Anthropology & Medicine

Vol. 12, No. 2, August 2005, pp. 129–134

Rejoinder: Genetic Research into the

Causes of Type 2 Diabetes Mellitus
Morris J. Birnbaum

Type 2 diabetes mellitus is a disease of substantial health and economic impact in

Western societies. Its prevalence has drastically increased in the latter half of the 20th
century, largely due to the ready availability of large quantities of calorie-rich foods and
the technology-driven reduction in routine, daily exercise. Yet basic scientists repeatedly
emphasize the prominent genetic basis for Type 2 diabetes mellitus, even though its
widespread occurrence has appeared at a rate too fast to be accounted for by genetic
change. This phenomenon emphasizes how the human gene pool has evolved as an
adaptation to an ancient societal condition and when the environment changes rapidly
both it and our genetic structure can contribute to the development of disease.

In the latter half of the 20th century, diabetes mellitus emerged as one of the most
important diseases in Western societies, both in terms of its adverse effect on the
health and lifestyles of vast numbers of people, as well as its economic impact
(Centers for Disease Control and Prevention 2004). More recently, there has been
a well-documented, startling increase in the rate of new diabetes in Asian and third
world countries. For this reason, the biomedical community is now initiating
a belated, determined effort to understand in some detail the root causes of diabetes
mellitus with the aim of developing novel therapeutics and preventative strategies
(Diabetes Research Working Group 1999).
Diabetes mellitus, which is defined by an inappropriate elevation in the levels
of the circulating blood sugar, has been divided into two primary forms. Type 1
diabetes mellitus, formerly known as juvenile onset disease, is a condition in which
the immune system mistakenly destroys the insulin-producing beta cells of the
pancreas, thus leading to an absolute deficiency of the hormone. Insulin is the
primary message that controls the body’s response to nutritional abundance, and one
of its major roles is to instruct specialized organs like muscle and fat cells to take
up the surplus glucose and turn it into long-term energy stores. For this reason,

Correspondence to: Morris Birnbaum, 415 Curie Boulevard, Clinical Research Building 322, Philadelphia,
PA 19104, USA. Tel.: þ1 215 898 5001; Fax: þ1 215 573 9138; Email: [email protected]
ISSN 1364-8470 (print)/ISSN 1469-2910 (online) ß 2005 Taylor & Francis Group Ltd
DOI: 10.1080/13648470500139908
130 M. J. Birnbaum
a deficiency of insulin leads to impairment of glucose uptake into these tissues while
blood glucose rises inappropriately and dangerously. Type 2 diabetes mellitus, which
was formerly known as adult onset diabetes, accounts for greater than 90% of the
cases diagnosed in the United States and its proportion is rising steadily, particularly
in childhood and adolescent populations. This disease is quite complex and its
etiology is imperfectly understood. However, it is quite clear that Type 2 diabetes
mellitus, at least in part, involves a major inability of insulin to carry out its actions
at target tissues as well as a relative or absolute deficiency in the production of the
hormone. As is true of most complex, common human disease, the study of diabetes
mellitus requires a concerted effort that includes observation and analysis of human
behavior and physiology as well as well-controlled investigation using experimental
in vivo and in vitro model systems. Given the current importance and prevalence of
diabetes research, a thoughtful analysis from a novel, anthropological perspective is
welcome. It is an axiom of research that the answers we find will be dictated by the
questions we ask, and, therefore, any serious analysis that challenges our assumptions
and biases advances the quality and impact of the research.
Nonetheless, there are several apparent assumptions in Dr. Rock’s paper, all
derived from the postulate that my presentation in at the Canadian Diabetes
Association Professional Conference represented a consensus view of the medical
community, that I feel are not well founded. First, and perhaps most importantly, I
was specifically asked to talk about the genetics of diabetes mellitus and it was to this
subject that I limited my comments. Had I been requested to discuss the origins of
diabetes mellitus, I would have delivered a much different address, including other
biological components as well as social, economic and environmental factors. There
is, I think, unanimous agreement in the diabetes research community that
environmental and behavioral factors contribute enormously to the development
of Type 2 diabetes mellitus (Sorensen 2000; Fisher et al. 2002). Indeed, the growing
prevalence of diabetes is occurring at a speed much too fast to be accounted for by
genetic changes. Thus, how else can we account for such a dramatic increase in the
incidence of a disease, unless we implicate environmental factors or an infectious
etiology? For Type 2 diabetes mellitus, there is no evidence for the latter. Almost
certainly, the changes in diabetes mellitus have in large part been caused by dramatic
alterations in the Western lifestyle, in particular, the quality and quantity of the diet
and the reduced activity level. Quite clearly, our interaction with our environment is
critical to how we respond to these new nutritional and metabolic stresses. Again,
I reiterate that I appreciate the importance of these factors in the development of
diabetes and left them out of my talk only because it did not fall within the domain
of the chosen subject.
The question of how inherited factors influence human reactions to environ-
mental exposures is an increasing area of study in genetics. Though usually not
explicitly viewed in this context, it is fair to consider the genetics of diabetes as falling
within this sub-discipline. The problem is that often in the study of complex disease
it is impossible to distinguish even the roles of environmental versus genetic factors.
In trying to resolve issues, it is important to define genetics and here I take minor
issue with Dr. Rock’s working definition as requiring ‘linkage to a specific stretch
 Anthropology & Medicine 131

of DNA’. In fact, it is often quite possible to be relatively certain of a genetic basis for
a trait by the pattern of inheritance without actually localizing the responsible gene
to even a large expanse of DNA. In addition, it deserves emphasis that inheritance of
a trait does not in itself prove the existence of a genetic basis for that phenotype.
There is another phenomenon called ‘epigenetics’, in which characteristics are passed
on to future generations by mechanisms that don’t involve mutations, that is,
changes in the sequence of the DNA that forms the genes (Jiang et al. 2004). One
the best studied mechanisms for this is methylation, in which there is chemical
modification of parts of the DNA that are carried on to daughter cells as the
maternal DNA copies reproduce themselves. In regard to diabetes mellitus, the more
important epigenetic event is the tendency of fetal nutritional insufficiency to
increase the likelihood of Type 2 diabetes mellitus in adulthood (Lindsay & Bennett
2001; Wells 2003). Even more remarkably, if the Type 2 diabetic is a woman, an
increased proclivity for the development of diabetes mellitus may be passed on to her
children. At times, genetic and epigenetic effects may be difficult to distinguish,
particularly when the genetic analysis relies exclusively on so-called ‘affected sib pair
analysis’. In this genetic strategy, a large population is queried as to whether the
incidence of diabetes is greater in siblings than would be expected by chance in
unrelated individuals (Freimer & Sabatti 2004). However, there are other types of
genetic analyses, particularly those involving the father, in which it is relatively
straightforward to differentiate genetic from epigenetic modes of inheritance.
A second aspect of the above statement that depends on a false assumption is that
my opinion is representative of the medical community. As science and medicine
become more intricate, there has been an unfortunate gap between those doing
cutting-edge research and individuals actively engaged in patient care. Though we all
continue to cling to the idealized model of the physician/scientist translating
scientific findings ‘from bench to bedside’, the complexities of academic and
intellectual life have made this more a myth than reality. Though I am clinically
trained, I do not treat patients and have not done so for many years. Thus, my
approach to diabetes, if representative of any group, is more that of the basic
researcher who chooses to think about diabetes as a fascinating, but potentially
solvable puzzle, rather than as an disease to be faced every day in the most practical
of ways. It is likely that an active clinician, even one asked to deliver a talk on the
genetic aspects of diabetes mellitus, would have given a very different talk
emphasizing very different aspects of the problem. As unfair as it would have
been to extrapolate that hypothetical physician’s perspective of diabetes to the
research community, it is implausible to assume that my views are in any way similar
to those of the practicing physician. In the current academic environment, even
clinical and basic researchers tend to view disease in very different ways. Perhaps this
derives from their very different immediate goals. A clinical researcher, i.e. one who
interacts directly with human subjects, usually has the primary aim of improving the
treatment or lifestyle of people afflicted with diabetes. This investigator is often
forced to compromise scientific rigor to the ethical or logistical constraints of human
research. Nonetheless, the investigator persists in his research because of the
potential immediate impact of the work. A basic researcher, in contrast, if guided
132 M. J. Birnbaum
by the necessity of obtaining clear, unambiguous answers, and prefers those experi-
ments that can be scrupulously controlled even if the clinical applications are
distant. Obviously, these stylistic differences have an enormous impact on not only a
scientist’s approach to a research but also his overall impression of the disease, even
in the context of its origins. My view is that of the basic researcher, which gives me
the luxury of considering diabetes in an evolutionary context, a view unlikely to be
shared by investigators dedicated to patient-oriented research.
Lastly, even considering my position as a member of the biomedical research
community, it is dangerous to consider the views that I expressed in Canada as
representative of the diabetes research establishment. As discussed by Dr. Rock
in the accompanying paper, much of my presentation concerned an evolutionary
perspective on the underpinnings of diabetes mellitus. In reality, it has only been in
recent years that the research community has begun to think of lower organisms
as in any way related to human beings, and there continues to be much doubt about
this proposition among many investigators. When I first started working on the
fruit fly, Drosophila melanogaster, as a model of metabolic disease, it was virtually
impossible to obtain funding for this effort as most senior scientists regarded this
invertebrate system as irrelevant to human disease. I am gratified that much of
the community has now begun to accept the potential rewards of such avenues of
research. Nonetheless, it would be unfair to characterize this position as the
establishment view. It is also quite likely that only a minority of those working in this
field holds some of the other perspectives I touched on in my talk. My approach to
research is to look for and emphasize the innovative and novel. Whereas my talk
certainly emphasized the state of genetics research into diabetes as a whole, it would
be unavoidable for my biases to not influence the presentation.
Though I have emphasized that the talk I gave several years ago highlighted
genetics because that was the assigned topic, nonetheless I feel strongly that such
investigations into molecular aspects of the etiology of diabetes are justified, in spite
of the wealth of data emphasizing the impact of the environment on the course of
the illness. In reality, the examples of disease that has been successfully managed
in large populations by manipulating the environment or the effected individual’s
response to it are few or non-existent. In an ideal world, the most pragmatic
approach to preventing diabetes would be to revert one’s lifestyle to that of 50 years
ago, avoiding elevators, the television, and cars, and equally important, fast food.
There is certainly merit in trying to understand why these negative external
influences have strongly pervaded our society, but thus far no one has effectively
designed strategies to alter their progression. However, research scientists have been
considerably more successful in developing therapeutics that affect our response to
the environment by altering fundamental biological processes. This emphasizes the
reality that even when a disease has a primary origin in environmental influences,
molecular events intrinsic to the affected individual strongly influence the response
to these outside forces. For example, obesity is a virtually obligate antecedent for the
development of Type 2 diabetes mellitus in Western societies. On the other hand,
only about 10% of obese individuals go on to develop Type 2 diabetes mellitus.
Though the reasons for this are not at present understood, a prevalent idea is that the
 Anthropology & Medicine 133

genetic makeup of the individual accounts for this protection. One could legitimately
ask whether the pervasive influence of genetics in current research laboratories drives
academics towards genetic explanation for most unexplained biomedical phenom-
enon, or the model is really derived predominantly by ‘objective’ data. In any case,
the dialogue among anthropologists, biomedical researchers and clinicians should
help uncover possibilities for study and understanding of such devastating human
diseases as diabetes mellitus.
Lastly, I feel compelled to comment on some aspects of Dr. Rock’s accompanying
paper that do not necessarily pertain to the analysis of diabetes research. Throughout
her paper, there are numerous comments characterizing the nature of my talk and
the response to it by the audience and moderator. In some cases, Dr. Rock assumes
the motive for my words and that of the individual who introduced me and
concluded the session. In addition, Dr. Rock frequently points out the spontaneous
reaction of the audience, particularly their frequent laughter. I must confess that I do
not understand the reason behind Dr. Rock’s mentioning these various aspects of
the presentation, though they often smack of derision or mockery. She does not
explain her rationale for these parts of her commentary in the latter discussion
section of the paper. In all my oral presentations, the October 2000 Halifax being no
exception, my major goal is to communicate my ideas and view of a complicated
field. To accomplish this it I try to engage the audience through humor, metaphor,
and even occasional modest exaggeration. I am pleased by Dr. Rock’s very accurate
recount of the event that took place in Nova Scotia five years ago, but I am
concerned that the tone and context may lead to its being misinterpreted. Reading
through the transcript of my talk, I must say that I am gratified by the audience’s
very appropriate reactions that indicate to me that they were indeed ‘engaged’.
Another goal of this talk was to be somewhat provocative, and I take Dr. Rock’s
paper as evidence that I was at least in part successful.

Acknowledgements
I thank Rebecca Huss-Ashmore and Fran Barg, both of the University of
Pennsylvania, for valuable advice and Barbara Reville for her insightful reading of
this manuscript.

References
Centers for Disease Control and Prevention. (2004) ‘National diabetes fact sheet: general informa-
tion and national estimates on diabetes in the United States, 2003’, US Department of Health
and Human Services, Centers for Disease Control and Prevention, pp. 1–8.
Diabetes Research Working Group. (1999) Conquering Diabetes: A Strategic Plan for the
21st Century, National Institutes of Health, London.
Fisher, E. B., Walker, E. A., Bostrom, A., Fischhoff, B., Haire-Joshu, D. & Johnson, S. B. (2002)
‘Behavioral science research in the prevention of diabetes: status and opportunities’, Diabetes
Care, vol. 25, pp. 599–606.
Freimer, N. & Sabatti, C. (2004) ‘The use of pedigree, sib-pair and association studies of common
diseases for genetic mapping and epidemiology’, Nature Genetics, vol. 36, pp. 1045–1051.
134 M. J. Birnbaum
Jiang, Y. H., Bressler, J. & Beaudet, A. L. (2004) ‘Epigenetics and human disease’, Annual Review of
Genomics and Human Genetics, vol. 5, pp. 479–510.
Lindsay, R. S. & Bennett, P. H. (2001) ‘Type 2 diabetes, the thrifty phenotype—an overview’,
British Medical Bulletin, vol. 60, pp. 21–32.
Sorensen, T. I. (2000) ‘The changing lifestyle in the world: body weight and what else?’,
Diabetes Care, vol. 23, no. 2, pp. B1–4.
Wells, J. C. (2003) ‘The thrifty phenotype hypothesis: thrifty offspring or thrifty mother?’,
Journal of Theoretic Biology, vol. 221, pp. 143–161.
letters to nature
27. Banfi, S. et al. Identification and mapping of human cDNAs homologous to Drosophila mutant genes western blotting of liver and skeletal muscle tissue extracts (Fig. 1b,
through EST database searches. Nature Genet. 13, 167–174 (1996).
28. Bernal, J., Lee, J.-H., Cribbs, L. L. & Perez-Reyes, E. Full reversal of Pb++ block of L-type Ca++ channels c). Analysis of IRS expression in these tissues showed comparable
requires treatment with heavy metal antidotes. J. Pharmacol. Exp. Ther. 282, 172–180 (1997). levels of IRS-1 in IRS-2−/− and wild-type animals and comparable
29. Lacerda, A. E., Perez-Reyes, E., Wei, X., Castellano, A. & Brown, A. M. T-type and N-type calcium
channels of Xenopus oocytes: evidence for specific interactions with beta subunits. Biophys. J. 66,
levels of IRS-2 in IRS-1−/− and wild-type mice (Fig. 1d). IRS-2−/−
1833–1843 (1994). neonates were 10% smaller than IRS-2+/− or wild-type littermates,
30. VanDongen, A. M. J. A new algorithm for idealizing single ion channel data containing multiple and this small difference in weight persisted during weaning and
unknown conductance levels. Biophys. J. 70, 1303–1315 (1996).
into adult life (Fig. 1e).
Acknowledgements. We thank J. Yang for expert technical support. We also thank the UK HGMP Blood sugar levels were analysed at various ages to establish
Resource Centre for the Genbridge 4 radiation hybrid mapping panel and the EUCIB mouse panel. This
work was supported by grants from the NIH, Potts Foundation and the Medical Research Council. E.P.-R. the role of IRS-2 in glucose homeostasis. Three days after
is an established investigator of the American Heart Association. birth, levels of randomly fed sugars were elevated in IRS-2−/− mice
Correspondence and requests for materials should be addressed to E.P.-R. (e-mail: [email protected]). The (166 6 13 mg dl 2 1 ) compared with in wild-type mice (1166
a1G sequences have the following Genbank accession numbers: rat, AF027984; human, AF029228 and 10 mg dl 2 1 ). Monitoring of blood glucose between 3 and 6 weeks
AF029229. Human gene results are at www.hgmp.mrc.ac.uk/cgi-bin/contig/rhmapper.pl. Mouse gene
results are coded as 97/MW/001 at www.hgmp.mrc.ac.uk/Mbx/MbxHomepage.html. of age showed the development of fasting hyperglycaemia in
IRS-2−/− mice, and at 6–8 weeks of age these mice exhibited marked
glucose intolerance during an intraperitoneal glucose-tolerance test
(Fig. 2a, b). At 10 weeks, IRS-2−/− mice were overtly diabetic with
Disruption of IRS-2 causes fasting glucose levels of 323 6 35 mg dl 2 1 , and if left untreated the
levels of fasting sugars rose progressively to .400 mg dl−1 at 12–16
type 2 diabetes in mice weeks of age. Male IRS-2−/− mice showed polydypsia and polyuria

Dominic J. Withers*†, Julio Sanchez Gutierrez*†,

Heather Towery*, Deborah J. Burks*, Jian-Ming Ren‡,
Stephen Previs‡, Yitao Zhang*, Dolores Bernal*,
Sebastian Pons*, Gerald I. Shulman‡, Susan Bonner-Weir*
& Morris F. White*
* Howard Hughes Medical Institute, Joslin Diabetes Center, Department of
Medicine, Harvard Medical School, Boston, Massachusetts 02215, USA
‡ Howard Hughes Medical Institute, Department of Medicine, Yale University
School of Medicine, New Haven, Connecticut 06520, USA
† These authors contributed equally to this work.
.........................................................................................................................

Human type 2 diabetes is characterized by defects in both insulin

action and insulin secretion. It has been difficult to identify a
single molecular abnormality underlying these features. Insulin-
receptor substrates (IRS proteins) may be involved in type 2
diabetes: they mediate pleiotropic signals initiated by receptors
for insulin and other cytokines1. Disruption of IRS-1 in mice
retards growth, but diabetes does not develop because insulin
secretion increases to compensate for the mild resistance to
insulin2,3. Here we show that disruption of IRS-2 impairs both
peripheral insulin signalling and pancreatic b-cell function. IRS-
2-deficient mice show progressive deterioration of glucose
homeostasis because of insulin resistance in the liver and skeletal
muscle and a lack of b-cell compensation for this insulin resis-
tance. Our results indicate that dysfunction of IRS-2 may
contribute to the pathophysiology of human type 2 diabetes.
Secretion of insulin from pancreatic b-cells tightly regulates
glucose homeostasis by stimulating use of glucose by peripheral Figure 1 Gene targeting of the IRS-2 locus. a, Restriction map of the targeting
tissues and inhibiting hepatic glucose production4. Development of vector (top), the IRS-2 gene (middle) and the homologous recombinant gene
type 2 diabetes is characterized by a particular breakdown of this (bottom). Targeted deletion of the entire IRS-2 gene and its replacement with the
system: hyperinsulinaemia compensates for the resistance to insulin PGK/neo cassette was done using the targeting vector. Restriction enzymes are
that is found in the early prediabetic state5. The subsequent devel- as follows: A, Apa1; B, Bcl1; C, Cla1; H, HindIII; K, Kpn1; S, Spe1; X, Xma1. Probes
opment of hyperglycaemia results from the failure of b-cells to A–C were used to screen for homologous recombination by Southern blotting. b,
secrete enough insulin for effective compensation5. The cellular Southern-blot analysis from homozygous mutant mice (IRS-2−/−), heterozygotes
response to insulin is mediated by tyrosine phosphorylation of (IRS-2−/+) and wild-type animals (WT). Tail DNA was digested with Spe1 and
several cytosolic docking proteins (IRS proteins), which couple the analysed by Southern blotting with a 0.3-kilobase (kb) external genomic fragment
insulin receptor to various effector molecules, including phospha- (probe A). A 6.7-kb fragment is indicative of the wild-type allele and a 2.8-kb
tidylinositol-3-OH kinase (PI(3)K), Grb2/SOS, SHP2, NCK and fragment is indicative of the recombinant allele. c, Immunoprecipitatin and
CRK1. Identification of IRS-2 (ref. 6) as an alternative insulin western-blot analysis of liver and muscle, showing lack of IRS-2 protein in IRS-2−/−
receptor substrate in IRS-1-deficient mice7 indicates that this may mice. Each lane represents tissue from one animal and represents three
have an important role in mediating glucose homeostasis. There- independent experiments. d, Immunoprecipitation and western-blot analysis
fore, to characterize the role of IRS-2 in insulin-mediated regulation of the liver and muscle expression of IRS-1 in wild-type and IRS-2−/ − mice,
of glucose metabolism, we inactivated this gene by homologous and of IRS-2 in wild-type and IRS-1−/− mice. Each lane represents tissue
recombination (Fig. 1a). from one animal and represents three independent experiments. e,
Heterozygous IRS-2+/− offspring survived to adulthood and were Mean weight 6 s:e:m: of WT, IRS-2+/− and IRS-2−/− animals on a
interbred to homozygosity. Mice lacking IRS-2 were identified by C57B16 3 129=Sv background at the indicated ages (days). Data are from
Southern blot analysis, and the absence of IRS-2 was confirmed by six litters, with a total of at least 15 animals per genotype.

900 NATURE | VOL 391 | 26 FEBRUARY 1998

 letters to nature
without ketosis, and died from dehydration and hyperosomolar techniques8. Six-to-eight-week-old wild-type and IRS-2−/− mice in
coma; female mice followed a similar disease progression but rarely the fasted state showed identical basal rates of glucose disposal and
died. Thus, deletion of IRS-2 progressively impaired glucose levels of production of hepatic glucose, despite the presence of
tolerance with greater than 95% penetrance, whereas heterozygous hyperinsulinaemia in the IRS-2−/− animals. However, low-dose
and wild-type animals were unaffected. insulin infused at a rate of 2.5 mU kg−1 min−1 did not increase
The progressive development of diabetes in IRS-2-deficient mice whole-body glucose disposal or suppress hepatic glucose release
suggested that deletion of this gene caused a combination of in IRS-2−/− mice, although this dose of insulin markedly enhanced
peripheral insulin resistance and inadequate compensatory insulin these effects in wild-type and IRS-2+/− mice (Fig. 2e, f). A higher
secretion because of relative b-cell failure. To determine whether insulin dose (20 mU kg−1 min−1) increased glucose disposal and
IRS-2−/− mice were insulin-resistant, we measured fasting serum suppressed hepatic glucose production in IRS-2−/− mice, confirming
insulin levels and determined insulin sensitivity in vivo. Insulin that there was profound insulin resistance in both skeletal muscle
levels in wild-type and IRS-2−/− neonates were not significantly and liver.
different, suggesting that IRS-2−/− mice were slightly insulin- IRS proteins are tyrosine-phosphorylated by the insulin receptor
resistant in light of their increased random glucose levels, men-
tioned above. However, after 6 weeks IRS-2−/− mice exhibited
threefold higher fasting insulin levels than wild-type animals, and
insulin-tolerance tests showed a significantly reduced hypoglycaemic
response to exogenous insulin (Fig. 2c, d).
To define the nature of the insulin resistance, we determined basal
and insulin-stimulated whole-body glucose disposal in conscious
mice using the euglycaemic hyperinsulinaemic clamp and tracer

Figure 3 Expression and insulin-stimulated tyrosine phosphorylation of the

insulin receptor (IR), insulin-stimulated activation of PI(3)K and IRS association
with p85, and expression of PI(3)K adaptor subunits in the liver and muscle of wild-
type (WT), IRS-1−/− and IRS-2−/− mice. Supernatants of muscle (a) or liver (b)
homogenates containing equal amounts of protein from untreated and insulin
Figure 2 Fasting blood glucose and glucose-tolerance test, fasting insulin levels (INS)-treated 4–6-week-old mice were immunoprecipitated with anti-IRb antibody
and insulin-tolerance test, and in vivo glucose disposal and hepatic glucose and blotted for either antiphosphotyrosine (aPY) or IRb (aIR). Data are
production. a, After a 15 h overnight fast, blood glucose levels were determined representative of data obtained from three animals per genotype. Supernatants
using a Glucometer Elite glucometer (Bayer). Results are mean values 6 s:e:m: of muscle (c) or liver (d) homogenates containing equal amounts of protein from
for at least eight animals per genotype, with ages as indicated. WT, wild type. b, untreated and insulin-treated 4–6-week-old mice were immunoprecipitated (IP) in
Glucose-tolerance tests after intraperitoneal loading with 2 g D-glucose per kg duplicate with the indicated antibody (aIRS-1 or aIRS-2); immunoprecipitates
were performed on 6-week-old animals of the indicated genotype. Results are were assayed in vitro for PI(3)K activity. Data are the mean values 6 s:e:m: of two
mean values 6 s:e:m: for at least eight animals per genotype. c, Serum insulin independent experiments and represent data from a total of eight wild-type, nine
levels were measured by radioimmunoassay on 4–6-week-old anaesthetized IRS-2−/− and four IRS-1−/− animals. Results are expressed as fold stimulation of
animals after a 15 h overnight fast. Data are the mean values 6 s:e:m: for at activity above that of non-insulin-treated controls (fold stimulation over basal).
least 12 animals per genotype. d, Insulin-tolerance tests were performed on fed Muscle (e) and liver (f) homogenates treated as above were immunoprecipitated
4–6-week-old animals. Results are expressed as percentage of initial blood with the indicated antibody (aIRS-1 or aIRS-2) and subjected to western analysis
glucose concentration and are the mean values 6 s:e:m: for at least eight animals to study association with p85a/b. Data are representative of data obtained from
per genotype. e, Glucose-disposal rate and f hepatic glucose production rate three animals per genotype. Ins, insulin. Muscle (g) and liver (h lysates from
were determined on fasted, conscious 8-week-old mice using the euglycaemic animals of the indicated genotypes were subjected to SDS–PAGE and western
hyperinsulinaemic clamp. Basal rates and those stimulated by infusion of insulin blotting with an anti-p85 SH2 domain antiserum, which recognize p85a/b and the
at a rate of 2.5 mU kg−1 min−1 (insulin 1) and 20 mU kg−1 min−1 (insulin 2) were p50/55 splice variants. Data are representative of those obtained from three
determined. Results are the mean values 6 s:e:m: for three animals per genotype. animals per genotype.

NATURE | VOL 391 | 26 FEBRUARY 1998 901

letters to nature
during insulin stimulation and bind to the Src homology 2(SH2) immunoprecipitates from the liver and muscle of IRS-1−/− animals
domains in various effector proteins, including PI(3)K (ref. 9). We was markedly enhanced when compared with the IRS-2-associated
examined the proximal steps in this signalling cascade in IRS-2−/−, PI(3)K activity in the tissues of wild-type animals, as previously
IRS-1−/− and wild-type mice to determine the potential contribution reported7. This pattern of PI(3)K activation in muscle and liver
of any defects in these signalling elements to the development of paralleled the insulin-stimulated association of the p85 subunit of
peripheral insulin resistance. Equivalent insulin-receptor expres- PI(3)K with IRS-1 in wild-type and IRS-2−/− animals and its
sion and tyrosine phosphorylation of the insulin-receptor b- association with IRS-2 in wild-type and IRS-1−/− mice (Fig. 3e, f).
subunit were seen in both the muscle and liver of IRS-2−/− and wild- However, no differences in expression of p85a/b, p55PIK and the
type mice (Fig. 3a, b). PI(3)K may be involved in mediating several splice variants p50/55 were detected in the liver and muscle of
insulin-regulated metabolic pathways, including glucose uptake10, animals of the three genotypes (Fig. 3g, h). Therefore, the functional
antilipolysis11, glycogen synthesis12 and the suppression of hepatic defects observed in insulin-stimulated PI(3)K activation in IRS-2−/−
gluconeogenesis through the regulation of phosphoenolpyruvate mice may underlie the abnormalities in glucose metabolism in these
carboxykinase (PEPCK) expression13. Abnormalities in the activa- animals.
tion of PI(3)K could explain defective glucose homeostasis in the Although insulin resistance is important in the early stages of type
IRS-2−/− mice. Therefore, the increase in PI(3)K activity after 2 diabetes in humans, the failure in adequate b-cell compensation
stimulation with insulin of IRS-1 immunoprecipitates from the leads to the progression to the diabetic state5. Compensation for
muscle and liver of IRS-2−/− mice was compared with the increase in insulin resistance can be achieved either by greater insulin secretion
PI(3)K activity in wild-type animals and contrasted with the PI(3)K per b-cell or by an increase in b-cell mass through neogenesis or
activity detected in IRS-2 immunoprecipitates from these tissues replication of the existing b-cells14. Morphometric analysis of
from IRS-1−/− mice and wild-type mice. The increase in PI(3)K pancreases from mice at 4 weeks of age, a time when there is
activity associated with IRS-1 upon stimulation with insulin was normally a significant increase in b-cell mass15, showed that IRS-2−/−
reduced by .50% in both muscle and liver of IRS-2−/− mice as mice had significantly reduced b-cell mass (0:278 6 0:04 mg) com-
compared with wild-type animals (Fig. 3c, d). The reduction in fold pared with wild-type mice (0:677 6 0:09 mg), but no significant
stimulation was due in part to an increase in basal PI(3)K activity in difference in non-b endocrine-cell mass (Fig. 4a, b, d and data not
IRS-1 immunoprecipitates from liver and muscle of IRS-2−/− mice. shown). In contrast, the b-cell mass of IRS-1−/− mice, which have
These findings suggest a potential defect in the ability of cells to insulin resistance without diabetes, was almost double that of wild-
appropriately regulate both basal and insulin-stimulated PI(3)K type mice (1:280 6 0:07 mg) (Fig. 4c, d). Subsequent examination
activity in the absence of IRS-2. of neonatal IRS-2−/− pancreas showed relative b-cell deficiency,
In contrast, the insulin-stimulated PI(3)K activity in IRS-2 suggesting that these changes are independent of long-term
metabolic effects (data not shown). Functional assessment of
insulin release in vivo during a glucose-tolerance test showed that
4-week-old wild-type mice (with a fasting glucose level of
79 6 mg dl 2 1 ) exhibited a twofold increase in circulating insulin
levels (from 11:75 6 1:4 international microunits (mIU) ml−1 to
24:4 6 2:7 mIU ml 2 1 ; n ¼ 4) 60 minutes after glucose
loading. Similarly, IRS-2−/− mice (with fasting glucose levels of
106 6 5 mg dl 2 1 ), despite fasting hyperinsulinemia, responded
with a 1.9-fold increase in insulin levels 60 minutes after
glucose loading (fasting, 23:5 6 4:7 mIU ml 2 1 ; 60 min, 42:16
5:8 mIU ml 2 1 ; n ¼ 4). These results indicate that, in the early
stages of the development of the diabetic phenotype, glucose-
stimulated insulin release may be nearly normal. However, as the
hyperglycaemia progresses we see attenuation of glucose-stimu-
lated insulin release, which may be attributed to glucose toxicity
towards b-cell function (data not shown).
To obtain further insights into the role of IRS-2 in b-cell function,
we studied the expression of this protein in the islets of wild-type
animals. Immunofluorescence staining showed that IRS-2 expres-
sion co-localized with insulin in the islets, indicating that it may be
present in b-cells not non-b cells (Fig. 4a–g). There was also
significant IRS-2 staining in the ductal epithelium, which is the
site of neogenesis of new islets from ductal precursor cells14. IRS-2
staining was absent from the islets and ducts of IRS-2−/− mice
(Fig. 4h, i). Thus, IRS-2-dependent signalling pathways may be
Figure 4 Islet morphology and analysis of b-cell mass in IRS-2−/−, IRS-1−/− and important for the cells that are involved in the proliferative and
wild-type mice, and expression of IRS-2 in islets of wild-type and IRS-2−/− mice. neogenic responses of the islets.
a–c, Immunostaining for non-b-cell hormones of pancrease sections from 4- Our results indicate that deletion of IRS-2 causes the progressive
week-old male animals. Representative islets from (a) IRS-2−/−, (b) wild-type and development of a type 2 diabetic phenotype in mice. IRS-2−/− mice
(c) IRS-1−/− animals are shown and demonstrate the relative differences in islet- exhibit mild peripheral insulin resistance and b-cell deficiency at
cell mass between the animals of each genotype (bar represents 100 mm). d, birth but have adequate compensatory insulin secretion for several
Quantification of b-cell mass by point-counting morphometric analysis on weeks. However, subsequent relative b-cell failure in the face of
sections from the same mice analysed blind for the genotype. Mean total continued peripheral resistance causes overt fasting hyperglycaemia
pancreatic weights ðgÞ 6 s:e:m: for each genotype were: wild-type, 178 6 19; IRS- without ketoacidosis, the common characteristic of human type 2
2−/−, 123 6 12; IRS-1−/−, 113 6 9 (n ¼ 4). e–g, Immunostaining in wild-type animals diabetes (ref. 5). The mechanisms of peripheral insulin resistance
for (e) insulin, (f) IRS-2, and colocalization of both proteins (g). h, i, Immunostaining are unknown at present, and the exact contribution of the observed
in IRS-2−/− animals for insulin (h) and IRS-2 (i). Representative fields showing both insulin resistance of muscle and liver to the progression of the
islets and pancreatic ducts are shown. disease requires further study.

902 NATURE | VOL 391 | 26 FEBRUARY 1998

 letters to nature
Our results provide insight into the potential differences in the Euglycaemic hyperinsulinaemic clamps were performed on fasted conscious
physiological roles of IRS-1 and IRS-2. IRS-2−/− mice show marked mice using a two-step clamp technique8. [3-3H]glucose was infused throughout
abnormalities in glucose homeostasis but minimal growth defects, the clamp study to determine glucose-turnover rate. After a priming dose,
whereas the opposite is the case for IRS-1−/− mice. Thus, IRS-1 and [3-3H]glucose was continuously infused at a rate of ,0.08 mCi min−1 for 5 h.
IRS-2 cannot be functionally interchanged to produce either IGF-1- For basal glucose turnover rate measurements, blood samples were collected at
stimulated mitogenesis, as has been suggested by previous in vitro 70 and 80 min after the initiation of [3-3H]glucose infusion. Insulin (Eli Lilly)
data16, or, as we demonstrate, to produce insulin-regulated meta- was infused at a rate of 2.5 mU kg−1 for 90 min while 20% dextrose was infused
bolism. The signalling specificity through IRS-1 and IRS-2 may be by variable infusion pump. Blood samples were collected from tail-tip bleeds
accomplished by specific expression patterns and distinct phos- for glucose estimation every 10 min. Plasma glucose was clamped at
phorylation patterns during interaction with various activated 100 mg dl−1. While glucose levels remained steady, two blood samples were
receptors17. Our observations of PI(3)K activity in muscle and taken for [3-3H]glucose-specific-activity determination. The insulin infusion
liver show functional differences in the ability of IRS proteins to rate was then increased to 20 mU kg−1 min−1 for 90 min. Analysis of
regulate PI(3)K and implicate IRS-2 as the more critical in vivo [3-3H]glucose measurements and calculation of glucose-disposal rates and
regulator of this signalling pathway, which mediates many of the hepatic glucose production rates were performed as described8.
metabolic effects of insulin. Immunoprecipitations, western blotting and PI(3)K assays. Liver and
The combination of reduced b-cell mass and a failure of islet muscle tissue lysates were removed and homogenized at 4 8C as described7. The
hyperplasia in the face of insulin resistance and hyperglycaemia homogenates were solubilized for 1 h at 4 8C and clarified by centrifugation at
distinguishes the IRS-2−/− mouse from other monogenic and poly- 15,000 r.p.m. for 30 min. Supernatants containing equal amounts of protein
genic models of type 2 diabetes. For example, IRS-1−/− mice, or mice were immunoprecipitated for 2 h with an anti-IRS-2 antibody raised against a
with compound heterozygous disruptions of IRS-1 and the insulin glutathione-S-transferase (GST)-fusion protein containing residues 619–746
receptor18, develop marked b-cell hyperplasia in response to insulin of murine IRS-2, anti-IRS-1 antibody raised against a GST-fusion protein
resistance; this is not seen in human lean type 2 diabetics19. Likewise, containing residues 735–900 of murine IRS-1, or an antibody against insulin
IRS-2−/− mice cannot compensate for insulin resistance in this receptor subunit b. Immune complexes were collected with 100 ml of a 50%
manner. Our results indicate a unique role for IRS-2 in the slurry of protein-A–sepharose resolved on 7.5% SDS–PAGE and transferred to
regulation of b-cell neogenesis, proliferation and survival. The nitrocellulose. The blots were probed with polyclonal antibodies against IRS-1
progressive nature of the diabetes in these mice is due to both and IRS-2 and anti-p85 SH2 domain, which recognizes p85a/b, p55PIK and the
insulin resistance and reduced b-cell mass, which prevents adequate p50/55 splice variants21. Subsequent detection was by either [125I]protein A or
compensation. As this combination of features is the hallmark of enhanced chemiluminescence. For assays of p85 association with IRS proteins
human type 2 diabetes, functional abnormalities in IRS-2 could be and PI(3)K enzymatic activity, 5 units of human insulin were injected as a bolus
involved in the pathogenesis of this human disease. M into the inferior vena cava of anaesthetized mice and the liver, gastrocnemius
.........................................................................................................................
and quadriceps muscles were removed at 1, 2.5 and 3 min after insulin
Methods injection. They were homogenized and solubilized, and supernatants contain-
Preparation of the construct for homologous recombination and genera- ing equal amounts of protein were immunoprecipitated for 2 h with anti-IRS-2
tion of IRS-2-deficient mice. We cloned the IRS-2 gene from a 129 mouse or with anti-IRS-1C antibodies. Immune complexes were collected and washed
genomic library17. To construct the targeting vector, two fragments of the extensively and the PI(3)K reaction was performed as described7. [32P]incor-
genomic DNA flanking the coding region were subcloned at convenient poration was quantified using a Phosphorimager (Molecular Dynamics).
restriction sites into the pPNT vector. We transfected linearized pPNT/IRS-2 Immunochemistry and immunofluorescence. Animals were killed by
into the R1 line of embryonic stem (ES) cells derived from 129 mouse administering an overdose of sodium amytal. The pancreases were removed,
blastocysts. Selection was performed with G418 and ganciclovir, and resistant cleared of fat and lymph nodes, weighed, and, for immunochemistry, were
clones were screened for homologous recombination by Southern blotting fixed in Bouin’s solution and embedded in paraffin. Sections (of 5 mm) were
using 59 and 39 external probes (probes A and B) and by an internal probe (C) immunostained for the endocrine non-b-cells, using a cocktail of antibodies
derived from the neomycin-resistance gene (neo) cassette (Fig. 1a). One cell (rabbit antibodies against bovine glucagon, against synthetic somatostatin and
clone fulfilled the requirements for homologous recombination. Blastocysts rabbit against bovine pancreatic polypeptide15). b-cell mass was determined by
from C57B1/6 mice were injected with 10–16 targeted ES cells and implanted point-counting morphometry as described22 at a final magnification of × 420,
into pseudopregnant CD-1 foster mothers as described20. Several chimaeric with at least 175 fields being quantified per animal. For immunofluorescence,
male pups were obtained and mated with C57B1/6 females. Germline trans- pancreases were snap-frozen, and 5-mm cryosections were cut and fixed in
mission was confirmed by Southern blotting and heterozygote offspring were paraformaldehyde. Sections were permeabilized with 0.2% Triton X-100 and
intercrossed to obtain animals homozygous for the deletion of IRS-2. These stained with anti-IRS-2 antisera and guinea pig antibodies against porcine
animals were used in subsequent studies with the animals maintained with a insulin, and detection was performed with rhodamine- and fluorescein-
C57B1/129sv hybrid background. The IRS-1 targeting vector was constructed conjugated secondary antibodies.
in a similar manner and IRS-1−/− animals were maintained on the same genetic
background as the IRS-2 animals to facilitate comparison of the IRS-1−/− and
Received 24 November; accepted 29 December 1997.

IRS-2−/− phenotypes. 1. Myers, M. G. Jr & White, M. F. Insulin signal transduction and the IRS proteins. Annu. Rev.
Pharmacol. Toxicol. 36, 615–658 (1996).
Metabolic studies. Animals were maintained on a normal light/dark cycle and 2. Araki, E., Lipes, M. A., Patti, M. E., Bruning, J. C., Haag, B. L. III, Johnson, R. S. & Kahn, C. R.
handled in accordance with Joslin Diabetes Center Animal Care and Use Alternative pathway of insulin signalling in mice with targeted disruption of the IRS-1 gene. Nature
372, 186–190 (1994).
Committee protocols. Glucose levels were determined from blood taken from 3. Tamemoto, H. et al. Insulin resistance and growth retardation in mice lacking insulin receptor
mouse tails using a Glucometer Elite glucometer (Bayer). Blood for plasma substrate-1. Nature 372, 182–186 (1994).
insulin levels was taken either by retroorbital bleeds from anaesthetized mice or 4. Warram, J. H., Rich, S. S. & Krolewski, A. S. Joslin’s Diabetes Mellitus (eds Kahn, C. R. & Weir, G. C.)
201–215 (Philadelphia, Lea & Febiger, 1994).
by tail bleeds. Immunoreactive insulin levels were measured by radio immuno- 5. DeFronzo, R. A., Bonadonna, R. C. & Ferrannini, E. Pathogenesis of NIDDM: a balanced overview.
assay using rat insulin (Linco) as a standard. Glucose-tolerance tests were Diabetes Care 15, 318–368 (1992).
6. Sun, X. J. et al. Role of IRS-2 in insulin an cytokine signalling. Nature 377, 173–177 (1995).
performed on animals after a 15-h overnight fast. Animals were injected with 7. Patti, M. E. et al. 4PS/IRS-2 is the alternative substrate of the insulin receptor in IRS-1 deficient mice. J.
2 g kg−1 of D-glucose intraperitoneally and blood glucose values were Biol. Chem. 270, 24670–246673 (1995).
determined at the times indicated. Insulin tolerance was tested with fed 8. Ren, J. M. et al. Overexpression of Glut4 protein in muscle increases basal and insulin-stimulated
whole body glucose disposal in conscious mice. J. Clin. Invest. 95, 429–432 (1995).
animals between 14:00 and 16:00. Animals were injected with 0.75 units per kg 9. Backer, J. M. et al. Association of IRS-1 with the insulin receptor and the phosphatidylinositol 39-
body weight with human crystalline insulin (Lilly) intraperitoneally. Blood was kinase. Formation of binary and ternary signaling complexes in intact cells. J. Biol. Chem. 268, 8204–
8212 (1993).
taken immediately before injection and at the times indicated. Results were 10. Cheatham, B. et al. Phosphatidylinositol 3-kinase activation is required for insulin stimulation of
expressed as percentages of initial blood glucose concentration. pp70 S6 kinase DNA synthesis and glucose transporter translocation. Mol. Cell Biol. 14, 4902–4911 (1994).

NATURE | VOL 391 | 26 FEBRUARY 1998 903

letters to nature
11. Okada, T., Kawano, Y., Sakakibara, T., Hazeki, O. & Ui, M. Essential role of phosphatidylinositol 3- in precursor (pre)-B cells with complete IgH rearrangements, and in
kinase in insulin-induced glucose transport and antilipolysis in rat adipocytes: studies with a selective
inhibitor wortmannin. J. Biol. Chem. 269, 3568–3573 (1994). surface (s) IgM+ B lymphocytes when compared to age-matched
12. Shepherd, P. R., Nave, B. T. & Siddle, K. Insulin stimulation of glycogen synthesis and glycogen heterozygous controls (Table 1). This developmental block ensues
synthase activity is blocked by wortmannin and rapamycin in 3T3-L1 adipocytes: evidence for the
involvement of phosphoinositide 3-kinase and p70 ribosomal protein-S6 kinase. Biochem. J. 305, 25–
not only from defective proliferation1,2, but also from an impediment
28 (1995). to IgH rearrangements whose basis is undetermined2.
13. Gabbay, R. A. et al. Insulin regulation of phosphoenolpyruvate carboxykinase gene expression does To examine IgH rearrangements, we purified fractions of imma-
ture B lymphocytes from IL-7R−/− animals and þ/− controls. As
not require activation of the Ras/mitogen-activated protein kinase signaling pathway. J. Biol. Chem.
271, 1890–1897 (1996).
14. Bonner-Weir, S., Scaglia, L., Montana, E., Juang, J. H. & Weir, G. C. in Diabetes 1994 (eds Baba, S. & only a few cells could be isolated by fluorescent cell sorting (Table 1),
Kareko, T.) 179–228 (Excerta Medica Int. Congress, 1995).
15. Scaglia, L., Cahill, C. J., Finegood, D. T. & Bonner-Weir, S. Apoptosis participates in the remodeling of
genomic DNA was analysed3 by polymerase chain reaction (PCR) to
the endocrine pancreas in the neonatal rat. Endocrinology 138, 1736–1741 (1997). provide a semiquantitative measure of the relative frequency of
16. Bruning, J. C., Winnay, J., Cheatham, B. & Kahn, C. R. Differential signaling by insulin receptor recombination events. PCR primers are separated in the germ line
substrate 1 (IRS-1) and IRS-2 in IRS-1 deficient cells. Mol. Cell Biol. 17, 1513–1521 (1997).
17. Sun, X. J. et al. The IRS-2 gene on murine chromosome 8 encodes a unique signaling adapter for (Fig. 1a) but are brought into sufficient proximity for PCR ampli-
insulin and cytokine action. Mol. Endrocrinol. 11, 251–262 (1997). fication only after a recombination event (Fig. 1b).
18. Bruning, J. C. et al. Development of a novel polygenic model of NIDDM in mice heterozygous for IR
and IRS-1 null alleles. Cell 88, 561–572 (1997).
D–JH recombination proceeds normally without IL-7Ra (Fig.
19. Kloppel, G., Lohr, M., Habich, K., Oberholzer, M., Heitz, P. U. Islet pathology and the pathogenesis of 1c). There is no difference between IL-7Ra−/− pro-B cells and þ/−
type 1 and type 2 diabetes mellitus revisited. Surv. Synth. Pathol. Res. 4, 110–125 (1985). controls in the frequency of (D)JH joins involving 13 of 15 known
20. Papaioannou, V., Johnson, R. in Gene Targeting: A Practical Approach (ed. Joyner, A.) Ch. 4, pp. 107–
146 (Oxford Univ. Press, Oxford, 1993). DH segments4.
21. Pons, S. et al. The structure and function of p55PIK reveals a new regulatory subunit for the In contrast, VH –(D)JH joining is impaired. VH segments in IgH
phosphatidylinositol-3 kinase. Mol. Cell. Biol. 15, 4453–4465 (1994).
22. Montana, E., Bonner Weir, S. & Weir, G. C. Transplanted beta cell response to increased metabolic
are grouped into homologous families arranged in clusters (Fig. 1a)
demand. Changes in beta cell replication and mass. J. Clin. Invest. 93, 1577–1582 (1994). across approximately a megabase of DNA upstream of D segments
and the J cluster. Most VH segments are located at the extreme 59 end
Acknowledgements. We thank A. Nagy for R1 cells; M. Ginsberg, M. Petruzelli and M. Taneja for
technical support; B. Cheatham for the anti-IRb antibody. Blastocyst injections were performed in the
of this region5,6 and belong to the large VHJ558 family (with an
Core Laboratory of the Diabetes and Endocrinology Research Center, Vanderbilt University. This work estimated 60–1,000 members). In contrast, the small VH7183 family
was supported by grants from the NIH to G.I.S., S.B.W. and M.F.W. D.J.W. is supported by an MRC (UK)
Clinician Scientist Fellowship. D.J.B. was supported by a grant from the JDFI. J.S.G. and D.B. were
containing only ,25 members is adjacent to DH segments at its 39
supported by grants from the Spanish Government. end. Rearrangements involving these VH segment families were
distinguished by a PCR assay whose specificity was verified (Fig. 2a)
Correspondence and requests for materials should be addressed to M.F.W. (e-mail: Whitemor@joslab.
harvard.edu).
using hybridomas harbouring known rearrangements. VH to (D)JH
joins involving VHJ558 segments are normally the most frequent
rearrangements detected in the sIgM+ B lymphocyte fraction7,8.
Surprisingly, VHJ558 rearrangements were barely detectable in
IL-7Ra−/−, sIgM+ cells compared to þ/− controls, despite
Impaired immunoglobulin
gene rearrangement in a
5'
VH segments DH JH CH
3'
~ 800 Kb 60 Kb 40 Kb

mice lacking the IL-7 receptor VHX24

VHJ558 VHJ606 VH3660 VH7183

VHGAM3 VHQ52
Anne E. Corcoran, Andrew Riddell, Danielle Krooshoop
& Ashok R. Venkitaraman b Germline : no PCR amplification
VHJ558 VH7183 DH JH1-4
5' 3'
Medical Research Council Laboratory of Molecular Biology,
Cambridge CB2 2QH, UK (D)JH1-4 PCR products
(D)JH rearranged
.........................................................................................................................
VHJ558 VH7183

To generate the full diversity of antibody heavy-chain genes, (D)JH1

hundreds of dispersed germline V segments must undergo recom- V(D)JH1-4 PCR products

bination following D–J segment joining. Here we report that this V(D)JH rearranged

process is regulated by the a-chain of the receptor for interleukin- V(D)JH1 JH2-4

7, a cytokine that stimulates B-cell lymphopoiesis1. D–J joining

occurs normally in immature B lymphocytes from mice lacking
the a-chain of the interleukin-7 receptor (IL-7Ra). But recombi-
nation of V segments is progressively impaired as their distance
increases upstream of D/J, causing infrequent rearrangement of
most V segments, which markedly reduces diversity. This is not
simply due to defective cell proliferation or impaired recombinase
expression. Rather, germline transcripts from distal, unrear-
ranged V segments, a marker of chromatin changes that precede
recombination, are specifically silenced. So too is expression of
Pax-5, which binds to heavy-chain locus control elements and
normally stimulates recombination, suggesting a mechanism for
these effects. Thus ligands of the interleukin-7 receptor deliver an
extrinsic signal that targets V segment recombination in the
heavy-chain locus by altering the accessibility of DNA substrates Figure 1 Normal (D)JH recombination in IL-7Ra−/− mice. a, The murine IgH locus6,
to the recombinase. This mechanism augments the recombina- showing approximate map positions of the different VH families. b, PCR assays
tional diversity of the primary antibody repertoire. used to quantify (D)JH and V to (D)JH rearrangements. Arrows indicate specific
B lymphopoiesis is impeded at an early stage in the bone marrow oligonucleotide primers. c, There is no difference between IL-7R−/− and IL-7R+/−
of mice lacking IL-7Ra1. The number of progenitor (pro)-B cells pro-B cells in the frequency of (D)JH joining. DSP2/FL16 primers used detect joins
undergoing immunoglobulin heavy-chain gene (IgH) rearrange- involving 13 of 15 known murine DH segments. Expected product sizes are
ment is normal (Table 1). But there is a severe (,10-fold) reduction indicated.

904 NATURE | VOL 391 | 26 FEBRUARY 1998

Copyright of Nature is the property of Nature Publishing Group and its content may not be copied or emailed to
multiple sites or posted to a listserv without the copyright holder's express written permission. However, users
may print, download, or email articles for individual use.
SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE
2023, VOL. 41, NO. 4, 392–399
https://doi.org/10.1080/02813432.2023.2255062

RESEARCH ARTICLE

Use of primary health care services and mortality in older patients with type
2 diabetes with or without comorbidities
E. Mellanena , T. Kauppilaa , H. Kautiainena,b,c , M. Lehtoa,d , O. Rahkonene , K. Pitk€al€aa and
M. K. Lainea,b
a
Department of General Practice and Primary Health Care, University of Helsinki and Helsinki University Hospital, Helsinki, Finland;
b
Folkh€alsan Research Centre, Helsinki, Finland; cPrimary Health Care Unit, Kuopio University Hospital, Kuopio, Finland; dCity of
Vantaa, Vantaa, Finland; eDepartment of Public Health, University of Helsinki, Helsinki, Finland

ABSTRACT ARTICLE HISTORY

Objective: This study aimed to examine primary health care (PHC) service utilization and mortal­ Received 15 May 2023
ity in older patients with type 2 diabetes (T2D) with or without comorbidities. Accepted 30 August 2023
Design and setting: A cohort study in PHC in the city of Vantaa, Finland. Follow-up period was
KEYWORDS
set between the years 2011 and 2018.
Type 2 diabetes mellitus;
Subjects: PHC patients aged 60 years or more with a T2D were included. mortality; service utilization;
Main outcome measures: Service utilization was defined as the number of face-to-face appoint­ primary healthcare;
ments and telephone contacts between a patient and general practitioner (GP) or nurse. The comorbidity; aged
presence of comorbidities was defined using the Charlson Comorbidity Index (CCI). Mortality
was assessed using hazard ratio (HR) and standardized mortality ratio (SMR).
Results: In total, 11,020 patients were included and followed for 71,596 person years. Mean age
of the women and men in the beginning of follow-up were 71 and 69 years, respectively. The
patients in the study cohort had a mean of eight appointments per person year to the GPs or
nurses. Patients with T2D with comorbidities had more appointments than patients with T2D
without comorbidities (incidence rate ratio (IRR) 1.44 [95% CI 1.39–1.49]). Increase in the number
of all appointments reduced mortality in patients with T2D with and without comorbidities.
Between patients with T2D with comorbidities and patients with T2D without comorbidities, the
age and sex adjusted HR for death was 1.50 (95% CI 1.39–1.62). The SMR was higher in patients
with T2D with comorbidities (1.83 [95% CI 1.74–1.92]) than in patients with T2D without comor­
bidities (0.91 [95% CI 0.86–0.96]).
Conclusions: In older patients with T2D, the presence of comorbidities was associated with
increased use of PHC services and increased mortality. Increase in the number of appointments
was associated with reduced mortality in patients with T2D with or without comorbidities.
KEY POINTS
� In older patients with T2D, it has not been studied whether and to what extend multimorbid­
ity affects use of PHC services and mortality.
� The presence of comorbidities according to the Charlson Comorbidity Index (CCI) was associ­
ated with increased use of PHC services.
� The number of appointments to GPs or nurses was associated with reduced mortality in
patients with T2D with or without comorbidities according to the CCI.

Introduction The global prevalence of T2D is estimated to increase,

Type 2 diabetes (T2D) is a major cause of morbidity, although some European and East Asian countries
and the health burden of T2D is increasing [1–3]. have shown stable or decreasing trends in the inci­
Diabetes is one of the 10 most important drivers of dence of T2D [1,4]. Furthermore, the burden of T2D in
increasing burden, with a 24% age-standardized older people is estimated to almost double by
increase in disability-adjusted life year (DALY) rates [3]. 2030 [2].

CONTACT E. Mellanen [email protected] Department of General Practice and Primary Health Care, University of Helsinki and Helsinki
University Hospital, Helsinki, Finland
Supplemental data for this article can be accessed online at https://doi.org/10.1080/02813432.2023.2255062.
� 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the
posting of the Accepted Manuscript in a repository by the author(s) or with their consent.
 SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 393

Globally, most patients with T2D are treated in pri­ electronic health care system (Finnstar, Logica,
mary health care (PHC), and T2D is a common reason Helsinki, Finland). This electronic health care system
for PHC appointment [5–9]. In Finland, T2D is the fifth includes data on the age and sex of the patients,
most common cause of general practitioner (GP) appointments to the GPs and nurses, diagnosis codes
appointments [10]. Older T2D patients seem to use according to the International Classification of
PHC services more often than age- and sex-matched Diseases tenth revision (ICD-10), and prescription drug
controls without diabetes [11]. Furthermore, the use of details with Anatomical Therapeutic Chemical (ATC)
PHC services in patients with T2D seems to increase codes.
with the duration of T2D [12,13]. In addition, the use Patients were included in our study cohort if they
of physicians, nurses and PHC services varies in T2D were at least 60 years old at the beginning of the fol­
patients depending on the context, organization of low-up period and had a diagnosis of T2D made dur­
health care and patient profile [14]. Multimorbidity is ing the inclusion period. This age group was selected
common in patients with T2D, and they have more as the study cohort because patients in this age group
comorbidities than those without diabetes [15]. mostly use public PHC services [19,20]. Patients were
A recent English cohort study in adult patients with defined as having T2D if they had an ICD-10 code E11
T2D found an association between additional PHC or at least one prescription of an antihyperglycemic
appointments and increased rates of mortality [16]. drug (ATC codes A10�) written by GPs. Data on sex,
However, the assessment of mortality lacked impor­ age at the beginning of the follow-up period, comor­
tant factors such as the use of a reference population bidities relevant to the calculation of the Charlson
and standardized mortality ratio (SMR). The extent to Comorbidity Index (CCI), and the number of appoint­
which multimorbidity in older patients with T2D ments to GPs or nurses during the follow-up period
affects PHC utilization and mortality is also unknown. were gathered. Age at the beginning of follow-up, 1
This study aimed to examine the association
September 2011, was used in all analyses. CCI was
between comorbidity and the use of PHC services and
defined according to the CCI [21] using data from the
the association between mortality and the use of PHC
inclusion period from 1 January 2002 to 31 August
services in patients with T2D aged 60 years or older.
2011. An exception in the calculation of CCI was made
with diabetes, which was excluded from the calcula­
Methods tion of CCI, and no age correction was performed.
Patients were defined as having T2D with comorbid­
This cohort study was conducted in the city of Vantaa,
Finland, which is located in the Helsinki metropolitan ities if they had a CCI value of one or more, and T2D
area. Vantaa is the fourth-most populated city in without comorbidities if they had a CCI value of zero.
Finland [17]. In 2011, Vantaa had a population of The comorbidity status acquired at the end of the
203,000 inhabitants of which 51% were women, 24% inclusion period was used in all analyses. Patients with
were 60 years or older and 11% had a foreign back­ T2D with comorbidities and patients with T2D without
ground [17]. Finland’s healthcare system is multifa­ comorbidities are later in this paper referred to as
ceted. The health care system is mainly publicly T2D-patients with comorbidities and T2D-patients
funded, and the public sector is divided into PHC and without comorbidities, respectively. The appointments
specialized health care. In addition to public health included in the analysis were face-to-face appoint­
care, there is a private sector and occupational health ments and telephone contacts, and the number of
care sector [18]. Furthermore, in Finnish PHC, specially appointments was reported per person year.
trained diabetes nurses, registered nurses and public Healthcare professionals whose appointments were
health nurses are an active part in the treatment of included were GPs and nurses. Patients were followed
patients with T2D to the extent where these nurses up until the end of the follow-up period or until
can even make changes to dosing of patients’ medica­ death, whichever occurred first. Data on the date of
tions independently. death were obtained from Statistics Finland (Helsinki,
Due to changes in the organization of PHC services Finland). Mortality was assessed using hazard ratio
locally in the city of Vantaa, we set the inclusion (HR) and SMR.
period between 1 January 2002 and 31 August 2011, Research permits for the data and this study were
and the follow-up period between 1 September 2011 granted by the Research Ethics Committee of the
and 31 December 2018. During the inclusion and fol­ Faculty of Medicine of the University of Helsinki
low-up periods, the PHC of Vantaa city used an (04/2019 and 06/2020), the city of Vantaa
394 E. MELLANEN ET AL.

(VD/8059/13.00.00/2016 and VD/4713/13.00.00/2019), Results

and Statistics Finland (TK-53-514-20).
Table 1 shows the characteristics of the study cohort.
Data are presented as means with standard devia­
The mean ages of the women and men were 71 (SD
tions (SDs) or counts with percentages.
7.9) and 69 (SD 6.8) years, respectively. Of the T2D-
Crude and standardized estimates of care service
patients included in the study cohort, 34% had comor­
incidence or incidence rate ratios (IRRs) were calcu­
bidities according to CCI. Among T2D-patients with
lated using Poisson’s regression models. A possible
comorbidities, the mean CCI was 2.9 (SD 1.3), and no
nonlinear relationship between the number of
difference was observed between men and women in
appointments and the HR of death was assessed
using 3-knot-restricted cubic spline regression. The the presence of comorbidities, with mean CCI values
length of the distribution of knots was located at the of 2.9 (SD 1.3) and 2.8 (SD 1.3), respectively.
25th, 50th and 75th percentiles. The knot locations Figure 1 shows the distribution of appointments
were based on Harrell’s recommended percentiles per person years to the GPs and nurses in T2D-
[22]. Kaplan–Meier’s survival analysis were performed patients with or without comorbidities. Altogether, the
to estimate cumulative all-cause mortality. Adjusted patients of the study cohort had a mean of 7.98
Kaplan–Meier cumulative mortality rates were esti­ appointments per person years to the GPs or nurses.
mated using inverse probability weighting (IPW). Age Table 2 shows the appointments per person years to
and sex were introduced into the models as covari­ the GPs and nurses. T2D-patients with comorbidities
ates when appropriate. The ratio between observed had more appointments to the GPs or nurses com­
and expected numbers, the SMR, was calculated pared to T2D-patients without comorbidities (IRR 1.44
using subject-years methods with 95% confidence [95% CI 1.39–1.49]).
intervals (CIs), assuming a Poisson distribution. The The SMR was higher in T2D-patients with comor­
expected number of deaths was calculated on the bidities than in T2D-patients without comorbidities
basis of sex-, age-, and calendar-period-specific mor­ (p < .001) (Table 1). Between T2D-patients with
tality rates in the Vantaa population (Official Statistics comorbidities and T2D-patients without comorbidities,
of Finland, Helsinki, Finland). The expected number the age and sex adjusted HR for death was 1.50 (95%
was determined by multiplying the person-years of CI 1.39–1.62). The number of GP and nurse appoint­
observation by the appropriate mortality rate in the ments reduced mortality in both T2D-patients with
general population according to categories of sex, 1- comorbidities and in T2D-patients without comorbid­
year age group and calendar period. The Poisson ities (Figure 2). In Figure 2, at the mean number of all
regression was tested using the goodness-of-fit test appointments (eight appointments per person years),
of the model, and the assumptions of overdispersion in T2D-patients without comorbidities and in T2D-
in the Poisson model were tested using the Lagrange patients with comorbidities, the HRs for death were
multiplier test. The proportional hazards assumption 0.82 (95% CI 0.75–0.89) and 0.75 (0.68–0.82),
was tested graphically and using a statistical test respectively.
based on the distribution of Schoenfeld residuals. All In comparison with the population of Vantaa, T2D-
analyses were performed using STATA 17 .0 patients with or without comorbidities had mostly
(StataCorp LP, College Station, TX). higher SMRs (Figure 3) at different PHC service

Table 1. Characteristics and mortality of the patients aged 60 years or more with type 2 diabetes with or without comorbidities
from the city of Vantaa, Finland.
Patients with type 2
diabetes without Patients with type 2
All patients comorbidities diabetes with comorbidities p Value
Number of persons 11,020 7267 3753 –
Women, n (%) 5503 (50) 3605 (50) 1898 (51) .34
Age, years, mean (standard 70 (7) 69 (7) 72 (8) <.001
deviation)
Number of deaths, n 2682 1195 1487
Person years followed up 71,596 49,103 22,493 –
Mortalitya (95% confidence 24.3 (23.5–25.2) 16.4 (15.6–17.3) 39.6 (38.1–41.2) <.001
interval)
Standardized mortality ratio 1.26 (1.21–1.31) 0.91 (0.86–0.96) 1.83 (1.74–1.92) <.001
(95% confidence interval)
Presence of comorbidities based on Charlson Comorbidity Index.
a
At the end of the follow-up period (Kaplan–Meier’s estimate).
 SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 395

Figure 1. Distribution of general practitioner (GP) and nurse appointments per person years in the patients aged 60 years or
more with type 2 diabetes with or without comorbidities from the city of Vantaa, Finland. Presence of comorbidities based on
Charlson Comorbidity Index.

Table 2. General practitioner and nurse appointments per person years in patients aged 60 years or more with type 2 diabetes
with or without comorbidities from the city of Vantaa, Finland.
All patients with Patients with Patients with
type 2 diabetes, type 2 diabetes without type 2 diabetes with
mean (SE) comorbidities, mean (SE) comorbidities, mean (SE) IRRa (95% CI)
All appointments
Women 8.46 (0.09) 7.54 (0.10) 10.42 (0.18) 1.38 (1.32–1.44)
Men 7.49 (0.09) 6.38 (0.10) 9.98 (0.19) 1.51 (1.43–1.58)
All 7.98 (0.07) 6.96 (0.07) 10.21 (0.13) 1.44 (1.39–1.49)
General practitioner appointments
Women 2.80 (0.03) 2.51 (0.04) 3.41 (0.06) 1.33 (1.27–1.39)
Men 2.34 (0.03) 2.01 (0.03) 3.07 (0.06) 1.44 (1.37–1.51)
All 2.57 (0.02) 2.26 (0.02) 3.25 (0.04) 1.38 (1.33–1.43)
Nurse appointments
Women 5.66 (0.07) 5.03 (0.07) 7.01 (0.14) 1.40 (1.34–1.47)
Men 5.15 (0.07) 4.37 (0.07) 6.91 (0.14) 1.54 (1.46–1.63)
All 5.41 (0.05) 4.7 (0.05) 6.96 (0.10) 1.47 (1.42–1.52)
IRR: incidence rate ratio; SE: standard error; CI: confidence interval.
Presence of comorbidities based on Charlson Comorbidity Index.
a
Between patients with type 2 diabetes without comorbidities and patients with type 2 diabetes with comorbidities, adjusted for age and sex.

utilization rates. Figure 4 shows the age-adjusted

Discussion
Kaplan–Meier survival figures in women and men
between T2D-patients with or without comorbidities. The present study found that T2D-patients with
In women and men, the age-adjusted HRs for death comorbidities had more GP and nurse appointments
between T2D-patients with comorbidities and T2D- than T2D-patients without comorbidities. Furthermore,
patients without comorbidities were 2.08 (95% CI it was observed that the number of appointments to
1.86–2.33) and 2.07 (95% CI 1.86–2.30), respectively. A a GP or nurse reduced mortality in T2D-patients with
directed acyclic graph (DAG) between an individual or without comorbidities. Mortality in T2D-patients
exposure (use of PHC services) and outcome (mortal­ without comorbidities was lower than in T2D-patients
ity) is shown in Supplementary figure 1. with comorbidities.
396 E. MELLANEN ET AL.

This study has several strengths. The study cohort

is a representative sample of patients with T2D aged
60 years or older because of the integrity of the elec­
tronic patient record system in a single city, a publicly
funded health care system, and inclusion of only older
patients. During the follow-up, the proportion of GP
appointments with a documented diagnosis in the col­
lected dataset was at a good level (approximately
90%), as shown in a previous study using the same
dataset where PHC visits and documented diagnosis
were examined [23]. Inclusion of all GP and nurse
appointments and not focusing on only T2D related
appointments gives a more complete view of the util­
ization of PHC services. The inclusion of nurse appoint­
ments is crucial in a nurse-driven T2D treatment
system like ours in Finland to obtain an appropriate
view of service utilization.
This study has some limitations. The organization
Figure 2. Association of the number of general practitioner and accessibility of healthcare services vary depending
and nurse appointments and mortality in patients aged on country and context, and since the data of the
60 years or more with type 2 diabetes with or without comor­ study was obtained from a single city, the city of
bidities from the city of Vantaa, Finland. Presence of comor­ Vantaa, the generalizability of observed associations
bidities based on Charlson Comorbidity Index. The number of
appointments is expressed per person years and mortality is and presented appointment numbers is limited. The
shown as age and sex adjusted hazard ratios. proportion of nurse appointments was high in this
study and therefore the results presented may not be
reproducible in health care system where appoint­
ments are concentrated to a GPs alone. The use of
PHC services was examined, but the total use of all
health care services remains unclear because the col­
lected data did not include information from special­
ized health care, the private sector or occupational
health care. In addition, continuity of care and its
effect on mortality was not studied. Potential media­
tors of the results presented in this paper were not
examined because the data of this study did not
include laboratory tests or imaging study results.
Further, the duration of comorbidities or T2D was
unknown, and thus no further subanalyses were done.
This study found that PHC patients aged 60 years
or more with T2D with comorbidities had an increased
number of appointments to GPs or nurses. These find­
ings are aligned with those of previous studies that
have reported higher annual PHC service utilization
rates in adult patients with T2D with comorbidities
[16,24–26]. However, previous studies have varied in
their study designs. All previous studies included
patients with T2D aged 18 years or older, in contrast
Figure 3. Comparison of standardized mortality ratio between to this study’s older study cohort [16,24–26]. Various
population aged 60 years or more of the city of Vantaa, definitions of comorbidity or multimorbidity have
Finland, and patients with type 2 diabetes with or without
comorbidities in different number of appointments per person been used. Two studies stratified patients with T2D
years. Presence of comorbidities based on Charlson according to their cardiovascular disease (CVD) status
Comorbidity Index. [24,26], one study used CCI [25], and one study
 SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 397

Figure 4. Age-adjusted Kaplan–Meier’s survival comparison in women and men with type 2 diabetes aged 60 years or more with
or without comorbidities from the city of Vantaa, Finland. Presence of comorbidities based on Charlson Comorbidity Index.

defined comorbidities based on existing quality out­ between PHC GP and nurse consultation rates and
come framework (QOF) conditions, including coronary mortality [27]. Contrary to this study’s results, another
heart disease [16]. Furthermore, there are differences English study by Hodgson et al. reported increased
regarding which healthcare professional groups and rates of mortality per additional PHC contact in
which types of appointments are included in the stud­ patients with T2D [16]. In addition to the distortion
ies. Two studies reported GP and nurse appointments caused by different healthcare systems between this
[24,25] and two other studies reported the total num­ study and the study by Hodgson et al., differences in
ber of contacts without specifying which healthcare study cohorts and determination of covariates might
professional contacts were considered [16,26]. One partially explain the different findings. This study’s
study reported face-to-face and telephone appoint­ study cohort was older, and even though Hodgson
ments, as was done in this study [25]. One study et al. used age adjustment in their analysis of mortal­
reported only in-person appointments [24], one study ity, the difference between the study cohorts age
reported in-person appointments, telephone appoint­ distribution is significant. Hodgson et al. included
ments and home visits [16] and one study reported in-home visits, which were not included in this study.
outpatient visits, emergency room visits and inpatient In this study, analysis was restricted to GP and nurse
hospital visits [26]. Despite differences in reported appointments only, whereas no specification of a
studies, T2D-patients with comorbidities tend to use health care professionals was done by Hodgson et al.
more PHC services than T2D-patients without Hodgson et al.’s study had a larger study cohort than
comorbidities. this study, but as a strength of this study, in addition
The more appointments T2D-patients with or with­ to HR, mortality was also examined using SMR. Use of
out comorbidities had to GPs or nurses, the lower was SMR expands assessment of mortality to the level of
the patient mortality rate. Further, T2D-patients with population of the city this study was conducted in
comorbidities had higher mortality than T2D-patients instead of only looking mortality inside a certain
without comorbidities. In addition, mortality in T2D- cohort. The lower SMR of T2D patients without comor­
patients without comorbidities was lower in relation bidities in relation to the population of Vantaa might
to the population of Vantaa. A previous English study be explained by the Finnish T2D treatment system,
by Lay-Flurrie et al. where no stratification by underly­ which is well resourced with specialized diabetes
ing diagnoses was done, found no association nurses focusing mostly on patients with diabetes, or
398 E. MELLANEN ET AL.

factors not included in the data, such as lifestyle, Acknowledgements

smoking status or morbidity of the population of
Open access is funded by the Helsinki University Library.
Vantaa. The frequency or number of PCH appoint­
ments in patients with T2D has been reported to
improve glycemic control in patients with T2D who Disclosure statement
were formerly dropped out of the treatment system The authors declare no conflicts of interest.
[28] and to reduce the occurrence rate of potentially
avoidable hospitalizations [29]. These reported benefi­
Funding
cial findings could provide some reasoning for the
presented results showing reduced mortality with a No financial support was received during the preparation of
this study.
higher number of PHC GP and nurse appointments in
patients with T2D. Furthermore, the present study sug­
gests topics for further research. Mortality seemed to ORCID
increase in Figure 3 when annual PHC appointments E. Mellanen http://orcid.org/0000-0003-4946-5768
increase from 0 to 1–4. If this increase in mortality T. Kauppila http://orcid.org/0000-0002-6155-0300
with low annual PHC appointments compared with H. Kautiainen http://orcid.org/0000-0003-0786-0858
not visiting health centers at all is true, further studies M. Lehto http://orcid.org/0000-0001-7315-2118
O. Rahkonen http://orcid.org/0000-0002-7202-3274
would be needed to identify mechanisms involved K. Pitk€al€a http://orcid.org/0000-0001-9659-6985
(other comorbidities not involved in our current study, M. K. Laine http://orcid.org/0000-0002-1848-1514
medicational aspects, issues related to organization of
health services, patients’ behavioral mechanisms
References
involved, etc.).
The present study observed that T2D-patients with 0[1] Khan MAB, Hashim MJ, King JK, et al. Epidemiology of
comorbidities used more PHC services than T2D- type 2 diabetes – global burden of disease and fore­
casted trends. J Epidemiol Glob Health. 2019;10(1):
patients without comorbidities. In a world of growing
107–111. doi: 10.2991/jegh.k.191028.001.
demands on healthcare systems, it would be beneficial 0[2] Prince MJ, Wu F, Guo Y, et al. The burden of disease
to focus on preventative measures in patients with in older people and implications for health policy and
T2D to postpone the development of comorbidities practice. Lancet. 2015;385(9967):549–562. doi: 10.
and potentially reduce the use of PHC services. This 1016/S0140-6736(14)61347-7.
study also demonstrated a reduction in mortality with 0[3] GBD 2019 Diseases and Injuries Collaborators. Global
burden of 369 diseases and injuries in 204 countries
a higher number of PHC appointments. Due to limited
and territories, 1990-2019: a systematic analysis for
resources in healthcare, it is not meaningful or pos­ the global burden of disease study 2019. Lancet.
sible to pursue more frequent PHC appointments for 2020;396(10258):1204–1222.
every patient with T2D. Instead, it might be beneficial 0[4] Miao K, Wang Y, Cao W, et al. Trends in incidence of
to target patients with T2D who do not use or use sel­ total or type 2 diabetes: systematic review. Twin Res
Hum Genet. 2023;366:1–8.
dom PHC services in order to return them among
0[5] Bigio J, MacLean E, Vasquez NA, et al. Most common
services [28]. Future research is needed to identify fac­ reasons for primary care visits in low- and middle-
tors and needs that shape patients’ use of PHC serv­ income countries: a systematic review. PLOS Glob
ices, clarify the effect of PHC service utilization on Public Health. 2022;2(5):e0000196. doi: 10.1371/jour­
mortality, and evaluate whether the assessment of nal.pgph.0000196.
continuity of care alters the results presented in this 0[6] W€andell P, Carlsson AC, Wettermark B, et al. Most
common diseases diagnosed in primary care in
study.
Stockholm, Sweden, in 2011. Fam Pract. 2013;30(5):
In T2D-patients aged at least 60 years, the presence 506–513. doi: 10.1093/fampra/cmt033.
of comorbidities was associated with an increased use 0[7] Mash B, Fairall L, Adejayan O, et al. A morbidity sur­
of PHC services measured as GP or nurse appoint­ vey of South African primary care. PLOS One. 2012;
ments. An increase in the number of GP or nurse 7(3):e32358. doi: 10.1371/journal.pone.0032358.
appointments was associated with a reduction in mor­ 0[8] Soler JK, Okkes I, Oskam S, et al. An international
comparative family medicine study of the transition
tality in both T2D-patients with or without comorbid­
project data from The Netherlands, Malta and Serbia.
ities. A lower SMR in T2D-patients without Is family medicine an international discipline?
comorbidities was observed in relation to the popula­ Comparing incidence and prevalence rates of reasons
tion of Vantaa. for encounter and diagnostic titles of episodes of
 SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 399

care across populations. Fam Pract. 2012;29(3):283– [20] Hannikainen K. Ik€a€antyneiden sosiaali- ja terveyspalve­
298. doi: 10.1093/fampra/cmr098. luiden tarve ja k€aytto
€ eroavat tulotason mukaan [The
0[9] Davidson JA. The increasing role of primary care need and use of social and health services in the eld­
physicians in caring for patients with type 2 diabetes erly differs according to income level]. Finnish
mellitus. Mayo Clin Proc. 2010;85(12):S3–S4. doi: 10. Institute for Health and Welfare; 2018. Available from:
4065/mcp.2010.0466. https://urn.fi/URN:ISBN:978-952-343-067-9
[10] Hauhio N, Puroharju T, Mo €ll€ari K. Perusterveyden- [21] Charlson ME, Pompei P, Ales KL, et al. A new method
huollon avosairaanhoidon vastaanoton asiakkaiden of classifying prognostic comorbidity in longitudinal
k€ayntisyyt vuonna 2020 [Reasons for visits by cli­ studies: development and validation. J Chronic Dis.
ents of primary health care outpatient clinics in 1987;40(5):373–383. [Database] doi: 10.1016/0021-
2020]. Finnish Institute for Health and Welfare; 9681(87)90171-8.
2021. Available from: https://urn.fi/URN:NBN:fi- [22] Fe H. Regression modeling strategies: with applica­
fe2021042820667 tions to linear models, logistic regression, and survival
[11] Aro AK, Karjalainen M, Tiihonen M, et al. Use of pri­ analysis. New York (NY): Springer; 2001.
mary health care services among older patients with [23] Lehto M, Mustonen K, Raina M, et al. Differences
and without diabetes. BMC Prim Care. 2022;23(1):233. between recorded diagnoses of patients of an emer­
doi: 10.1186/s12875-022-01844-2. gency department and office-hours primary care doc­
[12] Weng W, Liang Y, Kimball ES, et al. Longitudinal tors: a register-based study in a Finnish town. Int J
changes in medical services and related costs in a sin­ Circumpolar Health. 2021;80(1):1935593.
gle cohort of patients newly diagnosed with type 2 [24] Abner S, Gillies CL, Shabnam S, et al. Consultation
diabetes, 2006 to 2012. Clin Ther. 2016;38(6):1314– rates in people with type 2 diabetes with and without
1326. doi: 10.1016/j.clinthera.2016.03.032. vascular complications: a retrospective analysis of
[13] Sabale U, Bodegård J, Sundstro €m J, et al. Healthcare
141,328 adults in England. Cardiovasc Diabetol. 2022;
utilization and costs following newly diagnosed type-
21(1):8. doi: 10.1186/s12933-021-01435-y.
2 diabetes in Sweden: a follow-up of 38,956 patients
[25] Coles B, Zaccardi F, Seidu S, et al. Rates and esti­
in a clinical practice setting. Prim Care Diabetes. 2015;
mated cost of primary care consultations in people
9(5):330–337. doi: 10.1016/j.pcd.2015.01.001.
diagnosed with type 2 diabetes and comorbidities: a
[14] van Dijk CE, Hoekstra T, Verheij RA, et al. Type II dia­
retrospective analysis of 8.9 million consultations.
betes patients in primary care: profiles of healthcare
Diabetes Obes Metab. 2021;23(6):1301–1310. doi: 10.
utilization obtained from observational data. BMC
1111/dom.14340.
Health Serv Res. 2013;13(1):7. doi: 10.1186/1472-6963-
[26] Mehta S, Ghosh S, Sander S, et al. Differences in all-
13-7.
cause health care utilization and costs in a type 2 dia­
[15] Zghebi SS, Steinke DT, Rutter MK, et al. Eleven-year
multimorbidity burden among 637 255 people with betes mellitus population with and without a history
and without type 2 diabetes: a population-based of cardiovascular disease. J Manag Care Spec Pharm.
study using primary care and linked hospitalisation 2018;24(3):280–290.
data. BMJ Open. 2020;10(7):e033866. doi: 10.1136/ [27] Lay-Flurrie S, Mathieu E, Bankhead C, et al. Patient
bmjopen-2019-033866. consultation rate and clinical and NHS outcomes: a
[16] Hodgson S, Morgan-Harrisskitt J, Hounkpatin H, et al. cross-sectional analysis of English Primary Care Data
Primary care service utilisation and outcomes in type from 2.7 million patients in 238 practices. BMC Health
2 diabetes: a longitudinal cohort analysis. BMJ Open. Serv Res. 2019;19(1):219. doi: 10.1186/s12913-019-
2022;12(1):e054654. doi: 10.1136/bmjopen-2021- 4036-y.
054654. [28] Kauppila T, Eriksson JG, Honkasalo M, et al.
[17] Official Statistics of Finland (OSF): population struc­ Relationship between number of contacts between
ture [e-publication] [Internet]. Statistics Finland; 2023 previous dropouts with type 2 diabetes and health
[cited 2023 Feb 13]. Available from: http://www.stat.fi/ care professionals on glycaemic control: a cohort
til/vaerak/index.html study in public primary health care. Prim Care
[18] Keskimaki I, Tynkkynen LK, Reissell E, et al. Finland: Diabetes. 2019;13(5):468–473. doi: 10.1016/j.pcd.2019.
health system review. Health Syst Transit. 2019;21(2): 03.003.
1–166. [29] Georgescu V, Green A, Jensen PB, et al. Primary care
[19] Krakau I. Trends in use of health care services in visits can reduce the risk of potentially avoidable hos­
Swedish primary care district. A ten year perspective. pitalizations among persons with diabetes in France.
Scand J Prim Health Care. 1992;10(1):66–71. doi: 10. Eur J Public Health. 2020;30(6):1056–1061. doi: 10.
3109/02813439209014038. 1093/eurpub/ckaa137.
Copyright of Scandinavian Journal of Primary Health Care is the property of Taylor &
Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.