Risk Factors for Complications After Esophageal Cancer Resection

Abstract

Objective:

To identify risk factors for complications after resection for esophageal or cardia cancer.

Summary Background Data:

Knowledge of risk factors for complications after esophageal resection for cancer is sparse, and prospective population-based studies are lacking.

Methods:

A prospective, nationwide, population-based study was conducted in Sweden in April 2, 2001 through December 31, 2003. Details about tumor characteristics and stage, surgical procedures, and complications were collected prospectively from the Swedish Esophageal and Cardia Cancer register. Medical records and specific charts from surgical procedures, histopathology reports, and intensive care units were continuously scrutinized. Multivariable logistic regression analyses were used to estimate relative risks and their 95% confidence intervals.

Results:

Among 275 patients undergoing surgical resection for esophageal or cardia cancer, 122 (44%) had at least one predefined complication. Operation by low-volume surgeons (<5 operations annually) were followed by more anastomotic leakages than those by surgeons with higher volume (odds ratio, 7.86; 95% confidence interval, 2.13–29.00). Hand-sewn and stapled anastomoses did not differ regarding risk of anastomotic leakage. Among cardia cancer patients, transthoracic approach resulted in more respiratory complications compared with transhiatal (abdominal only) approach (odds ratio, 4.78; 95% confidence interval, 1.66–13.76). Older age, adjuvant oncologic therapy, and higher preoperative bleeding volume nonsignificantly increased the risks of complications, while no influence of sex or tumor stage was found.

Conclusions:

High-volume esophageal surgeons seem to lower the risk of anastomotic leakage. More large-scale studies are warranted to establish the roles of the other potentially important risk factors suggested in our study.

A prospective, population-based, nationwide study was conducted in Sweden in 2001 through 2003 to analyze risk factors for complications after esophageal resection for cancer. The main finding was that surgery by low-volume surgeons increased the risk of anastomotic leakage.

Radical esophageal resection alone remains the only established curative treatment of cancer of the esophagus and gastric cardia.^1–3 A major problem is that the required surgery with a curative intent is extensive and entails a considerable risk of serious complications, and although the short-term mortality seems to have decreased in recent years, the complication rate remains high.⁴ Recent results from our group indicate that the occurrence of acute postoperative complications is the main surgery-related factor with a negative influence on the quality of life after radical esophagectomy for cancer.^4a Quality of life is an outcome of particular importance after esophageal cancer surgery, since the majority of the surgically treated patients are not cured.^1,2 Furthermore, the occurrence of surgery-related complications per se might contribute to a worse long-term prognosis after surgical resection for esophageal cancer.^5,6 All these observations emphasize the need to identify factors that influence not only the survival, but also the risk of complications. Moreover, in a recent evaluation of studies addressing short-term survival as outcome, it was concluded that, owing to the low rates of short-term mortality, individual studies regarding esophageal cancer surgery generally lack statistical power to provide valid results using this outcome.⁷ Investigations that have addressed complications as the outcome measure have been limited.^8,9 We therefore carried out a prospective, population-based study in Sweden with the purpose of identifying factors that affect the occurrence of predefined complications after esophageal surgery for cancer. Prior to any analyses, we formulated several hypotheses, ie, that age, sex, tumor stage, neoadjuvant oncologic treatment, preoperative bleeding volume, surgical approach, surgeon volume, and hospital volume influence the occurrence of severe complications. Moreover, we hypothesized that the occurrence of anastomotic leakage is influenced by the anastomosis technique, surgeon volume, and hospital volume. The main aim was to evaluate how surgeon volume influences the risk of complications, however.

METHODS

Study Design

A prospective, population-based, nationwide observational study was conducted in Sweden. All patients with a newly diagnosed adenocarcinoma or squamous cell carcinoma of the esophagus or gastric cardia who underwent tumor resection in Sweden during the period April 2, 2001 through December 31, 2003 were eligible for the study.

Data Source

The data were collected from the Swedish Esophageal and Cardia Cancer register (SECC register), an almost complete nationwide register of esophageal and cardia cancer surgery in Sweden. The organization of this register is a continuation of a collaborative nationwide Swedish network of hospital departments and clinicians involved in the diagnosis or treatment of patients with cancer of the esophagus or gastric cardia. This network was built up during the early 1990s, initially for the purpose of a population-based case-control study, which addressed risk factors for esophageal and cardia cancer, conducted in 1995 through 1997, and described in detail elsewhere.^10–12 In the current study, we identified patients shortly after the diagnosis had been histopathologically confirmed through collaboration with all pathology departments in Sweden. Thereafter, we collected all information required after communication with contact doctors responsible for the local registration at each participating department (listed in the Acknowledgments). During the study period, a total of 174 of 179 (97%) relevant hospital departments of general surgery, thoracic surgery, oto-rhino-laryngology, oncology, and pathology in Sweden participated in the registration.

Data Collection, Definitions of Outcome, and Surgical Procedures

Details about tumor characteristics, surgical procedures, and complications were collected through review of medical records, including protocols from surgical procedures, histopathology reviews, endoscopic and radiologic examinations, and intensive care units. The complications that were deemed to be severe were defined by a group of leading Swedish esophageal surgeons prior to the inclusion phase of the study. These complications included any of the following occurrences within 30 days after surgery: mortality (independently of the cause), anastomotic leakage (causing clinical symptoms and verified by radiology or endoscopy), serious infections (intra-abdominal or intrathoracic abscess, sepsis with positive bacterial culture in the blood, or wound infection requiring intervention), respiratory insufficiency (need for reintubation, or severe pneumonia), cardiac failure (myocardial infarction, or arrhythmia with requiring for intervention), renal or liver failure (need for dialysis and jaundice, respectively), technical complications (postoperative bleeding >2000 mL or a need for reoperation, inadvertent damage to the recurrent laryngeal nerve or the thoracic duct), early anastomotic stricture (with severe dysphagia and a need for endoscopic intervention), and others (embolus, deep venous thrombosis, rupture of the wound, intestinal obstruction, or stroke, all with a need for intervention). The specific charts of surgical procedures provided detailed information on the surgical approach, macroscopic surgical radicality, type of substitute used, type of anastomosis performed, technical complications, preoperative bleeding volume, and details of any reoperations carried out. We defined the surgical approaches as follows: 1) Esophageal resection refers to removal of the main part of the esophagus with an anastomosis between an esophageal substitute (stomach, jejunum, or colon) and the proximal esophagus. 2) Cardia resection represents removal of the proximal part of the stomach and the distal part of the esophagus with an anastomosis between the remaining stomach and the remaining esophagus. 3) Extended total gastrectomy refers to removal of the entire stomach and the distal part of the esophagus with anastomosis between the jejunum and the esophagus. 4) Total gastrectomy and esophageal resection means that the entire stomach and the main part of the esophagus were removed with an anastomosis between an esophageal substitute (jejunum or colon) and the proximal esophagus. Our continuous review of all histopathologic reports provided us with further information regarding tumor location, histologic type, tumor stage, number of removed lymph nodes, and microscopic radicality of the tumor resection. The location of adenocarcinomas within the gastroesophageal junction was classified according to the Siewert definitions.¹³ Adenocarcinomas of Siewert types 2 and 3 were considered to be located in the cardia, while those of Siewert type 1 were classified as esophageal. To define the tumor stage, we used the TNM classification according to Union Internationale Contre le Cancer (UICC).¹⁴

Statistical Analyses

We used unconditional logistic regression model to estimate the relative risk of complications in the form of odds ratios (OR) with 95% confidence intervals (CI). In multivariable modeling, our basic model included adjustments for age (categorized into 3 groups: <60, 60–69, or ≥70 years), sex, and tumor stage (4 groups: 0–I, II, III, or IV). We also analyzed the variables in a more extensive multivariable model in which we also adjusted for all other covariates under study, including histologic type of cancer (categorized into 2 groups: adenocarcinoma or squamous cell carcinoma), neoadjuvant treatment (2 groups: yes or no), preoperative bleeding volume (3 groups: <500, 500–1000, or ≥1000 mL), surgical approach (2 groups: transhiatal [abdominal only] or transthoracic), surgeon volume (3 groups: <5, 5–10, or >10 operations per year), type of hospital (2 groups: university or nonuniversity), and type of anastomosis (2 groups: stapled or hand-sewn). We used 2 separate main outcome variables: 1) occurrence of any predefined complications and 2) occurrence of anastomotic leakage. Regarding complications, this outcome variable was analyzed in 3 levels: 1) occurrence of predefined complication, 2) at least 2 complications, and 3) at least 3 complications. All statistical analyses were performed using SAS 8.2, and the PROC GENMOD was used for the logistic regression analyses. No adjustment for multiple comparisons was made.

Ethical Considerations

Informed consent was obtained from each patient before inclusion in the study. The ethics committee at Karolinska University Hospital, Karolinska Institutet, Stockholm, Sweden, approved the study.

RESULTS

Patients

During the study period, 276 patients who underwent surgical resection for esophageal or cardia cancer were included. Of these patients, 1 was excluded because of missing data regarding the occurrence of complications. Thus, 275 patients remained for final analysis. Some characteristics of the included patients are presented in Table 1. The overall median age at the time of surgery was 67 years. A higher proportion of the cardia cancer patients were older than 70 years (46%) compared with those with esophageal cancer (36%). The majority of the patients were male (79%), and the sex ratio was similar between esophageal and cardia cancer patients. Among the patients with esophageal cancer, adenocarcinoma was more common (58%) than squamous cell carcinoma. About half (51%) of the patients had a tumor in stage III or IV.

TABLE 1. Characteristics of 275 Patients Undergoing Resection for Esophageal or Cardia Cancer

Type of Surgery

The type of surgery among the study patients is presented in Table 2. Patients with esophageal cancer were, with a few exceptions, operated with a transthoracic esophageal resection with a gastric tube as esophageal substitute, similar among high-volume surgeons and low-volume surgeons (Table 2). The type of surgery among patients with cardia cancer varied between esophageal resection, cardia resection, extended total gastrectomy, or total gastrectomy and esophageal resection (see definitions above). For the treatment of cardia cancer, high-volume surgeons preferred esophageal resection or a combination of total gastrectomy and esophageal resection (78%), while surgery including esophageal resection was considerably less common among low-volume surgeons (30%) (Table 2). The choice of substitute was similar between high-volume and low-volume surgeons, however. The colon was used as esophageal substitute in merely 7 patients (2.5%), and only one of these operations was transhiatal with a neck anastomosis.

TABLE 2. Type of Surgery Among 275 Patients Undergoing Resection for Esophageal or Cardia Cancer Among High-Volume and Low-Volume Surgeons

Frequency of Complications

The distribution of complications is presented in Table 3. Almost every second operation, ie, 122 of 275 patients (44%), led to at least one major predefined complication within 30 days after surgery, and 11% of the operations were followed by more than 2 complications. The overall 30-day mortality rate was 3.6% (10 patients). Three patients died of myocardial infarction, while the remaining deaths were due to anastomotic leakage, postoperative bleeding, sepsis, heart insufficiency, or lung embolus. The in-hospital mortality was 7.3% (20 patients) and the 90-day mortality was 8.4% (23 patients). The most frequent complications were respiratory insufficiency, severe pneumonia, heart complications, and serious infections (Table 3). Symptomatic anastomotic leakage occurred in 9% of the patients. The overall frequency and distribution of complications were similar in the patients with cancer of the esophagus and cancer of the cardia, except for technical complications, which were more common among in the esophageal cancer group (13%) than in the group with cardia cancer (4%). Three of the 7 patients in whom colon was used as an esophageal substitute had a complication (data not shown).

TABLE 3. Distribution of Complications After Resection for Esophageal and Cardia Cancer Among 275 Patients

Risk Factors for Complications

The results of the analyses conducted to test the influence of the defined hypothetical factors on the risk of other complications after esophageal or cardia cancer resection are presented in Table 4. In analyses using at least 2 complications as outcome, we found indications that higher age might be associated with an increased risk, and when at least 3 complications were used as outcome, age above 70 years carried a statistically significantly increased risk (OR, 4.30; 95% CI, 1.37–13.47). Moreover, the trend with increasing age was significant (P for trend = 0.004). Sex did not materially influence the risk of complications (Table 4). There was no evidence of an increased risk of complications with increasing tumor stage (Table 4). Similarly, no differences in the frequency of complications were identified between cancer located in the esophagus or the gastric cardia (Table 4). All risk estimates of complications were nonsignificantly increased among patients treated with radio- or chemotherapy prior to surgery. Similarly, a higher bleeding volume during surgery resulted in a nonsignificant tendency toward an increased risk of complications (Table 4). The frequency of cardiac complications was independent of preoperative bleeding volume (data not shown). In an analysis restricted to patients with cardia cancer, the risk of complications was found to be almost 5 times higher in patients operated on with a transthoracic approach than in those in whom a transhiatal (abdominal only) approach was used (OR, 4.78; 95% CI, 1.66–13.76). This higher risk was entirely attributable to an elevated risk of respiratory complications after surgery with the transthoracic approach (26%, 21 of 82 patients) compared with the transhiatal (abdominal only) approach (4%, 2 of 46 patients). Patients operated on by low-volume surgeons, ie, surgeons who performed fewer than 5 esophageal resections annually, had a tendency to sustain more complications compared with patients operated on by surgeons who performed at least 5 operations per year, but the difference in risk did not reach statistical significance (Table 4). In analyses including categories of surgeons performing more than 10 operations per year, these showed no appreciable difference regarding complications from surgeons performing 5 or more operations (data not shown). The risk of complications was not materially influenced by the type of hospital at which the operation was carried out, ie, university or nonuniversity hospital (Table 4). Generally, the relative risk estimates in the basic multivariable model (adjusted for age, sex, and tumor stage only) did not differ importantly from those in the full multivariable models (Table 4), including adjustment for all the variables listed in Methods, indicating lack of strong confounding by these additional variables. Assessing short-term mortality separately, 7 of the 99 patients (7%) who had been operated on were operated on by a low-volume surgeon died within 30 days after surgery, and the corresponding mortality rate among high-volume surgeons was 3 of 176 (2%). This rendered an OR of 4.02 (95% CI, 0.99–16.33) among low-volume surgeons compared with high-volume surgeons after adjustment for age, sex, and tumor stage.

TABLE 4. Risk of Any Severe Complication After Resection for Esophageal or Cardia Cancer Among 275 Patients, Expressed in Odds Ratios (OR) With 95% Confidence Intervals (CI)

Risk Factors for Anastomotic Leakage

The comparison of hand-sewn anastomoses with stapled anastomoses did not reveal any clear difference in the risk of anastomotic leakage after multivariable adjustments, including and excluding adjustment for surgeon volume (Table 5). Esophageal resections conducted by low-volume surgeons (<5 operations per year) increased the risk of anastomotic leakage nearly 8-fold compared with those performed by surgeons with a higher volume (OR, 7.86; 95% CI, 2.13–29.00). The type of hospital, ie, university hospital or not, did not influence the occurrence of anastomotic leakages, however (Table 5). The results based on the 2 main models used for adjustment for potentially confounding factors did not show any major differences.

TABLE 5. Risk of Anastomotic Leakage After Resection for Esophageal or Cardia Cancer Among 275 Patients, Expressed in Odds Ratio (OR) With 95% Confidence Intervals (CI)

DISCUSSION

This nationwide Swedish study indicates that almost every second esophageal resection for cancer entailed at least one severe complication. The main finding of the study was that surgery conducted by low-volume surgeons greatly increased the risk of anastomotic leakage, and possibly also for short-term mortality. Furthermore, cardia cancer patients were at increased risk for respiratory complications after surgery with a transthoracic compared with a transhiatal (abdominal only) approach. Higher age, low-volume surgery, preoperative chemo- or radiotherapy, and a high preoperative bleeding volume were potential or moderate risk factors for complications, while sex, tumor stage, and tumor location showed no influence in this respect. No difference in the risk of anastomotic leakage was found between hand-sewn and stapled anastomoses.

Some methodologic aspects deserve attention. A particular advantage of this study is the population-based design, which reduces selection bias and makes generalizability possible. Moreover, the data were collected prospectively, before the occurrence of complications, which would further reduce risk of selection bias. The registration of esophagectomies in Sweden was not entirely complete during the study period, however. A few surgical departments did not participate, and all patients did not give their consent to be registered; finally, during the initial workup phase of the registration, the registration frequency was more limited. The low short-term mortality rate in our study indicates that some degree of selection might have occurred, but such bias should not importantly have affected the main findings concerning relative risks. Furthermore, the frequency of anastomotic leakages was similar to that in most other Western countries,⁴ suggesting that selection bias was not a major problem in our study. The information regarding the tumors, operations, and complications was detailed and valid, thus diminishing misclassification of the exposure or outcome. The ability to adjust for a variety of potential confounding factors is another strength of our study. The lack of information regarding comorbidities and some other potentially important confounders, including tobacco smoking and alcohol consumption, is a weakness, however. We only included complications that occurred within 30 days after surgery, which meant that late complications, mainly anastomotic strictures, were not assessed. Furthermore, chance might have influenced our results through problems with sample size or multiple testing. The limited sample size prohibited statistical ascertainment of weak associations. The relatively small number of included patients despite the nationwide extent of the study is explained by 1) the low frequency of esophageal cancer resection in Sweden (approaching 25% of all esophageal cancer according to unpublished data from the SECC register), and 2) the limited inclusion rate during the initial phase of the registration. Our strict analyses, confined solely to the predefined hypothetical factors, should counteract spurious findings due to multiple comparisons. Finally, since the outcome under study was frequent, the calculation of our odds ratios would tend to overestimate the relative risks.

The frequency of complications in our study was higher than that in most previous reports (26%–41%),² which might be due to selection bias in studies that are not population-based. Of the factors reported to that reduce the risk of short-term mortality after esophageal cancer resection, the best documented are a high hospital volume and high surgeon volume of the procedure.^8,15–17 Among the limited number of studies that have addressed complications as the outcome after esophageal cancer surgery, a recent retrospective study showed that patients operated on in low-volume hospitals had an increased risk of surgery-related complications,⁹ a finding that is supported by our data. In our study, the surgeon volume seemed more important than the hospital volume with regard to anastomotic leakage. Moreover, although the number of patients who died shortly after surgery in our study was limited and prohibited valid multivariable adjusted analyses, our study suggests an increased risk of 30-day mortality among low-volume surgeons. This finding is well in line with the results of a large recent retrospective U.S. study, where the mortality rate after esophageal resections was higher among patients of low-volume surgeons than among those of high-volume surgeons, regardless of the surgical volume of the hospital in which they practiced.¹⁸ We found no decreased risk of overall complications at university hospitals compared with nonuniversity hospitals, however. This might be due to the fact that operations requiring extensive dissection and advanced reconstructions are usually carried out at university hospitals in Sweden, and such procedures more frequently entail complications compared with less advanced cases that can be operated on at smaller hospitals. This is supported by the finding that all patients with a colon substitute in our study were operated on at university hospitals. Moreover, patients with more comorbidity might more readily be selected for surgery in university hospitals rather than smaller hospitals. Such selection bias might explain the lack of improved short-term results with regard to occurrence of complications at university hospitals in our study.

Findings in previous studies regarding postoperative complications and short-term mortality after neoadjuvant radio- or chemotherapy are contradictory. This uncertainty is probably mainly due to the limited statistical power of the individual studies. Our study revealed indications of an increased risk of complications among patients who received neoadjuvant therapy. These results are compatible with those of a recent meta-analysis, which showed a statistically significant 2-fold increase in the risk of postoperative death after neoadjuvant therapy.¹⁹

Our study, like some previous studies,^1–3,20 suggests that transthoracic esophageal surgery is followed by an increased risk of respiratory complications compared with a transhiatal (abdominal only) approach. However, although no differences are yet firmly established,¹ there might be a survival advantage in the transthoracic group.²⁰ Moreover, the only way to perform an optimal radical tumor dissection with the intention to cure in the case of esophageal cancer includes a transthoracic approach.²¹ Hence the increased risk of respiratory complications should not prevent the use of a transthoracic approach whenever this procedure is possible and the surgery has a curative intent. For palliative resections, however, this increased risk of complications should be taken into consideration.

We hypothesized that cardiac complications might be more common after surgery with high preoperative bleeding volume, but we found no such evidence in our study. The finding of an increased risk of complications in older age groups is probably mainly due to the higher prevalence of comorbidity among the elderly. The lack of difference between the sexes with regard to complications is not surprising because there are no biologic reasons to expect any divergence. Some previous results indicate that a higher tumor stage might increase the risk of surgical complications,²² but we did not find any support for this possibility in the present study. Although the long-term prognosis is highly dependent on the tumor stage, the stage will probably play a minor role with regard to short-term results.

CONCLUSION

Our study indicates that the risk of anastomotic leakage after resection for esophageal or cardia cancer was considerably increased in patients operated on by low-volume surgeons. Moreover, among cardia cancer patients, a transthoracic approach carries a higher risk of respiratory complications than a transhiatal (abdominal only) method. Older age, low-volume surgery, preoperative oncologic therapy, and a higher preoperative bleeding volume might increase the risk of complications, while no influence of sex or tumor stage was found. No difference in anastomotic leakage was noted between hand-sewn and stapled anastomoses. A seemingly valid clinical implication of our findings is that the risk of complications of esophageal cancer surgery can be reduced if the operations are performed by high-volume esophageal surgeons.

ACKNOWLEDGMENTS

The authors thank Eja Fridsta for invaluable help with the fieldwork of the Swedish Esophageal and Cardia Cancer register (SECC) as well as all contact doctors who participated in the SECC register: Erik Aasberg, Gunnar Adell, Stig-Olof Almgren, Christer Andersson, Bengt Anesten, Ebbe Ankeraa, Lena Appelgren, John Blomberg, Bernt Boeryd, Poul Boiesen, Mogens Bove, Mats Broberg, Pawel Burian, Björn Bäckstrand, Tora Campbell-Chiru, Ulf Carlsson, Åke Dahlqvist, Jan Dalenbäck, Stefan Dedorson, Jan Dolata, Thomas Edekling, Kent Edin, Tom Ekstrand, Annika Elmhorn Rosenborg, Lars-Krister Enander, Mats Eriksson, Thomas Eriksson, Bo Erlandsson, Ivan Farago, Elisabeth Flashar Lindahl, Petra Flygare, Ulla Friberg, Signe Friesland, Ashok Gadré, Ulf Gerdes, Per Gullstrand, Olle Gunder, Gustav Gustavsson, Mehmet Gözen, Andrea Halasz, Marianne Hall Angerås, Ann Hammarlund, Jörgen Hansen, Poul Erik Hansen, Mats Hedberg, Lennart Hedenborg, Antti Heikkilä, Per Herlin, Leif Hoffmann-Laursen, Erik Holm, Fredrik Holmner, Hans-Olof Håkansson, Hans Högström, Leif Ingvarsson, Peter Jilcke, Bengt Johansson, Karl-Erik Johansson, Leif Johansson, Sven Johansson, Erik Johnsson, Göran Karlsson, Mats Karlsson, Staffan Kinman, Gerhard Kjellén, Wieslava Kranze, Hans Krook, Richard Kuylenstierna, Anders Kylebäck, Peter Lannes, Sven Larsson, Gustav Levin, Pontus Lidberg, Bengt Liedman, Håkan Liljeholm, Ramon Lillo-Gil, Daniel Lindahl, Gert Lindell, Christer Lindholm, Johan Lindholm, Christina Lindström, Gunilla Ljung, Anders Lundberg, Kjell Lundberg, Lars Lundell, Britta Lödén, Ulf Lönn, Ester Lörinc, Shabaz Majid, Janusz Matusik, Kai Melén, Lennart Mellblom, Hans Modig, Eva Munck-Wikland, Torbjörn Myrnäs, Ewa Mårtensson, Torben Navne, Anders Nihlberg, Erik Nilsson, Mats Nilsson, Roland Nilsson, Torgny Nordén, Tomas Norlander, Johan Nyman, Ingmar Näslund, Bengt Ohlander, Bo Ojen, Sante Olling, Bengt Olsson, Håkan Olsson, Mikael Olsson, Rickard Palmkvist, Fredrik Petersson, Johan Ramström, Gunnar Rimbäck, Pehr Rissler, Staffan Rosendahl, Claes Rudberg, Valter Ryd, Thure Rådmark, Björn Sandmark, Charles Silseth, Svante Sjöstedt, Stefan Spinnell, Roger Stenling, Dag Stockeld, Edith Stöckel, Magnus Sundblad, Erik Svartholm, Conny Svensson, Jan-Olov Svensson, Robert Szepesvari, Claes Söderlund, Claes Theander, Stefan Thelin, Maria Thorén Örnberg, Tibor Tot, Torbjörn Tuveson, Maciej Tytor, Sören Vallgren, Svend Vang Petersen, Anders Victorin, Torkel Wahlin, Kjell Wallin, Bruno Walther, Olle Wihlborg, Bo J Wilhelmsson, Åke Östh, and Görel Östberg.