Keywords breast cancer - certification - outcome quality - health care research - health care-related
data
Introduction
Breast cancer is the most common cancer affecting women in Germany. According to the
Robert Koch Institute, 71375 new cases with disease were recorded in 2019 (ICD-10
C50 Malignant neoplasm
of breast), and the annual standardized incidence rate (per 100000 persons, ESR) was
114.6 for women and 1.2 for men [1 ].
In addition, around 6000 new cases of carcinoma in situ (DCIS) are diagnosed every
year. This means that almost every 8th woman will develop breast cancer in her lifetime.
The outcomes for this
very large population are therefore highly relevant for the overall population, both
economically and in terms of health-care policies. The current 5-year overall survival
rate is 88% for women
and 84% for men [2 ]. Mortality rates of breast cancer patients have decreased continuously since 1990,
especially for women
between the ages of 55 and 69 years [2 ]. This is also the age group who experienced the introduction of comprehensive
mammography screening, which has led to fewer advanced tumors and more small tumors
and carcinomas in situ being detected compared to the era before general screening
[2 ].
Advances in breast cancer therapy have meant that the chances of long-term survival
have increased significantly. The last decade, in particular, has seen the introduction
of personalized
targeted multimodal therapies which take account of distinct tumor biologies. The
choice of systemic therapy used to treat breast cancer depends on the intrinsic subtype.
HER2-neu receptor
status, (steroid) hormone receptor status, grading and even the proliferation marker
Ki-67 are important factors when choosing an individualized therapeutic approach.
Prognostic and predictive
factors determine the choice of therapy. In addition to tumor biology, the stage of
disease is highly relevant for survival. There has been significant progress not just
in the treatment of
early breast cancer but also in the treatment of metastatic disease. Targeted therapies
often allow disease to become chronic. In addition to prolonging survival, the goal
is also to improve
patients’ quality of life by offering therapies which patients tolerate well.
In Germany, certification programs were set up to implement the goals of the National
Cancer Plan. The concept was that cancer patients would receive high-quality evidence-based
therapy in
accordance with current guidelines which would improve patient survival. In Germany,
organ-specific certification programs are mainly run by the German Cancer Society
(Deutsche Krebsgesellschaft e. V. , DKG). Centers can be DKG-certified irrespective of whether they are also ISO-certified
(e.g., DIN EN ISO 9001), and the two certification types should
not be confused, as ISO does not evaluate content-related oncological quality parameters.
The DKG introduced certification of breast cancer centers in 2003 [3 ]. There are currently 248 DKG-certified breast cancer centers in Germany [4 ]. Earlier studies reported varying results with regards to the outcomes of patients
treated in certified breast cancer centers compared to treatment in non-certified
hospitals [5 ]
[6 ]
[7 ]. But these early publications had certain limitations such as regional restrictions,
limited time frames, and low numbers of cases.
The analysis we present here investigated differences in the survival of patients
treated in DKG-certified breast cancer centers compared to patients cared for in non-certified
hospitals in
a large patient population. It was hypothesized that patients would benefit from treatment
in certified centers.
Material and Methods
WiZen study
The cohort study on the effectiveness of care in certified cancer centers (German
title: Wirksamkeit der Versorgung in onkologischen Zentren , WiZen) examined
whether treatment in DKG-certified cancer centers offered benefits with regards to
the survival of patients with different malignancies compared to patients treated
in non-DKG-certified
hospitals in Germany. The project received financial support from the Innovation Fund
of the German Federal Joint Committee (grant no. 01VSF17020). The study evaluated
patients with breast
cancer, colorectal cancer, pancreatic cancer, lung cancer, prostate cancer, tumors
of the head and neck, brain tumors and gynecological tumors. The results for breast
cancer patients are
presented below.
Data base
The used data were obtained, firstly, from the anonymized billing data of mandatory
health insurance companies (GKV) collected across all of Germany for persons insured
with the AOK in the
period 2006–2017 and were provided by the Scientific Institute of the AOK (Wissenschaftliches Institut der AOK , WIdO). Performance data and master data from the
subsets “master data of insured persons” in accordance with Sec. 284 SGB V, “outpatient
care” (Sec. 295 SGB V), “inpatient care” (Sec. 301 SGB V) and “computerized physician
order entries”
(Sec. 300 Para. 1 SGB V) were brought together for analysis. A phase without diagnoses
from 2006–2008 was used to determine the cancer incidence, leaving the period from
2009–2017 for
analysis.
Anonymized data were also obtained from clinical cancer registries (KKR) in Brandenburg,
Dresden, Erfurt, and Regensburg. The pooled datasets included initial diagnoses made
in the period
2006–2017 as well as personal information and disease-specific data. To allow these
data to be compared with the results of the analysis of the GKV data, the evaluated
cohort was limited to
the diagnostic years from 2009 to 2017.
Structural characteristics of hospitals obtained from publicly available structured
quality reports in accordance with Sec. 136 SGB V and data on the DKG-certification
of hospitals
including the date of certification were used in addition to the data from the GKV
and the KKR. The intervention group for the results presented here consists of patients
treated in centers
certified by the German Cancer Society (centers for specific cancer types and oncological
centers).
Data were anonymized at patient and hospital levels, and data transfers were encrypted.
The pseudonymization at both levels was carried out by the WIdO and the cancer registries
which
provided data, and the subsequent evaluations were carried out in the Center for Evidence-based
Health Care (ZEGV) of Hochschulmedizin Dresden and the Regensburg Tumor Center (TZR)
of the
Center for Quality Assurance and Care Research of Regensburg University. The WiZen
study was approved by the Ethics Commission of the TU Dresden (reference number: EK95022019)
and was
registered with ClinicalTrials.gov (ID: NCT04334239). Data processing and data analysis
was done in accordance with the Declaration of Helsinki and the General Data Protection
Regulation of
the European Union.
Inclusion and exclusion criteria
Patients who were at least 18 years of age at diagnosis and who received an initial
diagnosis of breast cancer (ICD-10-GM: C50, D05; cf. [8 ]) in the years 2009–2017 were included in the study. The choice of ICD-10 numbers
was decided by a panel of clinical experts. Patients whose
date of initial diagnosis was identical with their death date and patients where information
on confounders was lacking or implausible were excluded. When reviewing the GKV data,
patients who
were not continuously insured by the AOK or who did not have an inpatient primary
diagnosis (ICD-10-GM) of the investigated entity or who had an index treatment in
a hospital within one year
before the hospital received DKG certification were excluded. An index treatment was
defined as the first entity-specific inpatient treatment for a primary or secondary
diagnosis of the
respective entity.
Endpoints
Primary endpoint was the overall survival time from the date of the index treatment
(for GKV data) or initial diagnosis (for KKR data). The date of initial diagnosis
was the date of the
first histological confirmation of disease, excluding diagnoses which were registered
as recurrences (KKR). Survival times of patients without a death date or with a death
date after 2017 were
treated as right censored data up until the end of the observation period on 31 December
2017. If the last known dataset for a patient with no death date in the KKR data included
the
information that the patient was still alive at the end of 2017, this was used for
right censoring. Mean follow-up time was estimated using the reverse Kaplan-Meier
method [9 ].
Relocations of patients did not affect the completeness of the GKV data as these data
are collected nationally. With regards to the cancer registries, when a patient moves
from one
catchment area into the catchment area of another registry, the patient’s clinical
data is transferred to the other registry as part of the general exchange of data
between registries.
Recurrence-free survival was another endpoint. Other events in addition to death were
local, regional, and distant metastatic recurrence. If events occurred in succession,
the first
recurrence event was used. Clear identification of recurrence events was only possible
for KKR data, meaning that recurrence-free survival was only calculated for KKR data.
Intervention
An intervention was defined as treatment in a DKG-certified center. Patients whose
initial treatment was carried out in a center which was already certified at the time
of treatment were
defined as the intervention group and patients treated in a non-certified hospital
were the control group. This constituted so-called complex intervention [10 ]. The date of initial treatment, if recorded, was defined as the date of resection
for the primary diagnosis of the respective entity,
otherwise it was the date of the patient’s first stay in hospital. For the KKR data,
the DKG certification status of the treating hospital at the time of the initial diagnosis
was used if the
code of the institution had been recorded. Otherwise, we used the case-related variable
“treatment in center yes” used by all the registries, which is used in audit evaluations
provided by the
registries on regularly collected performance indicators. For hospitals groups and
hospitals spread across several locations, all hospitals/locations were given the
status of DKG-certified
center if one of the institutes had the status, as it was not possible to directly
assign the certification status to a specific institution.
Risk adjustment
Age group, sex, year of diagnosis or of index treatment and severity of disease were
included in the risk adjustment of estimated center effects on patients as they were
considered
influencing variables. For the GKV data, disease severity was operationalized using
the variables “distant metastasis,” “additional oncological disease” and “comorbidities.”
For the KKR data,
disease severity was operationalized using the following variables: invasive carcinoma
vs. carcinoma in situ, stage (UICC), grading, lymph node/vascular invasion, hormone
and HER2/neu receptor
status. The entity-specific choice of comorbidities was done based on the comorbidities
defined by Elixhauser et al. [11 ] and clinical expertise. For the GKV data, hospitals were classified according to
number of beds, function as a university hospital and/or teaching hospital as well
as the
hospitals’ funding bodies in structured quality reports. Different models with gradually
increasing numbers of variables were created. Only completely adjusted models (i.e.,
including all
possible variables) are presented here. A complete overview of all risk adjustment
variants is provided in the final report [8 ].
Statistical evaluation
To estimate center effects while taking the effects of possible explanatory variables/confounders
into account, overall survival was modelled using multivariable Cox regression analysis
and the calculated hazard ratios including 95% confidence intervals were recorded.
By including a random effect for hospitals when using the GKV data, the Cox models
also showed possible
correlations for patient outcomes in hospitals [12 ]. These models are referred to as shared frailty Cox models.
Results
Description of investigated population
The sample consisted of 143720 (GKV data) or 59780 (KKR data) patients with breast
cancer, who were treated in 1010 hospitals (280 DKG-certified cancer centers, 730
not DKG-certified
centers). 63.5% (n = 91269, GKV data) or 66.7% (n = 39859, KKR data) of patients were
treated in DKG-certified breast cancer centers ([Table 1 ], [Fig. 1 ]). No significant difference was found between certified and non-certified institutions
with
regards to patient characteristics (age, sex, clinical characteristics). However,
the percentage of unknown values in the records of the KKR for certified centers was
consistently lower than
for non-certified institutions. Larger institutions were more likely to be certified
than small hospitals. When considering whether DKG-certified institutions offer a
possible survival
benefit, the different characteristics of the hospitals also need to be considered.
Table 1
Analysis populations (breast cancer – C50/D05), certified/not certified, number and
percentages of attribute groups for all investigated entities and data
sources.
Unit of observation
Attribute
GKV data
KKR data
Treated in center
Yes
No
Yes
No
– = Value could not be determined for the respective data source; n.s. = not specified
Patients
Total (n)
91269
52451
39859
19921
Age 18–59 y (%)
33.4
30.2
42.9
39.5
Age 60–79 y (%)
51.1
49.6
46.3
47.3
Age 80+ y (%)
15.6
20.2
10.7
13.2
Female sex (%)
99.1
98.8
99.2
99.0
Distant metastasis C78/C79 (%)
13.4
11.6
7.2
6.7
In situ D05 (%)
8.7
7.1
9.3
8.4
Secondary oncological disease (%)
16.3
15.9
–
–
Grading G1/2 (%)
–
–
65.1
64.2
Grading G3/4 (%)
–
–
24.6
23.4
Grading GX/n.s. (%)
–
–
10.2
12.4
Lymph node invasion L0 (%)
57.2
51.4
Lymph node invasion L1 (%)
–
–
23.3
23.8
Lymph node invasion LX/n.s. (%)
–
–
19.5
24.8
Vascular invasion V0 (%)
–
–
76.2
70.2
Vascular invasion V1/2 (%)
–
–
3.5
4.1
Vascular invasion VX/n.s. (%)
–
–
20.3
25.7
Positive hormone receptor status (%)
–
–
84.0
75.1
Negative hormone receptor status (%)
–
–
10.0
5.8
No data on hormone receptor status (%)
–
–
6.0
19.0
Overall HER2/neu status positive (%)
–
–
13.4
11.9
Overall HER2/neu status negative (%)
–
–
80.7
76.0
Overall HER2/neu status n.s. (%)
–
–
5.8
12.1
Hospitals
Total (n)
280
730
–
–
1–299 beds (%)
21.8
64.5
–
–
300–499 beds (%)
32.1
24.2
–
–
500–999 beds (%)
30.7
10.1
–
–
1000+ beds (%)
15.4
1.1
–
–
Fig. 1
Flow chart of inclusion and exclusion criteria according to data source.
The percentage of patients with breast cancer treated in DKG-certified centers in
the period 2009 to 2017 increased from 57.4% to 67.8% (GKV data) or from 59.0% to
64.4% (KKR data, [Fig. 2 ]).
Fig. 2
Percentage of patients treated in certified centers (breast cancer – C50/D05) over
time according to the data source.
Overall survival
The mean follow-up time for the total patient population was 3.4 years (median 3.1).
The mean follow-up time was 3.5 years (median 3.2) for the cohort of patients treated
in DKG-certified
centers and 3.2 years (median 2.8) for the cohort of patients treated in hospitals
which were not DKG-certified.
For both data sources, the unadjusted overall survival rate of patients with breast
cancer treated in a certified center was significantly higher than the rates of patients
who did not
have an index treatment in a certified center ([Fig. 3 ]; GKV patients treated in a certified center: 5-year survival rate = 85.5%,
95% CI = [85.2%–85.7%] compared to patients not treated in a certified center: rate = 80.6%,
95% CI = [80.2%–80.9%]; KKR: 5-year survival rate = 79.0% [78.4%–79.6%] vs. 73.7%
[72.7%–74.7%].
Fig. 3
Overall patient survival (breast cancer – C50/D05) according to center status and
data source.
Point estimate values including the confidence intervals of the adjusted hazard ratios
for center effects on overall survival were less than 1 for both the GKV data and
the KKR data ([Table 2 ]; GKV: HR = 0.77, 95% CI = [0.74–0.81]; KKR: HR = 0.88, 95% CI = [0.85–0.92]). This
means that there were significant survival
benefits for patients treated in DKG-certified centers for both cohorts (GKV data:
23%; KKR data: 12%).
Table 2
Estimated center effects (breast cancer – C50/D05) according to data source, investigated
(sub-)group and endpoint.
Data source
(Sub-)groups
Endpoint
HR
95% CI
HR = hazard ratio, CI = 95% confidence interval, p value: *p < 5%, **p < 1%, ***p < 0,1%,
DCIS = Ductal Carcinoma in Situ
1 not adjusted
2 adjusted for age, sex, distant metastasis, other oncological diseases, Elixhauser
comorbidities, number of beds in the hospital, teaching hospital, university
hospital, hospital funding body, year of index treatment – dummy-coded (GKV data)
3 adjusted for sex, age at diagnosis, year of diagnosis, ICD-10 diagnosis, UICC stage,
grading, lymph node invasion, vascular invasion, hormone receptor status, HER2/neu
receptor status (KKR data)
GKV1
Total
Overall survival
0.63***
(0.59–0.67)
GKV2
Total
Overall survival
0.77***
(0.74–0.81)
KKR1
Total
Overall survival
0.75***
(0.72–0.78)
KKR3
Total
Overall survival
0.88***
(0.85–0.92)
GKV2
Hosp. with 1–299 beds
Overall survival
0.66***
(0.60–0.73)
GKV2
Hosp. with 300–499 beds
Overall survival
0.78***
(0.73–0.84)
GKV2
Hosp. with 500–999 beds
Overall survival
0.82***
(0.76–0.88)
GKV2
Hosp. with 1000+ beds
Overall survival
0.94
(0.80–1.10)
KKR3
UICC stage 0 DCIS
Overall survival
0.80
(0.62–1.04)
KKR3
UICC stage I–III
Overall survival
0.83***
(0.78–0.88)
KKR3
UICC stage IV
Overall survival
1.02
(0.94–1.11)
KKR3
UICC stage 0 DCIS with R0 resection
Recurrence-free survival
0.97
(0.76–1.23)
KKR3
UICC stage I–III with R0 resection
Recurrence-free survival
0.78***
(0.74–0.82)
A stratified analysis which took account of the number of beds in the respective institution
(1–299, 300–499, 500–999, 1000+) was carried out for the GKV data to test for possible
effect
modifications caused by the size of the treating hospital ([Table 2 ]). With the exception of the group of hospitals with 1–299 beds,
the 95% confidence intervals of the estimated hazard ratios for center status overlapped;
however, all point estimate values were less than 1 and a possible effect modification
did not
interfere with the basic statement. The KKR data were additionally analyzed to identify
whether certification effects depended on disease severity (UICC stage) and whether
the events for
overall survival also translated to recurrence-free survival. The survival benefit
from receiving treatment in a certified center was found to be more significant for
patients with locally
limited disease and locally advanced disease (I-III) compared to patients with advanced
stage IV disease. A significant survival benefit was found for patients with stage
I-III disease
(HR = 0.89; 95% CI = [0.85–0.93]) but not for stage IV patients with primary distant
metastatic disease (HR = 1.02; 95% CI = [0.94–1.11]).
Recurrence-free survival
Recurrence-free survival was investigated in patients with R0 resection who did not
have primary distant metastasis using the KKR data. The observed effects were even
more significant than
those identified for overall survival (HR = 0.78; 95% CI = [0.74–0.82]). Subgroup
analysis of the KKR data showed comparable estimators for overall survival for the
patient cohort which
included both patients for whom data on their stage of disease was lacking and patients
for whom data about disease severity was available. Additional analysis results for
all model
specifications are available in the final report of the WiZen study [8 ].
Subgroup analysis
The estimation results of the GKV data remained robust to stratifications based on
sex (male/female), other oncological diseases (yes/no), single hospital/hospital group,
distant
metastasis (yes/no), tumor resection (yes/no) and number of hospital beds (< 500/
>= 500) (online supplement Table S1 ). The significant survival benefit observed
for the KKR cohort was confirmed in most subgroup analyses, except for those patients
who were male, patients diagnosed before the age of 50, patients with stage IV disease
or negative hormone
receptor status and positive HER2/neu receptor status ([Table 2 ], online supplement Table S1 ). The
certification effect was even more pronounced for DKG-certified centers which had
been certified for longer: while the estimated HR for breast cancer centers certified
for less than one year
was 0.82 (95% CI = [0.75–0.89]), the HR for centers certified for 5 or more years
was 0.74 (95% CI = [0.71–0.78]) (online supplement Table S2 ).
Discussion
The aim of the National Cancer Plan introduced in 2008 was to develop cancer screening
and patient orientation further to optimize oncological care structures and associated
quality
assurance measures. Previously, only regional analyses on breast cancer were available
for Germany. These regional studies suggested that treatment in DKG-certified hospitals
could be associated
with a survival benefit [13 ]. In certified institutions, structural and content-related parameters are regularly
evaluated
and quality indicators are reviewed. The goal is to provide specialized interdisciplinary
quality-oriented therapy based on current guidelines which also offers patients the
option of
participating in studies. Annual certification is associated with high personnel and
financial costs [14 ]
[15 ]; it is therefore important to have reproducible data showing that this expenditure
is justified and that the benefit is
reflected in longer overall survival rates.
A recent review showed that guideline-based therapies and the implementation of consensus
recommendations leads to better survival rates for breast cancer patients [16 ]. Overall, a higher percentage of patients with breast cancer are treated in DKG-certified
centers than patients with other oncological entities
[8 ]
[17 ].
This analysis used large data volumes to provide a representative evaluation of the
treatment situation in Germany. Our study showed that, according to the KKR data,
information on patients’
hormone receptor status (in 19.0%) and HER2-neu status (in 12.1%) were more likely
to be missing if patients were treated in non-DKG-certified hospitals compared to
DKG-certified centers. If the
reported figures represent actual diagnostic workups, then they are inacceptable because
when information about these predictive parameters is missing it is impossible to
provide the optimal
therapy.
The treatment of breast cancer is becoming ever more complex; many factors already
need to be considered during the initial diagnosis in a non-metastatic situation to
ensure that the chosen
therapy is the optimal choice for the patient. Tumor size, lymph node status, grading
and proliferation marker Ki-67 all play a role in decision-making in addition to (steroid)
hormone receptor
status and HER2/neu receptor status. When early-stage breast cancer is initially diagnosed,
a decision will have to be taken on whether primary treatment should consist of surgery
followed by
adjuvant therapy or whether – if chemotherapy is indicated – it should be carried
out as neoadjuvant therapy or whether endocrine therapy is sufficient and chemotherapy
not required as the
hormone receptor-positive patient is low risk. Patients who are HER2/neu-positive
should receive anti-HER2-targeted therapy in addition to chemotherapy [18 ]. Additional immune therapy should be considered for cases with early high-risk triple
negative breast cancer [19 ]. There are strong indications that new treatment options are implemented relatively
quickly in DKG-certified centers once they have been
included in the S3-guideline on breast cancer (e.g., trastuzumab for patients who
are HER2-positive). Further analyses of data from breast cancer centers could show
that the improved study
outcomes in terms of pathological complete remission after neoadjuvant therapy for
HER2-positive and triple negative breast cancers can be reproduced in routine treatment
[20 ]
[21 ].
Our analysis confirmed a significant certification effect for hormone receptor-positive
and HER2/neu-negative breast cancer but not for patients who were HER2/neu-positive.
In addition to making the right diagnosis, the care of patients and management of
potential toxicities is decisive. This is why further training in drug-based tumor
therapies and specialist
training in gynecological oncology was added to the advanced training of gynecologists
in 2005 [20 ]. Specialist training
in gynecological oncology does not only focus on surgical expertise but also on providing
a detailed knowledge of systemic therapies with the aim of improving patient care
[20 ].
Our analysis was able to show that overall survival of breast cancer patients treated
in DKG-certified centers was significantly longer, demonstrating that the significant
expenditure
associated with certification is beneficial for patients. It is well known that certification
entails additional costs because of the need to adapt structures and processes and
carry out audits
[22 ]
[23 ]. But the additional costs associated with
certification can also yield economic benefits as was shown by Cheng et al. in their
cost-effectiveness analysis (CEA) for bowel cancer [24 ].
It is also worth pointing out that the quality of treatment provided even in certified
centers is continually improving. This is made evident by the fact that survival benefits
increased,
the longer the center had been DKG certified.
Limitations and strengths
Selectivity of the analyzed cohort was low as documenting GKV billing data is governed
by legal regulations and recording KKR data has been compulsory since the German law
on cancer
screening and registration (Krebsfrüherkennungs- und -registergesetz , KFRG) came into effect. Even before the KFRG was passed, KKR data were almost complete.
Both
data sources included extensive patient-specific risk factors such as comorbidities
or information about disease severity. Characteristics of the treating hospitals were
included in the
analysis of the GKV data. Unfortunately, some of the influencing variables were only
available for one of the data sources, and no information was available about socioeconomic
status. Most of
the information about certification was provided directly by the DKG and is therefore
highly valid for the question investigated here. As the German federal state of North
Rhine-Westphalia
uses a different certification system, known as Äkzert, to certify breast cancer centers,
data from North Rhine-Westphalia could not be used because the definitions and approaches
differed
from those used for DKG certification. This analysis still used data from all German
federal states (which included 21843 patients who were resident in North Rhine-Westphalia;
for more
details, see the WiZen final report [8 ]). It can therefore be assumed that the estimates shown here may even
underestimate the certification effect.
As patient volumes (i.e., the number of cases receiving relevant treatment per hospital)
can have an impact on relevant outcomes such as survival [21 ]
[25 ]
[26 ] and a minimum patient volume is required for DKG certification, some of the results
presented here could be ascribed to volume effects. As the GKV data were obtained
from
a single health insurance provider and not all patients included in the KKR could
be assigned to a specific main treating hospital, it was not possible to quantify
the total volume of patients
in the various treating hospitals and include this in the analysis. Extending the
DKG certification status of individual hospitals to the entire group of hospitals
may even result in a
conservative (i.e., low in absolute numbers) estimate of the center effect.
The difference in the strength of the center effect between the two data sources was
evident despite adjustment for several confounders. This could be due to differences
in the analyzed
populations, hospitals, and data generation or to natural variation. The fact that
both data sources qualitatively showed a statistically significant effect strengthens
the robustness of the
deduced inference of a survival benefit for patients treated in DKG-certified centers.
Overall, care should be taken before making any causal interpretations of the findings.
On the one hand, certification status represents a complex arrangement of interventions
at the level
of the treating institutions which are difficult to quantify. On the other hand, it
was not possible to randomize the cohort due to the structure of the certification
system and the use of
secondary data/data from cancer registries. It was nevertheless possible to carry
out a valid examination of the effect of DKG certification by using different data
sources and including
relevant patient data, tumor characteristics, and hospital features in the risk adjustment.
This minimized the risk of bias and allowed the certification effect to be compared
across different
types of cancers.
Conclusion
Our analysis provides robust evidence that patients treated in DKG-certified breast
cancer centers have better survival rates. There are many reasons for the longer overall
survival rates.
It can be assumed that certified centers are more likely to offer individualized and
guideline-based therapies as well as the option of early access to innovative therapies
[20 ]
[21 ].
Online Supplement
Supplement Table S1: Sensitivity analysis – hazard ratio for the adjusted certification effect for subgroups
according to the basic data.
Supplement Table S2: Sensitivity analysis – ratio for the adjusted certification effect according to how
long the institution has been certified (GKV
data).
