Oral presentations

Session 1 – Plenary

(E)¹ Address by Jennifer Madans, Associate Director for Science & Acting Co-Deputy Director, National Center for Health Statistics, USA

Session 2A – Nutrition

(E) Addressing food and nutrition data gaps: Canadian Community Health Survey, Cycle 2.2

Ronald Gravel, Health Statistics Division, Statistics Canada

In addition to the Canadian Community Health Survey (CCHS) 2.2, food and nutrition data are collected by a number of different Statistics Canada programs. Agriculture Division produces the Canada Food Statistics (CFS) Compendium, Income Statistics Division collects food expenditure data in the Food Expenditure Survey, and the Physical Health Measurures Division is currently collecting nutritional bio-markers in the Canadian Health Measures Survey. While all of these programs collect food or nutrition data, they all answer very different research questions. Data from CCHS 2.2 provides researchers with a unique data source that contains many data that have never before been collected by Statistics Canada. For example, the survey collected 24 hour dietary recalls and also direct measures of height and weight. As such, many challenges were encountered during the collection and processing of the data. This presentation will provide an overview of the CCHS 2.2, the challenges encountered, and opportunities for analysis. As well, tools and products available to assist analysts in their research will also be highlighted.

(E) Vitamin and mineral supplement data in CCHS 2.2: analytical challenges and strategies

Stacey Todd, Health Statistics Division, Statistics Canada

Use of vitamin and mineral supplements is widespread and experts maintain that dietary nutritional analyses are incomplete and even misleading if supplement data are not included. A module in CCHS 2.2 (Nutrition) asked about respondents' use of vitamin and mineral supplements during the prior month: which supplements were taken, how frequently they were taken, and how much was taken each time. However, combining supplement data with the food data will require special consideration due to the way in which the data were collected. For example, the reference periods for food and supplement consumption do not match. Furthermore, while there was a second dietary recall for food, this was not the case for supplements. Statistics Canada's position is that the method for combining food and supplement data will vary depending on the type of analysis being conducted. This paper will outline the CCHS 2.2 supplement module, present challenges analysts will encounter when using the data, as well as present various approaches researchers can use to integrate the food and supplement data.

(E) Reconciling food and nutrition statistics from two Statistics Canada data sources

Didier Garriguet, Health Information and Research Division, Statistics Canada

Before the release of CCHS 2.2 (Nutrition), the Canada Food Statistics (CFS) Compendium produced by Agriculture Division was the only source of food consumption data produced by Statistics Canada. Now that dietary recall data from the CCHS 2.2 are available, researchers can now analyze data that are true measures of food consumption in Canada. The main difference between the two data sources is that the CCHS 2.2 is a survey that collects a 24 hour dietary recall. In contrast, the CFS is a microsimulation model that uses data from a number of different data sources as input to the model. In addition, while the CCHS 2.2 is aimed at providing data in a population health context, the CFS is aimed at providing data on the agricultural sector. As such, the two data sources are geared towards different data users. Due to the differences between the data sources, researchers will obtain different estimates if they compare CCHS 2.2 consumption estimates with data from CFS. What accounts for the differences between the two estimates will be demonstrated to data users. As well, the strengths and limitations and appropriate uses of the two data sources will be presented.

Session 2B – Disease and Injury in Canada

(E) Survival analysis of liver transplant patients with HCV infection in Canada 1997-2002

Z. Hong², M. Dawood³, G. Smart³, K Kaita⁴, SW Wen⁵, J. Wu²

Rationale: Liver transplantation is an important health care issue for Canadians. Very few studies have assessed survival and determinants of survivals in liver transplant patients in Canada. Methods: We carried out an epidemiological analysis of 1-year survival and determinants in liver transplant patients, using Canadian Organ Replacement Registry data (1997-2003). Survival curves were plotted by Kaplan-Meier methods. Cox proportional hazard analysis was applied to evaluate the hazard ratios for different ethnicities, blood groups, donor type, pre-transplantation medical status, and HCV infection status. Results: A total of 2058 liver transplant patients were included in the analysis. The main indication for liver transplantation was chronic hepatitis C (23.13%), followed by alcoholic cirrhosis (13.12%). HCV infected patients (n=320) had significantly lower survival probability than HCV negative patients (n=929) with 1-year survival rates of 75% and 85% respectively. In comparison with Caucasians, the Aboriginals had 9% lower survival probability whereas Pacific Islands had 32% higher survival probability. Live organ recipients (n=117) had 73% higher survival probability than cadaveric organ recipients (n=1,945) with in-family receipts having higher survival probability than patients receiving organs from unrelated donors. Pre-transplantation medical statuses had significant influences on the survival rates of the patients with fulminant hepatitis, hospitalized patients at ICU units, and patients at home. Conclusion: Chronic hepatitis C virus infection was a main indication for liver transplantation in Canada. Types of organ donor, genetic similarities between donors and recipients, ethnicity and pre-transplantation medical status of the recipients played important roles for the survival probability status of the recipients.

(E) Approaches to selection of hospitalization records for national injury tabulation

Susan G. Mackenzie, Public Health Agency of Canada

Introduction: Injury hospitalization records classified using the International Classification of Disease contain variables for both injury diagnoses (e.g. fractured wrist) and external causes of injury (e.g. fall). It is thus possible to select injury records in at least two ways. Objectives: 1) To identify approaches used to select records for national injury tabulations, 2) To assess comparability of the sets of injury records retrieved using these approaches. Methods: Approaches were identified by surveying members of the International Collaborative Effort on Injury Statistics. Each approach was applied to a 2002-03 set of hospitalization data from the Public Health Agency of Canada's holding of the Canadian Institute for Health Information's Hospital Morbidity Database. The sets of records retrieved were compared. Results: Six distinct approaches were identified: three were based on diagnoses, three on external causes. The complete set of hospitalization records contained 414,369 records in which either the primary diagnosis was an injury or there was at least one external cause. Application of the six approaches to this set of possible injuries yielded six injury sets that ranged in size from 154,821 to 401,537 records. External cause-based approaches retrieved more records than diagnosis-based approaches. Overall differences and those for selected groups of external causes will be presented and discussed. Conclusions: Approaches used to select injury hospitalization records can have important effects on the sets of records retrieved. In the absence of a standard approach, comparison of injury hospitalization data from different sources should be undertaken with caution.

(E) Population health impact of cancer in Canada, 2001

Presenter: William M. Flanagan, Health Information and Research Division, Statistics Canada. Co-authors: Jane Boswell-Purdy, Hélène Roberge, Christel Le Petit, Kathleen J White, Jean-Marie Berthelot

Summary measures of population health that incorporate morbidity provide a new perspective for health policy and priority setting. Health-adjusted life years (HALYs) lost to a disease combine the impact of years of life lost to premature mortality and morbidity, measured as year-equivalents lost to reduced functioning. HALYs for 25 cancers were estimated from mortality and incidence in 2001 in Canada; population-attributable fractions were estimated for major risk factors contributing to these cancers. Results from this analysis indicate that Canadians would lose an estimated 905,000 health-adjusted years of life to cancer for 2001, including 771,000 to premature mortality and 134,000 to morbidity from incident cases (years discounted at 3%). Most of the estimated premature mortality was due to lung cancer; morbidity was largely due to breast, prostate and colorectal cancers. An estimated one quarter of HALYs lost to cancer were attributable to smoking and almost one quarter were attributable to alcohol consumption, lack of fruit and vegetables, obesity and physical inactivity combined. These results are a significant advance in measuring the population health impact of cancer in Canada because they incorporate morbidity as well as mortality.

(E) Measuring wait times for breast cancer care in Saskatchewan

Paul Cascagnette, Melanie Rathgeber; Health Quality Council, Saskatoon

Background: Improving access to health care by reducing long waits is a priority for Canadians. The goal of this project was to measure overall wait times for breast cancer care in Saskatchewan, from first clinical detection (screening or family physician visit) to the last mode of treatment (surgery or adjuvant therapy). Regional comparisons were also desired. An approach for comparison was needed that would take into account the variability in patient flow pathways based on different start and end events. Methods: Female breast cancer patients with surgeries in 2002-2004 were studied. Since outliers were present in the data, medians were chosen as the measure of overall wait time. Direct standardization of medians based on patient flow pathways was done using the provincial cohort of breast cancer patients as the standard population. Confidence intervals were generated by bootstrapping. Statistical tests for regional comparisons were performed by comparing hazard ratios in Cox regression. Regression weights were chosen to adjust the proportion of patients on the patient flow pathways for each region to be the same as under direct standardization. Weights were further normalized so that the sum of the weights for each region was equal to the population in that region. Results: The provincial median overall wait time was about four months. Differences were found in the distribution of patient flow pathways across regions. Weighted Cox regression showed significant differences in wait time hazards at the regional level.

Session 3A – New Findings from CIHI: Wait Times, Patient Flow and Other Highlights

(E) Understanding emergency department wait times

Greg Webster, Director of Research and Indicator Development, CIHI

Waiting in emergency departments (EDs) is a topic that appears often in the media. Most of the media stories are based on observations at single institutions and, in some cases, on the negative experiences of one person. Given the importance of EDs in the health system, their frequent use and the fact that over 1.5 million Canadians are admitted to acute care hospitals via EDs annually, the Canadian Institute for Health Information (CIHI) analyzed its ED and hospitalization databases and produced a three report series to improve the understanding of ED wait times. The intent of CIHI's three part series of reports on ED wait times was to provide a broad range of new data on ED wait times and to provide additional information about systems issues that are associated with ED wait times. This report series is available on the CIHI website. The first report in the ED Wait Times series, focused on the characteristics of patients visiting selected EDs in Canada and the overall length of time people spent in EDs. The second report, an Analysis-in-Brief, looked more closely at ED wait times in Ontario. Specifically, the report examined variations in overall time spent in EDs by type of hospital and geographic location, wait times to initial physician assessment, and variations by patient triage level and discharge disposition. The third report provides insight into factors that are associated with patient flow and wait times in the ED for admission to an inpatient bed. The Data Users 2007 presentation will include highlights from all three ED reports and provide participants with information on CIHI's ED-related datasources.

(E) Wait time for hip fracture surgery – a new indicator

Zeerak Chaudhary, Senior Analyst for Health Indicators, CIHI

An operative delay in patients with hip fracture is associated with higher risk of postoperative complications and mortality. CIHI's new wait time for hip fracture surgery indicator is intended to provide a comparable measure of timely access to hip fracture surgery across the country. Method: A risk-adjusted proportion was calculated for hip fracture patients aged 65 and older who underwent hip fracture surgery on the day of admission or the next day after being admitted to an acute care hospital. The logistic regression model was fitted with age, sex and selected pre-admission co-morbid conditions as independent variables, modeling the probability of having hip fracture surgery on the same day or the next day. Results: Analysis showed that of the cohort of 17,075 patients who had hip fracture surgery 65.2% had surgery on the day of admission or the next day. There was significant variation across the provinces, ranging from 53% to 78%. Similarly, variations in the wait time for hip fracture surgery were observed across health regions. In our cohort we found that 8% of hip fracture patients were transferred between acute care hospitals at least once. The likelihood of this happening varies across the country. Transfers were more common in Saskatchewan, British Columbia and Manitoba and less so for Nova Scotia, Newfoundland and New Brunswick. Other factors associated with wait time for hip fracture surgery were also analyzed. Summary: To our knowledge, for the first time, wait time for surgery following hip fracture was examined at national, provincial and regional levels and significant variations were observed. The presentation will highlight the results of the analysis and variations in wait time for hip fracture surgery.

(E) Analytical highlights from CIHI's health system databases

Brent Diverty, Director of Continuing and Specialized Care Information Services, CIHI

The Canadian Institute for Health Information (CIHI) presently maintains over 25 health system databases pertaining to health services, expenditures and human resources. The organization regularly turns data from these systems into information to support evidence-informed decisions by system managers, policy makers, researchers and front-line service providers and overall system accountability. Join Brent Diverty (Director, Continuing and Specialized Care Information Services) for an overview, using specific examples, of how these systems are making a contribution to priority areas including measuring access and outcomes, safety and quality improvement, patient pathways, health human resources and system efficiency and sustainability.

Session 3B – Social Determinants of Health

(E) Applying Statistics Canada LifePaths microsimulation model to project the health status of Canadian elderly

Jacques Légaré (Université de Montréal) and Yann Décarie (Université de Montréal and Université de Sherbrooke)

Complex population projections usually use microsimulation models; in Canada, Statistics Canada has developed a global dynamic microsimulation model named LifePaths in the Modgen programming language to be used in policy research. LifePaths provides a platform to build on for our research program, conjointly with Janice Keefe from Mount Saint Vincent University, on projections of the Canadian chronic homecare needs for the elderly up to 2031 and of the human resources required. Beside marital status, family networks and living arrangements, future health status of the elderly is a key variable, but an intricate one. We will use in the present research the current disability module of LifePaths from longitudinal data from the Canadian National Population Health Survey (NPHS). These new health status transitions are considering significant explicative variables like marital status, age of spouse, number of surviving children etc. We will then present projections of future Canadian elderly by health status. Our previous research has shown the importance of future disability level for the management of an elderly society. The main output of the present paper will produce, with new health scenarios, new estimates for Canada of elderly in poor health, for those aged 75 and over, useful especially for implementing new policies for the well-being of the Canadian elderly.

(E) Could social services save more lives? Using administrative data to analyse risk of mortality for older citizens in London Borough of Camden (UK)

Tavakoly, B⁶, Professor R.D. Wiggins⁶, Professor L. Mayhew⁷

This paper investigates the relationship between mortality and a number of factors drawn from existing administrative databases including gender, housing tenure, council tax bands (a proxy of wealth) and three popular causes of hospital admission (falls, strokes and ischemic-heart disease) for Camden residents aged 50 years and older. The study also includes the assessment of information on social service contact in order to identify the potential and/or the effectiveness of service delivery. Existing data sources are merged using a relational database management systems approach. Risks of mortality are examined for different combinations of factors (Risk Ladders) and a Geographical Information System (GIS) application is used to illustrate the findings across the Borough to identify high pockets of risk. The relative importance of risk factors are assessed by logistic regression and the effectiveness of these models are also evaluated by use of Receiver Operating Characteristic (ROC) Curves. The risk of mortality is more likely to occur for people in lower council tax bands (A-C) than higher tax bands (D-H) and for men rather than women. However, the effect of tenure varies for different age groups, gender and tax band. The risk of mortality significantly increases for those groups of individuals who had at least one hospital admission for any of the three causes during 2002-04. Our results show the extent to which contact with social services is aligned with mortality risk among this age group with consequent implications for how social services are organised and delivered.

(E) Measurement of socioeconomic status of immigrants to Canada

Farah N. Mawani, MSc, PhD Candidate, Department of Public Health Sciences, University of Toronto

Statement of Purpose/Problem: It is well established that socioeconomic status (SES) is an important determinant of health. It is particularly important to examine the influence of SES on the health of immigrant populations in Canada because: Canada accepts proportionately more immigrants than any other country; immigrants experience the well-documented ‘healthy immigrant effect'; and there has been a dramatic decline in the economic status of immigrants in Canada over the past twenty years, with a growing incongruence between their education and occupation levels. Methods: This paper will provide an overview of the socioeconomic profile of immigrants to Canada, examine the utility of standard SES measures for immigrants, and propose underemployment as an alternate SES measure that incorporates the potential negative correlation between immigrant education and occupation levels. The paper will then use data from the CCHS 1.1 to compare the association of SES to depression using underemployment (a constructed variable) compared to standard measures (education, occupation, income) of SES. Sample: The sample will be restricted to those aged 18–64 who are employed labor force participants. Analysis: Logistic regression analyses will be performed to estimate odds ratios associating SES with depression. First, analyses will be conducted using underemployment as the primary independent variable. Second, analyses using income, education and occupation as primary independent variables will be conducted. The results of these analyses will be compared to assess whether the association of SES to depression is measure-dependent; i.e. whether underemployment, is an alternate SES measure that is particularly relevant for immigrants.

(E) Measuring health and analyzing data with a gender lens: pointers and pitfalls from a case study on mental health

Mireille Kantiebo, MSc, PhDc, Senior Researcher & Analyst, Bureau of Women's Health and Gender Analysis, Health Policy Branch, Health Canada; Cara Tannenbaum, MD, MSc, Assistant Professor of Medicine, Université de Montréal; Centre de Recherche, Institut Universitaire de Gériatrie de Montréal

Context & Methods: This presentation will provide an overview of the scientific and policy context at Health Canada (HC) that supports the integration of a gender and diversity perspective in the development and use of health indicators for women. It will also highlight the results of one of the projects funded under HC's Health Policy Research Program, which sought to develop and validate new gender-sensitive mental health indicators using data from the Canadian Community Health Survey cycle 1.2 Mental Health Supplement 2002 and the Quebec RAMQ medical services claims database 2003. Three new potential mental health indicators were explored: 1) rates of psychotropic drug use; 2) proportion of physician billings for mental health conditions; and 3) rates of self-reported mental health symptoms. Also were analyzed the social determinants of women's mental health. Results: Higher levels of distress were associated with inadequate social support, lower income levels, unemployment, physical inactivity, lower educational attainment and single parenthood. Except for unemployment, all of these determinants had a greater impact on distress in women than men. Annual self-reported rates of psychotropic drug use were 8% for women compared to 4% for men. The proportion of annual claims made by physicians for mental health visits was 25% for women and 18% for men. These numbers underestimate mental illness as only 87% of adult women and 75% of adult men access a physician per year. Twenty percent of women and 14% of men admitted to feeling depressed or anxious. Discussion: The potential use of these indicators, policy implications as well as a number of data quality issues and limitations related to analyses on diverse populations and minority sub-groups will be discussed.

Session 4A – Introduction to the Canadian Health Measures Survey (CHMS)

To address longstanding limitations within Canada's health information system, Health Canada and the Public Health Agency of Canada have supported Statistics Canada in obtaining funding for a ‘direct measures' health survey. This support was announced in the 2003 federal budget as part of an extension of the Health Information Roadmap Initiative. The Canadian Health Measures Survey (CHMS) will collect key information relevant to the health of Canadians by means of direct physical measurements such as blood pressure, height, weight and physical fitness. As part of the CHMS, a clinical oral health examination will help to evaluate the association of oral health with major health concerns such as diabetes and respiratory and cardiovascular diseases. In addition, the survey will collect blood and urine samples to test for infectious diseases, nutrition and environment markers. Through household questionnaires, the CHMS will gather information related to nutrition, smoking habits, alcohol use, medical history, current health status, sexual behaviour, lifestyle and physical activity, the environment and housing characteristics, as well as demographic and socioeconomic variables. All this valuable information will create national baseline data on the extent of such major health concerns as obesity, hypertension, cardiovascular disease, exposure to infectious diseases, and exposure to environmental contaminants. In addition, the survey will provide clues about illness and the extent to which many diseases may be undiagnosed among Canadians. The CHMS will enable us to determine relationships between disease risk factors and health status, to explore emerging public health issues and to evaluate new measurement technologies.

(E) Overview of the Canadian Health Measures Survey

Marc St-Denis and Mark Tremblay, Physical Health Measures Division, Statistics Canada

This presentation will provide a general overview of the survey, its relevance as well as the challenges faced during its implementation such as the creation of mobile examination clinics for the collection of blood and urine and for the fitness testing. The main measures that are collected will be reviewed and demonstrations of some of the key physical measures will be provided at the end of the session. Findings from other international health measures surveys such as the American National Health and Nutrition Examination Survey will also be highlighted.

(E) Biomonitoring of environmental chemicals in Canadians

Douglas Haines, Environmental Health Surveillance, Risk Management Bureau, Safe Environments Program, Healthy Environments and Consumer Safety Branch, Health Canada

Biomonitoring is a powerful tool that can be used for direct measurement of environmental chemicals, their metabolites or reaction products in people for both surveillance and research purposes. Health Canada, in collaboration with Statistic Canada's Canadian Health Measures Survey (CHMS) will incorporate the first National biomonitoring study to measure human levels of environmental chemicals in Canadians. This biomonitoring component consists of a sample of 5000 Canadians from ages 6-79 with sub-sampling of some measures. Blood and urine specimens will be collected and analysed for the following contaminants: metals, phthalates, PCBs, organochlorine pesticides, brominated flame retardents, organophosphate pesticides, phenoxy herbicides, cotinine, perfluorinated compounds and bisphenol-A. A corresponding questionnaire will be administered to each respondent which will allow for the analysis of risk factors related to exposures to these environmental chemicals. The results of these analyses will establish the current population levels of each of these contaminants including differences by age and sex. This information will provide a baseline for emerging trends and allow comparisons with data from other countries as well as selected targeted Canadian research. The results of this biomonitoring study will provide a unique opportunity to assess the exposure of Canadians to environmental chemicals. The results will allow future research efforts to focus on the link between exposure and health outcomes and provide evidence to help prioritize health interventions and efforts of government regulators.

(E) Analysis and research planning

Sarah Connor Gorber, Health Information and Research Division, Statistics Canada

This section will briefly summarize the analytical planning process for the CHMS and will highlight some of the background research that is underway by the CHMS analytical team. Two main projects will be highlighted: 1) Validity of the CHMS accelerometers: The CHMS will be using 380 Actical accelerometers to assess physical activity in respondents. While the accelerometry function has been tested and found to be both reliable and a valid measurement tool, the pedometry or step count function has not been tested. In the fall of 2006 the CHMS team undertook a validity study to ensure that the Actical devices provided valid results. Thirty eight Statistics Canada employees (and their families) in 3 age groups: <20, 20-49 and 50+ were asked to participate. The study consisted of participants slow walking (3 km/h), fast walking (5 km/h) and jogging (8 km/h) in six minute trials. Each participant was wearing 18 devices: 8 Actical (used in CHMS), 8 Actigraph accelerometers and 2 Yamax pedometers (gold standard). Observers also visually counted steps allowing comparisons between 4 measures of step counts. Validity of the devices proved to be high at the normal walking and running speeds indicating that our devices provide valid measurement of step counts. 2) Systematic reviews comparing reported and measured values: A series of four systematic reviews comparing measured and reported values of physical activity, hypertension, obesity and smoking status were conducted to determine the error associated with self reported measurement of these indicators. Results of the reviews indicate that there is substantial under-reporting of key health indicators that are based on self reported values.

(F) Ethics and privacy issues in collection directly measured data

Renée Langlois, Physical Health Measures Division, Statistics Canada

The CHMS will store biological samples for further analyses of measures at a later date. The CHMS team is working closely with the Health Canada Research Ethics Board and the Office of the Privacy Commissioner of Canada in order to address privacy issues and to implement proper laboratory procedures. This presentation will outline the challenges faced by the CHMS team in regard to privacy while responding to the needs expressed by the Research Ethic Board. When developing this complex survey, the team had to respond to both requirements while developing data collection strategies that most often were new. This section will also outline the proposed strategy to access the stored biospecimens.

Session 4B – Expanding the Application of Data through Linkages

(E) Creating and using CCHS files to link survey responses to health service utilization data – Ontario's experience

Carol Paul, Elsa Ho, Jessie Wong, Judy Corazza, Knowledge Management and Reporting Branch, Ontario Ministry of Health and Long-Term Care, Toronto, Ontario

This session will focus on two aspects of linking: 1) Verifying the Health Numbers (HNs) for those who agreed to link. Steps used in Ontario to link survey respondents to the Registered Persons Database to identify/verify HNs. The importance of accurate personal identifying information such as HNs, last name, given name, date of birth, and sex for accurate linking will be examined. Data quality issues identified during the validation process will be discussed. 2) Linking the verified Ontario linking file to utilization data by HNs to answer policy questions. For example: (i) the relationship between self-rated health status or smoking status and the utilization of in-patient services and same day procedures; (ii) the association between socio-economic factors and health service utilization.

(E) Promise and pitfalls: using electronic medical record data to measure quality of care

Dr. Alan Katz, MMSF Clinical Research Professor in Population Medicine, Associate Professor, Departments of Family Medicine and Community Health Sciences, University of Manitoba; Tom Fogg, Consulting Director, Sierra Systems, Winnipeg, Manitoba,

Manitoba's Physician Integrated Network (PIN) Project is an approach to primary care renewal that utilizes pay-for-performance as a lever to improve patient access, provider quality of work life, access to and use of information, and demonstration of quality care. Essential to this approach is the ability to collect information from clinics' Electronic Medical Records (EMRs). This is used to measure clinics' performance against a subset of the CIHI Primary Care indicators. For the majority of these indicators the CIHI Primary Care Indicator report indicated that there are no current sources of data for the indicator; PIN is of interest in that it will provide early evidence on whether EMRs can effectively fill this gap. PIN is currently in a demonstration phase with four sites and approximately 100 clinicians participating, but important findings are already emerging. One of the criteria for inclusion in the PIN project is the current use of an EMR. This presentation will describe issues identified around EMR usage, data collection and interpretation challenges, that are vital to understand if EMRs are to become a useful source of data for health research, policy-making and program management. This presentation will provide specific examples of issues and associated impacts on quality, time and cost. Finally, we will discuss techniques we adopted to address these issues and reflect on their effectiveness.

(E) Using population data to measure outcomes of joint replacement

Claudia Sanmartin and Kimberlyn McGrail, Health Information and Research Division, Statistics Canada; Mike Dunbar, Dalhousie University; Eric Bohm, University of Manitoba

Background: There is growing consensus on the need to report on the performance of the health care system. Despite this, for most procedures we do not know, at the population level, how healthy patients are following surgery. Purpose: The objective of this work is to quantify patients' outcomes after hip and knee replacement surgery using existing data sources. The hypothesis is that people undergoing these surgical procedures will return to a health status similar to that of their age and sex peers. Methods: Data from the CCHS (1.1) were linked to hospital administrative data to identify joint replacement patients who responded to the CCHS between 6 and 60 months after surgery. We compared the health utility index (HUI) of this group to a comparable age/sex population cohort. Multi-linear regression models were used to identify factors associated with outcomes of care. Results: Overall, surgical patients reported 8% lower HUI than controls, after adjusting for other determinants of health. The differences were mostly in the dimensions of mobility and pain. Outcomes varied by diagnosis, ranging from 4% lower HUI for osteoarthritis patients to 20% lower for other types of arthritis. The presence of chronic disease, lower income, lower education, smoking and obesity were also associated with lower HUI following surgery. Conclusions: People who have undergone joint replacement surgery enjoy lower HUI than others, even after for controlling for other factors known to influence health status. Linking survey and administrative data can provide valuable information regarding outcomes of care at the population level.

(E) A new look at an old story: variations in health care services use and the influence of need

Kimberlyn McGrail and Claudia Sanmartin; Health Information and Research Division, Statistics Canada

Background: Area level variation in the use of health care services is well documented. Explanations include differences in practice patterns, in supply of health care services, and underlying differences in populations by region. Previous analyses have often not included measures of need beyond age and sex. The objective of this work was to exploit survey data linked to administrative data on acute inpatient care to better understand regional variations in hospital use. Methods: CCHS (1.1) data were linked to administrative data on acute hospital use for the two years following the survey. Analytic variables were divided into four blocks: age and sex; need variables; health care services use; and other “non-need” factors. Logistic regression models were used to construct standardized hospitalization rates for each of the 100+ health regions in Canada. Results: In the two years following the CCHS 1.1 survey, there were 77 acute inpatient admissions per thousand population, and a more than five-fold variation in this crude rate by health region. “Need” variables had relatively little influence in explaining this variation. Health care services use variables /increased /variation, and including non-need variables substantially reduced – but did not eliminate -- variation. Discussion and conclusions: The largest impact on standardization comes not from need variables but from adding in variables in the “other”, non-need category. Bringing higher use regions down to the national average would eliminate over 120,000 hospitalizations over the course of two years. Understanding regional variations must remain a concern for researchers and policy-makers.

Session 5 – Plenary

(E) Address by Richard C. Alvarez, President and Chief Executive Officer, Canada Health Infoway

Session 6A – Adding Socio-economic Variables to Health Files based on Administrative Data

(E) Exploiting postal codes and census geography in population-based research: strengths and weaknesses of small-area data

Russell Wilkins, Health Information and Research Division, Statistics Canada

Research using small-area data in surveys and administrative data sets must deal with a variety of potential problems. These include lack of variables at a suitable geographic level, unspecified vintage of geography, lack of or imprecision of documentation, imprecision of coding, misclassification, missing data (geographic and census), and changes over time in geographic coding and definitions. Various tools and techniques for dealing with or overcoming such problems will be discussed, including PCCF+, geographic attributes files, inter-censal EA/DA translations, and merging to small-area census data.

(E) Merging area-level census data with survey and administrative data

Denis Gonthier, Statistics Canada Research Data Centres Program, Montreal

This paper explains how to append census area-level summary data to survey or administrative data. It uses examples from survey datasets present in Statistics Canada Research Data Centres, but the methods also apply to external datasets, including administrative datasets. Four examples illustrate common situations faced by researchers: (1) when the survey (or administrative) and census data both contain the same level of geographic identifiers, coded to the same year standard ("vintage") of census geography (for example, if both have 2001 DA); (2) when the two files contain geographic identifiers of the same vintage, but at different levels of census geography (for example, 1996 EA in the survey, but 1996 CT in the census data); (3) when the two files contain data coded to different vintages of census geography (such as 1996 EA for the survey, but 2001 DA for the census); (4) when the survey data are lacking in geographic identifiers, and those identifiers must first be generated from postal codes present on the file. The examples are shown using SAS syntax, but the principles apply to other programming languages or statistical packages.

(E) Development of a tool to support health researchers in using census data – a Users' Guide to Census Data for Health Researchers

Dawn Mooney, Sandra Peterson, Chris Balma, Kerry Kerluke, Rachael McKendry, Charlyn Black; UBC Centre for Health Services and Policy Research, Vancouver

The Users' Guide to Census Data for Health Researchers gives researchers guidance in understanding and using census data for health research. Informed by our own experiences using census data and funded by the Canadian Institutes for Health Research (CIHR), the Users' Guide is designed to help others avoid some of the pitfalls we discovered along the way. The Users' Guide consists of a web-based series of linked questions and answers, hosted on the UBC Centre for Health Services and Policy Research (CHSPR) web site. To create the Users' Guide we identified important questions by reviewing census data and documentation products from several sources, including Statistics Canada, BC Stats and the UBC Data Library. We also consulted with experienced and potential users of census data. Answers were drafted in consultation with individuals at BC Stats and Statistics Canada, and the results were beta tested with junior and senior health researchers. The Users' Guide addresses complex questions around access to and cost of census data for different audiences, the parallel universes of census and health geographies (especially in BC), small area data quality, the potential for linking census data to other data sets, and strategies for accessing and understanding existing documentation. Throughout, we focus on bridging the gap between health and census geographies and data sets. Health researchers wishing to leverage the considerable socioeconomic and demographic information collected in the census now have a specially targeted starting point.

(E) Childhood leukemia and socioeconomic status in Canada

Marilyn J. Borugian, BC Cancer Research Centre, Vancouver

Leukemia is one of the most common potentially fatal illnesses in children, and its causes are not well understood. Although socioeconomic status (SES) has been related to leukemia in some studies, this apparent association may be attributable to ascertainment or participation bias. This study was undertaken to determine whether there is a difference in incidence of childhood leukemia for different levels of SES, as measured by neighborhood income, in an unselected population case group. All cases of childhood leukemia diagnosed in the years 1985–2001 were identified from population-based cancer registries in Canada. Postal codes for the place of residence at diagnosis were used to ascertain the census neighborhoods for cases. We constructed neighborhood-based income quintiles from census population data, and stratified the population at risk by sex and 5-year age groupings. Age-standardized incidence rates and 95% confidence intervals (CIs) were calculated. We used Poisson regression to compare incidence rate ratios (RRs) across income quintiles. A slightly lower relative risk of childhood leukemia was observed in the poorest quintile compared with the richest (RR _ 0.87; 95% CI _ 0.80–0.95). The lower risk in the poorest quintile was restricted to acute lymphoid leukemia (0.86; 0.78–0.95) and was strengthened slightly by restriction to urban areas (0.83; 0.74– 0.93). This analysis suggests that high SES is a true risk factor for childhood leukemia and that inconsistent results from other studies may be related to differences in case ascertainment or study participation.

Session 6B – National Surveys: Methodological Issues and Analytical Techniques

(E) Combining cycles of the Canadian Community Health Survey

Steven Thomas, Household Survey Methods Division, Statistics Canada

The Canadian Community Health Survey consists of two cross-sectional surveys conducted on an alternating annual cycle. The first survey collects information from over 130,000 respondents for production of general health estimates at the health region level. The second survey collects information from over 35,000 respondents for production of survey specific health estimates at the provincial level. Even with such large sample sizes, the surveys do not meet all the needs of users. Given that similar information is collected in each cycle, users are interested in combining the cycles of the CCHS to get the sample size required to improve the quality of the estimates, create estimates for small geographical domains, or to estimate for rare populations. Methods exist for combining surveys but the combining of CCHS cycles or any series of cross-sectional surveys can not be performed blindly. The researcher must have a clear understanding of the goal in combining cycles of the CCHS. This includes a clear understanding of the population being covered and the parameter to be measured. The clarification of the analysis will dictate the most appropriate method for combining. This paper will focus on the issues related to combining cycles of the CCHS including real examples from CCHS users. Possible methods to combine cycles will be outlined and guidelines will be proposed.

(E) Uses of longitudinal health data from the National Population Health Survey

Margot Shields, Health Information and Research Division, Statistics Canada

The National Population Health Survey (NPHS) provides unique information on the health of Canadians. Longitudinal data from the NPHS reveal the dynamics of health-related behaviours, health-illness transitions and the changes in health in relation to stressful exposures, for example. The NPHS yields information about the health of the Canadian population that cannot be obtained from other surveys or research. For example, analyses showing changes over time within individuals, such as smoking initiation or transition from acceptable weight to overweight, provide insights into the characteristics and conditions that give rise to such changes. Findings from the NPHS are thus a rich source of policy-relevant information. This presentation will showcase examples illustrating significant results based on analysis of the NPHS longitudinal data. The focus will be on examples where longitudinal data are essential to answer specific research questions. The presentation will also feature some of the analytical techniques that are currently being used for studies bases on NPHS data. The disadvantages and challenges associated with using NPHS longitudinal data will also be presented. It takes decades for longitudinal surveys such as the NPHS to reach their full analytical potential; interest in analysing NPHS data will likely increase in the years to come.

(E) Improvement of the income adequacy and body measurement variables for Canadian Community Health Survey (CCHS)

Xiaochuan (Sherry) Wang, Health Policy Research Division, Applied Research and Analysis Directorate, Health Policy Branch, Health Canada

Background: In the 2001-2002 and 2003 Canadian Community Health Survey (CCHS), the three derived variables on income adequacy were based on categorized household income and household size. Since respondents with household income over $80,000 were all put into one income category, the population distribution was disproportionately concentrated on the high end of income groups. The general surveys of CCHS provide self-reported height and weight, which other research has found to be over and under reported, respectively. Consequently, Body Mass Indexes (BMI) derived from these self-reported data are not very accurate. The nutrition survey (CCHS 2.2) provides data on both measured and self-reported height and weight making it possible to correct the self-reporting bias. This paper examines the accuracy of these derived variables, and provides analysis and suggestions to improve the measurement. Method: We used the equivalent scale (ES) method, which adjusts reported continuous income by household size, to redefine household income adequacy. For those who did not report their continuous household income, we imputed it by considering their personal and family characteristic such as age, marital status, education, occupation, and working hours. Using CCHSv 2.2, we compared self-reported height/weight against the measured height/weight, quantified the systematic reporting bias by estimating the relationship between self-reported height/weight and measured counterparts, and corrected the reporting error on prevalence rate of obesity in CCHS 1.1 and 2.1. Results: Analysis using current income adequacy variables resulted in almost 70% of population belonging to high-middle or highest income groups. Using the ES method, we are able to divide respondents into groups with approximately equal representative population in each, and thus facilitate better understanding of income as a determinant of health outcomes. After correcting the report bias on self-reported height and weight, the estimated prevalence rates of obesity are significantly higher for both men and women than those previously reported from CCHS 1.1 and 2.1.

(F) Measure of the underestimation of the energy intake in the nutrition component of the Canadian Community Health Survey (Cycle 2.2)

Didier Garriguet, Health Information and Research Division, Statistics Canada

Statistics Canada publishes two estimates of the intake of energy by the population. The first is based on a model that uses the disappearance of food and the second is based on the nutrition component of the Canadian Community Health Survey (CCHS), cycle 2.2 that relies on the 24 hour dietary recall. These estimates differ for a number of reasons, notably because they do not measure the same thing. Each estimate has its own limitations. The 24 hour dietary recalls are associated with an under estimation of the intake of energy even though tools have been developed to minimize this under estimation. Nor is this under estimation uniform across the population, being greater in certain subgroups. By using an estimate of energy requirements based on age, sex, height, weight and level of physical activity, and comparing it to the reported intake of energy, this article estimates the under estimation of energy intake for the CCHS 2.2 to explain part of the difference observed with the estimates from food statistics. It also informs all users of the CCHS 2.2 data about this potential bias, measures that bias and identifies the groups within the population that are most at risk of under estimating their energy intake.

Session 7A – Health Outcomes

(E) A framework for health outcomes research: a diabetes and depression case study

Greg Webster, Canadian Institute for Health Information; and Kimberlyn McGrail, Statistics Canada

Objective: To highlight the findings of the work related to the CIHI/STC joint work related to Health Outcomes. The presentation will provide an overview of conceptual framework for studying health outcomes, and present results of health outcomes analyses for persons with diabetes and depression in Canada. Presentation Overview: Conceptualizing Health Outcomes Research – definition, framework for conceptualizing health outcomes research, data and methods (with a focus on linked DAD/CCHS). Results of the Health Outcomes Analyses for Diabetes and Depression based on the framework: Canadian Organ Replacement Register (CORR) analysis which illustrates an examination of End Stage Renal Disease (ESRD) patients and treatments at the treatment inception and all consecutive, follow up treatment and outcome information from 1994 to 2005. Results illustrate that with aging of the general population leads to increase of number of ESRD patients especially with type 2 diabetes. Diabetes is a very strong adverse factor for ESRD patient survival regardless of treatment they had been receiving in the past: 1) A linked analysis that examines hospitalizations following responses to the CCHS among a diabetic cohort. Results illustrate that health system factors such as lack of a regular medical doctor and/or did not see a GP in the past year and patient factors such as leisure inactivity and smoking influence the likelihood for hospitalization; and 2) A linked analysis that examine hospitalizations following responses to the CCHS among a depression cohort. Results illustrate that health system factors such as contact with a health service provider and patient factors such as lack of sense of belonging and prior hospitalization influenced the likelihood for hospitalization. Concluding discussion around use of the framework for future research including data needs.

(E) Health outcome indicators for health regions

Yana Gurevich, Canadian Institute for Health Information

Outcome indicators are useful for performance improvement and monitoring quality of care at both the health region and provincial and territorial levels. This presentation will provide an overview of six outcome indicators: 30-day in-hospital mortality rates for heart attack (i.e. acute myocardial infarction) and stroke, and readmission rates for heart attack, asthma, hysterectomy and prostatectomy. Variations across health regions, provinces and territories will be highlighted. Examining variations in outcomes indicators is one way to identify potential areas for improvement.

(E) Revascularization and survival trends of Canadian acute myocardial infarction (AMI) patients, 1995/96 to 2003/04

Helen Johansen, Health Information and Research Division, Statistics Canada

Revascularization can be a key treatment for improving outcomes in heart attack/AMI patients. Trends in revascularization.and thirty day survival of AMI patients across nine Canadian provinces were investigated using Statistics Canada's linked hospital discharge database (HPOI). Revascularization rates within thirty-days of their attack increased in all provinces between 1995/96 to 2003/04. Male rates increased from 18% to 44%; female from 12% to 28%. Increases were more pronounced in the Western provinces and Quebec. Although coronary artery bypass graft surgery (CABG) rates increased, the revascularization increase was mainly due to percutaneous coronary intervention (PTCA) rates. At the same time, the average Charlson Index increased slightly, the length of hospital stay decreased and the time to the procedure decreased markedly. Overall, the percentage who survived thirty days increased, but the relationship between 30-day survival and revascularization showed much variability between health regions. The variation in cardiac procedures rates after AMI may represent potential inequalities in the treatment of AMI.

Session 7B – New Data Collection, Surveillance and Analysis Tools

(E) Tracking developmental changes in the smoking behaviours of Canadian youth: NLSCY data availability and gaps

Stefania Maggi and Kristy Callaghan, Centre for Early Education and Development Studies (CEEDS), Thompson Rivers University (TRU)

Thanks to the National Longitudinal Survey of Children and Youth (NLSCY) developmental changes in the smoking behaviours of Canadian youth can be tracked over time through critical periods of development from childhood to adulthood. Current knowledge of acquisition of smoking behaviours relies heavily on findings from cross-sectional studies and a limited number of longitudinal studies. It is of paramount importance to study change over time using proper designs and a representative sample. The NLSCY offers invaluable data at the individual level and information about a range of possible factors that influence smoking acquisition either positively or negatively. With this presentation we propose to discuss data availability in the NLSCY for the study of smoking behaviours; the type of analytical techniques that can be used to analyse the NLSCY longitudinal data; the challenges faced by applied researchers in analysing NLSCY data related to smoking behaviours; and gaps in the data that could enrich what is currently in the NSLCY. Specifically, we will present results from trajectory analysis obtained using growth mixture modelling of the changes in smoking behaviours of NLSCY participants spanning 10 years of development (from 10-11to 20-21 years of age). Challenges in conducting the analysis will be discussed such as the large drop in NLSCY participation throughout the study and the change in specific measures of smoking from one survey cycle to the next. In addition, theoretical and policy implications of the findings generated using the NLSCY to identify trajectories of smoking behaviours and their predictors will be discussed.

(E) What to expect from the Maternity Experiences Survey

Kristina Shimmons and Elizabeth Majewski, Special Surveys Division, Statistics Canada

The results of this first in Canada, national survey of maternity experiences will be available soon after the conference. It was carried out by Statistics Canada on behalf of the Public Health Agency of Canada in the fall of 2006. Close to 6,500 mothers of babies between five and ten months of age from the provinces and territories were interviewed about their pregnancy, birth and post-partum experiences. A comprehensive picture of the health care these women received, the practices related to giving birth, and information sources most popular during this time of women's lives emerges from the survey. The physical health of the mother, her reproductive history, stress levels before baby's birth, post-partum depression and basic socio-economic characteristics are part of this picture. The special features of the survey such the use of 2006 Census as a sample frame, overlap with other post-censal surveys in the territories, and sensitive content, together with information on data quality and analytical potential of the data set will be of interest to health analysts and survey takers.

(E) Comparability of health surveillance statistics in Europe: what can be learnt from their systems?

Tek-Ang Lim and Daniel Eilstein; Institut de Veille Sanitaire (InVS), 12 rue du Val d'Osne, 94415 Saint-Maurice Cedex, France

Objective: To review and assess the availability and quality of existing health statistics (HS) across the EU for the purpose of enabling integrated assessment of environmental health risks. This work will also enable to highlight the gaps in HS when comparing the data across the countries and propose challenging methods to cope with intercomparison difficulties. Methods: Three main kinds of datasets are reviewed and assessed: first, “official” data second, survey type data and last socio-economic data that are used in order to proxy the lack of indicator. Official datasets were reviewed and evaluated for six countries in the EU, whereas other types of datasets were carried at the EU level. Results: There are very rich varieties of HS available hence enabling to help health policy makers with providing them with relevant information. When working with the data across the countries three main sources of problems were identified: first, different disease coding standards are in used; second, HS are drawn from purpose designed datasets, hence when the statistics are used for other purposes they may not be the most relevant. Third, for policy decisions, HS are combined with other type of datasets (environmental or socio-economical data), as such datasets interoperability difficulties arise. Conclusion: This experience led in the EU might bring valuable information to providers and users in the HS field. Indeed, HS are gathered by many different national organizations with their own know-how and practices. It would be interesting to analyse how they overcome the challenges of combining the different health datasets and intercompare the data sources and methods.

(E) Evaluation of an online survey for hospital data collection

Jeanie Lacroix, Lisa Milgram, Joanne Tse; Canadian Institute for Health Information

Background: Using experiences from the 2007 Hospital Report System Integration and Change survey, we evaluate the benefits and limitations of using online technology to administer an indicator-based survey spanning various hospital sectors and compare results to the paper-based predecessor. Benefits of developing an online survey include improving process efficiency, data quality, data timeliness, and privacy along with reducing long-term administrative costs. Methods: A multi-sector survey consisting of 102 questions was designed to measure traditional areas of performance among hospitals. After the questions were entered into the survey tool, validation, skip logic, and workflow design were developed in the software. A pilot was conducted with eleven participating hospitals to receive feedback on question format and the online tool. The final survey was sent out to 122 Ontario hospitals via email. The participation rate was 96%. Participant satisfaction, ease of use, and data quality were assessed by various qualitative and quantitative feedback methods. Results: Compared to previous years' manual data entry process, the online tool eliminated the need to create a MS Access database for data entry and validation, hire and train staff for a six-week data entry period, and perform significant manual quality checks and follow-up calls to hospitals. The tool was also effective in reducing administrative costs such as mailing and printing. Conclusions: 81% of respondents were satisfied with the tool. Some of the reported challenges included a steep learning curve, technological uptake, process change, and IT compliance.

Session 8A – Health Services Utilization and Expenditure Research

(E) Estimating the public share for dental health care expenditures in Canada

Carlos Quiñonez, DMD, MSc⁸, Peter Cooney, BDS, DDPH, MSc, FAGD, FRCD(C)⁹, David Locker, BDS, PhD, FAHS, DSc⁸

In 2005, the Community Dental Health Services Research Unit, University of Toronto (CDHSRU), and the Office of the Chief Dental Officer, Health Canada (OCDO), conducted an environmental scan of publicly financed dental care across Canada. This work estimated the public share for dental care at 5.9% ($579.2M) of total dental expenditures, a difference of over one percentage point from the 4.7% ($459.9M) estimated by the Canadian Institute of Health Information (CIHI). Why this difference? This presentation will attempt to locate where this difference exists, and will discuss the challenges of tracking publicly financed dental health expenditures in Canada. It will demonstrate that CDHSRU/OCDO and CIHI efforts likely underestimate the public share due to the complexity of how Canadian governments finance public dental care. In short, it is argued that more refined data gathering mechanisms are needed to fully capture public expenditures, and that this is important for decision makers in an increasing environment of potential dental health care reform.

(E) Health care use at the end-of-life in western Canada

Anne McFarlane, Executive Director, Western Canada, CIHI; Francis Lau, Associate Professor, School of Health Information Science, University of Victoria; Verena Menec, Director, Centre on Aging, University of Manitoba; Bernie Paillé, Special Projects Lead, Canadian Institute for Health Information (CIHI); Jennifer Walker, Special Projects Lead, CIHI; Maria Hewitt, Special Projects Lead, CIHI

Purpose: Health care at the end-of-life is frequently intense and complex, with a substantial impact on the health care system, patients and their families. However, relatively little is known in Canada about the types of services provided at the end-of-life. The present study is a collaborative effort of CIHI and the ministries or departments of health from B.C., Alberta, Saskatchewan and Manitoba to examine the continuum of care at the end-of-life in western Canada. Methods: This retrospective study is based a cohort of individuals who died in 2003-04 in the four provinces, as identified from vital statistics data. Analyses consisted of two components. In the “trunk”, location of death, hospitalizations, palliative drug program registrations, and seniors' prescription drug use, derived from administrative databases, were compared across the four provinces. In the “branch” components, specific analyses were conducted for each of the four provinces on health care services such as home nursing care that, based on different provincial delivery mechanisms, were not interprovincially comparable. Findings: The majority of decedents died in an acute care hospital; however, location of death varied across provinces and regions. Hospital use varied across province and region, and by specific disease groupings, with decedents from terminal illnesses showing much higher use closer to death compared to those with organ failure or frail elderly individuals. This presentation will provide an overview of select findings from the study, as well as a discussion of the challenges faced in analyzing the full continuum of care with limited interprovincially comparable data.

(E) Obesity is associated with increased health care services utilization in Newfoundland and Labrador

L Twells, MSc, PhD(c)¹⁰, J Knight, MSc, PhD(c)^{10, 11}, R Alaghebandan, MD¹¹, D MacDonald, MSc, PhD(c) ^{10, 11}, T Bridger, MD, FRCPC¹¹ and Y Chen, MD, PhD¹².

Purpose: To examine the association of BMI level with health care service utilization (HSU) and costs in the province of Newfoundland and Labrador. Methods: In a secondary analysis of the provincial component of the Canadian Community Health Survey Cycle 1.1 (2000/01), survey records for respondents aged 20-64 (n=2345) were linked to 5 years of administrative physician and hospitalization data in order to obtain objective longitudinal HSU. Regression models were used to examine whether BMI level was an independent predictor of family physician (FP) visits, hospitalization and costs. Results: Of the study sample, 37%, 39%, 17% and 6% were classified as normal, overweight, obese, and morbidly obese respectively. The obese (including morbidly obese) were more likely to report having serious chronic conditions after adjusting for age and gender. Analyses identified the morbidly obese group (BMI ≥ 35kg/m2) as having a significantly higher number of FP visits over a 5-year period when compared with the normal weight group (BMI 18.5-24.9kg/m2), [median 22.0 vs. 17.0, p<.05], and as having significantly higher average costs of FP (p<.001) and specialist services (p<.05). Being morbidly obese remained a significant independent predictor of FP visits (p < .001) and total physician costs (p < .01), after adjusting for number of chronic conditions and other covariates, but was not a predictor hospitalization. Conclusion: Morbid obesity is independently associated with increased family physician utilization and total physician costs but not with hospital utilization. Future health promotion/education and weight loss interventions should be targeted at this high risk group.

Session 8B – A Growing Health Problem: Diabetes

(E) Patterns of health care utilization among individuals with diabetes

Farrah Prebtani, MPA, Michael Hillmer, MSc, Kira Leeb, MA, Diane Watson, PhD, MBA, Health Council of Canada

Background: Diabetes represents a significant health and economic concern. The purpose of this work was to estimate health care utilization among Canadians with diabetes and determine the extent to which co-morbidity was a key determinant of population patterns of use. Methods: The Canadian Community Health Survey, 2005 was used to calculate and compare health care utilization among Canadians, adults with diabetes and adults with diabetes and other chronic conditions. Results: Compared to Canadians overall, individuals with diabetes were twice as likely to stay overnight in a hospital/nursing home (7.6% v. 15.4%, respectively), three times more likely to use home care (2.5% v. 8.6%, respectively), twice as likely to consult a GP (3.1 v. 5.7, respectively), three times more likely to consult with other doctors (0.9 v. 1.5, respectively) and three times more likely to consult a nurse (0.7 v. 2.1, respectively). Adults with diabetes and three or more chronic conditions are three times more likely to use hospital/nursing homes and five times more likely to use home care than adults who have diabetes but no other chronic condition. They also visit GPs and nurses twice as often and other doctors three times as often. Conclusion: Individuals with diabetes used more health care than the Canadian average and co-morbidity is an important driver of health care utilization. Health promotion and disease prevention are important to limiting co-morbidity and chronic illness care should target populations with multiple chronic conditions to reduce the impact of morbidity on health and our health care system.

(E) Using laboratory reporting of A1C results for diabetes surveillance

Cynthia Driver¹³, Hadi Makki¹⁴, Megan Saynisch¹⁴, Diana Berger¹³, Lynn Silver¹³

Background: Effective January 15, 2006, laboratories reporting test results electronically to the New York City Department of Health and Mental Hygiene (DOHMH) were required to report hemoglobin A1C test results to the New York City A1C Registry (NYCAR). The purpose of the NYCAR is to establish public health surveillance of glycemic control in New York City and develop provider and patient interventions to improve diabetes outcomes. Methods: To assess implementation, we tracked the cumulative percent of individuals in the A1C registry by month and compared reported with expected volume among eligible laboratories. Eligible laboratories were those reporting results to the DOHMH electronically and performing A1C tests. Of the 105 laboratories reporting to the DOHMH via the electronic clinical laboratory reporting system (ECLRS), 56 were reporting via the file upload method and 38 performed A1C testing. Results: As of December 31, 2006, 28 of 38 eligible laboratories were reporting A1C test results to DOHMH representing 84% of anticipated volume. The remaining laboratories are working with DOHMH to enable reporting. NYCAR received 884,516 tests for 635,475 individuals. Data were received for 50% of individuals in the registry within 5 months after the mandate took effect, and 75% within 8 months. The median monthly volume of A1C tests was 76,321 (range 34,413 to 99,539). Conclusion: A high level of laboratory reporting was achieved in a short timeframe by building on pre-existing ECLRS and eliminating paper reporting. Electronic laboratory reporting is a feasible source of surveillance data for chronic conditions typically monitored by laboratory analyses.

(E) Stress, diabetes, and northern living

Shooshtari S., University of Manitoba; Young T.K., University of Toronto; Mayer T. , Health Child Manitoba; Buzdugan R., University of Manitoba

Introduction: Type 2 Diabetes is an increasingly important public health concern. Recently, stress has increasingly been linked to this disease. However, there has not been any evidence from large-scale population-based epidemiological studies that psychological stress does contribute to the onset of diabetes. Thus, the main goal of this study was to investigate the association between psychological stress and diabetes, with particular emphasis on place of residence, southern versus northern health regions. Methods: The data were obtained from the cross-sectional 2002/03 Canadian Community Heath Survey [CCHS, Cycle 2.1] and the first 5 cycles [1994/95 – 2002/03] of the National Population Health Survey (NPHS). The two dependent variables were diabetes prevalence (from the CCHS) and incidence (from the NPHS). The potential explanatory variables were age, sex, family history of diabetes, place of residence, level of education, household income level, marital status, body mass index, energy expenditure, drinking behaviour, smoking, and stress. The data were analyzed using multiple logistic regressions. The bootstrap re-sampling technique was applied to fully account for the survey design effect. Results at the 0.05 level were considered to be statistically significant. Results: High level of psychological stress was associated with the development of diabetes. There was no significant interaction effect between stress and place of residence. A number of other known diabetes risk factors were confirmed. Conclusions: Our results suggest “stress management” as one of the strategies to prevent diabetes.

Notes

1. The capital letter in front of the title indicates the language of the presentation: E = English, F = French
2. Blood-borne Pathogens Unit, Blood Safety Surveillance and Health Care Acquired Infection Division, Centre for Infectious Disease Prevention and Control, Public Health Agency of Canada
3. Cadham Provincial Laboratory, Manitoba Health, Winnipeg, Manitoba
4. Viral Hepatitis Investigative Unit, University of Manitoba, John Buhler Research Centre, Winnipeg, Manitoba
5. OMNI Research Group, the Clinical Epidemiology Program, Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, Ontario
6. Department of Sociology, School of Social Science, City University, Northampton Square, London, EC1V 0HB
7. CASS Business School, City University, 106 Bunhill Row, London, EC1Y 8TZ
8. Community Dental Health Services Research Unit, Faculty of Dentistry, University of Toronto
9. Office of the Chief Dental Officer, Health Canada
10. Faculty of Medicine, Memorial University of Newfoundland
11. Research and Evaluation Division, Newfoundland and Labrador Centre for Health Information
12. Faculty of Medicine, University of Ottawa
13. New York City Department of Health and Mental Hygiene: Bureau of Chronic Disease Prevention and Control
14. New York City Department of Health and Mental Hygiene: Bureau of Informatics and Information Technology