Introduction: The distribution of the causes of maternal deaths, and the impact of facility-based obstetric care access (obstetric access) at the individual and population level in India have not yet been quantified. Objectives: (i) estimate the physician agreement for coding of maternal deaths; (ii) explain the distribution of maternal mortality in India; and (iii) quantify the effect of obstetric access on the probability of maternal death, accounting for effect modification of state level skilled attendant coverage. Methods: I used the nationally representative Million Death Survey (MDS) of the causes of death from 2001-3. I identified context-specific risk factors from the field reports to provide information on care patterns before deaths. The 1096 MDS maternal deaths were matched to 147 001 controls of non-fatal deliveries from a representative fertility survey. Findings: 1.Inter-rater reliability was substantial for two physicians assigning a cause of death. 2. Three-quarters of India's maternal deaths were clustered in rural areas of poorer states, although these regions have only half the estimated births in India. The distribution of major causes of maternal deaths (most notably hemorrhage, obstruction, sepsis, abortion) did not differ between poorer and richer states. Two-thirds of maternal deaths died seeking healthcare, most in a critical medical condition. 3. The probability of maternal death decreased with increasing skilled attendant coverage, among both women who had and had not accessed obstetric care. The risk of death among women who had obstetric access was higher (at 50% coverage, OR=2.32) than among those women who did not. However at higher population levels of coverage of safe birth, obstetric access had no effect. This effect appears to be driven partially by reverse causality, in which critical illness is shown to confound the association between care seeking and death. Conclusions: Simple MDS methods enable measurement of the levels, determinants of maternal deaths. Currently, obstetric access in India appears to be an indicator for inequitable access and poor quality even though it is a life-saving intervention. Reduction in maternal mortality in India will require expanded population-based coverage of skilled birth attendance and improved and early access to high quality obstetric care.

Introduction: The distribution of the causes of maternal deaths, and the impact of facility-based obstetric care access (obstetric access) at the individual and population level in India have not yet been quantified. Objectives: (i) estimate the physician agreement for coding of maternal deaths; (ii) explain the distribution of maternal mortality in India; and (iii) quantify the effect of obstetric access on the probability of maternal death, accounting for effect modification of state level skilled attendant coverage. Methods: I used the nationally representative Million Death Survey (MDS) of the causes of death from 2001-3. I identified context-specific risk factors from the field reports to provide information on care patterns before deaths. The 1096 MDS maternal deaths were matched to 147 001 controls of non-fatal deliveries from a representative fertility survey. Findings: 1.Inter-rater reliability was substantial for two physicians assigning a cause of death. 2. Three-quarters of India's maternal deaths were clustered in rural areas of poorer states, although these regions have only half the estimated births in India. The distribution of major causes of maternal deaths (most notably hemorrhage, obstruction, sepsis, abortion) did not differ between poorer and richer states. Two-thirds of maternal deaths died seeking healthcare, most in a critical medical condition. 3. The probability of maternal death decreased with increasing skilled attendant coverage, among both women who had and had not accessed obstetric care. The risk of death among women who had obstetric access was higher (at 50% coverage, OR=2.32) than among those women who did not. However at higher population levels of coverage of safe birth, obstetric access had no effect. This effect appears to be driven partially by reverse causality, in which critical illness is shown to confound the association between care seeking and death. Conclusions: Simple MDS methods enable measurement of the levels, determinants of maternal deaths. Currently, obstetric access in India appears to be an indicator for inequitable access and poor quality even though it is a life-saving intervention. Reduction in maternal mortality in India will require expanded population-based coverage of skilled birth attendance and improved and early access to high quality obstetric care.

Rheumatoid arthritis (RA) is a chronic, destructive, inflammatory arthritis that places significant burden on the individual and society. This thesis represents the most comprehensive effort to date to determine the accuracy of administrative data for detecting RA patients; and describes the development and validation of an administrative data algorithm to establish a province-wide RA database. Beginning with a systematic review to guide the conduct of this research, two independent, multicentre, retrospective chart abstraction studies were performed amongst two random samples of patients from rheumatology and primary care family physician practices, respectively. While a diagnosis by a rheumatologist remains the gold standard for establishing a RA diagnosis, the high prevalence of RA in rheumatology clinics can falsely elevate positive predictive values. It was therefore important we also perform a validation study in a primary care setting where prevalence of RA would more closely approximate that observed in the general population. The algorithm of [1 hospitalization RA code] OR [3 physician RA diagnosis codes (claims) with ≥1 by a specialist in a 2 year period)] demonstrated a high degree of accuracy in terms of minimizing both the number of false positives (moderately good PPV; 78%) and true negatives (high specificity: 100%). Moreover, this algorithm has excellent sensitivity at capturing contemporary RA patients under active rheumatology care (>96%). Application of this algorithm to Ontario health administrative data to establish the Ontario RA administrative Database (ORAD) identified 97,499 Ontarians with RA as of 2010, yielding a cumulative prevalence of (0.9%). Age/sex-standardized RA prevalence has doubled from 473 per 100,000 in 1996 to 784 per 100,000 in 2010, with approximately 50 new cases of RA emerging per 100,000 Ontarians each year. Our findings will inform future population-based research and will serve to improve arthritis surveillance activities across Canada and abroad.

Rheumatoid arthritis (RA) is a chronic, destructive, inflammatory arthritis that places significant burden on the individual and society. This thesis represents the most comprehensive effort to date to determine the accuracy of administrative data for detecting RA patients; and describes the development and validation of an administrative data algorithm to establish a province-wide RA database. Beginning with a systematic review to guide the conduct of this research, two independent, multicentre, retrospective chart abstraction studies were performed amongst two random samples of patients from rheumatology and primary care family physician practices, respectively. While a diagnosis by a rheumatologist remains the gold standard for establishing a RA diagnosis, the high prevalence of RA in rheumatology clinics can falsely elevate positive predictive values. It was therefore important we also perform a validation study in a primary care setting where prevalence of RA would more closely approximate that observed in the general population. The algorithm of [1 hospitalization RA code] OR [3 physician RA diagnosis codes (claims) with ≥1 by a specialist in a 2 year period)] demonstrated a high degree of accuracy in terms of minimizing both the number of false positives (moderately good PPV; 78%) and true negatives (high specificity: 100%). Moreover, this algorithm has excellent sensitivity at capturing contemporary RA patients under active rheumatology care (>96%). Application of this algorithm to Ontario health administrative data to establish the Ontario RA administrative Database (ORAD) identified 97,499 Ontarians with RA as of 2010, yielding a cumulative prevalence of (0.9%). Age/sex-standardized RA prevalence has doubled from 473 per 100,000 in 1996 to 784 per 100,000 in 2010, with approximately 50 new cases of RA emerging per 100,000 Ontarians each year. Our findings will inform future population-based research and will serve to improve arthritis surveillance activities across Canada and abroad.

DKK1, SFRP1, WIF-1, and Wnt5a encode Wnt pathway genes that are frequently silenced by promoter hypermethylation in colorectal cancer. Despite attractive biological consequences of these events, it is unclear whether they contribute to patient prognostication or may influence tumour cell biology within distinct patient subsets. I sought to determine the prognostic roles of these methylation events in a large cohort of colorectal carcinomas from Ontario and Newfoundland. Methylation was quantified and associated with patient clinicopathlogical features. Methylation was present in cancer tissue. DKK1, Wnt5a, and SFRP1 were strongly and independently associated with tumour subtype in a manner that suggested subtype-specific activity of Wnt signaling. Methylation of DKK1 was a borderline prognosticator of favourable outcome. These results offer intriguing insight into subtype-specific biology and lead to a proposed model whereby methylation-induced Wnt bias may contribute to patient outcome.

Objectives: 1) To evaluate the quality of randomized clinical trials (RCTs) in rare diseases using Juvenile Idiopathic Arthritis (JIA) as an example, 2) to evaluate the time to treatment response in patients with rheumatic diseases, 3) to evaluate the power of the Randomized Placebo-Phase Design (RPPD) under various response time distributions, and 4) to examine the use of Value of Information (VOI) methodology in the optimal design of clinical trials for rare disease using hemophilia prophylaxis with factor VIII as an example. Methods. The methods include a systematic review, a secondary analysis of data from an RCT and from a patient registry, a computer simulation study, and an evaluation of hypothetical RCT scenarios with VOI methodology. Results. The quality of RCTs in JIA based on selected quality indicators was poor with some positive changes over time. In the data sets used for the assessment of hazard distributions, the response times followed mostly generalized gamma or lognormal distributions. The impact of time-to-event distribution on the power of RCTs was assessed in computer simulations. Based on the simulation results, the highest sample sizes were observed for response times following the exponential distribution. In most scenarios, the parallel groups RCT design had higher power than the RPPD. The conclusion of the VOI analyses indicated that at threshold values lower than 400,000 the current evidence supported the use of on-demand therapy. Threshold values higher than 1,000,000 supported the use of tailored or alternate day prophylaxis. At threshold values between 400,000 - 1,000,000 the optimal decision varied from on-demand to prophylaxis therapies. Conclusions. New, more powerful and acceptable designs should be developed for rare diseases. When time-to-event outcomes are used, investigators should use various sources of information to evaluate response time distributions before the new trial is designed, and consider this information in sample size calculation and analysis. VOI methodology should be used in the planning stage of studies to determine the relevant costs and benefits of future research, and to determine the optimal trial parameters that maximize the cost-benefit trade-off.

Background: We evaluated geographic patterns for immediate and delayed breast reconstruction (IBR, DBR) in Ontario. The influence of plastic surgeon availability on rates and service provision was determined. Methods: We examined IBR and DBR from 2002 through 2011 across Ontario counties. Regional availability of plastic surgeons was described. Geographic patterns were examined using funnel plots, random-effects models and migration indices. Results: Over ½ Ontario counties have no plastic surgeons. IBR ranged from 0 to 21.5% across counties and differences in plastic surgeon availability explained 41% of variation (p < 0.0001). For DBR there was less variation. 5/45 counties performed ¾ of BR, however rates among local residents were not highest. Interpretation: Nearly 1/3 of the population has limited access to plastic surgeons, contributing to low rates of BR. Geographic access is a major determinant of IBR but is less important for DBR, however service provision for both was highly regionalized.

Objectives: 1) To evaluate the quality of randomized clinical trials (RCTs) in rare diseases using Juvenile Idiopathic Arthritis (JIA) as an example, 2) to evaluate the time to treatment response in patients with rheumatic diseases, 3) to evaluate the power of the Randomized Placebo-Phase Design (RPPD) under various response time distributions, and 4) to examine the use of Value of Information (VOI) methodology in the optimal design of clinical trials for rare disease using hemophilia prophylaxis with factor VIII as an example. Methods. The methods include a systematic review, a secondary analysis of data from an RCT and from a patient registry, a computer simulation study, and an evaluation of hypothetical RCT scenarios with VOI methodology. Results. The quality of RCTs in JIA based on selected quality indicators was poor with some positive changes over time. In the data sets used for the assessment of hazard distributions, the response times followed mostly generalized gamma or lognormal distributions. The impact of time-to-event distribution on the power of RCTs was assessed in computer simulations. Based on the simulation results, the highest sample sizes were observed for response times following the exponential distribution. In most scenarios, the parallel groups RCT design had higher power than the RPPD. The conclusion of the VOI analyses indicated that at threshold values lower than 400,000 the current evidence supported the use of on-demand therapy. Threshold values higher than 1,000,000 supported the use of tailored or alternate day prophylaxis. At threshold values between 400,000 - 1,000,000 the optimal decision varied from on-demand to prophylaxis therapies. Conclusions. New, more powerful and acceptable designs should be developed for rare diseases. When time-to-event outcomes are used, investigators should use various sources of information to evaluate response time distributions before the new trial is designed, and consider this information in sample size calculation and analysis. VOI methodology should be used in the planning stage of studies to determine the relevant costs and benefits of future research, and to determine the optimal trial parameters that maximize the cost-benefit trade-off.

Background: We evaluated geographic patterns for immediate and delayed breast reconstruction (IBR, DBR) in Ontario. The influence of plastic surgeon availability on rates and service provision was determined. Methods: We examined IBR and DBR from 2002 through 2011 across Ontario counties. Regional availability of plastic surgeons was described. Geographic patterns were examined using funnel plots, random-effects models and migration indices. Results: Over ½ Ontario counties have no plastic surgeons. IBR ranged from 0 to 21.5% across counties and differences in plastic surgeon availability explained 41% of variation (p < 0.0001). For DBR there was less variation. 5/45 counties performed ¾ of BR, however rates among local residents were not highest. Interpretation: Nearly 1/3 of the population has limited access to plastic surgeons, contributing to low rates of BR. Geographic access is a major determinant of IBR but is less important for DBR, however service provision for both was highly regionalized.

Introduction: The Hawthorne effect, or behaviour change due to awareness of being observed, is believed to inflate directly observed hand hygiene compliance rates, but evidence is limited. Methods: A real-time location system tracked hospital hand hygiene auditors and recorded alcohol-based hand rub and soap dispenses. Rates of hand hygiene events per dispenser per hour within sight of auditors were compared to dispensers not exposed to auditors. Results: The event rate in dispensers visible to auditors (3.75/dispenser/hour) was significantly higher than unexposed dispensers at the same time (1.48) and in prior weeks (1.07). The rate increased significantly when auditors were present compared to five minutes prior to arrival. There were no significant changes inside patient rooms. Conclusions: Hand hygiene event rates increase in hallways when auditors are visible and the increase occurs after the auditors’ arrival, consistent with the existence of a Hawthorne effect localized to areas where auditors are visible.