Making Progress in Health and Health Care how

Making Progress in Health and Health Care · how do we know we are making progress? · need to distinguish two broad domains: · progress in population health · progress in health care services Michael Wolfson, Statistics Canada Denise Lievesley, UK NHS and ISI (please use “normal view” or “notes page” to see speaking text) OECD Istanbul June 2007

World’s Two Most Widely Used “Health” Indicators · Life Expectancy ( + other indicators based on mortality rates, e. g. infant mortality) · good as far as it goes; clearly fundamental · but leaves out how healthy people are while alive · Health Care Spending as % of GDP · very poor indicator · is more spending better or worse? · focuses on inputs to health care, rather than results · We can and should do better for our most basic measures of progress in health and health care OECD Istanbul June 2007

How do we know we are making progress in population health? · currently, a plethora of indicators · often a failure to distinguish “health” from · antecedents, e. g. risk factors like smoking, · correlates, e. g. bio-medical parameters like blood pressure, and · sequalae, e. g. social participation like work, mortality · simple idea: HALE = health-adjusted life expectancy · builds on already very widely use measure, life expectancy · progress ≡ “adding years to life” and/or “adding life to years” OECD Istanbul June 2007

Basic Definitions · LE = area under survival curve · HALE = “weighted” area under survival curve · where “weights” are levels of individual health status, ranging between zero (dead) and one (fully healthy) OECD Istanbul June 2007

UK LE and HALE (Simpler Method) OECD Istanbul June 2007

Measuring Functional Health Status in a Population · examples: Mc. Master Health Utility Index, Euroqol EQ-5 D, WHO World Health Survey · define a set of health domains · develop a parsimonious set of survey questions to elicit levels of functioning for each domain, and collect data for a representative sample · Budapest Initiative · apply a systematic method for eliciting values for various health states for another, typically smaller, sample · estimate a “valuation function” OECD Istanbul June 2007

Changes in Life Expectancy (LE) and Health-Adjusted Life Expectancy (HALE) by Cause, Canada LE HALE (Source: Manuel et al, ICES and Health Canada, NPHS) OECD Istanbul June 2007

Progress in Levels and in Differences – Health Inequality · old (statistical) adage: “beware of the mean” · HALE is fundamental for measuring overall progress in population health – analogous to “size of the pie” in income analysis · but HALE itself says nothing about “how the pie is divided” – about the distribution of health within a population OECD Istanbul June 2007

The Concept of Health Inequality · concept of health inequality is different · income inequality is “univariate” · e. g. what share of income goes to the top 1%; how many individuals are living on less than $1 per day? · health inequality is “bivariate”, i. e. about correlations, especially systematic associations with socio-economic status · e. g. how does health (HALE) vary from one region in a country to another; · how steep is the gradient – i. e. how much does health status improve as we move up the social ladder within a country OECD Istanbul June 2007

Life Expectancy (LE) and Health-Adjusted Life Expectancy (HALE), Canada 2001 at birth at age 65 males at birth females at birth income terciles (thirds) OECD Istanbul June 2007

An Almost Familiar World Map www. worldmapper. org; cartogram algorithm: Mark Newman OECD Istanbul June 2007

Area Proportional to Population www. worldmapper. org; cartogram algorithm: Mark Newman OECD Istanbul June 2007

Area Proportional to GDP 2002 www. worldmapper. org; cartogram algorithm: Mark Newman OECD Istanbul June 2007

Area Proportional to HIV (prevalence ages 15 – 49) www. worldmapper. org; cartogram algorithm: Mark Newman OECD Istanbul June 2007

Area Proportional to “Unhealthy Life” (LE – HALE, based on WHO estimates) www. worldmapper. org; cartogram algorithm: Mark Newman OECD Istanbul June 2007

National Income and Health, Correlated ? HALE (Sources: HALE – WHO; GDP – World Bank) GDP per capita, US $ at PPPs, 2002 OECD Istanbul June 2007

How do we know we are making progress in health care? · this is a far more popular question than progress in population health, but also not nearly so fundamental · simple reason: there is far more to the determinants of health than health care – e. g. poverty, lifestyle, hierarchy · progress in health care ≡ { health care interventions improved health of individuals treated } · n. b. most interventions are not well evaluated OECD Istanbul June 2007

Definition - Health Outcome health intervention health status “before” health status “after” other factors health outcome change in health status attributable to a health intervention (for an individual) OECD Istanbul June 2007

How NOT to Know Whether We are Making Progress in Health Care · try to use SNA (System of National Accounts) concepts to measure health care “outputs” · try to apply macro-economic concepts of aggregate productivity to the health care sector OECD Istanbul June 2007

SNA Approach: Treat Public Sector Activities the Same as the Private Sector Define (i. e. make up) “Outputs” “Profits” ? ? ? Outputs Inputs Commercial Sector Industries OECD Istanbul June 2007 Public Sector

Why the SNA Approach is Problematic · “outputs” do not exist naturally in publicly provided health care · we certainly can count “activities”, like numbers of vaccinations (probably all useful) and numbers of coronary procedures (see later slide!) · but outcomes of interventions should clearly be the objective of systematic and routine measurement · productivity is obviously important · but high “productivity” in doing useless or iatrogenic activities is bad · remember the three “E’s”: efficacy, effectiveness, and efficiency; no point measuring efficiency unless we know efficacy and effectiveness OECD Istanbul June 2007

n. b. virtually no differences in one year survival; but no data on differences in health-related Qo. L (Tu et al on Coronary Surgery) e. g. almost 17 x, with no benefits? OECD Istanbul June 2007

Heart Attack Patients in Large Health Regions – Treatment and 30 Day Mortality Rates (%) – 1995/96 to 2003/04 1995/96 2003/04 OECD Istanbul June 2007

What Does this Graph Tell Us? · we may be missing important data · treatments – e. g. nothing on thrombolysis, post AMI medication and rehabilitation · Framingham risk factors – smoking, obesity, physical activity · other risk factors – income, chronic stress · (n. b. age, sex and comorbidity included) · health care is driven by opinions · clinical judgment is not well-informed by rigorous and systematic evaluation · health system managers have no empirical bases for judging the effectiveness of their activities · aggregate SNA style measures of “productivity” miss the real issues OECD Istanbul June 2007

Concluding Comments · need to measure both progress in population health and in health care · for population health: HALE is fundamental · for health care: outcomes are fundamental · for both: a common metric for measuring individual health status is essential – propose Budapest Initiative short form questions (along with items covering many other facets of health) · using basic health information principles · incentive compatibility – providers of crucial health information should have a stake… · empowerment – information should enable both general public and providers (as well as health system managers) to improve outcomes / quality OECD Istanbul June 2007