Presentation is loading. Please wait.

Presentation is loading. Please wait.

DECREASING LOST TO FOLLOW-UP : AN INFORMATICS ASSISTED APPROACH (AND HOW TO BUILD A REPORTING APP) Daniel Ochieng Jonathan Dick.

Similar presentations


Presentation on theme: "DECREASING LOST TO FOLLOW-UP : AN INFORMATICS ASSISTED APPROACH (AND HOW TO BUILD A REPORTING APP) Daniel Ochieng Jonathan Dick."— Presentation transcript:

1 DECREASING LOST TO FOLLOW-UP : AN INFORMATICS ASSISTED APPROACH (AND HOW TO BUILD A REPORTING APP) Daniel Ochieng Jonathan Dick

2 AMRS HAS LOTS OF DATA… Majority of AMRS data in two tables: encounter and obs  4.8 million encounters, 180 million observations

3 BUT HOW DO YOU ACCESS YOUR DATA? How do you get data from AMRS right now?  Is there an intermediary who gets you your data?  If obtaining on your own, what is the process? Import into SAS?  Which dataset do you use?  What are your challenges?

4 A NEW WAY TO GET YOUR DATA…. Introducing AMRS Reporting v1.0 A web based application to assist with the reporting of data as well as the creation of tools to improve the management of interventions.

5 CREATING YOUR OWN AMRS REPORTING APP 1.Design the sql to make a derived table of the data you need using SQL. 2.Design the sql queries to get the data you need from your derived table and upload to AMRS Reporting 3.Design the templates you need to display your data. 4.Work with your team to integrate the application into the management of your project

6 CREATING YOUR OWN APP – STEP 1  Design the sql to make a derived table of the data you need using SQL.  Allows for faster querying compared to directly working on the AMRS tables  Will still take a bit of time to create the table – ideally though table creation should be < 1 hour  *May be possible to use another group’s data – will have repository available  Size of tables vary but generally 1 – 2 GB

7 CREATING YOUR OWN APP – STEP 2  Design the sql queries to get the data you need from your derived table and upload to AMRS Reporting  Should be able to be executed in <= 15 seconds  Provides the meat to be used for reporting/intervention

8 CREATING YOUR OWN APP – STEP 3  Design the templates you need to display your data.  Currently done using Django template system  Requires programming  Default system exists to automate display of report data

9 CREATING YOUR OWN APP – STEP 4  Work with your team to integrate the application into the management of your project  Perhaps the trickiest part!

10 CASE STUDY: LOST TO FOLLOW-UP  LTFU definition: Patient has not had contact with any AMPATH service provider in > 90 days from last RTC date  Current LTFU across AMPATH: 30%  Previous study to evaluate LTFU:

11 RESULTS Outcomes from chart reviewsAdults N=2179 Children N=361 Total N=2540 Deceased42 (1.9%)8 (2.2%)50 (2.0%) HIV-negative0 (0%)16 (4.4%)16 (0.6%) Has a recent visit at original AMPATH clinic39 (1.8%)6 (1.7%)45 (1.8%) Documentation of transfer out of AMPATH146 (6.7%)22 (6.1%)168 (6.6%) File missing/insufficient contact info83 (3.8%)8 (2.2%91 (3.6%) Total not tracked352 (16%)65 (18%)417 (16%) Outcomes of tracked patientsAdults N=1540 Children N=260 Total N=1800 Still in care at AMPATH clinic37 (2%)10 (4%)47 (3%) In care somewhere else286 (19%)33 (12%)319 (18%) Deceased363 (24%)12 (5%)375 (21%) Moved away403 (26%)55 (21%)458 (25%) Disengaged from care405 (26%)119 (46%)524 (29%) Other46 (3%)31 (12%)77 (4%)

12 CURRENT METHODS FOR FINDING LTFU  Currently there is no regular identification and follow-up of LTFU patients  Only patients who qualify as “no shows’ as per the program definition are followed up.

13 CURRENT CHALLENGES FOR FINDING LTFU PATIENTS  Appointment lists provided to clinics are incomplete  No efficient way to identify patients who have not been to clinic recently or are about to become LTFU  The program currently has no enrollment protocol in place  Lack of access to community resources even where they exist

14 REDUCING LTFU : WHAT INFORMATION DO WE NEED? Reducing existing LTFU  Obtain list of existing LTFU patients from AMRS Reducing about to be LTFU  Patients scheduled to come to clinic within 30 days  Manual Diaries  Patients scheduled to come to clinic after 30 days  AMRS data Provide a method to measure progress over time  Compare clinics  Measure work habits of outreach workers Incorporate geographical data : catchment from non-catchment

15 TIMELINE FOR DEVELOPING THE LTFU APP… Creation of derived table  ~ 1 – 2 weeks to come up with query  4.8 million rows where every row represents one encounter Creation of queries of the derived table  ~ 2 weeks  Key queries include:  Ampath level LTFU data; By clinic level LTFU data; Ranking query to assess the performance of clinics overtime; about to be LTFU patients Creation of templates to display data  ~ 1 week Total Development time: ~1 month Integration on the ground: about to be started

16 LTFU APP HOME PAGE

17 USING THE LTFU APP AT THE CLINICS PILOT  Focus on the 14 clinics which account for 90% of all LTFU  Will use a sample of these clinics as pilot sites  Pilot to take place over 1 year – expectation/hope is to reduce LTFU by 50%

18 USING THE LTFU APP AT THE CLINIC: CURRENT IDEAS  Focus on preventing LTFU using while slowly doing chart reviews of current LTFU patients  Restructuring of current outreach worker work plan  Focus on preventing LTFU : 70% of patients with recent clinic visits (in past 3-6 months) able to be contacted  “Extra time” to be used to do chart reviews. Will assess over initial month to help understand resource limitations.  Once Chart Review Complete  move to tracking in community

19 CURRENT IDEAS CONT  Development of AMRS forms to document outreach LTFU accounting  Development of AMRS Reporting app to track patients tracked by outreach team  Expand taxonomy to describe patients not coming to clinic  We need a way of categorizing and documenting patients’ multitude of legitimate reasons for not coming to clinic, e.g voluntary disengagement, as this represents a substantial percent of LTFU  Better death documentation

20 FUTURE DIRECTIONS FOR AMRS REPORTING Installation of AMRS Reports onto the AMPATH system (currently only runs on my computer) I am looking for clinicians/researchers who want to develop new apps I am looking for geeks who want to learn how to do code the sql and template design Find me:  jonathan.j.dick@gmail.com jonathan.j.dick@gmail.com  0724 679 898


Download ppt "DECREASING LOST TO FOLLOW-UP : AN INFORMATICS ASSISTED APPROACH (AND HOW TO BUILD A REPORTING APP) Daniel Ochieng Jonathan Dick."

Similar presentations


Ads by Google