Published Online, 31 March 2009, www.theannals.com, DOI 10.1345/aph.1K618a.
The Annals of Pharmacotherapy: Vol. 43, No. 4, pp. 793-794. DOI 10.1345/aph.1K618a
© 2009 Harvey Whitney Books Company.
Comment: Sensitivity of Patient Outcomes to Pharmacist Interventions. Part III: Systematic Review and Meta-Analysis in Hyperlipidemia Management
Theresa L Charrois, BScPharm MSc
Research Associate, Department of Medicine, University of Alberta
EPICORE Centre, Edmonton, Alberta, Canada
Ross T Tsuyuki, BScPharm PharmD MSc
Professor of Medicine, Cardiology, EPICORE Centre/COMPRIS, University
of Alberta, Suite 220 College Plaza, 8215-112 Street, Edmonton, Alberta, T6G
2C8 Canada, fax 780/492-6059,
ross.tsuyuki{at}ualberta.ca
Published Online, March 31, 2009. www.theannals.com, DOI 10.1345/aph.1K618a
TO THE EDITOR: We read with great interest the recently published paper by
Machado et
al.1 We
appreciate that pharmacy practice research has many challenges that, by
extension, may pose further challenges in conducting systematic reviews of
this
research.2
Typically, pharmacy practice research varies widely in design, quality, and
outcomes. In the Machado paper, this variability is evident even from a visual
inspection of Figures 2, 3, and 5. Given the wide range of outcomes presented,
we were surprised that the authors reported no heterogeneity among the studies
they combined.
Using the data that the authors provided in the article and a random
effects model, we recalculated the heterogeneity using Review Manager, version
5.0 (Cochrane Collaboration, Copenhagen). Our results are very different from
those of the authors. For the outcome of total cholesterol levels, the authors
reported a heterogeneity of p = 0.962. Our results show the heterogeneity as
an I2 of 52%
(moderate heterogeneity; p = 0.02). Similarly, while the authors reported
heterogeneity as nonsignificant for low-density lipoprotein cholesterol
(LDL-C) and triglycerides, we calculated heterogeneity as an
I2 of 84% (a
high degree of heterogeneity; p < 0.001), and 27% (moderate heterogeneity;
p = 0.21), respectively. These findings may alter the conclusions of the
research report, as it is considered inappropriate to combine studies with
significant heterogeneity.
We believe that systematic reviews can be more easily interpreted when
heterogeneity is presented in terms of both the Q-statistic and the
I2
value.3 The 2
previous articles in this series also reported heterogeneity with only the
Q-statistic.4,5
In light of the heterogeneity observed, we think that the results of the
Machado articles should be interpreted with caution. Further analyses to help
determine the sources of heterogeneity in the results would seem
warranted.
References
- Machado M, Nassor N, Bajcar JM, Guzzo GC, Einarson TR. Sensitivity
of patient outcomes to pharmacist interventions. Part III: systematic review
and meta-analysis in hyperlipidemia management. Ann
Pharmacother 2008;42:1195-207. Epub 5 Aug 2008. DOI 10.1345/aph.1K618[Abstract/Free Full Text]
- Charrois TL, Durec T, Tsuyuki RT. Systematic reviews of pharmacy
practice research: methodologic issues in searching, evaluating, interpreting,
and disseminating results. Ann Pharmacother 2009;43:118-22.
Epub 2 Dec 2008. DOI 10.1345/aph.1L302[Abstract/Free Full Text]
- Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring
inconsistency in meta-analyses. BMJ 2003;327:557-60.[Free Full Text]
- Machado M, Bajcar JM, Guzzo GC, Einarson TR. Sensitivity of patient
outcomes to pharmacist interventions. Part II: systematic review and
meta-analysis in hypertension management. Ann Pharmacother 2007;41:1770-81.
Epub 9 Oct 2007. DOI 10.1345/aph.1K311[Abstract/Free Full Text]
- Machado M, Bajcar JM, Guzzo GC, Einarson TR. Sensitivity of patient
outcomes to pharmacist interventions. Part I: systematic review and
metaanalysis in diabetes management. Ann Pharmacother 2007;41:1569-82.
Epub 21 Aug 2007. DOI 10.1345/aph.1K151[Abstract/Free Full Text]
