Just some thoughts off the top of my head:
Are any of the ideas that John puts forward similar to those ideas of equipment management that insurance-based schemes follow? If they are then they're positively discouraged by the NHS. As far as I see it when you increase the period of the service interval you have to accept that a point will be reached where you start to see failures due to wear and tear (assuming the previously used manufacturers recommended interval was appropriate and the prior methods were effective of course i.e. few wear & tear failures) before PPM is due - this is the feedback and the failures due to wear & tear are the indicators that the PPM period is too long. Random failures (mechanical and electrical) and FOB (operator) problems will always occur and will be related to quantity of equipment and training but not necessarily wear & tear.
It's a fact that to be certain that the selected PPM interval is effective that an additional period of time will have to elapse, whilst the current servicing interval is maintained for all devices. During this period the failure rate due to wear and tear will probably increase, not necessarily that rapidly but it will likely increase at a faster rate as more time elapses and more feedback is collected. The decision is when to change the servicing interval and how to deal with devices that have now obviously exceeded their "inherent service requirements" i.e. they're overdue PPM or more difficult, perhaps, when to increase the interval. There's always a tendency for too long a PPM interval since that's when feedback and the indicators are provided). Ideally devices still in service but beyond the newly determined service interval would have to be pulled-in for immediate servicing so that application of this servicing method could be maintained, effectively, without introducing additional risk. Hopefully then the reduced service interval would eradicate problems with failure due to wear and tear - the objective of all planned maintenance.
If all equipment were not PPM'd at change of PPM interval (if it were found to be too long) then there would be be a number of items of equipment that would not have been serviced (at the manufacturers or the newly defined period), due PPM (on paper) and will possibly be in a condition where they're imminently likely to fail due to wear and tear when this could have been preventable via appropriate PPM inteval recommended by the manufacturer. You have effectively set up a system that you can ocasionally produce the evidence to "hang" yourself or demonstrate that you're creating additional risk if you actually choose too long a PPM interval using this iterative process for determining PPM intervals. Fine as long as you don't have a fatal incident attributed to equipment failure and end up in an inquiry.
Using John's method how do we seperate the random failures from the routine failures associated with a particular make, model and manufacturer? I ask this since I think this method John suggests is only really possible using statistics and suchlike and reliable feedback from servicing. Using statistical methods implys that there will be a significant number of identical devices that are being serviced using this method; otherwise the information gathered is probably not going to be of much use in determining service intervals generally. Thus the process becomes complex when you consider the different makes, models and type of devices in a typical hospital. We all know that particular models, brands and manufacturers equipment tend to have their own failure traits - battery failures, mechanical failures, devication from calibration, etc, etc. So the method can only really be applied, on a batch-basis, to identical types of equipment I think.
The way I look at it, in theory, John's idea of using this method makes sense if you want to make the best use of resources, you are having problems with manufacturers service intervals i.e. failures due to wear & tear, you have lots of equipment of the same type and failure due to wear and tear does not present a significant risk (if you get the PPM interval wrong), either physically or financially (as a precaution just in case you do actually get service intervals wrong). However I wonder what evidence John can produce to prove that equipment is being "over-serviced" to justify using this method - there is no evidence that servicing is not currently working until you actually promote failures by increasing the interval to justify using a different method. This does not feel the right method of justification to me. Financially and operationally for EBME maybe but not technically, considering both financial and physical risk to ALL concerned.
This approach actually makes what may have been a reliable device that was routinely serviced (or "over-serviced") at the manufacturers recommended intervals into a device that has to be proved unreliable by increasing the PPM interval so you can justify increasing the PPM interval i.e you will probably actually increase the statistical risk of equipment failing due to wear & tear in use (there will be a apread in failures over time but less "leeway" possibly, compared to intervals associated with "over-servicing") if you increase the PPM interval. What you try to achieve with this method is an acceptable balance of risk versus failure. Multiply this by many times (applied to a number of batches of different types of high-volume equipment in the hospital) and if you get it wrong you could be asking for trouble.
This method can only really be effectively applied for large batches of equipment of the same type, make and model that have identical failure traits. You'd then have to monitor all failures to seperate user problems, random failures and recognised failures associated with a lack of servicing (wear & tear). Thus there is "noise" associated with the "wear & tear" indicator used to give feedback required to set the servicing interval. There is also the issue of usage of equipment. Failure due to wear & tear is the indicator for setting PPM intervals. However wear & tear is also dependent upon the amount of use equipment gets. Now equipment of the same type in different applications or areas gets different amounts of use and abuse hence wear & tear varies across the equipment population. This method fails to address this problem since it can only really be applied to the pool of equipment as a whole, otherwise you may as well micro-manage it. Once you start to fragment maintenance down to local areas or smaller quantities of equipment it becomes impossible to manage and less effective.
If too long a PPM interval is applied across all the equipment in the pool (even if it is of the same model and I have said that I think this method has a tendency to PPM intervals that are too long) the risk is that the high-usage equipment will tend to fail before the rest of it due to wear & tear. To overcome this you'd have to set the variable PPM intervals to ensure that the majority of equipment presented minimal risk - if you had a minority of the equipment being heavily used and suffering high rates of failure due to wear & tear, and the PPM interval was set appropriately to avoid this, then the rest of the equipment would still be "over-serviced" using this method. Potentially the same situation that John was trying to argue against. If the servicing interval is too short then you have equipment out of service uneccessarily, you don't necessarily prevent the FOB and random failures that occur anyhow but you can keep on top of "routine" problems associated with equipment that fails due to wear and tear and not have as many failures in-use.
Finally you have to consider the lifetime of the equipment as well. Typically for the large batches of equipment, where this method might be applied successfully, it's only going to be 7 years or so, if that. This means there won't be time for too much chopping and changing of service intervals for most of the high-volume types of equipment this would apply to. If you only select too long as service interval once or twice in that lifetime you would probably end up with a knock-on effect where you'd be forced to "reset" servicing i.e. get to service all the affected devices as soon as you find the rate of failures due to wear and tear increasing rapidly. Otherwise your servicing method actually contributes to failures before the next PPM is due and you're back to running a breakdown service anyhow.
The problems of setting too long a service interval may fall onto patients and users if the results of wear & tear due to selection of too long a PPM interval cannot be rectified quickly (taking up the EBME manpower and resources that you're trying to save) hence increasing risk both financial and physical. I thought we were around to avoid risk to operators and patiets if we could, not create additional risk then manage it on others behalf i.e. Trust board level. At least "over-servicing" at prescribed intervals gives predictable costs, predictable periods between downtime, meets the manufacturers recommendations. Though I'm all for having shorter PPM intervals, than the manufacturer recommends, if wear & tear is an issue and the manufacturers interval is found to be too long.
Johns ideas are probably feasible if there are large volumes of the same equipment and the service interval can be "reset", i.e. all the equipment is serviced, when the interval is found to be "too long" and failures start showing-up. It's the period during which failures show-up when the risk is being taken. Thus John's idea is not suited to certain risk categories of equipment either. Users probably wouldn't tolerate the inconvenience or risk of finding an appropriate interval since this process would in fact rely upon an increase in failures flagging problems with the interval. Not good news if the equipment is life-support for example. Or if you're a manufacturer with a product that is perceived as unreliable by operators if selected PPM intervals are inclined towards risk and saving EBME resources.
Just an afterthought - if you don't see problems with equipment failing and the intervals for PPM are extended tehn it's likely that the probability of catastrophic in use will be increased e.g. electric shock, failure to ventilate, infuse, etc, etc. This means that there would proably have to be a maximum limit placed on the service interval anyhow, depending upon the risk classification of the device being considered.