Device-Related Adverse Events During Percutaneous Nephrolithotomy: Review of the Manufacturer and User Facility Device Experience Database

Introduction

Percutaneous Nephrolithotomy (PCNL) is the preferred treatment for large and/or complex renal stone burdens. ¹ Stone-free clearance rates are as high as 95% in experienced hands, but may come at the cost of increased complications. ^2,3 The multi-institution, international PCNL Global Study found an overall complication rate of 15%. ⁴ Many of the most commonly reported complications can be severe and include bleeding (3.6%–12.9%), urosepsis (0.3%–9.3%), urinary extravasation, or injury to surrounding organs such as pleura, spleen, liver, and bowel (0.3%–1.0%). ^{5 –8}

Although overall complications of PCNL are well reported, the relative contribution of patient factors, operative technique or experience, and device-related events are less well understood. The complex nature of PCNL requires that multiple classes of instruments be used, each subject to malfunction. This unique aspect of PCNL increases the importance of evaluating device performance. Cho and colleagues compared ultrasonic and pneumatic lithotripters and discussed mechanical failures, including probe fracture, suction tube obstruction, and overheating. ⁹ While previous studies have focused on specific instruments, there is limited research on overall device performance, particularly at a multi-institutional level.

The Manufacturer and User Facility Device Experience (MAUDE) database is a Food and Drug Administration (FDA) maintained public database that contains anonymous medical device reports (MDRs) submitted when device associated incidents occur. These incidents are defined as undesirable experiences associated with the use of medical devices in patients. This voluntary self reporting database has a large potential for underreporting of events. However, due to its anonymous nature, the MAUDE database allows for reporting of any perceived unwanted events and allows room for practitioner awareness and improvement in patient safety. ¹⁰

A previous study by Gupta and colleagues considered device complications of robotic surgery and proposed and validated a novel classification system that was used to standardize complications from the MAUDE database. ¹¹ Modeled off the Clavien-Dindo classification system, this MAUDE classification was adapted for device-related adverse events, and each event was divided into four categories: Level I (none/mild)—with no harm occurring to the patient, Level II (moderate)—with harm to the patient requiring minor intervention, Level III (severe)—with harm to the patient requiring major intervention, and Level IV (life threatening/death). Furthermore, the MAUDE classification system was externally validated among 14 surgeons from different specialties and institutions.

In this study, we reviewed PCNL events reported to the FDA maintained MAUDE database of medical device malfunctions or patient injuries in the United States. We then applied the previously described MAUDE classification system to compare device complication rates and impact on patient outcomes in PCNL.

Materials and Methods

The MAUDE database was queried for any device-related malfunction or deviation, as well as complications that may have occurred as a result of using that device. Using the phrase “percutaneous nephrolithotomy” a search was performed over the 2006 to 2016 period, which resulted in 285 MDRs. Each report was then carefully reviewed for the device class, specific type of device, malfunction that occurred, as well as any potential complication or change in management of the patient. In addition, we also obtained data regarding the occupation of the informant, manufacturer device review, and the source of the defect. Duplicate and incomplete cases were excluded for a total of 218 MDRs that were used to compile our data. We then categorized resulting data into the described MAUDE classification system.

Results

A total of 218 MDRs were reviewed for PCNL device-related malfunctions and complications from 2006 to 2016. Sixteen device categories were used during PCNL with the five most common categories as follows: 53 lithotripters (24.3%), 43 wires (19.7%), 30 balloon dilators (13.8%), 28 occlusion balloons (12.8%), and 11 nephrostomy tubes (5.0%). A total of 42 complications were reported with the most common complications across all the device classes being: 12 second procedures (28.6%), 8 cases of bleeding (19.0%), 7 residual fragments (16.7%), 4 prolonged procedures (9.5%), and 3 conversions to open (7.1%). Table 1 lists the overall device complications and associated MAUDE level with 176 (80.7%) of cases as Level I or mild complications, 26 cases (11.9%) requiring minor intervention for a moderate complication (Level II), 15 cases requiring aggressive intervention for a severe complication (Level III), and no cases of death, but 1 case of a life-threatening event (Level IV) due to sepsis.

Table 2 describes the malfunctions for the most common devices used. The most common malfunctions for lithotripter and wire were detachment of the device (60.3% and 65.1%, respectively). Malfunctions were seen most commonly with use of ultrasonic lithotripter devices in our study. Balloon dilators and occlusion balloons were commonly associated with balloon rupture (43.3% and 64.3%, respectively). Of note, there was one case of sepsis (Level IV) resulting from use of a screw dilator, which broke off and obstructed the distal ureter.

We performed logistic regression analysis of the most common malfunctioning devices comparing cases with no harm (Level I) to those with moderate or severe harm (Level II–IV). As seen in Table 3, lithotripter malfunction was significantly less likely to cause severe harm to the patient (OR 0.122, CI: 0.029, 0.526, p = 0.005), while balloon dilator malfunction was over four times more likely to cause significant grade II–IV complications (OR 4.333, CI: 1.978, 9.493, p < 0.001). The majority of significant cases of bleeding and blood transfusion were seen primarily in malfunctioning balloon dilator use (six of eight cases, 75%).

The majority of MDRs were generated by physicians (65.6%), nurses (8.3%), risk managers (4.6%), and user facilities (3.2%). The malfunctioning device was evaluated by the manufacturer in 42.7% of cases. Within the devices that were evaluated, 54.8% were due to misuse by the operator. Among the lithotripter defects that were reviewed, 18 were due to operational/user error, 7 cases were unknown, and 1 was the result of shipping damage. Wire defects after review included 9 cases of operational/user error, 2 cases of shipping/handling damage, and 3 cases were unknown. Balloon dilator malfunction was attributed to 5 cases of operational/user error and 1 case of manufacturer defect. No association was seen between specific devices for each category and any particular manufacturers.

Discussion

We assessed the spectrum of reported device-related adverse events during PCNL over a decade derived from MAUDE reports. Furthermore, we utilized a recently proposed MAUDE classification system for standardized, severity-stratified event reporting. We found that a majority (81%) of device related complications caused no harm or change in management (Level I), although 16 of 218 reports described severe and/or life-threatening events (Level III and IV, respectively). The use of the recently proposed MAUDE classification system allows for standardized reporting of complications associated with device malfunction. Universal acceptance of such a system will allow for translation of MDRs and comparison among device classes.

PCNL tract dilation methods are often based on surgeon preference and available equipment. ¹² The PCNL Global Study noted significantly higher rates of bleeding and blood transfusion with the use of balloon dilators; however, this was not appreciated on multivariable analyses. ⁴ In contrast, a meta-analysis by Dehong and colleagues revealed that patients who receive balloon dilation have a lower blood transfusion rate and shorter surgical duration than serial dilation. ¹³ Our analysis showed that the misuse or malfunction of balloon dilators had an increased chance of a significant bleeding complication compared to other malfunctioning devices used during PCNL. Despite the demonstrated safety in the literature, we feel that clinicians still need to be well versed with use of this device and how to manage its complications during performance of PCNL as improper use or malfunction may lead to significant morbidity. Our findings highlight the importance of proper use and training in dilation techniques and surgeon experience to avoid complications. Surgeon preference of access techniques and their experience play a major role in PCNL complications and morbidity and require further study.

Our study examined all types of PCNL device complications as a whole and its effect on the procedure and patient. Effective use of a device or surgical instrument is the result of the balance between the functional strengths and weaknesses of the tool combined with the skills and experience of the surgeon. Higher case volumes and longevity in practice often translate to improved surgical skills and outcomes. ^14,15 Examination of the collection of MAUDE reports allows the practicing urologist to be aware of device limitations noted by other colleagues, but because the database does not correlate operator expertise, we hesitate to draw any conclusions about potential device limitations based on this analysis. It is noteworthy that 54.8% of cases reviewed by the manufacturers attributed the malfunction to surgeon error. This finding underscores the fact that PCNL remains a complex multistep procedure that requires a great deal of technical expertise and device familiarity to achieve success. Furthermore, it highlights the importance of continued collaboration in device development and training between urologists and device manufacturers to maximize the value of a tool, while minimizing misuse and subsequent patient harm.

The MAUDE database has proven to be a valuable resource for identifying adverse device-related events with several urologic surgeries and devices, including ureteroscopy, laser, robotic surgery, Hem-O-Lok clips, and midurethral slings. ^{10,11,16 –20} Review of the MAUDE adverse event rates in the endourology literature includes 8.3% for lasers and 3.2% for locked deflection with ureteroscopes, which is similar with our findings that the majority of device malfunctions do not cause significant harm to the patient. Alternatively, Hem-O-Lok clips and midurethral slings have demonstrated higher complication rates in the database compared with their reported rates in the literature. The MAUDE database has the potential to present findings that would otherwise be unknown if there was not an anonymous reporting system. Revision of the FDA system with use of the recently proposed classification system and also more objective reporting options may help improve the quality of the MAUDE database with future reports.

Several limitations of this study must be acknowledged. The MAUDE database relies on voluntary self-reporting usually by healthcare professionals and is thus nonstandardized. This results in variability in the data, especially with how the malfunction and complication itself are reported. Second, while we highlight reported adverse events, we do not know the total number of cases, thus we cannot make generalizations about the frequency of adverse events. Furthermore, follow-up information about the prognosis of the patient and care with regard to the complication is limited preventing a more longitudinal understanding. Patient data are unavailable within this database so additional demographic data and medical comorbidities cannot be studied. Similarly, this is true for the complexity of the case with regard to stone composition or burden. In addition, we do not have correlative surgeon data such as the impact of case presentation, case volume, overall experience, and training as a possible source of device malfunction. It is worth noting that the device manufacturers attributed malfunction to operator error in over 50% of cases when reviewed. We recognize that device malfunction is in many cases multifactorial and a combination of surgeon expertise, device engineering, and patient/anatomic factors. Reporting of specific devices and manufacturers was not standardized across the database and no specific conclusion could be drawn between complications and specific brand of devices. This may be related to the variety in reporting which includes nurses and risk managers who may not understand specific nuances with each device. These aspects require further study in a more structured manner.

Conclusion

Our study represents a new method to standardize complications which occur during PCNL based on a severity scale of Level I–IV. The majority of PCNL complications were Level I and caused no patient harm. However, serious events occurred in as much as 20% of PCNL complications, and operator misuse was found in a majority of reviewed devices. Our findings highlight the importance of training urologists in proper operation and management of surgical devices to maximize surgical efficiency and decrease morbidity. These findings are useful for benchmarking outcomes, improving surgical performance, patient counseling, and can be of use in future quality improvement studies.