monocausal explanation of causes of death is appropriate
in HIV-positive persons because several cofactors may sub-
stantially contribute to death, including comorbidities,
polypharmacological treatment, substance use  and
socioeconomic factors [36–38]. Also, it is an immanent
difﬁculty of observational cohort studies to obtain infor-
mation on participants who were lost to follow-up. The
latter problem was not the focus of this study, but linkage
of the SHCS with the SNC revealed an additional 9.4%
probable deaths among SHCS participants in the drop-out
To improve the quality of classiﬁcation of causes of
death, we used the CoDe protocol, which includes system-
atic collection of information speciﬁcally to classify causes
of death in HIV-positive persons; a classiﬁcation system
adapted to morbidity in HIV-positive persons; and adjudi-
cation of diagnoses . While, for HIV-positive persons,
the ICD-10 coding system often tends to relate deaths to
HIV also in non-HIV-related causes, the CoDe system uses
coding according to organ system or aetiology, and differ-
entiates among immediate, contributing and underlying
causes of death .
In our study, discordance between the routinely collected
ICD-10 coded diagnoses in the SHCS and the adjudicated
data according to the CoDe protocol was 28%. Reasons for
discordance were not systematically studied, but may
include incorrect use of the ICD-10 coding, neglect of cause
and effect of events leading to death, incorrect association
of causes of death with HIV infection, less careful collection
of necessary information used for coding, and lack of
adjudication. The discordance of 40% between the SHCS
and the SNC data may partially be explained by ICD-10
coding which overemphasizes the contribution of HIV to
death, and by the possibility that some death certiﬁcates
may have been completed by physicians not specialized in
HIV medicine. Also, a limitation of the SNC is its reliance on
routine death certiﬁcates [19,43,44], which has been found
to be satisfactorily reliable for clearly deﬁned diagnoses
(e.g. malignancies and accidents) but less accurate for less
well-deﬁned conditions (e.g. chronic obstructive pulmo-
nary disease and diseases of the nervous system) [45–47].
Strengths of our study include the prospective collection
of incident events using structured reporting forms, classi-
ﬁcation of causes of death according to the systematic CoDe
protocol, distinction between immediate and underlying
causes of death, and central adjudication of endpoints.
However, several limitations should be noted. First, the
autopsy rate was low, and coding of deaths that occurred
outside a medical institution was often not possible. Sec-
ondly, permission for record linkage to the national death
registry in Switzerland is limited to make use of anonymized
data only. Thirdly, some patients with end-stage disease may
have been lost to follow-up in the cohort because terminally
ill patients may prefer other institutions than study centres,
or may have died outside of Switzerland.
In conclusion, our data have important methodological
and clinical implications. First, the distribution of causes of
deaths varied signiﬁcantly according to data source and
coding system . Therefore, future efforts should rein-
force the awareness among HIV care physicians that inves-
tigation and accurate documentation of deaths are
important. Methods developed to harmonize coding algo-
rithms and classiﬁcation systems must be made widely
available , and experience gained from their applica-
tion disseminated [17,32]. Future development should
ensure that methods remain sensitive enough to detect rare
events (e.g. ART-related portal hypertension without cir-
rhosis ). Secondly, overall mortality and AIDS-related
causes of death were decreasing, while malignancies, liver-
related morbidities, non-AIDS-related infections and car-
diovascular diseases became the major causes of death.
Many of these causes of death were associated with modi-
ﬁable risk factors which require increased attention in
primary and specialized care.
This study was ﬁnanced in the framework of the Swiss HIV
Cohort Study, supported by the Swiss National Science
Foundation. The members of the Swiss HIV Cohort Study
are: J. Barth, M. Battegay, E. Bernasconi, J. Böni, H. C.
Bucher, C. Burton-Jeangros, A. Calmy, M. Cavassini, C.
Cellerai, M. Egger, L. Elzi, J. Fehr, J. Fellay, M. Flepp, P.
Francioli (President of the SHCS), H. Furrer (Chairman of the
Clinical and Laboratory Committee), C. A. Fux, M. Gor-
gievski, H. Günthard (Chairman of the Scientiﬁc Board), D.
Haerry (deputy of ‘Positive Council’), B. Hasse, H. H. Hirsch,
B. Hirschel, I. Hösli, C. Kahlert, L. Kaiser, O. Keiser, C. Kind,
T. Klimkait, H. Kovari, B. Ledergerber, G. Martinetti, B.
Martinez de Tejada, K. Metzner, N. Müller, D. Nadal, G.
Pantaleo, A. Rauch, S. Regenass, M. Rickenbach (Head of
Data Center), C. Rudin (Chairman of the Mother & Child
Substudy), P. Schmid, D. Schultze, F. Schöni-Affolter, J.
Schüpbach, R. Speck, P. Taffé, P. Tarr, A. Telenti, A. Trkola,
P. Vernazza, R. Weber and S. Yerly (Version, July 2011).
The Swiss National Cohort (SNC) is funded by the Swiss
National Science Foundation. The SNC was made possible
by the support of the Swiss Federal Statistical Ofﬁce. We
thank Fabio Valeri for the probabilistic record linkage
between the SHCS and the SNC.
Conﬂicts of interest: RW has received travel grants or
unrestricted educational grants from Abbott, Boehringer
Ingelheim, Bristol-Myers Squibb, Gilead Sciences, Glaxo-
SmithKline, Merck Sharp & Dome, Pﬁzer, LaRoche, TRB
10 R Weber et al.
© 2012 British HIV Association HIV Medicine (2012)