Concept: Health Status Indicators

Introduction

A set of population-based health status indicators was developed from multiple administrative data sources - see POPULIS for more information. These indicators were then used to compare the health status of 1 million Manitoba residents across 8 administrative regions over a 1-year period. Marked variations in health status were shown.

The strength of the various indicators was evaluated, and premature mortality emerged as the most useful indicator for future analyses. Indicators that purport to be sensitive to how well a health care system is performing showed patterns similar to those derived from classic measures (mortality, low birth weight). Furthermore, the "system sensitive indicators" did not appear to be sufficiently independent of utilization biases.

Notes:
• Potential years of life lost and life expectancy were not included in the original set of health indicators developed in POPULIS but have been included in this concept.
  
  
• Even though all of the original indicators developed for POPULIS are identified in this concept, MCHP has developed more types of health status indicators. Please see the Related Concepts section below for a list of additional "Health Status Indicator"-related concepts, and more detailed information on specific indicators.

Indicator Categories

Each of the major categories was further subdivided (e.g. mortality indicators include deaths due to injuries, chronic diseases, and cancer). Here we report highlights from the extensive analysis.

I. Demographic Profile - The Extremes of Age

The population distribution is sometimes suggested as a health status indicator since it reveals the more vulnerable groups as those at the extremes of age. We define youth as persons less than 24 years of age according to the World Health Organization ( WHO , 1986); the elderly are persons over 75, those who are most at risk for personal care home placement, disability or ill health. While this indicator was developed as part of the health status measures, in subsequent applications we have used the extremes-of-age measure (also see socio-economic indicators) as a potential indicator of the need for health care rather than as a measure of health status per se.

II. Low Weight (<2500 gm) Births

Miller et al. (1989) define low birth weight infants as those weighing less than 2500 grams. These infants are at higher risk for developmental delay, physical complications such as birth defects and death. The resource implications for low birth weight infants in terms of neonatal intensive care and continuing health problems are considerable.

Rates of low and high birth weight babies are based on all singleton (i.e. single baby) births over the most recent 3 years of Manitoba hospital data. Given that there are very few low birth weight newborns (approximately 4% of all singleton births) in each strata there should be in excess of 1,000 total births in order to obtain "statistically stable" results for low birth weight newborns. In most situations this will mean using more than 1 year of newborn data.

SAS Macro : _lowbwgt (internal access only) will calculate the rates of low and high birth weight babies by region. Birth weight is divided into three groups: 1 - < 2500g, 2- 2500-4000g, 3 - > 4000g.

III. Health Care System Sensitive Indicators

These indicators, derived from the literature, purport to reflect the need for health care. These indicators are aggregates of certain medical conditions for which medical treatment is believed to be effective in either preventing the condition, finding and treating the condition in an early phase to avoid major consequences, or treating the condition in a late phase, thus avoiding death or disability.


• Amenable Hospitalizations - Because most health care expenditures (physician and hospital) are devoted to the treatment of illness, it is reasonable to consider the impact that such expenditures have on "health." The list of medical conditions which a panel of physicians have agreed should prevent untimely death originates with a study by Charlton et al. (1983) from England and later modified by Poikolainen and Eskola (1986) and Desmeules and Semenciw (1991) . The indicators chosen were "intended to be used not to provide a definitive evaluation, but rather to indicate where a problem may exist and to stimulate further inquiry." Age limits were imposed for some of the conditions such as deaths from diabetes, acute respiratory infections and Hodgkin's disease. A list of the medical conditions included is given in Table 1 .
  
  
• Ambulatory Care Sensitive (ACS) Hospitalizations . "Diagnoses for which timely and effective outpatient care can help to reduce the risks of hospitalization by either preventing the onset of an illness or condition, controlling an acute episodic illness or condition, or managing a chronic disease or condition." The conditions were identified and assessed by a panel of U.S. physicians Billings et al. (1993) , and are listed in Table 1 .
  
  
• Avoidable Hospitalizations . Conditions for which hospitalization can be avoided if ambulatory care is provided in a timely and effective manner. Weissman et al. (1992) The conditions which are included in the grouping were those agreed upon by a panel of physicians and represent conditions which present important health problems, would be affected by appropriate ambulatory care, and have been used in previous studies. The 14 conditions and their corresponding ICD-9-CM codes are given in Table1 .
  
  
• Single event indicator conditions . List of medical conditions where death and often the disease itself are preventable or avoidable so that even one case is considered to be disturbing.( Carr et al. (1988) , World Health Organization (1986) ) These cases are considered as "sentinel" events whose occurrence is a marker that quality of care may need to be improved.
  
  
• Rate event indicator conditions . List of medical (sentinel) conditions which are considered to be of concern when there are sufficient numbers of events.( Carr et al. (1988) , World Health Organization (1986) )

Even though they contain overlapping elements no one of the five is superior to the others. These measures are derived both from mortality and from hospitalization data. They are not particularly sensitive to health care system performance; the indicators based on mortality were too infrequent to be used and those based on hospitalization data mimicked various other measures developed from hospital use data.

IV. Mortality Indicators: Population and Cause-Specific

• Premature mortality
  Premature mortality refers to deaths occurring among persons under 75 years of age and has been suggested by some health researchers as the best single indicator of health status and need for health care.( Carstairs and Morris (1989) , Eyles et al. (1991) )
  
  SAS Macro: _premort (internal access only) can be used to calculate premature mortality rates for various areas of Manitoba. This macro calculates age and sex standardized and crude rates per 1,000 population aged 0-74.
  
  
• Injury Deaths
  Injuries remain the greatest cause of death for adolescents and young adults and include unintentional injuries and suicide; in the U.S., homicide is also a major concern. Most unintentional injuries are attributable to motor vehicles. McGinnis et al. (1992)
  
  
• Cancer Deaths
  Cancer accounts for about one-quarter of all deaths ( Bisch et al. (1989) ) with lung cancer, breast cancer and cancer of the colon accounting for the most cases and deaths.( Muir and Sasco (1990) ) Other cancers such as bladder and kidney are associated with occupational exposures.( Andersen et al. (1987) )
  
  
• Deaths Due to Chronic Disease
  For adults in the mid-years, chronic diseases (heart disease, stroke and diabetes) are the main causes of death and disability. For the elderly, heart disease, stroke, chronic obstructive lung disease (emphysema), and diabetes are among the leading causes of death.( McGinnis et al. (1992) )
  
  

  Causes of death were determined from provincial vital statistics which uses Death Certificates (also known as the Medical Certificate of Death). The "main cause of death" was used in our rate calculations. Only one "cause" can be given even for persons with multiple health problems. In some circumstances it is difficult to know the cause of death precisely. Using Death Certificates may be unreliable for certain conditions where the cause of death is poorly known, for multiple conditions, or where conditions carry a social stigma.

• Life Expectancy
  MCHP computes life expectancy by gender and region, based on the mortality experience over the most recent 5 years of Manitoba vital statistics data. The calculations are based on information from Health and Welfare Canada (1992) . We use all-cause mortality for both genders, and five-year age group values after age 0.
  
  
  • Higher life expectancy is associated with better socio-economic and health conditions. Life expectancy varies with marital status, gender, income and geographical location but it is not affected by the age structure of the population.
    
  • Time series reveals a decrease in premature mortality.
    
  • Along with infant mortality, life expectancy is one of the most commonly used indicators of health status.
    
  • Mortality only gives information on fatal illnesses; it does not supply information on the number of sick individuals nor the importance of diseases that do not result in death.
    
  • Life expectancies calculated for a given period do not reflect only the mortality for that period. They may be influenced by past conditions or by the consequences of previous events (wars, epidemics) that caused a temporary increase in mortality in what are now the upper age groups.
    
  • The calculation is based on the hypothesis that the age-specific mortality observed is stable during a given period. When mortality decreases over time, the life expectancy obtained underestimates the true longevity.

    ( Source: Health & Welfare Canada, 1992 )

  SAS Macro: _life (internal access only) computes life expectancy based on the mortality experience over the most recent 5 years of Manitoba vital statistics.
  
  
• Potential Years of Life Lost
  Potential years of life lost is a measure of the number of years of life lost prior to a given age in the population (typically age=75). This measure is typically calculated on the most recent 3 years of Manitoba Vital Statistics data. A discussion of the methodology is available from Health and Welfare Canada (1992) . As a measure of premature mortality it gives more importance to the causes of early death than those that appear in old age. It varies with gender, socioeconomic status, cause of death, and geographical area.
  
  The choice of upper age limit may depend on the objectives of the study: end of adulthood, normal retirement age, etc. Infant deaths are excluded for 2 reasons: these deaths are due to causes that often have a different etiology from the deaths at later ages, and each infant death would contribute approximately 70 years of lost life, double the weighting of a death between ages 30 and 40.
  
  SAS Macro: _pyll (internal access only) will compute potential years of life lost by gender, region and optionally cause of death.

V. Indicators Based on Rates of Individuals Hospitalized*

• Injury hospitalizations
  
• Cancer hospitalizations
  
• Chronic disease hospitalizations
  
• Infectious disease hospitalizations
  
• Infectious disease mortality has declined from the onset of the century but is still associated with considerable morbidity. For the elderly, pneumonia and influenza are major causes of mortality and morbidity. Some infectious diseases reflect lifestyle: for example, pelvic inflammatory disease and AIDS. As will be seen, despite our use of rates based on individuals hospitalized rather than on rates of discharge, the indicators co-vary with utilization rates.
  
  * all hospitalizations were counted according to region of residence and included cases in which a person was hospitalized out of region.

VI. Indicators based on Physician Contacts

Indicators for various medical conditions associated with disability and social functioning.


• Disability among youth (0-24 years). The disabled represent a vulnerable population who may have been disadvantaged since birth. In developed countries, medical advances and management have enabled these children to survive beyond adolescence. ( World Health Organization (1986)) We included the rate of individuals identified on the basis of physician contacts for these conditions (i.e. two or more visits in one year for the disabling condition).
  
  The number of visits for some indicators was very small, making rates unstable and comparisons difficult. Further investigation looked at the advantage of using two years of physician contact data for identifying youths with these conditions. We found that these rates tended to co-vary with utilization patterns. Thus, we do not rely on these measures in further work.
  
  
• Functional status/role limitations/perceived health status (aged 75+)
  
  
• Restricted activity days (aged 75+)
  
  
• Functional status/role limitations/perceived health status (all ages)
  
  
• Restricted activity days (all ages)
  
  

  Physician visits included all Ambulatory Visits - Physician Physician claims must include the reason for the visit (ICD-9-CM code at the three digit level). The coding of diagnoses (reason for visit) from physician visits is less reliable than for hospitalizations. The simplified version of the coding system that is used does not discriminate well for some conditions. Although only one reason can be recorded in the database, a person may visit a physician with multiple problems. Thus, in this analysis, to ensure that a person was likely to have the health condition indicated, persons were counted if they saw a physician two or more times for one of the conditions listed.

Group Indicators and ICD-9-CM Codes

See Table 1