# paper.pdf ## Metadata - PDFFormatVersion=1.4 - Language=en-US - IsLinearized=false - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - Creator=Adobe InDesign 21.3 (Macintosh) - Title=Association of General Practice Continuity With Hospital Admissions and Costs: A Retrospective Study - Author=Annals of Family Medicine Inc - Producer=Adobe PDF Library 18.0 - Trapped=False - ModDate=D:20260622145516-05'00' - CreationDate=D:20260622145516-05'00' - xmp:createdate=2026-06-22T14:55:16-05:00 - xmp:modifydate=2026-06-22T14:55:16-05:00 - xmp:metadatadate=2026-06-22T14:55:16-05:00 - xmp:creatortool=Adobe InDesign 21.3 (Macintosh) - xmpmm:instanceid=uuid:1ff2b7cd-0498-264f-a978-12cdd20d1d8d - xmpmm:originaldocumentid=adobe:docid:indd:7de92d64-364b-11de-93a4-81f3cec3252f - xmpmm:documentid=xmp.id:b67ef9c4-22a7-4adf-8d98-c660d1e4ba90 - xmpmm:renditionclass=proof:pdf - xmpmm:derivedfrom=xmp.iid:47bd9b2c-810e-45aa-9713-dc69bd5c3eb5xmp.did:d398b682-0d79-4c45-a959-6ae39218f4b3adobe:docid:indd:7de92d64-364b-11de-93a4-81f3cec3252fdefault - xmpmm:history=convertedfrom application/x-indesign to application/pdfAdobe InDesign 21.3 (Macintosh)/2026-06-22T14:55:16-05:00 - dc:format=application/pdf - dc:rights=Downloaded from the Annals of Family Medicine website at www.AnnFamMed.org. Copyright © Annals of Family Medicine, Inc. For the private, noncommercial use of one individual user of the Web site. All other rights reserved. Contact copyrights@aafp.org for copyright questions and/or permission requests. - dc:creator=Annals of Family Medicine Inc - xmprights:marked=True - pdf:producer=Adobe PDF Library 18.0 - pdf:trapped=False ## Contents ### Page 1 Association of General Practice Continuity With Hospital Admissions and Costs: A Retrospective StudyABSTRACT PURPOSE Continuity of care has been associated with lower health care use and costs. When general practitioners (GPs) serve as gatekeepers to hospital care, having an ongoing relation- ship with a GP may promote more efficient care and reduce urgent referrals, thereby lower- ing costs. We examined the association of 2 conceptually distinct measures of continuity in general practice with both urgent hospital admissions and hospital costs.METHODS We conducted a retrospective cohort study using routinely collected health care data from 100,450 adults in 48 Dutch general practices (2013-2018) who were linked to national hospital admission and cost records (2019). Continuity was separately measured by duration of GP-patient relationship (time registered with the practice) and by density of GP contacts (Herfindahl-Hirschman index). We assessed associations with urgent admissions and log-transformed hospital costs using multilevel logistic and linear models, adjusting for sex, age, comorbidity, migration background, and income.RESULTS Compared with individuals registered with their practice for 0 to 5 years, those registered longer had 9% to 21% lower odds of urgent hospital admission, with the low- est odds seen in the group registered 15 to 20 years (adjusted odds ratio = 0.79; 95% CI, 0.70-0.88). No significant association was found between higher density of GP contacts and urgent admissions. All groups registered longer than 5 years had significantly lower hospital costs, with the largest model-based reduction (−39.7%) in the group registered 10 to 15 years. Higher density of GP contacts was also associated with lower costs, with an adjusted relative reduction of 6% to 7% in total hospital costs.CONCLUSIONS Long-standing continuity of care in general practice among adult patients is associated with reduced urgent hospital admissions and health care costs in the Netherlands. These findings highlight the potential value of sustained GP-patient relationships in improv- ing hospital outcomes.Annals Early Access articleAnn Fam Med 2026;24:Online. https://doi.org/10.1370/afm.250537INTRODUCTIONContinuity of care is considered one of the core values of general practice.1-3 It is defined as the ongoing, long-term relationship between patient and health care professional, which can and usually does arise from repeated contacts over time.4,5 Numerous studies on continuity have demonstrated improved health outcomes, including better communication between patient and physician,6,7 better medication prescribing and compliance,8-10 better uptake of preventive care,11,12 fewer emergency department visits and hospital admissions,13-15 and reduced mortality.16-19 Additional important benefits of continuity of care include lower health care use and costs.12,20 A 2023 systematic review confirmed these associations across the majority of outcomes examined20; however, it also revealed substantial variation in how continuity, use, and costs were defined and measured across studies, health care systems, and populations. This conceptual and methodologic heterogeneity highlights the need for further research to confirm these relationships in different settings and to support the development of consistent, evidence-based approaches in policy and practice. In addition, none of the studies examining the association between continuity and health care use and/or costs used more than a single con- ceptually distinct type of continuity measure within the same analysis. Similar to health care systems internationally, the Dutch health care system is increasingly challenged by population aging and workforce shortages.21 To ensureMarije T. te Winkel, MD1,2 Annemarie A. Uijen, MD, PhD3 Birgit I. Lissenberg-Witte, PhD4 Rob J. van Marum, MD, PhD2,5 Henk J. Schers, MD, PhD3 Pauline Slottje, PhD1,6 Jettie Bont, MD, PhD6,7 Otto R. Maarsingh, MD, PhD1,21Department of General Practice, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands 2Aging and Later Life, Amsterdam Public Health, Amsterdam, The Netherlands 3Department of Primary and Community Care, Radboud University Medical Center, Nijmegen, The Netherlands 4Department of Epidemiology and Data Science, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands 5Department of Medicine for Older People, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands 6Quality of Care, Amsterdam Public Health, Amsterdam, The Netherlands 7Department of General Practice, Amsterdam UMC, University of Amsterdam, Amsterdam, The NetherlandsConflicts of interest: authors report none.CORRESPONDING AUTHOROtto R. Maarsingh Department of General Prac- tice, Amsterdam UMC Van der Boechorststraat 7, 1081 BT Amsterdam, the Netherlands o.maarsingh@amsterdamumc.nlANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20261ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORGDownloaded from the Annals of Family Medicine website at www.AnnFamMed.org. Copyright © Annals of Family Medicine, Inc. For the private, noncommercial use of one individual user of the Web site. All other rights reserved. Contact copyrights@aafp.org for copyright questions and/or permission requests. ### Page 2 GENERAL PRACTICE CONTINUITY AND HOSPITAL OUTCOMESthe sustainability of health care, it is essential to focus on eliminating interventions that do not contribute to appro- priate care and support. General practitioners (GPs) act as gatekeepers for hospital care, generally allocating health care resources efficiently.22 In times of decreasing continu- ity,23,24 however, it remains unknown how the duration and density of the GP-patient relationship influence hospital care use and costs. The aim of this study was to examine the relationship between continuity in general practice and urgent hospital admissions and hospital costs. We hypothesized that care provided by a known GP and team over years or even decades is more appropriate—avoid- ing unnecessary hospital admissions, referrals, and follow-up, and ultimately reducing costs.25 Increasing continuity of care may therefore play a crucial role in addressing these sys- temic challenges.METHODS Study DesignWe conducted a retrospective observational cohort study in which patients could enter and leave the cohort over time, using ano- nymized routine care data extracted from the database of the Academic Network of General Practice Amsterdam, North-Holland, and Almere (AWH-ANHA). These 48 gen- eral practices were located in the urban area in and around Amsterdam and Haarlem, the Netherlands. Patients in the ANHA database were individually linked to national admin- istrative data from Statistics Netherlands (StatLine) to obtain information on hospital admissions and costs, country of birth, and household income. To study the GP-patient relationship, patients were included if they were registered at least 1 year with the same practice; had at least 5 contacts (ie, telephone calls, home visits, e-mails, and/or face-to-face consulta- tions) with their practice, of which at least 2 were with a GP; and were aged 18 years or older in 2013 (the starting point for observa- tion of contacts) and registered on January 1, 2019 (Figure 1). These GP contacts could occur with different GPs within the same practice, reflecting both individual and team- based care at the practice level rather than care delivered by a formally defined or fixed team. In the Dutch health care system, almost all individuals are registered with a general practice and they may change practices, but registration typically lasts for many years, often until death. Although Dutch GPs provide care for patients of all ages, including children, the present study was restricted to adults.Figure 2 is a schematic of our study timeline for collec- tion of data on continuity measures and collection of data on study outcomes, described below.Main OutcomesUrgent Hospital AdmissionsWe defined an urgent hospital admission as at least 1 urgent, unplanned, clinical hospital admission in 2019 (recorded asFigure 1. Flowchart of Patient Inclusion in the StudyGP = general practitioner.48 General practices269,478 Patients with ≥1 contact with practice during contacts observation period (Jan 1, 2013 to Dec 31, 2018 ) 190,886 Patients registered ≥1 year with ≥2 GP contacts during 2013-2018 Excluded: 152 Patients not identified by trusted third party 27,156 Patients with failed linkage to Statistics Netherlands 34,456 Patients aged <18 years on Jan 1, 2013 129,122 Adult registered patients with ≥2 GP contacts during 2013-2018 who were linked to Statistics Netherlands data Excluded: 2,903 Patients registered at multiple practices (only practice with highest number of contacts was retained) 531 Patients with unavailable information on household income 25,238 Patients deregistered before Jan 1, 2019 Excluded: 9,521 Patients registered <1 year in total 66,636 Patients with <5 contacts with practice 2,435 Patients with <2 GP contactsFinal study population100,450 Adult registered patients with ≥2 GP contacts during 2013-2018 who were linked to Statistics Netherlands data and registered on Jan 1, 2019ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20262ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ### Page 3 GENERAL PRACTICE CONTINUITY AND HOSPITAL OUTCOMESyes or no). Urgency was determined by the admitting physi- cian or nurse as requiring clinical admission within 24 hours after initial presentation, as opposed to a planned or elective admission.26Hospital CostsHospital costs were defined as the total hospital costs per patient in 2019. This included all health care costs submitted to standard health insurance, which residents of the Neth- erlands are legally required to have under the Dutch Health Insurance Act.27 Both GP and hospital care costs are expenses reimbursed by the health insurers.DeterminantsContinuity MeasuresNo consensus exists on the preferred way to measure conti- nuity of care within primary care. To optimize the validity and rigor of our results and allow for comparison to other studies, we decided to use 2 established continuity measures. First, we used duration of the GP-patient relationship, defined as the time between date of registration and Janu- ary 1, 2019 measured in years. We created 6 similarly sized groups based on duration of registration. To describe cohort composition, we examined the distribution of year of reg- istration within 5-year registration-duration groups in the pre-linkage data set (Supplemental Figure 1). Second, we used density of GP contacts, defined by the Herfindahl-Hirschman index, which captures the density and dispersion of contacts with a particular GP (definition and example in Supplemental Appendix 1). For each patient, we used all contacts that were registered in the electronic health record by any GP up to 6 years before January 1, 2019. Val- ues range from 0 to 1, with higher values denoting greater continuity. We categorized Herfindahl-Hirschman index val- ues into tertiles (low, intermediate, and high).Number of Chronic ConditionsWe defined chronic conditions as irreversible conditions without complete recovery or relatively long-lasting condi- tions as documented on January 1, 2019, based on International Classification of Primary Care (ICPC) codes in the electronic health record.28 Conditions were selected based on high prev- alence and a chronic and severe character from a standard list of chronic conditions for primary care. This resulted in 29 chronic conditions (definitions and prevalences are given inSupplemental Table 1). This ICPC-based chronic condition list captured the presence and number of chronic conditions but not their severity.Migration Background and Household IncomeOn the basis of information from Statistics Netherlands, we classified patients as having a “migration background” if they (or at least 1 parent) were born in a country other than the Netherlands (classified as either a European Union Plus country or some other non–European Union country).16,19,29,30 We used household income as a proxy for socioeconomic sta- tus,31 calculating the mean of the yearly disposable income of the household between 2013 and 2018.Statistical AnalysisWe used SPSS versions 25 and 28 (IBM Corp) for data cleaning, data transfor- mation, and creation of descriptive plots. Subsequent analyses were conducted in R version 4.3.2 (R Foundation) and used various R packages (haven, nlme, lme4, xlsx, dplyr, emmeans, and parameters). We performed natural log–transforma- tion of hospital costs because of their non-normal distribution. For other mea- sures, because the linearity assumptions were not met, we created categories, pre- sented as percentages and ranges. We estimated multilevel logistic and linear regression models for the outcomes of urgent hospital admis- sions and costs, respectively. Sex, age, comorbidity, migration background, and household income on the patient level were included as fixed effects. We suspected unobserved heterogeneity because of correlation among patients ofFigure 2. Timeline-Based Schematic for Collection of Data on Continuity Measures and OutcomesGP = general practitioner. a Time registered with the practice on January 1, 2019. b Tertile of Herfindahl-Hirschman index for contacts during 2013-2018.Duration of GP-patient relationshipa Density of GP contactsb January 1, 2019 Low, intermediate, or high in 6 preceding years ≤5, 5-10,10-15, 15-20, 20-25, or ≥25 years Urgent hospital admissions in 2019 Hospital costs in 2019ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20263ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ### Page 4 GENERAL PRACTICE CONTINUITY AND HOSPITAL OUTCOMESTable 1. Baseline Characteristics of the Study Population (N = 100,450) Characteristic Duration of GP-patient relationshipa Density of GP contactsb Total< 5 y 6-10 y 10-15 y 15-20 y 20-25 y ≥ 25 y Low Intermediate HighSex, No. (%) Male 9,127 (38.6) 7,631 (41.5) 7,323 (43.1) 5,857 (45.0) 6,014 (44.2) 6,451 (43.5) 13,307 (39.7) 14,079 (42.0) 15,017 (44.9) 42,403 (42.2) Female 14,498 (61.4) 10,766 (58.5) 9,651 (56.9) 7,150 (55.0) 7,591 (55.8) 8,391 (56.5) 20,177 (60.3) 19,478 (58.0) 18,392 (55.1) 58,047 (57.8) Age group, No. (%)≤35 years 10,321 (43.7) 4,175 (22.7) 1,511 (8.9) 1,081 (8.3) 1,594 (11.7) 1,364 (9.2) 7,764 (23.2) 6,812 (20.3) 5,470 (16.4) 20,046 (20.0) 36-50 years 6,635 (28.1) 7,687 (41.8) 7,201 (42.4) 3,537 (27.2) 1,984 (14.6) 1,685 (11.4) 10,238 (30.6) 9,655 (28.8) 8,836 (26.4) 28,729 (28.6) 51-65 years 3,465 (14.7) 3,723 (20.2) 5,090 (30.0) 4,961 (38.1) 5,309 (39.0) 5,170 (34.8) 8,860 (26.5) 9,163 (27.3) 9,695 (29.0) 27,718 (27.6)≥66 years 3,204 (13.6) 2,812 (15.3) 3,172 (18.7) 3,428 (26.4) 4,718 (34.7) 6,623 (44.6) 6,622 (19.8) 7,927 (23.6) 9,408 (28.2) 23,957 (23.8) Number of chronic conditions, No. (%) 0 13,603 (57.6) 9,992 (54.3) 8,281 (48.8) 5,506 (42.3) 5,079 (37.3) 4,909 (33.1) 16,534 (49.4) 15,907 (47.4) 14,929 (44.7) 47,370 (47.2) 1 5,914 (25.0) 4,903 (26.7) 4,756 (28.0) 3,867 (29.7) 4,065 (29.9) 4,219 (28.4) 9,109 (27.2) 9,285 (27.7) 9,330 (27.9) 27,724 (27.6)≥2 4,108 (17.4) 3,502 (19.0) 3,937 (23.2) 3,634 (27.9) 4,461 (32.8) 5,714 (38.5) 7,841 (23.4) 8,365 (24.9) 9,150 (27.4) 25,356 (25.2) Migration background, No. (%) Native Dutch 12,910 (54.6) 10,007 (54.4) 9,578 (56.4) 7,329 (56.3) 7,966 (58.6) 10,428 (70.3) 18,331 (54.7) 19,886 (59.3) 20,001 (59.9) 58,218 (58.0) EU Plus 3,710 (15.7) 2,225 (12.1) 1,605 (9.5) 1,180 (9.1) 1,289 (9.5) 1,147 (7.7) 3,740 (11.2) 3,652 (10.9) 3,764 (11.3) 11,156 (11.1) Other non-EU 7,005 (29.7) 6,165 (33.5) 5,791 (34.1) 4,498 (34.6) 4,350 (32.0) 3,267 (22.0) 11,413 (34.1) 10,019 (29.9) 9,644 (28.9) 31,076 (30.9) Income, No. (%) Low 7,413 (31.4) 5,730 (31.1) 5,389 (31.7) 4,282 (32.9) 4,908 (36.1) 4,955 (33.4) 11,213 (33.5) 10,738 (32.0) 10,726 (32.1) 32,677 (32.5) Intermediate 7,716 (32.7) 5,757 (31.3) 5,434 (32.0) 4,541 (34.9) 4,857 (35.7) 5,910 (39.8) 11,268 (33.7) 11,485 (34.2) 11,462 (34.3) 34,215 (34.1) High 8,496 (36.0) 6,910 (37.6) 6,151 (36.2) 4,184 (32.2) 3,840 (28.2) 3,977 (26.8) 11,003 (32.9) 11,334 (33.8) 11,221 (33.6) 33,558 (33.4) Urgent hospital admissions in 2019, No. (%) No 22,608 (95.7) 17,701 (96.2) 16,363 (96.4) 12,485 (96.0) 12,950 (95.2) 13,986 (94.2) 32,090 (95.8) 32,086 (95.6) 31,917 (95.5) 96,093 (95.7) Yes 1,017 (4.3) 696 (3.8) 611 (3.6) 522 (4.0) 655 (4.8) 856 (5.8) 1,394 (4.2) 1,471 (4.4) 1,492 (4.5) 4,357 (4.3) EU = European Union; EU Plus = European UnionPlus (another European country, the United Kingdom, the United States, Canada, New Zealand, or Australia). a Time registered with the practice on January 1, 2019. b Tertile of Herfindahl-Hirschman index for contacts during 2013-2018. Note: Baseline characteristics on January 1, 2019. Additional details on hospital admissions, including nonurgent cases and admission counts, are provided in Supplemental Table 7.the same practice (patients nested within practices) and therefore included practices as random intercepts. We checked the associations between outcomes and conti- nuity for confounding and effect modification by age, sex, and number of chronic conditions. We stratified the population if the associations were modified by a variable (interaction term between variable and conti- nuity determinant P <.001). Results are presented as crude and adjusted odds ratios (ORs), corresponding 95% CIs, and P val- ues. Finally, to increase understanding of the natural log–transformed costs, we used the multilevel linearANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20264 ### Page 5 GENERAL PRACTICE CONTINUITY AND HOSPITAL OUTCOMESregression models to calculate mean estimated costs per group with 95% CIs.Ethical ApprovalThe ANHA database is managed according to Dutch pri- vacy legislation and contains pseudonymized general prac- tice care data from all patients of the participating general practices, except for those who decline participation. The medical ethics committee of the Amsterdam UMC con- firmed that the Medical Research Involving Human Subjects Act does not apply to observational studies that are based on anonymized data from the ANHA database and Statistics Netherlands, and that this study was therefore exempted from the patient informed consent (protocol no. VUmc 2015-260 and Amsterdam UMC 2023.0624). The linkage of the ANHA and Statistics Netherlands data sets was based on unique personal identification numbers and executed by a trusted third party.RESULTS Patient CharacteristicsThe final population used in analyses consisted of 100,450 patients (Figure 1). At baseline (January 1, 2019), one-half of the population were aged younger than 50 years (Table 1). The majority, 57.7%, were female. The largest share of patients, 47.2%, had no chronic conditions at baseline, fol- lowed by 27.6% with 1 chronic condition and 25.2% with 2 or more. The most frequent chronic condition was asthma (9.6%), followed by diabetes (8.7%), anxiety (8.6%), and vision disorder (8.2%) (Supplemental Table 1).Urgent Hospital AdmissionsOverall, 4.3% of our cohort had at least 1 urgent hospital admission in 2019. Compared with patients registered with their GP’s practice for 5 years or less, all groups registered for longer had 9% to 21% lower adjusted odds of urgent hospital admission (Table 2). In contrast, we found no statistically sig- nificant association between a patient’s index-based density of GP contacts and urgent hospital admissions.Hospital CostsCompared with patients registered with their GP’s practice for 5 years or less, all patient groups registered for longer had statistically significant adjusted reductions in total hospital costs in that year, with the reduction ranging from 17% to 28% (Table 3, and stratified in Supplemental Table 2). Com- pared with the patient group having a low density of index- based contact, the intermediate- and high-density groups showed an adjusted relative reduction of 6% and 7% in total hospital costs, respectively. Model-based mean hospital costs showed that all groups with a longer registration duration consistently incurred lower costs compared with the reference group (those reg- istered 0-5 years, €112.39) (Table 4). The largest reduction was observed in the group registered 10 to 15 years, with a 39.7% decrease (€80.44), followed by the group registered 20 to 25 years (−32.0%, €85.15), the group registered 15 to 20 years (−27.5%, €88.12), and the group registered 25 years or longer (−20.8%, €93.08). Higher density of GP contacts was also associated with lower mean hospital costs. Compared with the low-density group (€133.12), the inter- mediate- and high-density groups had costs that were 13.6% lower (€117.17) and 27.9% lower (€104.12), respectively. In addition, patients who were older, had chronic conditions, and were female had higher model-based hospital costs; see details in Supplemental Tables 3 & 4)(duration analyses) andSupplemental Tables 5 & 6)(density analyses).DISCUSSION SummaryWe found that patients with higher continuity based on reg- istration with their GP’s practice longer than 5 years had sig- nificantly lower odds of urgent hospital admissions (9%-21% lower) and lower hospital costs (17%-28% lower) compared with those registered for 0 to 5 years, after adjusting for age, sex, number of chronic conditions, household income, and migration background. Although higher continuity based on density of GP contacts was not associated with urgent admis- sions, it was associated with relative reductions in hospital costs (6%-7%). These findings highlight the potential value of sustained GP-patient relationships in reducing future hospital health care use and costs in the Netherlands.Table 2. Multilevel Logistic Regression Models Showing Associations of Continuity Measures With Urgent Hospital Admissions in 2019 Continuity measure Urgent hospital admission, aOR (95% CI) P valueDuration of GP-patient relationshipa≤5 years (ref) … … 5-10 years 0.91 (0.82-1.00) .055 10-15 years 0.82 (0.74-0.92) <.001 15-20 years 0.79 (0.70-0.88) <.001 20-25 years 0.81 (0.73-0.91) <.001≥25 years 0.85 (0.77-0.94) .002 Density of GP contactsb Low (ref) … … Intermediate 1.00 (0.93-1.08) .99 High 0.96 (0.89-1.04) .32 aOR = adjusted odds ratio; GP = general practitioner; ref = reference. a Time registered with the practice on January 1, 2019. b Tertile of Herfindahl-Hirschman index for contacts during 2013-2018. Note: Duration analysis was adjusted for practice, sex, age, number of chronic conditions, household income, and migration background; for complete model, including crude values, see Supplemental Table 8. Density analysis was adjusted for the same variables; for com- plete model, including crude values, see Supplemental Table 10.ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20265ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ### Page 6 GENERAL PRACTICE CONTINUITY AND HOSPITAL OUTCOMESStrengths and LimitationsTo the best of our knowledge, this is the first study to exam- ine the association between continuity in general practice— assessed with multiple conceptually distinct measures of continuity—and both hospital use and costs within the same analysis. We studied continuity of general practice care from different perspectives: the practice (using duration of regis- tration as a proxy for GP-patient continuity) and the patient (using an established continuity index as proxy for density of GP-patient contacts). Use of these measures allowed us to capture continuity at different levels within contemporary primary care. In previous studies, we used multiple indices to calculate density of GP contacts, which showed associa- tions similar to those seen with the Herfindahl-Hirschman index.9,32 The study reported here was based on longitudinal real-life health care data of 100,450 patients from 48 general practices in and around Amsterdam and had similar conti- nuity levels compared with those in practices in other stud- ies.33,34 We therefore consider our results representative and generalizable, especially for urban Dutch general practices. Some limitations of our study should be acknowledged. First, we applied 2 major inclusion criteria: (1) patients had to have contacted their practice at least 5 times, includ- ing at least 2 contacts with a GP, to enable calculation of GP-contact density, and (2) patients had to remain regis- tered at a specific practice through December 31, 2018, to ensure they were not receiving primary care elsewhere. As a result, patients who avoided, declined, or did not require care—and thus had no GP contacts—were excluded. Consequently, we were unable to assess the associ- ation between zero continuity and hospital admis- sions or costs. Second, these inclusion criteria may have acted as a collider, particularly affecting recently registered patients. Although examination of the pre-linkage data set suggests that the group registered 0 to 5 years was not predominantly composed of recent registrants (Supplemental Figure 1), residual cohort construction effects cannot be fully excluded. Third, although we successfully matched 85.7% of patients to data from Statistics Netherlands using date of birth, sex, and postal code, some patients may have been inadvertently excluded. This is par- ticularly likely for individuals with missing or generic birth dates (eg, January 1 or July 1), which are often assigned to people with a first-generation migration background. As the achieved match rate is generally considered acceptable, however, selection bias is unlikely.35 Fourth, age is inherently correlated with length of continuity, which may introduce collinearity and complicate interpretation of their independent asso- ciations. Although analyses were adjusted for age, residual collinearity cannot be fully excluded. Fur- thermore, even though we adjusted for the number ofTable 4. Multilevel Linear Regression Models Showing Associations of Continuity Measures With Mean Hospital Costs in 2019 Continuity measure Estimated hospital costs,a mean (95% CI), € Difference, %Duration of GP-patient relationshipb≤5 years (ref) 112.39 (105.04-120.25) … 5-10 years 90.62 (84.43-97.28) −24.0 10-15 years 80.44 (74.73-86.58) −39.7 15-20 years 88.12 (81.57-95.20) −27.5 20-25 years 85.15 (78.83-91.98) −32.0≥25 years 93.08 (86.20-100.51) −20.8 Density of GP contactsc Low (ref) 133.12 (124.84-141.95) … Intermediate 117.17 (110.00-124.81) −13.6 High 104.12 (97.68-110.98) −27.9 GP = general practitioner; ref = reference. a Natural log of hospital costs, retransformed. b Time registered with the practice on January 1, 2019. c Tertile of Herfindahl-Hirschman index for contacts during 2013-2018. Note: Duration analysis was adjusted for practice, sex, age, household income, and migration background, without stratification by number of chronic conditions. Density analysis was adjusted for sex, age, number of chronic conditions, household income, and migration background. Further methodologic details are provided in Supplemental Tables 3-6.Table 3. Multilevel Linear Regression Models Showing Associations of Continuity Measures With Hospital Costs in 2019 Continuity measure Hospital costs, coefficienta (95% CI) P valueDuration of GP-patient relationshipb≤5 years (ref) … … 5-10 years 0.81 (0.76-0.86) <.001 10-15 years 0.72 (0.67-0.76) <.001 15-20 years 0.78 (0.73-0.84) <.001 20-25 years 0.76 (0.71-0.81) <.001≥25 years 0.83 (0.77-0.89) <.001 Density of GP contactsc Low (ref) … … Intermediate 0.94 (0.89-0.99) .01 High 0.93 (0.88-0.98) .008 GP = general practitioner; ref = reference. a The exponentiated regression coefficient, representing relative change in cost. Hospital costs were retransformed. b Time registered with the practice on January 1, 2019. c Tertile of Herfindahl-Hirschman index for contacts during 2013-2018. Note: Duration analysis was adjusted for practice, sex, age, household income, and migra- tion background; for complete model, including crude values, see Supplemental Table 9, and for analysis stratified by number of chronic conditions, see Supplemental Table 2.Density analysis was adjusted for practice, sex, age, household income, number of chronic conditions, and migration background; for complete model, including crude values, seeSupplemental Table 11.ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20266ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ### Page 7 GENERAL PRACTICE CONTINUITY AND HOSPITAL OUTCOMESchronic conditions, residual confounding by illness severity may persist. Other unmeasured factors, such as patient resi- dential mobility or physician turnover, may have influenced observed continuity patterns but could not be assessed in this study. Missing data on outcomes hospital costs or admissions are unlikely, because both are routinely reported to Statistics Netherlands by health insurers and hospitals.31 It is important to note that our study focused on hospital costs. Consequently, we are unable to draw conclusions about the relationship between continuity of care and costs that extend beyond hospital care. Regarding hospital admissions, our data included only admissions to hospitals within the Netherlands. We do not have access to data on admissions occurring abroad.Comparison With Existing LiteratureMore than 30 years ago, Hjortdahl36 suggested that it takes at least 5 years to create an “extensive knowledge base” between GP and patient. Although a lot has changed since, our study shows a similar trend, as patients who were reg- istered longer than 5 years had lower hospital costs. This suggests that longer enlistment may contribute to more effi- cient care, potentially through accumulated knowledge and increased trust. Additionally, we found that higher density of contacts was associated with cost reductions, reinforcing the value of seeing the same GP. We acknowledge, however, that these observed associations may not be fully attributed to continuity alone. Patients with a higher risk of hospital admissions and costs may be more likely to change GP or general practice for various reasons, ranging from relo- cating closer to caregivers to being dissatisfied with their previous GP. Although many studies have examined the relationship between continuity of care and health care use or costs, they have varied widely in design, continuity measures, populations, and outcomes. This heterogeneity, as described in a review by Bazemore et al in 2023,20 complicates direct comparisons. To our knowledge, only a single study has been published on the association between continuity and health care use using a Dutch general practice population.37 That study found that patients registered less than 1 year had a higher probability of receiving an antibiotic prescription as a proxy for health care use, although the study did not evaluate broader outcomes including hospital admissions or costs. A study from the United Kingdom using the Usual Provider of Care index found that a 0.2-point increase in continuity could reduce avoidable admissions by 6.2%.15 Continuity was measured over a 2-year period and admissions were limited to a predefined list of 22 conditions, however. In contrast, our study used a broader population and focused on all urgent admissions, which may be less influenced by conti- nuity because of their acute nature. We found no association between contact density and urgent hospital admissions. Per- haps, patients with medical reasons for more frequent visits have a higher contact density, potentially attenuating the protective effect of continuity. In addition, urgent admissions may be related to accidents or acute events, which may be independent of GP continuity.Implications for Practice and PolicyOur study suggests an association between continuity in gen- eral practice and hospital use and costs. Although there is an overwhelming amount of evidence regarding the benefits of continuity of care for both patients and GPs, this study shows that continuity is also associated with fewer urgent admissions and lower hospital costs. Providing continuity requires delib- erate and sustained effort from all stakeholders—patients, health care professionals, insurers, and policy makers. It is our shared responsibility to continue ensuring high continuity of care in general practice.Read or post commentaries in response to this article. Key words: continuity of care; health care costs; health care use; hospital costs; urgent admissions; health service delivery; primary care; practice-based research Submitted August 25, 2025; submitted, revised, February 17, 2026; accepted March 9, 2026.Funding support: Marije T. te Winkel and Otto R. Maarsingh were supported by the Stichting Beroepsopleiding Huisartsen in the Netherlands.Disclaimer: The funding organization was not involved in the design of the study; in the collection, analysis, and interpretation of the data; or in writing the manuscript.Acknowledgments: We are grateful to all participating patients and practices, and the data managers of the ANHA and Centraal Bureau voor de Statistiek (Statistics Netherlands) for their time and effort in contributing to routine care data.Supplemental Materials References1. Guthrie B, Saultz JW, Freeman GK, Haggerty JL. Continuity of care matters.BMJ. 2008; 337: a867. doi: 10.1136/bmj.a8672. Stokes T, Tarrant C, Mainous AG III, Schers H, Freeman G, Baker R. Continuity of care: is the personal doctor still important? A survey of general practi- tioners and family physicians in England and Wales, the United States, and The Netherlands. Ann Fam Med. 2005; 3(4): 353-359. doi: 10.1370/afm.3513. Haggerty JL, Reid RJ, Freeman GK, Starfield BH, Adair CE, McKendry R. Con- tinuity of care: a multidisciplinary review. BMJ. 2003; 327(7425): 1219-1221. doi: 10.1136/bmj.327.7425.12194. Saultz JW. Defining and measuring interpersonal continuity of care. Ann Fam Med. 2003; 1(3): 134-143. doi: 10.1370/afm.235. Starfield B, Horder J. Interpersonal continuity: old and new perspectives. Br J Gen Pract. 2007; 57(540): 527-529.6. Saultz JW, Albedaiwi W. Interpersonal continuity of care and patient satisfac- tion: a critical review. Ann Fam Med. 2004; 2(5): 445-451. doi: 10.1370/afm.917. Ridd M, Shaw A, Salisbury C. ‘Two sides of the coin’—the value of personal continuity to GPs: a qualitative interview study. Fam Pract. 2006; 23(4): 461- 468. doi: 10.1093/fampra/cml0108. Chen CC, Tseng CH, Cheng SH. Continuity of care, medication adherence, and health care outcomes among patients with newly diagnosed type 2 dia- betes: a longitudinal analysis. Med Care. 2013; 51(3): 231-237. doi: 10.1097/ MLR.0b013e31827da5b99. Te Winkel MT, Damoiseaux-Volman BA, Abu-Hanna A, et al. Personal conti- nuity and appropriate prescribing in primary care. Ann Fam Med. 2023; 21(4):305-312. doi:10.1370/afm.2994ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20267ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ### Page 8 GENERAL PRACTICE CONTINUITY AND HOSPITAL OUTCOMES10. Kim D, Cha J. Association between medical complications according to con- tinuity of care and medication adherence in patients with hypertension in Korea: a national population-based cohort study. BMJ Open. 2023; 13(6):e073404. doi:10.1136/bmjopen-2023-07340411. Bayliss EA, Ellis JL, Shoup JA, Zeng C, McQuillan DB, Steiner JF. Effect of con- tinuity of care on hospital utilization for seniors with multiple medical condi- tions in an integrated health care system. Ann Fam Med. 2015; 13(2): 123-129. doi: 10.1370/afm.173912. Saultz JW, Lochner J. Interpersonal continuity of care and care outcomes:a critical review. Ann Fam Med. 2005; 3(2): 159-166. doi: 10.1370/afm.28513. Kohnke H, Zielinski A. Association between continuity of care in Swedish primary care and emergency services utilisation: a population-based cross-sec- tional study. Scand J Prim Health Care. 2017; 35(2): 113-119. doi:10.1080/0281 3432.2017.133330314. Barker I, Steventon A, Deeny SR. Association between continuity of care in general practice and hospital admissions for ambulatory care sensitive con- ditions: cross sectional study of routinely collected, person level data. BMJ. 2017; 356: j84. doi: 10.1136/bmj.j8415. Sandvik H, Hetlevik Ø, Blinkenberg J, Hunskaar S. Continuity in general prac- tice as predictor of mortality, acute hospitalisation, and use of out-of-hours care: a registry-based observational study in Norway. Br J Gen Pract. 2022;72(715): e84-e90. doi: 10.3399/BJGP.2021.034016. Leleu H, Minvielle E. Relationship between longitudinal continuity of primary care and likelihood of death: analysis of national insurance data. PLoS One. 2013; 8(8): e71669. doi: 10.1371/journal.pone.007166917. Maarsingh OR, Henry Y, van de Ven PM, Deeg DJ. Continuity of care in primary care and association with survival in older people: a 17-year pro- spective cohort study. Br J Gen Pract. 2016; 66(649): e531-e539. doi: 10.3399/ bjgp16X68610118. Pereira Gray DJ, Sidaway-Lee K, White E, Thorne A, Evans PH. Continuity of care with doctors-a matter of life and death? A systematic review of conti- nuity of care and mortality. BMJ Open. 2018; 8(6): e021161. doi: 10.1136/ bmjopen-2017-02116119. Baker R, Freeman GK, Haggerty JL, Bankart MJ, Nockels KH. Primary medi- cal care continuity and patient mortality: a systematic review. Br J Gen Pract. 2020; 70(698): e600-e611. doi: 10.3399/bjgp20X71228920. Bazemore A, Merenstein Z, Handler L, Saultz JW. The impact of interpersonal continuity of primary care on health care costs and use: a critical review. Ann Fam Med. 2023; 21(3): 274-279. doi: 10.1370/afm.296121. Dutch Healthcare Authority. Stand van de zorg 2024. Published 2024. Accessed Aug 25, 2025. https:// puc.overheid.nl/doc/PUC_775407_2222. van der Linden MC, Lindeboom R, van der Linden N, et al. Self-referring patients at the emergency department: appropriateness of ED use and motives for self-referral. Int J Emerg Med. 2014; 7: 28. doi: 10.1186/s12245-014-0028-123. Tammes P, Morris RW, Murphy M, Salisbury C. Is continuity of primary care declining in England? Practice-level longitudinal study from 2012 to 2017. Br J Gen Pract. 2021; 71(707): e432-e440. doi: 10.3399/BJGP.2020.093524. Levene LS, Baker RH, Newby C, Couchman EM, Freeman GK. Ongoing decline in continuity with GPs in English general practices: a longitudinal study across the COVID-19 pandemic. Ann Fam Med. 2024; 22(4): 301-308. doi: 10.1370/ afm.312825. Sripa P, Hayhoe B, Garg P, Majeed A, Greenfield G. Impact of GP gate- keeping on quality of care, and health outcomes, use, and expenditure:a systematic review. Br J Gen Pract. 2019; 69(682): e294-e303. doi: 10.3399/ bjgp19X70220926. Statistics Netherlands. LBZbasistab. Published 2025. Accessed Aug 25, 2025. https://www.cbs.nl/nl-nl/onze-diensten/maatwerk-en-microdata/microdata-zelf- onderzoek-doen/microdatabestanden/lbzbasistab-ziekenhuisopnamen-lbz27. Statistics Netherlands. ZVWzorgkostentab. Published 2025. Accessed Aug 25, 2025. https://www.cbs.nl/nl-nl/onze-diensten/maatwerk-en-microdata/ microdata-zelf-onderzoek-doen/microdatabestanden/zvwzorgkostentab- zorgkosten-personen-basisverzekering28. van Oostrom SH, Picavet HS, van Gelder BM, et al. Multimorbidity and comorbidity in the Dutch population - data from general practices. BMC Public Health. 2012; 12: 715. doi: 10.1186/1471-2458-12-71529. Bos V, Kunst AE, Garssen J, Mackenbach JP. Socioeconomic inequalities in mortality within ethnic groups in the Netherlands, 1995-2000. J Epidemiol Community Health. 2005; 59(4): 329-335. doi: 10.1136/jech.2004.01979430. Statistics Netherlands. Persoon met een niet-westerse migratieachtergrond.Accessed Aug 25, 2025. https:// www.cbs.nl/nl-nl/onze-diensten/methoden/ begrippen/persoon-met-een-niet-westerse-migratieachtergrond31. Mackenbach JP, Kulhánová I, Artnik B, et al. Changes in mortality inequalities over two decades: register based study of European countries. BMJ. 2016; 353:i1732. doi:10.1136/bmj.i173232. Winkel MTT, Slottje P, de Kruif AJ, et al. General practice and patient charac- teristics associated with personal continuity: a mixed-methods study. Br J Gen Pract. 2022; 72(724): e780-e789. doi: 10.3399/BJGP.2022.003833. Bazemore A, Petterson S, Peterson LE, Bruno R, Chung Y, Phillips RL Jr. Higher primary care physician continuity is associated with lower costs and hospital- izations. Ann Fam Med. 2018; 16(6): 492-497. doi: 10.1370/afm.230834. Coma E, Mora N, Peremiquel-Trillas P, et al. Influence of organization and demographic characteristics of primary care practices on continuity of care:analysis of a retrospective cohort from 287 primary care practices cover- ing about 6 million people in Catalonia. BMC Fam Pract. 2021; 22(1): 56. doi:10.1186/s12875-021-01414-y35. Willenborg L, Heerschap H. Method series; Theme: matching. Statistics Neth- erlands. Published 2012. Accessed Aug 25, 2025. https:// www.cbs.nl/en-gb/ our-services/methods/statistical-methods/throughput/throughput/matching36. Hjortdahl P. Continuity of care: general practitioners’ knowledge about, and sense of responsibility toward their patients. Fam Pract. 1992; 9(1): 3-8. doi:10.1093/fampra/9.1.337. Jabaaij L, de Bakker DH, Schers HJ, Bindels PJ, Dekker JH, Schellevis FG. Recently enlisted patients in general practice use more health care resources.BMC Fam Pract. 2007; 8: 64. doi: 10.1186/1471-2296-8-64ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG ✦ VOL. 24, NO. 4 ✦ JULY/AUGUST 20268ANNALS OF FAMILY MEDICINE ✦ WWW.ANNFAMMED.ORG