INTEROPERABILITY TO SUPPORT PATIENT CARE

THE WESTERN CAPE PROVINCIAL HEALTH DATA CENTRE

1

Andrew Boulle

Provincial Health Data Centre

Department of Health, Western Cape

Outline

2

Provincial Health data Centre

The architecture and processes

Examples of outputs

Unique health ID in Western Cape Province

3

Hospital system (Present in 53/54 hospitals,

including all in metropolitan area)

Patient Master Index (PMI)

Primary care (Patient registration

systems in all 307 clinics)

Clinical systems recording electronic data

against shared registration

number (PMI)

• Laboratory

• Pharmacy

• ART registers

• TB registers

• Maternity system(s)

• Discharge summaries

Health

Info

rmatio

n E

xchange

PMI

Mortality and birth surveillance • Province-wide mortality surveillance system for all deaths and stillbirths

• Province-wide birth registration system

Clinical audit tools • Perinatal (PPIP)

• Child Health (CHIP) PMI

PMI

PMI

PMI

PMI

PMI

PMI

PMI

PMI

PMI

Disease

monitoring

systems

(eg HIV / TB)

Laboratory

and

pharmacy

data

Hospital and

primary care

registration

systems

Un

iqu

e id

en

tifie

r

Population

register

Many other

systems

Health

information

exchange

or

Data Centre

or

Whatever

you want to

call it

Clinical

viewing

Care

cascades and

operational

reports

Management

reporting

Alerting

engine (eg. NMC’s)

Epi analyses

Business

intelligence

Research

support and

stewardship

Outline

5

Provincial Health data Centre

The architecture and processes

Examples of outputs

Ho

ldin

g

Co

mm

un

ity

Faci

lity

Source archive

Hospital IS

PHC IS

TIER/ETR/EDR (disease IS)

Pharmacy

Laboratory

PACS/ECM

Mortality

Clin

ical

P

atie

nt

Stag

ing

Child death audits

Partner systems

mHealth Op

enH

IM

(rea

l-ti

me)

D

aily

or

pe

rio

dic

dat

abas

e p

ulls

o

r lo

adin

g o

f e

xpo

rts

Person

Place

Code

Core tables

Lookup tables

Mapping tables

Encounters

Episodes

Lin

kin

g vi

ews

(N

o c

linic

al d

ata)

A

nal

ysis

vie

ws

(an

on

ymo

us)

Sto

red

pro

ced

ure

laye

r (f

or

inte

rfac

e en

gin

es)

Op

enH

IM

Ap

plic

atio

ns

(e

g. S

PV

, PH

CIS

)

Op

erat

ion

al r

epo

rts

(eg.

Sh

arep

oin

t)

Bes

po

ke e

xtra

cts

(f

or

anal

yses

/req

ues

ts)

Pre

-pro

cess

ed

casc

ades

Rep

ort

ing

and

an

alys

is s

ervi

ces

HIGH LEVEL ARCHITECTURE OF THE PROVINCIAL HEALTH DATA CENTRE THE PHDC

Processes - curation

7

Facility Mapping Example

8

Sinjani ID Sinjani Name Source ID Source Name Data Source

82 Bishop Lavis CDC 990598 BISHOP LAVIS CHC CDU

82 Bishop Lavis CDC 73197343 BISHOP LAVIS CHC CDU

82 Bishop Lavis CDC 197343 BISHOP LAVIS CHC CDU

82 Bishop Lavis CDC 73197343 BISHOP LAVIS CHC CDU

82 Bishop Lavis CDC BL Bishop Lavis CHC JAC

82 Bishop Lavis CDC BFD5562A-88F0-410A-9947-799F4CC0B313 wc Bishop Lavis CDC TIER

82 Bishop Lavis CDC BLP bishop lavis cdc EKAPA

82 Bishop Lavis CDC 9 bishop lavis cdc EKAPA

82 Bishop Lavis CDC BISDH bishop lavis day hospital DISA

82 Bishop Lavis CDC 91B014 bishop lavis cdc wc blp TRAK

82 Bishop Lavis CDC BLP BLP MED REC CLINICOM

82 Bishop Lavis CDC 02FF8B45-22C9-4D07-8590-F9D227F3B080 WC BISHOP LAVIS CDC ETR.NET

82 Bishop Lavis CDC Bishop Lavis CDC Bishop Lavis CDC PHCIS

82 Bishop Lavis CDC BLP bishop lavis cdc PHCIS

82 Bishop Lavis CDC BLP BLP MED REC PHCIS

82 Bishop Lavis CDC 9 bishop lavis cdc PHCIS

82 Bishop Lavis CDC BLP Bishop Lavis CDC PHCIS/EKAPA

82 Bishop Lavis CDC 642333 wc Bishop Lavis CDC MOMCONNECT

SNAPSHOTS FROM THE DAILY LOAD 28-09 NHLS

LABS

Subsection Query Result

ROWS Number of rows imported into Staging 666 182

Number of rows imported into Clinical 182 070 Number of rows flagged as error rows and added to 0

staging.temp.labs_raw_errors:

Number of rows imported into Staging (Antibiotic resistance) 6 850

Number of rows imported into Clinical (Antibiotic resistance)

6 850

FAC Number of distinct facilities 434

Number of mapped facilities 433 (99.77 %)

Number of facilities sent for curation (unseen) 0

PMI Number of distinct patients 24 456

Number of patients mapped using pmi_links history (previously mapped)

21 406 (87.53 %)

Number of patients with exact matches 2 303 (9.42 %) Number of patients with highly probable matches 193 (0.79 %) Number of patients with possible matches 122 (0.50 %)

Number of unmapped patients 432 (1.77 %) Number of unmapped patients by environmental samples 17 (3.94 %)

Number of unmapped patients by PMI facility 110 (25.46 %)

Number of unmapped patients by non-PMI facility 306 (70.83 %) Number of distinct patients (Antibiotic resistance) 367

Number of patients mapped using pmi_links history (previously mapped) (Antibiotic resistance)

362 (98.6%)

Number of unmapped patients (Antibiotic resistance) 5 (1.4%)

Number of unmapped patients by PMI facility (Antibiotic resistance) 2 (40%)

Number of unmapped patients by non-PMI facility (Antibiotic resistance)

3 (60 %)

88% of records for existing patients

10.2 % of records match to PMI

1.8% unmapped, of which 71% from NON-PMI facilities

Adding 150k-200k lab results per day for 20k patients including 2000 new patients

Common Concept Example: TB Microscopy

10

Test Code

Test Description

Parameter Code

Parameter Description

Common Group

Common Concept

Common Method

TBA TB MICROSCOPY TBACO RESULT TB Microscopy Result Auramine

TBA TB MICROSCOPY TBAC1 NOT AVAILABLE TB Microscopy Result Auramine

TBCUL TUBERCULOSIS INVESTIGATION TBAUR AURAMINE TB Microscopy Result Auramine

M176 AURAMINE (WITH TB CULTURE) M0025 AFB (CONCENTRATED) TB Microscopy Result Auramine

M176 AURAMINE (WITH TB CULTURE) M0024 AFB (RAPID UNCONCENTRATED) TB Microscopy Result Auramine

M005 AURAMINE O STAIN M0024 AFB (RAPID UNCONCENTRATED) TB Microscopy Result Auramine

M005 AURAMINE O STAIN M0025 AFB (CONCENTRATED) TB Microscopy Result Auramine

TBZ TB MICROSCOPY TBZCO ZN RESULT TB Microscopy Result Ziehl-Neelsen

TBCUL TUBERCULOSIS INVESTIGATION TBZN ZIEHL - NEELSEN TB Microscopy Result Ziehl-Neelsen

M195

AFB (+VE RE-

DECONTAMINATE) M0201 ZN RESULT (POSITIVE CULTURE) TB Microscopy Result Ziehl-Neelsen

M185 AFB (POSITIVE MGIT CULTURE) M0201 ZN RESULT (POSITIVE CULTURE) TB Microscopy Result Ziehl-Neelsen

M185 AFB (POSITIVE MGIT CULTURE) M0252 ZN RE-INCUBATED TB Microscopy Result Ziehl-Neelsen

B160 NICD: AFB (POSITIVE CULTURE) M0201 ZN RESULT (POSITIVE CULTURE) TB Microscopy Result Ziehl-Neelsen

B161 NICD: DST (POSITIVE CULTURE) B0065 ZN RESULT TB Microscopy Result Ziehl-Neelsen

M181 TB CULTURE SOLID MEDIUM M0201 ZN RESULT (POSITIVE CULTURE) TB Microscopy Result Ziehl-Neelsen

M181 TB CULTURE SOLID MEDIUM M0252 ZN RE-INCUBATED TB Microscopy Result Ziehl-Neelsen

M010 ZIEHL-NEELSEN STAIN M0024 AFB (RAPID UNCONCENTRATED) TB Microscopy Result Ziehl-Neelsen

M010 ZIEHL-NEELSEN STAIN M0025 AFB (CONCENTRATED) TB Microscopy Result Ziehl-Neelsen

Disease Information Systems

11

• DIS data is received quarterly

• When received it is incorporated into existing data structures to supplement data received daily

• As the coverage of PHCIS/EKAPA increases we will start to receive the DIS data more real time

TIER ETR EDR PPIP

Processes - beneficiation

12

Data beneficiation and enrichment: encounters, episodes and

cascades

Encounters

13

• Facility visit ≈ Encounter

patient + facility* + date + time*

• Encounter types & Sources:

Admissions: One line per admission with admission & discharge date [CLINICOM]

OP Hospital Visits: One line per OP clinic, date & time [CLINICOM]

PHC Visits: One line per patient, facility & date (time not reliable) [PHCIS/EKAPA/PREHMIS*]

Dispensing: One line per patient, facility & date. Inferred visits from drug dispensing. [JAC]

Lab visits: One line per patient, facility & date. Inferred visits from taken labs. [NHLS]

DIS Visits: One line per patient, facility & date. Inferred visits from DIS treatment dates. [TIER/ETR]

Community: Self enrolment dates [MomConnect]/Community registrations [Catch&Match]

Linking community to facility: Catch and Match example

14

Referral

to Health

services,

social

services

etc.

*MATCHCHW mobile mHEALTH

platform

Interoperability

layer PHDC

Transversal Web

Interface: S/W, EHOClient

attended to?Referral

to Health

services,

social

services

etc.

CHW mobile mHEALTH

platform

Episodes

15

Drugs

dispensed

Lab tests Diagnosis

codes

Wards &

specialities

Disease

Information

Systems ?

• Evidence from clinical data enables us to infer health conditions

Health conditions

• Episodes can be acute/chronic

• We combine evidence from clinical data into a single record per

patient per episode

• Includes: First & Last evidence dates, treatment start dates and

outcome dates, as well as facilities, datasources & a list of

evidence

Datasources used to ascertain pregnancy

16

•Birth outcomes (e.g. live birth, still birth etc) Birth records

(CLINICOM, PHCIS and PPIP)

•Rhesus Testing, Pap Smear with pregnancy parameter

•Supporting evidence: BETA HCG, Syphilis

Laboratory

(NHLS)

•Terminations by mifepristone

•Supporting evidence: Iron & Folate, Metaclopramide Pharmacy

(JAC)

•ICD10 codes

•Supporting evidence: Ward or OP clinic type, Speciality codes, internal codes describing admission/discharge methods

Admissions &OP visits

(CLINICOM)

•Mom connect registrations Community

(MOM CONNECT, CATCH&MATCH)

•MOU Visits (weak-moderate) PHC Visits

(PHCIS, EKAPA, PREHMIS*, inferred visits from JAC/NHLS)

SOURCE EVIDENCE (bold=high confidence)

Episode evidence categories

17

•Strong enough evidence to start a pregnancy episode

•High confidence score

High confidence

•Evidence of an outcome for the episode

•High confidence score Outcome

•Able to start an episode, however more evidence is required to improve confidence in the episode

•>=5 High confidence score.

•<5 Low confidence score.

Weak-Moderate evidence

•Can only be appended to existing episode

•It is not strong enough to start an episode

•Improves confidence in existing episodes

Supporting only

Inferring pregnancy episodes

18

Evidence Source Category Date

Rhesus test performed Labs High confidence 2015-01-23

Iron & Folate dispensed Drugs Supporting 2015-04-03

Has live birth record Birth register Outcome 2015-06-16

Has diagnosis code indicating live

birth

Diagnosis codes Weak-Moderate 2015-06-16

Example evidence list

Patient Pregnancy

number

Start Date End Date Last

Contact

date

Evidence list End date

evidence list

Facility Confidence

xxx 1 2015-01-23 2015-06-16 2015-06-16 Birth Record,

Diagnosis

code, Rhesus

Test, Iron &

Folate

Birth Record,

Diagnosis

code

MMH 0.95 High

confidence

Rolled up into a single record

per pregnancy (using a sliding date range of at most 270 days between the first and last

contact date)

Rolling Episode counts (as of 2017-01-01)

19

Episode Type High confidence episodes [Total*]

HIV positive 347 550 [412 370]

HIV positive on ART 245 024

Diabetes Mellitus 246 387 [246 896]

Hypertension 407 738

Asthma or COPD 187 344

Epilepsy 30 471

Pregnancy 96 813 [114104]

TB positive 52 088 [58 646]

TB on treatment 33 443 *[39816 with extra rollback to 2016-06-01]

*All episodes where the high confidence count is different from the total count Note that for chronic episodes, counts are calculated based on patients who have no death record and have had any encounter within 2 years before the report date. For HIV, the dates were shifted earlier by 6 months to account for the lag in disease information systems. For HIV, the number on treatment is based on those with an ART start date and any encounter with TIER/EKAPA in the last 18 months. For Pregnancy & TB, counts are calculated based on patients who started their episode within 1 year before the report date. For TB, the dates were shifted earlier by 6 months to account for the lag in disease information systems. For TB, the number on treatment is based on those registered in EDR/ETR.

Cascades

Vital status outcomes

Episode-specific outcomes

Service utilisation

outcomes

Episode

of

Interest

• Start date • Treatment

date • Last date • Outcome

• Selected attributes

+ Other episodes (co-morbidities)

+ + +

=

Ca

sca

de

• Comprehensive patient level view of episodes, including key dates, evidence,

outcomes and co-morbidities relevant for patient management

• Provides us with a cohort in the absence of complete registers

• Optimised for operational and management purposes

• Easily aggregated by region or attributes

Cascades| Current & future

21

• Maternal and infant, including PMTCT

• HIV

• TB

• Diabetes

• Cervical health (Pap smear & Colposcopy)

• Antimicrobial resistance / Hospital acquired infections

• (Notifiable conditions)

• (Asthma / COPD)

• (Cancers)

• (Child health chronic conditions)

Processes - reporting

22

Report categories / generic types

Patient Mx report

Service Mx report

Dashboard Anonymised data

PHDC admin reports - episode logic - who has access to what - usage statistics - etc.

Single patient viewer for clinical care

Outline

24

The intervention – a Provincial Health data Centre

The architecture and processes

Examples of outputs

25

Examples of findings: HIV treatment

PMTCT drill down example, back up slides in case live demo fails. 1) Positive PCRs, most recent yesterday

28

2) SPV for 1st mother in the list

29

Graphical viewer of HIV episode. Third pregnancy probably artefact and a post-natal visit linked to second. Note Momconnect registration in green

30

3) SPV for child

Disease monitoring

systems (eg HIV / TB)

Laboratory and pharmacy data

Hospital and primary care registration

systems

Un

iqu

e id

en

tifi

er

Population register

Many other systems

Health information

exchange

or

Data Centre

or

Whatever you want to call it

Clinical viewing

Care cascades and operational

reports

Management reporting

Alerting engine (eg. NMC’s)

Epi analyses

Business intelligence

Research support and stewardship

Patient protection and data stewardship

+

Unique ID

1. National systems HPRS and NHLS

+ 3. Clinic systems in Province

and City 4. Hospital systems

including discharge summaries

5. Health Information Exchange (PHDC)

6. Cascade reports and Single Patient Viewer

2. Community Health Worker systems

Provincial / national alignment

Opportunities for public-private interface

For direct patient care Safely share fact of shared health record with private sector HIE or individual

health insurance schemes (no health data shared)

Create mechanism for data to be requested or accessed with patient consent

Share with public sector presence of shared health record on private sector

systems – patient can consent to provide access

Analyses of patient movement between public and private Eg. Immunisations, maternity care

One-way encryption of ID numbers, linked anonymously by trusted third-party

Province-wide burden estimation: HIV, diabetes, renal failure, etc.

Verification of service provision for reimbursement Eg. Chronic medicine delivery – can provide an API for private provider to

verify that a patient is scheduled to collect medicines through private provider;

notify public sector that collection has taken place with patient co-verification

Strengths and weaknesses around interoperability

34

Public-sector Strength

Platform-wide data is truly platform-wide, and single-provider

Some clinical systems include clinical data, not just claims data

Weakness

Not everything is digitized

Quality of coding suboptimal as usually not linked to reimbursement

Cannot insist on national ID numbers or equivalent as condition for service provision

Private sector Strengths

Claims data verified as basis of payment

High completion of verified identifiers for data linkage

Weaknesses

Lack of continuity of data, as patients move in and out of cover, and use other services, out-of-pocket or public

Limited population perspectives

Limitations of claims data

Challenges sharing data between providers and insurers

Thank you. Questions?

35

Acknowledgements Provincial Health Data Centre staff, collaborators and executive

Colleagues in WC DoH Information Management and Health Impact

Assessment, Chief Directorate Strategy and Support

Jembi Health Systems

Long-standing investments by many stakeholders in core information systems

over the past 20 years.

Additional funding: NICHD and BMGF