The data for this sheet was downloaded from the following web sites:

1) for individual device and patient outcome data for the years 2000 through 2011.

2) to group all events by the FDA product code.

3) for further classification information. This was parsed using Java to create separate table. The descriptions are informative but could not be included due to data storage limitations.

The device files from the first site were collated into a single file.

For each patient sequence, the most significant result was listed as patient outcome. Patient outcome had to be interpreted in cases where there was more than one device or patient which accounts for less than %5 of the entries. These did not appear to be entered consistently. Where one patient outcome was entered and multiple devices, that outcome was counted for all devices. Where one device was entered and multiple patients, each patient outcome was counted for the first device. Where multiple devices and patient outcomes reported, the records were cleaned made most sense. 3 devices 6 patient outcomes counted for 6 outcomes not 18.

A separate entry named 'maintained' was determined by the author. This was done using device names, the FDA's description in the matching 21 cfr entry, additional web searches and this author's 17 years experience servicing many of these devices. The criteria was to eliminate disposable and implanted devices. Those items could not be included in the data file due to storage limitations. It was named maintained as it is assumed that the institution will be responsible for its condition even though there may not require maintenance.

Conclusions: The data should be used as a guide for further investigation only. Alot of data is missing. The data does not indicate that certain devices are individually more dangerous any others. As an example, the higher numbers of fatal collisions in Toyota Camry's in 2010 vs Ford Model T's does not inidicate the model T is safer. Present Ford and Toyota automobiles have included lessons learned over 80 years that make both safer. It's just that there are more Camry's on the road. Becuase it is unknown how many of any model is in use, individual manufacturer data is omitted. What can be assumed is that your instition has approximately the same proportions of device types as the nation. I will use a comparison of beds and Autiomatic defibrillators as an example of how this data may be used.

From 2000-2011 there were approximately 600 physiologic and another 450 cardiac monitors implicated in patient deaths. Similarly there were 300 electric and another 150 mechanical beds involved. When deciding how to spend time, it seems that risk demands that patient monitors throughout the nation demand about twice the time that beds do. Any one institution probably has close to the same mix of equipment as the national average so should do the same. Your 200 monitors demands about twice the time analyzing safety risks in purchase decisions and maintenance planning as 1000 beds. If this is the mix in your institution, the best information you may have indicates a monitor has 10 times the risk of death as bed - (600 + 450)monitor incidents/200 monitors ) / ( (300+150)bed incidents/1000 beds). There are devices that are clearly higher, such as Intra Aortic Balloon Pumps.

To say that each monitor demands 10 times the maintenance hours is a little premature to say. While there is anecdotal evidence indicating individual issues that maintenance can alleviate, there is not enough data to ration resources. By rationing, I mean 1000 FTE's nationwide changing batteries produces the same reduction in patient harm as 100,000 performing electrical safety checks.

Personal observations I have seen are that lifts and beds seem to have the same risks as monitors. These are not devices that are often taken for granted. Surgical staplers struck me as a source of patient harm that I would not have noticed without this effort.

If you would like more information please use the "Contact me" page. I have larger data files that could not be published, some PostresSQL "*,sql" files with comments detailing methodology and can also generate more specific queries. I welcome any reader opinions and independent analysis Thank you for visiting.

Page that allows downloading