|
 |
|
|||||||
|
|||||||||
Electronic Letters to:
|
|
eLetters: Electronic letters published:
-
![[Read eLetter]](/icons/misc/sectionDOWN.gif)
Smoking in AIAN people with diabetes revisited: a lesson in ascertainment bias
- Kelly Acton MD MPH, Ann Bullock, MD (29 April 2008)
|
|
|||
|
Kelly Acton MD MPH, physician / Director IHS Division of Diabetes Treatment and Prevention Indian Health Service, Ann Bullock, MD
Send letter to journal:
kelly.acton{at}ihs.gov Kelly Acton MD MPH, et al.
|
April 20, 2008 Dear Editor, We read this month’s AJPH article entitled Current Smoking and Type 2 Diabetes Among Patients in Selected Indian Health Service Clinics, 1998–2003 (1) with great interest. The Indian Health Service Division of Diabetes has been measuring smoking status yearly in American Indians and Alaska Natives (AIAN) with diabetes since 1986 as part of our Annual Diabetes Care and Outcomes Audit (2). The IHS Standards of Care for Diabetes require smoking assessment as part of our annual diabetes care exam. Last year over 312 facilities participated in the audit of 54,415 charts, representing care to nearly 124,000 AIAN with diabetes. In 2007 91.8% of AIAN with diagnosed diabetes were assessed for tobacco use, compared with 66.0% in 1998, and 23.4% reported current tobacco use. Of these, 31.6% had been referred to counseling for tobacco cessation, as compared with 26.2% in 1998 (p < 0.01). Because no such organized emphasis has been applied in the IHS for AIAN without diabetes, the findings presented in Morton, et al’s paper are likely flawed from ascertainment bias. Two well-designed epidemiologic studies of cardiovascular risk factors have not found increased rates of smoking in American Indian patients with diabetes compared to those without diabetes. The Strong Heart Study, a population based sample from 13 American Indian communities in Arizona, Oklahoma and the Dakotas, found that current tobacco use rates in AIAN participants with diabetes were 27% compared to 41% in non- diabetic participants (p <0.001 for the difference between diabetic and non-diabetic participants). (3) The Inter Tribal Heart Project found that among 1376 Chippewa and Menominee Indians age 25 years and older in 1992- 94, current tobacco use was reported by 53.1% of participants with diabetes (diagnosed or undiagnosed), compared with 57.8% with pre-diabetes and 68.9% among those without diabetes or pre-diabetes. (4) In addition, population based telephone surveys of smoking and diabetes in AI have not found rates of smoking to be higher in AI with diabetes compared to those with no history of diabetes. CDC’s Behavioral Risk Factor Surveillance System (BRFSS) data for AIAN were examined for two time periods: 2001-2003 and 2005-2006. There were no significant differences in current smoking prevalence between adult AIAN with diabetes and those without, both in 2001-2003 (34.2% without diabetes vs 26.9% with diabetes) and in 2005-2006 (32.2% without diabetes vs 29.6% with diabetes). (5) The authors used an electronic record set to identify cases with diabetes, and used a criterion defined as “a diagnosis of type 2 diabetes at any time during the 5-year period. A diagnosis was counted if it was for any of the 10 levels of International Classification of Diseases, Ninth Revision, Clinical Modification–coded diagnoses (code=250)” that has high sensitivity but less desirable specificity (1). This may have included among the cases for the study some persons who did not have diabetes. These misclassified cases would not have been included in the diabetes population for which smoking assessment and referral is emphasized at IHS sites, but would have been counted as diabetes cases for the study. This type of error is particularly problematic in Indian health facilities because of the high frequency with which persons are screened in many Indian communities due to the high prevalence of diabetes. (6) Our program has conducted numerous chart reviews over the years on AIAN listed as having diabetes in the system, only to discover that many are coded for diabetes when they have been screened for diabetes. Thus, more valid electronic audit criteria for identifying cases of diabetes require at least 2 visits with a diabetes purpose-of-visit in a one year period, or all active clinical patients with a 250-250.93 diagnostic code who have had the diagnosis for at least one year, or a visit to pharmacy for diabetes medications. Since the findings presented by Morton et al are not consistent with a variety of other published studies, and are very likely invalid due to the biases described above, we urge readers to interpret the findings with great caution. Smoking is a very important health issue, for persons with and without diabetes. Resources should be devoted to identifying smokers and helping them quit. It would be desirable to have more resources available for all IHS programs that identify and help smokers. Before the distribution of currently limited IHS smoking cessation resources is modified, however, the evidence for redistributing those resources should be more substantial than that presented by Morton et al. Kelly Acton, MD, MPH, FACP Ann Bullock, MD 1. Morton D, Garrett M, Reid J, Wingard DL.. Am J Public Health. 2008 Mar;98(3):560. 2. Mayfield JA et al. Diabetes Care. 1994 Aug;17(8):918-23. 3. Personal communication Zhang Y and Lee E, Strong Heart Study, (unpublished data) 4. Personal communication, Burrows N and Geiss L, Center for Disease Control and Prevention, Atlanta, GA., 1996 5. CDC. Behavioral Risk Factor Survey, 2001-2006. Available online at http://www.cdc.gov/brfss/ 6. Harwell et al J Public Health Management and Practice 2005;11:537-541. 33 |
|||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH |