What are internists looking for?
JGIM LETTERS TO THE EDITOR Comorbidity and Length of Stay: Statistically Significant? To the Editor:--Matsni et al.1 report that comorbidity as mea- sured by the Char lson index was associated with a statistically signif icant increase in length of stay for pat ients hospital ized with chest pain. The part ia l R 2 for comorbidity in their mult ivar iate model suggests that about 0.3% of the total var iat ion in length of stay was independent ly explained by comorbidity. Moreover, of the var iat ion explained, comorbidity accounted for about 0.7% (.0034/.5103, from Table 4). Adjust ing for mult ip le compar isons (at least 18 in the mult ivar iate analysis), I would guess that the overall importance of comorbidity may not be stat ist ical ly signifi- cant. In fact, what is surpr is ing about these resu l ts is that comor- bidity independent ly explained such a smal l amount of the varia- t ion in length of stay. It seems that phys ic ians general ly ignored compl icat ing noncardiac medical factors in treat ing pat ients with chest pain. Whether this is good or bad is unclear. All in all, the data suggest that, for chest pain hospital izat ions, collecting infor- mat ion on comorbidity provides very little useful information for health services researchers or resource uti l ization managers . - - DAI~II~L J . CHlgR, MD, Palo Alto VA Health Care System, Calif. REFERENCES 1. Matsui K, Goldman L, Johnson P, Kuntz K, Cook E, Lee T. Comor- bidity as a correlate of length of stay for hospitalized patients with acute chest pain. J Gen Intern Med. 1996; 11:262-8. In reply:--Dr. Cher has apparent ly misread Table 4 and cited the part ial R 2 for "Rales on init ial examinat ion"- - the line below the information on Char lson index scores. Nevertheless, his ob- servation that comorbidity score seems to explain only a smal l percentage of the variabi l i ty in length of stay in this populat ion is worth addressing, since the correct part ial R 2 was .005. In the overall populat ion of pat ients with chest pain, the major "drivers" of length of stay were, not surpris ingly, the diagnosis of acute my- ocardial infarction (R 2 - . 12), performance of coronary artery by pass graft surgery (R 2 - . 19), and the occurrence of compl icat ions such as congestive heart failure (R 2 - .05) and recurrent ischemic pain (R 2 = .06}. Among uncompl icated patients, however, comor- bidity score accounted for about as much of the variabi l i ty in length of stay as a prior history of acute myocardial infarction, and a Char lson score of four or more was associated with an ad- jus ted increase in length of stay of 41% compared to pat ients with scores of 0 to 1. For example, this increase in comorbidity score would be associated with an increase in length of stay of 1.8 days compared to the mean length of stay of 4.5 days seen among pa t ients with Char lson scores of 0 or 1. We would note that our data also demonstrate a clear "dose-response" effect between comor- bidity score and length of stay. Therefore, we conclude that co- morbidity as measured with Char lson index can improve analy- ses of length of stay for pat ients with acute chest pain. Whether other indices of comorbid condit ions might provide even greater information remains to be determined. - -THoM~ H. LEE, MID. Partners Community HealthCare, Inc., Boston, Mass. 500 What are Internists Looking For? To the Editor;---Clinical practice guidel ines are proliferating. Although some phys ic ians worry about their impact on cl inical autonomy and satisfact ion with clinical practice, most physic ians appreciate their potential benefit. Only recently have investiga- tors begun to analyze what specific features make for effective practice guidelines. In their article, Hayward et al. nicely identify some of those features such as the importance of concise recom- mendat ions, the inclusion of a synopsis of support ing evidence and quanti f icat ion of benefit.1 One area that was not addressed in their s tudy is the impor- tance of adequate d isseminat ion of information. Practice guide- l ines can be of great benefit to cl inicians. They also imply a stan- dard of care. However, guidel ines are not helpful if phys ic ians do not know of their existence. Frequently, this is due to numerous societies publ ish ing guidel ines in their own subspecia l ty jour- nals, many or most of which general internists do not read on a regular basis. A Medline search of practice guidel ines for the year 1995 yielded 38 art ic les of potential importance to general inter- nists. These 38 art icles were distr ibuted among thirty-one differ- ent journals. Thirty three (87%) of the practice guidel ines were publ ished in subspecia l ty journals . The art icles cover important issues such as the management of diabetes, 2 osteoarthrit is, 3 and diverticulit is. 4 To address this problem, 1 would like to suggest that a widely-circulated journa l for general internists, such as JGIM, regularly publ ish an annotated bibl iography of cl inical practice guidelines. As more practice guidel ines are publ ished and their importayme grows, we cannot say that ignorance is b l i ss . - -B~ M. /kBOFF, MD, Medical Center of Delaware, Newark. REFERENCES 1. Hayward RS, Wilson MC, Tunis SR, Guyatt GH, Moore K, Bass EB. Practice guidelines, What are internists looking for? J Gen In- tern Med. 1996:11 : 176-8. 2. American Diabetes Association. Clinical practice recommenda- tions 1995. Diabetes Care. 1995;18(Suppl}: 1-96. 3. Hochberg MC, Altman RD. Brandt KD, et al. Guidelines for the medical management of osteoarthritis. Part 1. Osteoarthritis of the hip. Arthritis Rheum. 1995:38:1535-40. 4. Roberts P. Abel M, Rosen L. et al. Practice parameters for sigmoid diverticulitis. The Standards Task Force Amei-ican Society of Co- lon and Rectal Surgeons. Dis Colon Rectum. 1995;38(2]: 125-32. In reply:--The authors agree entirely with Dr. AbofFs obser- vation that general internists could benefit from a widely avail- able annotated bibl iography of clinical practice guidelines. In- deed, the Guidel ine Appraisal Project, which conducted the Practice Guidel ines survey upon which our article is based, has more recently init iated an effort to bui ld an inventory of evidence- based practice guidel ines together with s t ructured abstracts of as many as possible. This eflbrt is supported by a grant from the Na- tional Library of Medicine and is now gett ing underway. Informa- tion about the project and a rapidly growing inventory can be found on Internet at ht tp: / /h i ru .mcmaster .ca/CPG.- -ROBERT I - InYw~, MD, McMaster University, Hamilton. Ontario, Canada.