BMJ 2000;320:1468 ( 27 May )

Education and debate

Statistics Notes

The odds ratio

J Martin Bland, professor of medical statistics ^a,  Douglas G Altman, professor of statistics in medicine ^b. 

^a Department of Public Health Sciences, St George's Hospital Medical School, London SW17 0RE, ^b ICRF Medical Statistics Group, Centre for Statistics in Medicine, Institute of Health Sciences, Oxford OX3 7LF

Correspondence to: Professor Bland

In recent years odds ratios have become widely used in medical reportsalmost certainly some will appear in today's BMJ. There are three reasons for this. Firstly, they provide an estimate (with confidence interval) for the relationship between two binary ("yes or no") variables. Secondly, they enable us to examine the effects of other variables on that relationship, using logistic regression. Thirdly, they have a special and very convenient interpretation in case-control studies (dealt with in a future note).

The odds are a way of representing probability, especially familiar for betting. For example, the odds that a single throw of a die will produce a six are 1 to 5, or 1/5. The odds is the ratio of the probability that the event of interest occurs to the probability that it does not. This is often estimated by the ratio of the number of times that the event of interest occurs to the number of times that it does not. The table shows data from a cross sectional study showing the prevalence of hay fever and eczema in 11 year old children.¹ The probability that a child with eczema will also have hay fever is estimated by the proportion 141/561 (25.1%). The odds is estimated by 141/420. Similarly, for children without eczema the probability of having hay fever is estimated by 928/14 453 (6.4%) and the odds is 928/13 525. We can compare the groups in several ways: by the difference between the proportions, 141/561928/14 453=0.187 (or 18.7 percentage points); the ratio of the proportions, (141/561)/(928/14 453)=3.91 (also called the relative risk); or the odds ratio, (141/420)/(928/13 525)=4.89.

View this table: [in this window] [in a new window]   Association between hay fever and eczema in 11 year old children1

Now, suppose we look at the table the other way round, and ask what is the probability that a child with hay fever will also have eczema? The proportion is 141/1069 (13.2%) and the odds is 141/928. For a child without hay fever, the proportion with eczema is 420/13 945 (3.0%) and the odds is 420/13 525. Comparing the proportions this way, the difference is 141/1069420/13 945=0.102 (or 10.2 percentage points); the ratio (relative risk) is (141/1069)/(420/13 945)=4.38; and the odds ratio is (141/928)/(420/13 525)=4.89. The odds ratio is the same whichever way round we look at the table, but the difference and ratio of proportions are not. It is easy to see why this is. The two odds ratios are

which can both be rearranged to give

If we switch the order of the categories in the rows and the columns, we get the same odds ratio. If we switch the order for the rows only or for the columns only, we get the reciprocal of the odds ratio, 1/4.89=0.204. These properties make the odds ratio a useful indicator of the strength of the relationship.

The sample odds ratio is limited at the lower end, since it cannot be negative, but not at the upper end, and so has a skew distribution. The log odds ratio,² however, can take any value and has an approximately Normal distribution. It also has the useful property that if we reverse the order of the categories for one of the variables, we simply reverse the sign of the log odds ratio: log(4.89)=1.59, log(0.204)=1.59.

We can calculate a standard error for the log odds ratio and hence a confidence interval. The standard error of the log odds ratio is estimated simply by the square root of the sum of the reciprocals of the four frequencies. For the example,

A 95% confidence interval for the log odds ratio is obtained as 1.96 standard errors on either side of the estimate. For the example, the log odds ratio is log_e(4.89)=1.588 and the confidence interval is 1.588±1.96×0.103, which gives 1.386 to 1.790. We can antilog these limits to give a 95% confidence interval for the odds ratio itself,² as exp(1.386)=4.00 to exp(1.790)=5.99. The observed odds ratio, 4.89, is not in the centre of the confidence interval because of the asymmetrical nature of the odds ratio scale. For this reason, in graphs odds ratios are often plotted using a logarithmic scale. The odds ratio is 1 when there is no relationship. We can test the null hypothesis that the odds ratio is 1 by the usual ² test for a two by two table.

Despite their usefulness, odds ratios can cause difficulties in interpretation.³ We shall review this debate and also discuss odds ratios in logistic regression and case-control studies in future Statistics Notes.

Acknowledgments

We thank Barbara Butland for providing the data.

References

1.	Strachan DP, Butland BK, Anderson HR. Incidence and prognosis of asthma and wheezing illness from early childhood to age 33 in a national British cohort. BMJ 1996; 312: 1195-1199[Abstract/Free Full Text].
2.	Bland JM, Altman DG. Transforming data. BMJ 1996; 312: 770[Free Full Text].
3.	Sackett DL, Deeks JJ, Altman DG. Down with odds ratios! Evidence-Based Med 1996; 1: 164-166.

---

© BMJ 2000

This article has been cited by other articles:

• LaValley, M. P. (2008). Logistic Regression. Circulation 117: 2395-2399 [Full text]  
• Pasutto, F., Krumbiegel, M., Mardin, C. Y., Paoli, D., Lammer, R., Weber, B. H. F., Kruse, F. E., Schlotzer-Schrehardt, U., Reis, A. (2008). Association of LOXL1 Common Sequence Variants in German and Italian Patients with Pseudoexfoliation Syndrome and Pseudoexfoliation Glaucoma. IOVS 49: 1459-1463 [Abstract] [Full text]  
• Ke, S., Zhang, X. H.-F., Chasin, L. A. (2008). Positive selection acting on splicing motifs reflects compensatory evolution. Genome Res. 18: 533-543 [Abstract] [Full text]  
• Grimes, D. A., Schulz, K. F. (2008). Making Sense of Odds and Odds Ratios. Obstet Gynecol 111: 423-426 [Abstract] [Full text]  
• Gormley, F. J., MacRae, M., Forbes, K. J., Ogden, I. D., Dallas, J. F., Strachan, N. J. C. (2008). Has Retail Chicken Played a Role in the Decline of Human Campylobacteriosis?. Appl. Environ. Microbiol. 74: 383-390 [Abstract] [Full text]  
• Rowe, J. A., Handel, I. G., Thera, M. A., Deans, A.-M., Lyke, K. E., Kone, A., Diallo, D. A., Raza, A., Kai, O., Marsh, K., Plowe, C. V., Doumbo, O. K., Moulds, J. M. (2007). Blood group O protects against severe Plasmodium falciparum malaria through the mechanism of reduced rosetting. Proc. Natl. Acad. Sci. USA 104: 17471-17476 [Abstract] [Full text]  
• Remmers, E. F., Plenge, R. M., Lee, A. T., Graham, R. R., Hom, G., Behrens, T. W., de Bakker, P. I.W., Le, J. M., Lee, H.-S., Batliwalla, F., Li, W., Masters, S. L., Booty, M. G., Carulli, J. P., Padyukov, L., Alfredsson, L., Klareskog, L., Chen, W. V., Amos, C. I., Criswell, L. A., Seldin, M. F., Kastner, D. L., Gregersen, P. K. (2007). STAT4 and the Risk of Rheumatoid Arthritis and Systemic Lupus Erythematosus. NEJM 357: 977-986 [Abstract] [Full text]  
• Pandit, J. J., Satya-Krishna, R., Gration, P. (2007). Superficial or deep cervical plexus block for carotid endarterectomy: a systematic review of complications. Br J Anaesth 99: 159-169 [Abstract] [Full text]  
• Ni, X., Valente, J., Azevedo, M. H, Pato, M. T, Pato, C. N, Kennedy, J. L (2007). Connexin 50 gene on human chromosome 1q21 is associated with schizophrenia in matched case control and family-based studies. J. Med. Genet. 44: 532-536 [Abstract] [Full text]  
• Katz, K. A. (2006). The (relative) risks of using odds ratios.. Arch Dermatol 142: 761-764 [Abstract] [Full text]  
• Depp, C. A., Jeste, D. V. (2006). Definitions and Predictors of Successful Aging: A Comprehensive Review of Larger Quantitative Studies. AJGP 14: 6-20 [Abstract] [Full text]  
• Brett, P.M., Zygogianni, P., Griffiths, G.S., Tomaz, M., Parkar, M., D'Aiuto, F., Tonetti, M. (2005). Functional Gene Polymorphisms in Aggressive and Chronic Periodontitis. J. Dent. Res. 84: 1149-1153 [Abstract] [Full text]  
• Tanner, K., Sabrine, N., Wren, C. (2005). Cardiovascular Malformations Among Preterm Infants. Pediatrics 116: e833-e838 [Abstract] [Full text]  
• Kohlmeier, M., da Costa, K.-A., Fischer, L. M., Zeisel, S. H. (2005). Genetic variation of folate-mediated one-carbon transfer pathway predicts susceptibility to choline deficiency in humans. Proc. Natl. Acad. Sci. USA 102: 16025-16030 [Abstract] [Full text]  
• Knox, E G (2005). Oil combustion and childhood cancers. J. Epidemiol. Community Health 59: 755-760 [Abstract] [Full text]  
• Nelson, N. C., Evans, R. S., Samore, M. H., Gardner, R. M. (2005). Detection and Prevention of Medication Errors Using Real-Time Bedside Nurse Charting. J. Am. Med. Inform. Assoc. 12: 390-397 [Abstract] [Full text]  
• Peele, P. B., Siminoff, L. A., Xu, Y., Ravdin, P. M. (2005). Decreased Use of Adjuvant Breast Cancer Therapy in a Randomized Controlled Trial of a Decision Aid with Individualized Risk Information. Med Decis Making 25: 301-307 [Abstract]  
• Morgan, P B, Efron, N, Hill, E A, Raynor, M K, Whiting, M A, Tullo, A B (2005). Incidence of keratitis of varying severity among contact lens wearers. Br. J. Ophthalmol. 89: 430-436 [Abstract] [Full text]  
• Penarrubia, J., Fabregues, F., Manau, D., Creus, M., Carmona, F., Casamitjana, R., Vanrell, J. A., Balasch, J. (2005). Previous cycle cancellation due to poor follicular development as a predictor of ovarian response in cycles stimulated with gonadotrophin-releasing hormone agonist-gonadotrophin treatment. Hum Reprod 20: 622-628 [Abstract] [Full text]  
• Linke, R., Schroeder, M., Helmberger, T., Voltz, R. (2004). Antibody-positive paraneoplastic neurologic syndromes: Value of CT and PET for tumor diagnosis. Neurology 63: 282-286 [Abstract] [Full text]  
• Sette, A., Neder, J. A., Nery, L. E., Kavakama, J., Rodrigues, R. T., Terra-Filho, M., Guimaraes, S., Bagatin, E., Muller, N. (2004). Thin-Section CT Abnormalities and Pulmonary Gas Exchange Impairment in Workers Exposed to Asbestos. Radiology 232: 66-74 [Abstract] [Full text]  
• McMahon, B. T., West, S. L., Lewis, A. N., Armstrong, A. J., Conway, J. P. (2004). Hate Crimes and Disability in America. Rehabil Couns Bull 47: 66-75 [Abstract]  
• Leman, P, Kapadia, Y, Herington, J (2003). Randomised controlled trial of the onset of analgesic efficacy of dexketoprofen and diclofenac in lower limb injury. Emerg. Med. J. 20: 511-513 [Abstract] [Full text]  
• Wren, C, Birrell, G, Hawthorne, G (2003). Cardiovascular malformations in infants of diabetic mothers. Heart 89: 1217-1220 [Abstract] [Full text]  
• Raghunath, A., Hungin, A P. S, Wooff, D., Childs, S. (2003). Prevalence of Helicobacter pylori in patients with gastro-oesophageal reflux disease: systematic review. BMJ 326: 737-737 [Abstract] [Full text]  
• Hayes-Bautista, D. E., Hsu, P., Hayes-Bautista, M., Iniguez, D., Chamberlin, C. L., Rico, C., Solorio, R. (2002). An Anomaly Within the Latino Epidemiological Paradox: The Latino Adolescent Male Mortality Peak. Arch Pediatr Adolesc Med 156: 480-484 [Abstract] [Full text]  
• Leonard, H, Barrett, A M, Scott, J E S, Wren, C (2001). The influence of congenital heart disease on survival of infants with oesophageal atresia. Arch. Dis. Child. Fetal Neonatal Ed. 85: F204-206 [Abstract] [Full text]  

Rapid Responses:

Read all Rapid Responses

OR scope and constraints

J F Martin, et al.
bmj.com, 29 May 2000 [Full text]

Odds ratio and confidence intervals calculator

David J R Hutchon
bmj.com, 12 Jun 2000 [Full text]

Correction for Table: Hay Fever and Eczema

Deane L Burns
bmj.com, 2 Jun 2002 [Full text]