Improving Preventive Care for Older Americans (Text Version)

Slide Presentation from the AHRQ 2010 Annual Conference

On September 29, 2010, Elizabeth Eckstrom made this presentation at the 2010 Annual Conference. Select to access the PowerPoint® presentation (1 MB). Free PowerPoint® Viewer (Plugin Software Help).


Slide 1

Improving Preventive Care for Older Americans

Improving Preventive Care for Older Americans

Oregon Evidence-Based Practice Center/
Kaiser Permanente Center for Health Research

Principal Investigator: Evelyn P. Whitlock, MD, MPH
Elizabeth Eckstrom, MD, MPH
David Feeny, PhD
Jennifer Lin, MD, MCR
Rongwei Fu, PhD
Leslie A. Perdue, MPH
 

Slide 2

What are the most important challenges in synthesizing the evidence on prevention in older adults?

What are the most important challenges in synthesizing the evidence on prevention in older adults?

Slide 3

Which outcomes do you measure when lengthening life is no longer the primary goal?

Which outcomes do you measure when lengthening life is no longer the primary goal?

  • Focus shifts to Health Related Quality of Life and Function.
  • More reliance on self-reported measures: but are they reliable?
    • Test-retest reliability
    • Inter-observer reliability
  • Do these measures have evidence for validity in older people?
    • Criterion validity
    • Cross-sectional construct validity
    • Longitudinal construct validity (responsiveness)

Slide 4

The Pitfalls of "Effect Size"

The Pitfalls of "Effect Size"

  • Distribution based scales may be difficult to translate to clinically important change:
    • COPE trial-caregiver training for dementia subjects showed ".24" change in ADL/IADLs. This was statistically significant. But does it mean subjects got modest improvement in one ADL? Or a small improvement in multiple ADLs? Was the change enough to make the subjects more independent (i.e., less work for the caregiver)?
  • Anchor based criteria might be easier to translate.

Slide 5

Better Yet: Translate numbers to "Percent Improved"

Better Yet: Translate numbers to "Percent Improved"

Table 3. Clinical Significance of Main Outcomes at 4 Months

 Control Group, No. (%)a
(n = 107)
Intervention Group, No. (%)a
(n = 102)
Difference in Net Improvement
(95% CI)
P Value
ImprovedWorsenedNet ImprovementImprovedWorsenedNet Improvement
Overall functional dependenceb41 (39.8)11 (10.7)30 (29.3)51 (51.5)3 (3.0)48 (48.5)19.2
(2.7 to 36.0)
.02
IADL dependenceb52 (50.5)7 (6.8)45 (43.7)64 (64.6)3 (3.0)61 (61.6)17.9
(1.9 to 34.0)
.03
Activity engagemenet40 (37.4)42 (39.3)-2 (-1.9)44 (43.1)31 (30.4)13 (12.7)14.6
(-8.8 to 38.0)
.22
Perceived changed in well-being42 (39.3)21 (19.6)21 (19.6)58 (58.9)11 (10.8)47 (46.1)26.5
(7.2 to 45.8)
.007
Confidence using activitiesc29 (27.4)24 (22.6)5 (4.7)41 (41.0)10 (10.0)31 (31.0)26.3
(7.9 to 44.7)
.005

Abbreviations: CI, confidence interval; IADL, instruments of daily living.
a Not shown are the numbers (%) of those who stayed the same.
b This measure was assessed for 99 of 102 patients in the intervention group and 103 of 107 patients in the control group because 7 patients were placed in nursing homes and the caregivers were unable to assess functional dependence.
c This measure was assessed for 106 caregivers in the control group.

Clinical Significance of Main Outcomes at 4 Months. Gitlin, L. N. et al. JAMA 2010;304:983-991.

Slide 6

More Methodologic Challenges in Outcomes

More Methodologic Challenges in Outcomes

  • Large inter-study variation in outcome reporting (domains measured, domain-specific measures chosen and reported) within a body of evidence.
  • Floor and ceiling effect issues for measures.
  • How much time is needed for change to occur?
  • What is the minimum clinically important difference?

Slide 7

Challenges in Synthesizing Results: Considering a Constellation of Outcomes

Challenges in Synthesizing Results: Considering a Constellation of Outcomes

Example from Multifactorial Intervention and Management to prevent Functional Decline

Image: Chart shows a series of nested circles. The innermost is labeled "Mortality k=44"; the next surrounding it is labeled "Meta-Analysis k=51" and the outer circle is labeled "Total Trials k=63." To the right are four smaller circles, one of which is nested inside another. The inner of these smaller circles is labeled 'IADL k=10" and the outer "IADL k=12." The two other small circles are labeled "Hospitalization k=13" and "Institutionalization k=15."

Slide 8

Outcomes represent different bodies of evidence; how would you give one summary grade for the review?

Outcomes represent different bodies of evidence; how would you give one summary grade for the review?

Image: Chart shows three intersecting circles. The first is labeled "ADL (k=12), IADL (k=10)"; the second is labeled "Hospitalization (k=13)" and the third is labeled "Institutionalization k=15." In the intersected area between the first and second circles is the number 3. In the intersected area between the first and third circles is the number 1. In the intersected area between the second and third circles is the number 6.

Slide 9

Challenges in synthesizing intervention studies

Challenges in synthesizing intervention studies

  • Population: Baseline risk often varies study to study, unclear if "appropriate" risk population was identified for that intervention.
  • Inconsistent reporting of patient risk and use of mean differences without subgroup. explorations limit ability to determine if there are differential effects by subgroups.
  • Intervention: Often very broad and clinically heterogeneous array of interventions-difficult to fairly consider or categorize.
  • Comparison: Older people have a variable trajectory- harder to determine if groups similar-especially without randomization.
  • Outcomes: Not independent of each other.
  • Very small impact of some MFAM interventions on ADL and IADL.
  • Inconsistent reporting of a set of outcomes limits ability to determine true benefit-as may reduce hospitalizations but increase institutionalization.

Slide 10

How do you think these challenges are different for older adults than for younger adults?

How do you think these challenges are different for older adults than for younger adults?

Slide 11

Relative lack of Research

Relative lack of research

  • Older adults are relatively understudied as a group, and extrapolation from younger adults may not be accurate.

Slide 12

Health Trajectories

Health Trajectories

  • Older people with similar baseline health status may decline at markedly different rates.
  • Concept of "Compression of Morbidity."
  • Large, population-based cohort studies needed to ascertain range of possible trajectories and define predictors for those trajectories.

Slide 13

Figure 1. HUI 3 trajectories, by age, men aged 40 or older in 1994/1995, Canada excluding territories

Figure 1. HUI 3 trajectories, by age, men aged 40 or older in 1994/1995, Canada excluding territories

Image: Graph illustrates the selection biases associated with following only those who survive and remain in the community or only those who survive. Health-adjusted life expectancy at birth (2001) is age 68; life expectancy at birth (2002) is age 72; health-adjusted life expectancy at age 65 (2001) is 73; life expectancy at age 65 (2001) is 82. The UI scores for living (household), living (household and institution), and living and dead all begin at 0.9 at age 40, but begin to divide at age 65, with living and dead decreasing to 0.8. The divide widens at age 72; living (household) and living (household and institution) are now 0.8, but living and dead are 0.6. At age 82, living (household) and living (household and institution) are approximately 0.7, with living (household and institution) slightly lower; living and dead has decreased to 0.4. After age 82, all three lines decrease sharply; living (household) decreases to 0.1 by age 100, living (household and institution) to -0.2, and living and dead to -0.9.

Note: The "effect" of the "living and dead" for males illustrates their higher rates of mortality.

Health-Related Quality of Life: A 10-year Cohort Study. Health Reports, Vol. 20, No. 1, March 2009, pp 29-35.

Slide 14

Figure 2. HUI 3 trajectories, by age, women aged 40 or older in 1994/1995, Canada excluding territories

Figure 2. HUI 3 trajectories, by age, women aged 40 or older in 1994/1995, Canada excluding territories

Image: Graph illustrates the selection biases associated with following only those who survive and remain in the community or only those who survive. Health-adjusted life expectancy at birth (2001) is age 71; life expectancy at birth (2002) is age 82; health-adjusted life expectancy at age 65 (2001) is 79; life expectancy at age 65 (2001) is 84. The UI scores for living (household), living (household and institution), and living and dead all begin at 0.9 at age 40, but decrease to 0.8 at age 71 and begin to divide. The divide widens after age 79; living (household) and living (household and institution) are now approximately 0.7, with living (household and institution) slightly lower, but living and dead decrease to 0.6. By age 84, living (household) and living (household and institution) are approximately 0.6, with living (household and institution) still slightly lower; living and dead has decreased to 0.5. After age 84, all three lines decrease sharply; living (household) decreases to 0.0 by age 100, living (household and institution) to -0.2, and living and dead to -0.5.

Health-Related Quality of Life: A 10-year Cohort Study. Health Reports, Vol. 20, No. 1, March 2009, pp 29-35.

Slide 15

Assessing Net Benefit

Assessing Net Benefit

Measuring net benefit for functional outcomes

  • Many different functional scales (Katz ADL, Lawton/ Brody IADL, SF-36 Physical Function, etc).
  • Challenge to synthesize results into a single effect size:
    • Consider absolute magnitude of the change in function and defining proportion improved, not changed, or worsened.
  • Benefits and harms may be completely different constructs so how do you compare?
    • Benefit might be reduced caregiver burnout, harm might be a fall—how do you compare these?

Slide 16

Assessing Harms

Assessing Harms

  • Often inadequate assessment/ representation of harms:
    • Few RCTs of older people so harms not well quantified (need to utilize observational studies better).
    • Harms usually restricted to the screening itself; eg, pain of biopsy. Need to broaden scope (daughter must miss ½ day of work to take mom to mammogram).

Slide 17

Health Related Quality of Life

Health Related Quality of Life

  • HRQOL particularly important for older adults with limited life expectancy.
  • Many RCTs have not included HRQOL as an outcome measure, so we are stuck not being able to broaden our scope of benefit and harm to include HRQOL.

Slide 18

What's needed to enhance this field?

What's needed to enhance this field?

  • Consistent and complete ascertainment and reporting of study population baseline risk:
    • Understand population's natural history of progression or risk for functional decline.
    • Investigate impact of effectiveness of interventions in subgroups at higher risk for poor outcomes.
  • Standardized reporting of intervention descriptions to help better characterize complex interventions and to compare across different interventions:
    • More research needed to test consistent models or intervention components.
  • Detailed reporting about instruments for ADL/IADL and HRQOL measures:
    • Focused and consistent set of agreed upon measures and reporting of measures.
  • Consistent use and reporting of a constellation of outcomes that characterize net health benefit or harm.

Slide 19

Comments and Questions

Comments & Questions

Image: A photograph of a snow-capped mountain is shown.

Current as of December 2010
Internet Citation: Improving Preventive Care for Older Americans (Text Version). December 2010. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/news/events/conference/2010/eckstrom/index.html