1. Introduction
Around 30 years ago, Herzlinger [
1,
2] was the first to suggest that replacing traditional general hospitals with specialized hospitals emulating focused factories [
3] would improve quality and productivity in healthcare services. Since that time, there has been a strong emphasis on the general advantages of focus in healthcare delivery. Research has shown that focused units, which are specialized in a specific segment of patients or offering a narrow range of services, benefit from lower variability and fewer conflicting operational activities and consequently produce healthcare at a lower cost [
4], with lower mortality [
5,
6], and with a shorter length of stay [
6] than non-focused units. In practice, the number of specialized healthcare units, such as single specialty hospitals [
7,
8,
9] and ambulatory surgery centres [
10], has grown rapidly. Several empirical studies have also shown that specialized units outperform general hospitals both in cost efficiency [
8,
11] and in patient satisfaction [
12,
13].
Studies have demonstrated that the positive effect of focus may differ depending on the organizational level considered. In 2008, in a study of clinical trials, Huckman and Zinner [
14] discovered that there are benefits of focus not only at an aggregated organizational level (e.g., a hospital) but also at the department level. They concluded that their findings from the context of clinical trials align with the proposal by Skinner [
3] in manufacturing management, i.e., to structure industrial manufacturing facilities into focused divisions (‘plants-within-a-plant’) with adequate delineation organizationally as well as physically. This allows for concentration on a single task, without mixing production objectives in terms of quality and volume, and increasing clarity and focus on training, processes, and equipment [
14].
Following similar research propositions, but applied in the context of healthcare services, KC and Terwiesch [
6] undertook a study aiming to unravel the differences in focus outcomes between three organizational levels: hospital, department, and process. After controlling for selective patient admissions (or ‘cherry-picking’ admission practices), the authors found that hospital-level focus did not bring significant improvements in performance, which was measured as length of stay or mortality rate [
6].
However, recent research challenges the generalization of the positive relation between focus and outcomes. Kuntz et al. [
15] suggest that the benefits of focus and volume differ between patient segments and that the focus–outcome relationship needs to be determined at a more granular level. The authors distinguish between what they refer to as ‘routine and ‘complex’ patients, placing patients on a complexity spectrum that ranges from elective admissions of patients without comorbidities at one end to emergency admissions of frail patients with multiple comorbidities at the other end. Based on an analysis of in-hospital mortality in 60 German hospitals across 39 disease segments, they argue that for routine patients, focus is the main driver of service quality, while for complex patients, focus does not affect quality. Hence, they suggest that general hospitals should be reorganized as multi-specialty hubs for emergency and non-routine elective services, which are complemented by organizationally separate ‘focused factories’ for routine and elective services (similar hospital operational models are advocated by, e.g., Christensen et al. [
16] and Hopp and Lovejoy [
17]). The authors call for more research to verify their findings in the context of specific conditions, using condition-specific quality metrics, to be able to give more specific advice to hospital managers who wish to implement such reorganizations. For complex patients, this has been addressed by Miedaner and Sülz [
18] in a study of neonatal intensive care units, which confirmed that focus has no effect, or a negative effect, on length of stay in this context. The effect of focus has been extensively evaluated in the context of cardiac care, e.g., [
4,
5,
6,
19], but there are few studies in the context of other conditions, especially at the routine end of the complexity spectrum. Routine patients represent a large share of all patients. Comparing elective and emergency surgery, where emergency patients are more complex [
15], Prin et al. [
20] show that in developed countries, less than one emergency surgery is performed for every ten elective surgeries. In addition, timely and effective elective surgery might reduce the need for emergency services [
20]. Therefore, an efficient operational model for treating routine patients is central to curbing healthcare costs while delivering high-quality care.
The purpose of this study is to establish the relationship between focus and performance for [
17] patients undergoing obesity surgery. Obesity surgery, or bariatric surgery, is usually elective, and although technically advanced and associated with a risk of complications, it is a routine intervention where standardized protocols exist [
21]. Therefore, a study of bariatric surgery may be representative of other routine surgical procedures. Furthermore, organizing bariatric surgery in an efficient way is important from a managerial perspective, since it represents a common elective surgical procedure or, in US hospitals, even the most common elective surgical procedure [
22].
Using data from the Scandinavian Obesity Surgery Registry, complemented by a survey distributed to hospitals, this paper provides evidence for a strong association between hospital focus and performance in the context of bariatric surgery. We use a more comprehensive set of outcomes to measure process performance than previous studies of hospital focus. Our study adds to the existing literature on the link between focus and performance by supporting the notion that hospital focus is an important driver of quality and productivity for routine patients. Our findings also inform hospital managers in showing that bariatric surgery benefits from allocation to specialty hospitals or to organizationally separate hospital-within-hospitals focusing on this patient segment.
The rest of the paper is structured as follows.
Section 2 presents a brief theoretical background to the hypotheses investigated in our study.
Section 3 introduces the methods used, including data, variables, and analyses.
Section 4 describes the results. In
Section 5, we discuss the results and outline the contributions and limitations of our study.
Section 6 presents conclusions and implications for practice.
3. Methods
3.1. Data and Sample
Data on obesity surgeries in Sweden were retrieved from the Scandinavian Obesity Surgery Registry (SOReg) and a survey collecting complementary data from the hospitals that report to SOReg.
The SOReg database is a national quality registry constituting one of the most comprehensive databases of bariatric surgery in the world. The registry is continuously audited with less than 2% incorrect data entries [
31]. It contains national data from over 99% of the bariatric surgeries in Sweden since 2007 [
31], including those performed with (1) gastric bypass and (2) sleeve gastrectomy. The former includes rerouting nutrients to bypass the stomach, duodenum, and the proximal small intestine in order to reduce the sense of appetite. The latter involves the resection of a major part of the stomach, with the same of intent of reducing appetite and increasing the sensation of satiation. The registry compiles data on patient characteristics, surgical details, and outcomes including complications. Nineteen variables are mandatory for all patients and are recorded at the time of surgery and at several follow-up visits after surgery. Other variables are sometimes mandatory, depending on the type of surgical procedure. Additional optional variables, including quality-of-life-surveys completed by patients before and after surgery, are used in some hospitals. By the end of December 2016, the database included data on 58,417 patients. For this study, we collected data for all primary gastric bypass and sleeve gastrectomies performed between 1st January and 31st December 2016. Excluding repeat surgeries (
n = 197), we identified a total of 5152 adult patients who underwent surgery in 2016 and patients who had other surgical procedures performed at the same time as their bariatric surgery (
n = 265).
The SOReg registry contains comprehensive data on all obesity surgery patients, but in order to measure focus, we also needed information about other patients at the hospitals who were not covered by the registry. Hence, in order to measure the degree of focus in obesity surgery, we collected complementary data through a short survey sent to all Swedish hospitals reporting to SOReg. The survey consisted of a questionnaire asking for the total number of admissions at the hospital as well as the total number of surgical procedures at the department where obesity surgery was performed between 1st January and 31stDecember 2016. The questionnaire was sent by email to the SOReg coordinator at each hospital in March 2019 and was followed by reminders via telephone when needed. Of the 44 hospitals that performed obesity surgery in 2016, 39 responded to the survey, corresponding to a response rate of 89%. The non-respondents included two clinics that had been closed down. Two of the remaining hospitals were reached by phone and attempted to respond, but they could not deliver the historical aggregated data that was necessary for this study due to organizational and software changes. Only one of the hospitals did not respond, despite repeated reminders. There was no reason to assume a systematic bias in the missing responses.
The study was approved by the Regional Ethical Review Board in Stockholm, Sweden (registration number 2018/695-31/3).
3.2. Variables
We used a comprehensive set of outcomes as performance indicators: complications at three points in time, procedure time, and length of stay. Separate regressions were run for each of the five outcome variables.
Complications are considered the most reliable measure of quality in bariatric surgery, and they are preferable to other measures such as re-operations, which are too rare to be a reliable measure for low-volume hospitals, or mortality, which is too rare to be a reliable measure regardless of hospital volume [
32]. The data were filtered to include all complications that required surgical intervention under general anesthesia or intensive care (complication grade > 3a according to the Clavien–Dindo classification [
33,
34] as binary variables at three points of time: (1) during surgery (intraoperative complications); (2) within 30 days after surgery as reported at the follow-up visit approximately six weeks after surgery (30-day complications); and (3) occurring between 30 days and one year after surgery, as reported at the follow-up visit approximately one year after surgery (one-year complications).
Procedure time and
length of stay are indicators of both quality and cost in bariatric surgery. Procedure time, measured as operating time in minutes (time from first skin incision to completion of wound closure), is an established quality indicator, since shorter procedure times have been shown to be correlated with lower mortality [
35], fewer complications [
35,
36], and shorter length of stay [
36]. Shorter procedure times can also be a driver of lower cost per procedure, since they allow more procedures to be performed using the same operating facilities and staff. Similarly,
length of stay, measured as days from admission to discharge, is a widely used indicator of quality in bariatric surgery (see, e.g., the outcome measures included in the systematic reviews of Markar et al. [
37] and Zevin et al. [
38]). Longer postoperative hospital stay is also a cost driver, since this consumes more medical and physical resources and consequently increases the cost per patient.
Hospitalfocus and
department focus constituted the two explanatory variables. In line with previous studies [
4,
5,
6,
18], we conceptualize focus as an emphasis on a particular service segment (i.e., obesity surgery) and operationalize it as the number of obesity surgery patients in relation to the total number of patients. The annual number of obesity surgery admissions divided by the annual number of total admissions to the hospital was used as our measure of
hospital focus. The annual number of obesity surgery procedures divided by the annual number of all surgical procedures was used as our measure of
department focus.
Our sample included many low-focus hospitals and departments and relatively few high-focus units. To address the resulting positive skew of the distributions, hospital focus and department focus were normalized using the natural log of the variables [
39].
The following control variables were included as covariates to control for case mix: age; sex; pre-surgery BMI; and the comorbidities sleep apnea, hypertension, diabetes, and dyslipidemia. Type of procedure (gastric bypass or sleeve gastrectomy) was included as an additional covariate. We also controlled for procedural volume, measured as the number of primary bariatric surgical procedures performed in 2016 at the hospital where the surgery was performed.
3.3. Analyses
Multilevel regression analyses using complete case analysis were conducted using Stata version 14.1, with individuals nested within clinic-level random effects. Adopting a multilevel model structure has become increasingly common when comparing the performance of multiple healthcare providers [
40,
41] and are appropriate when observations are not independent of each other due to shared group characteristics. Multilevel regression models have been used in previous studies of focus in hospitals e.g., [
4,
18]. The multilevel modelling approach, as applied to patients in hospitals, is based on the assumption that there are hospital-specific random effects that are drawn from a common distribution. In our case, the predictor variables were measured at the hospital level, while the output variables were measured at the individual level, and this structure necessitated a multilevel modelling approach with random intercepts that allow different performance baselines for each hospital [
42].
Three regression models were developed and tested: the first using multilevel logistic regression, the second using multilevel linear regression, and the last using multilevel Poisson regression (following the guidelines by Rabe-Hesketh and Skrondal [
42,
43]). Logistic regression was used to assess complications, as this variable is binary. Linear regression was used to assess procedure time, as this is a continuous variable. Poisson regression was used to assess length of stay, as this is measured in whole days, i.e., discrete data with non-negative integer values. All regressions used the maximum likelihood (ML) estimation method (see, e.g., [
4]).
Furthermore, we employed a complete case analysis, using only observations with complete data in each regression [
39]. For most variables (with a few exceptions discussed below), less than 0.2% of the values were missing. The focus variables were missing for 8.9% of patients due to the survey response rate, as described in the section “Data and sample” above. Data on 30-day complications were missing for 3.9% of patients, reflecting a small number of individuals that either did not show up for the six-week follow-up visit or had a follow-up visit at another clinic that did not report to the registry. For one-year complications, 14% of the values were missing, presumably reflecting patients that did not show up for the one-year follow-up visit. Patients with serious complications could be assumed to have returned to the hospital; consequently, patients with a missing value for one-year complications are likely to belong to the great majority of patients without major complications.
5. Discussion
The present study reveals a positive effect of hospital focus and department focus on performance. Significant effects were found on the number of complications within 30 days and one year after surgery, as well as on procedure time, after controlling for the effects of different patient risk factors. The effect size was relatively small, with a 10% increase in focus corresponding to 2–4% fewer complications and a reduction in procedure time of 3–6 min. For obesity surgery, where the overall risk of complications is low, this may not be of great clinical significance. However, the relationship established in this study could have a far greater clinical significance for other types of elective surgeries with higher overall complication rates: e.g., certain cancer surgeries.
We found no statistically significant relationship between focus and length of stay. For length of stay, there are large inter-hospital differences that may be explained by different operational characteristics—or even traditions—not related to focus, such as local discharge criteria and geographical distance between the hospital and patients’ homes. We note that also previous studies show mixed results on the effects of focus on length of stay: Capkun, Messner, and Rissbacher [
24] reported that focus decreases the length of stay for hospitals overall, while KC and Terwiesch [
6] did not find a statistically significant association between focus at any organizational level and length of stay in the context of cardiac care.
Thus, the present study shows that hospital units with higher focus have better outcomes. There are two possible explanations of this relationship: focus might drive performance or performance might drive focus. In the former case, focus enables learning, specialization, and higher quality in healthcare procedures, which results in better outcomes. In the latter case, good outcome records result in stronger demand for obesity surgery in relation to other healthcare services, which, in the long run, makes the hospital more focused. In Sweden, the individual choice of healthcare provider is limited, and the majority of obesity surgery patients whose procedures are tax-funded are typically remitted to the closest provider. Given this empirical context, it is plausible to assume that it is focus that drives performance in this setting. However, the dynamics of causation might differ between different contexts and healthcare systems, which calls for these contextual factors to be taken into account in further research.
Our results are not consistent with the recent study by Miedaner and Sülz [
18], which found negative effects from focus in neonatal intensive care. A possible explanation of the incongruent results would be that patient characteristics moderate the impact of focus on performance. In that case, the different characteristics of obesity surgery patients and patients in neonatal care could explain the mixed findings. According to a recently presented contingency theory of hospital focus [
15], the differences in results may be explained by the fact that patients in the two studies are widely distributed on the scale between highly complex, unpredictable, emergency care, and highly predictable, routine, elective care. It would be unreasonable to assume that obesity surgery is on the extreme routine end of this scale, given that many obese patients have comorbidities that increase unpredictability and that many simpler elective procedures exist. However, we argue that the current standardized nature of the obesity surgery procedures means that patients undergoing obesity surgery can be characterized as relatively routine elective patients, and that this is a likely explanation of our results in relation to the results of Miedaner and Sülz [
18].
We also found that the benefits of focus are stronger at the more granular organizational level. This hypothesis was supported by all five outcome measures. Our results are consistent with the findings of KC and Terwiesch [
6] in the context of cardiac care, in which length of stay and mortality were reduced more by department-level (operating unit) and process-level focus than by hospital-level (firm) focus.
The present study adds to the emerging contingency theory of hospital focus by supporting the notion that hospital focus is an important driver of quality and productivity for patients located toward the routine end of the complexity spectrum [
15]. A key strength of this study is that it uses a more comprehensive set of relevant outcomes to measure process performance than previous studies looking at the link between focus and performance.
This study has some limitations. It uses data from the Scandinavian Obesity Surgery Registry, and this may limit the generalizability to other countries—in particular to developing countries—where conditions for quality and cost in healthcare are substantially different from those in Scandinavia. However, the study benefits from using one of the most comprehensive databases on obesity surgery available, and it is plausible that our results are valid for countries in North America and Europe with similar conditions as those found in Sweden. Our study is also limited to studying only one type of surgery; therefore, its generalizability for other surgical procedures could be questioned. Further studies could also include the effects of focus on the hospital as a whole or the spillover effects [
5] of the hospital’s focus on related patient segments, as well as further control variables, accounting for, e.g., seasonal and day-of-week effects. Moreover, since this study was limited to quality outcomes and productivity measures (procedure time and length of stay), studies that measure cost performance more explicitly are also called for.
6. Conclusions and Implications
This study addresses the relationship between focus and performance for patients undergoing elective surgery. Based on a study of obesity surgery, we conclude that focus has a positive effect on elective surgery performance at both the hospital level and the department level. Significant effects were found on the number of complications within 30 days and one year after surgery, as well as on procedure time after controlling for the effects of different patient risk factors.
This study has several implications for how obesity surgery should be organized to improve efficiency and patient safety. Firstly, the establishment of specialty hospitals and other types of specialized healthcare providers should be encouraged. Focus leads to improved performance at the hospital level, which could be explained by a lower level of distraction from other processes and patient segments. Secondly, general hospitals can emphasize particular areas in order to increase hospital focus. It may be appropriate to emphasize a particular service area in an operational way and thus achieve a higher proportion of these patients, even when a broad range of services is required from the hospital [
4]. Thirdly—and probably most feasible to implement—general hospitals should create hospitals-within-hospitals with a high level of department focus. Our study shows that for patients undergoing obesity surgery, a focused department is more important than a focused hospital, and focus is more important than procedural volume. These hospital-within-hospital departments can be complemented by flexible multi-specialty services that meet the needs of complex patients [
15,
16,
17]. An organizational redesign to create a hospital-within-a-hospital can be guided by case studies of successful implementations of focused hospital units e.g., [
25,
45].
Our findings clearly support the use of these three guidelines for organizing obesity surgery. They also indicate that similar organizational designs are appropriate for other elective surgeries that involve standardized procedures. However, more research is needed to provide advice for other specific medical conditions.