Re: Empirical evaluation of information graphics? pt 2

Hi Will and Café,

Here are 4 more references on information graphics for decision  
making. They are from a professional at Becton Dickinson who works in  
diabetes care.
1

BMC Med Res Methodol. 2008 Feb 25;8:8.  [ http://www.pubmedcentral.nih.gov/articlerender.fcgi?tool=pubmed&pubmedid=18298827 
  ]

The harvest plot: a method for synthesising evidence about the  
differential effects of interventions.Ogilvie D, Fayter D, Petticrew  
M, Sowden A, Thomas S, Whitehead M, Worthy G.

Medical Research Council Social and Public Health Sciences Unit,  
Glasgow, UK. [email protected]

BACKGROUND: One attraction of meta-analysis is the forest plot, a  
compact overview of the essential data included in a systematic review  
and the overall 'result'. However, meta-analysis is not always  
suitable for synthesising evidence about the effects of interventions  
which may influence the wider determinants of health. As part of a  
systematic review of the effects of population-level tobacco control  
interventions on social inequalities in smoking, we designed a novel  
approach to synthesis intended to bring aspects of the graphical  
directness of a forest plot to bear on the problem of synthesising  
evidence from a complex and diverse group of studies. METHODS: We  
coded the included studies (n = 85) on two methodological dimensions  
(suitability of study design and quality of execution) and extracted  
data on effects stratified by up to six different dimensions of  
inequality (income, occupation, education, gender, race or ethnicity,  
and age), distinguishing between 'hard' (behavioural) and  
'intermediate' (process or attitudinal) outcomes. Adopting a  
hypothesis-testing approach, we then assessed which of three competing  
hypotheses (positive social gradient, negative social gradient, or no  
gradient) was best supported by each study for each dimension of  
inequality. RESULTS: We plotted the results on a matrix ('harvest  
plot') for each category of intervention, weighting studies by the  
methodological criteria and distributing them between the competing  
hypotheses. These matrices formed part of the analytical process and  
helped to encapsulate the output, for example by drawing attention to  
the finding that increasing the price of tobacco products may be more  
effective in discouraging smoking among people with lower incomes and  
in lower occupational groups. CONCLUSION: The harvest plot is a novel  
and useful method for synthesising evidence about the differential  
effects of population-level interventions. It contributes to the  
challenge of making best use of all available evidence by  
incorporating all relevant data. The visual display assists both the  
process of synthesis and the assimilation of the findings. The method  
is suitable for adaptation to a variety of questions in evidence  
synthesis and may be particularly useful for systematic reviews  
addressing the broader type of research question which may be most  
relevant to policymakers.

PMID: 18298827 [PubMed - indexed for MEDLINE]

PMCID: PMC2270283

-------------------------

2

Stat Med. 2002 Sep 30;21(18):2641-52 [ http://www3.interscience.wiley.com/journal/98516224/abstract?CRETRY=1&SRETRY=0 
  ]

A graphical method for exploring heterogeneity in meta-analyses:  
application to a meta-analysis of 65 trials.Baujat B, Mahé C, Pignon  
JP, Hill C.

Institut Gustave Roussy, Département de Biostatistique et  
d'Epidémiologie, 39 rue Camille Desmoulins, 94805 Villejuif cedex,  
France.

Heterogeneity can be a major component of meta-analyses and by virtue  
of that fact warrants investigation. Classic analysis methods, such as  
meta-regression, are used to explore the sources of heterogeneity.  
However, it may be difficult to apply such a method in complex cases  
or in the absence of an a priori hypothesis. This paper presents a  
graphical method to identify trials, groups of trials or groups of  
patients that are sources of heterogeneity. The contribution of these  
trials to the overall result can also be evaluated with this method.  
Each trial is represented by a dot on a 2D graph. The X-axis  
represents the contribution of the trial to the overall Cochran Q-test  
for heterogeneity. The Y-axis represents the influence of the trial,  
defined as the standardized squared difference between the treatment  
effects estimated with and without the trial. This approach has been  
applied to data from the Meta-Analysis of Chemotherapy in Head and  
Neck Cancer (MACH-NC) comprising 10,850 patients in 65 randomized  
trials. The graphical method allowed us to identify trials that  
contributed considerably to the overall heterogeneity and had a strong  
influence on the overall result. It also provided useful information  
for the interpretation of heterogeneity in this meta-analysis. The  
proposed graphical method identifies trials that account for most of  
the heterogeneity without having to explore all possible sources of  
heterogeneity by subgroup analyses. This method can also be applied to  
identify types of patients that explain heterogeneity in the treatment  
effect. Copyright 2002 John Wiley & Sons, Ltd.

PMID: 12228882 [PubMed - indexed for MEDLINE]

-------------------------

3

J Clin Epidemiol. 2001 Oct;54(10):1046-55. [ http://www.jclinepi.com/article/S0895-4356(01)00377-8/abstract 
  ]

Funnel plots for detecting bias in meta-analysis: guidelines on choice  
of axis. Sterne JA, Egger M.

MRC Health Services Research Collaboration, Department of Social  
Medicine, University of Bristol, Canynge Hall, Whiteladies Road, BS8  
2PR, Bristol, UK. [email protected].

Asymmetry in funnel plots may indicate publication bias in meta- 
analysis, but the shape of the plot in the absence of bias depends on  
the choice of axes. We evaluated standard error, precision (inverse of  
standard error), variance, inverse of variance, sample size and log  
sample size (vertical axis) and log odds ratio, log risk ratio and  
risk difference (horizontal axis). Standard error is likely to be the  
best choice for the vertical axis: the expected shape in the absence  
of bias corresponds to a symmetrical funnel, straight lines to  
indicate 95% confidence intervals can be included and the plot  
emphasises smaller studies which are more prone to bias. Precision or  
inverse of variance is useful when comparing meta-analyses of small  
trials with subsequent large trials. The use of sample size or log  
sample size is problematic because the expected shape of the plot in  
the absence of bias is unpredictable. We found similar evidence for  
asymmetry and between trial variation in a sample of 78 published meta- 
analyses whether odds ratios or risk ratios were used on the  
horizontal axis. Different conclusions were reached for risk  
differences and this was related to increased between-trial variation.  
We conclude that funnel plots of meta-analyses should generally use  
standard error as the measure of study size and ratio measures of  
treatment effect.

PMID: 11576817 [PubMed - indexed for MEDLINE]

-------------------------

4

The European Journal of Public Health 1997 7(1):101-105; doi:10.1093/ 
eurpub/7.1.101  [ http://eurpub.oxfordjournals.org/cgi/content/abstract/7/1/101 
  ]

A graphical display useful for meta-analysis

F. JAVIER JIMÉNEZ1,2, ELISEO GUALLAR1, and JOSÉ M. MARTÍN-MORENO1,3

1 Department of Epidemiology and Biostatistks, National School of  
Public Health, ‘Carlos III’ Institute of Health Madrid, Spain

2 Department of Preventive Medicine, ‘San Carlos’ University Hospital  
Madrid, Spain

3 National Centre for Epidemiology, ‘Carlos III’ Institute of Health  
Madrid, Spain

Eliseo Guallar, MD, Dr PH, Departamento de Epidemiologla y  
Bioestadlstica, Escuela National de Sanidad, Sinesio Delgado 8, 28029  
Madrid, Spain, tel 34 1 3877873, fax 34 1 3877872

Graphical methods are frequently used in meta-analysis to summarize  
their results and to explore potential sources of heterogeneity across  
studies. In this paper, we illustrate a graphical method for meta- 
analysis of studies with dichotomous exposures and outcomes that  
complements other graphical and analytical approaches to meta- 
analysis. In prospective studies, the proportion of cases among the  
unexposed is plotted on the horizontal axis versus the proportion of  
cases among the exposed on the vertical axis. Contour lines for equal  
values of relative risk, odds ratio or risk difference and for the  
combined estimate of effect and its confidence interval are then  
superimposed on the graph. In case-control studies, the proportion of  
exposed controls is plotted on the horizontal axis versus the  
proportion of exposed cases on the vertical axis, although only the  
contour lines of equal odds ratios yield direct epidemiological  
interpretation. In these graphs, the distribution of the individual  
estimates of effect with respect to the contour lines offers a due as  
to the adequacy of the scale of measurement used (additive or  
multiplicative). This graphical method also permits direct inspection  
of the range of disease frequency in follow-up studies and of the  
range of exposure in case-control studies. Its use is illustrated with  
the aid of 3 examples derived from the literature.


Karen Schriver
KSA Communication Design & Research, Inc.
33 Potomac Street
Oakmont, PA 15139
412.828.8791
[email protected]

___________________________________________________________________

Use the following address to post a message to all subscribers: 
 [email protected]

To subscribe, unsubscribe or change your options, visit:
 http://list.InformationDesign.org/mailman/listinfo/infodesign-cafe

For all Information Design matters:
 http://InformationDesign.org

Problems? Write to:
 [email protected]
___________________________________________________________________