This is an observational, retrospective study of ACS patients treated in the Netherlands. We included patients aged 18 years and older undergoing cardiac catheterisation for ACS in the period 2010–2012. We used data from existing electronic databases in each of the individual interventional cardiac centres. At the time we started our study, 30 interventional cardiac centres existed in the Netherlands. The 24 centres with electronically available clinical data and an established interventional cardiology programme that offered full 24/7 service were asked to participate. For several reasons, 13 eligible centres were unable or unwilling to participate, and thus we present data from a total of 11 centres.

Data were extracted in each individual centre using a pre-specified list of variables, which was mainly based on quality of care indicators [8, 9]. Electronically abstracted data included patient demographics, cardiovascular risk factors and history and characteristics of cardiac procedures. In patients who underwent multiple procedures within the study period, only the clinical information of the first procedure was retained. For the purpose of this study, patients were not contacted, subjected to acts, and neither was any mode of behaviour imposed, otherwise than as part of their regular treatment. Data extracted from the medical records were processed anonymously, not traceable to individual patients. Therefore, in accordance with Dutch law, written informed consent was not required.

To enable data merging of different centres, broad definitions were used. We distinguished STEMI, non-STEMI (NSTEMI) and unstable angina pectoris (UAP) as ACS subclasses, according to the European Society of Cardiology guidelines [10]. Prevalent risk factors including hypertension, hypercholesterolaemia, diabetes mellitus, including insulin dependence, smoking and family history of cardiovascular disease were defined as such in the dataset of the concerning centre. This also applies to prior MI, percutaneous coronary intervention (PCI), coronary artery bypass grafting (CABG) and renal failure. Left ventricular ejection fraction (LVEF) was defined as good (>50%), moderately impaired (30–50%) and severely impaired (<30%). Segment involvement was created using the coronary artery segments mapping on the coronary angiogram according to the American Heart Association classification [11]. Severity of coronary artery disease (CAD) was expressed in number of segments and vessels involved with ≥50% stenosis. Multi-vessel obstructive disease was defined as ≥50% stenosis in left main artery or ≥50% stenosis in ≥2 separate epicardial coronary artery territories.

The availability of specific data items varied between the participating centres. Descriptive statistics were provided for all available data. Inferential statistics (including statistical modelling) was limited to variables that were available for at least 4 participating centres.

Continuous variables were expressed as means with standard deviations and medians with interquartile ranges. Categorical variables are presented as numbers with percentages. We used Mann-Whitney tests (continuous variable) and chi-squared tests (categorical) to evaluate differences between women and men.

Estimates of the cumulative incidence of mortality were obtained by the method of Kaplan-Meier, and differences between women and men were evaluated by the log-rank test. The relationship between sex and 1‑year mortality was further analysed using Cox proportional hazards regression models. We applied univariable and multivariable analyses, with adjustment for age and further adjustment for 1, potential, confounder, including body mass index (BMI), hypertension, hypercholesterolaemia, diabetes mellitus, smoking, family history of CVD, prior MI, prior PCI, prior CABG, LVEF, renal failure, acute MI, STEMI, access site, multi-vessel disease, segment involvement, multi-lesion procedure, multi-vessel procedure and use of drug-eluting stent. Full adjustment for multiple (potential) confounders was not possible as structural missing data precluded multiple imputation. We introduced a sex*age interaction term in the Cox model to study the relationship of sex, age and 1‑year mortality. As results showed a relevant sex*age interaction the clinical endpoints were also presented in 5 strata according to age. Sex differences were analysed in the entire ACS study population and after categorisation for STEMI and NSTEMI/UAP patients.

A total of 8,720 (29.8%) women and 20,545 (70.2%) men with ACS followed by coronary angiography were included. The number of patients included per centre ranged from 1,184 to 5,592. The participating centres had a widespread geographical distribution (Fig. 1).

Women and men had a different age distribution (Fig. 2). Women were on average 4.5 years older (mean age 68.5 vs 63.0 years, p < 0.001) and had a higher prevalence of traditional risk factors for CVD, including hypertension (62.7 vs 49.8%, p < 0.001) and insulin-dependent diabetes mellitus (9.6 vs 6.8%, p < 0.001), but a lower prevalence of current smoking (27.2 vs 36.0%, p < 0.001), family history of CAD (40.2 vs 42.1%, p = 0.037) and history of MI, PCI or CABG. Almost half of the patients had hypercholesterolaemia and about 5% of the patients had a severely impaired LVEF (Tab. 1). Renal failure was comparable between both groups with 7.4 vs 7.1% (p = 0.61).

Procedural data are listed in Tab. 2. A total of 65.1% women and 68.2% men with ACS underwent coronary angiography for acute MI, of which significantly less women (43.7%) than men (47.6%) were treated for STEMI. Fewer women than men underwent coronary angiography by radial access (52.5 vs 55.9%). Women had significantly less extensive CAD with relatively more single-vessel disease compared with men. Most lesions were located in the left anterior descending coronary artery (49.9%), followed by the right coronary artery (41.1%) in both sexes. In women, relatively fewer stents (per treated segment and in total) were implanted than in men (46.2 vs 49.4%). Drug-eluting stents were the most commonly implanted stents in both women and men.

Women had a higher cumulative incidence of all-cause mortality than men (7.3% vs 5.6%, hazard ratio [HR] 1.31 and 95% confidence interval [CI] 1.18–1.45, p < 0.001). After adjustment for age, the observed mortality difference disappeared. However, the sex*age*mortality relation appeared more complex, as we found a statistically significant interaction between age and sex with respect to 1‑year mortality (p < 0.001). At younger age, women had a higher mortality than men, whereas this pattern was reversed at elderly age (Fig. 3). Based on the logistic regression model, the age of 71 years appeared to be the turning point (Fig. 4). After adjustment for age, the sex*age interaction term and single variables, female sex was still associated with higher mortality (data not shown).

Sex-related differences in baseline and the majority of procedural characteristics of STEMI and NSTEMI/UAP patients were comparable with the entire ACS population (Supplementary data—Tables 1–3). Compared with STEMI patients, the risk profile of NSTEMI patients was more unfavourable, except for the risk factor smoking. Women and men with NSTEMI were respectively 1.2 and 3.0 years older, had a 21.2 and 22.8% higher incidence of hypertension, 24.5 and 26.7% higher incidence of hypercholesterolaemia, 10.1 and 8.5% higher incidence of diabetes mellitus and 6.0 and 6.7% more renal failure. The prevalence of current smoking was 16.3% lower for women and 14.5% lower for men with NSTEMI. Both women and men with NSTEMI more frequently had a prior cardiovascular history and had on average a worse LVEF.

In women and men with STEMI, a femoral approach was relatively more often applied than in NSTEMI patients. There were no significant sex-related differences in CAD severity in STEMI patients. However, in NSTEMI patients, women more often had non-obstructive and single-vessel disease.

In all age categories, STEMI patients had higher 1‑year mortality than NSTEMI patients. Women with STEMI had higher 1‑year mortality than men (9.4% and 6.8%; respectively p < 0.001). The difference was most striking at young (≤50 years) age (4.8% and 2.3%, respectively; p = 0.022) and was seen from 30-day follow-up. Women and men with NSTEMI had similar mortality (5.0% and 4.3%, respectively; p = 0.15).

We were able to collect and merge data of ACS patients of 11 experienced interventional centres with a widespread geographical distribution, and thus to construct a large data set that is unique for the Netherlands. The majority of the remaining 13 centres were sympathetic, but were not able to provide data. They had switched to a new electronic patient record system, collected data in free text fields or had no ability to extract data from the data system. Unlike in Sweden, until now Dutch centres have freedom of choosing their own electronic patient record system, which hampers data sharing. Some of the centres understandably refused participation because data were not validated or registered for scientific purposes. It took a disproportionate amount of time and energy to validate and merge datasets, whereas missing data was a serious concern. We recognise that for more detailed analyses—for example analyses of time trends—uniform, prospective patient cohorts are necessary. Clinical registries can be very useful for outcome monitoring and benchmarking with other European countries in order to improve the quality of care. In addition, clinical registries have proven their value in multi-centre observational research, and in registry-based randomised clinical trials [29, 30]. The Netherlands Society of Cardiology recognised that we should no longer lag behind other European countries, and initiated a national ACS registry. We expect the first results within 5 years from now.

Several limitations of our work warrant discussion. The retrospective design represents a serious drawback, but also highlights a sobering fact: a proper, prospective, nationwide ACS registry is still lacking in the Netherlands. We validated data (definitions and outliers) at hospital level, but validation on patient level was not feasible. We are aware of the selection bias that is created by studying only patients admitted to the cath lab, and thus neglecting conservatively treated patients. Remarkably and inexplicably, we found more missing data in women than in men. Due to missing data, we were not able to perform a full multivariable regression analysis. Also, we could not report on details of clinical and procedural characteristics, including angiographic data and time intervals. Especially time intervals would have been very interesting in view of previous described sex differences in the literature and the excellent logistics concerning STEMI care in the Netherlands. All in all, given that 46% of interventional hospitals participated, with satisfactory geographical distribution, and given the large sample size, our data provide a reasonable insight in ACS management, and sex differences in ACS management, in our country. It is reassuring to learn that our findings were similar to other national registries, including the United Kingdom and Sweden.