Using Analytics to Diagnose Ontario’s Health Care Delivery System

By Janet E. Hux, MD

When it comes to the role that ICES plays in Ontario’s healthcare system, we could be likened to a CT scanner. We perform ongoing studies, analyze data, and scan the system to determine how well it is working for Ontario’s 12 million residents in terms of quality, access, equity, and effectiveness.

With more than 100 projects on the go at any one time, one of our biggest challenges at ICES is to analyze massive volumes of data that range from information at the individual patient and neighbourhood level, to provincial and national census data. Working within a highly secure environment, ICES holds anonymous copies of data on a wide range of transactions within the healthcare system, from physician encounters and hospitalizations to medication prescriptions.

Analyzing all of this information to determine trends and patterns is an important part of our role in generating knowledge to support health policy. Some specific examples of how we serve the public and policy makers include conducting research projects to uncover side effects of medications that might not be revealed in clinical drug trials – which often include only a small sampling of relatively healthy patients – and finding new links between non-traditional disease factors and variables that past studies might not have considered. From emergency room wait times to rising rates of diabetes, ICES scientists and medical staff are conducting research into some of this province’s most important medical challenges.

Analytical software such as the products we obtain from SAS have played a key role in helping ICES sift through millions of data points and records in order to uncover previously hidden trends, some of which can have significant health implications for patients. SAS software also acts as the common language that we use to integrate and analyze data from different sources in order to find new cause and effect relationships. For example, we have filtered driving record data from the Ontario Ministry of Transportation with case reports from family physicians in order to draw correlations that have pointed to specific groups of medical patients as potentially unsafe drivers.

Diabetes mapping
In 2007, the ICES team published the findings of an extensive, Toronto-wide study that looked at how certain characteristics of neighbourhoods – including proximity to public services, availability of grocery stores, crime levels, immigration, and education levels – could be contributing to "diabetogenic" environments, and leaving some portions of the population at higher risk for developing diabetes than others. Examining the nature of the physical neighbourhoods themselves was a pioneering step as researchers tied in known genetic and socio-economic risk factors to see how they intersected with the ability to, for example, walk to pick up groceries and other amenities instead of having to drive.

ICES used SAS and other software to apply advanced statistical modeling to millions of pieces of information about diabetes rates, transit routes, retail outlets, doctors’ offices, recreation centres, physician service claims, drug prescription data, and census and demographic data. The result was an actual diabetes map that highlighted the areas of the city where people were most likely to develop the disease. This helped to paint a clear picture for policy makers about areas that were most in need of education and outreach programs.

Gatifloxicin and blood sugar levels
Gatifloxicin, a commonly used antibiotic, was pulled off the market in 2006, in large part due to an ICES study that revealed it had adverse effects on blood sugar levels in otherwise healthy patients being treated for minor infections like sore throats. Some patients were being hospitalized with highly abnormal blood sugar levels that could not have been caused by their initial infections.

Health Canada encourages doctors to send in case reports when they see unexpected results. Health Canada was receiving individual reports of abnormal blood sugar levels, but it was hard to put them into context within the larger population. It was equally hard to determine what was causing the abnormality since the number of variables was so high. For example, some patients taking gatifloxicin had other kinds of ailments that were affecting their blood sugar levels, and some patients might have stopped eating because they didn’t feel well, so their blood sugar levels might have dropped. So the drug could not necessarily be identified as the cause. ICES used SAS analytics to look at the health data from a much larger number of similarly sick patients taking different antibiotics in order to pin-point gatifloxicin as the cause of the blood sugar level inconsistencies.

Car accidents and psychotropic medications
In October of 2008, ICES released the findings of a new study that revealed that older drivers who are taking psychotropic medications like antidepressants and sleeping pills represent a higher risk for having car accidents. The study involved using SAS software to link province-wide accident data from the Ministry of Transportation with data from the Ministry of Health and Long-Term Care in order to connect driving and accident records with patient illness and medication records. The outcome of the study was to recommend streamlining the process whereby doctors report potentially unsafe drivers to the Ministry of Transportation, with additional proactive outreach to the patients’ family members and to the patients themselves to advise them of the increased risk.      

Next up – tackling ER wait times
We are most excited about a real-time evaluative study we are conducting that looks at ways to improve how emergency rooms process and treat patients. Emergency department wait times are an important issue for politicians, and an even more important issue for patients waiting for care, especially those with acute medical conditions. We are excited to be working directly with decision makers in the Ministry of Health and Long-Term Care in order to help understand more about what causes backlogs in emergency room departments, and evaluate the effectiveness of new patient care strategies in the ER.

Janet E. Hux, MD, is Chief Operating Officer with the Institute for Clinical Evaluative Sciences in Toronto. This story originally appeared in the March 2009 issue of Canadian Healthcare Technology.