Melior Data Overview

This vignette provides a comprehensive overview of the Melior electronic health record data used in the SEM cohort study. The data is organized into three temporal categories:

1. Pre-SEM Data: Data collected before patient inclusion in the SEM cohort
2. SEM Data: Data collected during the SEM cohort contact/hospitalization
3. Post-SEM Data: Data collected after discharge from the SEM cohort contact

Each dataset follows standardized variable naming conventions and processing procedures as outlined in the Variable Naming Standards.

Pre-SEM Data

These datasets capture patient history prior to the index SEM cohort contact, providing context for their clinical presentation.

Dataset	Description	Key Variables	Timeframe
Historical Diagnoses	Diagnoses recorded within 5 years prior to SEM contact	diagnosis_code, diagnosis_type, care_form	5 years before 2017-2018 contacts
Pre-SEM Medications	Medication prescriptions and administrations before SEM contact	medication_atc_code, prescription_start_date, medication_name	1 year before arrival

SEM Cohort Data

These datasets capture clinical information during the index SEM cohort contact/hospitalization period.

Dataset	Description	Key Variables	Timeframe
SEM Diagnoses	All diagnoses recorded during SEM contact	diagnosis_code, diagnosis_type, care_episode_start/end	2017-2018 contacts
ED Diagnoses	Preliminary diagnoses recorded in the emergency department	diagnosis_code, diagnosis_type	2017-2018 ED visits
SEM Interventions	Healthcare interventions performed during SEM contact	intervention_code, intervention_description, care_episode_start/end	2017-2018 contacts
Laboratory Tests (24h)	Lab test results obtained within 24 hours of arrival	lab_test_name, test_result_value, test_result_unit	First 24 hours from arrival
Medications (24h)	Medication prescriptions and administrations within 24 hours of arrival	medication_atc_code, medication_name, administration_method	First 24 hours from arrival

Post-SEM Data

These datasets capture healthcare utilization and outcomes after discharge from the index SEM cohort contact.

Dataset	Description	Key Variables	Timeframe
Post-SEM Care Episodes	Inpatient care episodes after discharge	care_episode_start/end, ward, care_episode_duration_days	After discharge through Dec 31, 2019
Post-SEM Diagnoses (30d)	Diagnoses recorded within 30 days after SEM contact	diagnosis_code, diagnosis_type, care_form	30 days after 2017-2018 contacts
Post-SEM Diagnoses (All)	All diagnoses recorded after discharge	diagnosis_code, diagnosis_type, care_form	After discharge through Dec 31, 2019
Post-SEM Inpatient Hours	Duration of inpatient care in hours	contact_id, patient_id, duration_hours	Within 30 days after discharge
Post-SEM Inpatient Stays	Inpatient hospital stays after discharge	duration_minutes, duration_days	After discharge through Dec 31, 2019
Post-SEM Interventions (30d)	Healthcare interventions performed within 30 days after SEM contact	intervention_code, intervention_description, care_form	30 days after 2017-2018 contacts
Post-SEM Interventions (All)	All healthcare interventions performed after discharge	intervention_code, intervention_description, care_form	After discharge through Dec 31, 2019
Return to ED	First return to the emergency department after discharge	discharged_to, admission_ward, discharge_date	Within 30 days after discharge

Data Relationships

The datasets can be linked through common identifiers:

• patient_id: Links all records for a specific patient across datasets
• contact_id: Links records related to a specific healthcare contact/encounter

Data Characteristics

Temporal Coverage

• Pre-SEM data: Primarily covers 2012-2017
• SEM data: Focuses on 2017-2018
• Post-SEM data: Covers 2017-2019

Data Types

The Melior datasets include several types of clinical information:

1. Diagnoses: Using ICD-10 codes
2. Interventions/Procedures: Using KVÅ codes
3. Laboratory Tests: With values and reference ranges
4. Medications: With ATC codes and administration details
5. Care Episodes: With timestamps and duration information

Data Processing Notes

• All datasets have been standardized to use consistent English variable names
• Original Swedish variable names are preserved in the documentation
• Swedish categorical values (e.g., care forms) have been translated to English in some datasets
• Dates and times are stored in POSIXct format
• Numeric fields have been validated and corrected where needed
• Missing or erroneous data is documented in each dataset’s documentation

Usage Guide

For most analyses, the recommended approach is to:

1. Start with the SEM cohort data that defines your study population
2. Link to pre-SEM data to understand patient history and risk factors
3. Link to post-SEM data to analyze outcomes and healthcare utilization

Each dataset is accompanied by detailed documentation explaining the variables, data quality issues, and examples for proper usage.