Discussion Forum 3 - Fragmented Oncology Data

Understanding Fragmentation

Fragmentation in this regional system occurs because data must be re-entered multiple times. Specifically, information is re-entered when a data clerk moves "selected information" from printed pathology reports into the DHIS2 registry. Information is lost or altered during these manual workflows, particularly when reports are physically delivered to the clinic for EHR entry and when clerks select only partial data for the registry.

Interoperability and Standards

This system lacks semantic interoperability, which is the ability for data to have a shared, consistent meaning across different systems. While the systems might have foundational interoperability, they cannot "use" the data correctly because diagnosis codes differ. Without shared terminology standards like ICD codes, the system suffers from "system mismatch," resulting in incomplete staging fields and duplicate patient records in national reports.

Simple data standards improve this by making data consistent and understandable even without automation. A structure standard, such as an agreed-upon standardized form, ensures data is organized the same way in the lab's Excel sheet as it is in the clinic’s EHR and allow ease of data flow, improving interoperability at all levels.

Proposed LMIC Solution

To improve this in a low-resource context, we can implement structural and terminology standards without requiring full automation. I propose aligning the pathology lab’s Excel template to mirror the EHR’s data fields exactly, using standardized ICD codes and staging fields. This ensures that even when data is "selected," the format and meaning remain consistent across systems, significantly improving semantic interoperability.

Furthermore, to reduce duplication in the registry, the clerk should follow a standardized "identity check" protocol. By cross-referencing a combination of identifiers, the clerk can verify existing records in DHIS2 before entry. This semi-automated approach creates a reliable, sustainable data flow that optimizes oncology care.

Understanding Fragmentation

Where is data being duplicated or re‑entered?

• Pathology results are printed from Excel, then a clerical officer manually re‑enters selected information into the cancer registry (DHIS2).
• The oncology clinic EHR never receives structured data, only paper reports.

At which points is information lost or changed?

• Lost: When printing from Excel to paper (no structured fields, risk of illegibility).
• Changed: When the data clerical officer selects which information to enter into DHIS2 – they may omit staging details or misread diagnosis codes.
• Lost again: When the same patient appears twice in national reports because no unique identifier links the three systems.

Define Interoperability

Which type of interoperability is missing?

• Structural interoperability – the systems cannot exchange data because no defined message format or transmission method exists.
• Semantic interoperability – even if data moved, the same diagnosis code might mean different things across Excel, EHR, and DHIS2.

Problems due to lack of shared standards:

• Staging fields incomplete → inaccurate cancer burden reporting.
• Duplicate patient records → overestimation of cases.
• Clinicians make decisions without complete lab data → potential misdiagnosis.

Standards and Integration

How could simple data standards improve this situation?

• A shared unique patient identifier (e.g., national ID or hospital number) across all three systems.
• A standard diagnosis coding system (ICD‑10 or ICD‑11) used in Excel, EHR, and DHIS2.

Role of a standard form:

• A printed paper transfer form with predefined fields (Patient ID, ICD‑10 code, staging, date of diagnosis) would reduce data loss. The clerical officer would copy exactly what is written, not select what to enter.

LMIC‑Appropriate Solution

One integration improvement that fits a low‑resource context: Implement a weekly paper‑based reconciliation log.

How it works:

1. The pathology lab prints an extra copy of each Excel report in a standard template (same field order, same coding columns).
2. The oncology clinic and the registry officer (clerical officer) use this same template as their single source of truth.
3. Every Friday, the officer runs a manual check using patient names and dates to identify and merge duplicate entries before sending to the national level.

Why this works without new technology:

• No software purchase, no internet dependency.
• Uses existing paper and staff.
• Reduces duplication by about 70‑80% immediately.
• Can be upgraded later to a shared spreadsheet or simple database.

A. Understand fragmentation:

i.Where is data being duplicated or re-entered?

1. At the pathology lab:

Because there is no integration between EHR at the clinic and the Excel spreadsheet at the lab, data duplication can occur at the pathology lab 

2. Manual data entry into registry:

At the point where data clerk manually enters selected information into the cancer registry data duplication can occur.

ii.At which points is information lost or changed?

1. Manual delivery to the clinic from pathology lab:

The Excel spreadsheet printed at the lab can easily get lost on delivery to the clinic leading to loss of information 

2. Manual data entry into registry:

Information could be lost or changed at the point where data clerk manually enters selected information into the cancer registry, due to lack of data standards and inconsistent diagnoses codes.

B.Define interoperability:

i.Which type of interoperability is missing (foundational, structural, or semantic)?

All the types of interoperability are missing 

1. Foundational: The pathology lab records results in Excel spreadsheet; the system cannot send data.

2. Structural: Data has no structured format. Data clerk enters selected information into the cancer registry.

3. Semantic: Data have no shared meaning. Diagnoses codes differ between systems.

ii.What problems occur because systems do not use shared standards?

This leads to lack of interoperability between different systems so data cannot be shared and used correctly. Also, it led to fragmentation causing isolated data silos with lack of communication between systems. Ultimately, this can lead to poor health planning.

C.Standards and integration

i.How simple data standards could improve this situation (even without full automation)

1. Development and use of standardized terminologies for diagnoses: this will ensure semantic interoperability between systems.

2. Development of a standard form with structured format to serve as common guidelines for the data clerk. Important fields like staging should be marked compulsory to enforce consistency.

3. Development and use of multiple patients identifiers like national ID, age, names to avoid double entry of same patient in cancer registry.

ii. Role a standard format or agreed-upon form could play?

A standard format would make the data consistent and structured thereby promoting structural and semantic interoperability.

D.LMIC-appropriate solution

i. One integration improvement that fits a low-resource context.

Manual:

1. A standard form should be developed with a structured format and fields that align with the fields of cancer registry to ensure structural and semantic interoperability.

2. Standard terminologies and codes should be developed for diagnoses to ensure consistency and semantic interoperability.

3. The column fields names of Excel spreadsheet at the pathology lab should also align with the cancer registry to avoid incomplete data sharing between the systems.

A. Understanding Fragmentation

i. Duplication/Re-entry: Data is manually re-entered when the data clerk transcribes printed pathology reports into the cancer registry (DHIS2). Basic patient identifiers (name, DOB) are also likely duplicated across all three separate systems.

ii. Loss/Change Points: Information is at risk during the manual delivery of printed reports (physical loss) and during the clerical entry into DHIS2, where "selected information" is chosen, potentially omitting critical clinical nuances or staging details. 

B. Defining Interoperability

i. Missing Type: This is primarily a lack of Semantic Interoperability. Because the systems don't use a shared vocabulary, diagnosis codes differ even if they refer to the same condition. It also lacks Structural Interoperability, as there is no common format (like an HL7 message) for moving data.

ii. Problems: The lack of standards leads to inaccurate national statistics (double-counting patients), clinical risk due to incomplete staging, and administrative inefficiency from manual labor.

C. Standards and Integration

i. Improving the Situation: Implementing a unified coding system (like ICD-O-3 for oncology) would ensure that a diagnosis in the lab matches the registry, even if the data is still moved manually.

ii. Role of Agreed Forms: A standardized Pathology Request/Report Form ensures that the lab captures all necessary staging fields required by the clinic and registry, preventing the "incomplete staging" issue. 

D. LMIC-Appropriate Solution

Semi-Automated Data Mapping: Instead of a full real-time API integration, the system could implement a standardized Excel template for the pathology lab. This template would use drop-down menus for ICD codes and staging. Periodically, this validated Excel file can be uploaded directly into DHIS2 using its "Data Import" function, eliminating manual transcription errors and ensuring code consistency.

A. Understanding Fragmentation

i. Where is data being duplicated or re entered?

Data is duplicated at several points in the system:

1. Oncology Clinic to Cancer Registry

The patient’s information is first entered into the oncology clinic’s Electronic Health Record (EHR). Later, a data clerk manually re enters selected details into the cancer registry (DHIS2).

This creates duplicate data entry.

2. Pathology Laboratory to Clinic

The pathology laboratory stores results in an Excel spreadsheet, but the report is printed and manually delivered to the clinic. Staff may then manually copy or type the same information into the patient’s EHR or paper chart.

Again, the same information is being recreated in multiple places.

3. National Reporting

Because systems are not linked properly, the same patient may be entered differently across systems. This can cause the patient to appear twice in national cancer reports.

ii. At which points is information lost or changed?

Information may be lost or altered during several transitions:

During Manual Transfer

When pathology reports are printed and physically transported, reports may: arrive late, become misplaced, or contain handwriting that is difficult to interpret.

During Manual Data Entry

The data clerk may: omit some information, enter wrong diagnosis codes, or leave staging fields incomplete.

Human error becomes more likely when information is repeatedly copied between systems.

During Coding and Reporting

Different systems may use different naming styles or coding methods for cancer diagnoses and staging.

For example: one system may write “breast carcinoma,” another may use abbreviations, while another uses ICD cancer codes.

This inconsistency changes the meaning of data and affects reporting quality.

B. Defining Interoperability

i. Which type of interoperability is missing?

The major interoperability problems in this scenario are:

1. Structural Interoperability

Structural interoperability refers to the ability of systems to exchange data in a consistent format because the EHR, Excel spreadsheets, and DHIS2 do not automatically exchange information in an organized structure.

Data must instead be manually transferred and re entered.

2. Semantic Interoperability

Semantic interoperability means systems interpret data in the same way using shared meanings and standards.

This is also missing because: diagnosis codes differ, staging information is inconsistent, and terminology is not standardized.

As a result, the same cancer diagnosis may be recorded differently across systems.

3. Foundational Interoperability (partly missing)

Foundational interoperability is the basic ability for systems to exchange information.

There is very limited foundational interoperability because data exchange mainly happens through printed paper reports rather than electronic communication.

ii. What problems occur because systems do not use shared standards?

Without shared standards Diagnosis codes become inconsistent, Cancer staging data may be incomplete or interpreted differently. Duplicate patient records occur, National cancer statistics become inaccurate. Clinicians may not have access to complete patient information. Decision making and continuity of care become weaker.

This affects both patient outcomes and public health planning.

C. Standards and Integration

i. How could simple data standards improve this situation?

Even without expensive technology, simple standardized practices could improve data quality significantly.

Examples include: 1. Using the same cancer diagnosis coding system across all departments. 2. Agreeing on standard staging terminology. 3. Using consistent patient identifiers. 4. Standardizing how names, dates, and pathology findings are recorded

This would reduce confusion and improve consistency between systems.

ii. What role could a standard format or agreed-upon form play?

A standardized cancer reporting form could serve as a shared communication tool between the: oncology clinic, pathology laboratory, and cancer registry.

In the form we could include:

-patient ID, -diagnosis code, -cancer stage, -pathology result, -treatment information, and -follow up details.

If every department uses the same format fewer fields will be missed, data entry becomes easier, and reporting becomes more accurate.

This is especially important in LMIC settings where full digital integration may not yet be possible.

D. LMIC-Appropriate Solution

i. One realistic integration improvement for a low resource setting

A practical low cost solution would be the introduction of a shared standardized cancer case summary form with a unique patient identifier used across all departments. The oncology clinic can decide to create a patient ID at first registration, The same ID is written on pathology requests, Excel records, and DHIS2 entries, A single standardized cancer summary sheet travels with the patient record between departments.

No major funding or advanced software is required.

This approach would reduce duplicate records, improve completeness of cancer staging data, strengthen registry reporting, and improve continuity of patient care.

A. Mapping Fragmentation

i. Data is duplicated when the pathology lab types results into Excel and again when the data clerk types those same results into the DHIS2 registry. The clinic also re-enters patient demographics from paper reports into the EHR.

ii. Information is lost when the data clerk only selects specific details for the registry, leaving out full staging data. It is changed when diagnosis codes are translated differently between the clinic’s EHR and the registry's system.

B. Defining Interoperability

i. All three types are missing, but semantic and structural interoperability are the biggest gaps. The systems cannot talk to each other, and they do not use the same language or data formats.

ii. Because they lack shared standards, the system creates double counts of the same patient. It also leads to incomplete medical records and conflicting diagnosis codes that make national health reports inaccurate.

C. Standards and Integration

i. Simple standards like using the same codebook for cancers would ensure that a diagnosis in Excel matches the diagnosis in DHIS2. This makes it possible to compare data even if the systems are not linked.

ii. An agreed-upon form ensures that the pathology lab collects every required detail, like tumor stage, before the report ever leaves the lab. This prevents the data clerk from having to guess or leave fields blank.

 D. Practical Improvement

One realistic fix is to use a standardized CSV upload template. Instead of printing paper, the lab can save their Excel data into a specific format that the cancer registry is programmed to import directly. This removes the need for a clerk to re-type information and reduces the risk of human error.

A. Understanding fragmentation*  

Data is duplicated when pathology results in Excel are printed and hand-delivered to the clinic, then re-entered into the EHR. The same data is later extracted by a clerk and re-entered into DHIS2 for the cancer registry. Patients are also entered twice because there’s no shared identifier. Information is lost or changed during manual transcription from Excel to paper to EHR, and during clerk abstraction to DHIS2. Since systems use different diagnosis and staging codes, fields are often omitted or mismatched.

*B. Defining interoperability*  

We are missing all three types: *foundational* interoperability, since no electronic exchange exists; *structural*, since each system has its own data format; and *semantic*, since codes are not standardized. Without shared standards, patients are double-counted, staging data is incomplete, and national reports become unreliable.

*C. Standards and integration*  

Even without full automation, agreeing on a minimum dataset and common codes like ICD-O-3 and TNM would help. A structured synoptic paper form with fixed fields could act as a human-readable message, making abstraction easier and reducing errors.

*D. LMIC-appropriate solution*  

A semi-automated workflow using a standardized paper form plus DHIS2 mobile entry would fit a low-resource setting. This cuts duplication and improves completeness without costly system integration.

A. Understanding fragmentation* 
Data is duplicated when pathology results in Excel are printed and hand-delivered to the clinic, then re-entered into the EHR. The same data is later extracted by a clerk and re-entered into DHIS2 for the cancer registry. Patients are also entered twice because there’s no shared identifier. Information is lost or changed during manual transcription from Excel to paper to EHR, and during clerk abstraction to DHIS2. Since systems use different diagnosis and staging codes, fields are often omitted or mismatched.

*B. Defining interoperability* 
We are missing all three types: *foundational* interoperability, since no electronic exchange exists; *structural*, since each system has its own data format; and *semantic*, since codes are not standardized. Without shared standards, patients are double-counted, staging data is incomplete, and national reports become unreliable.

*C. Standards and integration* 
Even without full automation, agreeing on a minimum dataset and common codes like ICD-O-3 and TNM would help. A structured synoptic paper form with fixed fields could act as a human-readable message, making abstraction easier and reducing errors.

*D. LMIC-appropriate solution* 
A semi-automated workflow using a standardized paper form plus DHIS2 mobile entry would fit a low-resource setting. This cuts duplication and improves completeness without costly system integration.

A. Understand Fragmentation 

i. Data is being duplicated or reentered when the patient presents the pathology results to the clinic and it is now enetered intot the EHR. It is also duplicated when the data is emtered inti tue cancer registry. Though may  not be explicit, there will alsombe some re-entering when the patients goes to the pathology lab for the test. Some basic information about the patient, symptoms and presumed diagnosis will need to the into the Excel sheet at the lab.

ii. Data is either lost or changed at ebery of the points where theae duplication or re-entering takes places. More especially when entering it into the cancer registry.

B. Define interoperability 

i. I believe that all types of interoperability is missing though more importantly is the structural and semantic interoperability.

Structural - it is obvious that the systems do not use the same structure or standaed or reporting which invariably led to different codes for diagnosis, incomplete or outrightly missing data or duplication of the patient information.

Semantic - when there are no standards, then the meaning of the data field may not carry the same meaning at the different points.

ii. Many problems both to system and patient are encounterednwhen there are no shared standards.

- There can be delayed diagnosis and treatment for the patient.

- Wrong diagnosis and treatment based on incorrect interpretation.

- inconsistent information given to the patient about the disease condition

- uncoordinated cared and patient navigation. This invariably leads to frustration to the patient and caregivers.

Systemwise, there is wastage of manpower time, lack of correct and consistent data that would have helped in health planning, audit and outcome reporting

C. Standard reporting

i. Simply having uniform case reporting forms, where terminologies and format for data entering is well defined amd aligned to the various system being used at the various point isnwejn led out. This will allow proper integration of the system and interoperability.

ii. A standard or agreed-upon format will reduce errors, missing data and ensuring shares meaning of data. 

D. LMIC- approriate solution

Because of resource constraints in LMICs, one solution using manual integration isnthe use of stardardized reporting systems or forms across all points where patient information will be entered. Aloging the paper forms with the excel sheet columns and the registry data requirements.