Advertisement
Canadian Journal of Cardiology

Primary Care Physician Visits by Patients With Incident Hypertension

Open AccessPublished:April 02, 2014DOI:https://doi.org/10.1016/j.cjca.2014.03.033

      Abstract

      Background

      Access to a primary care physician (PCP) improves health outcomes among patients with hypertension. The study objective was to compare PCP use among patients with incident hypertension with and without comorbidities.

      Methods

      Hypertensive patients newly diagnosed between April 1, 1998 and March 31, 2009 were identified using Alberta administrative databases. Three comorbidity subgroups were defined: (1) none, (2) vascular risk related, and (3) unrelated. The number of PCP visits was calculated using zero-inflation Poisson regression, with time trends compared using the χ2 test. A Cox model was used to assess the association between PCP use and clinical outcomes.

      Results

      Of 456,263 newly diagnosed hypertensive patients (mean age, 57.6 years; 50.6% men; 62.5% no comorbidity), 88% had seen a PCP in the year before diagnosis, and 94% had seen a PCP in the year after being diagnosed. Compared with before diagnosis, the mean number of PCP visits increased after diagnosis (none, 3.95 vs 6.15; vascular risk related, 6.45 vs 7.99; and unrelated, 6.76 vs 8.24). Over the study period, the frequency of PCP visits before diagnosis was constant, and there was a statistically significant decline in the adjusted mean number of visits after diagnosis. Those with higher PCP use were less likely to die but more likely to be hospitalized regardless of comorbidity.

      Conclusions

      The frequency of PCP visits was high before and after diagnosis. Increased PCP use was associated with a lower risk of death; however, it does increase the costs of caring for patients with hypertension. Therefore, future studies are necessary to determine the optimal level required to achieve cost-effective use of PCP resources.

      Résumé

      Introduction

      L’accès à un médecin de première ligne (MPL) améliore les résultats cliniques des patients souffrant d’hypertension. L’objectif de l’étude était de comparer le recours à un MPL parmi les patients souffrant d’une hypertension associée ou non à des comorbidités.

      Méthodes

      Les patients hypertendus ayant nouvellement été diagnostiqués entre le 1er avril 1998 et le 31 mars 2009 ont été trouvés en utilisant les bases de données administratives de l’Alberta. Trois (3) sous-groupes de comorbidités ont été définis : 1) aucune; 2) risque vasculaire associé; 3) non associée. Le nombre de visites chez le MPL a été calculé en utilisant la régression de Poisson à inflation zéro, avec les tendances temporelles comparées par le test du χ2. Un modèle de Cox a été utilisé pour évaluer le lien entre le recours à un MPL et les résultats cliniques.

      Résultats

      Parmi les 456 263 patients hypertendus nouvellement diagnostiqués (âge moyen, 57,6 ans; 50,6 % d’hommes; 62,5 % sans comorbidités), 88 % avaient rencontré un MPL au cours de l’année qui avait précédé le diagnostic, et 94 % avaient rencontré un MPL au cours de l’année après avoir reçu le diagnostic. Comparativement à ceux qui avaient rencontré le MPL avant le diagnostic, le nombre moyen de visites chez le MPL augmentaient chez ceux qui avaient rencontré le MPL après le diagnostic (aucune, 3,95 vs 6,15; risque vasculaire associé, 6,45 vs 7,99 et non associée, 6,76 vs 8,24). Au cours de la période d’étude, la fréquence des visites chez le MPL avant le diagnostic a été constante, et un déclin statistiquement significatif du nombre moyen ajusté de visites après le diagnostic a été observé. Ceux ayant le plus recours au MPL étaient moins susceptibles de mourir, mais plus susceptibles d’être hospitalisés indépendamment de la comorbidité.

      Conclusions

      La fréquence des visites chez le MPL a été élevée avant et après le diagnostic. L’augmentation du recours au MPL a été associée à un risque plus faible de mortalité. Cependant, il augmente le coût des soins des patients souffrant d’hypertension. Par conséquent, d’autres études sont nécessaires pour déterminer le niveau optimal requis pour atteindre un bon rapport coût-efficacité du recours aux effectifs du MPL.
      Hypertension is 1 of the most common chronic conditions worldwide,
      • Kearney P.M.
      • Whelton M.
      • Reynolds K.
      • et al.
      Global burden of hypertension: analysis of worldwide data.
      affecting 23% of the Canadian adult population.
      • Robitaille C.
      • Dai S.
      • Waters C.
      • et al.
      Diagnosed hypertension in Canada: incidence, prevalence and associated mortality.
      It is an important modifiable risk factor for cardiovascular disease and the most common attributable cause for mortality in the world.
      • Ezzati M.
      • Lopez A.D.
      • Rodgers A.
      • Vander Hoorn S.
      • Murray C.J.
      Selected major risk factors and global and regional burden of disease.
      • Kannel W.B.
      Blood pressure as a cardiovascular risk factor: prevention and treatment.
      More patients visit primary care physicians (PCPs) and receive prescriptions for the treatment of hypertension than for any other medical disorder,
      • Morgan S.
      Drug spending in Canada: recent trends and causes.
      • Baum C.
      • Kennedy D.L.
      • Forbes M.B.
      • Jones J.K.
      Drug use and expenditures in 1982.
      with more than 4 million prescriptions for antihypertensive medications written every month in Canada.
      • Campbell N.R.
      • McAlister F.A.
      • Brant R.
      • et al.
      Temporal trends in antihypertensive drug prescriptions in Canada before and after introduction of the Canadian Hypertension Education Program.
      The cost in high-income countries' health care systems to manage patients with hypertension and related complications is estimated to consume 10% of all health care spending.

      World Health Organization. 2008-2013 action plan for global strategy for the prevention and control of noncommunicable diseases, 2008. Available at: http://whqlibdoc.who.int/publications/2009/9789241597418_eng.pdf. Accessed April 7, 2014.

      • Cecchini M.
      • Sassi F.
      • Lauer J.A.
      • et al.
      Tackling of unhealthy diets, physical inactivity, and obesity: health effects and cost-effectiveness.
      A better understanding of the use health care resources for hypertension will enable targeted strategies to be developed to facilitate more cost-effective care. Even small changes in health care costs to manage hypertension will have a large overall impact given the high prevalence of hypertension in Canada. With an aging Canadian population, it is expected that there will be an increase in the burden of hypertension-related chronic illnesses, multiple comorbidities, and disability as well as the demand for health care services.
      • Hogan S.
      How will the ageing of the population affect health care needs and costs in the foreseeable future?.
      Hypertension Canada consists of policy makers and health care professionals who have made extensive efforts to improve hypertension detection and management through the development of ongoing national knowledge translation strategies.
      • Harris S.
      • Lank C.
      Recommendations from the Canadian Diabetes Association. 2003 guidelines for prevention and management of diabetes and related cardiovascular risk factors.
      • Khan N.A.
      • Hemmelgarn B.
      • Herman R.J.
      • et al.
      The 2008 Canadian Hypertension Education Program recommendations for the management of hypertension: part 2—therapy.
      Currently in Canada, there are no mechanisms in place to monitor hypertension management, hypertension-related outcomes, and health care resource implications. Inefficient care will lead to a significant burden on publically funded health care systems through either inefficient management or excess hypertension-attributable end-organ complications, with attendant implications on health and health care costs. Thus, at a time of unprecedented pressure on our health care system, health care leaders are driven to reduce waste and inefficiency. Identifying inefficiencies and factors associated with clinical care gaps in hypertension management is imperative for designing policy and intervention programs to close care gaps.
      To the best of our knowledge, no studies have documented the pattern of PCP visits after a new diagnosis of hypertension in the universal health care system in Canada and whether the frequency of these visits varies with the presence of other comorbid conditions and socioeconomic factors. Given the prevalence and importance of hypertension in the outpatient setting, particularly PCP offices, this study focused on evaluating the PCP use patterns among patients with incident hypertension with and without other comorbidities.

      Methods

      Data sources

      Administrative health databases from Alberta were linked using an anonymized unique personal identifier. The study databases included provincial health insurance registries, hospital discharge abstracts, and physician billing claims. The Alberta insurance registry contains demographic information including date of birth, sex, and mailing address for all residents of the province eligible to receive health services. Discharge abstracts contain clinical information for all hospitalized patients, with up to 25 diagnosis codes recorded for each hospitalization using the International Classification of Diseases (ICD), Ninth Revision (ICD-9) and the ICD-9 Clinical Modification (ICD-9-CM) before April 2002 or the ICD-10 Canadian Modification (ICD-10-CA) after April 2002.
      • Canadian Institute for Health Information
      International Statistical Classification of Disease and related health problems: tenth revision Canada [ICD-10-CA].
      • National Center for Health Statistics
      Centers for Disease Control and Prevention. International Classification of Diseases—Clinical modification, ninth revision.
      • World Health Organization
      International Classification of Diseases, ninth revision.
      Physician billing claims contain fee-for-service billing information that includes a unique physician identifier and ICD-9 codes for services provided. At least 1 and up to 3 ICD-9 codes are recorded, corresponding to the primary reasons for each physician visit. Physicians submit claims for payment or reporting services to provincial government insurance programs. The physician claims capture outpatient physician services and the majority of inpatient physician services, and each claim records the billing physician's specialist type.

      Study population

      We defined a cohort of patients newly diagnosed with hypertension between April 1, 1998 and March 31, 2009 (fiscal years 1998-2008). Hypertension cases were identified using a previously validated case definition for Canadian hospital discharge and physician claims administrative databases, which was 75% sensitive and 94% specific, with a positive predictive value of 81% when evaluated in the Alberta data sets in the same time frame as this study.
      • Quan H.
      • Khan N.
      • Hemmelgarn B.R.
      • et al.
      Validation of a case definition to define hypertension using administrative data.
      • Tu K.
      • Campbell N.R.C.
      • Chen Z.L.
      • Cauch-Dudek K.J.
      • McAlister F.A.
      Accuracy of administrative databases in identifying patients with hypertension.
      The first date a hypertension code was specified was defined as the date of diagnosis. We excluded patients with an index date between April 1, 1994 and March 31, 1997 (3-year washout period) to focus on patients with incident hypertension.

      Outcomes

      The primary outcome was the mean number of outpatient PCP visits in the year after the diagnosis of hypertension. Because the majority of patients with hypertension are managed in the community by a PCP (Fig. 1), only outpatient PCP services were analyzed in this study. Hypertension-related PCP visits and the number of patients with no visits were secondary outcomes. Hypertension-related PCP visits were identified using ICD-9 codes 401.x, 402.x, 403.x, 404.x, or 405.x. Patients with no PCP outpatient visits in the year after diagnosis were considered to have not used a PCP. To assess the association between PCP use and outcomes, we considered death, all-cause hospitalization, cardiovascular-specific hospitalization (ICD codes G45, G46, I60-I69, H340, I43, I50, I099, I110, I130, I132, I255, I420, I425-I429, P290, I21, I22, and I252 as the most responsible diagnosis) and hypertension-specific hospitalization (ICD codes 401.x, 402.x, 403.x, 404.x, or 405.x as the most responsible diagnosis). All outcomes were defined 1 year after the hypertension diagnosis.
      Figure thumbnail gr1
      Figure 1Patient flowchart. HTN, hypertension.

      Exposure variables

      Comorbidity status, defined using the Charlson Comorbidity Index, was the exposure variable. All comorbidities were derived from validated ICD-9/ICD-10 algorithms using both the hospital discharge abstract data and the physician claims database for the 3 years before the hypertension diagnosis.
      • Quan H.
      • Sundararajan V.
      • Hlafon P.
      • et al.
      Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.
      • Quan H.
      • Li B.
      • Saunders D.
      • et al.
      Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database.
      Three categories of comorbidity were considered: (1) no comorbidity, (2) vascular risk–related comorbidity (myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, diabetes with and without complications, and renal disease), and (3) unrelated comorbidity (unrelated to vascular risk, including dementia, chronic pulmonary obstructive disease, connective tissue disease, rheumatic disease, peptic ulcer disease, paraplegia/hemiplegia, liver disease, cancer/metastatic carcinoma, and acquired immunodeficiency syndrome/human immunodeficiency virus infection). In the analysis considering clinical outcomes, PCP use was the exposure variable and was categorized based on use within other health care settings as ≤ 2 visits per year, 2-5 visits per year and ≥ 5 visits per year.
      • LaCalle E.
      • Rabin E.
      Frequent users of the emergency departments: The myths, the data and the policy implications.

      Independent variables

      Age and sex at index date of hypertension diagnosis were defined using the registry data. Age was categorized into 6 groups: < 35 years, 35-45 years, 45-55 years, 55-65 years, 65-75 years, and > 75 years. Median income quintile was assigned using 2001 Statistics Canada Census data and each patient's postal code forward sortation area.

      Statistical analysis

      Baseline characteristics of patients with incident hypertension were analyzed descriptively. To provide a baseline measure of use, all outcomes were calculated for both the year preceding and the year after hypertension diagnosis. The number of PCP visits in the year preceding and the year after diagnosis was estimated using zero-inflation Poisson regression to account for the overdispersion resulting from excessive zero counts. If the Poisson assumption that variance is equal to the mean was not met, negative binominal regression was used. If hypertension was defined using physician claims, the 2 visits required to meet the definition of hypertension were not included in estimates of the mean number of visits. Both crude and adjusted means are reported. The adjusted model included age group, sex, income quintile, rural/urban residence, and year of diagnosis. To assess if use patterns changed over time, the mean number of PCP visits was calculated by year of diagnosis. A χ2 test for trend was completed for each comorbidity group over time. A proportional hazard Cox model was completed to assess the association between PCP use and clinical outcomes, including death and all-cause, cardiovascular, and hypertension-specific hospitalization. Because of the time-dependent nature of PCP exposure, only adjusted analyses were considered. A time-dependent variable was included to account for the relationship between PCP use and exposure time. All analyses were adjusted for age, sex, income quintile, rural/urban dwelling, and year of diagnosis. The statistical analyses were conducted using Statistical Analysis software (SAS), version 9.3 (SAS Institute, Cary, NC). Ethics approval was obtained from the Conjoint Health Research Ethics Board at the University of Calgary.

      Results

      In total, 456,263 patients (mean age, 57.6 years; 50.6% men) with newly diagnosed hypertension were identified in fiscal years 1998-2008, the majority of whom (62.5%) had no comorbidities at the time of the hypertension diagnosis (Table 1). Patients with newly diagnosed hypertension and no comorbidities were more likely to be younger, have a higher income, and be urban dwellers (Table 1). The vast majority of our cohort was diagnosed with hypertension on the basis of 2 outpatient clinic visits (90.1%).
      Table 1Baseline characteristics
      VariableOverallComorbidity status
      NoneVascular relatedUnrelated
      Total patients (n)45,626328,559996,46774,197
      Male (%)50.649.558.345.0
      Age (%)
       < 35 y5.87.12.55.3
       35-45 y13.916.57.012.7
       45-55 y25.829.517.322.9
       55-65 y23.223.422.623.1
       65-75 y17.815.223.720.4
       > 75 y13.48.326.915.6
       Mean age (SD)57.6 (14.7)54.9 (13.8)64.6 (14.7)59.0 (14.8)
      Income quintile (%)
       1 (lowest)20.318.524.721.7
       219.118.320.720.0
       318.718.419.119.4
       417.918.616.417.3
       5 (highest)17.419.313.615.5
       Missing6.67.05.76.1
      Region of residence (%)
       Rural18.817.721.020.1
       Urban74.775.373.273.8
       Missing6.67.05.76.1
      SD, standard deviation.
      In the year preceding the diagnosis of hypertension, 7,386,681 physician claims were submitted within our cohort (Fig. 1). Of those, 4,933,038 were excluded because they were for diagnostic and therapeutic services, in-patient services, day surgical procedures, or emergency department visits. An additional 220,346 claims (3.0% of the total number of claims) were excluded because the service provider specialty was an internal medicine specialist in an outpatient clinic rather than a PCP (eg, it was a visit with general internal medicine, cardiology, or nephrology physicians). The remaining 2,233,297 outpatient PCP claims were included in the analysis in the year preceding hypertension diagnosis. Similarly, in the year after the hypertension diagnosis, 8,888,886 physician claims were submitted. Of those, 5,499,698 were excluded because they were for diagnostic and therapeutic services, in-patient services, day surgical procedures, or emergency department visits, and 271,710 were excluded (3.1% of the total number of claims) because they were internal medicine specialist outpatient visits. The remaining 3,389,188 outpatient PCP claims in the year after diagnosis were included. Of note, only 4228 of our incident hypertension cohort (0.93%) were seen solely by an internal medicine specialist as an outpatient in the year after diagnosis and had no PCP visits.
      More than 88% of all patients had seen a PCP at least once in the year before the diagnosis of hypertension (Supplemental Table S1), and the frequency of PCP visits for any indication increased after the diagnosis of hypertension in all 4 comorbidity-defined subgroups. As expected, the mean number of visits among patients with comorbidities was higher than those with no comorbidities both before and after the diagnosis of hypertension. Generally, the mean number of visits was higher among women than men and among older age groups than younger age groups, but use patterns did not vary substantially across rural vs urban regions or by socioeconomic status. The majority of the increase in PCP visits observed from the year preceding diagnosis to the year after diagnosis was accounted for by hypertension-related visits (Supplemental Table S1). The absolute (and relative) increase was greater in those with no comorbidities compared with those with at least 1 comorbidity. Individuals younger than 35 years of age, men, and those from lower income quintiles had higher percentages of patients with no PCP visits (Supplemental Table S2). However, it should be noted that even in these “underserved” subgroups, more than 90% of patients had at least 1 PCP visit in the year after the diagnosis of hypertension.
      The mean number of PCP visits before the hypertension diagnosis remained stable from 1998-2008 across all comorbidity subgroups (Fig. 2) (no comorbidity, P = 0.47; vascular risk related, P = 0.93; unrelated, P = 0.41). The adjusted model included age, sex, urban/rural dwelling, and income quintile. However, the adjusted mean number of visits after diagnosis appeared to decline over time (Fig. 3), with a statistically significant decline among those with no comorbidities (P = 0.01), vascular risk–related comorbidities (P < 0.001), and unrelated comorbidities (P = 0.002).
      Figure thumbnail gr2
      Figure 2Adjusted mean number of primary care physician (PCP) visits in the year preceding hypertension diagnosis stratified by comorbidity subgroups, 1998-2008. Results are adjusted for sex, age, urban/rural dwelling, and income quintile.
      Figure thumbnail gr3
      Figure 3Adjusted mean number of primary care physician (PCP) visits in the year after hypertension diagnosis stratified by comorbidity subgroups, 1998-2008. Results are adjusted for sex, age, urban/rural dwelling, and income quintile.
      There were large variations in the risk of death across the comorbidity subgroups (Table 2). Compared with patients with ≤ 2 PCP visits, those with ≥ 5 visits had a significantly lower risk of death (ranging from a hazard ratio of 0.07; 95% confidence interval [CI], 0.06-0.07 for those with vascular-related comorbidities to a hazard ratio of 0.16; 95% CI, 0.15-0.18 for those with none). As consistently found in other clinical groups, those with higher PCP use have higher rates of hospitalization; high users of PCPs are high users of the health care system more generally.
      • LaCalle E.
      • Rabin E.
      Frequent users of the emergency departments: The myths, the data and the policy implications.
      • Glynn L.
      • Valderas J.
      • Healy P.
      • et al.
      The prevalence of multimorbidity in primary care and its effect on health care utilization and cost.
      • Weinberger M.
      • Oddone E.
      • Henderson W.
      for the Veterans Affairs Cooperative Study Group on Primary Care and Hospital Readmission
      Does increased access to primary care reduce hospital readmissions?.

      Yang J, Nijjar A, Quan H, et al. Utilization of health resources in South Asian, Chinese and White patients with diabetes mellitus [e-pub ahead of print]. Prim Care Diabetes doi: 10.1016/j.pcd.2013.11.001.

      Table 2Adjusted hazard ratios by primary care physician use category
      Adjusted for time, age, sex, income quintile, rural/urban residence, and year of diagnosis.
      CategoryComorbidity status
      NoneVascular relatedUnrelated
      ≤ 23-5≥ 5≤ 23-5≥ 5≤ 23-5≥ 5
      All-cause death in the year after diagnosis (95% CI)Reference0.31 (0.28-0.33)0.17 (0.16-0.18)Reference0.16 (0.15-0.18)0.07 (0.07-0.07)Reference0.21 (0.19-0.24)0.12 (0.11-0.13)
      All-cause hospitalization in the year after diagnosis (95% CI)Reference1.13 (1.07-1.19)1.57 (1.51-1.64)Reference1.42 (1.33-1.52)2.06 (1.96-2.16)Reference1.21 (1.11-1.32)1.67 (1.56-1.79)
      Cardiovascular-specific hospitalization in the year after diagnosis (95% CI)Reference1.09 (0.93-1.29)2.17 (1.90-2.48)Reference1.10 (0.97-1.24)1.51 (1.38-1.65)Reference1.01 (0.81-1.27)1.30 (1.09-1.54)
      Hypertension-specific hospitalization in the year after diagnosis (95% CI)Reference0.98 (0.52-1.86)1.55 (0.94-2.56)Reference0.78 (0.46-1.31)0.72 (0.49-1.05)Reference1.02 (0.20-5.05)2.78 (0.86-9.06)
      CI, confidence interval.
      Adjusted for time, age, sex, income quintile, rural/urban residence, and year of diagnosis.

      Discussion

      Among this population-based cohort of patients with incident hypertension, we found that > 88% of patients visited a PCP at least once in the year before diagnosis, and > 94% saw a PCP in the year after the hypertension diagnosis. The mean number of visits after diagnosis increased by approximately 2 (in patients without comorbidities) and approximately 1.5 (in patients with at least 1 comorbidity) compared with the preceding year, over and above the 2 visits required for meeting the hypertension case definition. The majority of the increase was accounted for by visits coded by the attending physician as being hypertension related. Patterns of prediagnosis PCP visits did not change appreciably over the 10-year study period; however, there was a small but real decline in the mean number of PCP visits in the year after diagnosis. Finally, increased PCP use was associated with a lower risk of death but increased risk of hospitalization.
      There are several possible explanations for the decrease in visits over time. First, patients may be more likely to be seen by specialists in more recent years. However, in an exploratory analysis, the crude mean number of visits to non-PCPs (including surgeons, internists, cardiologists, nephrologist, and so on) had decreased from 0.52 (0.50-0.53) in 1998 to 0.38 (0.37-0.39) in 2008 among those with no comorbidities and from 1.57 (1.49-1.65) in 1998 to 1.13 (1.08-1.20) in 2008 among those with any comorbidity. Thus, the decrease in PCP visits was mirrored by a decrease in specialist visits, arguing against a shift from PCP follow-up to specialist follow-up.
      Second, patients may be achieving better control of their blood pressure and thus require fewer visits. Although this study is unable to provide evidence to support or refute this claim, because we do not have patient-level blood pressure measurements, other studies have found no relationship between blood pressure control and the intensity of PCP use.
      • Birtwhistle R.V.
      • Godwin M.S.
      • Dleva M.D.
      • et al.
      Randomised equivalence trial comparing three month and six month follow up of patients with hypertension by family practitioners.
      Thus, the decline in the mean number of visits over time is unlikely to be attributable to better blood pressure control.
      Third, the decline over time may be related to changes in mortality over time, combined with a younger age of onset for the diagnosis of hypertension. If the 1-year death risk has significantly increased among patients with newly diagnosed hypertension, the mean PCP use may appear to decrease over time simply because of at-risk time. However, national Canadian data demonstrate that the risk of death has decreased over time,

      Public Health Agency of Canada. Report from the Canadian Chronic Disease Surveillance System: Hypertension in Canada, 2010. Available at: http://www.phac-aspc.gc.ca/cd-mc/cvd-mcv/ccdss-snsmc-2010/pdf/CCDSS_HTN_Report_FINAL_EN_20100513.pdf. Accessed April 7, 2014.

      and with improvements in identification, treatment, and education programs available, patients are diagnosed and treated at a younger age.
      • Campbell N.R.
      • McAlister F.A.
      • Quan H.
      for the Hypertension Outcomes Research Task Force
      Monitoring and Evaluating Efforts to Control Hypertension in Canada: why, how and what it tells us needs to be done about current care gaps.
      Finally, changes in models of care may have impacted the PCP use rates. However, primary care networks (models of care that provide comprehensive disease management programs that may result in patients being seen by a nurse as opposed to a PCP) were introduced in Alberta in 2002. Our study documents a constant decrease in PCP visits from 1998 onward in both complicated and uncomplicated patients with hypertension arguing that changes in care models cannot account for all the observed decrease and that the impact would have affected all groups equally.
      Our study suggests that PCP access at the population level is being achieved; the vast majority of patients had PCP visits before and after diagnosis, and use patterns did not vary substantially by rural/urban regions or by income quintile. This finding disproves 1 hypothesis for why an earlier Alberta-based study found markedly higher rates of ambulatory care–sensitive hospitalizations among uncomplicated hypertensive patients from rural areas or those with lower household income levels.
      • Walker R.L.
      • Chen G.
      • McAlister F.A.
      • et al.
      Uncomplicated hypertension: an ambulatory care sensitive condition.
      Given the emphasis in current hypertension guidelines on absolute cardiovascular risk and the need to address all cardiovascular risk factors, we had anticipated that patients with multiple cardiovascular risk factors would have exhibited a greater increase in PCP visit frequency after the diagnosis of hypertension than those without any other cardiovascular risk factors. Indeed, our finding that this was not true perhaps helps explain recent reports that other cardiovascular risk factors are not any better controlled in patients with hypertension than in those without hypertension.
      • Benetos A.
      • Thomas F.
      • Bean K.E.
      • Guize L.
      Why cardiovascular mortality is higher in treated hypertensive patients versus subjects of the same age, in the general population.
      • Gu Q.
      • Burt V.L.
      • Paulose-Ram R.
      • Yoon S.
      • Gillum R.F.
      High blood pressure and cardiovascular disease mortality risk among US adults: the Third National Health and Nutrition Examination Survey mortality follow-up study.
      • Gudmindsson L.S.
      • Johannsson M.
      • Thorgeirsson G.
      • et al.
      Risk profiles and prognosis of treated and untreated hypertensive men and women in a population-based longitudinal study: the Reykavik Study.
      • McAlister F.A.
      • Robitaille C.
      • Gillespie C.
      • et al.
      The impact of cardiovascular risk factor profiles on blood pressure control rates in adults from Canada and the United States.
      Alternatively, perhaps these patients had few reasons to see a PCP previously; however, with the diagnosis of hypertension, they are required to visit more frequently for the evaluation of possible underlying causes and risk factors (eg, hyperlipidemia, diabetes, or renal function impairment).
      Our finding that increased PCP use was associated with a decrease in the risk of death but an increase in the risk of hospitalization is similar to findings from other groups.
      • LaCalle E.
      • Rabin E.
      Frequent users of the emergency departments: The myths, the data and the policy implications.
      • Glynn L.
      • Valderas J.
      • Healy P.
      • et al.
      The prevalence of multimorbidity in primary care and its effect on health care utilization and cost.
      • Weinberger M.
      • Oddone E.
      • Henderson W.
      for the Veterans Affairs Cooperative Study Group on Primary Care and Hospital Readmission
      Does increased access to primary care reduce hospital readmissions?.
      There are several possible explanations for this finding. First, the result may be caused by the time-dependent nature of the PCP user definition. To be classified as a high user, a patient must survive long enough to complete 5 or more PCP visits. Second, there may be clinical and demographic differences between the different PCP use cohorts; patients who see their PCP more often may have different underlying health issues than those who do not. Indeed, it is likely that high PCP users are more likely to have multiple comorbidities and require more frequent health care visits to manage their diseases. However, the optimal PCP use to achieve maximum clinical benefit without incurring additional, perhaps unrequired, physician visit costs remains unknown. Future work should focus on assessing the appropriate number of PCP visits required to maximize health benefit.
      Our study has limitations. First, only visits with a PCP in an outpatient office are included. This definition excludes any visits to PCP clinics within a hospital. However, given that the majority (90%) of primary care physicians in Alberta work in outpatient clinics,

      The College of Family Physicians of Canada, Canadian Medical Association, The Royal College of Physicians and Surgeons of Canada. 2010 National Physician Survey. Available at: http://nationalphysiciansurvey.ca/surveys/2010-survey/. Accessed April 7, 2014.

      this is unlikely to introduce substantial bias. In addition, any outpatient visits to a specialist—such as an internist, nephrologist, or cardiologist—were not considered in our main analysis; however, this bias will be small because those visits accounted for only 3% of the physician claims in both the prediagnosis and postdiagnosis intervals. To be included in our cohort, patients had to meet the definition of hypertension using administrative data (either 2 physician claims or 1 hospitalization). The visits required to meet the definition of hypertension are excluded from the number of visits in all our analyses. Thus, the data we report is for the number of PCP visits over and above the 2 needed for meeting the case definition. Our analysis does not take into account physician visits for hospitalized patients, but only 1.6% of this cohort was hospitalized in the year after the initial diagnosis of hypertension. In addition, patients who died in the first year after diagnosis of hypertension contributed follow-up intervals that were shorter than 1 year. Because the death risk varied across comorbidity subgroups, we completed a sensitivity analysis calculating the mean PCP use in the year after diagnosis in the cohort of patients that survived the entire year of follow-up to ensure that death was not a source of differential bias across comorbidity subgroups. Our results were remarkably similar, demonstrating that death was not a source of significant bias in our analysis. Finally, we analyzed Alberta data, and the resource use patterns we observed may differ in other Canadian provinces. This study should be replicated in other jurisdictions.
      In conclusion, we found that the frequency of PCP visits for patients with a new diagnosis of hypertension was very high, with > 88% of patients visiting a PCP at least once before diagnosis, and > 94% of patients seeing a PCP at least once in the year after diagnosis, even after excluding diagnosis-related visits. We found a small but statistically significant decline over time in the frequency of PCP visits after the diagnosis of hypertension. Future studies should examine the optimal patterns of outpatient visits for patients with hypertension to achieve the most cost-effective use of scarce PCP resources.

      Acknowledgements

      The following are members of the Hypertension and Outcomes Surveillance Team of the Canadian Hypertension Education Program: Oliver Baclic, Gillian Bartlett, Debra Butt, Norm Campbell, Guanmin Chen, Sulan Dai, Brenda Hemmelgarn, Michael Hill, Helen Johansen, Nadia Khan, Lisa Lix, Finlay McAlister, Jay Onysko, Hude Quan, Mark Smith, Larry Svenson, Gary Teare, Karen Tu, Robin Walker, and Andy Wielgosz.
      This study is based in part on deidentified data provided by Canadian provincial health ministries . The interpretation and conclusions contained herein are those of the researchers and do not represent the views of these provincial governments. The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding sources. No official endorsement by Manitoba Health is intended or should be inferred.

      Funding Sources

      H.Q. and F.A.M. receive salary support from Alberta Innovates—Health Solutions. F.A.M. holds the University of Alberta Chair in Cardiac Outcomes Research. K.T. is supported by a Fellowship in Primary Care Research from the Canadian Institute for Health Research . N.R.C.C. holds the Heart and Stroke Foundation of Canada CIHR Chair in Hypertension Prevention and Control. B.R.H. is supported by the Roy and Vi Baay Chair in Kidney Research .

      Disclosures

      The authors have no conflicts of interest to disclose.

      Supplementary Material

      References

        • Kearney P.M.
        • Whelton M.
        • Reynolds K.
        • et al.
        Global burden of hypertension: analysis of worldwide data.
        Lancet. 2005; 365: 217-223
        • Robitaille C.
        • Dai S.
        • Waters C.
        • et al.
        Diagnosed hypertension in Canada: incidence, prevalence and associated mortality.
        CMAJ. 2012; 184: E49-E56
        • Ezzati M.
        • Lopez A.D.
        • Rodgers A.
        • Vander Hoorn S.
        • Murray C.J.
        Selected major risk factors and global and regional burden of disease.
        Lancet. 2002; 360: 1347-1360
        • Kannel W.B.
        Blood pressure as a cardiovascular risk factor: prevention and treatment.
        JAMA. 1996; 275: 1571-1576
        • Morgan S.
        Drug spending in Canada: recent trends and causes.
        Med Care. 2004; 42: 635-642
        • Baum C.
        • Kennedy D.L.
        • Forbes M.B.
        • Jones J.K.
        Drug use and expenditures in 1982.
        JAMA. 1985; 253: 382-386
        • Campbell N.R.
        • McAlister F.A.
        • Brant R.
        • et al.
        Temporal trends in antihypertensive drug prescriptions in Canada before and after introduction of the Canadian Hypertension Education Program.
        J Hypertens. 2003; 21: 1591-1597
      1. World Health Organization. 2008-2013 action plan for global strategy for the prevention and control of noncommunicable diseases, 2008. Available at: http://whqlibdoc.who.int/publications/2009/9789241597418_eng.pdf. Accessed April 7, 2014.

        • Cecchini M.
        • Sassi F.
        • Lauer J.A.
        • et al.
        Tackling of unhealthy diets, physical inactivity, and obesity: health effects and cost-effectiveness.
        Lancet. 2010; 376: 1775-1784
        • Hogan S.
        How will the ageing of the population affect health care needs and costs in the foreseeable future?.
        Commission of the Future of Health Care in Canada, Ottawa, CanadaOctober 2002 (Available at:) (Accessed April 7, 2014)
        • Harris S.
        • Lank C.
        Recommendations from the Canadian Diabetes Association. 2003 guidelines for prevention and management of diabetes and related cardiovascular risk factors.
        Can Fam Physician. 2004; 50: 425-433
        • Khan N.A.
        • Hemmelgarn B.
        • Herman R.J.
        • et al.
        The 2008 Canadian Hypertension Education Program recommendations for the management of hypertension: part 2—therapy.
        Can J Cardiol. 2014; 24: 465-475
        • Canadian Institute for Health Information
        International Statistical Classification of Disease and related health problems: tenth revision Canada [ICD-10-CA].
        Canada, Ottawa, Ontario2003
        • National Center for Health Statistics
        Centers for Disease Control and Prevention. International Classification of Diseases—Clinical modification, ninth revision.
        National Center for Health Statistics, Atlanta, GA1978
        • World Health Organization
        International Classification of Diseases, ninth revision.
        World Health Organization, Geneva, Switzerland1976
        • Quan H.
        • Khan N.
        • Hemmelgarn B.R.
        • et al.
        Validation of a case definition to define hypertension using administrative data.
        Hypertension. 2009; 546: 1423-1428
        • Tu K.
        • Campbell N.R.C.
        • Chen Z.L.
        • Cauch-Dudek K.J.
        • McAlister F.A.
        Accuracy of administrative databases in identifying patients with hypertension.
        Open Med. 2007; 1: 3-5
        • Quan H.
        • Sundararajan V.
        • Hlafon P.
        • et al.
        Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.
        Med Care. 2005; 43: 1130-1139
        • Quan H.
        • Li B.
        • Saunders D.
        • et al.
        Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database.
        Health Serv Res. 2008; 43: 1424-1441
        • LaCalle E.
        • Rabin E.
        Frequent users of the emergency departments: The myths, the data and the policy implications.
        Ann Emerg Med. 2010; 56: 42-48
        • Glynn L.
        • Valderas J.
        • Healy P.
        • et al.
        The prevalence of multimorbidity in primary care and its effect on health care utilization and cost.
        Fam Pract. 2014; 28: 516-523
        • Weinberger M.
        • Oddone E.
        • Henderson W.
        • for the Veterans Affairs Cooperative Study Group on Primary Care and Hospital Readmission
        Does increased access to primary care reduce hospital readmissions?.
        N Engl J Med. 1996; 334: 1441-1447
      2. Yang J, Nijjar A, Quan H, et al. Utilization of health resources in South Asian, Chinese and White patients with diabetes mellitus [e-pub ahead of print]. Prim Care Diabetes doi: 10.1016/j.pcd.2013.11.001.

        • Birtwhistle R.V.
        • Godwin M.S.
        • Dleva M.D.
        • et al.
        Randomised equivalence trial comparing three month and six month follow up of patients with hypertension by family practitioners.
        BMJ. 2004; 204: 328
      3. Public Health Agency of Canada. Report from the Canadian Chronic Disease Surveillance System: Hypertension in Canada, 2010. Available at: http://www.phac-aspc.gc.ca/cd-mc/cvd-mcv/ccdss-snsmc-2010/pdf/CCDSS_HTN_Report_FINAL_EN_20100513.pdf. Accessed April 7, 2014.

        • Campbell N.R.
        • McAlister F.A.
        • Quan H.
        • for the Hypertension Outcomes Research Task Force
        Monitoring and Evaluating Efforts to Control Hypertension in Canada: why, how and what it tells us needs to be done about current care gaps.
        Can J Cardiol. 2013; 29: 564-570
        • Walker R.L.
        • Chen G.
        • McAlister F.A.
        • et al.
        Uncomplicated hypertension: an ambulatory care sensitive condition.
        Can J Cardiol. 2013; 29: 1462-1469
        • Benetos A.
        • Thomas F.
        • Bean K.E.
        • Guize L.
        Why cardiovascular mortality is higher in treated hypertensive patients versus subjects of the same age, in the general population.
        J Hypertens. 2003; 21: 1635-1640
        • Gu Q.
        • Burt V.L.
        • Paulose-Ram R.
        • Yoon S.
        • Gillum R.F.
        High blood pressure and cardiovascular disease mortality risk among US adults: the Third National Health and Nutrition Examination Survey mortality follow-up study.
        Ann Epidemiol. 2008; 18: 302-309
        • Gudmindsson L.S.
        • Johannsson M.
        • Thorgeirsson G.
        • et al.
        Risk profiles and prognosis of treated and untreated hypertensive men and women in a population-based longitudinal study: the Reykavik Study.
        J Hum Hypertens. 2004; 18: 615-622
        • McAlister F.A.
        • Robitaille C.
        • Gillespie C.
        • et al.
        The impact of cardiovascular risk factor profiles on blood pressure control rates in adults from Canada and the United States.
        Can J Cardiol. 2013; 29: 598-605
      4. The College of Family Physicians of Canada, Canadian Medical Association, The Royal College of Physicians and Surgeons of Canada. 2010 National Physician Survey. Available at: http://nationalphysiciansurvey.ca/surveys/2010-survey/. Accessed April 7, 2014.