Data Classification Methodology: The Way Forward to Data-Enabled Healthcare in Egypt

Abstract

A robust data management system is crucial for healthcare entities to get value from the exponential volume of generated and stored data. Given the impact of data on the decision-making process at the individual, organizational, and institutional levels, appropriate management of data should be considered a priority. Data management is referred to as the activities and processes that are directly related to the actual management of data, whereas data governance is a strategic framework concerned with managing rules, responsibilities, processes, and practices of managing the data. Data management can turn this data classification framework into executable policies and procedures. The main aim of this article is to formulate a proposal or road map for data classification methodology, with a primary focus on the first step of data classification, which is categorizing different types of data collected and released from various ecosystems within the Egyptian healthcare system.

INTRODUCTION

Different organizations have started to recognize the value of data, leading to effective data management becoming a priority. Data management refers to the processes and systems involved in collecting, storing, organizing, and using data efficiently and securely. It includes data classification, governance, and quality control, all of which are essential for extracting valuable insights and enabling evidence-based decision-making. Given the urgent need for innovative solutions to address the problems the Egyptian healthcare system faces, maintaining sustainability without investing in the digitization of the healthcare system is nearly impossible.[1]

Data security management is the data management function that is responsible for ensuring that data are properly protected. Policies and procedures must be formulated for achieving data security, including data classification, access, authorization, data auditing, and treatment of sensitive data. Classifying data is a crucial step in managing the enormous amount of data generated by the Egyptian healthcare sector for better data security, privacy, compliance with regulations, and enhanced usage and integration of data for better patient outcomes and decision-making processes. Data classification identifies the most sensitive information that must be secured from unauthorized access and selects the most appropriate means of access control.

Compared to the private sector, there is a trend of “overclassifying” data in the public sector, which hinders sharing information within and across organizations, leading to irrational decisions and increased costs of data security and classification. This article outlines the data classification steps, handling, and processing within the healthcare sector through shedding light on different data types within the Egyptian healthcare system, and the level of sensitivity of data generated and stored in each to reach data-enabled healthcare.

A framework that outlines the accountability and guidelines for managing various types of data while protecting data security and privacy is crucial for overcoming the barriers to data exchange among these bodies and improving decision-making. Classifying data is usually the first step in maintaining a data-enabled healthcare system. Data classification is the process of organizing data into categories based on their content, type, sensitivity, and value. Data classification is a key aspect of data governance, which is a comprehensive approach to ensuring the quality, availability, integrity, security, and utilization of data in personalized medicine and other areas to generate patient centric outcomes.

In alignment with Egypt Vision 2030, the Government has embarked on building an all-encompassing vision of a “Digital Egypt,” establishing the groundwork for facilitating digital transformation across diverse sectors in Egypt. A key dimension of this initiative involves the formulation of a data classification methodology to facilitate the commencement of the digitalization process, which is emphasized by the Ministry of Health (MoH) as a key aspect of data governance. According to this methodology, data classification pertains to the compartmentalization of data into three distinct categories based on their content, type, sensitivity and value, which is vital for implementing appropriate security measures and ensuring compliance with data protection standards.

Support for this initiative is provided by the Personal Data Protection Law (PDPL; Law No. 151 of 2020), which creates a legal framework for the protection of personal data processed electronically. It lays out essential guidelines for the collection, processing, and storage of personal data, empowering individuals to maintain control over their own information. A significant aspect of this legal framework is the establishment of the Personal Data Protection Center, which is the executive branch tasked with overseeing compliance, providing guidance, and ensuring the effective implementation of data protection measures across various sectors. This legal framework draws inspiration from international standards, particularly the European General Data Protection Regulation (GDPR), which is renowned for its rigorous approach to privacy rights and data security. By instituting a comprehensive legal framework for data protection, an executive body to oversee compliance, and a classification methodology for data collection and storage, the Ministry of Communications and Information Technology (MCIT) has developed a multifaceted approach to ensure a secure and effective digitalization process in Egypt.[2] The main aim of this analysis is to formulate a proposal/road map for data classification methodology, with a primary focus on the first step of data classification, which is categorizing different types of data collected and released from various ecosystems within the Egyptian healthcare system, based on their sensitivity, and assigning them to the most appropriate classification level.

DATA CLASSIFICATION IN THE EGYPTIAN HEALTHCARE SYSTEM

The data classification process can be summarized into five key steps as illustrated in Figure 1.[1,2] The most common data classification model is usually based on the principle of data confidentiality and the risk of its disclosure. This model is the most widely used classification in the public sector and relevant to the Egyptian sector, validated by MCIT and MoH. Data are typically categorized into public, internal, confidential (or highly confidential), regulatory, restricted, or top-secret categories based on their sensitivity and risk of disclosure.[2] Following the aforementioned methodology for data classification, healthcare data classification starts with identifying various types of generated data within the different organizations in Egypt (Figure 2).

Figure 1.

Data classification process.

Figure 2.

The main public healthcare organizations in Egypt.

EDA: Egyptian Drug Authority; EHA: Egyptian Healthcare Authority; GAHAR: The General Authority for Healthcare Accreditation and Regulation; HIO: health insurance organization; MoH: Ministry of Health; UHIA: Universal Health Insurance Authority; PTES: Program for Treatment at the Expense of the State; UPA: The Egyptian Authority for Unified Procurement, Medical Supply, and Management of Medical Technology.

According to the Egyptian PDPL, personal data are defined as any data pertaining to a natural person who has been identified or can be potentially identified, either directly or indirectly, by referring to other identifiers such as the name, voice, image, among others. On the other hand, sensitive personal data are referred to as data including the subject’s psychological, mental, or physical health; financial data; religious doctrines; political thoughts; or children’s data (Figure 3 shows the different public data types in Egypt, their uses, and their multiple sources).

Figure 3.

The different data types and sources in Egypt.

CROs: clinical research organizations; EDA: Egyptian Drug Authority; EHA: Egyptian Healthcare Authority; HIO: health insurance organization; MoH: Ministry of Health; NGOs: nongovernmental organizations; PTES: Program for Treatment at the Expense of the State; UHIA: Universal Health Insurance Authority; UPA: The Egyptian Authority for Unified Procurement, Medical Supply, and Management of Medical Technology.

It is crucial to highlight that oversight of data integrity and security does not fall under the MoH but rather under the purview of MCIT, necessitating collaboration between both institutions. Both data security and integrity pose significant challenges to the establishment of a data-enabled healthcare system (discussed later) and require robust regulation for progress. The following section details the data types and their classification level that are recommended to improve their current practice, as most of these data are underused in Egypt.

DATA TYPES IN EGYPT AND THEIR CLASSIFICATION LEVEL

Research Data

Clinical research encompasses all sorts of research that involves human beings. Medical research types include epidemiologic, behavioral, health services, or clinical research.[3] Clinical research data can be classified as clinical trial data or real-world data. Clinical trial data are generated throughout the clinical trial, whereas real-world data refer to medical data gathered during actual clinical practice from sources other than randomized controlled trials.

Clinical trials, also known as interventional clinical studies, are a type of prospective medical research conducted when a new medicinal product is tested on human subjects. The generated data could include safety or efficacy information.[3]

The individual participants’ data include raw data and analyzable data sets. The raw data are observations of the individual participants that are used by the researchers, such as weight, heart rate, blood pressure, among others. These raw data must go through a process of cleaning, quality control, and assurance in a database for further processing and analysis. After the trial is completed and no further editing or cleaning is performed, the data will be transferred into the analyzable data file, where no extra changes will be made. It is referred to as analyzable data rather than analyzed data because a large proportion of these data are never used. The final analyzable data set contains many components, including participant characteristics and primary outcomes, prespecified secondary and tertiary outcomes, adverse event data, and exploratory data.

Metadata and additional documentation

Clinical researchers need extra documentation and metadata, other than individual participants' data, so as to use clinical trial data (e.g., to perform primary analysis). This includes the trial registration data set, data-sharing plans and all amendments, study protocols and all amendments, and a full statistical analysis plan.

Summary data

These are the documents generated from the analysis of the clinical trial data. They include publications, results summaries in registries, and clinical study reports.

Medical Data/Real-World Data

Real-world data are defined as the patient’s health status and/or service provision data collected from different sources, including electronic health records, medical claims, health technology applications, disease registries, among others. Medical data could be generated from medical records (electronic and paper), laboratory results, diagnostic imaging, pathology results, radiographs, medical data in claims and prescriptions, or from technologic applications and wearable devices.[4]

Genomics data refer to data holding information on the base sequence of an individual’s genome. Genomic data are nowadays an integral part of biomedical research. The GDPR prohibits the processing of genetic data as one of the personal data categories, except in particular conditions described in Article 9 GDPR.[5]

Biometric data means “personal data resulting from specific technical processing relating to the physical, physiological, or behavioral characteristics of a natural person that allows or confirms the unique identification of that natural person, such as facial images or dactyloscopy data.”[6] Vast platforms are now considered sources of pharmacovigilance data, including electronic health records, clinician-reported outcomes, patient-reported outcomes, wearable devices, and social media.

Claims and Invoice Data

Claims data, also known as administrative data, are derived from billing codes that are delivered by doctors, pharmacists, and healthcare providers to payers for reimbursement.[7] Claims data provide information about financial metrics, such as costs, charges, and reimbursement amounts. Despite being defined as administrative data, claims data also contain patient information.

Claims include data that are directly related to the patient’s health status, such as diagnosis, treatments, and interventions performed on the patient. As previously mentioned, health data are sensitive personal data and should be protected from unauthorized access or disclosure.

Claims include administrative data that may be a predictor of the patient’s health outcomes, such as the refill pattern in chronic diseases. They are generated during every visit/hospital stay, such as patient demographics, outpatient services’ provision, inpatient admission and discharge dates, prescribed medications, prescriber ID, dispensing pharmacy, procedure codes, disease codes, and billing codes. Administrative data could be analyzed to study drug use, such as prescribing and dispensing trends and therapy adherence.[8]

Resource Utilization Data (Drug Supply/Device Data or Any Healthcare Intervention)

The Egyptian Authority for Unified Procurement, Medical Supply, and Management of Medical Technology (UPA) generates an enormous amount of data. These data are described below.

Pre-tender data

Pre-tender data encompass information regarding the market goods, services, and potential tenderers. Collected data include information that is not available to the public, such as references from potential tenderers, preliminary pricings, and technical aspects and limitations that are not disclosed by the tenderers to the public. In addition, public organization may request some personal information about the work history of tenderers’ employees.

Tender data

They include contract notice, invitation to tender, contract draft, and attachments on prices, references, or customer statements. These data are usually available online and accessed freely, and are classified as public data. Similar to pre-tender data, employees’ personal information should not be disclosed to the public. However, justifications for the tender award and related documents should be disclosed to the public in the final phase of the tender process.

Post-tender data

Data in the post-tender phase are focused on the contractual terms and operations. This phase starts with a purchase request from the internal customer, and it usually includes the following information: requester, approval flow, project, supplier, product, product category, quantity, price, delivery address, required delivery time, price list, or quotation. All post-tender data are confidential, as they include sensitive data related to business transactions.

Epidemiologic Data

These data include information within a record that could identify an individual—such as age, sex, residence area, method of admission and discharge, and codes of procedure and investigation—and are pseudonymized or replaced by artificial identifiers. Information used in epidemiologic analysis should be stored in a secure database that is only available to authorized statisticians.[9] The analysis results and final information about population health patterns and different indicator rates are designated as unclassified data and should be made available to the public via various platforms.

Nongovernmental Organizations’ Health Data

In Egypt, nongovernmental organizations (NGOs) own 242 hospitals and 154 clinics and medical centers in Egypt, constituting 2800 beds, 270 operation theaters, 530 dialysis units, 1500 incubators, and 160 emergency rooms.[10] NGOs’ health data include administrative data on service types, services' evaluation (outcome indicators and coverage rates), as well as the number of drugs dispensed and the length of stay.

The Recommended Data Classification Types

The guiding principles when managing the many types of data mentioned above and generated by any public or private organization should be as follows: sensitive personal data should only be accessed with the subjects’ explicit consent; administrative data should be shared with researchers and other stakeholders, based on their classification level, which is determined by the risks associated with the sharing and disclosure of the involved data. Aligned with the classification levels covered by MCIT, the recommended data classification types considered in this analysis are showed in Table 1. When classifying data, it is crucial to consider that there are subsets of the data for each type. For example, tender data pertaining to contract notices, pricing attachments, contract drafts, and the rationale behind tender awards should not be categorized at the same level as tender documents that contain personal information about employees.

Table 1.

The recommended data classification types

Classification Level	Data Type	Reason Behind Each Classification (Settings)
Public data	Research publications (all types of clinical studies, including epidemiologic studies) Tender data	It is available in public domain
Internal use	Pre-tender data	It is available and used in the Egyptian public and private hospitals, and in UPA
Patient data	Individual participants’ data of clinical trials Clinical study report of clinical trials Medical data in real-world population (medical data in all sources other than clinical trials, including medical records, medical registries, patient-reported outcomes, medical equipment, laboratory results, wearable devices, pharmacovigilance data) Medical data in claims Medical data in NGO registries	It is sensitive information, needs de-identification and anonymization of patient information, available in all Egyptian healthcare settings
Administrative data (depends on the type of data being processed)	Administrative data in claims and invoices Administrative data in NGO registries Pre-tender data	It is critical information, needs access control and de-identification, available in all Egyptian healthcare settings
Research data (risk mitigation measures should be taken)	Metadata and additional documents of clinical trials Summary data	It need access and use control, available in EHA, UHIA, HIO, PTES, UPA, MoH, semigovernmental sector, and university hospitals
Restricted data	Genomics data Biometric data Post-tender data	It is confidential data, needs restricted access and use control, available in MoH and UPA

EHA: Egyptian Healthcare Authority; HIO: health insurance organization; MoH: Ministry of Health; NGO: nongovernmental organization; PTES: Program for Treatment at the Expense of the State; UHIA: Universal Health Insurance Authority; UPA: The Egyptian Authority for Unified Procurement, Medical Supply, and Management of Medical Technology.

CHALLENGES IN EXECUTING DATA CLASSIFICATION METHODOLOGY

Data Availability and Quality

The available data remain in a nascent stage, with certain data being collected but inaccessible by other entities operating in the healthcare ecosystem (and often within the same entity). Moreover, overall quality is inadequate for practical use owing to inadequate data entry personnel and processes lacking the necessary expertise and standards in Egypt.

Data Usability

Data usability remains a significant concern. Even when data are available and accessible, their usability is limited owing to absence of a clear data governance framework, which exacerbates challenges in effective data management and utilization.

Data Integrity

Data integrity is underscored by the absence of a standardized reference guide outlining principles of data integrity, which raises liability issues. There is ambiguity regarding risk levels and awareness among stakeholders, which creates reluctance by key data holders to incorporate a data-based framework. This highlights a practical challenge in defining and addressing risks associated with data management and security.

Data Security

Policies remain unclear, despite the inception of the Data Privacy Law No. 151/2020, which lacks definitive implementation measures to date. There is uncertainty surrounding the classification of sensitive data, owing to fear about handling data with potential legal consequences. Additionally, varying levels of risk necessitate different degrees of data sharing, further complicating the regulatory landscape and accountability measures in data shareability.

KEY CONSIDERATIONS

The following should be considered before implementing data classification:

1. Data security management is a multisegmental function involving many authorities. Cooperation between all key players is important for defining roles and assigning responsibilities for a well-established, implementable classification policy.

2. A proper risk assessment plan is crucial for appropriate mitigation measures. The tendency toward overemphasizing security levels of data negatively affects the security system, increases the costs of data security and protection, and burdens practitioners by blocking their access to necessary data.

3. Data classification is one element in a data management system that incorporates other functions, including data quality management, which is of great significance to the system.

4. Training staff within the policy scope should be informed and trained on their roles and the sensitivity of the data they handle and process.

5. Continuous monitoring of the process is the key pillar of sustainability. Evaluation measures of the short-, medium-, and long-term outcomes should be an integral part of the initially designed plan for data classification.

CONCLUSION

It is crucial to classify healthcare data by focusing on appropriate data classification as a key step toward data governance, aligning with the Egyptian government’s strategic goals. Identifying distinct categories of data will help improve the positioning of each category within the right classification level, enabling the implementation of suitable privacy and protection measures to establish a data-driven healthcare system in Egypt.