[Cidrz] Conference Call at Dimagi Office at 4:30 PM

[Leo Anthony Celi]

Thanks.  I will be updating the document.

Andrés Monroy-Hernández <andresmh at media.mit.edu> wrote:  Here are my notes from the meeting

Doctors involved in the project
~~~~~~~~~~~~~~~~~~~~~
dr. goresbeck parham, University of Alabama
mulindi mwahamuntu, head research at ciderz
(dr. moulinda, miniser of health)


Scenarios:
1. normal pic
2. abnormal pic
option A) can be treated by the nurse at the clinic
option B) patient needs to be sent to the hospital

Q. what pictures should be stored?
A: both A and B) and maybe even 1?

Q. can we handle numbers of referrals?
A. the referrals will not increase at the same rate because the nurses
will get better and our system will help to refer less

number of patients that a nurse refers now: 10/week
with more experience: current referral 10/100 moves to -> 10/1000

Q. how often do you meet face to face?
every week
as the program moves to rural areas it will hard to meet that often

Q. are you OK in not meeting and instead provide feedback using
text-based communication? (sms, email, web-based messages)
yes

Q. opinions on papsmear as the only way to screen?
citology has failed in 3rd world nations
requires infrastructure, difficult to sustain,
not 1 citologist in Zambia, 1-3 years to train, they usually leave, to
UK, SA, Middle East
pathologists, 8 yrs to train, brain drain too

citologist - 50% accuracy in the 1st world, accurate because patient
keeps going every year

visual inspection, acilic acid, screeneed and treated, done by nurses...

* we need criteria (5 to 6 points) for doctors on call to tell the
nurse, why dr feels is abnormal or normal
1. changes compatible invasive cancer
2. if you can see the entire abnomal legion, inside cervical canal
3. seems too large, > 75% cervix, do not treat in clinic, referr
4. _
5. _
6. _

info from nurse to doctors:
age, treated before? failed treatment? hiv status...

Q. metrics to evaluate?
tracking progress of nurses.
how close they come to making full diagnosis, comparable to the expert say?
maybe decrease in # of patients they refer, track their ability put
the cervical lesions in the same boxes, in the same categories, than
the physicians...
digital pathology, italian pathologist w/o borders, sending them some
of the tissue removed,
we can tell their nurses how accurate they are in predicting cancer
based on what the pathologists see on the slide


More tracking:
patients are asked to go back 1 month after treatment to make sure
they are not infected, and cervix is healing,
then 6 months, 12 months...for evaluation

about 20-25% patients follow up
track them down is hard,

as of jan 2006, 2 years, screened 14K patients, 14 clinics, 1 big eval
center, 15 nurses


*someone needs to investigate, intl laws, privacy: we need to ask people!!

2008/3/20, Daniel Myung :
> comments interspersed below, i left some notes i was annotating down so
> ignore or glean from them as you please.
>
> On 3/16/08, Leo Anthony Celi wrote:
> >
> > Hi,
> > My flight is at 6:30 PM so I was planning to get to the airport by
> 4:30 PM. Even if I get to the airport by 4 PM, I'm probably queueing up for
> about an hour given it's an international flight.
> > Go ahead without me. My main questions are:
> > 1. How important is real-time decision support functionality to them? If
> they think it is key, who will provide the on-call service? Doctors in
> Lusaka? Doctors in the US? How will this be possible given the time
> difference between Zambia and the US?
>
> it's mixed. It might not be a high percentage of cases. but the dr.
> reiterated that they want feedback within 30-60 minutes of an image sent to
> teh central server.
>
> text information to patient personal mobiles.
> Additional QA process in addition to just comments on images?
> UNAIDS on sending images internationally?
>
>
> 9+ countries from pepfar, will
>
> >
> > 2. How do the doctors on-call want to be notified of the need for
> decision support? The workflow needs to be granularized fully. Will the
> doctor on-call just want to ring the nurse who sent the image? A use case
> diagram will be helpful. How do we capture and document the
> decision/information flow? This is improtant for QA.
>
> I'm thinking nothing is going to stop a doctor from calling a nurse
> directly. There's a paper from a SUNY dermatoliist who does his whole
> remote derm. work from the ER vai MMS and phone calls. So there is a
> precedence for this kind of stuff.
>
> But anyway, the scope I'd say for the initial implementation is just going
> to be this, the software is only 1 way, nurse to server, server messages Doc
> of info via MMS and/or SMS alert. Doctor will call nurse for their
> findings. Doctor must then follow up on server since the server will show
> that it was not updated on the server side, and the server can't know a call
> was made, so a doc must go in and self acknowledge that he made a call and
> put the diagnosis in.
>
> Future implementations I'd say that the doctor will have special software on
> the phone. We can mandate this since CIDRZ is in full control of its docs,
> and have the doc be able to electronically update his findings on the phone
> and msg them back to the nurse.
>
> >
> > 2. What other functionalities do they want? How do they want to carry on
> the weekly image reviews that they have now? Are the doctors planning to
> continue travelling to each area of the country as the programme is rolled
> out? Or do they envision a tele-medicine type image reviews?
>
> This is a tough question from this vantage point. Once again, nothing is
> going to prevent the docs or other nurses from going afield. This will have
> to be discussed as part of their institutional regimen. However, as far as
> the electronic system goes, I'd say that we should be writing the server
> component as if it was the primary means of QA/QI by having docs and nurses
> discuss over their images. It's not the scope of this implementation to
> look into screen sharing or whiteboards for conferences to discuss the same
> image. Assume asynchronous discussion based communication.
> 'q's
> changes compatible w/ invasive cancer
> where does the lesion go? inside/out?
> > 75% cervix area
> demog
> treatment history?
> hiv status
> cd4 count?
>
>
> how do we define metrics?
> evaluation of nurses. track their progress over time. how close they come
> to the expert.
>
> num of patients they refer
>
> put cervical lesions as same categories as attendings.
>
> digital pathology italian pathologists w/o borders
> use as comparative for cancer accuracy for histology
>
>
> >
> > 3. How are they envisioning implementation? I presume they'll pilot the
> real-time functionality in a small area? I've presumed from my previous
> question that it will be a roll-out implementation across the country.
>
> Yeah. As far as my experience has gone in Zambia, the rollout will probably
> start local and/or trusted nurses. In particular one nurse who's really
> good is moving to the other side of the country. She'd be an ideal
> candidate for remote rollout.
>
> >
> > 4. What other QA processes should we put in place for the pilot apart
> from a record of the exchange and transfer of information? for the actual
> implementation?
>
> I think this might be part of how we build up the phone software to include
> 2 way communication back? Electroncially capturing the transaction in
> realtime will probably help QA/QI in the long term.
>
> >
> > 5. We should also come up with both process and outcome metrics that we
> will track as we implement it from one area to another. These are important
> if CIDRZ really wants to become established as the center of training in
> cervical cancer screening. This is my area of expertise and would be happy
> to work with the CIDRZ folks.
>
> Metrics as discussed in the call were discussed as thus. For a pilot
> evaluation we will document the ratio of diagnoses the nurses do. Positive,
> negative, indeterminate. Next, we will also track the accuracy of the
> diagnoses the nurses put as compared with the doctors. Presumably maybe
> there needs to be a QA/QI process that evaluates nurse/doctor findings and
> assigns a score for each diagnosis after the fact?
>
> >
> > 6. Are they going to use the same nurses as we roll out from one area of
> the country to another? If not, how many nurses need to be trained? I
> presume the core group of nurses will be responsible for training the nurses
> in other areas of the country? Have they given this a thought?
>
> The training and such is CIDRZ' responsibility. They have already a well
> defined 8 week training program for nurses. Their rollout for training and
> other things are contingent on funding and other in country issues.
> However, the training of practitioners and experts in countrywide
> deployments is CIDRZ' expertise in country.
>
> >
> > 7. What sort of operational and technical risks are involved in the
> project? How can we minimize these risks? Do we have contingency plans for
> those that are most likely to be encountered?
>
> One risk/uncertainty I'm concerned about is the reliability of the
> cellular/internet infrastructure. For keeping the ministries of health
> pleased (as well as general privacy concerns appeased), we may have
> additional overhead on our data that may make communication difficult. So,
> finding contingency plans for this sort of thing looks to be important as
> well.
>
>
>
> >
> > Andres, I leave the oral presentation in your hands. Daniel
> Myung, Andres will be presenting on Tuesday in case you haven't been
> informed yet. Andres, please post a blog regarding the conference call and
> the comments/suggestions of the instructors and the rest of the class. I'll
> make it up when I get back.
> > Cheers,
> > Leo
> >
> >
> >
>
> _______________________________________________
> Cidrz mailing list
> Cidrz at mit.edu
> http://mailman.mit.edu/mailman/listinfo/cidrz
>
>


-- 
Andrés

_______________________________________________
Cidrz mailing list
Cidrz at mit.edu
http://mailman.mit.edu/mailman/listinfo/cidrz

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/cidrz/attachments/20080322/8edf296d/attachment.htm