Figures
Abstract
Background
Digital rectal examination (DRE) has been traditionally recommended to evaluate acute appendicitis, although several reports indicate its lack of utility for this diagnosis. No meta-analysis has examined DRE for diagnosis of acute appendicitis.
Data Sources
Cochrane Library, PubMed, and SCOPUS from the earliest available date of indexing through November 23, 2014, with no language restrictions.
Study Selection
Clinical studies assessing DRE as an index test for diagnosis of acute appendicitis.
Data Extraction and Synthesis
Two independent reviewers extracted study data and assessed the quality, using the Quality Assessment of Diagnostic Accuracy Studies 2 tool. Bivariate random-effects models were used for the pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio (DOR) as point estimates with 95% confidence intervals (CI).
Main Outcomes and Measures
The main outcome measure was the diagnostic performance of DRE for diagnosis of acute appendicitis.
Results
We identified 19 studies with a total of 7511 patients. The pooled sensitivity and specificity were 0.49 (95% CI 0.42–0.56) and 0.61 (95% CI 0.53–0.67), respectively. The positive and negative likelihood ratios were 1.24 (95% CI 0.97–1.58) and 0.85 (95% CI 0.70–1.02), respectively. The DOR was 1.46 (0.95–2.26).
Citation: Takada T, Nishiwaki H, Yamamoto Y, Noguchi Y, Fukuma S, Yamazaki S, et al. (2015) The Role of Digital Rectal Examination for Diagnosis of Acute Appendicitis: A Systematic Review and Meta-Analysis. PLoS ONE 10(9): e0136996. https://doi.org/10.1371/journal.pone.0136996
Editor: Lorna Marson, Centre for Inflammation Research, UNITED KINGDOM
Received: April 5, 2015; Accepted: August 11, 2015; Published: September 2, 2015
Copyright: © 2015 Takada et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The authors received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Acute appendicitis is one of the most common diagnoses associated with acute abdominal pain, with lifetime prevalence of 7% [1]. Its diagnosis can be challenging because the clinical presentation is often atypical and overlaps with other conditions [2]. Delay of accurate diagnosis could result in rupture of the appendix, which is associated with a worse prognosis [3]. Therefore, a prompt and accurate diagnosis of appendicitis is crucial. Careful history taking and physical examination have an important role in the correct diagnosis [4].
In the physical examination for evaluation of appendicitis, digital rectal examination (DRE) has been considered a necessary investigation [5]. This traditional teaching that DRE should be performed routinely in all patients with abdominal pain has been supported in most surgical textbooks [6,7]. On the other hand, several reports found that DRE rarely provides useful information for diagnosis of acute appendicitis [8,9] and it often induces mild to severe discomfort in patients [10,11].
To date, some review articles investigated the role of DRE for diagnosis of acute appendicitis [8,9,12–15], but none of them used the recommended methods for a meta-analysis of diagnostic studies [16]. Accordingly, the aim of the current study was to perform a meta-analysis of studies that report the diagnostic role of DRE for acute appendicitis to determine the accuracy of this traditional examination.
Methods
We followed standard guidelines for a systematic review of diagnostic studies [16].
Data sources and searches
We conducted a literature search of MEDLINE via PubMed, Scopus, and Cochrane Library via Wiley Online Library from the earliest available date of indexing through November 23, 2014. See S1 Text for protocol and searching strategy. The literature search identified potential studies in all languages. We translated the non-English language papers and fully assessed them for potential inclusion in the review as necessary. We checked reference lists of all included studies and relevant systematic review articles for additional references. We contacted authors of additional studies identified that were missed in the original electronic searches.
Study selection
Studies were included if they met the following conditions: (1) DRE was assessed as an index test (the test under investigation) to evaluate the diagnosis of appendicitis; (2) the reference standard (the criterion standard) of appendicitis was defined as histologically proven acute inflammation in the appendix; (3) exclusion of appendicitis was histologically proven or evaluated in the clinical follow-up with careful observation; and (4) absolute numbers of true-positive, false-positive, false-negative, and true-negative results were reported in the articles, or these data were derivable from the published results.
Two authors (TT and NH) examined the titles and abstracts of references identified by the electronic search strategies described above to check whether the study was likely to be relevant. Studies that were considered potentially relevant in the search were obtained as full articles and independently assessed for inclusion by the same two authors. In the case of discordance, resolution was sought by discussion between the two authors. The discordance in the selection of studies was evaluated by quantifying both the percentage of agreement and Cohen’s Kappa (k). Values of kappa between 0.40 and 0.59 have been considered to reflect fair agreement, between 0.60 and 0.74 to reflect good agreement, and 0.75 or more to reflect excellent agreement [17].
Data extraction and quality assessment
Two authors used a structured, pilot-tested, Excel data collection form to independently extract the data from the included studies. Extracted data included study characteristics (design, one/two-gate, number of participants, number of excluded participants, index test, reference test, blinding), patient characteristics (setting, age, sex). Two authors also independently assessed methodological quality using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool [18]. Any disagreement was resolved by discussion between the two authors. When relevant information regarding design or outcomes was unclear, the original study authors were contacted for clarification.
Data synthesis and analysis
To evaluate the diagnostic performance, we constructed 2 × 2 tables. Measures of the diagnostic performance, including sensitivity, specificity, positive likelihood ratio (LR+), negative likelihood ratio (LR-), and diagnostic odds ratios (DORs), were reported as point estimates with 95% CI. A DOR can be calculated as the ratio of the odds of positivity in a disease state relative to the odds of positivity in the nondisease state, with higher values indicating better discriminatory test performance. The value of the DOR ranges from zero to infinity, with higher values indicative of better discriminative performance. A value of 1 indicates that the test does not discriminate between people with and without the disease/condition [19]. Publication bias was assessed using the effective sample size funnel plot and associated regression test of asymmetry [20]. We used the bivariate random-effects model for analysis and pooling of the diagnostic performance measures across studies [21]. The bivariate model was used to estimate pairs of logit transformed sensitivity and specificity from studies, incorporating the correlation that might exist between sensitivity and specificity. We used the hierarchical summary receiver operating characteristic (HSROC) curves to estimate the area under the curve [22]. Between-study statistical heterogeneity was assessed using the I2 test on the basis of the random-effects analysis [23]. To explain heterogeneity, we performed meta-regression to identify potential sources of bias. Pooled estimates were also calculated for subgroups of studies that were defined according to specific study designs. The following variables were selected a priori as potential sources of heterogeneity: publication year, number of participants, study design (prospective vs. retrospective), age of participants (children vs. adults vs. all ages), sex, inclusion criteria (abdominal pain vs. suspected appendicitis/appendectomy), definition of positive DRE (right-sided tenderness vs. the others), reference standard (appendectomy vs. appendectomy/follow-up). The definition of children was adopted in accordance with each article. Two-sided P < .05 was considered statistically significant. Statistical analyses were performed with commercial software programs (STATA, version 13.1 SE; StataCorp LP, College Station, Texas).
Results
Study selection
We identified 12 articles from the Cochrane Library, 130 from PubMed, 429 from SCOPUS, and 23 from hand-searching. Seventy-nine full-text articles were assessed for eligibility, and 19 studies were included in the final analysis [10,24–45]. The literature search process is shown in Fig 1. The percentage of agreement in the selection of studies was 91.8%, and Cohen’s Kappa (k) was 0.79, which is regarded as excellent agreement [17].
Characteristics of included studies
A total of 7511 patients were included in the 19 studies. The characteristics and results of studies are summarized in Table 1 and Fig 2. Nine studies were prospective, and 10 were retrospective. Thirteen studies included all age groups, 4 included only adults, and 2 included only children. Inclusion criteria were patients with appendectomy in 8 studies, suspected appendicitis in 8, and abdominal pain in 3 studies. The definition of “positive” DRE was right-sided tenderness for diagnosis of appendicitis [6]. Eight studies defined right-sided tenderness as positive, while 11 studies had no apparent definition, or right-sided tenderness alone was not regarded as positive. With regard to exclusion of appendicitis, 10 studies used a histologic diagnosis, and 9 studies used careful observation for exclusion of appendicitis.
Quality of included studies
S1 Fig and S2 Fig present an overview of the quality of included studies evaluated by QUADAS-2. In terms of patient selection, 3 studies had inappropriate exclusions. The index test was judged as unclear in 6 studies that did not describe the definition of “positive” DRE. In terms of flow and timing, 9 studies that excluded appendicitis by histology or careful observation were regarded as having a high risk of bias. Most studies did not describe whether the result of histopathology and final diagnosis in follow-up were interpreted without knowledge of results of DRE. In S3 Fig, the funnel plots and regression test indicate no significant publication bias (P = 0.47), although it seemed that more studies reported a low DOR.
Overall diagnostic accuracy
The pooled sensitivity was 0.49 (95% CI 0.42–0.56), the pooled specificity was 0.61 (95% CI 0.53–0.67), the pooled LR+ was 1.24 (95% CI 0.97–1.58), the pooled LR- was 0.85 (95% CI 0.70–1.02), and the DOR was 1.46 (0.95–2.26). In Fig 3, sensitivity and specificity of DRE shows great variation. The HSROC curve is close to the line of discrimination.
Investigation of heterogeneity
Fig 2 shows that the heterogeneity among studies was large, with I2 for sensitivity and specificity of 94% and 93%, respectively. The result of subgroup analyses is shown in Table 2. No significant differences were observed for the heterogeneity of sensitivity. On the other hand, whether the reference standard was only appendectomy or included follow-up with careful observation had a significant effect on heterogeneity in specificity. The subgroup analyses for age and sex of participants could not be performed because sufficient data for these analyses were not reported.
Discussion
Overall, our analysis showed poor diagnostic performance of DRE for acute appendicitis. Fig 4 shows the relationship between pre- and post-test probability based on the pooled LR+ and LR-. The change between pre- and post-test probability is minimal regardless of the results of DRE. These findings are consistent with the report that DRE results did not change physicians’ decisions regarding a diagnosis of appendicitis [46–48]. Considering the discomfort induced by DRE [10,11], the opinion that DRE is an indispensable, routine examination for evaluation of appendicitis should be reconsidered. On the other hand, the role of DRE should not be denied entirely. In Schwartz’s Principles of Surgery, it has been described as “When the appendix hangs into the pelvis, abdominal findings may be entirely absent, and the diagnosis may be missed. Right-sided rectal tenderness is said to help in this situation [49].” Thus, the DRE could have an important role when the inflamed appendix cannot be evaluated through abdominal examination. In our review, no study investigated such a specific situation. Instead, a report described the relationship between the sensitivity of DRE and the positions of inflamed appendixes [27]. The sensitivity for pelvic appendicitis was 0.38, while the sensitivity for all appendicitis was 0.44. The specificity was unknown. These data are not enough to conclude the role of DRE for pelvic appendicitis and further investigation is warranted.
Our analysis had several limitations. First, in an investigation of heterogeneity, the heterogeneity of sensitivity could not be explained by subgroup analyses. Many factors possibly contributing to heterogeneity could not be evaluated because they were not reported, such as the technique and experience of the physician and the position of the appendix. Second, we could not perform subgroup analyses defined a priori for age and sex because sufficient data for analyses were not obtained. Almost half of the retrieved studies were performed before the 1990s, and most authors contacted for further information did not have their data. It is known that the diagnosis of appendicitis is challenging in children younger than age 3 and elderly patients [50,51]. In women, there are a variety of other diagnoses, mostly related to gynecologic diseases, that mimic features of appendicitis. Therefore, the specificity of DRE could be lower. Thus, the diagnostic performance of DRE for these groups should be assessed separately. S1 Table shows some data for these groups, although there is not enough information to perform a meta-analysis. Third, it is preferable that all participants receive the same reference standard in diagnostic research. Partial verification bias may exist when a nonrandom set of patients does not undergo the reference standard. It usually leads to overestimation of sensitivity, and the effect on specificity varies [15]. However, it is ethically difficult to perform surgery on all participants, especially when the possibility of appendicitis is considered low. Therefore, we included studies that excluded appendicitis by clinical follow-up. In our review, 9 studies excluded appendicitis by appendectomy and clinical follow-up. Patients who do not undergo surgery and who do not develop appendicitis during follow-up are assumed not to have had appendicitis, but several studies suggest that spontaneous resolution is common [52,53]. Therefore, this might also lead to a high risk of bias. The sensitivity and specificity were 0.50 and 0.68 in the appendectomy/follow-up group, whereas they were 0.47 and 0.52, respectively, in the appendectomy group alone.
The strength of our systematic review is that this is the first systematic review and meta-analysis following a standard protocol and comprehensive search strategy on this topic. There have been several reviews without meta-analyses [8,9,12–15], and a review with a meta-analysis that did not follow a standard protocol [15]. The review with the meta-analysis investigated the value of various history, examinations, and laboratory tests, was not focused on DRE, and included only 5 studies [15].
In conclusion, acute appendicitis cannot be ruled in or out through the results of DRE. Reconsideration is warranted for the traditional teaching that DRE should be performed routinely in all patients with suspected appendicitis. Situations in which the DRE has an important role should be further investigated.
Supporting Information
S1 Fig. Summary of Risk of bias and applicability concerns.
Review authors' judgements about each domain for each included study. The figure was generated using Review Manager Version 5.3.
https://doi.org/10.1371/journal.pone.0136996.s001
(TIF)
S2 Fig. Risk of bias and applicability concerns graph for included studies.
Review authors' judgements about each domain presented as percentages across included studies. The figure was generated using Review Manager Version 5.3.
https://doi.org/10.1371/journal.pone.0136996.s002
(TIF)
S3 Fig. Funnel plot for publication bias assessment.
https://doi.org/10.1371/journal.pone.0136996.s003
(TIF)
S1 Table. Diagnostic performance of DRE for woman and elderly in diagnosis of appendicitis.
https://doi.org/10.1371/journal.pone.0136996.s004
(DOCX)
S1 Text. Study protocol and searching strategy.
https://doi.org/10.1371/journal.pone.0136996.s005
(DOCX)
Author Contributions
Conceived and designed the experiments: TT HN YY YN SNF. Performed the experiments: TT HN YY YN SGF SY. Analyzed the data: TT HN YY YN SGF SY SNF. Contributed reagents/materials/analysis tools: TT HN YY YN SGF SY SNF. Wrote the paper: TT HN YY YN SNF. Study supervision: YN SY SNF.
References
- 1. Irvin TT. Abdominal pain: a surgical audit of 1190 emergency admissions. Br J Surg. 1989;76: 1121–1125. pmid:2597964
- 2. Birnbaum BA, Wilson SR. Appendicitis at the millennium. Radiology. 2000;215: 337–348. pmid:10796905
- 3. Gwynn LK. The diagnosis of acute appendicitis: clinical assessment versus computed tomography evaluation. J Emerg Med. 2001;21: 119–123. pmid:11489398
- 4. Park JS, Jeong JH, Lee JI, Lee JH, Park JK, Moon HJ. Accuracies of diagnostic methods for acute appendicitis. Am Surg. 2013;79: 101–106. pmid:23317620
- 5. Chandrasekaran TV, Johnson N. Acute appendicitis. Surgery. 2014;32: 413–417.
- 6.
Silen W. Cope’s early diagnosis of the acute abdomen. 22th ed. New York: Oxford University Press; 2010.
- 7.
Mary EK, Abdulhameed A, Ankit B, Amy CF, Matthew RP. The Washington Manual of Surgery. 6th ed. Philadelphia: Lippincott Williams & Wilkins; 2011.
- 8. Brewster GS, Herbert ME. Medical myth: a digital rectal examination should be performed on all individuals with possible appendicitis. West J Med. 2000;173: 207–208. pmid:10986194
- 9. Kessler C, Bauer SJ. Utility of the digital rectal examination in the emergency department: a review. J Emerg Med. 2012;43: 1196–1204. pmid:22883714
- 10. Dickson AP, MacKinlay GA. Rectal examination and acute appendicitis. Arch Dis Child. 1985;60: 666–667. pmid:4026364
- 11. Macias DJ, Sarabia MJ, Sklar DP. Male discomfort during the digital rectal examination: Does examiner gender make a difference? Am J Emerg Med. 2000;18: 676–678. pmid:11043620
- 12. Khoshnood A, Ekelund U. Routine rectal palpation in acute abdomen has no proven value. A review of the literature. Lakartidningen. 2014;111: 89–90. pmid:24552011
- 13. Muris JW, Starmans R, Wolfs GG, Pop P, Knottnerus JA. The diagnostic value of rectal examination. Fam Pract. 1993;10: 34–37. pmid:8477890
- 14. Jesudason EC, Walker J. Rectal examination in paediatric surgical practice. Br J Surg. 1999;86: 376–378. pmid:10201782
- 15. Andersson RE. Meta-analysis of the clinical and laboratory diagnosis of appendicitis. Br J Surg. 2004;91: 28–37. pmid:14716790
- 16. Leeflang MM, Deeks JJ, Gatsonis C, Bossuyt PM. Systematic reviews of diagnostic test accuracy. Ann Intern Med. 2008;149: 889–897. pmid:19075208
- 17. Bland JB, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1: 307–310. pmid:2868172
- 18. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included insystematic reviews. BMC Med Res Methodol. 2003;3: 25. pmid:14606960
- 19. Glas AS, Lijmer JG, Prins MH, Bonsel GJ, Bossuyt PM. The diagnostic odds ratio: a single indicator of test performance. J Clin Epidemiol. 2003;56: 1129–1135. pmid:14615004
- 20. Deeks JJ, Macaskill P, Irwig L. The performance of tests of publication bias and other sample size effects in systematic reviews of diagnostic test accuracy was assessed. J Clin Epidemiol. 2005;58: 882–893. pmid:16085191
- 21. Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol. 2005;58: 982–990. pmid:16168343
- 22. Rutter CM, Gatsonis CA. A hierarchical regression approach to metaanalysis of diagnostic test accuracy evaluations. Stat Med. 2001;20: 2865–2884. pmid:11568945
- 23. Thompson SG. Why sources of heterogeneity in meta-analysis should be investigated. BMJ. 1994;309: 1351–1355. pmid:7866085
- 24. Smith PH. The diagnosis of appendicitis. Postgrad Med J. 1965;41: 2–5. pmid:14253734
- 25. Bonello JC, Abrams JS. The significance of a "positive" rectal examination in acute appendicitis. Dis Colon Rectum. 1979;22: 97–101. pmid:428284
- 26. Nase HW, Kovalcik PJ, Cross GH. The diagnosis of appendicitis. Am Surg. 1980;46: 504–507. pmid:7416631
- 27. Chee CP, Somasundaram K. Acute appendicitis: the significance of "positive" rectal examination and the position of the vermiform appendix. Singapore Med J. 1982;23: 265–268. pmid:7157006
- 28. Berry J Jr, Malt RA. Appendicitis near its centenary. Ann Surg. 1984;200: 567–575. pmid:6385879
- 29. Rehák F, Pafko P, Schützner J, Votoček J. Akutní apendicitis–Diagnostický problém. Prakt Lek. 1984;10: s371.
- 30. Alvarado A. A practical score for the early diagnosis of acute appendicitis. Ann Emerg Med. 1986;15: 557–564. pmid:3963537
- 31. Garcia JC. What is new in the diagnosis of appendicitis? Borborygmi in the cecal region. J Chir (Paris). 1989;126: 630–632.
- 32. Dixon JM, Elton RA, Rainey JB, Macleod DA. Rectal examination in patients with pain in the right lower quadrant of the abdomen. BMJ. 1991;302: 386–388. pmid:2004144
- 33. Koudelka J, Preis J, Králová M. Diagnostic value of the rectal examination and the difference in axillo-rectal temperatures in acute appendicitis in childhood. Rozhl Chir. 1991;70: 36–41. pmid:1925781
- 34. Izbicki JR, Knoefel WT, Wilker DK, Mandelkow HK, Müller K, Siebeck M, et al. Accurate diagnosis of acute appendicitis: a retrospective and prospective analysis of 686 patients. Eur J Surg. 1992;158: 227–231. pmid:1352137
- 35. Lee HY, Jayalakshmi P, Noori SH. Acute appendicitis—the University Hospital experience. Med J Malaysia. 1993;48: 17–27. pmid:8341168
- 36. Eskelinen M, Ikonen J, Lipponen P. Clinical diagnosis of acute appendicitis A prospective study of patients with acute abdominal pain. Theor Surg. 1992;7: 81–85.
- 37. Eskelinen M, Ikonen J, Lipponen P. A computer-based diagnostic score to aid in diagnosis of acute appendicitis A prospective study of 1333 patients with acute abdominal pain. Theor Surg. 1992;7: 86–90.
- 38. Eskelinen M, Ikonen J, Lipponen P. Sex-specific diagnostic scores for acute appendicitis. Scand J Gastroenterol. 1994;29: 59–66. pmid:8128177
- 39. Eskelinen M, Ikonen J, Lipponen P. Acute appendicitis in patients over the age of 65 years; comparison of clinical and computer based decision making. Int J Biomed Comput.1994;36: 239–249. pmid:7960209
- 40. Eskelinen M, Ikonen J, Lipponen P. The value of history-taking, physical examination, and computer assistance in the diagnosis of acute appendicitis in patients more than 50 years old. Scand J Gastroenterol. 1995;30: 349–355. pmid:7610351
- 41. Jahn H, Mathiesen FK, Neckelmann K, Hovendal CP, Bellstrøm T, Gottrup F. Comparison of clinical judgment and diagnostic ultrasonography in the diagnosis of acute appendicitis: experience with a score-aided diagnosis. Eur J Surg. 1997;163: 433–443. pmid:9231855
- 42. Andersson RE, Hugander AP, Ghazi SH, et al. Diagnostic value of disease history, clinical presentation, and inflammatory parameters of appendicitis. World J Surg. 1999;23: 133–140. pmid:9880421
- 43. Gil VF, Peinado E, Obrador E, Reyes P, Pérez Barba C, Merino J. Validez de las pruebas diagnósticas para confirmaro descartar una apendicitis aguda. Med Clin (Barc). 2000;114: 48–51.
- 44. Özdogan M, Gürer A, Gökakin AK, Ersoy E, Kulaçoglu H, Aydin R. Rectal examination in the differential diagnosis of acute appendicitis in adults. Turkish Journal of Surgery. 2005;21:175–178.
- 45. Sedlak M, Wagner OJ, Wild B, Papagrigoriades S, Zimmermann H, Exadaktylos AK. Is there still a role for rectal examination in suspected appendicitis in adults? Am J Emerg Med. 2008;26: 359–60. pmid:18358949
- 46. Dunning PG, Goldman MD. The incidence and value of rectal examination in children with suspected appendicitis. Ann R Coll Surg Engl. 1991;73: 233–234. pmid:1863045
- 47. Manimaran N, Galland RB. Significance of routine digital rectal examination in adults presenting with abdominal pain. Ann R Coll Surg Engl. 2004;86: 292–295. pmid:15239875
- 48. Quaas J, Lanigan M, Newman D, McOsker J, Babayev R, Mason C. Utility of the digital rectal examination in the evaluation of undifferentiated abdominal pain. Am J Emerg Med. 2009;27: 1125–1129 pmid:19931762
- 49.
Brunicardi F, Andersen D, Billiar T, Dunn D, Hunter J, Matthews J, et al. Schwartz's Principles of Surgery. 10th ed. New York: McGraw-Hill Professional; 2014.
- 50. Daehlin L. Acute appendicitis during the first three years of life. Acta Chir Scand. 1982;148: 291–294. pmid:7136432
- 51. Horattas MC, Guyton DP, Wu D. A reappraisal of appendicitis in the elderly. Am J Surg. 1990;160: 291–293. pmid:2393058
- 52. Barber MD, McLaren J, Rainey JB. Recurrent appendicitis. Br J Surg. 1997;84: 110–112. pmid:9043472
- 53. Cobben LP, de Van Otterloo AM, Puylaert JB. Spontaneously resolving appendicitis: frequency and natural history in 60 patients. Radiology. 2000;215: 349–352. pmid:10796906