Minimum volume standards in German hospitals: do they get along with procedure centralization? A retrospective longitudinal data analysis

Background Compliance with minimum volume standards for specific procedures serves as a criterion for high-quality patient care. International experiences report a centralization of the respective procedures. In Germany, minimum volume standards for hospitals were introduced in 2004 for 5 procedures (complex esophageal and pancreatic interventions; liver, kidney and stem cell transplantations), in 2006 total knee replacement was added. This study explores whether any centralization is discernible for these procedures in Germany. Methods A retrospective longitudinal analysis of secondary data serves to determine a possible centralization of procedures from the system perspective. Centralization means that over time, fewer hospitals perform the respective procedure, the case volume in high-volume hospitals increases together with their percentage of the annual total case volume, and the case volume in low-volume hospitals decreases together with their percentage of the annual total case volume. Using data from the mandatory hospital quality reports for the years 2006, 2008 and 2010 we performed Kruskal Wallis and chi-square tests to evaluate potential centralization effects. Results No centralization was found for any of the six types of interventions over the period from 2006 to 2010. The annual case volume and the number of hospitals performing interventions rose at differing rates over the 5-year period depending on the type of intervention. Seven percent of esophagectomies and 14 % of pancreatectomies are still performed in hospitals with less than 10 interventions per year. Conclusions For the purpose of further centralization of interventions it will be necessary to first analyze and then appropriately address the reasons for non-compliance from the hospital and patient perspective.


Background
Minimum volumes are still deemed to be a criterion for better quality in patient care for numerous surgical interventions, since many studies and systematic reviews document better outcomes in high-volume hospitals compared to low-volume hospitals [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16]. A shift of cases from lowvolume hospitals to high-volume hospitals is a conclusion of most studies, and in one review is even quantified as number needed to treat [17]. The volume outcome relationship and the centralization of these procedures are therefore two sides of the same coin. In the United States, recommendations of the National Cancer Policy Board of the Institute of Medicine [18] or initiatives such as Leapfrog on Evidence-based Hospital Referral [19] promote centralization towards high-volume hospitals. Numerous studies also assess this incipient centralization as a result of findings concerning the volume outcome relationship. For numerous types of interventions they describe a notable centralization process in US-American health care over a period of several years [20][21][22][23][24][25]. In other countries the findings on the volume outcome relationship have fed back into the development of medical care towards centralized and specialized care structures. The effects of such planned centralization strategies are the object of studies on health systems mainly in Europe and Japan [26][27][28][29][30][31][32][33][34][35][36]. These studies confirm without exception the positive correlation between high case volume and better treatment results. Nevertheless there is still some controversial debate on the pros and cons of centralization for specific interventions [37,38]. Effects of prevalent centralization on patients are also a topic for research, with a focus on longer travel time for most and possible impairment of social support [22,[39][40][41][42][43].
Mandatory minimum volumes for certain defined surgical procedures were introduced in Germany in 2004 on a federal level. To date there are no studies into the implications of a possible centralization of care for these interventions. The minimum volume standards on the federal level that are mandatory for all approximately 2000 German acute-care hospitals apply to 6 types of surgical procedures, (1) complex esophageal interventions, (2) complex pancreatic interventions, (3) total knee replacement (TKR), and transplantations of (4) kidney, (5) liver and (6) stem cells. Between 2004 and 2006 these standards were raised in part and since then have remained the same (Table 1). Pertinent key figures for surgeries and procedures for each intervention are exactly defined in minimum volume regulations [44]. In Germany the Federal Joint Committee decides on minimum volume regulations. Represented in this body are the major actors of the health system, health insurance funds, physicians and hospitals; they form the joint selfgovernment of the German health care system. Minimum volume regulations do not imply any scheduled structural revision of patient care with regard to the specific interventions, such as a systematic formation of centers and centralization. If hospitals perform minimum volume procedures they are obliged to register the case numbers in their quality reports as the only format to document compliance with minimum volume regulations.
Quality reports are standardized nationwide, and all hospitals are legally required to publish their quality reports following the regulations on nature, scope and data format as defined by the Federal Joint Committee [45]. The quality report of each hospital in each reporting year consists of two data sets. One data set comprises structural and process data of the hospital, which are to be reported by the individual hospital following the regulations. These reported data are completely selfreported. A second data set for each hospital provides data on quality indicators as reported by each hospital to the official federal external quality assurance institution. These quality indicator data are validated in a structured manner including different statistical checks and a random sample of hospitals where data are verified during an on-site inspection each year. The data on minimum volume standards belong to the first data set without an external validation procedure. Hospital reports are publicly available on the Internet. An electronic data set of each reporting year can be obtained on application from the Federal Joint Committee. Reports on every second year of operation were mandatory from 2004 to 2012, and annual mandatory reports were introduced in 2013. Hospital quality reports for 2012 have been published in the summer of 2014.
In a preliminary evaluation of the consequences of the introduction of minimum volumes in German hospitals the authors reported earlier how many German hospitals performed the 5 procedures stipulated for 2004, how many complied with minimum volume standards, and indicated possible implications for geographically equal access to health care if all non-complying hospitals were excluded [46][47][48][49]. In 2014 the authors published a study updating the results on the research question how many hospitals complied with the minimum volume standards

Methods
The retrospective longitudinal analysis of secondary data is based on mandatory hospital quality reports for the years 2006, 2008 and 2010. Minimum volume data reported by hospitals are compared separately for the 6 types of interventions in the three years under review. Centralization is operationalized for the purpose of this study as change over time in three respects: (1) the number of hospitals performing the respective procedure decreases over time; (2) the volume, i.e. number of cases, in high-volume hospitals and their percentage in the annual entire volume increase over time; (3) the volume in low-volume hospitals and their percentage in the annual entire volume decrease over time.
Hospital data on minimum volume procedures were exported from quality reports to EXCEL files for each year under review, whereby the number of performing hospitals and cases were descriptively registered for each minimum volume procedure and each year under review. For a comparison of annual volumes the hospitals were divided in quintiles (esophagus, pancreas, total knee replacement), and into tertiles for the three transplantation procedures. Hospitals were ranked according to number of performed procedures per type of intervention and year, and divided into exact quintile and tertile groups. A possible change in case distribution due to changing annual volumes and in the number of performing hospitals was taken into account in the formation of quintile and tertile groups by individual new rankings of groups for the years 2006, 2008 and 2010 respectively.
Volumes of quintile and tertile groups were analyzed statistically for significant changes within each quintile and tertile over the three years under review (intra quartile / intra tertile differences). Subsequently the relative percentage ratios between quintiles / tertiles of one year were analyzed over the three one-year periods (inter quartile / inter tertile differences). The intra quartile / intra tertile volume change over the three years was submitted to test-statistical analysis using the Kruskal Wallis test for independent, non-parametric testing of samples, since the ranked quintile and tertile groups were assumed to be not normally distributed. Test-statistical analysis of inter quartile/inter tertile differences in case numbers over the three years under review was checked with a chi-square test for each minimum volume. The statistics software IBM SPSS Statistics Version 21 was used for test-statistical analysis.

Ethics statement
The authors declare that their study does not need ethical approval. All data used in this study are publicly available and usable due to legal obligations of hospitals to provide a high standard of transparency and offer information for patients, hospitals, health insurers, research and politics. Single hospital quality reports can be accessed via: http://www.g-ba-qualitaetsberichte.de/. Electronic data sets (XML format) of the quality reports can be requested from the Federal Joint Committee via: https://www.g-ba.de/institution/themenschwerpunkte/ qualitaetssicherung/qualitaetsbericht/xml-daten/.

Number of hospitals per minimum volume and compliance with minimum volume standards over time
The basic data on the number of hospitals and cases per year and minimum volume are listed in Tables 2 to 7 in the upper part respectively. From 2006 to 2010, the number of hospitals performing pancreatic, esophageal and total knee replacement interventions increased by 4 to 11 %, and the annual volume increased by 8 to 25 %. For all three minimum volumes the number of hospitals complying with the minimum volume standards remained constant or increased slightly. A major difference is notable in the percentage of complying hospitals between pancreatic interventions (70 %), esophageal interventions (55 %) and total knee replacement (90 %). For the volume, the percentage of cases in hospitals meeting minimum volume standards is between 85 and 99 %. The number of hospitals for stem cell and liver transplantations is constant over time, whereas a slight decrease has been registered for kidney transplantations. The percentage of hospitals complying with minimum volume standards is over 90 % for kidney and liver transplantations, but for stem cell transplantations is 75 % with a downward trend. For the cases treated in these hospitals the percentage is over 95 % respectively. Tables 2, 3, 4, 5, 6 and 7 present results for caseload changes in each quintile/tertile over the three years under review. For pancreatic interventions (Table 2), the caseloads increased significantly over time from the 2nd quintile (hospital group with the second most frequent volumes) up to the 4th quintile (hospital group with second-lowest volume); in the 1st quintile (hospital group with most frequent volume) the percentage of

Discussion
No centralization of care was found in the German hospital sector for the 6 minimum volume interventions from 2006 to 2010. The three criteria for centralizationless hospitals perform an intervention, growing case load and higher percentage of cases in high-volume hospitals, and declining case load and lower percentage of cases in low-volume hospitalsdo not apply to any of the six types of interventions. On the contrary, the percentage of hospitals performing the respective minimum volume procedures rose for four of the procedures by between What are possible reasons why no centralization occured in any of the introduced minimum volume interventions? We assume different contributing factors in each of the procedures. Liver and kidney transplantation, with 23 and 40 delivering centers nationwide respectively, were already highly centralized before introducing minimum volume standards and no further centralization was expected [46]. Stem cell transplantations are a special case in this context, since they are not subject to regulations for organ transplants as liver and kidney transplantation. Their indication is governed by dynamic developments in medicine, as may be seen from volatile figures for hospitals and case loads over the three years under study, with a tendency towards more hospitals, and cases performed without meeting minimum standards. The high reimbursements may be an additional incentive for performing such interventions. Total knee replacements are very frequent interventions, performed in every second German hospital, and have increased by 25 % within 5 years. The currently applicable minimum volume of 50 procedures per year must be considered as low to medium, based on the systematic review by Marlow [13]. The caseloads in German hospitals not meeting minimum volume standards are with 1 to 2 % -as low as only those for centralized liver and kidney transplantations. In contrast, pancreatic and esophageal interventions had the highest percentage of hospitals not complying with minimum volume standards in 2006 with 36 and 43 % respectively, and a centralization was expected [46,51,52]. Our findings on these two procedures can be contrasted by international studies quoted earlier. Numerous US studies explore the introduction of minimum volumes and the centralization processes involved for these two interventions among others. Based on a nationwide inpatient sample, Learn [20] reports that 67 % of pancreatectomies were performed in hospitals with less than 10 such interventions per year from 1997 to 1999; the percentage declined to 51 % from 2004 to 2006, comparable to esophagectomies with an initial percentage of 69 % in hospitals with less than 6 cases per year where the percentage declined to 53 %. For these two intervention categories Stitzenberg [39] also  [32] report the successful implementation of a minimum volume of 10 pancreatectomies per year, with a decrease in performing hospitals from 48 to 30, and an increase from 53 to 91 % in patients operated upon in hospitals with 10 and more interventions. Centralization criteria have been clearly met in this case. From the south of the Netherlands Gooiker [33] and Lemmens [34] give an impressive description of an inter-hospital, coordinated centralization of esophagectomies and pancreatectomies, leading to performance in 2 instead of 9 hospitals and to double-digit volumes.
These international experiences with introducing minimum volume standards for esophagectomies and pancreatectomies can help understanding the differing development in Germany. The low percentage of esophagectomies and pancreatectomies in German hospitals with less than 10 interventions per year reflects -in international comparison -a high degree of centralization existing prior to the introduction of minimum volume standards. Linked to this is a not equally promoted compliance with minimum volume standards on the part of medical societies, national specialist institutes or initiatives such as Leapfrog with an influence on reimbursement criteria in other countries. German minimum volume regulations oblige hospitals to comply but do not specify any sanctions. A 2006 survey among affected German hospitals not complying with minimum volumes for esophagectomies and pancreatectomies, and also among national umbrella organizations of all German sickness funds revealed that non-compliance has so far been irrelevant in the annual budget negotiations [53]. Another factor with impact on centralization processes in Germany is its federal structure. The 16 federal states in Germany are independently in charge of designing and implementing the structural planning of inpatient care, and not the Federal Joint Committee that defines nationwide minimum volume standards. This might be one important factor why on the state level in North-Rhine Westphalia breast cancer treatment was successfully concentrated in breast cancer centers, thereby reducing the number of hospitals with breast cancer treatment from about 250 to 100 [54]. This approach is comparable to The Netherlands where relocation of procedures as basic step toward centralization has been successfully fostered by specifying care structures and by cooperation in regional quality associations [31][32][33].
However, factors such as promoting initiatives of medical societies, reimbursement strategies, sanctions, administrative directives, and regional quality associations focus primarily on a system perspective. It might be helpful to address as well the hospital perspective to achieve stricter compliance with minimum volume standards. Especially hospitals not complying with minimum volume standards should be analysed why such procedures are performed. Factors to be studied are the relevance of and options for emergency procedures, attitudes towards and possibilities of regional cooperation and last but not least the patient view, since proximity to the place of residence and therefore to social support have a high priority for which patients are willing to accept higher risks [55].

Limitations
A limitation of this study is that the validity of data used from hospital quality reports is not quantifiable. The data are unevaluated and self-reported. It has been established that the reported data have had no practical consequences in the form of sanctions to date. But the possibility cannot be ruled out that data reported specifically from hospitals not meeting minimum volumes are inaccurate. The study does not investigate the temporal, intrahospital volume progression that would be of interest for individual hospital developments in complementing the systemic perspective of centralization, nor does it consider the influence of cooperation between hospitals on volume progression in detail. From the perspective of the individual hospital, the emergence of other regional health care structures may constitute important developments after the introduction of minimum volume standards, but has not been explored for the purposes of this study.