The future of GP computer systems is in Windows (or X-Windows, BeOS or
other things that _look_ just like Windows (TM))

Existing Windows systems lack usability as a result of following the
Windows paradigm too far, and ignoring the better features of the
better DOS/UNIX/Console mode programs.  Improvements are possible.

The natural tendency in Windows programming is to pop up forms as new
windows, medical records are complex but particular areas tend to have
predictable sequences of operations applied to them (eg entering an
immunisation record in surgery; writing a prescription) and this can
rapidly lead to a proliferation of independently floating windows, and
confusion.

For reasons of database design and control of transactions most systems
do not permit unlimited use of the various windows, in other words if a
form to add a morbidity code is open and in use in a patient record, it
is unlikely that the system will permit a prescription to be printed at
the same time - first the morbidity code entry must be finished.

Flattening the layers of windows by using elements on the same
form/window can be used to reduce confusion.  This is not dissimilar to
the DOS database interfaces used for instance in Surgery Manager and in
at least one of the MCS products, and which represent a highly evolved
and functional concensus that extends beyond medical applications. 
This is over and above being CUA compliant in key strokes and mouse
clicks.

For instance, while entering a new immunisation record there is much to
commend simulating a pop-up form by drawing a shadow over the
immunisation form, then drawing a frame in front of that, and drawing
text boxes on that frame with the buttons and keyboard accelerators to
complete the entry.  This keeps the entry form firmly linked to the
area in which the information is displayed, and facilitates disabling
the relevant part of the program interface until the task is completed.

Using successive boxes on a simulated form instead of a series of pop
up forms or message boxes to control the entry of data (eg name of imm,
date, why, batch number, comment, claim (if allowed) details) is
preferable and speeds entry.

The use of tabbed interfaces in the main part of the patient record and
in many areas of adjustment is so desirable as to be mandatory for the
next version of all programs.

Another lesson from the DOS and for instance Novell Netware design is
to use the whole screen for many things.  In the field of GP records
this becomes using the whole of the area of the multiple document
interface (MDI) within which the patient's details are held, but the
temptation to be avoided is to use little windows popping up and
floating unconnected in front of the notes.  Again, the use in Novell
of menus which come mainly in front of the preceding menu is a good
design model, and can well be replicated inthe individual screens or
windows of a Windows system.

Complexity reduces usability, current Windows GP systems are too
complex in use, their usability is reduced by it and this must account
for a significant part of the disatisfaction recorded in a recent user
survey with the undoubtedly powerful and effective systems Meditel
S6000 and Vamp Vision.  
Torex Premiere has the advantage of design in the tabbed era, giving a
very clean initial feel, but like the others it requires significant
work on delayering its windows to increase the usability of the
multitude of useful features they all contain.

In writing my experimental WIndows interface to the data files of
Surgery Manager I applied the principles above wherever I could, with
the exception of tab containers, and am happy to show examples.  The
code is currently of no commercial value.




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%