Page 48 - Clinical Small Animal Internal Medicine

16 Section 1 Evaluation and Management of the Patient

the question: of what value is the incidence odds ratio if it interpretative latitude in extending such equivalence to
VetBooks.ir is interpretatively inferior to other ratio measures? incidence rate ratios [8].
The answer is that in case–control studies, where
study groups are selected based on whether or not they
possess (cases) or do not yet possess (controls) the out- Conclusion
come of interest, it is not possible to directly calculate
measures of incidence because the ratio of cases to con- This chapter addresses many, but certainly not all, of the
trols is determined by the study investigator. Without the most common statistical interpretive issues that face
ability to calculate individual conditional incidence readers of the veterinary medical literature. Appreciating
measures in subgroups, investigators are therefore simi- these issues is an important avenue to understanding the
larly unable to calculate cumulative incidence or inci- basis for evaluating clinical research articles. But statisti-
dence rate ratio measures of association that depend on cal analyses only represent one component of medical
knowing individual ones. Fortunately, case–control research; not addressed here are the myriad issues that
studies are often reserved for studying rare diseases, arise in the course of designing and implementing exper-
even conditional on the levels of other variables such as imental and nonexperimental studies. Because study
potential risk factors, so using the aforementioned rarity design is also inextricably linked to precision and validity
condition that interpretatively equilibrates the cumula- issues, statistical inference complements and cannot be
tive incidence and the incidence odds, the incidence divorced from causal inference in evidence‐based veteri-
odds ratio can essentially be understood as the cumula- nary medicine. Topics involving the design of studies to
tive incidence ratio. Subtle modifications in case–con- generate data for clinical decision making are the subject
trol study designs can also sometimes provide more of the following chapter.

References

1 Doll R, Bradford Hill A. The mortality of doctors in mast cell tumors: a validation study. Vet Pathol 2009;
relation to their smoking habits. 1954; 328(7455): 46(2): 362–4.
1451–5. 5 Kass PH, Romansik EM, Reilly CM, Moore PF, London
2 Terry L. Smoking and Health: Report of the Advisory CA. Letter to the Editor: Author Response. Vet Pathol
Committee to the Surgeon General of the United States. 2009; 46(2): 364–5.
Public Health Service Publication No. 1103. Washington, 6 Kass PH. Evidence‐based veterinary medicine:
DC: Department of Health, Education, and Welfare, evaluation of nonexperimental studies. Pulse 2014;
1964. 58(5): 16–17.
3 Romansik ER, Reilly CM, Kass PH, Moore PF, London 7 Buhles W, Kass PH. Understanding and evaluating
CA. Mitotic index is predictive for survival. Vet Pathol veterinary clinical research. J Am Anim Hosp Assoc 2012;
2007; 44(3): 335–41. 48(5): 285–98.
4 Elston LB, Sueiro FAR, Cavalcanti JN, Konradin M. 8 Weng HY, Messam LL. Making inferences from a
Letter to the editor: the importance of the mitotic index case‐control study: implications and sampling. Vet J
as a prognostic factor for survival of canine cutaneous 2012; 194(3): 282–7.