About the General Practicioner data
Data from general affiliated practices are available for research on the I-Drive and within certain projects in the CBS-RA environment. General practitioners (GPs) maintain electronic patient records (EPR) of all their patients through software systems called a Huisartsen Informatie Systeem (HIS). In the ELAN region, the HIS system Medicom from Pharmapartners is predominantly used. Each HIS enforces the use of structure and coding, facilitating data accessibility and uniformity. The episode of care forms the foundation of the EPR structure, linking patient encounters to specific health problems. Interventions and diagnostics can also be associate with particular episodes of care, a process referred to as episode-oriented registration (episode gericht registreren). The GP code of conduct for adequate use of the EPR is available here.
Data Sources and Availability
These data are available in both the internal environent and, for certain projects, the external environment. Snapshots of the datasets listed below are uploaded periodically: annually for the external environment and quarterly for the internal environment.
| Property | Abbreviation | Description |
|---|---|---|
| Actions | ACT | Actions (operations) declared by the GP, cross-checked with NHG-table 15. |
| Contraindications | CIA | Contraindications, logged at multiple levels. |
| Correspondence | COR | Incoming and outgoing correspondence, including referrals. Note: not available for HIS Promedico. |
| Episodes | EPS | Episodes of care per patient, registered symptoms and conditions, coded with the ICPC. Includes start and possible end dates. |
| Journalen | JRN | Records every consultation between a patient and a practitioner. |
| Lab measurements | LAB | Measured values/readings in accordance with NHG table 45, including labt tests, BMI, and blood pressure. |
| Medication | MED | Prescriptions of medication, coded using ATC. |
| Patients | PAT | Characteristics of all registered patients, including those who have moved or passed away. |
| Indicators | RUI | Indicators for selecting groups of patients, categorized based on NHG table 16. An example of an indicator is the code ‘GV,’ which indicates whether a patient is eligible for an annual flu vaccination. |
The NHG tables can be accessed here.
Data Structure and Coding Practices
The records of a consultation in the journaal dataset are structured by GPs using the SOEP system, which stands for:
- Subjective: Patient’s concerns and reasons for visiting the GP.
- Objective: Observations or symptoms noted by the GP.
- Evaluation: Possible diagnoses.
- Plan: Proposed actions, such as further tests (e.g. blood analysis, ECG, or X-ray), referrals to specialist, or medication.
Below is an example. Each record is referenced by its dSOEPCode; for instance, the first record is an S-record, and the last is a P-record.
| Extractiedatum | Systeem | StartDate | EndDate | PATNR | EpisodeID | dDatum | dSOEPCode | Omschrijving |
|---|---|---|---|---|---|---|---|---|
| 2020-01-01 | 1 | 2020-01-01 | NA | 1 | 1 | 2019-12-12 | S | cold |
| 2020-01-01 | 1 | 2020-01-01 | NA | 1 | 1 | 2019-12-12 | O | |
| 2020-01-01 | 1 | 2020-01-01 | NA | 1 | 1 | 2019-12-12 | E | |
| 2020-01-01 | 1 | 2020-01-01 | NA | 1 | 1 | 2019-12-12 | P |
Typically, a first consultation will result in S, O, E and P records, whereas a follow-up consultation may only include an S or P record. Using episode-oriented registration, the related health condition can be found in the episodes dataset via the episode ID. The episodes dataset contains the most up-to-date ICPC-coded health condition, whereas the E-record in the journal dataset may reflect earlier symptoms or diagnoses (working hypotheses). For example, a journal with ICPC R03 (wheezing) might be linked to an episode with ICPC R96 (asthma).
Variability per Practitioners
There is significant variability among GPs in how patient records are maintained, particularly in using episode-oriented registration. For instance, one GP might diagnose a cold when a patient reports sore throat, while another might code it as a symptom. When data appears incomplete, consider alternative coding methods.
Here’s another example: a patient with chronic headaches presents with morning vomiting and dipplopia. Suspecting a brain tumour, the GP records these symptoms in the consultation and refers the patient to a neurologist. If the tumour is not confirmed, the previous e-record may be retroactively updated, but this is not guaranteed. In some cases, the initial record might still list migraines rather than the symptoms.
The StartDate column is not always reliable. For example, a practitioner might use the date of the first migraine, the consultation date, or even a fictional date. Additionally, a new GP will only record conditions reported by the patient after the transfer of care.
Data Snapshots
As noted earlier, data is collected quarterly. Patient data evolves over time, so StartDate and EndDate fields are included in the internal environment.
- For the initial data retrieval, StartDate is left empty, and EndDate is set to the firt retrieval date.
- For subsequent retrievals:
- If no changes are detected, the EndDate is updated.
- If new records are added, the new record’s StartDate is set to the day after the previous record’s StartDate.
- If records are modified, the original remains unchanged, and a new record is created with the day after the retrieval date.
Background on Data Standardisation
GP data is collected by STIZON, a trusted third party, which standardises it before sharing it. Standardised attributes start with a ’d’; researchers are encouraged to use these fields, though non-standard attributes may also be useful.
We strongly recommend researchers review the codebook for guidance. Additional information on challenges and potential solutions can be found here.