Article Text


Original article
The unnatural history of an atrial septal defect: Longitudinal 35 year follow up after surgical closure at young age
  1. Judith A A E Cuypers1,
  2. Petra Opić1,
  3. Myrthe E Menting1,
  4. Elisabeth M W J Utens2,
  5. Maarten Witsenburg1,
  6. Wim A Helbing3,
  7. Annemien E van den Bosch1,
  8. Mohamed Ouhlous4,
  9. Ron T van Domburg1,
  10. Folkert J Meijboom5,
  11. Ad J J C Bogers6,
  12. Jolien W Roos-Hesselink1
  1. 1Department of Cardiology, Erasmus MC, Rotterdam, The Netherlands
  2. 2Department of Child and Adolescent Psychiatry and Psychology, Erasmus MC-Sophia, Rotterdam, The Netherlands
  3. 3Department of Pediatric Cardiology, Erasmus MC-Sophia, Rotterdam, The Netherlands
  4. 4Department of Radiology, Erasmus MC, Rotterdam, The Netherlands
  5. 5Department of Cardiology, University Medical Center Utrecht, Utrecht, The Netherlands
  6. 6Department of Cardio-thoracic Surgery, Erasmus MC, Rotterdam, The Netherlands
  1. Correspondence to Dr Judith A A E Cuypers, Department of Cardiology, Erasmus University Medical Center, Thoraxcenter, Room BA 308, s'Gravendijkwal 230, Rotterdam 3015 CE, The Netherlands; j.cuypers{at}


Objective To describe the very long-term outcome after surgical closure of an atrial septal defect (ASD).

Design Longitudinal cohort study of 135 consecutive patients who underwent surgical ASD repair at age <15 years between 1968 and 1980. The study protocol included ECG, echocardiography, exercise testing, N-terminal prohormone of brain natriuretic hormone, Holter monitoring and cardiac MRI.

Main outcome measures Survival, major events (cardiac reinterventions, stroke, symptomatic arrhythmia or heart failure) and ventricular function.

Results After 35 years (range 30–41), survival status was obtained in 131 of 135 patients (97%): five died (4%), including two sudden deaths in the last decade. Fourteen patients (16%) had symptomatic supraventricular tachyarrhythmias and six (6%) had a pacemaker implanted which was predicted by early postoperative arrhythmias. Two reoperations were performed. One ischaemic stroke occurred. Left ventricular (LV) and right ventricular (RV) ejection fractions (EF) were 58±7% and 51±6%, respectively. RVEF was diminished in 17 patients (31%) and in 11 (20%) the RV was dilated. Exercise capacity and quality of life were comparable to the normal population. No clear differences were found between ASD-II or sinus venosus type ASD.

Conclusions Very long-term outcome after surgical ASD closure in childhood shows good survival and low morbidity. Early surgical closure prevents pulmonary hypertension and reduces the occurrence of supraventricular arrhythmias. Early postoperative arrhythmias are predictive for the need for pacemaker implantation during early follow-up, but the rate of late pacemaker implantation remains low. Although RVEF was unexpectedly found to be decreased in one-third of patients, the functional status remains excellent.

  • Congenital Heart Disease
  • Cardiac Function

Statistics from


While transcatheter closure with a device is the current method of choice for correction in patients with a suitable atrial septal defect (ASD), surgical closure was first choice treatment until the 1990s and is still the only treatment for large and non-centrally located defects. After successful operation at a young age, the natural history of patients with ASD improved dramatically with life expectancy similar to that of the general population.1–3

The retrospective study of Murphy et al1 was the first to demonstrate the benefits of early surgical closure. The incidence of postoperative atrial arrhythmias appeared to be related to the age of the patient at the time of repair. Benefits of closure in childhood have been confirmed by others,3–7 although persistent ventricular dilation has been reported in some but not all studies.

Early and mid-term follow-up of transcatheter ASD closure shows excellent results.8 For a comparison of both closure techniques, information on outcome beyond 30 years is crucial.

The aim of the present study was to provide data on mortality, morbidity and ventricular function up to 40 years after surgery. The investigation is based on a unique longitudinal cohort of consecutively operated patients who we examine in hospital every 10 years using the same protocol.


Study population

All patients who underwent surgical closure of secundum (ASD-II) or sinus venosus type ASD in our institution between 1968 and 1980 at an age <15 years were included in this study. The first follow-up study was performed in 1990, the second in 2000–2001 and the current third study in 2010–2011. Detailed information on the surgical procedure and earlier results have been published previously.3 ,5 Current patient survival status was obtained from the Dutch National Population Registry. All living patients who participated in one or both of the earlier studies were approached for participation in the current study. The in-hospital cardiac examination included medical history, physical examination, standard 12-lead ECG, 24-h ambulatory ECG (Holter), echocardiography, bicycle ergometry with oxygen consumption measurement, N-terminal prohormone of brain natriuretic hormone (NT-pro-BNP), cardiac MRI (CMR) and the 36-item Short Form Health Survey (SF-36) questionnaire. If the patient was unwilling or unable to visit the clinic, a written questionnaire was sent to obtain information on morbidity and permission to use available information from their medical records.

Major events

Survival was compared with that of the normal age-matched Dutch population. Major events were defined as cardiac reinterventions, stroke, symptomatic arrhythmia, heart failure or endocarditis. Arrhythmias were defined as symptomatic if antiarrhythmic medication was prescribed, cardioversion or catheter-based or surgical ablation had been applied, or pacemaker implantation was performed.

ECG and Holter monitoring

Standard 12-lead surface ECGs and Holter recordings were analysed as described previously.3 ECGs with pacemaker rhythm were excluded for comparison of conduction times. Heart rate variability (HRV) was compared with that of 10 years earlier using the SD of all normal RR intervals (SDNN).


Complete two-dimensional (2D) and colour flow Doppler imaging as well as pulsed-wave and continuous-wave Doppler echocardiography was performed using an iE33 xMATRIX X5-1 echocardiograph (Philips Medical Systems, Best, The Netherlands). All up-to-date techniques and definitions were used following the current guidelines.9–15 For right ventricular (RV) function the following parameters were used: fractional area change (FAC) (normal if ≥35%), S′ of the tricuspid annulus (normal if ≥10 cm/s) and tricuspid annular plane systolic excursion (TAPSE) (normal if ≥16 mm). For left ventricular (LV) function, 2D ejection fraction (EF) according to the Simpson rule was obtained. For comparison with the two previous studies, ventricular function was also assessed visually. Dilation was defined as left atrial (LA) dimension >45 mm and LV end-diastolic dimension >58 mm.3 ,5

Bicycle ergometry

Maximal exercise capacity and maximal oxygen consumption (Vo2 max) were assessed by bicycle ergometry with gradual increments of workload of 20 Watts/min. Exercise capacity and peak VO2 were compared with that of normal individuals corrected for age, gender, body height and weight. Exercise capacity and peak Vo2 <85% of the predicted value were considered to be decreased. Performance was considered maximal when a respiratory quotient (RER) of ≥1.1 was reached.


After a 30 min rest, venous blood samples were collected for determination of NT-pro-BNP. Standard kits to determine NT-pro-BNP levels were used (Roche Diagnostics, Basel, Switzerland) with a cut-off value for elevation of 14 pmol/L.16

CMR imaging

CMR imaging was performed using a Signa 1.5 Tesla whole body scanner (General Electric Medical Systems, Milwaukee, Wisconsin, USA) using dedicated phased-array cardiac surface coils. Details of the MR sequence used have been reported previously.17 For CMR analysis, a commercially available Advanced Windows workstation (GE Medical Systems) was used, equipped with Q-mass (V.5.2, Medis Medical Imaging Systems, Leiden, The Netherlands). The ventricular volumetric data set was quantitatively analysed using manual outlining of endocardial borders in end-systole and end-diastole. Biventricular end-diastolic volume (EDV), end-systolic volume (ESV), SV, EF and the regurgitation fractions of the valves were calculated. The results were compared with reported normal values: RVEF ≤49% and LVEF ≤54% were considered decreased and RV EDV >107.5 mL/m2, RV ESV > 47.2 mL/m2, LV EDV >102.5 mL/m2 and LV ESV >38.7 mL/m2 were considered enlarged.18

Health status assessment and family history

The results of the SF-36 questionnaire were compared with those in the normal Dutch population.19 Patients were also asked for the presence of a congenital heart defect in one of their first- or second-degree relatives.

Statistical analysis

Continuous data are presented as mean±SD unless indicated otherwise. Categorical variables are represented by frequencies and percentages. Comparison of continuous variables between independent groups was made by Student t tests. In the case of a skewed distribution in paired groups, the Wilcoxon signed rank test was performed. When comparing frequencies, the χ2 test or Fisher exact test was used where applicable, and for paired categorical data the McNemar test. For quantifying associations between two variables the Pearson correlation test was applied. Univariate and multivariate analysis of predictors of survival and major events was performed with the Cox proportional hazards model. A priori the following variables were selected: age at operation, ASD subtype, preoperative shunt size, cardiopulmonary bypass time and postoperative arrhythmia. HRV in 2001 was tested as a predictor for arrhythmias thereafter. The probability of survival and event-free survival over time was displayed as a Kaplan–Meier plot. The level of significance was chosen at p<0.05 (two-sided).



Information on survival status was obtained in 131 of the 135 patients. Four patients were lost to follow-up before 2001. During the total follow-up period, five patients died (4%). Two patients (both ASD-II) were found dead in bed 28 and 36 years after surgery. In neither of the patients was an autopsy performed. One of them had complained of palpitations 2 weeks before he died but he had not consulted a physician. The other was known to have diminished LV function attributed to longstanding RV pacing. Both patients had no risk factors for coronary disease. The other three patients died from non-cardiac causes (cancer in two, suicide in one). Cumulative survival after ASD surgery was 100% after 10 and 20 years, 98% after 30 years and 91% after 40 years (figure 1). This is comparable to the normal Dutch population.20 Survival in the ASD-II and sinus venosus subgroups was not significantly different.

Figure 1

Survival and survival free of major events (defined as reoperation, arrhythmia, heart failure, endocarditis, pacemaker implantation, hospitalisation and death). All-cause mortality, n=5; cardiac surgery, n=2; stroke, n=1; symptomatic arrhythmia, n=16. Survival of the normal population is shown from the age of 7.5 years, the mean age at operation of the cohort. ASD, atrial septal defect.

Study population

The original study cohort consisted of 135 patients (105 ASD-II and 30 sinus venosus type). Table 1 demonstrates their characteristics. Age at operation was 7.5±3.5 years (range 0–14). There was no in-hospital or early mortality. Of the 105 eligible patients (alive and participating in the earlier studies), 85 (81%) agreed to participate in the third follow-up study after a mean follow-up of 35 years (range 30–41). Of the 21 patients with sinus venosus defect, 19 had an anomalous pulmonary venous drainage; one patient had a persistent left superior caval vein.

Table 1

Patient characteristics

Major events

Event-free survival is plotted in figure 1.


Before 2001 one additional operation was performed for closure of a patent arterial duct. Between 2001 and 2011, one cardiac reoperation was necessary for symptomatic mitral valve regurgitation. Both operations were in patients with ASD-II.


Until 2001, 10 patients had symptomatic arrhythmias and, in the last 10 years, six additional patients developed symptomatic arrhythmias. One showed frequent symptomatic sinus arrests necessitating permanent pacemaker implantation. Three were treated medically for atrial flutter (n=1) or atrial fibrillation (n=2), one needed electrical cardioversion for atrial flutter and afterwards underwent catheter ablation and one patient was treated medically for supraventricular tachycardia (SVT). During the total follow-up period 16% of the patients developed symptomatic arrhythmias.

Heart failure

None of the patients developed heart failure.


An ischaemic stroke occurred in one patient with sinus venosus type ASD. No evidence of residual shunt or arrhythmia was found.

ECG and Holter monitoring

Twelve-lead ECG data are presented in table 2. Atrial flutter was found in two patients, of whom one was new. The QRS duration increased significantly during the last 10 years, but only one new case of right bundle branch block was found.

Table 2

Standard 12-lead ECG, 24-h Holter, bicycle ergometry and echocardiographic parameters in 1990, 2001 and 2011

On the 24-h Holter no sustained ventricular tachycardia (VT) was found in 1990, 2001 or 2011 (table 2). There was a trend towards more supraventricular arrhythmias over time. Atrioventricular (AV) conduction disturbances were observed in nine patients.

Heart rate variability

The mean SD of all normal RR intervals (SDNN) was 166.7±57.9 ms in 2001 and 143.4±42.0 ms in 2011 (p=0.2). There was no difference in HRV between patients with ASD-II and sinus venosus types.


Echocardiographic findings are summarised in tables 2 and 3. No residual shunts were found. A normal diastolic LV function was observed in 88% of patients, mild abnormal relaxation in 9% and pseudonormal diastolic filling in 3%. TAPSE was decreased in 22% of the patients, FAC was reduced in 10% and S′ was diminished in 43%. Comparing patients with ASD-II and sinus venosus defects, there were no significant differences except that S′ and TAPSE were significantly lower in patients with sinus venosus type whereas RV FAC was higher.

Table 3

Echocardiographic parameters 2011

Bicycle ergometry

The results of exercise testing in 1990, 2001 and 2011 are listed in table 2. Eighteen patients (27%) had a diminished exercise capacity at last follow-up. Mean peak Vo2 was 96±25% (range 60–194%). One-third had a mean peak Vo2 of <85%, one-third 85–100% and one-third performed >100%. The threshold RER of 1.1 was reached in 97%. Patients with reduced exercise capacity or peak Vo2 did not differ significantly from those with a normal test result with regard to preoperative shunt size, age at the time of operation or findings by echocardiography or CMR (dimensions and ventricular function). There were no significant differences between patients with ASD-II and sinus venosus defects but there was a trend towards a better workload in the patients with sinus venosus type ASD (p=0.07)


An increased NT-pro-BNP level (>14.0 pmol/L) was present in 53% of the patients. Values varied from 1.5 to 116 pmol/L with a mean of 15.0 pmol/L. No correlation between NT-pro-BNP and age, age at surgery, ASD type or LV and RV dimensions or function was found.

CMR imaging

CMR was performed in 57 patients. Reasons for not having a CMR investigation were patients’ refusal, prematurely ended because of claustrophobia and two patients had a pacemaker. One study could not be analysed due to technical problems. The results of CMR are summarised in table 4. There were no differences in mean ventricular volumes or EFs between patients with ASD-II and sinus venosus type defects. However, there was a trend towards more dilated RV volumes in the sinus venosus group (dilated ESV in 57% vs 27%, p=0.054).

Table 4

Cardiac MRI

Health status assessment and family history

The mean scores on the SF-36 survey for the patients and the normal population are shown in figure 2. Having had an event did not adversely affect SF-36 scores.

Figure 2

Results of the Short Form Health Survey questionnaire for patients with both types of atrial septal defect (ASD) and the normal Dutch population. a=significant difference between patients with ASD-II and sinus venosus type ASD; b=significant difference between patients with ASD-II and the normal Dutch population; c=significant difference between patients with sinus venosus type ASD and the normal Dutch population. Higher scores reflect higher levels of functioning or well-being.

A self-reported family history of congenital heart disease was found in 23% of patients (n=16); of these, 56% were in first-degree relatives. The majority of family members also had an ASD (n=8). Other defects were a ventricular septal defect (n=3), patent ductus arteriosus (n=2), univentricular heart (n=1) and valvular heart disease (n=2). Some had more than one defect. Information on genetic testing was not recorded.

Predictors for late events

Univariate analysis identified only the occurrence of arrhythmias in the early postoperative period as a predictor for the need for permanent pacing during follow-up (HR=16 (95% CI 3 to 70); p<0.0001). Multivariate analysis was not performed because the number of events was too low.


In this cohort of patients who underwent surgical closure of a haemodynamically significant ASD 30–41 years ago we found very good survival, no pulmonary hypertension, good exercise performance and a low incidence of supraventricular arrhythmias.

Total survival was comparable to the normal Dutch population. However, two patients in the ASD-II group suffered from sudden death in the last decade at the age of 36 and 40 years. In these patients arrhythmia as the cause of death cannot be excluded and is even very likely in one.

Signs of pulmonary hypertension, a major problem that may occur in patients without closure of the ASD,1 ,21 were not found. In addition, the clinical condition of the patients appeared excellent with normal exercise capacity in most patients. Neither heart failure nor residual shunts were observed. One patient developed an ischaemic stroke but a clear relation to the heart defect could not be found. However, 6% of patients required pacemaker implantation during total follow-up.


Of the 41% (n=36) of patients with some form of asymptomatic atrial arrhythmia on Holter recording 10 years ago,3 8% had developed new symptomatic arrhythmias in the past 10 years. This was comparable to the incidence in patients with a negative Holter recording in 2001. Thus, the predictive value of the asymptomatic arrhythmias seen on Holter monitoring seems limited. The prevalence of atrial fibrillation in our patients (12%) is higher than in the general population (<0.5%)22 but substantially lower than that reported in natural history studies of patients with ASD (up to 50%) and after surgical closure at adult age.1 ,4 ,6 ,7 ,21 ,23 This indicates that early closure is beneficial in the very long term. Surgical scars do not seem to induce atrial flutters on a large scale. The lower incidence of atrial fibrillation appears to be related to earlier termination of the left-to-right shunt, thereby preventing right-sided pressure overload and further right atrium dilation which would have increased the vulnerability to atrial arrhythmias.7 ,24 Whether device ASD closure will also prevent the occurrence of SVTs in the long term or will make patients prone to arrhythmias by direct stretching of the atrial septum or circular scar formation around the device remains to be established.

Ventricular dimensions, function and pressure

On echocardiography we found a mildly impaired RV function in one-third of our patients. This is somewhat unexpected as RV function was normal in all patients 10 years ago. This finding may be due to advances in the assessment of RV function as nowadays we use more sophisticated parameters than the mere visual assessment which was the only available method used in our earlier studies. We found no association between RV function and changes in QRS duration. The findings of FAC and S′ wave as well as the CMR results confirm abnormal RV function in a quarter of our patients. CMR was not performed in our earlier studies but RV dilation has also been described by others after shorter follow-up. De Koning et al25 described this in younger patients who had their surgical correction in a more recent era. In contrast to others and our results, no RV dilation on CMR was reported by Bolz et al even though the same normal values were applied and the mean age at operation is comparable (6.5 vs 7.6 years).26–28 However, their follow-up period was shorter.

Mild LV function impairment was found in one-fifth of our patients on CMR. This might be a result of the preoperative volume overload due to ventricular interaction. Literature on LV function after surgical ASD repair is lacking, but there are some reports of improvements of LV systolic function after transcatheter ASD closure.29 ,30

ASD-II versus sinus venosus type ASD

We did not find any significant difference in survival, occurrence of events, incidence of arrhythmias, ECG (QRS duration) and Holter-ECG or exercise capacity between patients undergoing surgery for ASD-II and sinus venosus defects. However, TAPSE and S′ were slightly reduced on echocardiography in patients with sinus venosus defects. On the other hand, RV FAC was higher in this group. Integrating all these results, patients with a sinus venosus defect may have a worse longitudinal RV function with probably a compensatory preserved circumferential function. New studies with larger patient groups are needed to confirm these findings.

Predictors for late events

The occurrence of arrhythmias in the early postoperative period showed a strong relation with the need for permanent pacing during early follow-up. Most pacemakers had been implanted in the early years after surgery, most are probably related to surgical damage to the conduction system for two of the three patients (66%) requiring pacemaker implantation within 4 years after surgery had a bradycardia as their postoperative arrhythmia. However, two patients needed a pacemaker for AV block very late after surgery. This is an unexpected finding for which we do not have a clear explanation.

We were not able to identify any other predictors for mortality, late events or ventricular dysfunction.

Study limitations

As in most studies in this field, the numbers are relatively low and therefore all results should be interpreted with caution. The number of patients lost to follow-up was limited with regard to information on survival status, but is larger considering study participation: only 65% of the survivors of the total cohort participated in the last follow-up. The patients considered ineligible for participation were either lost to follow-up, had moved far abroad or did not participate in the previous studies. We believe, however, that after a follow-up of almost 40 years, participation of 80% of eligible patients is very acceptable and can be regarded as a representative sample.


Very long-term (30–41 years) outcome after surgical ASD closure in childhood shows excellent survival and low morbidity, although two sudden unexplained deaths occurred at 28 and 36 years after surgery. No pulmonary hypertension and a low incidence of supraventricular arrhythmias were found. The general health and exercise capacity of the patients are excellent and comparable with the normal Dutch population. Persistent RV dilation, which was observed in our earlier follow-up studies in this cohort of patients, was confirmed by CMR. Although EFs of both RV and LV are unexpectedly decreased and RV volumes remain enlarged in some of the patients, their clinical condition remains sound.


We thank Jackie McGhie, Marian Blüm and Ellen Wiegers for their excellent echocardiographic assistance and Celeste Manley and Tineke van der Kolk for their valuable administrative work.


View Abstract


  • Contributors JAAEC: executive investigator, drafting the article, design and execution of the study, data gathering, data entry, statistical analyses, interpretation of results. PO: data gathering, data entry, statistical analyses, interpretation of results, revising the article. MEM: data gathering, data entry, statistical analyses, interpretation of results, co-drafting and revising the article. EMWJU: concept and design, interpreting the results, revising the article. MW, WAH, AEvdB, MO: scientific input, data gathering, revising the article. RTvD: supervision of statistical analyses, revising the article. FJM: scientific input, data gathering, concept and design, revising the article. AJJCB: concept and design, scientific input, revising the article. JWR-H: principal investigator, concept and design, data gathering, interpreting the results, revising the article.

  • Funding This research is financially supported by a grant from the Dutch Heart Foundation, Den Haag, The Netherlands (grant number 2009-B-073).

  • Competing interests None.

  • Ethics approval Erasmus MC Medical Ethical Committee approved the study (METC number 2010-015).

  • Patient consent Obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.