lung disease, liver disease, Coronary Artery Bypass Grafting, anaemia, cancer, dementia); and other medication (beta blocker and mineralocorticoid receptor antagonist)

DISCUSSION

Women and older patients were less exposed to RASb than men and younger patients. Exposure to RASb was associated with a reduction in mortality for women and men alike. Interaction analysis showed a stronger association between survival and RASb for younger vs older patients.

Treatment

In our data, reflecting RASb treatment in Sweden 2005 to 2010, 58% were exposed to RASb. Treatment guidelines have consistently recommended RASb as first line therapy in HFrEF since 2005; however, treatment patterns could have changed since our data collection. A study of drug treatment quality in the Stockholm region in 2012 reported that 68% of HF patients in primary and secondary care were treated with RASb.30 According to the Swedish Board of Health and Welfare's national statistics of health care quality indicators, 61% of all hospitalised HF patients were treated with RASb in 2016 compared with 55% in 2008.31 Thus, available data indicate that RASb is still underused although the trend may be towards higher use.

Mortality

The overall all‐cause mortality of 28% at 1 year and 37% at 2 years in our study is similar to other observational studies.10, 11, 13, 32, 33

Treatment effects

Our overall adjusted HR of 0.72 for all‐cause death among patients exposed to RASb is comparable with a systematic overview of randomised trials reporting an OR for death of 0.74 for ACEI compared with placebo.4 As for observational data, Keyhan et al reported similar HRs for ACEI treatment (HR 0.80 for women and 0.71 for men),11 and the Euro Heart failure study found an OR of 0.5 for death within 12 weeks for RASb among hospitalised patients.34

RASb was also associated with a lower risk for HF death, with an HR of 0.85 (0.77‐0.93). In the SOLVD study of enalapril efficacy, the greatest mortality reduction was found among deaths attributable to progressive HF.2

Gender

In our study, as in other observational studies, half of the patients were women.9, 11, 13 Also consistent with previous reports, women had a higher adjusted overall survival than men.11 This may be partly due to the higher proportion of HF with preserved ejection fraction (HFpEF) among women and conversely to the lower prevalence of ischemic heart disease. Alternative explanations include pathophysiology, sex hormones, and remodelling in HF in men and women.14

The finding that RASb is associated with decreased mortality for both male and female HF patients is noteworthy, as earlier research has been conflicting.4, 35 A review summarises that although ACEIs prevent cardiovascular death and readmissions, clinical trials have failed to show significant all‐cause mortality benefit for women, possibly due to underpowered trials.14 Our data, with a representative proportion of women, reflect a survival benefit for women on RASb in agreement with other observational studies.9, 11, 16 Keyhan et al reported a greater risk reduction associated with RASb for men than for women.11 In contrast, we found a somewhat higher risk reduction among women, although not statistically significant.

As in previous studies, we found less RASb exposure among women15, 16, 17, 18 and older patients.10, 36 Whether this constitutes inequity in treatment is unclear, as the gender difference in treatment could be due to a higher prevalence of HFpEF among women.37

Age

RASb exposure decreased with higher age. RASb was associated with better survival in all age groups, but the association weakened with higher age, perhaps partially due to a higher prevalence of HFpEF among older patients.37 Our results are relevant for numerous patients, as the HF population is old. Ageing populations are a reality, notably in middle‐ and high‐income countries. Hence, not only healthy ageing and quality of life are increasingly relevant considering equity but also health economics. HF accounts for 1% to 2% of hospital admissions in high‐income settings. In Sweden and the United Kingdom, 2% of the health care budget goes to HF care.38 Health costs are double in HF patients compared with the general population, largely attributable to institutionalised care. RASb reduces hospitalisations4 and is cost‐effective.39

Strengths and limitations

Our individual level data represent the entire Swedish population. The validity for HF as primary diagnosis in the patient register is 95%.24 In the prescribed drug register, RASb registration is around 99% complete. The validity of HF as underlying cause of death has not been studied however, and there is uncertainty concerning the validity of the outcome HF death. Our comorbidity data are limited to inpatient hospitalisation data within 1.5 years preceding the index hospitalisation, and comorbidities were likely underestimated. Notably for HF, the lack of primary care data would lead to underreporting and misclassification bias.40 The prevalence of several comorbidity diagnoses are lower than in some other studies9, 11, 13, 41 although Shafazand et al reported comorbidity levels similar to ours, also from Swedish patient register data.33 Some comorbidities are found in similar proportions for certain groups by Keyhan et al.11

As for misclassficiation, we believe that our cohort was actually less affected by some comorbidity, eg, hypertension, as we excluded patients with a recent RASb dispensation, and RASb is a treatment for hypertension. Renal dysfunction is a potential confounder in our study, being associated with increased HF mortality as well as decreased RASb exposure. Gasparini et al have shown that only 12% of patients with chronic kidney disease are recognized and diagnosed in primary and secondary care.42 Underestimating renal dysfunction could introduce bias towards overestimating the association between RASb and survival. Previous studies of HF patients report renal dysfunction among 1.3% to 56%, and mostly around 15% to 30%.9, 11, 13, 34, 43, 44 Those studies vary considerably regarding exhaustivity of data in time, register vs clinical data, hospitalisation data vs other care levels, de novo vs previous HF, previous RASb use, population age, renal dysfunction definitions, exclusion by serious renal dysfunction, etc. Our prevalence of renal dysfunction was 8.5% overall, and 13% among patients exposed to RASb vs 5.2% among those unexposed. Our cohort resembles a de novo HF population and an RASb‐naïve one and could be less affected by comorbidity including renal dysfunction than others. Furthermore, the relation between renal dysfunction and mortality in HF is graded; the more advanced renal dysfunction, the higher the mortality.43 Clinically significant, ie, more advanced, renal dysfunction should be less likely to be underreported in the patient register. Patients with comorbid heart disease were more likely to be diagnosed with chronic kidney disease in the study by Gasparini et al. The overall HR for patients with renal dysfunction in our study was 0.78 (data not shown), which is close to the overall HR for the whole cohort of 0.72. While we cannot refute misclassification bias in comorbidity, the consistency with other studies in overall mortality and RASb‐associated mortality reduction also supports the overall validity of our results.

A limitation shared with other register studies is the lack of characterisation of HFrEF/HFpEF, as the patient register does not contain data on ejection fraction (EF). RASb has not been shown to increase survival in HFpEF patients and is not recommended for HFpEF in guidelines. We were unable to validate true eligibility for RASb.

Survival bias in pharmacoepidemiology typically leads to overestimation of effects.45 A strength of our study is that we addressed this by using a time‐dependent exposure covariate.

Interaction was estimated on both the multiplicative and the additive scales and risks calculated for each exposure category with a common reference, as recommended in epidemiological literature.28 While one study suggested that the RERI is valid only for risk factors and not preventive exposures,46 a later publication concluded that it may be used for preventive factors when the sample size is sufficient (more than 1000 subjects).47 We thus regard our RERI estimates as valid.

Obviously, this observational study cannot assert causation.

CONCLUSIONS

We suggest that RASb is effective in hospitalised female and male HF patients of different ages, including very old patients. We found no evidence for a differential effect by gender. Exposure to RASb was associated with decreased mortality in all studied patient groups, although somewhat less in older patients. Yet, RASb exposure was low, and more pronounced so among older and female patients. There is a room for improvement in adherence to treatment guidelines. Further studies including data on EF would expand knowledge regarding equity in RASb treatment.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

ETHICS STATEMENT

The study conforms to the declaration of Helsinki and was approved by the Swedish Central Ethical Review Board (reg. no. Ö 29‐2011).

Figure S1. Directed acyclic graph showing assumptions for the association between measured and true comorbidity and the association of comorbidity with exposure (RASb) and outcome (death). RASb, renin‐angiotensin system blockade

Click here for additional data file.