Effects of Pediatric Asthma Education on Hospitalizations and

							Effects of Pediatric Asthma
Education on Hospitalizations and
Emergency Department Visits:
A Meta-Analysis
      Janet M. Coffman, PhD, Michael D. Cabana, MD, MPH,
            Helen A. Halpin, PhD, Edward H. Yelin, PhD
                  University of California, San Francisco
                    University of California, Berkeley
  Institute for
  Health Policy
  Studies                        June 3, 2007
Background and Rationale
 NHLBI guidelines recommend asthma
  education for all patients
 Latest meta-analysis only assessed studies
  published prior to 1999
 A number of additional studies have been
  published over the past eight years
 Innovations in treatment of asthma
 Dissemination of NHLBI guidelines
                                               2
Research Question
 Compared to usual care, does the provision
  of asthma education to children and their
  parents reduce
   Asthma ED visits?
   Asthma hospitalizations?




                                               3
Methods
 Research Design: Meta-analysis
 Databases:
     Cochrane Database of Systematic Reviews
     Cochrane Register of Controlled Trials
     PubMed
     Cumulative Index of Nursing and Allied Health
      Literature (CINAHL)


                                                      4
Methods
 Inclusion Criteria
   Enrolled children aged 2-17 years with a clinical
    diagnosis of asthma
   Conducted in the United States
   Compared asthma education to usual care
   Included a control or comparison group
   Examined ED visits and/or hospitalizations for
    asthma
                                                        5
Methods
 Calculated pooled findings for
   Odds of an event
        ED visit
        Hospitalization
   Mean Number of events
        ED visits
        Hospitalizations



                                   6
Methods
 Analysis
   Estimated fixed effects models for all outcomes
   Conducted Chi-Square test to determine whether
    results of the studies pooled are heterogeneous
   Where results were heterogeneous (i.e., p<0.1 for
    Chi-Square test), estimated random effects
    models
   Small number of studies precluded performing
    meta-regression to explore sources of
    heterogeneity
                                                        7
Results of Literature Search
 174 abstracts reviewed
 23 articles met the inclusion criteria
 Research design
   19 studies (83%) were RCTs or cluster RCTs
   4 (17%) were nonrandomized studies




                                                 8
Study Characteristics
 Demographics: in 16 studies (70%) most of the
  children enrolled were low-income
 Target of intervention: 57% provided education to
  both children and parents
 Types of education: included individual counseling,
  group classes, telephone calls, and educational
  computer games
 Types of settings: included outpatient clinics/
  physician offices, emergency departments, schools,
  and homes

                                                    9
Odds of ED Visit
Education vs. Usual Care – Fixed Effects
                                                    Odds ratio
          Study                                     (95% CI)           % W eight


 Butz                                               0.71 (0.33,1.52)         6.9
 Farber                                             0.97 (0.30,3.14)         2.5
 Guendelman                                         0.48 (0.16,1.39)         4.5
 Harish                                             0.57 (0.28,1.17)         8.9
 JosephMild                                         0.95 (0.55,1.66)        11.4
 JosephModSev                                       1.12 (0.49,2.57)         4.7
 Lukacs                                             1.21 (0.71,2.08)        10.8
 Persaud                                            0.29 (0.07,1.21)         3.1
 Shields                                            1.49 (0.75,2.95)         6.0
 Sockrider                                          0.62 (0.31,1.23)         9.2
 Teach                                              0.55 (0.38,0.80)        32.0


 Overall (95% CI)                                   0.77 (0.63,0.94)



                    .1            1                10
                                         Odds ratio


    Test of OR = 1: z = 2.61, p = 0.009;
    Test of Heterogeneity: χ2 =14.59 (df = 10), p = 0.148
                                                                                   10
Odds of Hospitalization
Education vs. Usual Care – Random Effects
                                                      Odds ratio
           Study                                      (95% CI)             % W eight



  Butz                                                0.62 (0.16,2.39)           6.3

  Evans1999                                           0.71 (0.49,1.04)          24.3
  Farber                                              8.79 (0.43,180.63)         1.5
  Guendelman                                          4.07 (0.44,37.50)          2.6
  Harish                                              1.03 (0.47,2.26)          13.4
  Lukacs                                              2.50 (0.96,6.54)          10.4
  Morgan                                              0.76 (0.49,1.15)          22.9
  Teach                                               0.51 (0.29,0.90)          18.8


  Overall (95% CI)                                    0.87 (0.60,1.27)



                     .1   1         10
                                         Odds ratio


  Test of OR = 1: z = 0.70, p = 0.482;
  Test of Heterogeneity: χ2 =13.31 (df = 7), p = 0.065
                                                                                       11
Mean ED Visits
Education vs. Usual Care – Random Effects
                                                       Standardised Mean diff.
            Study                                      (95% CI)            % W eight

  Alexander                                            -1.09 (-2.02,-0.16)       2.0
  Bartholomew                                          0.06 (-0.28,0.40)         8.9
  Christiansen                                         0.10 (-0.53,0.73)         3.9
  Clark                                                -0.16 (-0.43,0.12)       10.9
  Fireman                                              -0.78 (-1.58,0.02)        2.6
  Harish                                               -0.44 (-0.79,-0.09)       8.7
  JosephMild                                           0.00 (-0.20,0.20)        13.9
  JosephModSev                                         -0.06 (-0.43,0.30)        8.3
  Kelly                                                -0.45 (-0.90,0.00)        6.4
  La Roche                                             -0.37 (-1.22,0.47)        2.4
  McNabb                                               -1.09 (-2.24,0.06)        1.4
  Morgan                                               -0.06 (-0.19,0.08)       16.3
  Persaud                                              -0.76 (-1.44,-0.08)       3.4
  Shields                                              0.09 (-0.18,0.37)        11.0


  Overall (95% CI)                                     -0.17 (-0.31,-0.03)



                     -3          0                   3
                                     Standardised Mean diff.


    Test of SMD = 0: z = 2.40, p = 0.016;
    Test of Heterogeneity: χ2 =24.48 (df = 13), p = 0.027
                                                                                       12
Mean Hospitalizations
Education vs. Usual Care – Random Effects
                                                     Standardised Mean diff.
              Study                                  (95% CI)            % W eight




      Bartholomew                                    -0.10 (-0.44,0.24)       26.4

      Christiansen                                   -0.37 (-1.00,0.27)       13.1

      Clark                                          -0.17 (-0.43,0.10)       31.6

      Fireman                                        -0.79 (-1.59,0.01)        9.2

      Kelly                                          -0.77 (-1.23,-0.31)      19.7




      Overall (95% CI)                               -0.35 (-0.63,-0.08)



                         -3    0                   3
                                   Standardised Mean diff.


  Test of SMD = 0: z = 2.53, p = 0.012;
  Test of Heterogeneity: χ2 =7.68 (df = 4), p = 0.104
                                                                                     13
Possible Reasons for Heterogeneity
 Although there are not enough studies for
  meta-regression, findings for effects on ED
  visits appear to differ based on
   Type of education: individual education more
    effective than group education
   Setting: providing education in clinical settings
    more effective than providing in school


                                                        14
Limitations
 Only assessed effects on ED visits and
  hospitalizations
 Lack of consistent measures of severity of
  asthma symptoms
 Potential publication bias
 Results may not generalize to
   Upper- and middle-income children
   Children outside the USA

                                               15
Conclusions and Implications
 Pediatric asthma education reduces
      Odds of an ED visit
      Mean ED visits
      Mean hospitalizations
 However, in our sample, pediatric asthma education
  does not affect odds of hospitalization
 Health plans should provide incentives for pediatric
  asthma education
                                                         16
Thank You
 Co-authors
   Michael D. Cabana, MD, MPH, UCSF
   Edward H. Yelin, PhD, UCSF
   Helen A. Halpin, PhD, UC-Berkeley
 Funders
   California Health Benefits Review Program
   National Institutes of Health (#HL70771)

                                                17
QUESTIONS?

             18
Opportunities for Research
   Cost-effectiveness of pediatric asthma
    education
   Identification of the most important
    components of asthma education
   Which children benefit most from asthma
    education


                                              19
Why
Limit Meta-Analysis to US Studies?
•   Interested in effect of pediatric asthma education on
    ED visits
•   ED utilization depends in part on a country’s health
    care system
•   In the US, many low-income children have poor
    access to primary care
•   Including studies from countries with universal
    health care may have confounded the results

                                                            20
Comparisons of Different
Educational Interventions
 Comparisons of different educational
  interventions suggest that greater reductions
  in hospitalizations and ED visits were
  associated with
     More sessions
     More comprehensive education
     More interactive modes of education

                                                  21