Ten-year analysis of 17,639 thoracolumbar instrumentation failures using the manufacturer and user facility device experience (MAUDE) database

Abstract

Background

This study utilized the FDA-maintained Manufacturer and User Facility Device Experience (MAUDE) database, an extensive resource capturing medical device reports (MDRs) of adverse events, to perform a cross-sectional analysis of failure mechanisms reported by manufacturers.

Methods

All thoracolumbar pedicle screw system failures reported in the MAUDE database between January 2014 and December 2023 were identified using FDA product codes. Failures were classified as intraoperative or postoperative and grouped into 9 and 5 clusters, respectively. Following de-identification, the 8 most frequently reported manufacturers were analyzed individually. Reports specifying failed components were categorized as screw, set screw, rod, instrument, system, other, or unknown. The “system failure” category lacked specificity regarding the failed component. Descriptive statistics summarized failure mechanisms.

Results

A total of 17,639 MDRs related to thoracolumbar fixation failures were reported over the past decade, with 15,234 postoperative and 2,405 intraoperative events. The most frequent intraoperative issues were positioning problems (27.9%), device-device compatibility (26.4%), and difficulty in separation/removal (11.8%), while the most common postoperative failure mechanisms were breakage (45.2%), migration/detachment (42%), and deformation (10.1%). The distribution of failed components varied by the associated failure mechanism. Among 4,676 reported breakages, screws (20.8%), set screws (15.5%), and rods (11.8%) were most frequently involved. In the 4,871 migration/detachment events, set screws accounted for 31.6%, screws for 21%, and rods for 2.3%. Of 1,325 deformation events, the majority were linked to set screws (76.5%), with fewer attributed to system-level components (11.2%) and screws (6%).

Conclusions

Failure profiles for pedicle screw systems vary significantly between manufacturers. The MAUDE database is a useful resource for the evaluation of reporting patterns associated with pedicle screw systems. However, limitations include reporting biases and the absence of total sales data, which prevents the calculation of true failure rates.

Background

The number of thoracolumbar spinal fusion procedures involving pedicle screw systems performed in the United States has been steadily increasing over the past decade [1,2]. Despite advances in materials development, hardware failure remains a persistent concern from both clinical and cost-efficiency perspectives [3]. Failures of screws, rods, or the rod/screw interface can lead to motion at the fixation site and may result in nonunion, impaired spinal stability, deformity, and pain, often requiring subsequent reoperations [4]. Data on failure rates of thoracolumbar fixation systems and their underlying mechanisms is primarily derived from institutional studies and is limited to relatively small series [[5], [6], [7], [8]].

The Manufacturer and User Facility Device Experience (MAUDE) database, maintained by the FDA, has recorded medical device reports (MDRs) of adverse events across the United States since the early 1990s. Manufacturers, importers, and device user facilities are required to submit reports about device-related adverse events as mandated by regulatory guidelines. Despite its inherent limitations, this extensive open access postmarket database is the most inclusive resource available for spine hardware malfunctions. Given the sheer quantity of data available, and the appropriate interest of many spine surgeons to participate in the development and refinement of spinal instrumentation systems, the objective of this study was to utilize this resource to conduct a cross-sectional, descriptive analysis of reported thoracolumbar hardware failures.

Methods

Data collection

All thoracolumbar pedicle screw system failures reported in the MAUDE database from January 2014 to December 2023 were identified using specific FDA Product Codes and included in this study (detailed definitions of the product codes are provided in Supplementary Appendix 1). The system components included pedicle screws, rods, and the rod/screw interface.

Data classification

A total of 182 types of product problems were identified and classified as intraoperative or postoperative, with 9 and 5 distinct clusters identified for intraoperative and postoperative mechanisms, respectively (eg, “break” and “fracture” were combined under “break” for postoperative categories). Clusters are presented in Table 1. Representative examples of device problems are illustrated in Fig. 1.

Table 1.

MDRs with product codes associated with pedicle screw system malfunctions.

	N	%
Intraoperative	2,405	–
Positioning problem	672	3.8
Device-device compatibility	634	3.6
Difficult to separate/remove	284	1.6
Other operational malfunction	212	1.2
Malposition	206	1.2
Mechanical jam	171	1.0
Difficult to engage	83	0.5
Other defect in device	77	0.4
Defective packaging	66	0.4
Postoperative	15,234	–
Break	6,885	39
Migration/Detachment	6,392	36.2
Deformation	1,532	8.7
Patient compatibility	172	1.0
Other/Unspecified	253	1.4

Fig. 1.

Representative radiographs illustrating failure mechanisms in pedicle screw systems. (A–B) lateral radiograph displaying migration/detachment of a set screw resulting in rod dislodgement. (C–D) AP radiograph presenting rod migration. (E) Bilateral rod breakage leading to severe spinal malalignment.

All manufacturers listed in the MAUDE database were de-identified for the analysis. The 8 manufacturers with the most frequently reported failures were analyzed individually, while the rest were grouped into a single “other” category. To account for industry changes over the study period, manufacturers that had merged in recent years were retrospectively treated as a single entity for the entire 10-year interval, with the failures of individual companies prior to the merger consolidated under the current merged company’s name.

Data processing

Data were processed from the perspective of failure events, not from the perspective of individual patients. Thus, in instances where a single patient experienced multiple events, each event was counted separately. To avoid duplications, all events falling under the same problem category were grouped and counted as a single occurrence. However, events classified under different problem categories were counted separately. For example, if a single patient record included device “break” and “malposition,” this qualified as both a “break” and a “malposition” in this study. However, if a single patient record included 3 instances of device “break,” this would only count as a single “break” in this study.

A total of 133 variables archived with the MAUDE database were identified. Manufacturer-specific details, such as geographical location and contact information, were excluded. Only clinically relevant variables were analyzed, including patient demographics, problem type, manufacturer information, and report date.

Failure component analysis

To further characterize failure mechanisms, we extracted data specifying the affected component within the pedicle screw system. Each report was reviewed for component-level details, which were classified into the following categories: screw, set screw, rod, instrument, system, other, and unknown.

Following initial classification of reports based on failure component, the “system failure” category was disproportionately overrepresented. For example, 89.6% (n = 1,983) of 1 manufacturer's reports were classified as “system failures.” To better understand the underlying failure mechanisms, we manually conducted a free-text analysis on a random subsample of 500 system failure reports, identifying and quantifying the specific failed components.

Statistical analysis

Descriptive statistics were used to present the types of malfunctions reported across manufacturers. Results are reported as mean ± standard deviation (SD) and median with interquartile range (IQR), as appropriate. A box plot representation was used to illustrate the distribution of patient age across different intraoperative and postoperative failure mechanisms. The cross-sectional nature of this study precluded any analysis of causality.

Results

A total of 17,639 MDRs related to thoracolumbar pedicle screw system failures were reported between January 2014 and December 2023. Of these, 15,234 and 2,405 cases were classified as postoperative and intraoperative, respectively (Table 1). Fig. 2, Fig. 3 present the count plot of MDRs across both intraoperative and postoperative phases. Manufacturer-specific comparisons between surgical phases, normalized by manufacturer, are shown in Fig, 4. Among the MDRs with available sex data, 50.9% of the patients were female. The mean patient age was 53.0 ± 21.0 years.

Fig. 2.

Count plots of MDRs across the intraoperative phase.

Fig. 3.

Count plots of MDRs across the postoperative phase.

Fig. 4.

Manufacturer-specific comparisons between surgical phases, normalized by manufacturer.

System and component failure analyses

The most commonly reported failure component was system, followed by set screw and screw. This distribution remained consistent in subanalyses of both intraoperative and postoperative complications. Table 2 presents the distribution of failed components categorized by the associated failure mechanism. Among the 4,676 reported breakages, the most commonly affected components were screws (20.8%), set screws (15.5%), and rods (11.8%). Of the 4,871 migration/detachment events, set screws accounted for 31.6%, screws for 21%, and rods for 2.3%. Deformation events (n = 1,325) were most often associated with set screws (76.5%), followed by system-level failures (11.2%) and screws (6%).

Table 2.

Distribution of failed components by postoperative failure mechanism (top 8 manufacturers cohort).

Component	Break	Migration/Detachment	Deformation	Unspecified integrity problem	Patient compatibility	Total (n, % of total)
Screw	972 (20.8%)	1,021 (21.0%)	79 (6.0%)	52 (31.5%)	17 (15.7%)	2,141 (19.2%)
Set screw	727 (15.5%)	1,541 (31.6%)	1,014 (76.5%)	36 (21.8%)	25 (23.1%)	3,343 (30.0%)
Rod	551 (11.8%)	110 (2.3%)	5 (0.4%)	27 (16.4%)	5 (4.6%)	698 (6.3%)
Instrument	263 (5.6%)	27 (0.6%)	45 (3.4%)	0 (0.0%)	7 (6.5%)	342 (3.1%)
Other	97 (2.1%)	92 (1.9%)	24 (1.8%)	20 (12.1%)	0 (0.0%)	233 (2.1%)
System	1,741 (37.2%)	1,687 (34.6%)	149 (11.2%)	22 (13.3%)	32 (29.6%)	3,631 (32.6%)
Unknown	325 (7.0%)	393 (8.1%)	9 (0.7%)	8 (4.8%)	22 (20.4%)	757 (6.8%)

Based on manual free-text analysis, the most commonly identified failing component was screw, followed by set screw, rod, and instrument. Table 3 presents the distribution of failed components by postoperative failure mechanism in the 500-case subsample of system-reported failures. Of the 215 breakage events in the 500-sample subset, the majority involved screws (55.8%), followed by rods (20.5%), instruments (13.5%), and set screws (5.6%). Among the 204 migration or detachment events, set screws were most commonly implicated (42.6%), followed by screws (36.8%) and rods (9.3%). Of the 23 deformation events, 39% involved set screws, 26% involved other components, 17% involved screws, and none were attributed to rods.

Table 3.

Distribution of failed components by postoperative failure mechanism (random system failure subsample, n = 450).

Component	Break	Migration/Detachment	Deformation	Unspecified integrity problem	Patient compatibility	Total (n, % of total)
Screw	120 (55.8%)	75 (36.8%)	4 (17.4%)	1 (33.3%)	3 (60.0%)	203 (45.1%)
Set screw	12 (5.6%)	87 (42.6%)	9 (39.1%)	0 (0.0%)	2 (40.0%)	110 (24.4%)
Rod	44 (20.5%)	19 (9.3%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	63 (14.0%)
Instrument	29 (13.5%)	8 (3.9%)	3 (13.0%)	0 (0.0%)	0 (0.0%)	40 (8.9%)
Other	4 (1.9%)	2 (1.0%)	6 (26.1%)	1 (33.3%)	0 (0.0%)	13 (2.9%)
System	6 (2.8%)	13 (6.4%)	1 (4.3%)	1 (33.3%)	0 (0.0%)	21 (4.7%)
Unknown	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)

Discussion

This study evaluated pedicle screw system malfunction patterns by utilizing failures reported in the FDA’s open-access MAUDE database over the past decade. The rate of overall intraoperative and postoperative MDRs was 13.6% and 86.4%, respectively. Some manufacturers reported a significant portion of complications occurring intraoperatively. For example, 1 manufacturer had 23.2% of all malfunctions occur intraoperatively, while another reported only 5.8%, representing nearly a fourfold difference. To date, most studies on intraoperative complications have focused on screw placement accuracy and cage positioning [[9], [10], [11]]. However, reporting patterns in the MAUDE database show that device-device compatibility issues (26.4%) are almost as frequent as positioning problems (27.9%). Less commonly considered mechanisms, such as difficulty in separation/removal, account for up to 11.8% of intraoperative failures.

Intraoperative versus postoperative failure mechanisms

Intraoperatively, positioning problems were the most frequently reported issue. It's important to note that we distinguished between “positioning problem” and “malposition.” Malposition refers specifically to placing instrumentation in an undesired location, but the rate reported here does not reflect the overall incidence of misplaced screws. Instead, it captures malposition events explicitly linked to device malfunction. The “malposition” category includes the MDR classifications of “malposition of device” and “difficult to position.” In contrast, “positioning problem” includes ”failure to align,” “positioning problem,” “difficult to insert,” and “positioning failure.” “Difficult to engage” was also prevalent for certain manufacturers, accounting for 31.4% and 25.5% of intraoperative reports for manufacturer G and H, respectively.

The most common postoperative failure mechanisms were breakage (45.2%), migration/detachment (42%), and deformation (10.1%). These reporting patterns correspond with previous institutional reports on pedicle screw system failures [12]. Jutte et al. reported rates of 12.1% for screw breakage and 2.8% for rod migration [5]. In a study of 100 patients with thoracolumbar instrumentation malfunctions [6], screw fractures (34%), rod fractures (24%), rod loosening (22%), screw loosening (16%), and combined rod and screw failures (4%) were observed. In adult deformity populations [7], rod fractures were reported at 9.3%, rising to 16.2% following pedicle subtraction osteotomy. Even with multiple rods, fusions spanning 4 or more levels showed a rod fracture rate of 11.8% [8].

Component failures

Component failure analysis focused on identifying the specific failed element within the pedicle screw system. The failed component varied notably by failure mechanism (Table 2). For breakage events, screws accounted for nearly twice as many reports as rods (20.8% vs. 11.8%, respectively). One possible explanation is that, in a typical construct, a greater number of screws are implanted compared to rods, increasing the likelihood of screw-related failures being reported. In deformation events, 76.5% were attributed to set screws, while only 6% involved screws. This distribution may reflect differences in how deformation is interpreted and reported in MDRs, or it may indicate design-specific mechanical vulnerabilities associated with set screws. Notably, “system” was the most frequently reported failure component despite its lack of specificity. The proportion of failures categorized under System varied widely across manufacturers, ranging from 0% to 95.5%. To further characterize System failures, a manual review of free-text comments associated with structured reports was performed. Given the impracticality of reviewing thousands of reports, a random subsample of 500 system failures was analyzed to extract the specific failed component. This analysis revealed that screw and set screw failures were the predominant contributors to events initially categorized as System failures. In this subsample, the distribution of failed components differed from the overall cohort. For instance, screws accounted for 55.8% of breakage events- substantially higher than the 20.8% observed in the full dataset. Whether these components directly caused system-wide failure or this classification reflects variability in reporting practices cannot be determined from this cross-sectional analysis.

Each component demonstrated a distinct failure profile across the dataset. Screws most commonly failed due to breakage, accounting for 45.4% in the full cohort and 59% in the 500-sample. Migration or detachment was also frequently reported in screw-related failures (47.7% in the full cohort; 36.9% in the subsample), while deformation was uncommon (3.7% and 2.0%, respectively). Set screws showed a different pattern: migration or detachment was the leading issue (46.1% to 79.1%), followed by deformation (30.3% in the full cohort; 8.2% in the subsample). Breakage accounted for a smaller proportion of set screw failures, ranging from 21.7% in the full cohort to 10.9% in the subsample. Rod failures were primarily attributed to breakage (79% in the full cohort; 70% in the subsample), with migration or detachment representing a smaller share (15.8% to 30%). Deformation of rods was rarely reported.

Reporting variability and data analysis

Reporting patterns across manufacturers showed significant variability in both intraoperative and postoperative malfunctions. For example, deformation accounted for 30.5% of postoperative reports for 1 manufacturer, over threefold higher than the second highest manufacturer. This gap could potentially indicate a consistent malfunction in material manufacturing or engineering and will be more thoroughly examined in a separate manuscript (in preparation).

The decision to separate intraoperative and postoperative complications reflects a clinical perspective, recognizing that some devices are easier to use intraoperatively, while others may offer better outcomes despite their complexity. For the manufacturer analysis, companies were grouped by their current merger status. While this approach is practical given the industry changes over the decade, it may introduce some inaccuracies due to the consolidation of data from previously independent companies.

Study limitations

This study has several key limitations. First, as a cross-sectional study, it does not allow for establishing causality between independent variables and instrumentation failure. Second, without a control group of patients with intact instrumentation and due to the lack of detailed clinical information such as patient comorbidities, type of surgery, and other relevant factors, it is not possible to identify risk factors for failure. In addition, the absence of overall sales data on total number of implantations over this 10-year period prevents calculating the true incidence or prevalence of events. By definition, even an idealized version of the MAUDE dataset would represent only “numerator data,” in isolation from the totality of instrumentation placed during the study time period. Furthermore, reporting may be inconsistent as data is submitted by a range of stakeholders, including manufacturers, importers, and institutions. This data may also contain inaccuracies as it is not subject to objective validation. Finally, there is potential for under-reporting and undetected late failures in asymptomatic patients without routine radiographic follow-up, suggesting that hardware failures may be underrepresented.

Conclusions

The MAUDE database is a useful resource for the evaluation of reporting failure patterns of pedicle screw systems. However, limitations include reporting biases and the absence of total sales data, which prevent the calculation of true failure rates. Furthermore, there appear to be significant differences in failure patterns between individual manufacturers. This will be further explored in a subsequent evaluation. To our knowledge, this is the first publication to systematically analyze failure profiles in a large series of patients associated with pedicle screw systems.

Funding

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.