The aim of this study was to analyse whether the travel distance from the patient’s residential address to the medical facility performing the first cancer diagnostic investigation is associated with the GP’s chosen diagnostic strategy and the GP’s satisfaction with the access to and waiting time for diagnostic investigations.
Study design and setting
This nationwide observational study based on combined questionnaire- and registry data was conducted in Denmark with a population of 5.8 million in 2018. The healthcare system is tax-funded and offers free access to healthcare for all citizens. The GP acts as a gatekeeper serving as the first point of contact to the healthcare system, and 99% of the Danish population is registered with a general practice, which must be consulted for medical advice [19]. GPs can refer patients to investigations and treatment at hospitals and private practicing specialists. CPPs have been implemented for more than 30 types of cancer to ensure standardised national guidelines on the diagnostics and treatment of cancer [20].
Participants
The study population was defined as patients aged 30–99 years recorded with an incident cancer diagnosis (excluding ICD-10: C44) between 1 July and 20 December 2016 in the Danish National Patient Register (NPR) [21].
Patients were eligible if a first registered cancer related investigation could be assessed in Danish nationwide registers and two data sources were used. First, patients undergoing a first diagnostic investigation at a specialised private practice (gynaecology, ear-nose-throat specialist, eye specialist or dermatologist) were identified in the National Health Insurance Service Register (NHISR) (specialty codes: 04, 07, 15, 16, 19, 21, 39, 41). These specialties serve as filter functions in a number of CPPs often referred by a GP. Contacts up to three months prior to diagnosis were assessed in the NHISR, and the first registered contact in this period was selected. Second, patients for whom no diagnostic investigations had been recorded in the NHISR were identified through hospital contacts recorded in the NPR [21]. We identified contacts to the hospital for up to three months prior to the date of diagnosis and selected the first relevant contact (an ICD-10 DC or DZ code, excluding Z08-Z09, Z20-Z29, Z30-Z39, Z55-Z65, Z70-Z76).
The GPs of the included patients received a questionnaire between 28 April 2017 and 10 January 2018 for each patient who had given consent to contact their GP. If the patient had deceased shortly after the diagnosis, permission to contact was granted by the Danish Patient Safety Authority [22]. Each GP was asked to fill in the questionnaire based on the medical records. The questionnaire focussed on the following themes: milestone dates in the cancer trajectory, diagnostic strategy, routes to diagnosis and satisfaction with diagnostic procedures [23]. A GP could fill in questionnaires for more than one patient.
The inclusion criteria were: 1) the GP had completed the questionnaire, 2) the GP was involved in the diagnostic process. Thus, GP responses of 3,455 patients were eligible for inclusion in the study.
Data sources
All data sources were linked through the patient’s unique civil registration number (CRN), which is allocated to all Danish residents and used at every contact with the healthcare system [24]. Information on the GP’s diagnostic strategy and level of satisfaction with diagnostic investigations was obtained from the GP questionnaire.
Variables
Diagnostic strategy was assessed from the question, “Which actions did you/your practice take in the time from when the patient first contacted your practice until you/your practice referred the patient for further investigations for the first time?” The following actions were studied in the analyses: 1) wait-and-see approach (yes/no), 2) medical treatment (yes/no) and 3) referral to further diagnostic investigation on the same day (yes/no). Referral can either be to a CPP, diagnostic unit, specialised private practice, diagnostic imaging or other laboratory test at the hospital. For this study, referral was investigated in general terms and not distinguished between different referral modalities as this depend on e.g. the patients symptoms and the GPs options for referral.
Satisfaction with waiting time and available diagnostic investigations was assessed from the two following questions regarding the diagnostic workup of the patient in general practice: ”How satisfied were you with the availability of diagnostic investigations?” “How satisfied were you with the waiting time for diagnostic investigations?” The response categories “very satisfied” and “satisfied” were combined into “satisfied”, and “dissatisfied” and “very dissatisfied” were combined into “dissatisfied”. The response category “do not know/not relevant” was omitted. The availability of diagnostic investigations vary across regions in Denmark and the GP answered these questions based on his or her availability.
Travel distance (shortest road distance) between patient’s residential address on the date of diagnosis and the first diagnostic investigation in the cancer trajectory was calculated by ArcGIS Network Analyst [25]. This information was obtained from the Danish Civil Registration System [24]. Distance to the first diagnostic investigation was calculated for 3,231 patients (first contact to a hospital: 87%; first contact to a specialised private practice: 13%).. To avoid possible outliers or inclusion of erroneous registrations, it was chosen to exclude patients with a distance of more than 100 km to the first diagnostic investigation (n = 76), as it is unlikely in Denmark that the first relevant diagnostic investigation is so far from the patients residence. Thus, 3,155 patients were included in the analysis.
Confounders
The following variables were included as potential confounders based on data from Statistics Denmark: age, sex, patient’s education categorised according to UNESCO’s International Standard Classification of Education [26] (low: ≤10 years, middle: >10 ≤ 15 years and high: >15 years) and patient’s marital status (married/cohabiting or living alone). Information on cancer type was obtained from the NPR and included as a potential confounder categorised into: 1) breast cancer, 2) gynaecological cancer, 3) cancer in male genitals, 4) cancer in the digestive organs, 5) cancer in the respiratory system, 6) malignant melanoma, 7) haematological cancer or lymphomas and 8) other types of cancer.
Statistical analysis
Generalised linear models (GLMs) with prevalence rate ratios (PRRs) and 95% confidence intervals (CI) were applied to study the association between travel distance from patient’s residence to the first specialised diagnostic investigation and the GP’s diagnostic strategy and satisfaction with available for diagnostic investigations. Distance to the medical facility performing the first specialised diagnostic investigation was categorised into the 25%, 50%, 75% and 90% centiles, corresponding to 0–6 km, > 6–18 km, > 18–34 km, > 34–49 km and > 49 km.
The association between the travel distance to the first diagnostic investigation and the GP’s diagnostic strategy and satisfaction with the waiting time was studied using an unadjusted model, which was followed by a model adjusted for patient’s sex, age, education, marital status and cancer type. To assess if the association between travel distance and GP’s diagnostic strategy was modified by the GP’s suspicion of cancer, we further stratified the analyses on “wait-and-see”, “medical treatment” and “same-day referral” according to whether or not the GP suspected cancer or serious illness. Prior to this, it was tested if distance was associated with the GPs suspicion of cancer or serious illness, which was not the case.
Stata statistical software, release 15.0, was used for all analyses.