TORONTO -- Out-of-pocket health expenses rose sharply from 1997 to 2009, with low-income households taking the hardest hit, a new study from Statistics Canada reveals.

Low-income households saw their health-care-related costs rise by 63 per cent over that period, compared to an increase of between 36 and 48 per cent for higher earners, the report says.

In 2009, households with the lowest incomes spent about $1,000 on health care, compared to almost $3,000 for top earners.

While those numbers may not seem enormous, for a significant portion of families in the lowest income tier that outlay represented more than five per cent of the household's after-tax annual income.

In 2009, nearly 40 per cent of households in the two lowest income brackets reported spending more than five per cent of after-tax income on health costs compared with 14 per cent of households at the top end of the income spectrum.

The report says out-of-pocket health-care spending represented 5.7 per cent of after-tax income for the lowest income households, compared to 2.6 per cent for the families at the highest end.

The study raises concerns about whether expenditures at that level are sustainable for people in the lower income tiers, questioning whether they may lead to lower use of dental care, for instance.

It also notes that there is recent Canadian evidence that shows that almost 10 per cent of people who receive a prescription either don't fill it, don't renew it or do something to stretch out the drugs because of cost constraints.

The three biggest components of out-of-pocket health-care spending were dental care, prescription drugs and health-related insurance premiums. How a household distributed its out-of-pocket spending among those three components depended on where on the income spectrum the family was.

In 2009 lower-income households spent more on prescription drugs than they did on dental care or insurance. For higher-income households, insurance was the biggest expenditure of the three.

The authors of the study note that it has some limitations, including the fact that the survey from which the information was drawn collected self-reported data. Self-reported information is subject to what's called recall bias -- memory problems.