Examining Implant Safety in the US

Posted on May 2, 2023

Medical advancements have made great strides in the last century. With the new understanding of materials science, implantable medical devices have become increasingly popular. The United States holds the world's most significant medical device market, with sales making up 40% of worldwide revenue. Roughly 32 million Americans, about 10 percent, have an implanted medical device in them. For most people, it’s a literal lifesaver, but it does pose some risks.

According to data recorded by the U.S. Food & Drug Administration (FDA), medical devices have been linked to more than 80,000 deaths and 1.7 million injuries in the last decade. The FDA collects data voluntarily from manufacturers, doctors, and patients, often leading to incomplete reporting. The American Database for Medical Implant Transparency (ADMIT) was created to close the gap between different government sources that contained information about implantable medical devices approved by the FDA.

Methodology

This project analyzes a dataset from the American Database for Medical Implant Transparency (ADMIT) stored in the Harvard Dataverse. With over 300k records of implantable devices, this data set contains information about the type of device, classification, recalls, injuries, or deaths caused, among other data.

This analysis aims to answer these questions:

  • Which FDA panel reviews the most implantable medical device applications?
  • Which companies produce the most devices, and which medical industry do they target?
  • How does the FDA classify these devices, and how does it affect approval?
  • Which device class has the most reported recalls, malfunctions, injuries, deaths, and adverse events?
  • How does a clinical trial impact study sponsorship?

Analysis

Data inspection

The initial inspection of the dataset showed that the last few columns had a considerable amount of missing values (Nan); however, the owner explained that the last 7 columns starting with Study_Sponsor, only contain data if a clinical trial was required, which in the majority of the cases, is not. The plot below shows a summary of the non-missing values in the dataset.

What FDA panel receives the most implantable medical device applications?

Before a device can be marketed, manufacturers must submit an FDA application for approval that classifies it under one of the 16 medical specialty panels. The FDA provides specific regulations and requirements for each panel on the FDA website. All the device classification panels can be found at: https://www.fda.gov/medical-devices/classify-your-medical-device/device-classification-panels.

An outstanding number of implantable medical devices (99.5%) target the orthopedics industry, with plastic surgery and cardiovascular industries in second and third place, respectively. Even though the dataset doesn't contain information about the device name or specific function, the high number of orthopedic devices might be because orthopedic devices require more parts to be implanted. For example, in the case of an implantable knee, each type of screw, nut, and pad can be listed as different entries.

The figures below summarize the number of FDA applications per specialty.

Which companies produce the most devices, and which medical industry do they target?

The treemap below shows the top companies producing the most implantable devices. Nuvasive and GBS Commonwealth each make over 20,000 devices targeting the orthopedics market. The orthopedics industry is so large that plastic surgery, cardiovascular, gastroenterology, urology, and gynecology, don't even appear in the graph.

To compare the top 5 players in each specialty area, I grouped by medical_specialty, selected the top 5, and sorted by medical specialty name. By studying this data frame, we can see the striking difference in the number of implantable devices produced by the orthopedic industry compared to the others. The table and figure below show the top manufacturers in each sector.

How does the FDA classify these devices, and how does it affect approval?

The FDA classifies all medical devices into Class I, II, and III that determines its specific approval process, as summarized below:

  • Class I: These low-risk devices, e.g. toothbrushes and bandaids, can get approval by either Exemption or a 510(k) application. 
  • Class II: These moderate-risk devices,  e.g. catheters and blood pressure cuffs, can also be approved by Exemption or a 510(k) application.
  • Class III: These high-risk devices, e.g. pacemakers and high-frequency ventilators, can be approved by a 510(k) application, a Pre-Market Approval (PMA), or an Alternative Pathway. 

The figure below shows the FDA regulatory pathways:

Approval process description

  • Exemption: This path is used when the FDA doesn't require to provide reasonable assurance of safety and effectiveness for the device, though general controls are still needed.
  • 510k: This is the process for demonstrating that the new device is "substantially equivalent" to one already approved by the FDA. Safety testing requirements are waived if successful, although general controls are still required. This process can take 1 to 9 months.
  • Pre-Market Approval (PMA): This is the most stringent FDA approval path, used for devices that are high risk or don't have a "substantially equivalent" part. Manufacturers must meet general and special controls in addition to submitting clinical data. This process can take 9 to 36 months.
  • Alternative Pathway: Processing times can be expedited for humanitarian, De Novo, or Custom Device Exemptions.

The data shows that most Class 2 devices apply for 510k clearance, which makes sense as it is faster than pursuing PMA approval. However, for class 2 devices to qualify for 510k clearance, they must prove that they are equivalent to an already FDA-cleared device. Failing to show this will require a class 2 device to apply for PMA approval. The data shows that only 2 Class 2 devices applied for PMA, while 318,340 went through the 510k route.

Looking at the Class 3 devices, 275 Class 3 devices were cleared by 510k. These represent 17.1% of the total Class 3 devices. The FDA has been largely criticized for clearing these high-risk devices by the 510k route. The other 82.9% of Class 3 devices were approved through PMA application, which means they were subject to more stringent requirements, including clinical testing. The table and figure below summarize these findings:

Which device class has the most reported recalls, malfunctions, injuries, deaths, and adverse events? 

We know now that all medical devices are classified into three classes from safest to highest risk. The ADMIT data set contains information about recalls, malfunctions, injuries, deaths, and reported adverse effects. None of these events sound like something we would like to happen in a device implanted inside us. Ideally, we want these events to be close to zero. The analysis below shows the percentage of devices in each class associated with recalls, malfunctions, injuries, deaths, and overall adverse effects.

The plot above shows no specific pattern regarding the percentage of recalls, malfunctions, injuries, deaths, and overall adverse effects. While Class 2 devices have a higher rate of recalls, class 3 devices have a higher rate of malfunctions, injuries, deaths, and overall total adverse effects. This data is counterintuitive to our prediction that adverse effects would be rarer in class 3 devices due to the more stringent approval process. However, the data indicates otherwise. This raises questions about the PMA approval process and the clinical trials these devices underwent. Why did the 72.6% of Class 3 devices that cause injuries still come to market? Shouldn’t they have been identified and pulled on the basis of their clinical trials? 

As we've seen, Class 3 devices can get approval through a 510k application or a PMA application. We now investigate if a PMA led to fewer adverse effects than a 510k application. As we recall, a PMA application has more rigorous guidelines than a 510k application. Therefore it should minimize the negative effects. The plot below shows that a PMA application led to fewer recalls, malfunctions, and deaths than a 510k application, but a higher percentage of injuries and total adverse effects.

Although all adverse effects of an implantable medical device can be detrimental to our health, injuries and deaths impact patients directly. To draw more insight, we can calculate the ratio of injuries to total adverse effects and deaths to total adverse effects for Class 2 and Class 3 devices.

The analysis above shows that the ratios of injuries and deaths to total adverse effects for Class 2 and Class 3 devices are similar.

The data we have analyzed so far has focused on events reported to the FDA voluntarily by patients, doctors, and manufacturers. As we've seen, up to 72.6% of Class 3 medical devices have caused some injury. We can now compare that information with the data available during clinical trials for devices that obtained FDA approval through a PMA application, mainly Class 3 devices.

The analysis below focuses on the devices with clinical trial data recorded, that is, 1,103 devices.

The percentage of serious adverse events patients report during medical trials is 6.09%, and the percentage of other adverse effects is 10.39%.

A closer look at the data shows that some devices have more adverse effects than others. For 2.93% of devices for which medical trial data was available, 100% of patients reported adverse effects. It's worrisome that these devices still got FDA approval and are marketable in the US.

How does a clinical trial impact study sponsorship?

Clinical trials can be costly. Literature suggests the cost per patient averages $41,000. The data set contains the feature "Study Sponsor" for devices for which the manufacturer paid for the clinical trials.

The output above shows that manufacturers sponsored 100% of the clinical trials, which agrees with the findings reported in [2]. There is a significant conflict of interest regarding clinical trials and study sponsorships. Clinical trials are expensive, which raises the question of who should pay for them. If manufacturers pay for them, studies can be biased in favor of manufacturers. Otherwise, American taxpayers could face a hefty bill if the FDA funds the clinical studies.

Summary

As we've seen through this study, many implantable devices have been detrimental to patients' lives.  The percentage of serious adverse events patients reported during medical trials is 6.09%, and the percentage of other adverse effects is 10.39%. A closer look at the medical trial data showed that 2.93% of devices had 100% of patients that reported adverse effects. Out of all the  Class 3 devices 17.1% gained approval through a 510k application. The 82.9% that followed the PMA route correlated with a lower rate of recalls, malfunctions, and deaths, though they also correlated with a higher rate of injuries.

There are no formal requirements for manufacturers, doctors, and patients to inform the FDA of any unwanted effects caused by implantable medical devices. It's necessary for the FDA to collect this information and make it public so U.S. citizens can make more informed decisions regarding their health.

Future Work

Implantable medical devices can improve a patient's health; thus, it is essential to analyze the data to minimize the possibility of negative side effects. Multiple questions arise from the analysis performed in this study that we can look into further in future work. Here are some notions:

  • Can the specific requirements for the PMA hint at why the PMA process leads to more injuries?
  • Can we predict what would happen if all Class 3 high-risk devices file a PMA instead of a 510k application? How would the adverse effect change?
  • Can we obtain more specific data on what each product does and associate its function with the adverse effect reported?
  • Is there a correlation between clinical trials and total adverse events reported?

Resources

[1] Dataset: https://dataverse.harvard.edu/file.xhtml?fileId=4771681&version=1.0

[2] Babeu, J., Benoit, Z., Boulanger, A., & Turcotte, A. (2021). Introducing ADMIT : A First Step in Uncovering the FDA’s Deceit [STEM Fellowship Big Data Challenge 2021].

About Author

neizalazo

Neiza is a data scientist passionate about uncovering the story behind the data. Neiza worked 10 years as a CPU designer at Intel. She earned a Bachelor and a Master of Science in Electrical Engineering from the University...
View all posts by neizalazo >

Leave a Comment

No comments found.

View Posts by Categories


Our Recent Popular Posts


View Posts by Tags

#python #trainwithnycdsa 2019 2020 Revenue 3-points agriculture air quality airbnb airline alcohol Alex Baransky algorithm alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus ames dataset ames housing dataset apartment rent API Application artist aws bank loans beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep boston safety Bundles cake recipe California Cancer Research capstone car price Career Career Day citibike classic cars classpass clustering Coding Course Demo Course Report covid 19 credit credit card crime frequency crops D3.js data data analysis Data Analyst data analytics data for tripadvisor reviews data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization database Deep Learning Demo Day Discount disney dplyr drug data e-commerce economy employee employee burnout employer networking environment feature engineering Finance Financial Data Science fitness studio Flask flight delay gbm Get Hired ggplot2 googleVis H20 Hadoop hallmark holiday movie happiness healthcare frauds higgs boson Hiring hiring partner events Hiring Partners hotels housing housing data housing predictions housing price hy-vee Income Industry Experts Injuries Instructor Blog Instructor Interview insurance italki Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter las vegas airport lasso regression Lead Data Scienctist Lead Data Scientist leaflet league linear regression Logistic Regression machine learning Maps market matplotlib Medical Research Meet the team meetup methal health miami beach movie music Napoli NBA netflix Networking neural network Neural networks New Courses NHL nlp NYC NYC Data Science nyc data science academy NYC Open Data nyc property NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time performance phoenix pollutants Portfolio Development precision measurement prediction Prework Programming public safety PwC python Python Data Analysis python machine learning python scrapy python web scraping python webscraping Python Workshop R R Data Analysis R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn seafood type Selenium sentiment analysis sentiment classification Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau teachers team team performance TensorFlow Testimonial tf-idf Top Data Science Bootcamp Top manufacturing companies Transfers tweets twitter videos visualization wallstreet wallstreetbets web scraping Weekend Course What to expect whiskey whiskeyadvocate wildfire word cloud word2vec XGBoost yelp youtube trending ZORI