Secular decline in mortality from coronary heart disease in adults with diabetes mellitus: cohort studyBMJ 2008; 337 doi: https://doi.org/10.1136/bmj.39582.447998.BE (Published 01 July 2008) Cite this as: BMJ 2008;337:a236
All rapid responses
The authors reply to the rapid response: "There seems to be apparent computation errors in the paper. They should be clarified or corrected."
Based on the number of participants reported in other publications
from the HUNT Study, Kumarasena (1) has calculated the number of person
years that may be obtained during 9.5 years of follow up, as in our recent
paper in BMJ (2). She claims that the person years that we report for the
age group 80 years and older must be incorrect (i.e. the number given in
the paper is 21480 person years for women without diabetes, and Kumarasena
suggests that this number should not exceed 18810 person years).
In the analysis mortality rates were calculated using the stsplit
command in Stata. This procedure expands the individual survival record
into information on time at risk for an event to occur for different
categories of time (in the present study, that means age categories;
younger than 60 years, 60-69, 70-79, and 80 years and older). As a
consequence, the number of person years for individuals in a certain age
group at baseline cannot be simply calculated as the sum of the
observation time for people within that category. For example, consider a
woman who was 77 years at baseline, and seven years later, at the age of
84, died from coronary heart disease. In the analysis, this woman will
contribute three years of person time to the denominator for the age-group
70-79 years, and four years for the age group 80 years and older. The
event (i.e. her death from coronary heart disease) will be counted in the
numerator of the age group 80 years and older, and not in the age group 70
-79 years, although she belonged to the latter category at baseline.
This approach is widely regarded as an appropriate way to handle time
at risk data, especially for long follow-up periods, and we hope this
clarification is satisfactory.
Ane Cecilie Dale, Lars J Vatten, Tom Ivar Nilsen, Kristian Midthjell,
(1) Arya K Kumarasena. There seems to be apparent computation errors
in the paper. They should be clarified or corrected. Published 16 July
(2) Dale AC, Vatten LJ, Nilsen TI, Midthjell K, Wiseth R. Secular
decline in mortality from coronary heart disease in adults with diabetes
mellitus: cohort study. BMJ 2008 Jul 1;337:a236. doi:
Competing interests: No competing interests
The paper provides the person years of males and females with and
without diabetes for both surveys HUNT1 and HUNT2 . Highest number of
deaths and morality rates are associated with 80 + person groups.
Research publications  and  provides further details on HUNT
surveys .According to them there were 2143 females in 80+ group in HUNT2
database(1906:80-89group, 237:90+group). Out of them 263 persons were
diabetic patients (12.83%:80-89 group, 12.24%:90+group).Therefore there
should be 1980 others.
As the HUNT2 survey started in August 1995 and person year
computation was based on the 31st December 2004 survivors, person years
contribution from each and every person will be less than 9.5 years. (Late
recruits and deceased will have much shorter periods).
Therefore total number of person years of this category cannot exceed
18810 years (1980*9.5). How ever the figure indicated in the paper is
21480 years, which shows an apparent computation error? Similar scenario
exists for all groups of 80+ non diabetic persons.
As the increase in person years will directly reduce the morality
rate, apparent error will effect the final conclusion. Therefore
clarification or correction seems to be necessary.
Indication of Mortality rate (95%CI) derived from only one death and
application of statistical techniques to correlate mortality rates of two
groups having around 18 years difference in mean age values may also
provide results with very little meaning.
 Published 1 July 2008, doi:10.1136/bmj.39582.447998.BE
Secular decline in mortality from coronary heart disease in adults with
diabetes mellitus: cohort study
Ane Cecilie Dale, research fellow, medical doctor, Lars J Vatten,
professor, Tom Ivar Nilsen, associate professor, Kristian Midthjell,
associate professor, Rune Wiseth, professor
 K Midthjell, O Kruger, J Holmen, A Tverdal, T Claudi, A Bjorndal,
and P Magnus
Rapid changes in the prevalence of obesity and known diabetes in an
adult Norwegian population. The Nord-Trondelag Health Surveys: 1984-1986
Diabetes Care 22: 1813-1820.
 Holmen J, Midthjell K, Krüger Ø, Langhammer A, Holmen TL,
Bratberg GH, et al. The Nord-Trøndelag health study 1995-97 (HUNT 2):
objectives, contents, methods and participation. Nor J Epidemiol
Competing interests: No competing interests