Co- and multimorbidity patterns in primary care based on episodes of care: results from the German CONTENT project
BMC Health Services Research
Co- and multimorbidity patterns in primary care based on episodes of care: results from the German CONTENT project
Gunter Laux 0
Thomas Kuehlein 0
Thomas Rosemann 0
Joachim Szecsenyi 0
0 Address: Department of General Practice and Health Services Research, University of Heidelberg , Germany
Background: Due to technological progress and improvements in medical care and health policy the average age of patients in primary care is continuously growing. In equal measure, an increasing proportion of mostly elderly primary care patients presents with multiple coexisting medical conditions. To properly assess the current situation of co- and multimorbidity, valid scientific data based on an appropriate data structure are indispensable. CONTENT (CONTinuous morbidity registration Epidemiologic NeTwork) is an ambitious project in Germany to establish a system for adequate record keeping and analysis in primary care based on episodes of care. An episode is defined as health problem from its first presentation by a patient to a doctor until the completion of the last encounter for it. The study aims to describe co- and multimorbidity as well as health care utilization based on episodes of care for the study population of the first participating general practices. Methods: The analyses were based on a total of 39,699 patients in a yearly contact group (YCG) out of 17 general practices in Germany for which data entry based on episodes of care using the International Classification of Primary Care (ICPC) was performed between 1.1.2006 and 31.12.2006. In order to model the relationship between the explanatory variables (age, gender, number of chronic conditions) and the response variables of interest (number of different prescriptions, number of referrals, number of encounters) that were applied to measure health care utilization, we used multiple linear regression. Results: In comparison to gender, patients' age had a manifestly stronger impact on the number of different prescriptions, the number of referrals and number of encounters. In comparison to age ( = 0.043, p < 0.0001), multimorbidity measured by the number of patients' chronic conditions ( = 0.51, p < 0.0001) had a manifestly stronger impact the number of encounters for the observation period. Moreover, we could observe that the number of patients' chronic conditions had a significant impact on the number of different prescriptions ( = 0.226, p < 0.0001) as well as on the number of referrals ( = 0.3, p < 0.0001). Conclusion: Documentation in primary care on the basis of episodes of care facilitates an insight to concurrently existing health problems and related medical procedures. Therefore, the resulting data provide a basis to obtain co- and multimorbidity patterns and corresponding health care utilization issues in order to understand the particular complex needs caused by multimorbidity.
Based upon technological progress and improvements in
medical care and health policy, a growing number of
patients survive medical conditions that used to be fatal
formerly. As a result of this, an increasing proportion of
mostly elderly primary care patients presents with
multiple coexisting medical conditions. A large number of
epidemiological studies from several countries support this
estimate [1-4]. It could be shown that the risk of avoidable
admissions and preventable complications increases
dramatically with the number of chronic conditions .
Generally, it becomes more and more important to
understand the particular complex needs caused by
multimorbidity. However, to properly assess the current
situation of multimorbidity and to draw recommendations for
improvement, valid scientific data are indispensable.
In the German health care system, the GP (general
practitioner) has some kind of gate-keeper role since patients
who visit a specialist, without visiting the GP in advance
have to pay an additional fee. In consequence, most visits
to specialists are preceded by a GP consultation. The
medical insurances cover all costs of routine care, including all
visits to GPs as well as to specialists. Only some
complementary and alternative medical treatments are not
covered. A medical un-insurance does not exist, since a
recently enacted law forces everyone to insure himself. In
Germany everyone has free and unlimited access to the
medical system and the vast majority of people uses the
GP as entry into this system.
Electronic patient records in Germany are predominantly
used for billing purposes. Thus, hitherto existing German
routine data are unlikely to yield a realistic and
differentiated picture of morbidity and health care utilization in
CONTENT (CONTinuous morbidity registration
Epidemiologic NeTwork) is an ambitious project in Germany to
establish a system for adequate record keeping and
analysis in primary care. A scientific network was established
consisting of participating surgeries, scientists, and
statisticians. The aims are strictly scientific and the underlying
hypothesis is that the knowledge-gaining process can be
accelerated by combining the experience of many,
especially with respect to complex interactions of factors and
the analysis of rare events. The CONTENT EPR 
(Electronic Patient Record) is based on the ICPC-2-R 
(International Classification of Primary Care, 2nd Revision) and
allows a documentation in an episode of care structure over
time. ICPC was accepted by the World Health
Organization (WHO) as a related classification to be used for
health information recording in primary care. The
electronic version of the 2nd ICPC edition (ICPC-2-E) is
available for the use in electronic medical records .
Generally, there is a broad consensus that ICPC exactly
meets the needs in primary care both in research as well as
in practice and will add knowledge about morbidity
patterns in this field.
An episode of care is defined as health problem from its
first presentation by a patient to a doctor until the
completion of the last encounter for it or presumably death, if
the focal problem still exists . An episode of care (in
this study) actually includes all GP encounter elements
none from specialists. Medical documentation in an
episode of care character facilitates an insight to concurrently
existing health problems and related medical procedures
(e.g. prescriptions, referrals and hospitalization) and
therefore provides a basis to obtain multimorbidity
patterns. The CONTENT database has already yielded
analyses that were impossible to achieve from German routine
health care data .
We used the term multimorbidity to describe the
cooccurrence of two or more chronic conditions as defined
by van den Akker et al. . The term comorbidity is used
to describe the co-occurrence of medical conditions
additional to an index disease as defined by Feinstein .
This study aims to describe co- and multimorbidity as well
as health care utilization based on episodes of care, taking
into account age and gender for the study population of
the first 17 participating general practices.
A software module was developed to enable the coding of
reasons for encounter, diagnoses and medical procedures
with ICPC and assigning these issues to episodes of care.
The module was integrated in an existing practice software
to be used by voluntarily participating GPs. The extended
practice software features a special function for data
export based on XML (eXtensible Markup Language). The
resulting data files are sent to the center in Heidelberg via
email or upload to a dedicated server. In addition to the
actual patient data the files contain meta data with
information about the observation period and the surgery. To
assure data quality the practices obtain feedback reports at
regular intervals containing operating figures about
episode based data entry with ICPC (e.g. the percentage of
encounters without a documented reason for encounter).
Moreover, these figures are discussed in periodic meetings
with the GPs in order to continuously improve the data
As a basic principle, only anonymized data are
transmitted. For each patient, the CONTENT EPR contains a case
number, the year of birth and the gender but not patients'
names or addresses. Thus, it is not possible to determine
a patient's identity and the implementation of extensive
data security mechanisms is not needed. Moreover, the
German Data Protection Act allows the transmission of
anonymized patient data for scientific purposes without
an explicit compliance of the patients. The study protocol
was approved by the ethics committee of the University of
Heidelberg (approval number 442/2005).
The data stem from a total of 39,699 patients in a Yearly
Contact Group (YCG) out of 17 general practices with a
total of 24 GPs located in 4 different federal states in West
Germany with a concentration in Baden-Wrttemberg
and Hessen. The YCG can be considered as an appropriate
denominator since it is a good approximation of the
"attending patients" in health systems with a patient list
ICPC and episode based data entry was performed
between 1.1.2006 and 31.12.2006. For these patients data
about age, gender and episode based diagnoses were
available as well as the corresponding medical procedures
(different prescriptions and referrals). The number of
different prescriptions per patient was determined at the 4th
level of the ATC (Anatomical Therapeutic Chemical
Classification). The 4th level determines the chemical or
therapeutic or pharmacological subgroup. This is the level
usually used to count "number of different drugs" as it is
the level which aggregates drugs just above their
descriptive chemical substance. The underlying referral list
included all referrals to specialists (incl. repeat referrals)
for the observed YCG.
On the basis of ICPC codes for the presented sample it was
possible to define chronic conditions by using the concept
of O'Halloran et al. that regards diagnoses as well as few
chronic symptoms and complaints .
In order to model the relationship between the
explanatory variables (age, gender, number of chronic
conditions) and the response variables of interest (number of
different prescriptions, number of referrals, number of
encounters) we used multiple linear regression. Odds
ratios were calculated by logistic regression. On account of
the cluster sample study design the calculations were
adjusted for the cluster (i.e. practice) on the basis of
calculated ICCs (IntraCluster-Correlations). Statistical
calculations were performed with SPSS version 14.0 and SAS
The study protocol was approved by the ethics committee
of the University of Heidelberg (approval number 442/
For 39,699 patients in the YCG a total of 76,428 different
episodes of care with an average of 1.87 0.02 chronic
conditions per patient were processed. 40.4% of the
patients were male and 59.6% were female. The average
age of the patients was 48.8 0.17 years.
To get an impression of multimorbidity regarding age and
gender we considered the average number of different
chronic conditions per patient in a YCG. Moreover, we
considered the number of different prescriptions as well
as the number of referrals per patient stratified for age and
gender (Table 1 and Figures 1A to 1C). The figures already
point in the direction that patient's age appears to play an
important role in context with the above mentioned
variables for both male and female patients.
To examine more precisely the dependence of the
response variables (number of different prescriptions,
number of referrals, number of encounters) on age,
gender and number of chronic conditions we performed a
multiple linear regression. Table 2 shows the
Generally, in comparison to gender, the influence of age
on the response variables was notably stronger, displayed
by the regression coefficient , that was standardized for
the particular range. It could be observed that the impact
of gender on the number of referrals ( = 0.053, p <
0.0001) was predominantly associated with referrals of
women to gynaecologists.
In comparison to age ( = 0.043, p < 0.0001),
multimorbidity had a manifestly stronger impact on the number of
encounters ( = 0.51, p < 0.0001) within our regression
model. Moreover, we could observe that the number of
patients' chronic conditions had a significant impact on
the number of different prescriptions ( = 0.226, p <
0.0001) as well as on the number of referrals ( = 0.3, p <
The CONTENT database also facilitates to describe the
cooccurrence of medical conditions additional to an index
disease. We wanted to describe the prevalence and the
extent of comorbidities for the following highly prevalent
chronic diseases: hypertension (K86/87), chronic
ischemic heart disease (K74-76), diabetes mellitus (T89/
90), and osteoarthrosis (L89-91). It has to be regarded
that the prevalence estimates are based on the entry of the
selected problem within the YCG in the observation
period and do not refer to patients' lifetime. Table 3 shows
that these chronic diseases feature a strong cohesion both
for male and female patients. Moreover, the highly
prevalent diseases lipid disorder (T93) and back syndrome
(L84) were associated with these diseases.
Age Group (Years) Gender Patients (%)
Average number of
chronic conditions per
patient in YCG ( SE)
Average number of
per patient in YCG ( SE)
Average number of
referrals per patient in
YCG ( SE)
As an answer to our research question we found a strong
correlation between age, gender, multimorbidity and
health care utilization. These findings are not surprising
and do not stand in contrast to comparable findings in the
international scientific literature [1-4,13]. Nonetheless,
our study is the first approach to this phenomenon in our
country with an international classification developed for
Generally, when addressing multimorbidity issues in
order to compare the results of different studies, possible
differences concerning the research question, the data
sources and the definition of multimorbidity have to be
taken into account. Fortin et al. collected prevalence
estimations of multimorbidity in Europe, the Middle East,
the United States, and Canada. Since research questions,
information collection, and multimorbidity measures
differed, major differences in the results were observed .
However, there is a broad consensus that multimorbidity
and its high prevalence is an important issue in family
practice that deserves more scientific research .
Van den Aker et al. concluded that multimorbidity,
although it increases with age, is a frequent phenomenon
among all ages . This phenomenon was also observed
within our study sample. 12.8% of the patients younger
than 50 years featured 2 or more chronic conditions.
Therefore, research into multimorbidity should not only
focus on the elderly, who are especially at risk.
Multimorbidity as defined by routinely collected data in
electronic patient records not only offers an
epidemiological overview of morbidity patterns for the scientist, but
can also help the GP to identify patients with an increased
likelihood of needing more attention . We observed a
typical clustering of specific health problems (e.g.
diabetes, hypercholesterinemia and hypertension, Table 3).
These clusters can be easily identified by the GP on the
basis of the EPR in order to apply an appropriate medical
Proportion of explained Adjusted regression coefficient
variance (adjusted R2)
41/ 8/ 3696- 2741/ moc.l art necde moi b. www//: ptt h
H D h s c
care and to initiate specific interventions (e.g. lifestyle
Generally, a potential selection bias must be admitted
since the GPs' participation is voluntary and by now
mainly focuses on Southwest Germany. Moreover, the
number of 17 practices is still too small to draw strong
In order to assess morbidity, there are several detailed and
validated morbidity indexes . For example, the
"Cumulative Illness Rating Scale" (CIRS)  index
additionally regards the severity of each condition and was
also validated for the use to quantify multimorbidity for
primary care patients . However, since we had no
information of the condition severity within the
CONTENT EPR, we could not calculate this index for our study
and had to limit on disease counts. Moreover, it would
have been challenging to analyse the influence of
sociodemografic factors (e.g. education, profession, income) on
multimorbidity. However, sociodemografic information
was only available for a small fraction of the sample.
The definition of a specific chronic condition on the basis
of ICPC codes is often ambiguous. For example, we
defined Osteoarthrosis (OA) by inclusion of ICPC codes
L89 (Osteoarthrosis of hip), L90 (Osteoarthrosis of knee)
and L91 (Osteoarthrosis, other). L84 (Back syndrome
without radiating pain) is not included in our selection
but also includes OA of the back. However, L84 also
includes diseases that are not related to OA (e.g. back
strain). Moreover, L91 includes 'arthritis unspecified' and
'traumatic arthropathy' that are not directly related to OA.
This general problem could be solved by using a more
specific terminology level which would allow grouping of
all osteoarthrosis (no matter the site) from all applicable
As mentioned above, the CONTENT project is the first
approach in Germany based on episodes of care and ICPC
that facilitates detailed long term analyses of co- and
Especially, the continuous registration of patients'
presented symptoms is new in comparison to hitherto
existing German EPRs. Thus, CONTENT data enable to analyse
the correlation between presented symptoms and
resulting diagnoses in consideration of existing comorbidities.
Moreover, age, gender as well as seasonal and regional
differences have to be taken into account. In the long run, for
every ICPC symptom (SY) it will be possible to determine
a list L of resulting diagnoses D1,....., Dn and
corresponding probabilities P1,....., Pn taking into account the above
L(SY ; A, G, S, R, Ci) =
Dn : Pn
mentioned constraints (A: age, G: gender, S: season, R:
region, C1,....., Cm: existing comorbidities), as the
following formal description shows:
This detailed model represents an extension of the model
presented by Lamberts et al. .
We could observe a strong correlation between age,
gender, multimorbidity and health care utilization in our
study sample. Generally, documentation in primary care
on the basis of episodes of care facilitates an insight to
concurrently existing health problems and related
medical procedures. Therefore, the resulting data provide a
basis to obtain multimorbidity patterns and
corresponding health care utilization issues. The continuously
growing number of patients and practices has the potential to
facilitate detailed long term analyses of co- and
The increasing application of ICPC- and episode-based
EPRs all over the world will allow challenging
international comparisons in order to see national differences
and regional distinctions and to discover what is generic
in family practice and independent from local or national
conditions. Further analyses will subsequently be based
on the continuously expanding database and have the
potential to shed light on complex epidemiological and
health economics research questions.
The author(s) declare that they have no competing
GL conceived and designed the study, organized and led
data collection, analyzed and interpreted the data, and
drafted the manuscript. TK assisted in data interpretation.
TR assisted in study design and critically reviewed the
manuscript. JS conceived and potentiated the
superordinated project CONTENT. All authors read and approved
the final manuscript.
The cooperation of the participating family doctors is greatly
acknowledged. The superordinated project CONTENT was financed by the
German Ministry of Education and Research (Bundesministerium fr Bildung
und Forschung, BMBF, grant number 01GK0301).
1. vd Akker M , Buntinx F , Metsemakers JF , Roos S , Knottnerus JA : Multimorbidity in general practice: prevalence, incidence, and determinants of co-occurring chronic and recurrent diseases . J Clin Epidemiol 1998 , 51 : 367 - 375 .
2. Fortin M , Lapointe L , Hudon C , Vanasse A , Ntetu AL , Maltais D : Multimorbidity and quality of life in primary care: a systematic review . Health Qual Life Outcomes 2:51. 2004 Sep 20 ;
3. Schellevis FG , vd Velden J, vd Lisdonk E , v Eijk JThM , v Weel C : Comorbidity of chronic diseases in general practice . J Clin Epidemiol 1993 , 46 : 469 - 473 .
4. Wolff JL , Starfield B , Anderson G : Prevalence, expenditures, and complications of multiple chronic conditions in elderly . Arch Intern Med 2002 , 162 : 2269 - 2276 .
5. Laux G , Koerner T , Rosemann T , Beyer M , Gilbert K , Szecsenyi J : The CONTENT project: a problem-oriented, episode-based electronic patient record in primary care . Inform Prim Care 2005 , 13 : 249 - 255 .
6. ICPC-2-R: International Classification of Primary Care . Revised Second Edition . Oxford: Oxford University Press ; 2005 .
7. Okkes IM , Jamoulle M , Lamberts H , Bentzen N : ICPC-2-E. The electronic version of ICPC-2. Differences with the printed version and the consequences . Fam Pract 2000 , 17 : 101 - 106 .
8. Laux G , Rosemann T , Krner T , Heiderhoff M , Schneider A , Khlein T , Szecsenyi J : Detailed data collection regarding the utilization of medical services, morbidity, course of illness and outcomes by episode-based documentation in general practices within the CONTENT project . Gesundheitswesen 2007 , 69 ( 5 ): 284 - 91 .
9. Feinstein A : The pre-therapeutic classification of co-morbidity in chronic disease . J Chron Dis 1970 , 23 : 455 - 68 .
10. De Loof J : Practice size. A fraction of the yearly attending group as practice size indicator . Allg Med Int 1983 , 12 : 127 - 128 .
11. Krogh-Jensen P : Estimation of the practice population . The denominator in health systems with fee for services . Allg Med lnt 1983 , 12 : 129 - 34 .
12. O'Halloran J , Miller GC , Britt H : Defining chronic conditions for primary care with ICPC-2 . Fam Pract 2004 , 21 : 381 - 86 .
13. Schellevis FG , Van de Lisdonk EH , Van der Velden J , Hoogbergen SH , Van Eijk JT , Van Weel C : Consultation rates and incidence of intercurrent morbidity among patients with chronic disease in general practice . Br J Gen Pract 1994 , 44 : 259 - 262 .
14. Fortin M , Bravo G , Hudon C , Vanasse A , Lapointe L : Prevalence of multimorbidity among adults seen in family practice . Ann Fam Med 2005 , 3 : 223 - 8 .
15. Fortin M , Lapointe L , Hudon C , Vanasse A : Multimorbidity is common to family practice: is it commonly researched? Can Fam Physician 2005 , 51 : 244 - 5 .
16. Kadam U , Croft P : Clinical multimorbidity and physical function in older adults: a record and health status linkage study in general practice . Fam Pract 2007 in press.
17. de Groot V , Beckerman H , Lankhorst GJ , Bouter LM : How to measure comorbidity. a critical review of available methods . J Clin Epidemiol 2003 , 56 : 221 - 29 .
18. Linn BS , Linn MW , Gurel L : Cumulative illness rating scale . J Am Geriatr Soc 1968 , 16 : 622 - 626 .
19. Hudon C , Fortin M , Vanasse A : Cumulative Illness Rating Scale was a reliable and valid index in a family practice context . J Clin Epidemiol 2005 , 58 : 603 - 608 .
20. Lamberts H , Oskam SK , Okkes IM : The clinical relationship between symptoms and the final diagnosis in general practice, determined by means of posterior probabilities calculated on the basis of the Transition Project . Ned Tijdschr Geneeskd 2005 , 149 : 2566 - 72 .