Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
ORIGINAL ARTICLE Emergency Department Visits and Primary Care Among Adults With Chronic Conditions Jane McCusker, MD, DrPH,*† Danièle Roberge, PhD,‡§ Jean-Frédéric Lévesque, MD, PhD,¶ Antonio Ciampi, PhD,*† Alain Vadeboncoeur, MD,㛳 Danielle Larouche, MSc,‡ and Steven Sanche, MSc* Background: An emergency department (ED) visit may be a marker for limited access to primary medical care, particularly among those with ambulatory care sensitive chronic conditions (ACSCC). Objectives: In a population with universal health insurance, to examine the relationships between primary care characteristics and location of last general physician (GP) contact (in an ED vs. elsewhere) among those with and without an ACSCC. Research Design: A cross-sectional survey using data from 2 cycles of the Canadian Community Health Survey carried out in 2003 and 2005. Subjects: The study sample comprised Québec residents aged ⱖ18 who reported at least one GP contact during the previous 12 months, and were not hospitalized (n ⫽ 33,491). Measures: The primary outcome was place of last GP contact: in an ED versus elsewhere. Independent variables included the following: lack of a regular physician, perceived unmet healthcare needs, perceived availability of health care, number of contacts with doctors and nurses, and diagnosis of an ACSCC (hypertension, heart disease, chronic respiratory disease, diabetes). Results: Using multiple logistic regression, with adjustment for sociodemographic, health status, and health services variables, lack of a regular GP and perceptions of unmet needs were associated with last GP contact in an ED; there was no interaction with ACSCC or other chronic conditions. Conclusions: Primary care characteristics associated with GP contact in an ED rather than another site reflect individual characteristics (affiliation with a primary GP and perceived needs) rather than From the *Department of Clinical Epidemiology and Community Studies, St Mary’s Hospital, Montreal, QC, Canada; †Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, QC, Canada; ‡Centre de Recherche de l’Hôpital Charles LeMoyne, Longueuil, QC, Canada; §Department des sciences communautaires de la Faculté de Médecine, Université de Sherbrooke, QC, Canada; ¶Centre de recherche du CHUM et Institut National de Santé Publique du Québec, Montréal, Canada; and 㛳Emergency Medicine Services, Montreal Institute of Cardiology, Montreal, QC, Canada. Supported by Fonds de la Recherche en Santé du Québec. Presented at the Annual Meeting of the Canadian Association for Health Services and Policy Research; May 11–14, 2009; Calgary, Alberta. Reprints: Jane McCusker, MD, DrPH, Department of Clinical Epidemiology and Community Studies, St. Mary’s Hospital, Montreal, Québec H3T 1M5, Canada. E-mail: [email protected]. Copyright © 2010 by Lippincott Williams & Wilkins ISSN: 0025-7079/10/4811-0972 972 | www.lww-medicalcare.com the geographic availability of healthcare, both among those with and without chronic conditions. Key Words: emergency department, primary care, chronic disease, cross-sectional study (Med Care 2010;48: 972–980) T he emergency department (ED) plays multiple functions for populations with chronic conditions. It provides a source of care in medical, surgical, and traumatic emergencies, but also functions as an alternative source for primary medical care.1 Certain characteristics of primary care (eg, lack of a regular physician, unmet needs for healthcare, poor continuity of care, perceived lack of rapid access to care) have been associated with an increased likelihood of making an ED visit,2– 8 although the evidence is inconsistent.9 Even under Canada’s universal health insurance program there may be differential access to primary care; certain population subgroups (recent immigrants, those with serious mental health problems) may be more likely to seek care in the ED than in other ambulatory care settings.10 Certain chronic medical conditions (eg, diabetes, asthma, congestive heart failure) are considered to be particularly sensitive to access to primary care services; hospitalization rates for these conditions are used in many jurisdictions as an indicator of limited access to primary care.11 Better primary care management of these conditions may improve the process of care and clinical outcomes12–14 and is hypothesized to reduce the probability of complications or clinical deterioration that could lead to an ED visit or hospitalization.11,15 Most research on ambulatory care sensitive chronic conditions (ACSCC) has examined ACSCC-hospitalization rates in relation to access to care,16 and among population groups with lower access to care, including racial or ethnic minority groups,17 and low income populations.18 However, the evidence relating ACSCC hospitalizations to characteristics of primary care is inconsistent and requires further validation.9,19 In US studies of those aged less than 65, rates of ACSCC-related ED visits are higher among Medicaid beneficiaries and racial minority groups.20 Among Medicaid beneficiaries, rates of ACSCC-related ED visits are higher among those without a regular source of care,21 and those who receive treatment outside community health centers.22 Medical Care • Volume 48, Number 11, November 2010 Medical Care • Volume 48, Number 11, November 2010 In this research, we aimed to extend this research to a population with universal health insurance, including all age groups, in the province of Québec, Canada. International comparative studies indicate that Québec has among the highest rates of ED visits, the lowest proportion with a regular family doctor, poor perceived accessibility by telephone to a family doctor, and the longest waiting times for an appointment.23 In a telephone survey, more than one-third of those who reported an ED visit considered that the problem could have been treated by their family physician, if available.1 We used pooled data from 2 population health surveys in Québec to investigate the relationship between primary care characteristics and use of the ED among those with and without ACSCC. Specifically, the surveys provide self-reported information on the location of last contact with health professionals, in an ED or elsewhere. We hypothesized that the relationships between characteristics of primary care and last general physician (GP) contact in an ED would be stronger in the population with ACSCC than among those with other or no chronic conditions. METHODS Data were derived from the combined Canadian Community Health Survey cycles 2.1 and 3.1, carried out in 2003 and 2005, respectively.24 The study sample comprised Québec residents aged ⱖ18. The Canadian Community Health Survey (CCHS) is a probability sample of Canadian residents aged 12 and over, excluding those in Indian Reserves, Crown lands, institutions or certain remote regions, and full-time members of the Canadian Forces. In cycle 3.1 (similar to 2.1) 3 sampling frames were used: an area frame (49%), a telephone frame (50%), and a random digit dialing frame (1%). A multistage sample was selected from the area frame, stratifying on socioeconomic and geographical variables, while clustering on household groups. Sampling from the telephone frame was random within health regions. The interviews were computer assisted and were achieved in person for the sample coming from area frame, and by telephone for the rest of the total sample. The Québec response rates were of 82.6% for the area frame and 72.5% for interviews over the telephone, for an overall response rate of 76.4%. A priori oversampling as well as a posteriori weighting adjustments corrected for nonresponse. As the 2 survey cycles include identical questions, they were combined for increased statistical power. Note that there was a very low probability of the same individual participating in both surveys. Outcome Variables Participants were asked a series of 3 questions: (1) “In the past 12 months, have you been a patient overnight in a hospital, nursing home, or convalescent home?” (2) “Not counting when you were an overnight patient, in the past 12 months, how many times have you seen or talked on the telephone about your physical, emotional, or mental health with a family doctor or general practitioner?” (3) among those with at least one such contact: “Where did the most recent contact take place?” The primary outcome (last GP contact in an ED) was defined, among participants with one © 2010 Lippincott Williams & Wilkins Emergency Department Visits or more GP contacts, as the location of last GP contact: in an ED versus all other locations. This measure was highly correlated with an equivalent measure derived from the provincial physician claims database, among 20 subgroups defined by 5 age and 4 rural-urban residence groups (Pearson’s r ⫽ 0.89; unpublished data). This cross-validation revealed a systematic effect of age but not of residence, suggesting greater underreporting of last GP contact in an ED among those aged ⱖ65 compared with other age groups. In Québec, the great majority of physicians who work in EDs are GPs. However, as patients are probably unaware of the specialty of the physician seen at an ED visit, we analyzed 2 secondary outcomes: location of last contact with specialists and nurses, based on a similar sequence of questions as the primary outcome. Specialist contacts were those with “any other medical doctor, such as a surgeon, allergist, orthopedist, gynecologist, or psychiatrist”; nurse contacts were those with “a nurse for care or advice.” Because it was impossible to determine whether or not the last ED contact resulted in hospitalization, we restricted our analyses of place of last contact to participants who reported no hospital admission during the past 12 months. We also conducted sensitivity analyses of the primary outcome that included those with a history of hospitalization. Independent Variables Sociodemographic Variables Age was analyzed as a 5-level categorical variable (18 –24, 25–34, 35– 44, 45– 64, 65⫹). Marital status was grouped in 3 categories: never married; divorced, separated, or widowed; common-law or married). Immigration status was grouped in 4 categories: born in Canada, 5 or fewer years, 6 to 10, and more than 10 years. Educational attainment was classified as follows: no secondary diploma, secondary diploma, some postsecondary studies, postsecondary diploma); income was defined as family income adjusted for family size, in quartiles. The area of residence was a 7-level variable measuring the influence of large urban core areas on residence and place of work from the highest (residence inside or close to a major urban core) to the lowest (areas with no major urban core influence).25 We aggregated levels 2, 3, and 4 because (1) the primary outcome was similar in these 3 levels, and (2) this led to the best fit (see statistical methods). Health Status Chronic conditions diagnosed by a doctor were classified into 3 categories: ACSCC (one or more of 5 conditions— diabetes, hypertension, asthma, chronic obstructive pulmonary disease, heart disease)16; other chronic conditions only; or no chronic conditions. We also created a measure of the total number of chronic conditions. Respondents reported their perceived general health status (fair or poor vs. good, very good, or excellent), health change during the past 12 months (worse vs. the same or better). Activity limitation was the extent to which the individual was limited in their usual activities (never, sometimes, often). Dependence in activities of daily living was a 3-category hierarchical variable: none, dependence only in instrumental activities (preparing meals, www.lww-medicalcare.com | 973 McCusker et al Medical Care • Volume 48, Number 11, November 2010 getting to appointments, housework, heavy household chores, personal finances), and dependence in basic activities of daily living (moving around inside the house, personal care). A measure of psychologic distress was used to identify respondents with potentially serious mental illnesses.26 High alcohol consumption was defined as reporting 5 or more consecutive drinks at least once a week. Health Services Two measures of primary care were defined as dichotomous variables: a regular or family physician (yes, no), unmet healthcare needs (yes, no). We also analyzed the perceived availability of healthcare in the community (excellent, good, or fair vs. poor); this variable was only available in cycle 2.1. Measures of health services utilization included the following receipt of homecare services (yes, no), and numbers of contacts during the past 12 months with GPs (none, 1, 2–3, 4, or more), specialists (none, 1–2, 3, or more), and nurses for care/advice (none, 1, 2, or more). Statistical Methods Combination of the 2 CCHS surveys involved sampling and bootstrap weights. These weights were divided by 2, reflecting equal importance of the 2 surveys.27 Sampling weights were used in the calculation of the estimates to correct for the sampling design and nonresponse. We used bootstrap weights to obtain the variance of the estimates because no exact formula is available.28 We used multiple logistic regression models for predicting the primary outcome, last contact with a GP in an ED, using the independent variables. The initial models reported were adjusted for age and sex. Further logistic regression modeling of the primary outcome was conducted in 2 stages with the following independent variables: (1) age, sex, number of contacts with a GP, chronic conditions, and health services variables; (2) all independent variables. Interactions between chronic conditions and the other independent variables were also examined. For all multivariate modeling stages, we selected variables using Bayesian model averaging (BMA),29 a procedure accounting for the uncertainty in the model choice by selecting all likely models. If the BMA procedure suggested more than 1 model, the models with a posteriori probabilities of 5% or more were investigated. Age, sex, chronic conditions, and number of contacts with a GP were preselected in each model. Because of the complex nature of the CCHS sampling methodology, all the BMA selected models were further analyzed using logistic regression for complex surveys, as implemented in the SUDAAN software; this provides correct usage of bootstrap weights, as described above. Although we examined all models selected by the BMA, we used the minimum Akaike Information Criterion and minimum Bayesian Information Criterion to select 2 final models; one model with and one without interactions. All likely models without interactions exhibited similar parameter estimates and confidence intervals (CIs), and had similar Akaike Information Criterion and Bayesian Information Criterion values. Therefore, we present the most complete model containing all variables considered in the BMA output. When considering models with interactions, as 974 | www.lww-medicalcare.com FIGURE 1. Derivation of study sample. CCHS indicates Canadian Community Health Survey; ED, emergency department; GP, general physician. *Includes overnight stay in hospital, nursing home, or convalescent home. there was little parameter variability between these 2 chosen models, we chose to present the simpler of the 2 while describing interactions of interest. We used similar methods for the sensitivity analyses that included the sample with a history of hospitalization. The secondary analyses of place of last contact with specialists and nurses were adjusted for age and sex only, because of the more restricted sample sizes. RESULTS Figure 1 shows the steps in the selection of the sample of participants with at least one GP contact during the previous 12 months (n ⫽ 33,491), of whom 1677 (5.0%) reported their last contact in an ED. Table 1 shows the results of the initial and multivariate logistic regression models of the primary outcome. In the initial age- sex adjusted models, sociodemographic variables associated with last GP contact in an ED included: age under 35, male, a low level of education, and living in an area with a lower metropolitan influence. Among the health status measures, absence of a chronic condition, deteriorating health, and high alcohol consumption were associated with last GP contact in an ED. Health services variables associated with last GP contact in an ED were nurse contacts in last 12 months, lack of a regular GP, and unmet healthcare needs. Notably, neither the perceived availability of health care in the community nor the total number of contacts with a GP were related to the place of last contact. For simplicity, we have not presented results for the following variables that were not significant in initial models and were not selected for the © 2010 Lippincott Williams & Wilkins Medical Care • Volume 48, Number 11, November 2010 Emergency Department Visits TABLE 1. Odds Ratios and 95% Confidence Intervals for Associations With Last General Physician Contact in an Emergency Department, in the Past 12 Months (n ⫽ 33,491) Initial Models† Variable Sociodemographic variables Age 18–24 25–34 35–44 45–64 65⫹ Sex Female Male No. years living in Canada Born in Canada 5 or less Between 6 and 10 More than 10 Education No secondary diploma Secondary diploma Some post secondary studies Post secondary diploma Area of residence CMA CA or strong CMA/CA influence Moderate CMA/CA influence Weak CMA/CA influence No CMA/CA influence Health status Chronic condition No chronic condition Has an ACSCC Has a chronic condition but not an ACSCC No. chronic conditions No chronic condition 1 or 2 chronic conditions 3 or more chronic conditions Perceived health status Fair or poor Excellent, very good, or good Perceived health changes during the past 12 months Worse or much worse Much better, better, or about the same High alcohol consumption No Yes Health services No. contacts with a GP (past 12 mo) 1 2 or 3 4 or more © 2010 Lippincott Williams & Wilkins Final Model %* OR 95% CI OR 95% CI 10.6 14.7 19.8 37.3 17.6 1.00 0.80 0.46 0.25 0.11 — (0.63, 1.00) (0.35, 0.60) (0.19, 0.31) (0.08, 0.17) 1.00 0.84 0.59 0.40 0.18 — (0.65, 1.08) (0.43, 0.80) (0.29, 0.53) (0.11, 0.29) 54.5 45.5 1.00 1.61 — (1.35, 1.92) 1.00 1.25 — (1.03, 1.50) 88.7 1.7 1.4 8.2 1.00 0.79 0.26 0.83 — (0.43, 1.48) (0.10, 0.73) (0.51, 1.33) 13.4 8.8 5.4 72.4 1.44 0.96 1.36 1.00 (1.03, 2.01) (0.72, 1.29) (0.97, 1.92) — 67.3 17.7 10.8 3.6 0.6 1.00 1.64 1.82 2.88 4.23 — (1.35, 2.00) (1.40, 2.36) (2.20, 3.76) (2.02, 8.86) 1.00 1.94 2.38 3.27 5.42 — (1.56, 2.40) (1.80, 3.16) (2.44, 4.40) (2.24, 13.11) 26.5 31.2 42.3 1.00 0.68 0.81 — (0.54, 0.87) (0.66, 0.99) 1.00 0.72 0.80 — (0.55, 0.93) (0.65, 0.99) 26.9 46.7 26.4 1.00 0.72 0.92 — (0.59, 0.88) (0.71, 1.17) 11.2 88.8 0.94 1.00 (0.68, 1.30) — 11.9 88.1 1.45 1.00 (1.10, 1.90) — 1.39 1.00 (1.02, 1.89) — 97.7 2.3 1.00 1.65 — (1.03, 2.64) 1.00 1.52 — (0.96, 2.38) 38.9 37.8 23.3 1.00 1.18 0.90 — (0.97, 1.45) (0.70, 1.16) 1.00 1.23 0.92 — (0.98, 1.53) (0.68, 1.25) (Continued) www.lww-medicalcare.com | 975 Medical Care • Volume 48, Number 11, November 2010 McCusker et al TABLE 1. (Continued) Initial Models† Variable No. contacts with a specialist (past 12 mo) No contact 1 or 2 3 or more No. contacts with a nurse for care or advice (past 12 mo) No contact 1 2 or more Regular GP No Yes Unmet healthcare needs No Yes Perceived availability of services in the community‡ Poor Excellent, good, fair Final Model %* OR 95% CI OR 95% CI 66.2 25.6 8.2 1.00 0.94 0.89 — (0.74, 1.19) (0.62, 1.28) 84.7 8.1 7.2 1.00 1.38 1.55 — (1.04, 1.83) (1.21, 2.00) 1.00 1.36 1.48 — (0.99, 1.87) (1.11, 1.99) 15.8 84.2 4.54 1.00 (3.70, 5.56) — 4.58 1.00 (3.67, 5.72) — 86.4 13.6 1.00 1.64 — (1.31, 2.06) 1.00 1.38 — (1.06, 1.80) 14.5 85.5 1.33 1.00 (0.88, 1.99) — *Estimated percent of the Québec population, aged 18 and over, with one or more GP contacts in the past 12 months. † Odds ratios are adjusted for age and sex. ‡ Sample restricted to respondents from CCHS cycle 2.1 (2003) only. ACSCC indicates Ambulatory care sensitive condition; CMA, Census metropolitan area; CA, Census agglomeration; GP, General physician; OR, Odds Ratios; CI, Confidence Intervals. final models: marital status, income, activity limitations, dependence in activities of daily living, psychologic distress, and receipt of home care. In the final multivariate model without interactions, variables that continued to be statistically significant predictors of last GP contact in an ED were: age younger than 35, male sex, residence further from a metropolitan area, absence of a chronic condition, worsening health, 2 or more contacts with a nurse, absence of a regular physician, and unmet healthcare needs. The associations with absence of a regular GP physician and with chronic conditions were similar to those in the initial models, whereas that with unmet needs was somewhat attenuated. The best-fitting model with interactions (not shown) included a strong interaction between chronic conditions and sex (P ⬍ 0.01), indicating that the association of male sex with ED contact was restricted to those without a chronic condition (odds ratio 关OR兴: 1.75 关95% CI: 1.26; 2.41兴). In further exploration of this interaction, we found that inclusion of injuries in the model resulted in attenuation of the OR for male sex in the group with no chronic conditions (data not shown). A second strong interaction was between lack of a regular GP and area of residence (P ⬍ 0.01). The ORs for lack of a regular GP for all except the most rural subgroup were strong and statistically significant, whereas the OR for the most rural subgroup was weaker and nonsignificant (OR: 1.87, 关95% CI: 0.40, 8.82兴). A weaker interaction was found with number of nurse contacts (P ⫽ 0.07), indicating that, among those with no chronic conditions, a single nurse contact was associated with last GP contact in an ED (OR: 2.23 关95% CI: 1.37; 3.64兴). The BMA procedure selected an 976 | www.lww-medicalcare.com interesting but nonsignificant interaction (P ⫽ 0.33), suggesting that unmet needs are associated with ED contact only among those with an ACSCC or other chronic conditions (OR: 1.69 关95% CI: 0.96; 2.97兴 and OR: 1.64 关95% CI: 1.17; 2.30兴, respectively, compared with OR: 1.00 关95% CI: 0.58; 1.73兴 among those without any chronic condition). There was no interaction between lack of regular GP and chronic conditions (P ⫽ 0.63); the ORs were similar and strong among participants with and without ACSCC or other chronic conditions. Sensitivity analysis that included hospitalized individuals yielded very similar results to those presented above (data not shown). In the main model, the effects of lack of regular GP (OR: 4.23, 95% CI: 3.43, 5.21) and unmet needs (OR) 1.28 (95% CI: 1.01, 1.63) remained significant. Interestingly, there was an interaction between unmet needs and hospitalization, such that the association with GP last contact in an ED was observed only for individuals without hospitalization, supporting our decision to exclude hospitalized persons in our primary analysis. Table 2 shows the results of the secondary analyses of place of last contact with specialists and with nurses for care/advice. In general, the associations of the sociodemographic and health status variables with these outcomes were similar to those found for the primary outcome, whereas there were notable differences for some of the health service variables. The OR for lack of a regular GP was lower for both secondary outcomes, although it remained significant for last nurse contact. Unmet needs were associated with place of last nurse contact but not last specialist contact. Two or 3 GP contacts were associated with a lower odds of last specialist © 2010 Lippincott Williams & Wilkins Medical Care • Volume 48, Number 11, November 2010 Emergency Department Visits TABLE 2. Odds Ratios* and 95% Confidence Intervals for Associations With Last Specialist and Last Nurse Contact in an Emergency Department, in the Past 12 Months Specialist† (n ⴝ 13,624) Variable Sociodemographic variables Age 18–24 25–34 35–44 45–64 65⫹ Sex Female Male No. years living in Canada Born in Canada 5 or less Between 6 and 10 More than 10 Education No secondary diploma Secondary diploma Some post secondary studies Post secondary diploma Area of residence CMA CA or strong CMA/CA influence Moderate CMA/CA influence Weak CMA/CA influence No CMA/CA influence Health status Chronic condition No chronic condition Has an ACSCC Has a chronic condition but not an ACSCC No. chronic conditions No chronic condition 1 or 2 chronic conditions 3 or more chronic conditions Perceived health status Fair or poor Excellent, very good, or good Perceived health changes during the past 12 months Worse or much worse Much better, better or about the same High alcohol consumption No Yes Health services No. contacts with a GP (past 12 mo) No contact 1 2 or 3 4 or more © 2010 Lippincott Williams & Wilkins Nurse‡ (n ⴝ 6962) OR 95% CI OR 95% CI 1.00 0.94 0.86 0.63 0.80 — (0.61, 1.44) (0.56, 1.32) (0.40, 0.98) (0.52, 1.23) 1.00 0.94 0.54 0.59 0.27 — (0.62, 1.41) (0.36, 0.82) (0.40, 0.87) (0.16, 0.45) 1.00 1.90 — (1.49, 2.42) 1.00 2.32 — (1.75, 3.08) 1.00 1.16 0.77 0.48 — (0.41, 3.30) (0.11, 5.63) (0.24, 0.97) 1.00 0.86 0.44 1.11 — (0.24, 3.14) (0.02, 10.38) (0.43, 2.84) 2.02 1.17 1.20 1.00 (1.21, 3.37) (0.78, 1.75) (0.64, 2.26) — 1.62 1.62 1.53 1.00 (0.90, 2.94) (1.01, 2.59) (0.62, 3.80) — 1.00 1.72 2.12 2.67 2.69 — (1.18, 2.50) (1.48, 3.04) (1.75, 4.05) (1.39, 5.21) 1.00 1.41 2.10 3.48 0.61 — (1.00, 2.00) (1.39, 3.19) (2.16, 5.61) (0.11, 3.26) 1.00 0.81 0.69 — (0.58, 1.14) (0.49, 0.96) 1.00 1.02 0.66 — (0.67, 1.56) (0.45, 0.97) 1.00 0.75 0.66 — (0.54, 1.05) (0.48, 0.93) 1.00 0.78 0.68 — (0.54, 1.14) (0.44, 1.05) 0.86 1.00 (0.60, 1.22) — 0.91 1.00 (0.55, 1.50) — 1.25 1.00 (0.91, 1.72) — 1.43 1.00 (0.95, 2.14) — 1.00 1.44 — (0.68, 3.06) 1.00 0.76 — (0.36, 1.60) 1.00 0.88 0.69 0.95 — (0.64, 1.21) (0.48, 0.99) (0.68, 1.34) 1.00 0.96 0.98 0.89 — (0.63, 1.45) (0.65, 1.49) (0.60, 1.31) (Continued) www.lww-medicalcare.com | 977 Medical Care • Volume 48, Number 11, November 2010 McCusker et al TABLE 2. (Continued) Specialist† (n ⴝ 13,624) Variable No. contacts with a specialist (past 12 mo) No contact 1 or 2 3 or more No. contacts with a nurse for care or advice (past 12 mo) No contact 1 2 or more Regular GP No Yes Unmet healthcare needs No Yes Perceived availability of services in the community§ Poor Excellent, good, fair Nurse‡ (n ⴝ 6962) OR 95% CI OR 95% CI — 1.00 0.82 — — (0.60, 1.11) 1.00 0.88 0.66 — (0.63, 1.24) (0.40, 1.10) 1.00 1.25 1.42 — (0.88, 1.78) (0.96, 2.10) — 1.00 0.71 — — (0.53, 0.95) 1.23 1.00 (0.92, 1.69) — 1.49 1.00 (1.06, 2.08) — 1.00 1.13 — (0.82, 1.56) 1.00 1.67 — (1.17, 2.38) 1.22 1.00 (0.77, 1.93) — 1.54 1.00 (0.73, 3.27) — *Odds ratios are adjusted for age and sex. † Sample consists of respondents with one or more specialist contacts in the last 12 months. ‡ Sample consists of respondents with one or more nurse contacts for care or advice in the last 12 months. § Sample restricted to respondents from CCHS cycle 2.1 (2003) only. ACSCC indicates Ambulatory care sensitive condition; CMA, Census metropolitan area; CA, Census agglomeration; GP, General physician; OR, Odds Ratios; CI, Confidence Intervals. contact in an ED; more than 2 nurse contacts was associated with a lower odds of last nurse contact in an ED. DISCUSSION We pooled data from 2 population health surveys conducted in the province of Québec, Canada, to investigate the relationships between characteristics of primary care and last doctor or nurse contact in an ED, among those with and without ambulatory care sensitive or other chronic conditions. Four features of this study make new contributions to knowledge. First, we focused on ambulatory ED visits (that did not lead to hospitalization), which may be more sensitive to primary care services than are ED visits with hospitalization.1 Second, the sample was not limited to those under 65 or with particular types of health insurance, but was a representative population sample of all adults aged 18 and over. Third, the study was conducted in a population with universal health insurance, which may have implications for the expansion of health insurance coverage in the US. Fourth, unlike previous research, we investigated ED visits in the population with ACSCC, and not ACSCC-related ED visits. Diagnostic information collected at ED visits may not be valid; a more complete picture of ED utilization avoids this bias. Finally, we compared individuals with selected ACSCC with those with other or no chronic conditions. In general, the outcomes were similar in the 2 chronic condition groups, suggesting that ACSCC are not more sensitive than other chronic conditions to limited access to primary care. 978 | www.lww-medicalcare.com We found that 2 measures of inadequate primary care were associated with last GP contact in an ED, even after adjustment for covariates including sociodemographics, health status, and number of GP contacts: lack of a regular GP and perceived unmet needs. Lack of a regular GP was strongly associated with last GP contact in an ED, both among those with and without chronic conditions. Having a regular or primary physician is associated with better perceived access to care9,30 and greater utilization of preventive services.8,31 Primary care physicians may also play a role in reducing unnecessary health service utilization and costs.8,32,33 We found a similar but weaker association between lack of a regular GP and place of last contact with a nurse, but not with place of last contact with a specialist. Patients may be more likely to recall that they saw a specialist in the ED whether they are referred to an ED by a specialist. Perceptions of unmet healthcare needs increased the odds of last GP contact in an ED, but only among those without prior hospitalization. This finding suggests that ambulatory ED visits are more likely to substitute for regular primary care than do those at which the patient is hospitalized.1 After adjustment for covariates, the relationship of unmet needs with place of last GP contact was attenuated and appeared to be limited to subgroups with ACSCC and other chronic conditions (although only the CIs for non-ACSCC chronic conditions exclude unity.) Unmet needs are increasing in Canada. Although multiple factors contribute to perceptions of unmet need, in Québec, this variable appears to © 2010 Lippincott Williams & Wilkins Medical Care • Volume 48, Number 11, November 2010 reflect mainly a perceived lack of accessibility to health care (eg, long waiting times).34 As reported elsewhere, we found important rural-urban differences in use of the ED.7 A more rural residence was associated with an increased likelihood of contacting any health professional (GP, specialist, or nurse) in an ED. Furthermore, the effect of lack of a regular GP was strong in all regions except the most rural. In a previous Québec study, greater continuity of care with a primary physician was associated with fewer ED visits; this relationship was stronger in urban versus rural areas.2 Rural primary care differs from that in urban areas; rural physicians are more likely to practice in multiple locations, including the ED, which becomes an additional site for primary care.7 Overall, these results provide little support for the hypothesis that ambulatory ED visits among adults with ACSCC are more sensitive to characteristics of primary care than are ED visits in those with other or no chronic conditions. Researchers in the United States have also questioned the validity of this hypothesis in relation to hospitalization data.9,19 Differences in our study results from those of prior research20 –22 may be due to differences in methodology (eg, measurement of ACSCC) or the study population (eg, representative population sample, healthcare system differences, inclusion of all age groups). Limitations of this study include those of any crosssectional survey, including the potential for recall bias. Several potential limitations implicit in the survey design should also be noted. First, our primary measure of place of last GP contact was limited to those who reported GP contacts, 62.7% of the survey population aged 18 and over. Although participants may not have recalled correctly the type of health professional they contacted, and probably underreported health professional contacts in an ED, our cross-validation suggests that this measure is valid in general. The relatively greater underreporting of the outcome among those aged 65 and over may be explained by impaired memory as well as higher rates of physician contact. Second, no information was available on a regular site of care; having a regular physician may be less important among those with a site of care.9 Third, no information was collected on the reasons for the ED contact. However, there is no reason to think that the frequency of ED visits for problems unrelated to chronic conditions (eg, trauma) would differ among those with and without chronic conditions. There are several implications of the study for health policy. First, the proportion of GP contacts in the ED may be a useful indicator of the accessibility of primary health care services.10 Use of this indicator should take into account the differences in primary care practice in rural versus urban populations. Second, primary care characteristics associated with whether care is obtained in an ED rather than another site reflect individual characteristics (affiliation with a primary GP and perceived needs) rather than the geographic availability of healthcare. Prior research in Québec seniors found that although increased provider continuity was associated with decreased ED visits, greater availability of physicians in an area (the physician-population ratio) was para© 2010 Lippincott Williams & Wilkins Emergency Department Visits doxically associated with greater ED utilization.2 Therefore, primary care reform should aim to increase affiliation with a regular physician, or at least a regular source of care, if the goal is to reduce the use of the ED as a source of primary care.23 Additional strategies to support this goal include group practice, multidisciplinary teamwork, and mechanisms to coordinate services.35 Notably, the experiences of Kaiser Permanente and the Veterans Administration suggest that improving the integration of primary care with other levels of care make it possible to reduce utilization of hospital and ED services.36,37 ACKNOWLEDGMENTS While the research and analysis are based on data from Statistic Canada, the opinions expressed do not represent the views of Statistics Canda. REFERENCES 1. Roberge D, Larouche D, Pineault R, et al. Hospital emergency departments: substitutes for primary care? Results of a survey among the population of Montréal and Montérégie, 2007. Direction de santé publique, Agence de la santé et des services sociaux de Montréal. 2. Ionescu-Ittu R, McCusker J, Ciampi A, et al. Continuity of primary care and emergency department utilization among elderly people. Can Med Assoc J. 2007;177:1362–1368. 3. Levesque JF, Pineault R, Simard B, et al. L’expérience de soins de la population: Portrait des variations intra-régionales à Montréal et en Montérégie. 2007. Direction de santé publique de Montréal, Institut national de santé publique du Québec, Centre de recherche de I’hõpital Charles-Lemoine. ISBN: 978-2-89494-572-8. 4. McCusker J, Verdon J. Do geriatric interventions reduce emergency department visits?: A systematic review. J Gerontol A Biol Sci Med Sci. 2006;61:53– 62. 5. Gill JM, Mainous AG III, Nsereko M. The effect of continuity of care on emergency department use. Arch Fam Med. 2000;9:333–338. 6. Lowe RA, Localio RA, Schwarz DF, et al. Association between primary care practice characteristics and emergency department use in a medicaid managed care organization. Med Care. 2005;43:792– 800. 7. Haggerty JL, Roberge D, Pineault R, et al. Features of primary healthcare clinics associated with patients’ utilization of emergency rooms: urban-rural differences. Healthc Policy. 2007;3:72– 85. 8. McCusker J, Karp I, Cardin S, et al. Determinants of emergency department visits by older adults: a systematic review. Acad Emerg Med. 2003;10:1362–1370. 9. Lambrew JM, DeFriese GH, Carey TS, et al. The effects of having a regular doctor on access to primary care. Med Care. 1996;34:138 –151. 10. Mustard CA, Kozyrskyj AL, Barer ML, et al. Emergency department use as a component of total ambulatory care: a population perspective. Can Med Assoc J. 1998;158:49 –55. 11. Brown AD, Goldacre MJ, Hicks N, et al. Hospitalization for ambulatory care-sensitive conditions: a method for comparative access and quality studies using routinely collected statistics. Can J Public Health. 2001; 92:155–159. 12. Weingarten SR, Henning JM, Badamgarav E, et al. Interventions used in disease management programmes for patients with chronic illness-which ones work? Meta-analysis of published reports. BMJ. 2002;325:1– 8. 13. Ofman JJ, Badamgarav E, Hennings JM, et al. Does disease management improve clinical and economic outcomes in patients with chronic diseases? A systematic review. Am J Med. 2004;117:182–192. 14. Tsai AC, Morton SC, Mangione CM, et al. A meta-analysis of interventions to improve care for chronic illnesses. Am J Manag Care. 2005;11: 478 – 488. 15. Sanderson C, Dixon J. Conditions for which onset or hospital admission is potentially preventable by timely and effective ambulatory care. J Health Serv Res Policy. 2000;5:222–230. 16. Bindman AB, Grumbach K, Osmond D, et al. Preventable hospitalizations and access to health care. JAMA. 1995;274:305–311. www.lww-medicalcare.com | 979 McCusker et al 17. Bindman AB, Chattopadhyay A, Osmond DH, et al. The impact of medicaid managed care on hospitalizations for ambulatory care sensitive conditions. Health Serv Res. 2005;40:19 –37. 18. Billings J, Anderson GM, Newman LS. Recent findings on preventable hospitalizations. Health Aff. 1996;15:239 –249. 19. Gill JM, Mainous AG III. The role of provider continuity in preventing hospitalizations. Arch Fam Med. 1998;7:352–357. 20. Oster A, Bindman AB. Emergency department visits for ambulatory care sensitive conditions: insights into preventable hospitalizations. Med Care. 2003;41:198 –207. 21. Falik M, Needleman J, Wells BL, et al. Ambulatory care sensitive hospitalizations and emergency visits: experiences of Medicaid patients using federally qualified health centers. Med Care. 2001;39:551–561. 22. Falik M, Needleman J, Herbert R, et al. Comparative effectiveness of health centers as regular source of care: application of sentinel ACSC events as performance measures. J Ambul Care Manage. 2006;29:24 –35. 23. Levesque JF, Benigeri M. L’expérience de soins des personnes présentant les plus grands besoins de santé: le Québec comparé. Résultats de l’enquête internationale sur les politiques de santé du Commonwealth Fund de 2008. 2009. Le Commissaire à la santé et au bien-être. 24. The Canadian Community Health Survey. (Canadian Health Survey Website) Jan 1, 2000. Available at: http://www.statcan.gc.ca/concepts/ health-sante/index-eng.htm. Accessed November 15, 2009. 25. McNiven C, Puderer H, Janes D. Census Metropolitan Area and Census Agglomeration Influence Zones (MIZ): A Description of the Methodology. Geography Working Series No 2000 –2. Ottawa, Canada: Geography Division, Statistics Canada; 2000. Catalogue No 92F0138MPE. 26. Kessler RC, Barker PR, Colpe LJ, et al. Screening for serious mental illness in the general population. Arch Gen Psychiatry. 2003;60:184 –189. 980 | www.lww-medicalcare.com Medical Care • Volume 48, Number 11, November 2010 27. Thomas S, Wannell B. Combining cycles of the Canadian Community Health Survey. Health Rep. 2009;20:53–58. 28. Rust KF, Rao JN. Variance estimation for complex surveys using replication techniques. Stat Methods Med Res. 1996;5:283–310. 29. Hoeting J, Madigan D, Raftery A, et al. Bayesian model averaging: a tutorial. Stat Sci. 1999;382– 417. 30. Stewart AL, Grumbach K, Osmond DH, et al. Primary care and patient perceptions of access to care. J Fam Pract. 1997;44:177–185. 31. Ettner SL. The relationship between continuity of care and the health behaviors of patients: does having a usual physician make a difference? Med Care. 1999;37:547–555. 32. Franks P, Fiscella K. Primary care physicians and specialists as personal physicians. Health care expenditures and mortality experience. J Fam Pract. 1998;47:105–109. 33. Forrest C, Starfield B. The effect of first-contact care with primary care clinicians on ambulatory health care expectations. J Fam Pract. 1996; 43:40 – 48. 34. Sibley LM, Glazier RH. Reasons for self-reported unmet healthcare needs in Canada: a population-based provincial comparison. Healthc Policy. 2009;5:87–104. 35. Castonguay C, Marcotte J, Venne M. En avoir pour notre argent Rapport du groupe de travail sur le financement du systeme de santé. 2008. Gouvernement du Québec. 36. Armstrong B, Levesque O, Perlin JB, et al. Reinventing Veterans Health Administration: focus on primary care. Healthc Q. 2006;9: 80 – 85, 84. 37. Feachem RG, Sekhri NK, White KL. Getting more for their dollar: a comparison of the NHS with California’s Kaiser Permanente. BMJ. 2002;324:135–141. © 2010 Lippincott Williams & Wilkins