A simple method for analyzing and presenting health data

Document Sample
A simple method for analyzing and presenting health data Powered By Docstoc
					   A SIMPLE METHOD FOR
ANALYZING AND PRESENTING
       HEALTH DATA
         Charles J. Vukotich, Jr.
          cvukotich@achd.net




 http://info.co.allegheny.pa.us/services/achd/
• This lecture is one of series produced by the
  Allegheny County Health Department (PA),
  Bethlehem Health Bureau (PA) and the City
  of Elizabeth Department of Health &
  Human Services (NJ).

• The organizers of this project are scholars in
  the Northeast Regional Public Health
  Leadership Institute, Class of 2000. For
  information contact:
  dcw01@health.state.ny.us
• Charles J. “Chuck” Vukotich, Jr.
  – Assistant Deputy Director
  – 25 years in public health in a variety of
    management Roles
  – Carnegie Mellon University
     • BS, Chemistry
     • MS, Public Management and Policy, H.
       John Heinz School of Public Management
       and Policy
     • Member of Tau Beta Pi, and Phi Kappa Phi
• Chuck Vukotich has spent many years of
  his professional career applying
  management tools to public health. He
  developed the first program plans for the
  Allegheny County Health Department. He
  has worked in improving budgeting and
  fiscal forecasting. He has led the
  reorganization of several programs, and
  participated in department wide
  reorganization. He is presently working on
  planning, including the applications of
  MAPP, benchmarks and indicators.
    LEARNING OBJECTIVES
• To understand the importance of data for
  planning and decision making.
• To understand the concept of racial and
  ethic disparities.
• To understand basic concepts of data
  presentation.
• To understand the basic concept of fitting a
  straight line to data.
PERFORMANCE OBJECTIVES
• Know how to calculate an average.
• Know how to do a linear regression using
  EXCEL.
• Know how to create a graph in EXCEL.
• Know how to determine if a racial disparity
  exists.
• Know how to calculate rate of change.
           Data is Important
• Decision making, strategic Planning and
  program design should be data driven.
• Data must be communicated in a way that is
  understandable by all, especially members
  of the community.
• This is one way to present health data using
  readily available tools, basic analysis and
  simple presentation.
      Look at 5 Years of Data
• A change in one year usually doesn‟t mean
  anything.
• Collect the 5 most current years of data and
  look at this.
• Use age adjusted data, where available.
• Use rates when possible.
Has this Disease Increased or
 Decreased Over 5 Years ?
                      16


                      14
  Age Adjusted Rate




                      12


                      10


                      8                                White
                                                       Black

                      6                                Total


                      4
                           1994   1995   1996   1997           1998
                                         Year
        Method
Health/Disease Indicators
                                                                           Rate of
                    16
                            Positive rate of                               Change
                            change indicates                               3.3
                    14
                            that these lines are                           2.9
                                                                           6.9
Age Adjusted Rate




                    12      all increasing
                    10


                    8                                       White
                                                            Black

                    6                                       Total


                    4
                         1994       1995      1996   1997           1998
                                             Year
Regression Analysis –EXCEL Screen




    Add numbers and
     divide by 5, it‟s
    just the average.
                    Regression Output
SUMMARY OUTPUT

Regression Statistics             This tells how good the data fits a
Multiple R 0.884566
R Square 0.782457                  straight line; think of this as a
            0.709943
Adjusted R Square
Standard Error
            1.299615
                                  percent, 100% is perfect; 78% is
Observations        5                           good.
ANOVA
              df        SS       MS        F   Significance F
Regression         1    18.225   18.225 10.79041 0.046256
Residual           3     5.067    1.689
Total              4    23.292

                    Standard Error t Stat
          Coefficients                    P-value Lower 95%Upper 95%          Upper 95.0%
                                                                    Lower 95.0%
Intercept    2721.14 820.3052 3.317229 0.045147 110.5603 5331.72 110.5603 5331.72
X Variable 1    -1.35 0.410974 -3.28488 0.046256 -2.65791 -0.04209 -2.65791 -0.04209

                                    The X Variable Coefficient.
         RATE OF CHANGE

  RATE OF              X Variable Coefficient
                  =       5 Year Average
  CHANGE
  This Rate of Change is the annual percentage
 increase or decrease of the disease or condition.
A negative rate of change means that the disease
  is decreasing; a positive rate means that it is
                  increasing.
                Plot the Data
•   People like to see the data plotted out.
•   It gives them a „feel‟ for the data.
•   It makes them comfortable.
•   It‟s easy to do using Excel.
•   See screen print on next slide.
Colorectal Cancer
black
                                      EXCEL Screen
         1994      30.5
                                             Colorectal Cancer
         1995      26.3
         1996        26
                          35
         1997        26
         1998      23.9
5 Year Mean       26.54   30



                          25
Linear rate      -5.09%
change
                          20

HP2010             13.9
                          15
predicted 2010     7.64
value
                          10



                          5



                          0
                               1994   1995           1996        1997   1998
          Determine Disparities
• You can determine racial/ethnic disparities by
  comparing the 5 year averages between groups.

              (Racial/Ethnic Group Average - Overall Average)
Disparity =           Overall Average

• There may be many reasons for disparities, but a
  gap of 30% or greater indicates that the gap is real,
  and not just from random forces.
• A DISPARITY MAY NOT EXIST.
 Can We Meet the 2010 Goal?
• We can project using regression analysis
  data
 Projected          Intercept Coefficient +
2010 Value =     (X Variable Coefficient x 2,010)
Example:
  7.64      = 2,721.14 + (-1.35 x 2,010)
   Can We Meet the 2010 Goal?
• HP 2010 Goal for Colorectal Cancer is 13.9
  deaths per 100,000
• Projected Value is 7.64 deaths per 100,000
• Assumes that current trend will continue.
• Projection may not be statistically valid.
• BUT, it‟s a good “what-if” for strategic
  purposes, but just don‟t take it too seriously.
        PUT IT ALL TOGETHER
Colorectal Cancer                                   Age Adjusted Death Rates for Colorectal Cancer
                                                    (Per 100,000 population)
          1994                    1995              1996          1997          1008          5 year    Rate of
                                                                                              Average   Change
 Total         21.8       23.1      22.0    19.0                                         18.1 20.8      -5.5
 White         21.1       22.8      21.7    18.2                                         17.8 20.3      -5.5
 Black         30.5       26.3      26.0    26.0                                         23.9 26.5      -5.1
HP 2010 Goal: 13.9       NOT MET
Overall goal will be met at current trend.
Racial Disparity: Yes, Black, Risk 31% greater

                                        35

                                                                                 White
                                                                                 Black
                                        30
                    Age Adjusted Rate




                                                                                 TOTAL



                                        25




                                        20




                                        15
                                             1994   1995   1996          1997        1998
                                                           Year
          PUT IT ALL TOGETHER
                         CAUSE OF DEATH

CAUSE OF DEATH Goal Met Currently?   Goal Met in 2010? Disparity?
Heart Disease No                     Yes, but not Blacks Black: 18%
Stroke            No                 Yes, overall        Black: 27%
Lung Cancer       No                 No                  Black: 30%
Breast Cancer Yes, but not Blacks    Yes, but not Blacks Black: 20% Trend Increasing.
Colorectal Cancer No                 Yes, overall        Black: 31%
“Goal” is Healthy People 2010 Goal.

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:120
posted:4/10/2008
language:English
pages:20