How to diagnose diabetes by smithhaleey


									the WHI (20 × 0.77% = 15.4%). We                               cine Teaching Tips Working Group. Tips for
                                                               learners of evidence-based medicine: 1. Relative
could use both the approaches described                        risk reduction, absolute risk reduction and num-
above. For the crude approach, the risk                        ber needed to treat. CMAJ 2004;171(4):353-8.
difference is now approximately 15.4%                     DOI:10.1503/cmaj.1041709
× 0.32 or 4.93% and the NNT 100/4.93
or just slightly above 20. Using the haz-
ard ratio approach for this patient also                  How to diagnose diabetes
yields an NNT of just over 20.
   As we have shown here, differences
between naïve approaches to calculat-
ing NNT based on event rates and
more sophisticated approaches based
                                                          I  n their commentary on the impact of
                                                             new guidelines for glucose tolerance
                                                          testing, Andrew Lyon and associates1
on survival analysis may not be large                     argue against increased use of the oral
enough to change clinical decisions.                      glucose tolerance test (OGTT) on the
We suggest that clinicians who are in-                    grounds of poor reproducibility, cum-
terested in using the NNT to help                         bersomeness and questionable cost-ef-
guide their practice should not be                        fectiveness. They rightly conclude that
overly concerned about inaccuracies                       devoting resources to programs that
that may arise from estimating the                        can help patients to modify their risk
NNT from event rates, especially when                     for diabetes is preferable to performing
using data from large, randomized trials                  more OGTTs. However, it would have
with high rates of follow-up. What they                   been appreciated if they had considered
must avoid is applying NNTs from trial                    the simplified or abbreviated version of
data without considering how their pa-                    the glucose tolerance test2,3 in terms of
tient’s baseline risk may differ from that                its suitability for detecting new cases of
of the patients in the trial. That mistake                diabetes; this form of the test would be
could lead to serious miscalculations of                  both reliable and less expensive.
the NNT that would have implications
for clinical decision-making.                             Gurusamy Sivagnanam
                                                          Asian Institute of Medicine, Science and
Alexandra L. Barratt                                       Technology
School of Public Health                                   Kedah, Malaysia
University of Sydney
Sydney, Australia                                         References
                                                          1.   Lyon AW, Larsen ET, Edwards AL. The impact
Peter C. Wyer                                                  of new guidelines for glucose tolerance testing
Columbia University College of                                 on clinical practice and laboratory services.
 Physicians and Surgeons                                       CMAJ 2004;171(9):1067-9.
                                                          2.   Posner NA, Silverstone FA, Breuer J, Heller M.
New York, NY                                                   Simplifying the intravenous glucose tolerance
Gordon Guyatt                                                  test. J Reprod Med 1982;27(10):633-8.
                                                          3.   Forrest RD, Jackson CA, Yudkin JS. The abbre-
Departments of Medicine and of Clinical                        viated glucose tolerance test in screening for dia-
 Epidemiology and Biostatistics                                betes: the Islington Diabetes Survey. Diabet Med
McMaster University                                            1988;5(6):557-61.
Hamilton, Ont.                                            DOI:10.1503/cmaj.1041707
Judy M. Simpson
School of Public Health
University of Sydney
Sydney, Australia                                         A    ndrew Lyon and associates1 point
                                                               out that the Canadian Diabetes As-
                                                          sociation’s new clinical guidelines may
                                                          increase the burden on laboratories be-
1.   Altman DG, Andersen PK. Calculating the              cause of increased use of the OGTT. I
     number needed to treat for trials where the out-     would like to add that the diagnosis of
     come is time to an event. BMJ 1999;319:1492-5.
2.   Writing Group for the Women’s Health Initia-         diabetes is mainly initiated by family
     tive Investigators. Risks and benefits of estrogen   doctors, but they may be too busy to
     plus progestin in healthy postmenopausal
     women. Principal results from the Women’s            implement any screening or to follow
     Health Initiative randomized controlled trial.       up appropriately once diabetes has been
     JAMA 2002;288:321-33.
3.   Barratt A, Wyer PC, Hatala R, McGinn T, Dans
                                                          identified.2 The burden on family prac-
     AL, Keitz S, et al; for the Evidence-Based Medi-     titioners to initiate mass testing and

manage patients afterward (with                              Hamman RF, Lachin JM, Walker EA, et al; Dia-            At the heart of the matter are issues
                                                             betes Prevention Program Research Group. Re-
lifestyle advice and oral therapy3,4) is po-                 duction in the incidence of type 2 diabetes with     critical to both patient care and clinical
tentially huge.                                              lifestyle and metformin. N Engl J Med 2002;346       research. Both of these activities are
    The prospect of targeted screening                  5.   Rolka DB, Narayan KM, Thompson TJ, Gold-             dependent upon public trust, which
(as supported by Lyon and associates)                        man D, Lindenmayer J, Alich K, et al. Perfor-        must be earned through openness and
                                                             mance of recommended screening tests for undi-
warrants consideration. Screening tools                      agnosed diabetes and dysglycemia. Diabetes Care      integrity.
with different predictive abilities (75%                     2001;24:1899-903.
                                                        6.   Lindstrom J, Tuomilehto J. The diabetes risk
to 80% sensitivity and 50% to 76%                            score: a practical tool to predict type 2 diabetes
                                                                                                                  Gerald A. Klassen
specificity5,6) are available, which could                   risk. Diabetes Care 2003;26(3):725-31.               Retired Physician
be used anywhere in the community.                      7.   Engelgau MM, Narayan KMV, Herman WH.                 Centreville, NS
                                                             Screening for type 2 diabetes. Diabetes Care
These tools take into account major                          2000;23(10):1563-80.
risk factors such as family history, exer-              8.   Peters AL, Davidson MB, Schriger DL, Hassel-         References
                                                             blad V. A clinical approach for the diagnosis of     1.   Bioethics when undertaking clinical trials. In:
cise levels, age, body mass index, waist                     diabetes mellitus: an analysis using glycosylated         Ethics of clinical trials for research ethics boards:
circumference, dietary habits, medica-                       hemoglobin levels. Meta-Analysis Research                 proceedings of a national workshop. NCBHR
                                                             Group on the Diagnosis of Diabetes Using Gly-             Communiqué 1991;2(2):19-23.
tion history and history of dysglycemia;                     cated Hemoglobin Levels [published erratum           2.   Ferris LE, Naylor CD. Physician remuneration
however, they perform poorly as stand-                       appears in JAMA 1997;277:1125]. JAMA 1996;                in inudstry-sponsored clinical trials: the case for
                                                             276:1246-52.                                              standardized clinical trial budgets [editorial].
alone tests.7                                           9.   Kaczorowski J, Karwalajtys T, Chambers L,                 CMAJ 2004;171(8):883-6.
    A 2- or 3-stage screening test (e.g.,                    Levitt C, Dolovich L, McDonough B, et al.            3.   Yusuf S. Randomized clinical trials: Slow death
                                                             Community strategies to monitor high blood                by a thousand unnecessary policies? [editorial].
the combination of a questionnaire and                       pressure among older adults [abstract]. Phar-             CMAJ 2004;171(8):889-92.
random capillary blood glucose testing,                      macy Practice Research Symposium; 2002 Feb 1;        4.   Ferris LE, Naylor CD. Rebuttal [editorial].
which yields 58% sensitivity and 94%                         Toronto.                                                  CMAJ 2004;171(8):892-3.
                                                                                                                  5.   Tri-Council policy statement: ethical conduct for re-
specificity8) might be a more efficient                 DOI:10.1503/cmaj.1041724                                       search involving humans. Ottawa: Medical Re-
use of resources, ensuring that OGTTs                                                                                  search Council of Canada; Natural Sciences and
                                                                                                                       Engineering Research Council of Canada; Social
are not performed unnecessarily. Other                                                                                 Sciences and Humanities Research Council;
combinations of near-patient tests and                  Clinical trial budgets                                         1998. Available:
                                                                                                                       /code_2/ (accessed 2005 Jan 5).
scoring tools that might be used in
community settings should be studied,                                                                             Competing interests: Dr. Klassen is a past president of
                                                                                                                  Canadian Society for Clinical Investigation and Na-
similar to the successful assessment in
local pharmacies of people at risk of
hypertension. 9 It would be entirely
                                                        I  n May 1991, Ian Rusted chaired a 2-
                                                           day workshop sponsored by the Na-
                                                        tional Council on Bioethics in Human
                                                                                                                  tional Council on Ethics in Human Research.


possible, using a mixture of commu-                     Research (now the National Council on
nity-based measurements such as scor-
ing tools for diabetes risk, fasting capil-
lary blood glucose readings and
                                                        Ethics in Human Research) entitled
                                                        “Ethics of Clinical Trials for Research
                                                        Ethics Boards.”1 The participants were
                                                                                                                  W        e commend Salim Yusuf for his
                                                                                                                           reply 1 to the call by Lorraine
                                                                                                                  Ferris and David Naylor2 for additional
near-patient testing of hemoglobin A1c                  representatives of the pharmaceutical                     monitoring of clinical trials. Yusuf’s
to target individuals who should un-                    industry, the Medical Research Council                    point on the increasing complexity of
dergo an OGTT. This might reduce                        of Canada, Health and Welfare Canada                      regulation for clinical trial research is
the potential burden on both laborato-                  and the Royal College of Physicians                       well taken, as is the point that comply-
ries and family physicians.                             and Surgeons of Canada, as well as                        ing with complex regulations creates
                                                        members of research ethics boards                         significant costs. Although one of us
Gina Agarwal                                            from across Canada.                                       (J.A.C.D.) has previously argued against
Assistant Professor                                        On reading the viewpoint by Lor-                       an excessive reliance on clinical trials,3 it
Department of Family Medicine                           raine Ferris and David Naylor, 2 the                      is clear that they represent the modern
McMaster University                                     spirited response by Salim Yusuf 3 and                    gold standard. Given this reality, it is
Hamilton, Ont.                                          the rebuttal by Ferris and Naylor,4 I ex-                 essential that we not choke off this im-
                                                        perienced a sense of déjà vu: the points                  portant type of research.
References                                              of view expressed in this exchange mir-                       Increasing costs through the re-
1.    Lyon AW, Larsen ET, Edwards AL. The impact        ror the conclusions of the 1991 work-                     quirement to deal with nontransparent
      of new guidelines for glucose tolerance testing
      on clinical practice and laboratory services.     shop. Unfortunately, although the Tri-                    and complex regulations actually makes
      CMAJ 2004;171(9):1067-9.                          Council drafting committee had access                     it harder for independent researchers to
2.    Harris MI, Klein R, Wellborn TA, Knuiman
      MW. Onset of NIDDM occurs at least 4–7            to the workshop recommendations for                       do research. We have recently seen the
      years before clinical diagnosis. Diabetes Care    financial accountability and conflict of                  consequences of restricting clinical tri-
3.    Chiasson JL, Josse RG, Gomis R, Hanefeld M,       interest, they were not incorporated in                   als to large drug companies4 rather than
      Karasik A, Laakso M. Acarbose for prevention of   the Tri-Council policy statement.5 The                    independent academic investigators. It
      type 2 diabetes mellitus: the STOP-NIDDM
      randomised trial. Lancet 2002;359(9323):2072-7.
                                                        authors and CMAJ are to be com-                           would seem more appropriate to have
4.    Knowler WC, Barrett-Connor E, Fowler SE,          mended for revisiting the subject.                        well-trained auditors who could iden-

616                                            JAMC • 1er MARS 2005; 172 (5)

To top