Chapter 4: Methods Used in This Project

Chapter 4: Methods Used in this Project

Priority Medicines for Europe and the World
"A Public Health Approach to Innovation"

Background Paper

Methods Used in this Project

By Warren Kaplan, Ph.D., JD, MPH

7 October 2004

4. Methods Used in this Project 3

4.1 Overview 3

Evidence-Based Framework for Prioritization 3

"Predictive" Prioritization Framework (middle series of boxes in Figure 4.1) 4

Ethical/Moral Prioritization Framework (middle series of boxes in Figure 4.1) 4

4.2 Burden of Disease (STEP 1) 4

Disability Adjusted Life Years (DALYs) 5

Calculating DALYs 5

Mortality 6

Measuring Mortality 6

Defining the Relationship Between Disability and Mortality 7

Demographics 9

4.3 Clinical Efficacy (STEP 2) 11

Systematic Reviews, Cochrane and Others 11

Limitations of Cochrane Reviews 12

Relative Risk as an Estimator of Clinical Efficacy 13

Calculating Clinical Efficacy with Relative Risk Ratios 14

Modifying the Data for Graphical Presentation: Using a Convention 16

Limitations of Relative Risk Calculations 16

Adverse Events 18

4.6 Reviewing the Preliminary List: In Depth Reviews- Based on Global Forum for Health Research (STEP 3) 19

4.7 Conclusions 20

4.8 Reading List on Meta-Analysis and Systematic Reviews 20

4.9 Endnotes 22

Annexes

4. Methods Used in this Project

4.1 Overview

The methodology aims to determine pharmaceutical “gaps” and to create a public health-based research agenda for the EU. As discussed in Chapter 3, our prioritization scheme employs several different conceptual frameworks. Our "evidence based" framework is described in this section. Our "ethical/moral" conceptual framework for prioritization is summarized in Chapter 3. Our "predictive" framework is described below.

Evidence-Based Framework for Prioritization

The evidence-based method uses analyses of several different factors: burden of disease, demographics, and clinical efficacy (upper two boxes in Figure 4.1).

Figure 4.1 is a Schematic of the Methodology

1. We conducted a ranking exercise using burden of disease information (disability adjusted life years and mortality) to generate the major disease conditions which account for the majority of the total DALY burden and the total mortality burden in both the expanded EU ("EU25") and the rest of the world. See Tables 5.3 and 5.4 in Background Chapter 5.

2. For these diseases, we looked to the Cochrane Database of Systematic Reviews for information regarding the clinical efficacy of any pharmacological interventions used to treat these conditions. From these systematic reviews, (provided such reviews existed) we determined in a quantitative manner if there were any pharmaceutical "gaps" in the treatment of these disease conditions. If so, we developed a Preliminary List of diseases for which there may be putative pharmaceutical gaps. The Cochrane library was accessed via the search engine of the internal WHO website (http://www.who.int). We used other analyses of clinical trials found in peer-reviewed articles as well as volume 10 of Clinical Evidence published by the BMJ publishing group (www.clinicalevidence.com).

"Predictive" Prioritization Framework (middle series of boxes in Figure 4.1)

For many major disease conditions, there is little to be found in the Cochrane Reviews (e.g., diabetes). Observational and clinical evidence was used to further define a Preliminary List. For example, we know from demographic studies (See Chapter 5) that multiple drug resistant TB is an ongoing public health problem in Eastern Europe. Given the recent concern over SARS and avian influenza, we have included pandemic influenza in our Preliminary List. (See also Chapter 6.2) Given the importance of TB, especially drug resistant TB, as a global public health threat, we have included both antibacterial resistance and TB in our Preliminary List. We know that diabetes, a major global public health concern, is poorly represented on the Cochrane Database. We have placed this on the Preliminary List based on our understanding of unmet therapeutic needs as they were revealed in the detailed review (See Background Chapter 6.4). We also know from demographic predictions, that the population is aging rapidly and that the disproportionate share of elderly will be elderly women. Thus, we have included in our Preliminary List diseases of the elderly (See also Chapter 7.2 of this Report). Pronouncements of the World Health Assembly (WHA: the governing body of the World Health Organization), while having no legal significance, are based on global political consensus and therefore should be reviewed. We have relied, in part, on the most recent WHA announcements (WHA57.1: May 2004) regarding Buruli ulcer and HIV/AIDS (WHA57.14:May 2004) for inclusion in our list of "neglected diseases" and HIV, respectively.

Ethical/Moral Prioritization Framework (middle series of boxes in Figure 4.1)

We have also used a framework based on previous ideas about equity and distributional justice to inform decisions about priority medicines. Chapter 3 discusses this in more detail but clearly, "neglected" diseases and neglected patient groups (women, children, elderly) have been placed on the list as the result of this type of review. There is some overlap between the conditions generated using this method and the "predictive" framework method. The Preliminary List contains 16 priority conditions for which "priority medicines" are needed.

3. We next reviewed this Preliminary List to update information on therapeutic interventions using a standard set of questions to further collect information on the Preliminary List of particular diseases having a pharmaceutical gap. These detailed reports make up Chapter 6 of this report.

In the following sections, we provide more detailed information on the components of our method.

4.2 Burden of Disease (STEP 1)

The WHO has defined “burden of disease” as the significance of a particular disease for society beyond the immediate cost of treatment.[1] Burden of Disease measurements are based on a comprehensive and consistent set of estimates of mortality and morbidity by age, sex and geographic region that have been carried out by the World Health Organization. The primary quantifier of burden of disease is the disability adjusted life year (DALY). The DALY is a summary measure of population health that combines in a single indicator years of life lost from premature death and years of life lived with disabilities. One DALY can be thought of as one lost year of ‘healthy’ life and the burden of disease as a measurement of the gap between current health status and an ideal situation where everyone lives into old age free of disease and disability. In addition, we report data on mortality by cause, sex and age category. We have used data on mortality and DALYs from the World Health Organization[2] (including the European Health for All database[3] for our estimates of burden of disease for both acute and chronic conditions

The WHO global burden of disease methodology is not perfect but it is probably the best single tool available for the audience of this Report- strategic planners and decision makers. It does provide a single summary measure of ill health, a fundamental tool for policy makers when considering the relative benefits of different policy options. Its methodology is open to challenge but it is standardized around the world. The burden of disease approach is well established in Europe as it can be broken down to show the relative contributions of different conditions to the overall burden of disease and it can show the burden of disease that can be attributed to known risk factors.

Disability Adjusted Life Years (DALYs)

The key health ranking system used by this Report is that of the WHO and is based upon the Disability-Adjusted Life Year (or DALY).[4] DALYs are a way of aggregating the number of life years lost by sufferers from each disease with the amount of disability suffered while they are still alive. These two amounts are combined in a complex manner to give the overall burden of that disease. Disease burdens are thus measured in DALYs lost due to each disease.

Consistent and meaningful DALY estimates depend on a clear definition of the condition under consideration in terms of case or episode, and severity level or disease stage. It is then necessary to ensure that the disability weight and the population incidence or prevalence data relate to the same case definition. Thus far, there appears to be a scarcity of empirical data on health state valuations. The DALY concept has generated a debate and a literature of its own within the public health community.[5],[6],[7],[8],[9],[10],[11]

Calculating DALYs

Data on morbidity and mortality are converted into DALYs through four sets of weights.

1.The number of years lost due to premature mortality is weighted by life expectancy at the age of death. The highest values are used (i.e., Japanese women with a life expectancy of 82.5 years at birth adjusted to allow for lower survival rates of males).

2. Future life years are discounted. This gives future years less weight than current years and is consistent with discounting of costs.

3. Life years are valued differently at different ages. A year of young or middle-aged adult life is valued more highly than a year of life lived by young children or the elderly.

4. Weighting by disability weight allows for years of life lost and years lived with a disability to be measured on the same scale. The disability weights use indicator conditions and from these weights, measurements for all other diseases and disabilities are obtained. To quantify the full loss of healthy life, the World Bank and the World Health Organization undertook a joint exercise in which they classified diseases into 109 categories on the basis of the International Classification of Diseases( tenth revision). These categories cover all possible causes of death and about 95 percent of the possible causes of disability. Using the recorded cause of death where available, and expert judgment when records were not available, the study assigned all deaths to these categories by age, sex, and demographic region. For each death, the number of years of life lost was defined as the difference between the age at death and the expectation of life at that age in a low-mortality population. For disability, the incidence of cases by age, sex, and demographic region was estimated on the basis of community surveys or, expert opinion. The number of years of health life lost was then obtained by multiplying the expected duration of the condition (to remission or to death) by a severity weight that measured the severity in comparison with loss of life.

Diseases were grouped into six classes of severity . The death and disability losses were then combined, and allowance was made for a discount rate of 3 percent (so that future years of healthy life were valued at progressively lower levels) and for age weights (so that years of life lost at different ages were given different relative values). The value for each year of life lost, rises steeply from zero at birth to a peak at age 25 and then declines gradually with increasing age. These age weights reflect a consensus judgment, but other patterns could be used-for example, uniform age weights, with each year of life having the same value, which would increase the relative importance of childhood diseases. The combination of discounting and age weights produces the pattern of DALYs (disability-adjusted life years). The sum across all ages, conditions, and regions is referred to as the Global Burden of Disease (GBD).

Mortality

The number of deaths, by age and sex, provides an essential “envelope” which constrains individual disease and injury estimates of deaths. The sum of deaths from all specific causes for any sex-age group must sum to the total number of deaths for that age-sex group. From the estimated age-specific mortality rates, life tables for the populations can be derived using standard methods. The WHO Global Burden of Disease studies estimate global and regional mortality. The WHO contacts Member states directly on a routine basis to obtain their latest data on cause-of-death from their vital registration sources. In the absence of complete and accurate vital registration system, countries are requested to submit data from other reliable sources as well. For countries without exploitable vital registration data, other sources of adult mortality such as survey and census were used to estimate the level of adult mortality. Those data submitted by Member states become part of WHO’s historical data base on causes of death which contains data as far back as 1950. Computerization of data at country level and electronic transmission to WHO have considerably improved the timeliness of information received.

Measuring Mortality

At the first level, overall mortality is divided into thee broad groups of causes: Group I, consisting of communicable diseases, maternal causes, conditions arising in the perinatal period and nutritional deficiencies, Group II, encompassing the non-communicable diseases; and Group III, comprising intentional and unintentional injuries.

Deaths and health states are categorically attributed to one underlying cause using the rules and conventions of the International Classification of Diseases. In some cases, diseases may act as risk factors for other diseases, and the total burden attributable to a disease may be greater than that assigned under the ICD conventions. For example, the total attributable burden for the conditions listed below include the following:

Hepatitis Include attributable burden of liver cancer and renal failure

Diabetes Include attributable burden of cardiovascular disease and renal failure

Depression Include attributable burden of suicide

For many countries, the vital registration systems is good enough to provide a time series of annual life tables (adjusted if the registration level is incomplete). For small countries with population size below 500 000, moving averages are used to smooth the time series. Certain life table parameters, survivors at age 5 (l5) and at age 60 (l60) are projected using a weighted regression model giving weight to recent data, depending on availability. For the GBD work by WHO in 2000, projected values of l5 and l60 are then applied to a life table model using the most recent national data as the standard, to predict the full life table in 2000.

In other countries with incomplete or sample vital registration systems, demographic techniques are used to estimate the level of completeness of death recording and to adjust the data accordingly. These adjusted levels of child mortality and adult mortality excluding HIV/AIDS deaths where necessary, are used to predict levels of two life table parameters (l5, l60) in 2000 assuming a constant pace of mortality change to what was observed in the 1990s. These predicted values of l5 and l60 in 2001 were then applied to a life table model, using a global standard, to predict the full life table in 2001. HIV/AIDS deaths were added to total mortality rates where necessary.