The Indications for Laparoscopic Pancreatectomy
Results on 550 patients in the German Laparoscopic Pancreas Registry in comparison to other registries
; ; ; ; ;
Background: Laparoscopic pancreatectomy is not yet established as a routine procedure everywhere in Germany or in other countries. Few data are available on its short- and long-term outcomes.
Methods: From 2008 onward, a working group initiated by 10 centers and currently comprising 34 centers has gathered data on all cases of laparoscopic pancreatectomy. Procedures in which laparoscopy was converted to open surgery are also included.
Results: The registry now contains 550 data sets representing 267 cases of benign disease, 244 malignancies, and 39 borderline tumors. The most common procedure was laparoscopic left pancreatectomy, followed by resection of the head of the pancreas and tumor enucleation. The most common intraoperative complication was hemorrhage, with a frequency of 3%. The rate of conversion to open surgery was 35%; if minilaparotomies are excluded, the conversion rate was only 16%. 39% of patients developed a pancreatic fistula after surgery (usually grade A or B, with 1.5% grade C) and 3% underwent reoperation because of postoperative hemorrhage. The procedure-related mortality was 1.3%. 91% of the patients reported only very mild postoperative pain. 6.7% newly developed diabetes mellitus after the procedure.
Conclusion: The patient cohort in the registry consists of persons who were selected to undergo laparoscopic pancreatectomy by the participating hospital teams, and the data are thus inherently affected by selection bias. The operative procedures that they underwent reflect the current practice of laparoscopic pancreatectomy in Germany. The complication rates are similar to those of open surgery. Selection bias can be avoided only by a randomized trial.
Laparoscopic procedures on many different organs have replaced open surgery over the past 20 years. This development has markedly reduced operative trauma, enhanced the preservation of bodily integrity, and shortened post-surgical recovery times (1). Resection of part or all of the pancreas is among the more complex procedures in visceral surgery, with a relatively high burden of complications; the use of laparoscopy in this area is motivated by the desire to lessen procedure-related trauma.
After the pioneering initial work of Gagner and Cuschieri in 1992, in which the procedures often involved assistance by hand, the use of laparoscopy in pancreatic surgery declined for a few years (2, 3). After the introduction of dedicated instruments for laparoscopic surgery, including ultrasonic scissors, sealing instruments, and special staplers, laparoscopy again became a viable concept for surgery on the pancreas.
The role of laparoscopy in pancreatic surgery was still a controversial topic in discussions at the annual meeting of the German Society for General and Visceral Surgery (Deutsche Gesellschaft für Allgemein- und Viszeralchirurgie, DGAV) in 2007. In 2008, surgeons with an interest in this area joined together to create a data registry. At the first consensus conference, held in 2011, criteria for the selection of especially suitable patients were determined and then published (4). Promising findings were already available from registries of laparoscopic pancreatic surgery in other countries (5, 6).
As stated in the IDEAL criteria (7), when a new technique is introduced, the outcomes after its use should be documented quantitatively so that its potential can be properly assessed. In particular, its advantages (e.g., easier recovery or less pain for the patient, with comparable final outcomes to those of standard technique) must be weighed against its disadvantages (e.g., the development of diabetes mellitus).
By the time the second consensus conference was held in 2013, the registry already contained information on 270 patients. It was decided then that the technique would be evaluated in accordance with the IDEAL criteria once data from 500 patients had been compiled. In the meantime, the number of participating centers has risen steeply to 34, and the laparoscopic pancreatic surgery registry now contains data on 550 patients. As far as we can tell from the published international literature, this is the largest such database anywhere in the world.
The results of laparoscopic surgery as determined from this registry, in terms of complication rates, surgical outcomes, and length of hospital stay, must be compared with results from other registries, e.g., those of Belgium and Spain (8).
Materials and Methods
In 2008, the registry for laparoscopic pancreatic surgery was established and was evaluated and approved by the Ethics Committee of the University of Ulm. The registry is a platform for the documentation of all laparoscopic procedures on the pancreas. Participation is voluntary; data are acquired by being entered onto a standardized form. The registry has been presented multiple times at major surgical congresses in Germany (e.g., the annual meetings of the societies for general surgery and visceral medicine), and these presentations were exploited as opportunities to acquire new participating institutions. All surgeons who perform laparoscopic pancreatectomy were invited to participate voluntarily in the registry. The registry has also been presented to DGAV working groups such as those for the liver, biliary pathways, and pancreas (Chirurgische Arbeitsgemeinschaft Leber-Galle-Pankreas, CALGP) and for minimally invasive surgery (Chirurgische Arbeitsgemeinschaft Minimal-Invasive Chirurgie, CAMIC). Subsequently, the presidents of these working groups recommended participation in the registry.
The progress of the registry was reported on once a year at the annual meetings of each of the two major German surgical societies (Fachgesellschaft fur Chirurgie, Deutsche Gesellschaft für Chirurgie). Data on laparoscopic procedures that were converted to open surgery were also included. The data were acquired by entry onto standardized forms for the documentation of key pre-, intra-, and postoperative information (e.g., histopathology) and the patients’ further course (e.g., length of hospital stay, pain score, and complications). Only complete data sets were accepted, with the further requirement that the data must pass a plausibility test.
The authors of the present review are from the four centers that have contributed the largest number of patients to the registry; all four centers have participated in the registry since its inception. The other highly active participating centers and surgeons are listed by name at the end of the article.
The main features of the patient cohort are listed in Table 1. 267 patients (48.5% of the total) had benign lesions. These comprised mainly
- Serous cystadenomas (58/10.5%),
- Neuroendocrine tumors (47/8.5%),
- Intraductal papillary-mucinous neoplasia (43/7.8%),
- Mucinous-cystic neoplasia (36/6.5%),
- Chronic pancreatitis (35/6.4%), and
- Pseudocysts (16/2.9%).
Rarer benign lesions accounted for the remaining 32 cases.
39 patients (7.1%) had borderline tumors, which, in turn, were further broken down into neuroendocrine tumors (23/4.2%), intraductal oncocytic papillary neoplasia (IPMN) (11/2%), and mucinous-cystic tumors (5/0.9%).
244 patients (44.4%) had cancerous lesions. The most common type was ductal adenocarcinoma (110/20%), followed by neuroendocrine carcinoma (79/14.4%). Rarer entities included metastases of other types of cancer (12/2.2%) and the following:
- Carcinoma of the papilla of Vater (8/1.5%)
- Mucinous cystadenocarcinoma (6/1.1%)
- Adenenocarcinoma on the background of intraductal papillary-mucinous neoplasia (6/1.1%)
- Acinar cell carcinoma (5/0.9%)
- Bile duct carcinoma (3/0.5%)
- Ampullary carcinoma (3/0.5%)
- Serous cystadenocarcinoma (2/0.36%)
- B-cell lymphoma (2/0.36%)
- Intestinal-type adenocarcinoma (2/0.36%)
- Solid pseudopapillary carcinoma (2/0,36%).
There was one case each (0.18%) of malignant pheochromocytoma, carcinoid of the ampulla of Vater, duodenal carcinoma, and carcinosarcoma.
Among the types of laparoscopic procedure that were performed, left pancreatectomy was the most common (65%), followed by pylorus-preserving resection of the head of the pancreas (23%) and lesional enucleation (9%). Isolated segmental and pancreas-head resections were also performed, as well as five total pancreatectomies.
Laparoscopic left pancreatectomy
Laparoscopic left pancreatectomy was performed in 359 cases (65%) and was the most common procedure. The spleen was removed at the same procedure in 56% of these cases. Of the 201 cases in which preservation of the spleen was intended at the outset (e.g., in patients with benign diseases), the spleen was in fact preserved in 157 cases (78%). The mean duration of the procedure was 220 minutes, and the mean size of the resected pancreatic specimen was 9.0 cm (standard deviation, 3.5 cm).
The instrumentation used for dividing the pancreas is an important matter, as one of the goals of surgery is to prevent the postoperative development of a pancreatic fistula. The EndoGIA stapler was used in 59% of the patients, the Ligasure vessel sealer and divider in 25%, and the Ultracision Harmonic Ace Scalpel in 6%. No further information on laparoscopic instruments was entered into the registry.
Laparoscopic resection of the head of the pancreas
Laparoscopic resection of the head of the pancreas was the second most common type of procedure (127 cases, 23%). It was associated with 0% mortality within 60 days, as reflected by the data entered into the standardized questionnaire. The median length of hospital stay was 15 days (range, 7–59). Procedure-associated postoperative complications were experienced by 97 patients.
In 48 cases, lesions were laparoscopically enucleated (9%). The mortality associated with these procedures was also 0%. The median length of hospital stay was 11 days (range, 4–67). 23 patients had postoperative complications.
Intra- and postoperative course of laparoscopic pancreatectomy
In 16% of all patients—not counting those who underwent a laparoscopically assisted, pylorus-preserving pancreaticoduodenectomy (PPPD)—a laparoscopic procedure was converted intraoperatively to open surgery. The median time from the beginning of the procedure to the conversion was 135 minutes. The reasons given for conversion were:
- Spread of cancer (n = 7)
- No tumor found (n = 5)
- Splenic hemorrhage (n = 4)
- Necrotizing pancreatitis (n = 1)
- Pancreatitis (n = 1)
- Dysfunction of an EndoGIA stapler (n = 1)
- Adhesions (n = 8)
- Obesity (n = 3)
- Inflammation involving the head of the pancreas (n = 3)
- Pancreatectomy (n = 4)
- Iatrogenic injury of the mesentery of the transverse colon (n = 1).
The reasons for conversion are listed in greater detail in Table 2.
As for intraoperative complications, hemorrhage beyond the amount typically expected (500 mL per procedure) was documented in 3.1% of patients. Further complications, occurring in one patient each, were iatrogenic injuries to the mesentery of the transverse colon, the splenic vein, the common bile duct, and the diaphragm. The spleen was injured intraoperatively in four cases.
As for postoperative complications, postoperative hemorrhage not requiring reoperation arose in 27 patients (4.9%), and postoperative hemorrhage leading to reoperation arose in 17. Thirty-nine percent of all patients developed a pancreatic fistula (mostly grade A and B fistulae, with 8 grade C fistulae in the entire patient cohort). 5.5% of patients developed an abscess, and 1.1% developed a splenic infarct. Seven patients died for reasons related to the procedure (1.3%); 3.5% had a postoperative wound infection. A more detailed listing of all intra- and postoperative complications is given in Table 3.
The median length of hospital stay was 14 days (range, 4–93 days).
As for postoperative pain, 91% of the patients had a Visual Analog Score (VAS) less than 4. Diabetes mellitus was already present preoperatively in 15.6% of all patients and newly arose postoperatively in a further 6.7% of all patients.
The three most common types of laparoscopic surgery are compared in Table 4. The 57–76% morbidity mentioned in the table may seem very high at first but is accounted for by the conscientious recording of cases of grade A pancreatic fistula or grade A gastric emptying disturbance, both of which are generally clinically irrelevant conditions.
This German-language registry, which now contains data on 550 patients, is currently the world‘s largest database on laparoscopic pancreatic surgery. The 34 participating centers are to be commended for their unprecedented collaboration and mutual solidarity under the aegis of the German Society for General and Visceral Surgery.
We are well aware that the registry data presented here are not free of bias. The advantages and disadvantages of clinical registry studies in comparison to prospective randomized controlled trials are well known: even though the latter are now considered the gold standard for testing a new technique, the former are much better at reflecting the overall current care situation in the geographical region of interest (9). Further advantages of registry studies are the following:
- No prior restriction on the number of patients
- Heterogeneous patient group
- Heterogeneous conditions and centers
- Open inclusion
- Safety and goal-directedness
- Choice of either a prospective or a retrospective design.
The registries in other countries that are known to us from the published literature consist of data derived from no more than 3–9 centers per registry.
It is clear that the participating centers selected patients for laparoscopic pancreatic surgery whom they considered to be highly suitable for the procedure (sample criteria: thin patient, small and easily accessible tumor). The percentage of laparoscopic operations among all pancreatic procedures in each participating hospital was 5–20%, which, of course, implies that 80–95% of the procedures were conventional open surgery. In our patient cohort, just as in the cohorts of earlier publications (8), left pancreatectomy was performed more often by laparoscopy. This is due to selection bias, as resections of the head of the pancreas predominate in conventional pancreatic surgery (10). Tumors located in the left side of the pancreas are less common, and this would make a randomized controlled trial difficult to perform, as the small group of candidates for left pancreatectomy would have to be subdivided into two separate, even smaller treatment groups (one for laparoscopic and one for open surgery).
Malignant tumors of the pancreas account for a higher percentage of the lesions resected laparoscopically in the German registry than in a comparable Belgian study (8). The reason is that, as surgeons become more experienced in the minimally invasive technique, it is increasingly used for malignant tumors as well. This is confirmed by a trend over time in the German registry: in 2011, only 20% of the laparascopic procedures were performed in patients with malignant disease (4), but, in the meantime, this figure has risen to 44.4%.
As shown in Table 5, extensive case series of pancreatectomies are now available from multiple industrialized countries. In some of them, the morbidity and mortality associated with laparoscopic pancreatic surgery are low and comparable to those associated with conventional resection. In 2015, Adam et al. reported on a series of 7061 patients—a subset of the National Cancer Database (11), which comprises 75% of newly diagnosed cases of cancer in the USA—among whom 983 (14%) underwent laparoscopic pancraticoduodenectomy. The conversion rate was 30%. The mortality of laparoscopic surgery was 4.8%, compared to 3.7% for open pancreaticoduodenectomy. Thus, in this study, the procedure was found to be widely applicable, though still associated with a somewhat higher mortality than conventional surgery. The German-language registry, in contrast, revealed no difference in mortality. This may be due to the numerical predominance of left pancreatectomies in the registry, with right-sided resections only recently being performed at an appreciable rate. Two current publications by Sulpice et al, and Riviere et al. show no clear superiority of either technique (laparoscopic vs. conventional surgery) for left-sided resections (12, 13).
The registry data can also be compared with those of the Initiative for Quality in Medicine (Initiative Qualitätsmedizin, IQM), a consortium of 387 hospitals jointly pursuing objective quality control. The mortality among the 4845 cases reported to the IQM in 2015 was 8.3% (14). The 1.3% mortality documented by the German registry is markedly lower. On the other hand, the 39% rate of postoperative pancreatic fistula formation in the German registry is higher than the 30% rate reported in the literature (15).
The reported rates of newly arising diabetes mellitus after conventional Whipple procedures vary from 10% to 37% (16, 17). This value rises to as high as 60% after surgery for chronic pancreatitis (10). The much lower rate of 6.7% observed in the German registry cohort may be due to selection bias.
Laparoscopic pancreatectomy is not yet listed as a standard procedure in the German S3 guideline that was issued in October 2013 (18). In the guideline, it is recommended that all laparoscopic resections should be documented in a registry such as the one presented in here, so that new therapeutic techniques can be adequately assessed. We endorse this recommendation.
The quantifiable advantages of laparoscopic pancreatic surgery, particularly laparoscopic left pancreatectomy, for patients in comparison to conventional procedures include a low complication rate—in particular, a low rate of wound infections (19)—along with low mortality and shorter hospital stays. As the next step in the objective evaluation of laparoscopic pancreas surgery, according to the principles of evidence-based medicine, these conclusions derived from registry data will have to be tested in prospective randomized controlled trials.
We thank all of the employees of the institutions responsible for the planning and implementation of the registry, and we thank all employees of the study centers for their support in patient recruitment. We also thank the German Society for General and Visceral Surgery, the Surgical Working Group on the Liver, Biliary Pathways, and Pancreas, and the Surgical Working Group on Minimally Invasive Surgery.
Participating study centers
Universitätsklinikum Giessen/Marburg (Detlef K. Bartsch, Peter Langer, C. Schicker); Universitätsklinikum Freiburg (Uwe Wittel, Tobias Keck, Ulrich Hopt); Kliniken Essen-Mitte (Martin K. Walz, Jakob Hinrichs), Evangelische Kliniken Bonn (Andreas Türler); Universitätsklinikum Schleswig-Holstein Campus Lübeck (Uwe J. Roblick, Ulrich Wellner, Tobias Keck, Hans-Peter Bruch); Krankenhaus der Elisabethinen Linz (Reinhold Függer); Universität Greifswald (Maciej Patrzyk, C.-D. Heidecke); Klinikum Südstadt Rostock (Kaja Ludwig); Universitätsmedizin Mainz (Thomas J. Musholt); St. Josef-Hospital Bochum (Kirsten Meurer, Orlin Belyaev, Waldemar Uhl); Klinikum St. Elisabeth Straubing (Robert Obermaier); Klinik Nürtingen (Svenja Haaga, Klaus Kraft), Universitätsklinikum Heidelberg (Jens Werner, Markus W. Büchler); Krankenhaus Dresden-Friedrichstadt (Olaf Pridöhl, Helmut Witzigmann); Klinikum Augsburg (Bernd Geissler, Matthias Anthuber); Lukas Krankenhaus Bünde (Steffen Krummbein); Schön Klinik Hamburg-Eilbeck (Lutz Steinmüller); Schön Klinik Fürth (Thomas C. Böttcher); Zentralklinik Bad Berka (Volker Weisse, Merten Hommann); St. Vinzenz-Hospital, Cologne (Thomas Wilhelm); Universitätsklinikum Hamburg-Eppendorf (Philipp Busch, Jakob R. Izbicki); Helios Kliniken Schwerin (Jörg-Peter Ritz); Kliniken der Stadt Köln, Cologne (Dirk Bulian, Markus M. Heiss); Klinikum Stuttgart (Jörg Köninger); Krankenhaus Zweibrücken (Dieter Birk); Kantonsspital Baselland (Andreas Zerz); Rotkreuzklinikum, Munich (Peter Schwendtner, M.H. Schoenberg); Klinikum Dritter Orden, Munich (Peter Kapfhammer, Detlef Krenz); Klinikum Böblingen, prev. Nagold (Stefan Benz); Universität Magdeburg (Benjamin Garlipp, Christiane Bruns); Caritas Klinikum Saarbrücken (Frank Schütze, Ralf Metzger); Immanuel Klinik Rüdersdorf (Jens Burghardt); Universität des Saarlandes (Matthias Glanemann)
Conflict of interest statement
The authors declare that no conflict of interest exists.
Manuscript received on 12 October 2016; revised version accepted on 14 December 2016.
Translated from the original German by Ethan Taub, M.D.
Prof. Dr. med Marco Siech
Chirurgie I, Ostalb-Klinikum Aalen
Im Kälblesrain 1, 73430 Aalen, Germany
Department of Surgery I, Ostalb Klinikum Aalen: Prof. Dr. med. Siech, Peter Strauss, Dr. med. Huschitt
Department of General and Visceral Surgery, Medical Center—University of Freiburg: Prof. Dr. med. Wittel
Department of Visceral, Thoracic and Vascular Surgery University of Marburg: Prof. Dr. med. Bartsch
Department of Surgery, University Medical Center-UKSH, Lübeck: Prof. Dr. med. Keck
|1.||van der Pas MH, Haglind E, Cuesta MA, et al.: Laparoscopic versus open surgery for rectal cancer (COLOR II): short-term outcomes of a randomised, phase 3 trial. Lancet Oncol 2013; 14: 210–8 CrossRef MEDLINE|
|2.||Cuschieri A: Laparoscopic surgery of the pancreas. J R Coll Surg Edinb 1994; 39: 178–84 MEDLINE|
|3.||Gagner M, Pomp A: Laparoscopic pyloruspreserving pancreatoduodenectomy. Surg Endosc 1994; 8: 408–10 CrossRef|
|4.||Siech M, Bartsch D, Beger HG, et al.: Indications for laparoscopic pancreas operations: results of a consensus conference and the previous laparoscopic pancreas|
register. Chirurg 2012; 83: 247–53 CrossRef MEDLINE
|5.||Vijan SS, Ahmed KA, Harmsen WS, et al.: Laparoscopic vs open distal pancreatectomy: a single-institution comparative study. Arch Surg 2010; 145: 616–21 CrossRef MEDLINE|
|6.||Weber SM, Cho CS, Merchant N, et al.: Laparoscopic left pancreatectomy: complication risk score correlates with morbidity and risk for pancreatic fistula. Ann Surg Oncol 2009; 16: 2825–33 CrossRef MEDLINE|
|7.||McCulloch P, Altman DG, Campbell WB, et al.: No surgical innovation without evaluation: the IDEAL recommendations. Lancet 2009; 374: 1105–12 CrossRef|
|8.||Mabrut JY, Fernandez-Cruz L, Azagra JS, et al.: Hepatobiliary and Pancreatic Section (HBPS) of the Royal Belgian Society of Surgery; Belgian Group for Endoscopic Surgery (BGES); Club Coelio. Laparoscopic pancreatic resection: results of a multicenter European study of 127 patients. Surgery 2005; 137: 597–605 CrossRef|
|9.||Stey AM, Russell MM, Ko CY, Sacks GD, Dawes AJ, Gibbons MM: Clinical registries and quality measurement in surgery: a systematic review. Surgery 2015; 157: 381–95 CrossRef MEDLINE|
|10.||Keim V, Klar E, Poll M, Schoenberg MH: Postoperative care following pancreatic surgery: surveillance and treatment. Dtsch Arztebl Int 2009; 106: 789–94 VOLLTEXT|
|11.||Adam MA, Choudhury K, Dinan MA, et al.: Minimally invasive versus open pancreaticoduodenectomy for cancer: practice patterns and short-term outcomes among 7061 patients. Ann Surg 2015; 262: 372–7 CrossRef MEDLINE|
|12.||Sulpice L, Farges O, Goutte N, et al.: Laparoscopic distal pancreatectomy for pancreatic ductal adenocarcinoma: time for a randomized controlled trial? Results of an all-inclusive national observational study. Ann Surg 2015; 262: 868–73 CrossRef MEDLINE|
|13.||Riviere D, Gurusamy KS, Kooby DA, et al.: Laparoscopic versus open distal pancreatectomy for pancreatic cancer. Cochrane Database Syst Rev 2016 CrossRef|
|14.||Martin J, Rink O, Zacher J: Handbuch IQM konsequent transparent. 1st edition. Berlin: IQM Initiative Qualitätsmedizin und MWV Medizinisch Wissenschaftliche Verlagsgesellschaft 2014.|
|15.||Diener MK, Seiler CM, Rossion I, et al.: Efficacy of stapler versus hand-sewn closure after distal pancreatectomy (DISPACT): a randomised, controlled multicentre trial. Lancet 2011; 377: 1514–22 CrossRef|
|16.||Jimenez RE, Fernandez-del Castillo C, Rattner DW, Chang Y, Warshaw AL: Outcome of pancreaticoduodenectomy with pylorus preservation or with antrectomy in the treatment of chronic pancreatitis. Ann Surg 2000; 231: 293–300 CrossRef|
|17.||Hamilton L, Jeyarajah DR: Hemoglobin A1c can be helpful in predicting progression to diabetes after whipple procedure. HPB (Oxford) 2007; 9: 26–8 CrossRef MEDLINE PubMed Central|
|18.||Seufferlein T, Porzner M, Becker T, et al.: S3-guideline exocrine pancreatic cancer. Z Gastroenterol 2013; 51: 1395–440 MEDLINE|
|19.||Okano K, Hiaro T, Unno M, et al.: Postoperative complications after pancreatic resection. Brit J Surg 2015; 102: 1551–60 CrossRef MEDLINE|
|20.||Jayaraman S, Gonen M, Brennan MF, et al.: Laparoscopic distal pancreatectomy: evolution of a technique at a single institution. J Am Coll Surg 2010; 211: 503–9 CrossRef MEDLINE|
|21.||Song KB, Kim SC, Park JB, et al.: Single-center experience of laparoscopic leftpancreatic resection in 359 consecutive patients: changing the surgical paradigm of left pancreatic resection. Surg Endosc 2011; 25: 3364–72 CrossRef MEDLINE|
|22.||Zhou ZQ, Kim SC, Song KB, et al.: Laparoscopic spleen-preserving distal pancreatectomy: comparative study of spleen preservation with splenic vessel resection and splenic vessel preservation. World J Surg 2014; 38: 2973–9 CrossRef MEDLINE|
|23.||Croome KP, Farnell MB, Que FG, et al.: Total laparoscopic pancreaticoduodenectomy for pancreatic ductal adenocarcinoma: oncologic advantages over open approaches? Ann Surg 2014; 260: 633–8 CrossRef MEDLINE|
|24.||Song KB, Kim SC, Hwang DW, et al.: Matched case-control analysis comparing laparoscopic and open pylorus-preserving pancreaticoduodenectomy in patients with periampullary tumors. Ann Surg 2015; 262: 146–55 CrossRef MEDLINE|
|25.||Lee SY, Allen PJ, Sadot E, et al.: Distal pancreatectomy: a single institution‘s experience in open, laparoscopic, and robotic approaches. J Am Coll Surg 2015; 220: 18–27 CrossRef MEDLINE|