Neal lesh-1202742298252135-3 (5)

Neal Lesh

Computer science applications to improve health delivery in low-income countries.

My Story

Mid-thirties computer researcher seeks more fulfilling career. Goes back to school then off to Africa. Discovers things are more simple and more complex than he originally imagined. Can't imagine doing anything else...

Outline

• BackgroundThe simplicity and complexity of global inequity

• Two examplesPatient record systems for AIDS treatmentMedical algorithms on handhelds

• Conclusion

Risk Factorfor surviving the Titanic.

0

10

20

30

40

50

60

70

1st 2nd 3rd

class of service

% s

urvi

ved

Poverty as a

Global Health

Simple Story$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

$

Infant mortality: 5 per 1,000 births

Maternal mortality: 8 per 100K births

Life expectancy: 78 years

Infant mortality: 95 per 1,000 births

Maternal mortality: 500-1000 per 100K

Life expectancy: 45 years

300-540

57

69

Simple Story$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

$

Simple Story$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

$$$$ Infant

mortality: 95 per 1,000 births

Maternal mortality: 500-1000 per 100,000 b

Life expectancy: 45 years

Simple Story

“We are the first generation that can end poverty.”

-Eveline Herfkens, UN Millennium Campaign

Complexity• Corruption, careerism, tax write-offs• 5-star poverty alleviation meetings• Unintended consequences, e.g., paying volunteers• Imperialism & foreign experts

“If you want to build a ship, don't drum up people to collect wood and don't assign them tasks and work, but rather teach them to long for the endless immensity of the sea.”– Antoine de Saint-Exupery

Information as Care

• Study: rigorous application of standard treatment protocols reduced in-hospital mortality in children’s malaria cases by 50%

• Clinician’s complaint: where are my lab results?!

• Patient Knowledge Example: five danger signs for seeking care during and after labor.

Outline

• BackgroundThe simplicity and complexity of global inequity

• Two examplesPatient record systems for AIDS treatmentMedical algorithms on handhelds

• Conclusion

One year later

AIDS Treatment in Rural Rwanda

One year later

Improving Health Systems

One year later

Connecting to the Internet

Electronic Medical Record (EMR)

Patient MonitoringReports

Clinicians & Patients

Managers

EMR Staff

Paper forms

Program MonitoringReports

Funder & governmentreports

$

Re-allocate resources

Patient Monitoring

Missed-Visit List

ICT task: satisfy reporting requests

OpenMRS

• Open source framework for medical record systems

• www.openmrs.org

Data Quality

• Mistyped IDs• Missing &

conflicting data• Backlog

Potential solution: point-of-care systems

Challenges & Opportunities

• Keep up with demand• Increased impact on decision making

– Inform to Improve (I2I) teams

• Integration of lab and pharmacy components• Detecting important trends in data

Outline

• BackgroundThe simplicity and complexity of global inequity

• Two examplesPatient record systems for AIDS treatmentMedical algorithms on handhelds

• Conclusion

Rural Dispensary in Tanzania

Standardized Care (IMCI)

Standardized Care (IMCI)

Standardized Care (IMCI)

10

15

20

25

30

35

1999-00 2001-02

An

nu

al

mo

rta

lity

ra

te

Morogoro (IMCI) Rufiji (IMCI)

Ulanga Kilombero

Tanzania: underfive mortality was 13% lower in the two IMCI districts

Source: Schellenberg J et al

Full IMCIin HF

End ofstudy

13% difference95% CI: -7%, 30%

Significant impact on stunting

Deploying IMCI

• IMCI – Shown to reduce mortality and morbidity– Adopted by ~100 countries

• But uptake not as good as hoped– Training expensive– Correct use tapers off over time– Supervision challenging

Why Automate IMCI?

Why Automate IMCI?

• Improve adherence• Improve

supervision• Easier to update• More sophisticated

protocols• Reduced training

Field Work

Results to be published in CHI’08

How Automate IMCI?

Exploratory Study

• Pretesting & rapid iteration• Structured interviews• Observed trials w/

additional clinician to:– Ensure safety– Record adherence to IMCI– Record time

Viral Training

Key Findings

• Must be – Fast– Flexible– Improve adherence to IMCI

• Must address intentional deviation from IMCI– Temperature, respiratory rate– Advice

Adherence ResultsInvestigation

Current practice adherence

e-IMCI adherence

p-value

Vomiting 66.7% (n=24) 85.7% (n=28) -

Chest indrawing 75% (n=20) 94.4% (n=18) -

Blood in stool 71.4% (n=7) 100% (n=3) -

Measles in the last 3 months

55.6% (n=9) 95.2% (n=21) < 0.05

Tender ear 0% (n=1) 100% (n=5) -

All 61% (n=299) 84.7% (n=359) < 0.01

Triaging patients on treatment for AIDS

(Study ongoing in South Africa)

Counselors ask a series of questions leading to a patient assessment.

e-CTC for HIV screening

CommCare

Start House Hold Visit

Plan Day

Explore Data

Exit

House Hold Visit (Task Queue)

000:04:56

Register Birth

Investigate Diarrhea of Sick Child

Review malaria bed nets

Topic of month: nutrition during pregnancy

END VISIT

Day Planning

MKWERA : TB Referral (2 wks)

MKEA: Severe diarrhea (3 days)

CHUMA: late HH visit (3 months)

KAIGILE: routine HH visit

MGANDA: routine HH visit

EXIT

Outline

• BackgroundThe simplicity and complexity of global inequity

• Two examplesPatient record systems for AIDS treatmentMedical algorithms on handhelds

• Conclusion

Conclusion

• Key points– Must understand context– Much potential, many challenges– Keep it simple

• Challenges– Evaluation, local ownership, I2I, duplication of

effort, …

Thank [email protected]