- Email: [email protected]
Agreement between reported use of interventions for liver diseases and research evidence in Cochrane systematic reviews
Journal of Hepatology 43 (2005) 984–989 www.elsevier.com/locate/jhep
Agreement between reported use of interventions for liver diseases and research evidence in Cochrane systematic reviews* Pia Ku¨rstein1,2, Lise L. Gluud2, Marlene Willemann1, Kim R. Olsen1, Jakob Kjellberg1, Jes Sogaard1, Christian Gluud2,* 2
1 DSI Danish Institute for Health Services Research, Copenhagen, Denmark The Cochrane Hepato-Biliary Group, Copenhagen Trial Unit, Centre for Clinical Intervention Research, Copenhagen, Denmark
Background/Aims: This study evaluates the agreement between reported use of interventions for patients with liver diseases and research evidence in Cochrane systematic reviews. Methods: In July 2002, the Cochrane Hepato-Biliary Group had completed 28 systematic reviews on 36 interventions that were available in Denmark. Based on the reviews, three interventions (n-acetylcysteine for paracetamol overdose, terlipressin for bleeding oesophageal varices, and antibiotics for patients with cirrhosis and gastrointestinal bleeding) with significant beneficial effects on clinical outcomes were classified as ‘evidence-based’, whereas 19 were classified as ‘possibly evidence-based’, and 14 as ‘not evidence-based’. Questionnaires on reported use and perceived intervention effects were mailed to 108 physicians practising in Danish hospitals. Sixty-six returned their questionnaire. Results: The proportion of physicians who reported that they never used the three evidence-based interventions varied considerably (2, 62, and 57%, respectively). The perceived intervention effect, duration of clinical experience, employment as head of department, and university hospital employment were significant predictors of more frequent use of evidence-based interventions. Physicians also reported that they used the interventions that were not evidencebased more often if they were employed at a university hospital. Conclusions: Considerable disagreements between reported use and research evidence were identified. Additional research on methods to introduce evidence-based medicine in practice seems warranted. q 2005 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. Keywords: Evidence-based medicine; Systematic reviews; Randomised trials; Meta-analyses; Clinical practice
1. Introduction Evidence-based medicine combines clinical experience with the most valid research evidence [1,2]. Although evidence-based medicine has an increasing number of
Received 20 January 2005; received in revised form 15 May 2005; accepted 6 June 2005; available online 12 July 2005 * All authors declare that they have no competing interests with regard to the submitted work. * Corresponding author. Tel.: C45 3545 7175; fax: C45 3545 7101. E-mail address: [email protected] (C. Gluud).
supporters, several disagreements seem to exist between clinical practice and research evidence [3–5]. The disagreements may be related to the clinical setting or lack of familiarity with the evidence [6]. Identification of disagreements between clinical practice and research evidence may help target initiatives to promote evidence-based clinical practice. Generally, systematic reviews of high quality randomised clinical trials provide the most valid evidence for health care interventions [7– 10]. The present study evaluates the agreement between reported use and research evidence in systematic reviews of interventions for patients with liver diseases.
0168-8278/$30.00 q 2005 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhep.2005.06.017
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
2. Methods 2.1. Participants In 2002, six university departments and 61 non-university departments in Denmark were responsible for treatment of patients with liver diseases. We selected all university departments and randomly selected 19 of the non-university departments by drawing of lots from an opaque container. Descriptions of the study and consent forms were sent to the administrative head of the selected departments. Twenty-four departments agreed to participate. One university department declined participation. Through correspondence with the selected departments, we listed all physicians who were employed at the participating departments. The physicians were listed and classified as specialists or non-specialists based on their title and speciality. Specialists were head of department and gastroenterologists or hepatologists or had the overall responsibility for treatment of patients with liver disease. Non-specialists were senior registrars or consultants who were not specialists in gastroenterology or hepatology. We aimed at including three specialists and three non-specialists from each department. Some departments employed fewer than three specialists or non-specialists. Other departments employed more than six potentially eligible participants. For those departments, we randomly selected three specialists and three non-specialists by drawing lots from an opaque container. In total, 54 specialists and 54 non-specialists from the 24 included departments were invited to participate.
2.2. Questionnaire The Cochrane Library has been available free of charge for all Danish physicians since 1999 [11–13]. The Cochrane Hepato-Biliary Group had completed 28 systematic reviews in July 2002 [14–42]. Some of the reviews described more than one intervention. In total, the reviews evaluated 37 interventions for prevention or treatment of 16 liver diseases. Thirty-six of these interventions were available in Denmark. A questionnaire on reported use and perceived effects of these interventions was developed. For each disease, the interventions were listed in random order. All questions were closed. Physicians were asked whether they used the interventions always, often, sometimes, rarely, or never for patients who fulfilled the indication and had no contraindications. Physicians were also asked whether they believed that the intervention affects mortality or morbidity (perceived intervention effects).
2.3. Statistical analysis Lists of the most important clinical and surrogate outcome measures were prepared for each of the evaluated liver diseases. Evidence of intervention effects, point estimates, and 95% confidence intervals (CI) for the listed outcome measures was extracted from the systematic reviews. Effects were classified as significant if the CI did not overlap one. None of the evaluated interventions had significant harmful effects. Interventions with significant beneficial effects on clinical outcomes were classified as ‘evidence-based’. Interventions with significant effects on surrogate outcomes or less beneficial effects than other interventions were classified as ‘possibly evidence-based’. Interventions for which there was no evidence of significant effects on clinical or surrogate outcomes were classified as ‘not evidence-based’. Physicians who responded were included in our analyses. Proportions were compared using chi-square and are presented as means with 95% CI. Regression analyses were performed to identify potential predictors of reported use of evidence-based interventions and interventions that are not evidence-based. The dependent variable (reported use) was entered as an ordinal scale [43]. Results are presented as regression coefficients (b) with P values. A negative b indicate more frequent use of evidence-based interventions. A positive b indicate more frequent use of interventions that were not evidence-based. The regression models were adjusted for the type of intervention. The potential predictors included the physicians’ perceived intervention effects, type of hospital (employment at university hospital), duration (years) of clinical experience, position (employment as head of department), clinical speciality (hepatology, gastroenterology, or other), and geographical region (location of hospital). Responses from physicians at university departments were weighted with one third to adjust for overrepresentation [44]. Analyses were performed in Stata version 8.0 for Windows (Stata Corp., Texas, USA) and SPSS version 10 software (SPSS Inc., Chicago, IL).
3. Results Questionnaires were sent to 108 physicians. Sixty-six (61%) physicians agreed to participate and returned their questionnaires. Forty-six of the participating physicians (70%) were employed at a university hospital and 34 (52%) were specialists. The mean duration of clinical experience was 19 years (SD 8 years).
Table 1 Interventions classified as ’possibly evidence based’ Disease
Intervention
Hepatitis B
Yeast derived hepatitis B vaccine [15] Glucocorticosteroids and interferon [36] Ursodeoxycholic acid [24] Chinese medical herbs [38,39] Interferon [35] Ursodeoxycholic acid [17] Interferon [34] Interferon plus ribavirin [40] Active charcoal [22] Ventricular lavage [22] Ipecacuanha [22] Ursodeoxycholic acid [17] Antibiotics [28] Sclerotherapy [14] Somatostatin [18] Surgical anastomosis [19] Transjugular intrahepatic portosystemic shunt [19] Liver support systems [41] Branched chain amino acids, intravenous [23]
Hepatitis C, acute Hepatitis C, chronic
Paracetamol overdose
Primary biliary cirrhosis Spontaneous bacterial peritonitis Bleeding oesophageal varices
Liver failure Hepatic encephalopathy
985
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
986
Table 2 Interventions classified as ’not evidence-based’ Disease
Intervention
Hepatitis B
Alpha-fetoprotein screening [27] Ultrasonography [27] Alpha-fetoprotein screening plus ultrasonography [27] Chinese medical herbs [37] Chinese medical herbs [37] Haemoperfusion [22] Colchicine [31] Anabolic-androgenic steroids [29] Propylthiouracil [30] S-adenosyl-L-methionine [32] Benzodiazepine receptor antagonists [25]
Hepatitis C, acute Hepatitis C, chronic Paracetamol overdose Cirrhosis Alcoholic cirrhosis
Hepatic encephalopathy Hepatocellular carcinoma
More frequent reported use of evidence-based interventions was significantly associated with university hospital employment (bZK0.97, PZ0.03), position as head of department (bZK0.99, P!0.001), duration of clinical experience (bZK0.20; PZ0.02), and perceived intervention effects (bZK0.61, P!0.001). The perceived intervention effect was a significant predictor of less frequent reported use of interventions that were not evidence-based (bZK0.45, P!0.001). Physicians at university hospital departments reported that they used the evidence-based interventions more often that the physicians at nonuniversity departments did (bZ1.57, P!0.001). None of the remaining variables were significantly associated with reported use.
Branched-chained amino acids, peroral [23] Neoadjuvant and adjuvant therapy [21] Radiofrequency thermal ablation [20]
Based on the Cochrane systematic reviews, three interventions were classified as ‘evidence-based’, 19 interventions as ‘possibly evidence-based’ (Table 1), and 14 interventions as ‘not evidence-based’ (Table 2). The ‘evidence-based’ interventions were n-acetylcysteine for paracetamol overdose [22], terlipressin for bleeding oesophageal varices [26], and antibiotic prophylaxis for patients with cirrhosis and bleeding oesophageal varices [42]. The proportion of physicians who reported that they never used the evidence-based interventions was 2, 62, and 57%, respectively (Table 3). The mean proportion of physicians who reported that they never used the interventions was 33% (95% CI: 22–46%) for ‘evidence-based’ interventions compared with 62% (95% CI: 49–74%) for interventions that were ‘possibly evidencebased’, and 80% (95% CI: 69–89%) for interventions that were ‘not evidence-based’ (P!0.01) (Table 4). The proportion of physicians who believed that the evidence-based interventions did not reduce mortality was 1, 54, and 71%, respectively. On average, 31% (95% CI: 20– 43%) of the physicians believed that the evidence-based interventions did not reduce mortality compared with 59% (95% CI: 46–71%) for the interventions that were possibly evidence-based and 87% (95% CI: 76–94%) for interventions that were not evidence-based (P!0.01).
4. Discussion The present study identified three interventions with significant beneficial effects on clinically relevant outcome measures in Cochrane systematic reviews of randomised clinical trials. The interventions were n-acetylcysteine for paracetamol overdose, terlipressin for bleeding oesophageal varices, and antibiotic profylaxis for patients with cirrhosis and gastrointestinal bleeding [22,26,42]. The evidence presented in the systematic reviews suggests that these three interventions should be considered as the treatment of choice for patients who fulfil the indication criteria and none of the contraindication criteria. Most of the physicians reported that they used n-acetylcysteine, but several reported that they never used terlipressin or antibiotic profylaxis. Physicians were more likely to report that they never used the interventions that were not supported by research evidence than the evidence-based interventions. Still, there were exceptions. Some physicians reported that they always or often used the interventions that were not evidence-based. The present study cannot identify the underlying reasons for the disagreements between reported use and research evidence. Additional research on methods to increase implementation of Cochrane systematic reviews in clinical practice seems warranted. Our results suggest that valid research evidence does not seem to be enough to implement interventions. Previous questionnaire studies have found that barriers to physician guideline adherence
Table 3 Reported use of ’evidence-based’ interventions
N-acetylcysteine for paracetamol overdose [22] Terlipressin for bleeding oesophageal varices [26] Antibiotic prophylaxis for patients with cirrhosis and gastrointestinal bleeding [42] a
Presented data are proportions (nZ66 physicians)
Reported frequency of usea Always
Often
Sometimes
Rarely
Never
88 17 3
6 13 10
4 7 13
0 0 16
2 62 57
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
987
Table 4 Reported use of interventions classified as ‘evidence-based’, ‘possibly evidence-based’, or ‘not evidence-based’ Reported usea Always Often Sometimes Rarely Never a
Interventions ’Evidence-based’
’Possibly evidence-based’
’Not evidence-based’
39 (28-52) 9 (6-24) 5 (3-17) 4 (2-15) 33 (22-46)
7 (4-21) 6 (3-19) 5 (3-17) 7 (4-21) 62 (49-74)
2 (0.1-8) 6 (2-15) 6 (2-15) 6 (2-15) 80 (69-89)
Presented data are mean proportions with 95% CI (nZ66 physicians).
include lack of awareness, familiarity, or agreement, selfefficacy, outcome expectancy, inertia of previous practice, and external barriers [6]. Similar barriers are likely to exist for implementation of systematic reviews on liver diseases in clinical practice. The present study found that Danish physicians were more likely to use an intervention if they believed that it had a beneficial effect. A national questionnaire survey of American gastroenterologists [45] asked how often they used ablation therapy for Barrett’s oesophagus. Few found that ablation is effective, but a significant proportion had adopted its use. The strongest predictor of ablation use was the belief that colleagues used it. In a similar study, 195 Swiss gastroenterologists were asked how often they used antibiotic prophylaxis for digestive endoscopy [46]. When compared to the local guidelines, the physicians seemed to use antibiotic prophylaxis too often. The authors argued that the overuse was partly explained by the weak strength of the scientific evidence in the guidelines [46]. In the present study, we found that antibiotic prophylaxis for patients with cirrhosis and gastrointestinal bleeding may be under-used although the scientific evidence is strong [42]. Terlipressin for bleeding oesophageal varices may also be used too infrequently [26]. Several factors may affect the extent to which the results of randomised trials and systematic reviews are used in clinical practice. Both the internal and the external validity of randomised trials are important. Sometimes, the internal validity is questionable and the results are likely to be seriously flawed. In that case, the evidence cannot be used in clinical practice and the assessment of external validity becomes irrelevant. On the other hand, even if a trial has high internal validity, clinicians may not be able to use the trial results because the evidence cannot be extrapolated to their patients. This situation often occurs when the inclusion and exclusion criteria in the trial are too restrictive. Understanding the predicaments of a patient is a key factor that must be considered in the choice of the best treatment. In our questionnaire survey, we specified that our questions concerned typical patients that fulfilled the indication and had no contraindications. Still, in clinical practice, there may be special circumstances, which affect the choice of treatment. There may be some discrepancies between reported use and observed clinical practice, but overall, discrepancies are
likely to be small or moderate. The present study suggests physicians seemed to use the evidence-based interventions more often if they were head of department, employed at a university hospital, or had a long clinical experience. Physicians at university departments also used the interventions that were not evidence-based more than physicians at non-university departments did. Initiatives for implementation of evidence-based clinical practice may need to consider these aspects. It is possible that younger physicians need more definitive guidelines than the more experienced physicians. In general, Cochrane systematic reviews present the research evidence as objectively as possible. Specific recommendations for clinical practice are often not included in the review conclusions because of insufficient or inconsistent evidence regarding the optimal dose or duration of treatment. A certain amount of baseline knowledge is required to implement interventions based on Cochrane reviews alone. Evidence-based guidelines may be useful. Guidelines may provide the additional information that the younger physicians cannot extract from Cochrane reviews. Often, guidelines are prepared and used locally. Regional, national, or even international guidelines may help decrease some of the discrepancies in clinical practice found in the present study. In a review of questionnaire studies on barriers for adherence to clinical practice guidelines, the median response rate for reporting barriers was 54.5% (range 26– 95%) [6]. The risk of bias increases with the proportion of non-responders because non-responders may be systematically different from responders. In the present study the response rate was 61%. Physicians were more likely to respond if they were employed at a university department. Responders also tended to have more clinical experience than nonresponders. It is likely that physicians with a positive attitude to evidence-based medicine were more likely to respond. Our survey may therefore underestimate the gap between research evidence and clinical practice. Evidence-based health care incorporates knowledge from clinical practice and research. Opponents may argue that reliance on research evidence leads to ‘cook-book’ medicine that overlooks the importance of tacit clinical knowledge. These arguments may explain lack of adherence to evidence-based guidelines [6]. Other potential barriers include both internal and external barriers such as physician knowledge or logistic considerations. Use of electronic
988
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
medical records and regular performance reports improved adherence to clinical practice guidelines for cardiovascular disease and stroke prevention [47]. Educational interventions were found to improve adherence to guidelines and yielded a better outcome for patients with a high-risk of acute coronary syndrome [48]. Education may also increase agreement between clinical practice and the evidence presented in Cochrane reviews. Few Danish physicians use The Cochrane Library regularly although it is available through free Internet access [49]. Internet access relies on passive dissemination of the results of the systematic reviews. Possibly, active rather than passive dissemination may increase the influence of Cochrane reviews on clinical practice [50]. Acknowledgements Supported by grants from The Danish Medical Research Council, The 1991 Pharmacy Foundation, DSI Danish Institute for Health Services Research, and Copenhagen Trial Unit, Centre for Clinical Intervention Research, Denmark. We thank the physicians who participated in our survey and Thorkild Thorsen and Erik Christensen for valuable comments on the design of our survey. We also thank the authors and peer reviewers who were involved in the development and interpretation of the included systematic reviews and the editors and the staff at the Editorial Team Office of The Cochrane-Hepato-Biliary Group.
References [1] Sackett D, Rosenberg WM, Gray JA, Haynes RB, Richardson WS. Evidence based medicine: what it is and what it isn’t. BMJ 1996;312: 71–72. [2] Younossi Z, Guyatt G. Evidence-based medicine: a method for solving clinical problems in hepatology. Hepatology 1999;30: 829–832. [3] Canty SJ, Shepard GJ, Ryan WG, Banks AJ. Do we practice evidence based medicine with regard to drain usage in knee arthroplasty? Results of a questionnaire of BASK members. Knee 2003;10: 385–387. [4] Carne CA, Foley E, Rowen D, Kell P, Maw R. Variation in clinical practice in genitourinary medicine clinics in the United Kingdom. Sex Transm Infect 2003;79:240–242. [5] Cundari AD, Ker N. Inclusion of evidence-based medicine in colleges of osteopathic medicine and suggestions for implementing evidencebased medicine into osteopathic medical school curricula. J Am Osteopath Assoc 2003;103:557–564. [6] Cabana MD, Rand CS, Powe NR, Wu AW, Wilson MH, Abboud PA, Rubin HR. Why don’t physicians follow clinical practice guidelines? A framework for improvement JAMA 1999;282:1458–1465. [7] Imperiale F. Meta-analysis: when and how. Hepatology 1999;29: 26S–31S. [8] Kjaergard LL, Villumsen J, Gluud C. Reported methodologic quality and discrepancies between large and small randomized trials in metaanalyses. Ann Intern Med 2001;135:982–989. [9] Kjaergard LL, Frederiksen SL, Gluud C. Validity of randomized clinical trials in Gastroenterology from 1964 to 2000. Gastroenterology 2002;122:1157–1160.
[10] Kjaergard LL, Gluud C. Funding, disease area, and internal validity of hepatobiliary randomized clinical trials. Am J Gastroenterol 2002;97: 2708–2713. [11] Gluud C, Cochrane Hepato-Biliary Group. In: Bircher J, Benhamou J P, McIntyre N, Rizzetto M, Rodes J, editors. Oxford textbook of clinical hepatology. Oxford: Oxford Medical Publications; 1999. p. 2145–2148. [12] www.cochrane.org/index0.htm. [13] Gluud C, Als-Nielsen B, D’Amico G, Gluud LL, Khan S, Klingenberg SL, et al. Hepato-Biliary Group. About The Cochrane Collaboration (Collaborative Review Groups (CRGs)) 2005, Issue 2. Art. No.: LIVER. [14] D’Amico G, Pietrosi G, Tarantino I, Pagliaro L. Emergency sclerotherapy versus medical interventions for bleeding oesophageal varices in cirrhotic patients. The Cochrane Library 2002, Issue 1. Oxford: Update Software 2002. [15] Jefferson T, Demicheli V, Deeks J, MacMillan A, Sassi F, Pratt M. Vaccines for preventing hepatitis B in health-care workers. The Cochrane Library 1997, Issue 3 1997. Oxford: Update Software. [16] Chen W, Deeks JJ, Gluud C. Vaccines for preventing hepatitis B in health care workers. The Cochrane Library 1997, Issue 3. Oxford: Update Software. [17] Gluud C, Christensen E. Ursodeoxycholic acid for primary biliary cirrhosis. The Cochrane Library 2002, Issue 1. Oxford: Update Software. [18] Gøtzsche PC. Somatostatin analogues for acute bleeding oesophageal varices. The Cochrane Library 1997, Issue 3. Oxford: Update Software. [19] Khan S, Tudur SC, Williamson P, Sutton R. Shunts versus endoscopic therapy for long-term management of variceal haemorrhage. The Cochrane Library 1997, Issue 4. Oxford: Update Software. [20] Galandi D, Antes G. Radiofrequency thermal ablation versus other interventions for hepatocellular carcinoma. The Cochrane Library 2002, Issue 3. Oxford: Update Software. [21] Chan ES, Chow PK, Tai BC, Machin D, Soo KC. Neoadjuvant and adjuvant therapy for operable hepatocelluar carcinoma. The Cochrane Library 1999, Issue 2. Oxford: Update Software. [22] Brok J, Buckley N, Gluud C. Interventions for paracetamol (acetaminophen) overdoses. The Cochrane Library 2003, Issue 3. Oxford: Update Software. [23] Als-Nielsen B, Koretz RL, Kjaergard LL, Gluud C. Branched-chain amino acids for hepatic encephalopathy. The Cochrane Library 2003, Issue 2. Oxford: Update Software. [24] Chen WD, Liu JP, Gluud C. Bile acids for viral hepatitis. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [25] Als-Nielsen B, Kjaergard LL, Gluud C. Benzodiazepine antagonists for acute and chronic hepatic encephalopathy. The Cochrane Library 2001, Issue 4. Oxford: Update Software. [26] Ioannou G, Doust J, Rockey DC. Terlipressin for acute esophageal variceal hemorrhage. In: The Cochrane Library 2003, Issue 1. Oxford: Update Software. [27] Wun YT, Dickinson JA. Alpha-fetoprotein and/or liver ultrasonography for liver cancer screening in patients with chronic hepatitis B. The Cochrane Library 2003, Issue 2. Oxford: Update Software. [28] Soares-Weiser K, Brezis M, Leibovici L. Antibiotics for spontaneous bacterial peritonitis in cirrhotics. The Cochrane Library 2001, Issue 3. Oxford: Update Software. [29] Rambaldi A, Iaqinto G, Gluud C. Anabolic-androgenic steroids for alcoholic liver disease. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [30] Rambaldi A, Gluud C. Propylthiouracil for alcoholic liver disease. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [31] Rambaldi A, Gluud C. Colchicine for alcoholic and non-alcoholic liver fibrosis and cirrhosis. The Cochrane Library 2001, Issue 3. Oxford: Update Software.
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989 [32] Rambaldi A, Gluud C. S-adenosyl-L-methionine for alcoholic liver diseases. The Cochrane Library 2001, Issue 4. Oxford: Update Software. [33] Myers RP, Poynard T. Interferon for interferon nonresponding and relapsing patients with chronic hepatitis C. The Cochrane Library 2000, Issue 4. Oxford: Update Software. [34] Myers RP, Regimbeau C, Thevenot T, Leroy V, Mathurin P, Opolon P, et al. Interferon for interferon naive patients with chronic hepatitis C. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [35] Myers RP, Regimbeau C, Thevenot T, Leroy V, Mathurin P, Opolon P, et al. Interferon for acute hepatitis C. The Cochrane Library 2002, Issue 1. Oxford: Update Software. [36] Mellerup MT, Krogsgaard K, Mathurin P, Gluud C, Poynard T. Sequential combination of glucocorticosteroids and alfa interferon versus alfa interferon alone for HBeAg-positive chronic hepatitis B. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [37] Liu JP, Manheimer E, Tsutani K, Gluud C. Medicinal herbs for hepatitis C virus infection. The Cochrane Library 2001, Issue 4. Oxford: Update Software. [38] Liu JP, McIntosh H, Lin H. Chinese medicinal herbs for chronic hepatitis B. The Cochrane Library 2000, Issue 4. Oxford: Update Software. [39] Liu JP, McIntosh H, Lin H. Chinese medicinal herbs for asymptomatic carriers of hepatitis B virus infection. The Cochrane Library 2001, Issue 2. Oxford: Update Software. [40] Kjaergard LL, Krogsgaard K, Gluud C. Ribavirin plus interferon versus interferon for chronic hepatitis C. The Cochrane Library 2002, Issue 1. Oxford: Update Software. [41] Liu JP, Kjaergard LL, Als-Nielsen B, Gluud C. Artificial and bioartificial support systems for liver failure. The Cochrane Library 2002, Issue 2. Oxford: Update Software.
989
[42] Soares-Weiser K, Brezis M, Tur-Kaspa R, Leibovici L. Antibiotic prophylaxis for cirrhotic patients with gastrointestinal bleeding. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [43] Long JS. Ordered logit and ordered probit analysis. In: Scott Long J, editor. Regression models for categorical and limited dependent variables. London: Sage Publications; 1997. [44] Woolridge JM. Sample selection, attrition, and stratified sampling. Econometric analysis of cross section and panel data. Cambridge, London: The MIT Press; 2002. [45] Gross CR, Cruz-Correa MCMI, McNeil-Solis C, Valente TW, Powe NR. The adoption of ablation therapy for Barrett’s esophagus: a cohort study of gastroenterologists. Am J Gastroenterol 2004;97: 279–286. [46] Ronduit J, Jornod P, Dorta N, Blum AL, Dorta G. Antibiotic prophylaxis of infective endocarditis during digestive endoscopy: over- and underuse in Switzerland despite professed adherence to guidelines. Endoscopy 2002;34:322–324. [47] Ornstein S, Jenkins RG, Nietert PJ, Feifer C, Roylance LF, Corley S, Dickerson L, Bradford WD, Litvin C. A multimethod quality improvement intervention to improve preventive cardiovascular care: a cluster randomized trial. Ann Intern Med 2004;141:523–532. [48] Vikman S, Airaksinen KE, Tierala I, Peuhkurinen K, MajamaaVoltti M, Tuunanen H, et al. Improved adherence to practice guidelines yields better outcome in high-risk patients with acute coronary syndrome without ST elevation: findings from nationwide FINACS studies. J Intern Med 2004;256:316–323. [49] Olivieri R, Gluud C, Wille-Jorgensen P. Hospital doctors’ self-rated skills in and use of evidence-based medicine - a questionnaire survey. J Eval Clin Pract 2004;10:219–226. [50] Lomas J, Anderson GM, Dominick-Pierre K, Vayda E, Enkin MW, Hannah WJ. Do practice guidelines guide practice? The effect of a consensus statement on the practice of physicians. N Engl J Med 1989;321:1306–1311.
Agreement between reported use of interventions for liver diseases and research evidence in Cochrane systematic reviews* Pia Ku¨rstein1,2, Lise L. Gluud2, Marlene Willemann1, Kim R. Olsen1, Jakob Kjellberg1, Jes Sogaard1, Christian Gluud2,* 2
1 DSI Danish Institute for Health Services Research, Copenhagen, Denmark The Cochrane Hepato-Biliary Group, Copenhagen Trial Unit, Centre for Clinical Intervention Research, Copenhagen, Denmark
Background/Aims: This study evaluates the agreement between reported use of interventions for patients with liver diseases and research evidence in Cochrane systematic reviews. Methods: In July 2002, the Cochrane Hepato-Biliary Group had completed 28 systematic reviews on 36 interventions that were available in Denmark. Based on the reviews, three interventions (n-acetylcysteine for paracetamol overdose, terlipressin for bleeding oesophageal varices, and antibiotics for patients with cirrhosis and gastrointestinal bleeding) with significant beneficial effects on clinical outcomes were classified as ‘evidence-based’, whereas 19 were classified as ‘possibly evidence-based’, and 14 as ‘not evidence-based’. Questionnaires on reported use and perceived intervention effects were mailed to 108 physicians practising in Danish hospitals. Sixty-six returned their questionnaire. Results: The proportion of physicians who reported that they never used the three evidence-based interventions varied considerably (2, 62, and 57%, respectively). The perceived intervention effect, duration of clinical experience, employment as head of department, and university hospital employment were significant predictors of more frequent use of evidence-based interventions. Physicians also reported that they used the interventions that were not evidencebased more often if they were employed at a university hospital. Conclusions: Considerable disagreements between reported use and research evidence were identified. Additional research on methods to introduce evidence-based medicine in practice seems warranted. q 2005 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. Keywords: Evidence-based medicine; Systematic reviews; Randomised trials; Meta-analyses; Clinical practice
1. Introduction Evidence-based medicine combines clinical experience with the most valid research evidence [1,2]. Although evidence-based medicine has an increasing number of
Received 20 January 2005; received in revised form 15 May 2005; accepted 6 June 2005; available online 12 July 2005 * All authors declare that they have no competing interests with regard to the submitted work. * Corresponding author. Tel.: C45 3545 7175; fax: C45 3545 7101. E-mail address: [email protected] (C. Gluud).
supporters, several disagreements seem to exist between clinical practice and research evidence [3–5]. The disagreements may be related to the clinical setting or lack of familiarity with the evidence [6]. Identification of disagreements between clinical practice and research evidence may help target initiatives to promote evidence-based clinical practice. Generally, systematic reviews of high quality randomised clinical trials provide the most valid evidence for health care interventions [7– 10]. The present study evaluates the agreement between reported use and research evidence in systematic reviews of interventions for patients with liver diseases.
0168-8278/$30.00 q 2005 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhep.2005.06.017
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
2. Methods 2.1. Participants In 2002, six university departments and 61 non-university departments in Denmark were responsible for treatment of patients with liver diseases. We selected all university departments and randomly selected 19 of the non-university departments by drawing of lots from an opaque container. Descriptions of the study and consent forms were sent to the administrative head of the selected departments. Twenty-four departments agreed to participate. One university department declined participation. Through correspondence with the selected departments, we listed all physicians who were employed at the participating departments. The physicians were listed and classified as specialists or non-specialists based on their title and speciality. Specialists were head of department and gastroenterologists or hepatologists or had the overall responsibility for treatment of patients with liver disease. Non-specialists were senior registrars or consultants who were not specialists in gastroenterology or hepatology. We aimed at including three specialists and three non-specialists from each department. Some departments employed fewer than three specialists or non-specialists. Other departments employed more than six potentially eligible participants. For those departments, we randomly selected three specialists and three non-specialists by drawing lots from an opaque container. In total, 54 specialists and 54 non-specialists from the 24 included departments were invited to participate.
2.2. Questionnaire The Cochrane Library has been available free of charge for all Danish physicians since 1999 [11–13]. The Cochrane Hepato-Biliary Group had completed 28 systematic reviews in July 2002 [14–42]. Some of the reviews described more than one intervention. In total, the reviews evaluated 37 interventions for prevention or treatment of 16 liver diseases. Thirty-six of these interventions were available in Denmark. A questionnaire on reported use and perceived effects of these interventions was developed. For each disease, the interventions were listed in random order. All questions were closed. Physicians were asked whether they used the interventions always, often, sometimes, rarely, or never for patients who fulfilled the indication and had no contraindications. Physicians were also asked whether they believed that the intervention affects mortality or morbidity (perceived intervention effects).
2.3. Statistical analysis Lists of the most important clinical and surrogate outcome measures were prepared for each of the evaluated liver diseases. Evidence of intervention effects, point estimates, and 95% confidence intervals (CI) for the listed outcome measures was extracted from the systematic reviews. Effects were classified as significant if the CI did not overlap one. None of the evaluated interventions had significant harmful effects. Interventions with significant beneficial effects on clinical outcomes were classified as ‘evidence-based’. Interventions with significant effects on surrogate outcomes or less beneficial effects than other interventions were classified as ‘possibly evidence-based’. Interventions for which there was no evidence of significant effects on clinical or surrogate outcomes were classified as ‘not evidence-based’. Physicians who responded were included in our analyses. Proportions were compared using chi-square and are presented as means with 95% CI. Regression analyses were performed to identify potential predictors of reported use of evidence-based interventions and interventions that are not evidence-based. The dependent variable (reported use) was entered as an ordinal scale [43]. Results are presented as regression coefficients (b) with P values. A negative b indicate more frequent use of evidence-based interventions. A positive b indicate more frequent use of interventions that were not evidence-based. The regression models were adjusted for the type of intervention. The potential predictors included the physicians’ perceived intervention effects, type of hospital (employment at university hospital), duration (years) of clinical experience, position (employment as head of department), clinical speciality (hepatology, gastroenterology, or other), and geographical region (location of hospital). Responses from physicians at university departments were weighted with one third to adjust for overrepresentation [44]. Analyses were performed in Stata version 8.0 for Windows (Stata Corp., Texas, USA) and SPSS version 10 software (SPSS Inc., Chicago, IL).
3. Results Questionnaires were sent to 108 physicians. Sixty-six (61%) physicians agreed to participate and returned their questionnaires. Forty-six of the participating physicians (70%) were employed at a university hospital and 34 (52%) were specialists. The mean duration of clinical experience was 19 years (SD 8 years).
Table 1 Interventions classified as ’possibly evidence based’ Disease
Intervention
Hepatitis B
Yeast derived hepatitis B vaccine [15] Glucocorticosteroids and interferon [36] Ursodeoxycholic acid [24] Chinese medical herbs [38,39] Interferon [35] Ursodeoxycholic acid [17] Interferon [34] Interferon plus ribavirin [40] Active charcoal [22] Ventricular lavage [22] Ipecacuanha [22] Ursodeoxycholic acid [17] Antibiotics [28] Sclerotherapy [14] Somatostatin [18] Surgical anastomosis [19] Transjugular intrahepatic portosystemic shunt [19] Liver support systems [41] Branched chain amino acids, intravenous [23]
Hepatitis C, acute Hepatitis C, chronic
Paracetamol overdose
Primary biliary cirrhosis Spontaneous bacterial peritonitis Bleeding oesophageal varices
Liver failure Hepatic encephalopathy
985
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
986
Table 2 Interventions classified as ’not evidence-based’ Disease
Intervention
Hepatitis B
Alpha-fetoprotein screening [27] Ultrasonography [27] Alpha-fetoprotein screening plus ultrasonography [27] Chinese medical herbs [37] Chinese medical herbs [37] Haemoperfusion [22] Colchicine [31] Anabolic-androgenic steroids [29] Propylthiouracil [30] S-adenosyl-L-methionine [32] Benzodiazepine receptor antagonists [25]
Hepatitis C, acute Hepatitis C, chronic Paracetamol overdose Cirrhosis Alcoholic cirrhosis
Hepatic encephalopathy Hepatocellular carcinoma
More frequent reported use of evidence-based interventions was significantly associated with university hospital employment (bZK0.97, PZ0.03), position as head of department (bZK0.99, P!0.001), duration of clinical experience (bZK0.20; PZ0.02), and perceived intervention effects (bZK0.61, P!0.001). The perceived intervention effect was a significant predictor of less frequent reported use of interventions that were not evidence-based (bZK0.45, P!0.001). Physicians at university hospital departments reported that they used the evidence-based interventions more often that the physicians at nonuniversity departments did (bZ1.57, P!0.001). None of the remaining variables were significantly associated with reported use.
Branched-chained amino acids, peroral [23] Neoadjuvant and adjuvant therapy [21] Radiofrequency thermal ablation [20]
Based on the Cochrane systematic reviews, three interventions were classified as ‘evidence-based’, 19 interventions as ‘possibly evidence-based’ (Table 1), and 14 interventions as ‘not evidence-based’ (Table 2). The ‘evidence-based’ interventions were n-acetylcysteine for paracetamol overdose [22], terlipressin for bleeding oesophageal varices [26], and antibiotic prophylaxis for patients with cirrhosis and bleeding oesophageal varices [42]. The proportion of physicians who reported that they never used the evidence-based interventions was 2, 62, and 57%, respectively (Table 3). The mean proportion of physicians who reported that they never used the interventions was 33% (95% CI: 22–46%) for ‘evidence-based’ interventions compared with 62% (95% CI: 49–74%) for interventions that were ‘possibly evidencebased’, and 80% (95% CI: 69–89%) for interventions that were ‘not evidence-based’ (P!0.01) (Table 4). The proportion of physicians who believed that the evidence-based interventions did not reduce mortality was 1, 54, and 71%, respectively. On average, 31% (95% CI: 20– 43%) of the physicians believed that the evidence-based interventions did not reduce mortality compared with 59% (95% CI: 46–71%) for the interventions that were possibly evidence-based and 87% (95% CI: 76–94%) for interventions that were not evidence-based (P!0.01).
4. Discussion The present study identified three interventions with significant beneficial effects on clinically relevant outcome measures in Cochrane systematic reviews of randomised clinical trials. The interventions were n-acetylcysteine for paracetamol overdose, terlipressin for bleeding oesophageal varices, and antibiotic profylaxis for patients with cirrhosis and gastrointestinal bleeding [22,26,42]. The evidence presented in the systematic reviews suggests that these three interventions should be considered as the treatment of choice for patients who fulfil the indication criteria and none of the contraindication criteria. Most of the physicians reported that they used n-acetylcysteine, but several reported that they never used terlipressin or antibiotic profylaxis. Physicians were more likely to report that they never used the interventions that were not supported by research evidence than the evidence-based interventions. Still, there were exceptions. Some physicians reported that they always or often used the interventions that were not evidence-based. The present study cannot identify the underlying reasons for the disagreements between reported use and research evidence. Additional research on methods to increase implementation of Cochrane systematic reviews in clinical practice seems warranted. Our results suggest that valid research evidence does not seem to be enough to implement interventions. Previous questionnaire studies have found that barriers to physician guideline adherence
Table 3 Reported use of ’evidence-based’ interventions
N-acetylcysteine for paracetamol overdose [22] Terlipressin for bleeding oesophageal varices [26] Antibiotic prophylaxis for patients with cirrhosis and gastrointestinal bleeding [42] a
Presented data are proportions (nZ66 physicians)
Reported frequency of usea Always
Often
Sometimes
Rarely
Never
88 17 3
6 13 10
4 7 13
0 0 16
2 62 57
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
987
Table 4 Reported use of interventions classified as ‘evidence-based’, ‘possibly evidence-based’, or ‘not evidence-based’ Reported usea Always Often Sometimes Rarely Never a
Interventions ’Evidence-based’
’Possibly evidence-based’
’Not evidence-based’
39 (28-52) 9 (6-24) 5 (3-17) 4 (2-15) 33 (22-46)
7 (4-21) 6 (3-19) 5 (3-17) 7 (4-21) 62 (49-74)
2 (0.1-8) 6 (2-15) 6 (2-15) 6 (2-15) 80 (69-89)
Presented data are mean proportions with 95% CI (nZ66 physicians).
include lack of awareness, familiarity, or agreement, selfefficacy, outcome expectancy, inertia of previous practice, and external barriers [6]. Similar barriers are likely to exist for implementation of systematic reviews on liver diseases in clinical practice. The present study found that Danish physicians were more likely to use an intervention if they believed that it had a beneficial effect. A national questionnaire survey of American gastroenterologists [45] asked how often they used ablation therapy for Barrett’s oesophagus. Few found that ablation is effective, but a significant proportion had adopted its use. The strongest predictor of ablation use was the belief that colleagues used it. In a similar study, 195 Swiss gastroenterologists were asked how often they used antibiotic prophylaxis for digestive endoscopy [46]. When compared to the local guidelines, the physicians seemed to use antibiotic prophylaxis too often. The authors argued that the overuse was partly explained by the weak strength of the scientific evidence in the guidelines [46]. In the present study, we found that antibiotic prophylaxis for patients with cirrhosis and gastrointestinal bleeding may be under-used although the scientific evidence is strong [42]. Terlipressin for bleeding oesophageal varices may also be used too infrequently [26]. Several factors may affect the extent to which the results of randomised trials and systematic reviews are used in clinical practice. Both the internal and the external validity of randomised trials are important. Sometimes, the internal validity is questionable and the results are likely to be seriously flawed. In that case, the evidence cannot be used in clinical practice and the assessment of external validity becomes irrelevant. On the other hand, even if a trial has high internal validity, clinicians may not be able to use the trial results because the evidence cannot be extrapolated to their patients. This situation often occurs when the inclusion and exclusion criteria in the trial are too restrictive. Understanding the predicaments of a patient is a key factor that must be considered in the choice of the best treatment. In our questionnaire survey, we specified that our questions concerned typical patients that fulfilled the indication and had no contraindications. Still, in clinical practice, there may be special circumstances, which affect the choice of treatment. There may be some discrepancies between reported use and observed clinical practice, but overall, discrepancies are
likely to be small or moderate. The present study suggests physicians seemed to use the evidence-based interventions more often if they were head of department, employed at a university hospital, or had a long clinical experience. Physicians at university departments also used the interventions that were not evidence-based more than physicians at non-university departments did. Initiatives for implementation of evidence-based clinical practice may need to consider these aspects. It is possible that younger physicians need more definitive guidelines than the more experienced physicians. In general, Cochrane systematic reviews present the research evidence as objectively as possible. Specific recommendations for clinical practice are often not included in the review conclusions because of insufficient or inconsistent evidence regarding the optimal dose or duration of treatment. A certain amount of baseline knowledge is required to implement interventions based on Cochrane reviews alone. Evidence-based guidelines may be useful. Guidelines may provide the additional information that the younger physicians cannot extract from Cochrane reviews. Often, guidelines are prepared and used locally. Regional, national, or even international guidelines may help decrease some of the discrepancies in clinical practice found in the present study. In a review of questionnaire studies on barriers for adherence to clinical practice guidelines, the median response rate for reporting barriers was 54.5% (range 26– 95%) [6]. The risk of bias increases with the proportion of non-responders because non-responders may be systematically different from responders. In the present study the response rate was 61%. Physicians were more likely to respond if they were employed at a university department. Responders also tended to have more clinical experience than nonresponders. It is likely that physicians with a positive attitude to evidence-based medicine were more likely to respond. Our survey may therefore underestimate the gap between research evidence and clinical practice. Evidence-based health care incorporates knowledge from clinical practice and research. Opponents may argue that reliance on research evidence leads to ‘cook-book’ medicine that overlooks the importance of tacit clinical knowledge. These arguments may explain lack of adherence to evidence-based guidelines [6]. Other potential barriers include both internal and external barriers such as physician knowledge or logistic considerations. Use of electronic
988
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989
medical records and regular performance reports improved adherence to clinical practice guidelines for cardiovascular disease and stroke prevention [47]. Educational interventions were found to improve adherence to guidelines and yielded a better outcome for patients with a high-risk of acute coronary syndrome [48]. Education may also increase agreement between clinical practice and the evidence presented in Cochrane reviews. Few Danish physicians use The Cochrane Library regularly although it is available through free Internet access [49]. Internet access relies on passive dissemination of the results of the systematic reviews. Possibly, active rather than passive dissemination may increase the influence of Cochrane reviews on clinical practice [50]. Acknowledgements Supported by grants from The Danish Medical Research Council, The 1991 Pharmacy Foundation, DSI Danish Institute for Health Services Research, and Copenhagen Trial Unit, Centre for Clinical Intervention Research, Denmark. We thank the physicians who participated in our survey and Thorkild Thorsen and Erik Christensen for valuable comments on the design of our survey. We also thank the authors and peer reviewers who were involved in the development and interpretation of the included systematic reviews and the editors and the staff at the Editorial Team Office of The Cochrane-Hepato-Biliary Group.
References [1] Sackett D, Rosenberg WM, Gray JA, Haynes RB, Richardson WS. Evidence based medicine: what it is and what it isn’t. BMJ 1996;312: 71–72. [2] Younossi Z, Guyatt G. Evidence-based medicine: a method for solving clinical problems in hepatology. Hepatology 1999;30: 829–832. [3] Canty SJ, Shepard GJ, Ryan WG, Banks AJ. Do we practice evidence based medicine with regard to drain usage in knee arthroplasty? Results of a questionnaire of BASK members. Knee 2003;10: 385–387. [4] Carne CA, Foley E, Rowen D, Kell P, Maw R. Variation in clinical practice in genitourinary medicine clinics in the United Kingdom. Sex Transm Infect 2003;79:240–242. [5] Cundari AD, Ker N. Inclusion of evidence-based medicine in colleges of osteopathic medicine and suggestions for implementing evidencebased medicine into osteopathic medical school curricula. J Am Osteopath Assoc 2003;103:557–564. [6] Cabana MD, Rand CS, Powe NR, Wu AW, Wilson MH, Abboud PA, Rubin HR. Why don’t physicians follow clinical practice guidelines? A framework for improvement JAMA 1999;282:1458–1465. [7] Imperiale F. Meta-analysis: when and how. Hepatology 1999;29: 26S–31S. [8] Kjaergard LL, Villumsen J, Gluud C. Reported methodologic quality and discrepancies between large and small randomized trials in metaanalyses. Ann Intern Med 2001;135:982–989. [9] Kjaergard LL, Frederiksen SL, Gluud C. Validity of randomized clinical trials in Gastroenterology from 1964 to 2000. Gastroenterology 2002;122:1157–1160.
[10] Kjaergard LL, Gluud C. Funding, disease area, and internal validity of hepatobiliary randomized clinical trials. Am J Gastroenterol 2002;97: 2708–2713. [11] Gluud C, Cochrane Hepato-Biliary Group. In: Bircher J, Benhamou J P, McIntyre N, Rizzetto M, Rodes J, editors. Oxford textbook of clinical hepatology. Oxford: Oxford Medical Publications; 1999. p. 2145–2148. [12] www.cochrane.org/index0.htm. [13] Gluud C, Als-Nielsen B, D’Amico G, Gluud LL, Khan S, Klingenberg SL, et al. Hepato-Biliary Group. About The Cochrane Collaboration (Collaborative Review Groups (CRGs)) 2005, Issue 2. Art. No.: LIVER. [14] D’Amico G, Pietrosi G, Tarantino I, Pagliaro L. Emergency sclerotherapy versus medical interventions for bleeding oesophageal varices in cirrhotic patients. The Cochrane Library 2002, Issue 1. Oxford: Update Software 2002. [15] Jefferson T, Demicheli V, Deeks J, MacMillan A, Sassi F, Pratt M. Vaccines for preventing hepatitis B in health-care workers. The Cochrane Library 1997, Issue 3 1997. Oxford: Update Software. [16] Chen W, Deeks JJ, Gluud C. Vaccines for preventing hepatitis B in health care workers. The Cochrane Library 1997, Issue 3. Oxford: Update Software. [17] Gluud C, Christensen E. Ursodeoxycholic acid for primary biliary cirrhosis. The Cochrane Library 2002, Issue 1. Oxford: Update Software. [18] Gøtzsche PC. Somatostatin analogues for acute bleeding oesophageal varices. The Cochrane Library 1997, Issue 3. Oxford: Update Software. [19] Khan S, Tudur SC, Williamson P, Sutton R. Shunts versus endoscopic therapy for long-term management of variceal haemorrhage. The Cochrane Library 1997, Issue 4. Oxford: Update Software. [20] Galandi D, Antes G. Radiofrequency thermal ablation versus other interventions for hepatocellular carcinoma. The Cochrane Library 2002, Issue 3. Oxford: Update Software. [21] Chan ES, Chow PK, Tai BC, Machin D, Soo KC. Neoadjuvant and adjuvant therapy for operable hepatocelluar carcinoma. The Cochrane Library 1999, Issue 2. Oxford: Update Software. [22] Brok J, Buckley N, Gluud C. Interventions for paracetamol (acetaminophen) overdoses. The Cochrane Library 2003, Issue 3. Oxford: Update Software. [23] Als-Nielsen B, Koretz RL, Kjaergard LL, Gluud C. Branched-chain amino acids for hepatic encephalopathy. The Cochrane Library 2003, Issue 2. Oxford: Update Software. [24] Chen WD, Liu JP, Gluud C. Bile acids for viral hepatitis. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [25] Als-Nielsen B, Kjaergard LL, Gluud C. Benzodiazepine antagonists for acute and chronic hepatic encephalopathy. The Cochrane Library 2001, Issue 4. Oxford: Update Software. [26] Ioannou G, Doust J, Rockey DC. Terlipressin for acute esophageal variceal hemorrhage. In: The Cochrane Library 2003, Issue 1. Oxford: Update Software. [27] Wun YT, Dickinson JA. Alpha-fetoprotein and/or liver ultrasonography for liver cancer screening in patients with chronic hepatitis B. The Cochrane Library 2003, Issue 2. Oxford: Update Software. [28] Soares-Weiser K, Brezis M, Leibovici L. Antibiotics for spontaneous bacterial peritonitis in cirrhotics. The Cochrane Library 2001, Issue 3. Oxford: Update Software. [29] Rambaldi A, Iaqinto G, Gluud C. Anabolic-androgenic steroids for alcoholic liver disease. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [30] Rambaldi A, Gluud C. Propylthiouracil for alcoholic liver disease. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [31] Rambaldi A, Gluud C. Colchicine for alcoholic and non-alcoholic liver fibrosis and cirrhosis. The Cochrane Library 2001, Issue 3. Oxford: Update Software.
P. Ku¨rstein et al. / Journal of Hepatology 43 (2005) 984–989 [32] Rambaldi A, Gluud C. S-adenosyl-L-methionine for alcoholic liver diseases. The Cochrane Library 2001, Issue 4. Oxford: Update Software. [33] Myers RP, Poynard T. Interferon for interferon nonresponding and relapsing patients with chronic hepatitis C. The Cochrane Library 2000, Issue 4. Oxford: Update Software. [34] Myers RP, Regimbeau C, Thevenot T, Leroy V, Mathurin P, Opolon P, et al. Interferon for interferon naive patients with chronic hepatitis C. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [35] Myers RP, Regimbeau C, Thevenot T, Leroy V, Mathurin P, Opolon P, et al. Interferon for acute hepatitis C. The Cochrane Library 2002, Issue 1. Oxford: Update Software. [36] Mellerup MT, Krogsgaard K, Mathurin P, Gluud C, Poynard T. Sequential combination of glucocorticosteroids and alfa interferon versus alfa interferon alone for HBeAg-positive chronic hepatitis B. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [37] Liu JP, Manheimer E, Tsutani K, Gluud C. Medicinal herbs for hepatitis C virus infection. The Cochrane Library 2001, Issue 4. Oxford: Update Software. [38] Liu JP, McIntosh H, Lin H. Chinese medicinal herbs for chronic hepatitis B. The Cochrane Library 2000, Issue 4. Oxford: Update Software. [39] Liu JP, McIntosh H, Lin H. Chinese medicinal herbs for asymptomatic carriers of hepatitis B virus infection. The Cochrane Library 2001, Issue 2. Oxford: Update Software. [40] Kjaergard LL, Krogsgaard K, Gluud C. Ribavirin plus interferon versus interferon for chronic hepatitis C. The Cochrane Library 2002, Issue 1. Oxford: Update Software. [41] Liu JP, Kjaergard LL, Als-Nielsen B, Gluud C. Artificial and bioartificial support systems for liver failure. The Cochrane Library 2002, Issue 2. Oxford: Update Software.
989
[42] Soares-Weiser K, Brezis M, Tur-Kaspa R, Leibovici L. Antibiotic prophylaxis for cirrhotic patients with gastrointestinal bleeding. The Cochrane Library 2002, Issue 2. Oxford: Update Software. [43] Long JS. Ordered logit and ordered probit analysis. In: Scott Long J, editor. Regression models for categorical and limited dependent variables. London: Sage Publications; 1997. [44] Woolridge JM. Sample selection, attrition, and stratified sampling. Econometric analysis of cross section and panel data. Cambridge, London: The MIT Press; 2002. [45] Gross CR, Cruz-Correa MCMI, McNeil-Solis C, Valente TW, Powe NR. The adoption of ablation therapy for Barrett’s esophagus: a cohort study of gastroenterologists. Am J Gastroenterol 2004;97: 279–286. [46] Ronduit J, Jornod P, Dorta N, Blum AL, Dorta G. Antibiotic prophylaxis of infective endocarditis during digestive endoscopy: over- and underuse in Switzerland despite professed adherence to guidelines. Endoscopy 2002;34:322–324. [47] Ornstein S, Jenkins RG, Nietert PJ, Feifer C, Roylance LF, Corley S, Dickerson L, Bradford WD, Litvin C. A multimethod quality improvement intervention to improve preventive cardiovascular care: a cluster randomized trial. Ann Intern Med 2004;141:523–532. [48] Vikman S, Airaksinen KE, Tierala I, Peuhkurinen K, MajamaaVoltti M, Tuunanen H, et al. Improved adherence to practice guidelines yields better outcome in high-risk patients with acute coronary syndrome without ST elevation: findings from nationwide FINACS studies. J Intern Med 2004;256:316–323. [49] Olivieri R, Gluud C, Wille-Jorgensen P. Hospital doctors’ self-rated skills in and use of evidence-based medicine - a questionnaire survey. J Eval Clin Pract 2004;10:219–226. [50] Lomas J, Anderson GM, Dominick-Pierre K, Vayda E, Enkin MW, Hannah WJ. Do practice guidelines guide practice? The effect of a consensus statement on the practice of physicians. N Engl J Med 1989;321:1306–1311.