In a prospective registry that included all consecutive patients referred to our syncope unit, we have previously described the diagnostic tests performed to exclude a cardiac cause.10 All the patients included in this study had a previous syncope. In the syncope unit of our hospital, we use a standardized protocol following the recommendations in the European guidelines on syncope.1,2,11 Patient enrolment started on April 1, 2006 and ended on January 31, 2018.

With the aim of achieving as close a follow-up as possible in these 1058 patients, for this retrospective analysis we included only those patients referred to our unit who were resident in our health care area. Cardiac etiology was always excluded by the patient's treating physician using the standard diagnostic tests that they considered necessary (that is, echocardiography, stress testing, perfusion imaging, electrophysiology study, coronary angiography, or implantable Holter). All tests were reviewed in the unit and have been reported previously.10 The patients were then referred to our unit for further assessment including a head-up tilt test (supplementary data). All patients gave written informed consent. As this was a registry and not an interventional study, we did not request hospital ethics committee approval.

To ensure that follow-up would be at least 6 months in all patients, follow-up started on April 1, 2006 and ended on July 31, 2018. A prespecified protocol was used for clinic follow-up visits, telephone calls, and review of electronic medical records: all patients had at least 1 clinic visit and annual telephone follow-up. In all patients, the data recorded at each follow-up visit included: presence or absence of recurrence and date, number of recurrences, death and its cause, injuries, need for device-based treatment, hospital admissions, and emergency department visits. The electronic medical records in our public health system helped achieve and complete 100% adherence to follow-up, as we were able to check the visits to the emergency department if the patient told us about them in clinic or over the phone. For patients who died at home or of an unknown cause, we contacted their families to assess the probability of sudden cardiac death.

Qualitative variables are presented as absolute and relative frequencies, and quantitative variables are presented as mean ±standard deviation or median [interquartile range] depending on the distribution of the variable, evaluated using the Kolmogorov-Smirnov test. Qualitative variables were compared using the chi-square test or Fisher exact test, and quantitative variables were compared using the Student t test, Mann-Whitney test, or Kruskal-Wallis test, as appropriate. To assess the event-free survival time (recurrent syncope, death), Kaplan-Meier survival analysis using the log-rank test (Mantel-Cox) was used to compare the different groups classified according to the final diagnosis reached in the unit, which was that used for follow-up analysis. The time was calculated from the date the patient was seen in the syncope unit until the date of the first recurrence. Cox proportional hazards regression models were used to analyze the risk of recurrence of syncope and of death within each specific etiology (1, VVS; 2, OHS; 3, SUE). All clinical variables were included to assess their possible relationship with outcome. Receiver operating characteristic (ROC) curve analysis was used to predict syncope recurrence as an outcome. Analysis was performed using the SPSS V24.0 program. P values < .05 were considered statistically significant.

We were able to complete 100% follow-up of the population. After a median follow-up of 52 [28-89] months, 220 patients (37.4%) had at least 1 recurrence of syncope, with a time to first recurrence of 46 [21-82] months. The median number of recurrences was 2 [1-3]. The annual cumulative probability of recurrence was 8.7%. In 42% of the patients who had recurrence, the recurrence was an isolated episode, while the other 58% had 2 or more episodes of recurrence (figure 1). The cumulative probability of syncope recurrence was higher in women than in men (P = .001), and women also had a higher number of recurrences (P=.0001) (table 1).

Figure 1.

Histograms of patients grouped by number of syncopal episodes experienced during follow-up.

(0.07MB).

The recurrence-free survival for the 220 patients who had recurrence was 82.2 (95% confidence interval [95%CI], 78-86.4) months. We identified different cutoff values for prediction of recurrence. According to the ROC analysis, 5 previous episodes gave the best sensitivity (0.605) and specificity (0.661), with an area under the curve of 67.2%±0.2%. The cumulative probability of recurrence was lower in patients with fewer than 5 episodes than in those with 5 or more previous episodes: 26.3% vs 51.6% (P=.0001) (figure 2).

Figure 2.

Cumulative probability of recurrence in patients with fewer than 5 previous syncopal episodes (blue line), syncope-free survival time, 94.3 (95%CI, 89-99.6) months, compared with those with 5 or more (red line), syncope-free survival time, 67.2 (95%CI, 61.2-73.2) months (log-rank test,P <.0001). 95%CI, 95% confidence interval.

(0.19MB).

The number of patients with recurrence of syncope was higher in the VVS group than in the OHS or SUE groups (P=.006). There were also differences in the number of recurrences of syncope depending on the etiology (P=.001) and in recurrence-free survival time (log-rank test,P <.001) (table 1, figure 3A). Factors identified as independent variables associated with recurrence of syncope were age (hazard ratio [HR],0.65; 95%CI, 0.49-0.86; P=.002), female sex (HR,2.81; 95%CI, 1.9-3.9; P <.0001) and the number of previous episodes (< 5 vs ≥ 5, HR,2.98; 95%CI, 2.1-4.2; P <.0001).

Figure 3.

A: cumulative probability of recurrence of syncope in patients with VVS, syncope-free survival time, 78.8 (95%CI, 73.9-83.4) months (blue), compared with those with OHS, syncope-free survival time, 84.3 (95%CI, 72.8-95.7) months (red) and SUE, syncope-free survival time, 92.4; (95%CI, 83.2-101.6) months (green) (log-rank test, P <.001), B: cumulative probability of mortality in patients with VVS, mortality-free survival time, 121 (95%CI, 118.4-123.5) months (blue) compared with those with OHS, mortality-free survival time, 102.32 (95%CI, 91.9-112.5) months (red) and SUE, 108.7 (95%CI,101.2-116.2) months (green) (log-rank test, P <.0001). 95%CI, 95% confidence interval; OHS, orthopedic hypotension syncope; SUE, syncope of unknown etiology; VVS, vasovagal syncope.

(0.25MB).

There were 39 deaths (6.6%), with an annual cumulative probability of death of 1.53%, with no differences between men and women (P=.25). Only 4 deaths were due to cardiovascular causes and none were sudden cardiac death. There was a significant difference in age between those who died and those who survived (P <.0001). Patients who died had a higher cumulative probability of hypertension (P=.021) and diabetes (P=.0001); however, they had fewer recurrences of syncope (P=.004) (table 2). Patients with recurrence of syncope had a lower cumulative probability of death (P=.004), and age and diabetes were associated with mortality (table 3).

The median time until death was 51 [27-85] months, with no differences between men and women (P=.19). Survival time was 117 (95%CI, 114.2-119.7) months. When analyzed by etiology, there were differences in mortality (P=.0001) and event (death)-free survival (log-rank test [Mantel-Cox], P <.0001) (table 1, figure 3B). On multivariate analysis, the diagnoses of OHS and SUE were identified as independent markers associated with mortality (table 3).

The annual syncope recurrence rate was 8.7% over a follow-up of 4.3 years, and recurrences occurred in 37.4% of the patients; 76.6% of these recurrences occurred in patients with VVS and OHS, findings that are similar to those published by other authors,4,15,16 although it is important to note that our population had a high cumulative probability of syncope recurrence and, therefore, was very symptomatic. This is in contrast to the population cohort with syncope in the Framingham study, who had a much lower cumulative probability of recurrence, of 21.6% or 1.27% per year after a mean follow-up of 17 years. In addition, it was reported that 17.6% of patients with recurrence had only 1 episode of recurrence and only 4% had more than 1.5 The recurrence-free survival time in the 220 patients who had recurrence was certainly long, at 82.2 (95%CI, 78-86.4) months. In previous studies by our group this time was shorter, at 50.1 (95%CI, 46-54) months, but the data related to the whole population included in the study, not only those with recurrence.4

The number of syncope recurrences was higher than that previously reported by other investigators, albeit with shorter follow-up times.7 Our findings indicate that one third of the patients sent for further assessment for syncope may need targeted therapies to avoid recurrence, and we believe that future research in this field should be encouraged.

Past syncope burden has been identified in other studies as a marker of future recurrence.4,6,17 In our population, having 5 or more syncopal episodes prior to inclusion in the registry was independently associated with recurrence. Another interesting factor associated with recurrences during follow-up was sex, as confirmed by other authors.4,6,16,18 In our series, women had a higher cumulative probability of recurrence and a risk of recurrence 3 times higher than that of men.

We found slightly higher mortality in men, driven mainly by noncardiac causes. The reasons for this finding remain undetermined. However, it is plausible that syncope, especially if it is recurrent and not of vasovagal etiology, may simply be a marker of frailty that identifies older patients with multiple comorbidities who are at risk of mortality from all causes. The low mortality and inverse association between recurrence of syncope and mortality would suggest that the relationship between syncope and mortality should be established with caution. Few studies have analyzed the cumulative probability of mortality in outpatients with noncardiogenic syncope who have not been admitted to hospital. Yasa et al.8 analyzed a prospective cohort based on a population of 30 528 middle-aged individuals who were discharged with a diagnosis of unexplained syncope or OHS, although they had a higher incidence of cardiovascular disease and mortality. It should be noted that this population, although they were the same age (62±7 vs 63±7 years), had been admitted to hospital for syncope. In a series of patients with orthostatic hypotension, Naschitz et al.19 reported a higher risk of death probably related to the severity of comorbidities, which was double that of the total study population. However, this association has not been reflected in other studies.5,20 In VVS, a lower mortality was found, which is in line with previous publications and supports the clinical observation included in the European guidelines that recurrent episodes are a marker of benign outcome.1,2,9

The role of age in mortality has previously been described in patients with syncope of different etiologies, but not specifically in patients with noncardiogenic syncope.21 Diabetes was independently associated with mortality in a study of patients admitted to hospital due to syncope; over a follow-up of 27 months, the cumulative probability of mortality was 12% and the factors independently associated with mortality were diabetes (HR,2.7; 95%CI, 1.4-5.2), smoking (HR,2.8; 95%CI, 1.4-5.5), and atrial fibrillation (HR,2.4; 95%CI, 1.9-5.4).19 Our finding is in agreement with a univariate meta-analysis that reported that the effect of noncardiac or unexplained syncope on mortality increases with age and prevalence of diabetes and hypertension.9

Finally, SUE was independently associated with mortality, as Ungar et al.20 already described in patients admitted to hospital for syncope (odds ratio [OR],4.6; 95%CI, 1.002-21.727; P <.059). In a population-based study, Sule et al.22 also reported the diagnosis of SUE as an independent predictor of mortality (HR,1.32; 95%CI, 1.09-1.60; P <.01). Likewise, in our study, OHS had a higher cumulative probability of mortality, an increase of almost 3 times that with VVS; this association was previously described by Masaki et al.,23 who reported a high mortality in the population with OHS and proposed the frailty of this population as driving the possible relationship with mortality. While it is true that the association between SUE and OHS with mortality may not be causal, it could be a marker (more than a factor) of risk of general (not cardiac) mortality. A study more in line with this objective should be designed to properly evaluate this.

Perhaps the most important conclusion from our descriptive study comprising 12 years of patient inclusion is that it has demonstrated the usefulness of syncope units—extremely common in other countries—for the assessment and management of this population, yet there are few in Spain. The very low cardiac mortality observed demonstrates that the initial diagnostic workup aimed at excluding cardiac syncope was appropriate, highlighting the importance of a good initial assessment and the enormous usefulness of these units.

Our study has several limitations: one is the prospective inclusion of all consecutive patients referred to our syncope unit, which indicates a potential selection bias. The inclusion period from 2006 to 2018 introduces an obvious limitation, as during these 12 years there have been substantial changes in the diagnostic workup of patients with syncope. The data analysis was retrospective. This was a single-center study of a sample selected from patients sent to a syncope unit after exclusion of a potential cardiac cause. Another limitation or bias is having selected only the population pertaining to our health care area. It would have been of great interest to include a series of patients with similar characteristics without a history of syncope to compare the outcomes. The diagnosis of cardiac syncope had been excluded through clinical assessment and diagnostic tests performed by the patients’ physicians before they were referred to the syncope unit, so the group will certainly not be homogeneous in terms of the tests performed or their interpretation in the specialized syncope unit. The reproducibility of the head-up tilt test has been questioned, but it is in line with our previous publications and the European guidelines on syncope. An area under the curve of 0.67 with the cutoff of 5 episodes of syncope must be recognized as relatively low. We tried to judge the cause of unknown deaths, but we cannot rule out the possibility of underestimating sudden cardiac deaths. We believe that the use of an implantable Holter monitor in our population with SUE could have improved the identification of the cause of syncope; however, 92.5% had a normal ECG and 95% had no evidence of structural heart disease, so an implantable Holter monitor was probably not indicated in most of them and would have offered very little added benefit to the diagnosis.

In a survey carried out in 2015 on hospitals in the USA, Kachalia et al.24 describe that 83% of patients with syncope underwent an excessive number of diagnostic tests. The excessive use of medical resources for patients with low-risk syncope is increasingly recognized as a problem.25 The findings of our study are applicable to patients referred for further assessment of unexplained syncope to an experienced syncope unit, and may or may not reflect the situation of a population of outpatients seeking medical advice following syncope.