Aim
The aim of this study was to determine the prevalence of common chronic diseases and comorbidities in a in whole primary care population and to analyse the associations between a) the number of diseases and primary care utilisation, b) adherence to guideline-based pharmacotherapy and continuity of care.
Setting
Primary care in Sweden
Swedish Family Physicians or General Practitioners (GPs) have undergone 5 years of specialisation after 5½ years of university studies and 18–21 months of internship. About 15–20% of all specialists are GPs. Three consultations with a specialist per inhabitant and year is average (26); half of these are with a GP. Consultations with GPs are, on average, 20 minutes.
GPs often work at primary health care centres (PHCC) in close collaboration with practice nurses and other health care personnel; however, there are also a few single doctor practices. There are roughly 1200 PHCCs (or GP practices) in Sweden (population 9.6 million). In Sweden, the specialists in primary care are almost exclusively specialists in Family medicine, i.e. GPs. However, in our calculations we included all doctors at all health centres.
Almost all primary health care is publicly financed through taxes. Around 40% is privately produced and 60% publicly produced. The reimbursement systems differ between the 21 county councils (regions) in Sweden, but in each county, the reimbursement is by law the same for privately and publicly produced primary health care. The most common reimbursement system is capitation, i.e. that PHCCs are financed in proportion to the number of patients registered at the PHCC. A small share can also be related to number of visits and to achieved quality goals. The capitation is often related to age, Adjusted Clinical Groups (ACG) (27)(28) and Care Need Index (CNI) (29) of the listed patients. The ACG system reflects burden of disease of the registered patients by the combinations of their diagnoses. All inhabitants are free to choose any PHCC, and the patient fees are low and the same for privately and publicly produced primary care. While most PHCCs try to provide a high continuity between patients and specific doctors, not all PHCC are able to achieve this goal.
Design and Material
This is a retrospective registry study based on data from EMRs concerning patients registered with any of the PHCCs in a county in Sweden (Jönköping County) from 2011 to 2015. The entire population in the county was 347 837 persons 2015, of which 345 916 persons (99,4%), were registered with a PHCC, e.g. data on morbidity etc was accessible. All the PHCC used EMRs for registration of patients and patient contacts, morbidity recording, and prescriptions during the study period.
Patients fulfilling the following criteria were selected of further analysis:
a) Diagnosed with one or more of 10 selected chronic diseases 2011: dementia, depression, anxiety, diabetes, atrial fibrillation, heart failure, ischaemic heart disease, COPD, stroke/TIA and/or vascular diseases other than ischaemic heart disease and stroke/TIA (but included in the definition of CHA2DS2VASc). (For diagnosis codes, see Appendix 1).
b) Visited a general practitioner (GP) at least 3 times during the study period.
c) Still registered with a PHCC in the county at the end of May 2015.
The selection of the diagnoses was made to include the most common chronic diseases that are regularly monitored in Swedish primary care. Patients with these diseases normally visit their GP for at least an annual check-up when e.g. health status is evaluated, medication is adjusted and lifestyle interventions are discussed. The selection of diseases and diagnoses was made in consensus by three of the investigators, all GPs.
The patients’ diagnoses were registered in EMRs in connection with each visit to the GPs.
For each patient only one diagnosis from each disease group was counted, e.g. type 2 diabetes mellitus without complications (E119), and type 2 diabetes mellitus with unspecified complications (E118) in the same patient was counted as one disease.
At least three visits in the study period (2011–2015) were considered as an indication of need of continuous care.
For patients with chronic diseases fulfilling the inclusion criteria, the following data were collected from the EMRs from January 2011 up to May 2015: new diagnoses for any of the selected chronic diseases (including the diagnosis anxiety), number of visits to a GP, number of different doctors, which GP that was visited each time, prescriptions of anticoagulants, beta-blockers and statins (for the ATC codes see Appendix 2). In addition, CHA2DS2VASc was calculated for patients with atrial fibrillation (30), (Appendix 3).
Data from 2011-01-01–2015-05-15 was extracted from the EMRs at the end of May 2015 and thus the study included in total material from 52 months (4 years and 4 months).
Patient identities were deleted and replaced by code numbers immediately after data extraction. This was performed by a data manager independent from the research group. Consequently, the research group did not have access to patient identities.
For calculation of continuity of care, two indices that measure the dispersion of continuity were chosen: MMCI (Modified Modified Continuity Index) and COC (Continuity Of Care) (31). These indices reflect a managing perspective and quantify the proportion of visits with distinct GPs in relation to all GPs involved. Therefore, these indices focus particularly on the common request that a multimorbid patient should preferably see the same GP most of the visits, which can be described as interpersonal continuity. MMCI takes into account the number of GPs seen and the number of primary care visits, while COC additionally takes into account the proportion of visits made to each GP. COC is the most used dispersion measure and MMCI is a modification of the first one.
Statistical analysis
Interval and ratio scaled variables were described using means, standard deviations, and various percentiles for different subgroups in the population. Ordinal variables were described using percentiles. Number of patients within different subgroups were described using frequency and percentage. Correlations were calculated between the number of diseases and the degree of health care utilisation and between degree of continuity of care and the adherence to guideline-base pharmacotherapy. Multivariate modelling analysis was calculated for the number of diseases and the degree of health care utilisation.
The analyses did not control for confounding factors as e.g. health literacy, education level, health behaviour or distance to the PHCC as these data were not available on the individual level. The statistical programs used were Stata version 14.3 and Excel version 2011.
Patient and Public Involvement
Patients were not involved in development of the research question, design of the study and outcome measures.