Adrian, You observation holds true for all aspects of the health system, and I've been using the same observation (50-75% of all patients in PHC clinics in developing countries are presenting with what we often refer to as "minor ailments") to argue that introduction of EMR systems in developing countries (at least) should target the 20-20% patient encounters that are serious cases: - HIV/AIDS - TB - STIs - Chronic cases (hypertension, diabetes, epilepsy, etc) - Trauma - Cholera and other serious communicable diseases
By focusing your EMR efforts on such encounters you are not only making the introduction of EMR easier - far fewer patients to painstakingly type into the PC, for one thing. In a recent pilot, we found that a computer literate doctor took 1.5 minutes on average to capture one TB patient's key data, whereas some of the nurses took up to 16 minutes to capture the same (PHC care in South Africa is largely nurse-driven). Many "complex" cases are in fact also long term cases where continuity of care is crucial for final outcome (HIV patients on anti-retrovirals are a good example - lack of continuity might result in resistant HIV strains etc etc). The experience with "big bang" introductions of EMR, where you want to collect everything about everybody from day 1, is pretty lousy - it usually don't work. Your observation is interpreted from a different perspective, but I think at least one issue overlaps: The simple and easy stuff is not really that important, i.e. you can also relax on all the cross-checking and follow-up routines you might want with more complex/serious cases. Some mistakes in the notes about patients with in-grown toenails or a snotty nose (URI) or a few stitches don't matter much - mistakes in diagnosing or treating TB do. What we tend to forget is that all our information system efforts only address a small percentage of the total amount of communication between health workers: an australian study presented and Medinfo 2001 indicated about 4% (50% is verbal communication between people...). Best regards Calle ********************************************* Calle Hedberg 3 Pillans Road, 7700 Rosebank, SOUTH AFRICA Tel/fax (home): +27-21-685-6472; Cell: +27-82-853-5352 ********************************************* ----- Original Message ----- From: "Thomas Beale" <[EMAIL PROTECTED]> To: "Open Health List" <[EMAIL PROTECTED]> Sent: Monday, October 20, 2003 2:58 PM Subject: Re: Some user stuff: 2) the 80:20 rule and interface speed > Adrian Midgley wrote: > > >Simple thing really - doctors using EMR systems, doctors doing doctoring. > > > >80% of what we do we do simply, easily and quickly, and are communicating > >through the notes with others who know it was simple, quick and easy. > > > >But much of the model of what we do is about the other 20% - where we work > >from gathering information, forming and then testing a series of > >hypotheses and using external information to determine the currently best > >treatment based on evidence where available. > > > >Usability and acceptability demands that the work done on making the 20% > >safer and better and even arguably quicker does not slow down the 80% to > >the same rate. > > > I think this is a key statement and probably should be considered for > inclusion in major EHR requirements works like the ISO 18308 TS. > > > > >I suspect that it may even justify having two modes, but it would be nice > >if the system ran in one mode but spotted the point of change. > > > Adrian, I've come to suspect the same thing by my conversations with > doctors about how they work. I am interested to know if you really think > 2 modes would make sense, even if they are secretly chosen by the > software. Would you characterise the 2 modes as: > - "simple patient" / "complex patient" (e.g. based on consult time, nr > consultations) > - "simple problem" / "complex problems" (e..g based on number of > comorbidities) > - "simple treatment" / "complex treatment" (e.g. advanced diabetic > versus child's ear infection) > > could the software try and guess what kind of patient you have based on > number of consults & consult time? Or would more sophisticated queries > be needed? Another way might be to configure a system to detect all > patients of particular pre-specified profiles. > > What would the two modes look like in terms of user interface & workflow? > > - thomas > > >
