Does hospital volume affect outcomes after abdominal cancer surgery: an analysis of Swiss health insurance claims data

Providing high quality treatment is a key goal in cancer care and is characterized by complex interaction of clinical judgment, surgical skills and multidisciplinary care [1, 2]. Over the last decades, quality of oncological care was enhanced by progresses in the understanding of cancer biology, pathology, radiation oncology, interventional radiology and systemic therapy in combination with the rapid advances of molecular analysis and imaging techniques [3, 4].

Numerous international studies have examined the volume-quality relationship for high-risk abdominal cancer surgery [5‐23] and the large majority showed clear, beneficial effects of increasing volume on outcomes (primarily mortality and length of stay). At the structural level, high volume hospitals are assumed to contribute to better patient outcomes by offering expertise through multidisciplinary teamwork and enhanced implementation of evidence-based guidelines [24‐26]. At the individual level, a surgical team is able to minimize blood loss and operative time as well as intra- and postoperative complications by gaining experience [9, 20, 27‐29]. Studies indicating a positive relationship between hospital volume and treatment quality provide the scientific fundament for minimum case numbers introduced as part of quality control in inpatient care in Switzerland [30‐32]. Hence, hospitals must demonstrate a defined number of operations in order to be awarded service contracts. However, the applicability of previous studies on volume and outcome to current practice is controversial. First, many studies on volume and outcome are outdated. Given that surgical mortality for many procedures has decreased substantially since these studies were conducted [33‐38], the relative impact of the volume effect of procedures may have decreased. Second, studies showing strong association between hospital volumes and better outcomes have been predominantly conducted in larger countries (i.e., US or UK). However, these results are not easily transferrable to healthcare systems with relatively small geographic units, small patient numbers and / or a small-spatial dispersion of hospitals. Consequently, the focus of our work was 1) to describe distribution and temporal changes in hospital volumes regarding colon, gastric, rectal and pancreatic resections between 2014 and 2018 in Switzerland and 2) to investigate whether high volume hospitals perform better with respect to postoperative mortality, length of hospital stay and cost.

Throughout the entire observation period, colon resections were the most common interventions (n = 1′690, 59%), whereas gastric resections were the least frequent procedures (n = 188, 7%). Over the observation period, the yearly average number of performing hospitals was highest for colon resections (n = 93), followed by hospitals for rectal resections (n = 78) and gastric resections (n = 62), with the lowest average number of performing hospitals for pancreatic resections (n = 35). Of all resections, 4% lead to death in the hospital, another 12% within twelve months after hospital discharge and 84% were not fatal after 12 months. Of the 2′859 abdominal resections, 47% were performed on women. The mean age of operated patients was 71 years in men and to 72 years in women. Table 1 depicts the patient characteristics of the study population.

This study gives further insight into the link between hospital volumes and outcomes based on real-life data of patients with four different types of highly incident abdominal cancer.

The first result is that during the observed time period, a concentration of service provision could be observed for colon, gastric, and pancreatic cancer resections. As second result, we found a less clear and statistically insignificant relationship between increasing hospital volume and better outcomes as opposed to most previous studies.

Our analysis showed no clear relationship between hospital volume and mortality. In general, the present analysis is based on low hospital volumes compared to other countries and very low mortality rates (Table 1). This probably hampered the detection of significant differences. For example, Nimptsch & Mansky 2017 showed a negative association between hospital volume and hospital mortality for resections of the colon in Germany [45]. This effect is shown with convincing significance only from about 90 cases per year compared with the reference category of about 20 cases. Hospital volumes in the present study are remarkably smaller, and hospitals with 90 cases per year are at the upper limit of the hospital volume distribution. Another study from England shows an effect in gastric cancer of ≥80 versus < 20 cases per year [11]. Our hospital volumes range up to a maximum of 33 cases within this service provision area, well below the hospital volumes in the Coupland et al. study.

The observed concentration of service provision is consistent with the existing literature in the Swiss context [7, 30, 44]. Hence, a Swiss study makes clear that the concentration of medical services is almost exclusively observed in services with defined minimum case numbers and in the areas of highly specialized medicine (HSM) [44]. HSM concerns the area of rare, complex and expensive interventions and therapies (e.g. severe burns in children) [46].

The cited study is in line with our findings, as resections of abdominal cancer are linked to minimum case numbers and have been part of HSM since 2016 [47].

Such concentration trends in hospital care are a double-edged sword. From an economic point of view, higher hospital volumes can be advantageous, e.g., due to purchasing advantages, increased capacity utilization, and lower average cost (economies of scale) [43]. On the contrary, a concentration of treatment cases at fewer sites might be detrimental to accessibility. Despite this potential negative effect, a comprehensive analysis commissioned by the FOPH rated hospital accessibility as “very good” in 2016 [44].

A beneficial experience effect on treatment outcomes is documented in a large body of the literature. For instance, Güller et al. showed a significant negative effect of hospital volume on in-hospital mortality rate after esophageal, gastric, rectal, and pancreatic resection [7]. The discrepancy between the Güller study and our results may be explained by a different observation period (1999–2012) as compared to our study (2014–2018), different patient identification methods (ICD-10) as compared to our study (IQI-specifications), and a more rigorous and comprehensive control for patient-sided variables that are likely to highly correlate with outcome in our study. Similarly, Krautz et al. showed a clear correlation between hospital volume and in-hospital mortality and postoperative complications for pancreatic surgery in Germany [40]. However, the relationship between incident abdominal cancer cases and the number of hospitals is markedly different in Germany as compared to Switzerland. Thus, the incident abdominal cancer cases are 150 times higher in the German study (5′403/year) as opposed to our study (45/year), whereas the average number of hospitals in one observation year is 5 times higher (178 vs. 35). Furthermore, a systematic review based in 19 studies conducted by Giwa et al. indicates reduced postoperative length of stay with increasing hospital volume [48]. However, the results are not directly comparable to our findings, as Giwa and colleagues included studies that were subject to different contextual conditions, such as a higher overall incidence of abdominal cancer resections and higher hospital volumes. There are two possible scientifically supported explanations for the gradual hospital volume-quality relationship: At the individual level, surgeons can reduce blood loss, operative time, and intraoperative and postoperative complications by gaining more experience in the field [20, 27, 49, 50]. From a structural perspective, hospitals with high volumes can contribute to better postoperative outcomes by providing expertise through multidisciplinary teams [12, 24, 28]. Studies have shown that collaboration between radiation oncologists, radiologists, medical oncologists, gastroenterologists, interventional radiologists, and pathologists is critical for the treatment of colon, rectal, gastric, and pancreatic cancers [2, 50].

Our finding that small hospitals can demonstrate comparable outcomes to larger hospitals is in line with individual studies. Thus, there is literature suggesting no significant differences in mortality between hospital volume categories in the first month after surgery [22, 23]. In line with our findings, further studies observe no relationship between hospital volume and postoperative morbidity [41‐46], indirect proxy of corresponding cost [51‐56]. To the best of our knowledge, there are no studies available so far that would allow an adequate comparison of the association between hospital volume and postoperative cost in the Swiss context. There are also research-based explanations for the finding that smaller hospitals can demonstrate similar results to larger hospital entities: several studies make clear that not necessarily only the volume per se, but rather the degree and extent of clinical infrastructure, resources and expertise of hospitals are decisive for treatment success [22, 23]. Thus, smaller regional hospitals with small volumes but excellent facilities and care processes can ensure equally good outcomes as hospitals with larger volumes. Our results imply that it is unlikely that patient outcomes will systematically improve if only unidimensional measures such as minimal hospital volumes are implemented, at least in healthcare settings with comparably low case numbers and a high dispersion of hospital infrastructure. Although this approach allows differentiating groups of hospitals with different levels of service provision, individual hospitals with low hospital volumes may also achieve outcomes above average outcomes. Therefore, switching from a categorization of hospitals based on hospital volumes to a categorization based on outcome quality could avoid this issue, which has already been shown to be successful in a Dutch study [57].

The results of this study should be interpreted in the context of a few important limitations. Firstly, the IQI system is not static and its specifications are constantly being further developed. Thus, in the given observation period, the versions for abdominal cancer have also changed, which might also contribute to temporal changes of hospital volumes to a small extent. Nevertheless, these diagnosis specifications provide an accurate categorization of the patient population and determination of hospital volumes with only little variation.

Secondly, compared to other countries, Switzerland faces generally low hospital volumes. This is particularly the case for gastric resections. Due to the relatively recent introduction of the DRG system and initial data inconsistencies, on which the given IQI specifications are based on, the observation period could, however, not be further extended. Thirdly, hospital volumes provided by the FOPH were aggregated at the hospital level, although some hospitals include multiple sites (e.g., Hirslanden Group, Insel Group). Rather, it is recommended to summarize hospital volumes at the hospital unit level to approximate the experience of a hospital unit more accurately. Fourthly, a limitation of the present investigation arises from information, which cannot be ascertained from the database, such as surgeon experience, presence of multidisciplinary teams, or existing infrastructure. It is therefore not possible to determine to what extent these factors may have influenced our analysis. To the best of our knowledge, however, the FOPH dataset is the most comprehensive, national dataset containing relevant information around hospital volumes in Switzerland, despite those minor shortcomings.

In summary, the results indicate concentration trends of the hospital landscape, however, they lack evidence for a hospital volume-outcome relationship with respect to individual treatment outcomes. Thus, focusing on minimal case numbers as a single regulatory instrument for improving outcomes in cancer surgery is not efficient in healthcare settings with low absolute incident case numbers, high dispersion of hospital infrastructure and small-scale regional entities. Future studies should additionally take into account supply-related factors (i.e., the accessibility of hospitals, the availability of medical staff, medical equipment) to shed light on which further aspects affect treatment quality in hospitals, allowing for optimal hospital planning.