Need for adjusted odds ratios for LLIN coverage/use in multivariate analysis
Thomas Kesteman, Institut Pasteur de Madagascar
2 September 2014
We read with great interest your article. We were nevertheless surprised not to find a variable for Long Lasting Insecticidal Nets (LLIN) ownership or use in the multivariate analysis (in Table 1), since both have been previously shown to offer a protective effect against incidence of clinical malaria [1–3]. It is recommended by most authors to force variables known as determinants in multivariate models even if no significant univariate association is found [4]. Could you provide the two revised versions of the multivariate model displaying the adjusted odds ratios of LLIN ownership or use?
Best regards,
Thomas KESTEMAN1 and Christophe ROGIER2
1 MD, MPH - Malaria Research Unit, Institut Pasteur de Madagascar
2 MD, MPH, PhD - Institut Pasteur de Madagascar
References
1. Lengeler C: Insecticide-treated bed nets and curtains for preventing malaria. Cochrane Database Syst Rev 2004:CD000363.
2. Mathanga DP, Campbell CH, Taylor TE, Barlow R, Wilson ML: Reduction of childhood malaria by social marketing of insecticide-treated nets: a case-control study of effectiveness in Malawi. Am J Trop Med Hyg 2005, 73:622–625.
3. Kamya MR, Gasasira AF, Achan J, Mebrahtu T, Ruel T, Kekitiinwa A, Charlebois ED, Rosenthal PJ, Havlir D, Dorsey G: Effects of trimethoprim-sulfamethoxazole and insecticide-treated bednets on malaria among HIV-infected Ugandan children. AIDS 2007, 21:2059–66.
4. Greenland S: Modeling and Variable Selection in Epidemiologic Analysis. Am J Public Health 1989, 79:340–349.
Need for adjusted odds ratios for LLIN coverage/use in multivariate analysis
2 September 2014
We read with great interest your article. We were nevertheless surprised not to find a variable for Long Lasting Insecticidal Nets (LLIN) ownership or use in the multivariate analysis (in Table 1), since both have been previously shown to offer a protective effect against incidence of clinical malaria [1–3]. It is recommended by most authors to force variables known as determinants in multivariate models even if no significant univariate association is found [4]. Could you provide the two revised versions of the multivariate model displaying the adjusted odds ratios of LLIN ownership or use?
Best regards,
Thomas KESTEMAN1 and Christophe ROGIER2
1 MD, MPH - Malaria Research Unit, Institut Pasteur de Madagascar
2 MD, MPH, PhD - Institut Pasteur de Madagascar
References
1. Lengeler C: Insecticide-treated bed nets and curtains for preventing malaria. Cochrane Database Syst Rev 2004:CD000363.
2. Mathanga DP, Campbell CH, Taylor TE, Barlow R, Wilson ML: Reduction of childhood malaria by social marketing of insecticide-treated nets: a case-control study of effectiveness in Malawi. Am J Trop Med Hyg 2005, 73:622–625.
3. Kamya MR, Gasasira AF, Achan J, Mebrahtu T, Ruel T, Kekitiinwa A, Charlebois ED, Rosenthal PJ, Havlir D, Dorsey G: Effects of trimethoprim-sulfamethoxazole and insecticide-treated bednets on malaria among HIV-infected Ugandan children. AIDS 2007, 21:2059–66.
4. Greenland S: Modeling and Variable Selection in Epidemiologic Analysis. Am J Public Health 1989, 79:340–349.
Competing interests
None.