Design
Retrospective analysis of routine patient data from a practice based registration network.
Setting
This study was performed in the practice based registration network FaMe-net (Radboud University Medical Center Nijmegen; www.famenet.nl) consisting of 30 general practitioners in the Netherlands, serving a population of approximately 40,000 patients. The population is a bit younger than the Dutch population, but comparable in gender and social class composition.(8) FaMe-net has been recording routine data since 1967. In the Netherlands, patients are listed with a GP, who serves as the first point of access to health care, provides care for the large majority of health problems, and coordinates access to specialized care.(7) GPs record one or more diagnoses and related interventions for every contact (diagnostic testing, referrals, prescribed medication), under regular review of reliability of coding/classification.(9) FaMe-net classifies each condition in the International Classification of Primary Care (ICPC) and the International Classification of Disease (ICD-10). As a unique feature, FaMe-net registers presented symptoms to general practice as the reason for encounter (RFE), which is the literal utterance of the reason why a person enters the consultation room. These include symptoms, diagnoses and cognitions or fears of illness.(10) Since 2016, as well, the patient’s estimation of the duration of symptoms before the first presentation is coded for every new episode of care in hours, days, week or months.
The Radboudumc Technology Center Health Data provides support for extraction and secure storage of routine data from the affiliated practices. It adheres to the regulations of Dutch and European laws and has gained ethical approval from the Radboudumc Medical Ethics Review Committee for this procedure (CMO number 2020-6871). Under Dutch and European privacy laws, it is not necessary to gain informed consent for retrospective studies with anonymised patient data.
Presented symptoms
We selected data from March 16th (the start of the COVID-19 pandemic in the Netherlands) until December 31th 2020, and, in order to contrast the presented symptoms during the pandemic to the preceding years, we also selected data for the years 2016-2019 (March 16th to December 31th ). We selected the presented symptoms as a reason to contact the GP for all new episodes of care (ICPC chapter codes 1-99. For further analysis, we selected the four groups of health symptoms as described: COVID-19 related symptoms, cancer-related symptoms, mental health symptoms and musculoskeletal symptoms. We also distinguished between contact modes; practice visits, home visits, telephone consultations as well as e-mail and video-consultations.
For COVID-19 related symptoms, we selected the most commonly presented: cough (R05), dyspnea (R02), thoracic pain (L04, K01, K02, A11), sore throat (R21), fever (A03), common cold (R74), loss of smell/taste (N16), and contacts starting with the question “do I have COVID-19” (R83).
Furthermore, we selected the symptoms that proved to have predictive value for the diagnosis of cancer in general practice:(8) lump in breast (X19), rectal bleeding (D16), postmenopausal bleeding (X12), hematuria (U06), weight loss (T08), hemoptysis (R24), swallowing problems (D21) and change of feces/ bowel habits (D18). For mental health symptoms we distinguished between anxiety/ stress (P01,P02), feeling depressed (P03), sleeping problems (P06), childhood/ adolescent problems (P22, P23), and other mental health symptoms (P04, P05, P07-21). For musculoskeletal symptoms we categorized upper limb symptoms (L08-12), lower limb symptoms (L13-17), back symptoms (L01-L03).
For diagnostic procedures initiated by the general practitioner we used all ICPC codes 33-43, including laboratory testing, pathology, microbiology, endoscopy, radiology, EKG, and the category “other”. Referrals included all referrals to secondary care. We excluded referrals within primary care.
Patient and doctor delay
For each new episode of care, starting with an RFE from the four groups (i.e. COVID-19 related symptoms, alarm symptoms for cancer, mental health symptoms and musculoskeletal symptoms), we extracted the registered duration of symptoms at the first presentation to the practice. This comparison may be considered a proxy for patient delay. Also, we calculated the time from the first presentation to the practice until GPs decided to refer for diagnostic testing or to refer to a medical specialist, also both for 2020 and 2016-2019. This time may be seen as a proxy for potential doctor’s delay.
Analysis
The presented health problems were expressed as incidence rates per 1000 patient years. The difference between 2020 and the period 2016-2019 was tested by the incidence rate ratio and 95 % confidence interval. For referrals and diagnostic testing we compared data from 2020 with data from the period 2016-2019. The numbers of diagnostic testing and referrals were related to the number of presented health symptoms to the practice and will be presented as percentages of presented symptoms. The patient’s estimation of duration of symptoms at the first visit, was categorized in < 48 hour; 3-7 days; 8-28 days; 29-90 days; and > 90 days. Also the time from the first presentation to diagnostic testing or referral was categorized: same day; 2-7 days; 8-28 days; 29-90 days; > 90 days. A Chi-square test was used to test the difference in distribution over time categories between 2020 and the period 2016-2019. In order to analyze differences in delay during the first lock-down from March 16th until May 15th, compared to the following period until December 31th 2020, we used Chi-square tests for observed differences. A p-value of < 0.05 was considered to be statistically significant, based on two sided tests
Analysis were performed with SPSS version 25 and the fmsb package from R version 3.6.2.(11)