The simplest way to locate repairs that could be prevented by periodic maintenance is to make a spreadsheet showing all repairs, sort it by device type, then control number and date, and scan the page looking for relevant information; the "manual" approach. Depending on the size of the institution, the numbers of repairs in your list, and the amount of time you have to spare, this task will either be doable, or not-so-doable.

More selective approaches for dealing with larger amounts of data would involve either using some form of data-mining software to sift through technician work order data entries, or would involve codifying much of the data you collect, then subjecting the batch to sorts and searches. Both of these complex methods require specific forms of management.

Let's assume we want to sort and search.

When we have large numbers of devices and repairs to analyze, what we want is to avoid having to read through quantities of individual repair summaries (written by different technicians, with different writing abilities). By creating discrete work order entries, we can then use search and sort functions to find relevant information.

Sorting categorizes makes it possible to assess actions taken on a particular device type or on a specific model, e.g. all portable defibrillators require regular battery changes, and all model XX defibrillators have paddle connectors that break after 5 years. Sorting like this works well in identifying parts failure patterns because parts are readily identifiable (they can be numbered). But identifying other technician actions is not as easy.

Consider the fact that a minor maladjustment, in certain cases, may render a device ineffective. Because of this, in a comprehensive approach, minor actions as well as parts replacement must be noted. And because of this, organizing work order information into discrete items increases the complexity of analysis.

The results are not easy to pick out. Unlike parts replacement, actions are not as easy to specify. "Replaced" "power supply" is easy to quantify, but how do you delineate and categorize "tightened retaining nut of ESU monopolar output connector?" An example of a work order designed to accommodate several actions on one device during one repair, might look something like this:

To guarantee data uniformity for sorting, column entries would require a "pick list" format standardizing the items a technician could select. Relying on structured work orders like this, you can then begin sorting columns of data looking for relevance.

When you consider that breaking out all actions into separate items so you can sort, has created many entries for each work order and long columns representing many work orders, effective analysis will likely require other sorts such as looking at the most critical devices, and making these searches more frequently.