Long‐Term Trends in Cardiac Erosion After Transcatheter Closure of Atrial Septal Defects With the Amplatzer Septal Occluder

ABSTRACT

Background

Trends in cardiac erosion after closure of secundum atrial septal defects (ASD) with the Amplatzer Septal Occluder (ASO) have not been assessed since updates were made to the Instructions for Use in 2012 and 2018. This study evaluates the incidence and demographics of ASO erosion over time since its introduction in 1996.

Methods

Data were analyzed for all cardiac erosions reported through March 2025 after ASD closure with an ASO device from 1996 to 2024, as adjudicated and characterized by the AGA Erosion Board or Abbott Medical Affairs. The number of implants annually was estimated as the number of units sold in the calendar year. Overall and time‐dependent erosion rates were calculated as the ratio of reported erosions to the estimated number of implants during each calendar year.

Results

As of March 2025, 198 erosions had been reported and confirmed for ASO devices implanted from 1996 through 2024, which corresponds to an estimated erosion rate of 0.037%. Overall, 31% of erosions occurred within 72 h of implant, 55% within 3 months, and 77% within 1 year, while 13% occurred >5 years after implant. The 1‐month, 1‐year, and 3‐year erosion rates were 0.016%, 0.029%, and 0.031%, respectively. The overall erosion rate decreased significantly from 1996 to 2012 to 2013 to 2024 (0.056% to 0.020%, p < 0.001), as did 3‐day, 1‐month, 1‐year, and 3‐year erosion rates (all p < 0.001). The erosion‐related mortality rate decreased from 0.0040% to 0.0014% from the early era to the later era (p = 0.061).

Conclusions

Although limitations in the method of estimating implant numbers and ascertaining erosions must be acknowledged, ASO erosion rates and related mortality have decreased significantly since 2012.

1. Introduction

In December 2001, the Amplatzer Septal Occluder (ASO; Abbott Vascular, Chicago, IL) was approved by the Food and Drug Administration (FDA) for transcatheter closure of secundum atrial septal defects (ASD). The most notable complication of this procedure has been erosion of the device through the left or right atrial wall into the pericardial space and/or the aortic root [1, 2, 3]. Following the first reported case of erosion in 2002 [4], an independent Erosion Board was established to adjudicate and monitor such events [1]. In response to additional reported cases and evaluation by the Erosion Board, AGA Medical Corporation (which manufactured the ASO at the time) issued a Technical Note on January 2, 2006, raising awareness of device erosion and providing new guidance on defect/device sizing, specifically, recommending stop‐flow sizing and advising against implanting a device size > 1.5 times the static ASD diameter.

In January 2012, after discussion with the FDA, St. Jude Medical (the manufacturer of the ASO at that time) made additional modifications to the Instructions for Use (IFU), most notably specifying that implantation of an ASO was contraindicated in patients with echocardiographic evidence of absence or deficiency (<5 mm) of the retroaortic ASD rim. In response to the ongoing occurrence of adverse events, including but not limited to erosion, the FDA convened an Advisory Committee meeting in May 2012 to review the safety of ASD closure devices from all manufacturers. Statements were solicited from multiple professional societies, almost all of which recommended classifying a deficient retroaortic rim as a high‐risk feature rather than a contraindication to ASD closure. On January 17, 2013, St. Jude Medical issued a physician letter summarizing the feedback from the FDA Advisory Committee meeting, along with changes made to the IFU to remove deficient retroaortic rim as a contraindication and to provide additional guidance to further reduce the risk of erosion. The post‐procedure instructions were updated to include recommendations for clinical and echocardiographic follow‐up and instructions for patients to seek immediate medical attention for signs/symptoms of hemodynamic instability. In September 2013, the FDA‐mandated a prospective study to evaluate risk factors for erosion after ASO implant. The mandate for the prospective study was subsequently satisfied by identifying relative risk factors for erosion based on a case‐control study by McElhinney et al. that included 125 erosions reported between 2002 and 2014 matched to 250 controls, which was incorporated into the IFU by Abbott (the manufacturer of the ASO at that time) in 2018 [3].

With the IFU changes in 2012 and subsequent physician letter, ASO erosions became a focus of talks and debates at national meetings and were analyzed in multiple publications [5, 6, 7, 8]. There is evidence that implanter practices also changed in response to this growing awareness, with a propensity to pursue operative ASD closer over transcatheter ASD closure in some patients who were deemed to be at high‐risk for erosion [9]. It has been a decade since ASO erosion data were reviewed [3], and trends in erosion rates after ASO implant have not been re‐assessed during that time. The present study was undertaken to evaluate the incidence and demographics of erosion since the introduction of the ASO device in 1996 through 2024.

2. Methods

All reported erosion events, as adjudicated and characterized by the AGA Erosion Board or Abbott Medical Affairs for ASO implants performed between 1996 and 2024, with follow‐up through March 31, 2025, were included, ensuring a minimum event occurrence window of 3 months for 2024 implants. The AGA Erosion Board was established in 2004 and independently adjudicated all suspected erosion events through the end of 2015. Following satisfactory completion of the FDA‐mandated prospective study to evaluate risk factors for erosion, the Erosion Board was discontinued, and all subsequent events were adjudicated by Abbott Medical Affairs. Adjudication was based on all available objective clinical evidence, which included surgical findings, imaging studies, autopsy results, presence of an aortic‐to‐atrial fistula, or clinical presentation and physician reporting consistent with possible erosion. The source of the reported erosion events included individual physician/patient reports to the company or to the FDA Manufacturer and User Facility Device Experience (MAUDE) Database, scientific publications, and events reported through industry‐sponsored clinical trials. All events suggestive of possible erosion (e.g., pericardial effusion, unexplained hemodynamic collapse, or death) were systematically reviewed to ensure no erosion events were missed. Erosion events were characterized according to details available in the adverse event report and/or the published article describing the event. Erosions were categorized both by the year the erosion occurred and by the year the device was implanted.

The number of ASO implants annually was estimated as the number of units sold in the corresponding calendar year, which serves as an approximate surrogate for procedural volume. This approach does not account for potential discrepancies between sales and actual implant timing related to factors such as inventory management, stocking practices, or device utilization patterns. However, in the absence of a comprehensive global registry of implanted devices, sales data provide the only available metric to approximate implant volume at scale and were used consistently across all years to enable evaluation of temporal trends. Specific data on erosion dates and details, along with estimated sales data, were available to the authors. Informed consent and investigative review committee approval were not required for this study.

To calculate the annual incidence rate of erosion, the number of erosions reported for each implant year was divided by the estimated number of ASO implants (units sold) to yield an overall erosion rate. Because erosion is a time‐dependent event that can occur many years after implant, overall erosion rates will be underestimated for more recent implants. Therefore, erosion rates were also calculated for various follow‐up durations, based on the number of erosions identified within the specified duration after implant: 3 days, 1 month, 1 year, 3 years, and 5 years, with each erosion rate only calculated for cases with an appropriate duration of risk (i.e., a 5‐year risk was not calculated for implants after 2019). Erosion rates were reported as percentages, representing the estimated percentage of implants after which an erosion event occurred. Three‐day, 1‐month, 1‐year, and 3‐year erosion rates were compared between implant eras (1996−2012, 2013−2017, 2018−2024) using chi‐square analysis. Incidence rates were evaluated according to various demographic and device‐related factors, including year/era of implant, duration after implant, patient age, patient sex, and device size. Erosion‐related outcomes, including death and erosion into the aorta (vs. only through the atrium), were also ascertained and reported.

3. Results

3.1. Erosion Rates

From 1996 to 2024, it is estimated that more than 500,000 ASO devices were implanted worldwide. Estimated annual number of implants derived from worldwide sales increased from 1996 to just over 20,000 in 2005, where they remained relatively stable through 2012 (Figure 1). In 2013, the number of implants rose to around 24,000 and remained at that level through 2019. After dropping by ~15%−20% during the COVID‐19 pandemic (2020−2021), they returned to ~24,000 in 2022 and then rose to a peak near 29,000 in 2024.

Figure 1.

(A) This line graph illustrates the annual number of implants (sales) and reported erosions from 1996 to 2023. (B) This line graph illustrates the overall, 1‐year, and 1‐month erosion rates for implants performed from 1996 to 2024. “F” indicates the timing of the FDA advisory committee meeting and subsequent IFU updates; “I” indicates the timing of the IFU update to include data on the relative risk factors for erosion. (C) This area chart depicts time‐dependent erosion rates by year of implant from 1996 to 2024. (D) This column chart depicts the time‐dependent erosion rates for each era (1996−2012, 2013−2017, 2018−2023); the 1 erosion reported for 2024 was not included because there was insufficient follow‐up to calculate the erosion rate beyond 3 months. [Color figure can be viewed at wileyonlinelibrary.com]

As of March 2025, a total of 198 erosions had been reported and confirmed after ASD closure with an ASO device implanted through 2024. The 2016 publication by McElhinney et al. included 125 of these erosions reported through 2014 [3]. The distribution of erosions over time is depicted in Figure 1. No erosions were reported for devices implanted from 1996 to 1999; since 2000, erosions were reported for up to 18 devices implanted per year. During this time interval, the estimated 1‐month, 1‐year, and 3‐year erosion rates were 0.016%, 0.029%, and 0.031%, respectively. Erosion rates varied from implant year to implant year, as did the distribution of time‐dependent erosion rates (Figure 1). There was a significant decrease in the 3‐day, 1‐month, 1‐year, and 3‐year erosion rates in patients implanted in 2013 or later compared to those implanted in 2012 or earlier (all p < 0.001) (Figure 1). On further evaluation, there was also a significant decrease in the 3‐day, 1‐month, 1‐year, and 3‐year erosion rates from 1996 to 2012 to 2013 to 2017 (p < 0.001 p = 0.004, p < 0.001, p < 0.001). There was another apparent decrease in the incidence of erosions from 2018 onward, but only the 3‐year rate differed significantly (p = 0.050) from the 2013 to 2017 period (3‐day rate p = 0.36, 1‐month rate p = 0.055, 1‐year rate p = 0.13) (Figure 1).

Overall, 31% of erosions were identified within 72 h of the implant, 55% within 3 months, and 77% within 1 year, while 13% were documented > 5 years after the implant (Figure 2). The latest reported erosion was identified 24 years after the implant. Demographic and outcome details overall and for patients implanted through 2012 and from 2013 onwards are summarized in Table 1. There were no significant demographic differences between erosions reported from 1996 to 2012 and those reported from 2013 to 2024 (Table 1). The distributions of patients with a reported erosion are depicted by age and by age and device size in Figure 3.

Figure 2.

This graph depicts the distribution of reported erosions after implant in terms of duration after implant. [Color figure can be viewed at wileyonlinelibrary.com]

Table 1.

Demographic and outcome variables overall and by era of ASO implant.

Variable	All erosions N = 198	1992−2012 N = 140	2013−2024 N = 58	p value
Age (years) [24]	27 (14.44)	28 (14.43)	26 (14.44)	0.81
Age ≤ 18 years [24]	62 (36%)	45 (35%)	17 (39%)	0.63
Age group (years) [24]
≤10	30 (17%)	23 (18%)	7 (16%)
11−20	41 (24%)	28 (22%)	13 (30%)
21−30	21 (12%)	15 (11%)	6 (14%)
31−40	29 (17%)	24 (19%)	5 (11%)
41−50	22 (13%)	17 (13%)	5 (11%)
>50	31 (18%)	23 (18%)	8 (18%)
Male sex [27]	55 (32%)	38 (30%)	17 (38%)	0.35
US Implant	85 (43%)	63 (45%)	22 (38%)	0.36
Erosion duration after implant [1]				0.28
<3 days	61 (31%)	48 (34%)	13 (23%)
3 days to 1 month	24 (12%)	16 (11%)	8 (14%)
1−3 month	24 (12%)	16 (11%)	8 (14%)
3 month to 1 year	44 (22%)	27 (19%)	17 (30%)
1−5 years	18 (9%)	11 (8%)	7 (12%)
5−10 years	18 (9%)	15 (11%)	3 (5%)
>10 years	8 (4%)	7 (5%)	1 (2%)
Death	14 (7%)	10 (7%)	4 (7%)	0.95
Aortic erosion [36]	104 (64%)	74 (65%)	30 (63%)	0.77
Treatment				0.74
Surgery	181 (91%)	125 (89%)	56 (97%)
Percutaneous closure	4 (2.0%)	3 (2.1%)	1 (1.7%)
Pericardiocentesis only	3 (1.5%)	3 (2.1%)	0 (0%)
No intervention—resolved	3 (1.5%)	3 (2.1%)	0 (0%)
Died without treatment	7 (3.5%)	6 (4.3%)	1 (1.7%)

Note: The number of patients with missing data are indicated in brackets. Percentages are calculated based on the number of patients with available data.

Figure 3.

(A) This histogram depicts the distribution of patients with reported erosions by age at implant in 5‐year increments. (B) This scatterplot demonstrates the relationship between age and device size in patients who developed erosions. Patients who died as a result of the erosion are indicated with open red symbols. Patients with unknown age or device size are not included. [Color figure can be viewed at wileyonlinelibrary.com]

3.2. Device Size

Erosions have been reported with ASO devices ranging from 11 to 38 mm, with over two‐thirds occurring in patients who received a device ≥ 20 mm (Figure 4). No erosions have been reported in patients who received 4−10 mm or 40 mm ASO devices. As observed in prior reports [1, 3], there was a high concentration of erosions in patients who underwent ASD closure with a 26 mm device (21%) (Figure 4). After 2013, there was a reduction in the number of erosions of 26 mm devices compared to 2012 and earlier, despite no change in the number of implants, and corresponding significant declines in the 1‐month (p = 0.015), 1‐year (p = 0.020), and 3‐year (p = 0.024) erosion rates (Figure 4). The overall reported erosion rate of the 26 mm device decreased significantly from 1996 to 2012 to 2013 to 2024 (0.210% vs. 0.058%, p < 0.001).

Figure 4.

These columns graphs depict the (A) percent of implants and (B) number of reported erosions by device size in earlier (1996−2012) and later (2013−2024) eras, (C) the overall erosion rate by device size, and (D) number of erosions during varying intervals post‐implant by device size. [Color figure can be viewed at wileyonlinelibrary.com]

Erosions tended to occur within 1 month or 1 year across the full range of device sizes (11−38 mm). Similarly, late erosions (>5 years after implant) were reported across the spectrum of device sizes, although the percentage of erosions identified after 5 years was particularly high in patients with a 30 mm or larger device (41% vs. 8% for smaller devices; p < 0.001).

3.3. Outcomes

Fatalities occurred in 7% of patients who experienced an erosion. Fifty‐seven percent of fatalities occurred within 1 month of implant (1‐month erosion‐related mortality rate 0.0015%), and 29% occurred between 1 month and 1 year following implant, resulting in a 1‐year erosion‐related mortality rate of 0.0022%. Overall, 3.5% of patients with a reported erosion died without treatment, and the erosion was diagnosed on autopsy. The remaining patients underwent treatment, as summarized in Table 1. An aortic erosion was present in all patients who died and in whom the erosion location was known (p = 0.006). An aortic‐to‐atrial fistula was present in 12% of patients with a reported erosion, none of whom died. There were no significant differences in age, sex, device size, year of implant, or duration after implant according to these two outcomes (aortic erosion or death). Only 29% of reported deaths have occurred since 2013. From 1996 to 2012 to 2013 to 2024, the overall erosion‐related mortality rate per implant decreased from 0.0040% to 0.0014% (p = 0.061), although the mortality rate per erosion was similar (6.2% vs. 6.9%; p = 0.86).

4. Discussion

4.1. Historical Context

Following the 2013 FDA recommendations, the ADVANCE ASO AMPLATZER Atrial Septal Occluder Post Market Surveillance Study (ClinicalTrials.gov Identifier: NCT02353351) was initiated in June 2014 with a target enrollment of 8,000 patients and estimated completion in 2019. Enrollment in that study was slow, making it difficult to achieve the target enrollment goal. In 2016, a case‐control analysis was performed comparing 125 erosion cases reviewed by the Erosion Board between 2002 and 2014 and 250 controls drawn from the ASO Post‐Approval Study [3]. In that article, relative risk factors for erosion were identified, including deficiency of any ASD rim, device size > 5 mm larger than ASD diameter, and smaller weight: device size ratio. In 2018, the IFU was modified to include the relative risk factors identified in that report, and the FDA allowed the ADVANCE trial to be halted, recognizing the insights provided by that article and that an 8000‐patient trial would be unlikely to provide further insight into the erosion issue, given the low absolute incidence. While the IFU changes in 2012−2013 prompted a focus on the ASO erosions, the implications of these modifications and of heightened concern for erosions over the past decade have yet to be assessed.

4.2. Findings of This Analysis

In this updated review of data ascertained by the AGA Erosion Board and Abbott Medical Affairs related to ASO erosion after ASD closure, several potentially important trends were identified. Overall and time‐dependent erosion rates declined significantly after 2012 and to a lesser degree after 2017. In addition, although a large majority of erosions (78%) were reported within 1 year of implant, including almost one‐third within the first 3 days, erosions were documented more than 2 decades after ASO implant, and erosions beyond 5 years post‐implant comprised 13% of reported cases. It was also notable that the likelihood of an erosion progressing to a fatality remained similar, at 7%, in earlier and later periods despite the updated guidance for physicians to advise patients to seek immediate medical attention for new signs or symptoms of hemodynamic instability. With the decreasing erosion rate, the estimated erosion‐related mortality rate among patients undergoing ASD closure has been approximately 14 per 1,000,000 implants since 2013. It is tempting to speculate that the warning added to the IFU in 2012 led to changes in practice and reduced risk, and that the addition of relative risk factors identified by McElhinney et al. [3] to the IFU in 2018 had an additional effect, but the true explanation may not be so straightforward.

4.3. Understanding Declining Erosion Rates

Factors that could plausibly contribute to decreasing erosion rates over time include changes in the device and/or its delivery system, changes in patient selection, changes in implant techniques or practices, the availability of other devices, or changes in ascertainment of erosions. The composition and structure of the ASO device have not changed, so this explanation can be eliminated. Although the introduction of the Trevisio delivery system in 2019 [10] may have favorably influenced ASO deployment and assessment of device position, particularly with larger devices, there is no suggestion that the use of the Trevisio delivery cable affected erosion rates.

It is highly likely that awareness of erosions and/or modifications to the IFU contributed to changes in practice, including patient selection, ASD sizing, selection of device type and/or size, and potentially implant techniques. Analysis of 2007−2015 data from the Pediatric Health Information Systems database found that the adjusted probability of operative versus transcatheter ASD closure in pediatric patients gradually fell from 2007 to 2012 but reversed course and rose from 2013 to 2015 [9], suggesting that practitioner behavior changed in response to the publicity and/or IFU changes in 2012−2013. Indeed, that would be the expected consequence of an IFU contraindication or warning about implanting in patients with a deficient or absent retroaortic rim. In fact, we regularly receive referrals for device closure in patients with a deficient retroaortic rim who were told by interventionalists at other centers that a percutaneous approach was not feasible, suggesting that the outdated IFU contraindication continues to dictate practice among some implanters. However, given that deficiency of the retroaortic rim is common among patients undergoing transcatheter ASD closure [11, 12, 13], major changes in practice related to this feature might be expected to result in a concomitant decrease in the overall number of implants. But the number of implants did not drop after the 2012 labeling changes; on the contrary, they increased. While it is possible that there was disproportionate growth in the closure of defects without a deficient retroaortic rim, this explanation seems implausible. On the contrary, it is possible that there has been a redistribution of such cases to implanters who are comfortable with the nuances of device closure in patients with a deficient retroaortic rim.

Patient selection aside, it is difficult to determine whether implant practices changed sufficiently to account for the decrease in erosion rates, as we do not have data on device size relative to patient or ASD size in patients who did not develop erosion. It is likely that minor changes in patient selection combined with changes in device sizing together contributed to the observed decline in erosion rates, but there may have been other confounders that could not be identified in this analysis that contributed to this trend as well.

Another potential factor that may have influenced temporal trends in erosion rates is the availability of alternative transcatheter ASD occluders and their perceived or reported safety profiles. Devices from other manufacturers have been available internationally for many years, including Gore septal occluder systems (W. L. Gore & Associates, Flagstaff, AZ, USA), which have been commercially available in Europe since the late 1990s and in the United States since 2006, with subsequent iterations introduced in 2011 and 2019. More recently, Occlutech ASD occluders (Occlutech GmbH, Jena, Germany) have also been increasingly adopted internationally following their introduction in Europe in the late 2000s, with subsequent widespread global use over the past decade, and more recent approval in the United States in 2023 and commercial use beginning in 2024. It is plausible that the availability of alternative devices, particularly those perceived to have lower erosion risk, may have influenced operator decision‐making, with implanters preferentially selecting non‐ASO devices in patients considered to be at higher risk for erosion. Although it is difficult to assess the contribution of this factor to the observed decline in erosion rates, the fact that ASO implants did not decrease substantially over time suggests that large‐scale diversion of putatively high‐risk patients to other device types did not occur to the extent necessary to drive down erosion rates on its own.

The possibility that changes in ascertainment of erosions contributed to lower incidence rates must also be considered. Erosions were ascertained by the Erosion Board and Abbott Medical Affairs through a variety of avenues, including published articles, reports to the MAUDE database, and direct reporting from implanters or other physicians. It is highly likely that some erosions occurred during this period that were not captured due to underdiagnosis or underreporting. Erosions may have been underdiagnosed if there were patients who died from this complication without postmortem examination or who developed an erosion that resolved without treatment, as with the patients in this series who did not undergo intervention. However, the rare erosion‐related deaths were spread evenly across eras, so it is unlikely that there was a significant era‐dependent change in underdiagnosis for this reason. It is also possible that some diagnosed cases of erosion were not reported to Abbott or regulatory bodies or published in the literature. Any of these scenarios would result in undercounting cases and consequent underestimation of the true erosion rate.

While there was likely ascertainment bias in estimating erosion rates, it is difficult to determine the extent to which changes in reporting may have impacted overall estimates and relative erosion rates over time. For example, it is possible that reporting tendencies evolved over time as erosion became a more widely recognized and less novel phenomenon, thereby potentially reducing incentives to publish, particularly for clinically straightforward cases or those presenting in the early post‐implant period. Indeed, more recent publications have tended to focus on the less common observation of erosions presenting late after implant [14, 15, 16, 17]. If true, a blunted impetus to report erosions could have contributed to some degree of undercounting in more recent eras. However, heightened awareness of erosion risk might also be expected to improve surveillance, ascertainment, and reporting. Ultimately, although the relative contribution of incomplete ascertainment cannot be precisely defined, the magnitude, temporal alignment, and consistency of the decline in estimated erosion rates suggest that reporting changes alone are unlikely to account fully for the observed trends.

Beyond general issues of reporting and ascertainment, an additional and distinct consideration relates to the adjudication process itself. While erosion events through 2015 were reviewed by an independent Erosion Board, subsequent events were adjudicated by Abbott Medical Affairs, which is not an independent entity. This introduces the potential for ascertainment or classification bias. However, several considerations suggest that this limitation alone is unlikely to account fully for the observed temporal trends. First, the most significant decline in erosion rates occurred between 2012 and 2013, which preceded the transition away from independent adjudication. Second, erosion events were ascertained from multiple sources, including published reports, clinical trials, and direct physician reporting, rather than solely from manufacturer submissions. Third, adjudication of more recent events relied on objective clinical evidence such as intraoperative findings or autopsy results, which reduces subjectivity in event classification. Finally, the proportion of fatal outcomes remained stable across eras, arguing against systematic under‐recognition of clinically significant events. Nonetheless, incomplete ascertainment and potential reporting bias remain important limitations and may result in underestimation of the true incidence of erosion, particularly in more recent years.

Another potential source of error in estimating erosion rates relates to the use of device sales as a surrogate for implant frequency. The number of annual implants was estimated as the number of units sold during the same calendar year, which serves as a reasonable approximation but remains imperfect for a variety of reasons. For example, devices may sit on the shelf for a period of time, ordering may be triggered or delayed by changes in par level or irregularities in inventory management that do not reflect the timing of implants, ASO devices may be used for other off‐label procedures, and devices may need to be reordered due to expiration rather than use. Thus, it is possible (1) that there were fewer ASO implants than units sold overall and/or in any given year if all purchased devices were not implanted to close an ASD, and (2) that the year‐to‐year distribution of implants did not align perfectly with sales, even if all purchased devices were used for ASD closure. Accordingly, the assumption that annual sales are equal to annual implants is a likely source of confounding in estimating erosion incidence rates. However, there is no clear reason that this confounding would be expected to change systematically over time, and therefore, although it may contribute to some inaccuracy in estimated erosion rates, it is unlikely to explain the observed changes in erosion rates after 2012 and 2017.

Additional considerations further support the conclusion that variability in estimated implant volume is unlikely to account for the observed trends. Inventory and distribution practices for the ASO have remained generally consistent over time, reducing the likelihood of systematic temporal changes in the relationship between units sold and devices implanted. A review of annual sales trends demonstrates early growth, subsequent stabilization, and more recent increases in implant volume, with only a transient decrease in 2020 consistent with the COVID‐19 pandemic. These patterns do not plausibly account for the decline in erosion rates observed after 2012. Although increasing procedural volume alone could contribute to lower calculated event rates, the number of reported erosion events also declined in absolute terms after 2012 despite stable or increasing volume, supporting a true reduction in erosion occurrence. The magnitude and timing of this decline, together with its consistency across multiple time‐dependent metrics, suggest that it is unlikely to be explained solely by denominator variability.

4.4. Other Limitations

Many limitations related to the data used for this analysis have been discussed above. Although relative changes in device‑size selection were observed over time, the absence of patient‑level anatomic and sizing data in patients without reported erosion precludes definitive conclusions regarding the causal contribution of these practice changes to declining erosion rates. In addition, it is important to note that comparison of erosion rates for devices implanted over a nearly 30‐year period is complicated by the fact that patients remain at risk for erosion over time, and erosions have been reported as long as 24 years after ASO implant. It is necessary to account for this time‐dependence when assessing changes in erosion risk over time. Although we considered assessing factors associated with changes in erosion rate by estimating cumulative incidence functions, we elected not to, as we believed that would entail further confounding without significantly enhancing the utility or accuracy of the estimates, as our primary interest was in evaluating era‐related differences rather than continuous changes.

4.5. Conclusions

Analysis of the available data suggests that reported ASO erosion rates have decreased significantly since 2012 and even more so since 2018. Although it is difficult to determine the extent to which the reduction in erosion rates over the past decade represents a true decline in incidence as opposed to a decrease in the case ascertainment rate, the findings of this study suggest that the drop in rates is at least partially due to a decreasing incidence and potentially to changes in physician practice.

Disclosures

Dr Levi serves as a consultant and proctor for Abbott. Dr McElhinney serves as a consultant and proctor fosr Abbott.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.