Just like the cheapest part of having a child is the actual hospital delivery costs, so too the cheapest part of obtaining a PACS is the actual equipment costs. Ongoing service and support costs can easily exceed the initial purchase price costs over a 5-year return on investment (ROI) period. Because virtually all imaging purchases routinely come with service agreements, most PACS are bought with service agreements. The problem is that, in more than 95% of the cases, PACS service agreements are not needed.

Virtually every PACS sold today carries a minimum 99%-plus uptime guarantee on critical components (eg, servers, interfaces). This means that no more than 6.4 hours of downtime per month are allowed before the vendor is in default of their uptime guarantee, based on a 24/7 system availability basis. If this is based on a 40- hour week (which, unfortunately, many service agreements are), the allowable downtime is reduced to just over 2 hours per month. While 6 or even 2 hours seem like a lot, the reality is that downtimes will not even come close to this unless there is a catastrophic systemic failure. In that case, unless you have a redundant server, you are in serious trouble with or without a service agreement.

Most other PACS components, including radiologists’ diagnostic workstations and the radiology information system (RIS) interface, surprisingly enough, are typically exempt from the uptime guarantees and service response time clauses. In fact, most service agreements identify that if 75% of the diagnostic workstations are up (two of eight workstations nonfunctioning, for example), then the entire system is considered “up.” Explaining to the two radiologists who cannot readand the other six who have to pick up their counterparts’ slackthat the system is technically “up” might be a slight challenge, but that is all part and parcel of what is known as “vendor logic” with regard to service agreements.

The economics of service agreements also seem to defy logic. If indeed the maximum allowable downtime is 77 hours per year total (99% uptime), why then would a facility want or need to spend 14% or 16% of the system list price to ensure that uptime? If you priced out 77 hours even at $500 per hour, that equates to less than $40,000 per year in allowable service costs before the vendor would be in default of their uptime guarantee. Even a sub-million-dollar PACS (list price) would cost $140,000 per year in service agreement costs, yet afford the facility little benefit for the extra $100,000 in cost over obtaining service on a time and materials basis. The one exception would be if that cost includes software obsolescence protection, yet not all vendors include this in their agreements.

Service agreement proponents would say that service agreement customers get priority service. There is some logic to that (after all, you are paying for it), yet if I am buying a multimillion-dollar system, I certainly am going to negotiate my service response time with a much more critical eye than I ever would system uptime, as response time is much more critical to the overall system performance.


There are a number of fallacies regarding service agreements as well. Uptime guarantees are typically measured quarterly, not monthly. What this means is that a PAC system that performs flawlessly for 6 months can be completely down for 30 hours in a row (as was the case in one recent PACS outage) and still have the vendor say they met their uptime guarantee. How? Because they were afforded 38.5 hours of cumulative downtime in the 6 months prior that were never used. Even though that exceeded the monthly allotment, it did not exceed the allotment cumulatively. Downtime should be measured monthly, with the clock starting again the first of every month.

Another misnomer is that the quoted service agreement price covers everything. A customer should budget at least an additional 30% over and above quoted service agreement costs for noncovered service, including travel time (which is not always covered), after-hours service, and noncovered parts. Always read the fine print. Most service agreements cover service only for Monday through Friday, from 8 am to 5 pm or, best case, 8 am to 9 pm (although at an added cost). Service obtained after that, including holidays and weekends, is not only extra, but at a premium price. That said, PACS service that addresses less than 25% of a hospital’s operational hours is not really service at all. In addition, even if you do opt for extras like software obsolescence protection, the new software will be loaded only during…you guessed it…normal service hours. No information technology manager in his or her right mind would ever consider doing that any more than they would consider bringing up new versions of software on a fully operational system (although once again many hospitals buying PACS still rarely consider the purchase of test servers for this purpose, let alone the purchase of redundant servers).


Why then do hospitals routinely buy service agreements if they are such a bad deal? Habit, for one. Virtually every piece of equipment in most radiology departments has a service agreement on it and for good reason. Very few places can service a CT or MRI scanner. For every hour a scanner is down, the hospital loses at least $1,000 or more in collected revenue. The same can not be said for PACS, as PACS does not generate revenue directly like a modality does, even though it does impact technologist productivity, which impacts revenue indirectly by improving throughput. The modalities can continue to scan if PACS is down, it is just that reading may be delayed slightly (although in a properly designed system, images can be routed so at least some images can be read)

There is one other huge distinction that needs to be considered. CT, MRI, ultrasound, and other devices are all highly complex medical devices with specialized boards. A PACS, even though it too requires the Food and Drug Administration’s 510(k) clearance as a medical device, is technically nothing more than applications software running on off-the-shelf hardware. That is why many PACS vendors allow a hospital to buy the software and then integrate it with hardware that the facility has purchased. The truth of the matter is that very few, if any, components in a PACS are specialized, making service on the hardware a commodity at best.

Software, however, is another story. Purchasing software-only support agreement is almost a requirement, especially if it includes software obsolescence (upgrades, not just updates). The cost of software upgrades can often exceed the cost of a software-only support agreement, so negotiating a software-only support agreement makes sense if the vendor does not offer software-only support on a time and materials basis (which almost none do).

Why software-only support? Most vendors will tell you that more than 90% of all service calls are handled by telephone support. With the exception of larger entities, most facilities have only one PACS administrator. That person, unfortunately, can not be on call 24/7. Remote telephone support not only picks up the slack when the PACS system administrator can not be there in person, but also assists the system administrator in troubleshooting and diagnosing problems.


The alternatives to not buying a service agreement are actually more than one would imagine. As previously stated, purchasing software-only support for a PACS should be a given and as absolute as the decision to have a PACS administrator. Expect to pay between 4% and 6% of the total system list price for software-only support if the vendor offers a complete package (hardware and software) or between 12% and 14% of the software-only price for software-only support. You should also seriously consider a separate service agreement on computed radiography systems if they are part of the PACS purchase, since they are specialized modalities.

If you absolutely feel the need to have on-site hardware support, you may be able to negotiate purchasing a prediscounted block of service at 15% to 30% off the standard time and materials rates. While the service block will have an expiration date, the costs savings would be of value, even if using it only for preventive maintenance checks (which service agreements typically provide only once a year anyway at 30 minutes per device). You can also consider buying a spare parts kit (best when done in conjunction with biomedical training so they can provide second tier service to the PACS administrator’s first tier support), but a better route may be to negotiate an extra software license and keep a “cold spare” for critical devices like diagnostic workstations that can be deployed when required.

The decision to get a service agreement should be based on logic and not on vendor arm twisting. As with any clinical system, a PACS will require service. One just needs to look at the value of the service agreement relative to the projected need and make a decision that is in the best financial and operational interest of the facility.

Michael J. Cannavo is president of Image Management Consultants based in Winter Springs, Fla