Family Practice Vol. 16, No. 6, 621-626
© Oxford University Press 1999
The diagnostic value of digital rectal examination in primary care screening for prostate cancer: a meta-analysis
The Medion group for meta-analysis of diagnostic studies.
a Department of General Practice, University of Maastricht, The Netherlands, and Leuven, Belgium and
b Department of Epidemiology, University of Maastricht, The Netherlands.
Prof. Dr F Buntinx, University of Maastricht, Department of General Practice, PO Box 616, 6200 MD Maastricht, The Netherlands.
Hoogendam A, Buntinx F and de Vet HCW. The diagnostic value of digital rectal examination in primary care screening for prostate cancer: a meta-analysis. Family Practice 1999; 16: 621–626.
Received 9 October 1998; Revised 7 June 1999; Accepted 22 June 1999.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
Objective. This systematic review examines the diagnostic value of the digital rectal examination (DRE) for the diagnosis of prostate cancer.
Method. Only studies relating to unselected populations and using either biopsy or surgery as the reference standard were included. The methodological quality of the studies was used in an attempt to explain differences between studies.
Results. Fourteen studies were eligible for selection, of which five complied with the predetermined list of ‘good-quality’ requirements. Between study heterogeneity was high, even within the group of high-quality studies, and could not be explained by the registered indicators of methodological quality.
Conclusions. In this setting, the DRE appears to be a test with a high specificity and negative predictive value, but a low sensitivity and positive predictive value. Neither a positive nor a negative test result is sufficient to enable conclusions without further confirmation.
Keywords. Predictive value of tests, primary health care, prostate neoplasms, sensitivity and specificity.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
The usefulness of screening for prostate cancer is still under discussion. Crucial to early detection is the availability of valid screening tests. The oldest and least invasive test is digital rectal examination (DRE). Studies aimed specifically at determining the value of DRE for the detection of prostate cancer are rare. However, the development of new diagnostic tests has enabled the conduction of numerous studies in which DRE is compared with other tests.
We performed a systematic review of the literature in order to summarize the data on the value of DRE for screening purposes in primary care. Recently, methods for evaluating diagnostic tests by meta-analysis have been developed. Meta-analysis can be used not only to summarize overall diagnostic accuracy of different tests, but also to determine whether diagnostic accuracy differs among patient subgroups or with varying study designs.1
Using such techniques, we assessed the accuracy of DRE in primary care and its relation with various methodological characteristics of the studies.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
Searching the literature
To identify studies that provide data on the value of DRE, we performed a MEDLINE search from 1983 to 1995, using MESH terms as well as free text searching. Older studies were not searched because of the technological improvements that have changed the diagnostic capacities in recent years. Additionally, Famli, a specialized database for studies in family practice was searched and some GP journals were searched manually for diagnostic studies. Thereafter, the references of all retrieved studies were checked for relevant citations. No language restrictions were used.
Inclusion criteria
Studies were included if DRE was compared with biopsy or surgery as a reference standard. The study population had to be unselected with respect to prostate-related signs and symptoms. In addition, true positive and true negative rates as well as false negative and false positive rates had to be presented or it had to be possible to calculate them from the published data. For this reason, in some studies only part of the total study population could be used, yielding results different from those reported in the original papers.
Assessment of methodological quality
Methodological aspects of all studies were assessed using a list of criteria proposed by the Cochrane Methods Working Group on Meta-analysis of Diagnostic and Screening Tests. This list is based on the recent literature2–4 and regarded as the most recent consensus on criteria for systematic reviews on diagnostic studies. It includes criteria with respect to internal validity, applicability of the results and description of test procedures, as well as indirect measures to estimate study quality. The complete list is presented in the appendix.
|
Analysis |
|---|
|
TopAbstractIntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
Sensitivity (sens), specificity (spec), positive predictive value (ppv) and negative predictive value (npv) with their 95% confidence interval (95% CI) were extracted from the papers or calculated on the basis of the published data.
We tested for the possibility of different implicit cut-off points between studies by correlating sensitivity and (1–specificity)5 and for heterogeneity by using the chi-square test for homogeneity. Statistical pooling was based on a random effects model,6–8 using FASTPRO version 1.7.9 Meta-analyses were performed including all studies, and for special subgroups separately. The influence on the diagnostic indicators of setting and methodological characteristics of each individual study was studied using multiple linear regression. Each indicator was used as a dependent variable, while setting (each patient received a personal invitation versus general publicity only), quality of the test procedure description (good if the presence of either induration, asymmetry or nodularity or a combination of these was described), prevalence of prostate cancer in the study population and duration of follow-up were used as independent variables.
Results of high-quality studies were analysed separately. For this group, the following inclusion criteria applied.
- DRE was performed on a complete population (no pre-test selection).
- All patients underwent testing with either prostate-specific antigen (PSA) or transrectal ultrasonography (TRUS) or both.
- All patients with a positive DRE, and in cases with a negative DRE result at least those who scored positive on one other test (PSA or TRUS), were eligible for the reference test.
- The reference test was performed on more than 90% of the people eligible for that test.
|
Results |
|---|
|
TopAbstractIntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
Study selection
Forty-nine studies could be identified in which the diagnostic value of DRE was studied. Twenty-one10–30 of them concerned primary care based screening studies. Two papers reported the results of one study.22,30 The last one was the most recent;30 therefore the other22 was not included. In three studies,20,24,28 DRE was only performed in cases of an increased PSA. In three additional studies,15,17,27 only follow-up by a GP was stated as the reference standard, with no presentation of further data about what was done. These six studies were excluded from the analysis. Ultimately, 14 studies fulfilled the inclusion criteria of this meta-analysis, totalling some 22 000 patients.
Characteristics and quality of the studies
Characteristics of the 14 selected studies are presented in Table 1
. All were published after 1980. Most patients were over age 50 years. Prevalence rates of detected cancer ranged from 1.2 to 7.3%. The number of patients lost to follow-up was more than 20% in six studies,16,21,23,26 but very low in the remainder. The percentage of patients eligible for a reference test in which no such test was performed, however, was less than 10% in six studies. The number was not reported in three studies and raised to maximally 39%.21 Only five studies10,13,18,19,25 complied with the criteria for a good-quality study, as mentioned previously. In some studies, patients were invited personally by the physician, in others they were attracted by advertisements on television or in local newspapers. None of these studies presented data on the presence of complaints in the screening population.
|
In most of the studies, it was difficult to determine whether benign prostate hypertrophy was considered a normal or abnormal result.11,13,14,16,18,19,21,25,26,29,30 To be as consequent as possible we considered every non-enlarged, smooth, symmetrical prostate with a normal consistency as normal. If enough data were presented, we recalculated test results according to this principle. However, some studies reported DRE positive or negative, without defining the criteria that were used.11,13,14,21,25,26,30
Moreover, no study mentioned the influence of the experience of the examiner or the reproducibility of DRE on the results of the studies.
Many studies failed to report other quality parameters that are on the Cochrane scoring list for diagnostic studies (see Appendix).
The spearman correlation coefficient of sens and 1–spec was 0.12 (and statistically non-significant), indicating absence of a substantial cut-off-point effect.5 Between-study heterogeneity, however, was highly significant for almost all indicators, even if only high-quality studies were considered.
Linear regression resulted in none of the independent variables showing any significant relation with any of the diagnostic indicators that were studied.
Pooling of the results of the 14 studies, as well as of the results of good-quality studies only, revealed high specificity (0.94) and npv (0.99), low sensitivity (0.59) and very low ppv (0.28) (Table 2).
|
Five studies satisfied our criteria for a good quality study. When only good-quality studies were included in the meta-analysis, ppv (0.47), sensitivity (0.64) and specificity (0.97) were somewhat higher. However, subtantial heterogeneity remained.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
Diagnostic studies are rather rare, especially in a general practice setting. We therefore were impressed that for this subject 14 studies could be included, most of them published in the 1990s.
Most of them were not designed to investigate the diagnostic value of the DRE, but examined a whole range of methods to diagnose prostate cancer, and only mentioned the results of the DRE to compare its effectiveness with newer techniques. Many studies failed to report a number of quality parameters that were on the Cochrane scoring list. This hampers the judgement of the validity of the results. It seems advisable to standardize the reports of future studies according to these criteria, in the same way as was done for RCTs after the start of the RCT-meta-analysis boom.31
Setting
The description of the setting and the procedure of patient recruitment is often poor. Therefore, the judgement of the presence of selection bias and referral filter is difficult. In many screening populations people are not invited personally, but attracted by advertisements on television or in local newspapers. In these cases the population may be biased through self-selection, and may be underrepresenting, but more probably overrepresenting, people with prostate-related complaints. If this were the case, we would not be dealing any more with screening of a symptom-free population and the relative large range of prevalence rates per study would not be very reassuring at this point. As described by Knottnerus,32 this could influence all diagnostic parameters.
Test description
There are several ways to perform a DRE and to present its results, e.g. the position of the patient is relevant to the accessibility of the prostate gland. Therefore, the characteristics which are scored and the definition of a DRE positive result should be mentioned explicitly.
In most of the studies this was not the case, at least with respect to the classification of benign prostate hypertrophy. Some studies reported DRE positive or negative only, without mentioning the criteria that were used. This was especially so in studies which were directed primarily at the diagnostic value of other tests for detecting prostate cancer.
Reference standard
In some studies,16,21,23 up to 30% of patients eligible for biopsy were not biopsied, probably due to the invasiveness of the test. Three studies reported that patients with positive DRE results in a screening were sent to their own GP for follow-up.15,17,27 As no further data on the reference standard were presented, these studies were not included in this meta-analysis.
Applicability of DRE in screening
The DRE appears to be a test with a high specificity and a high negative predictive value. False negative test results are rare; this largely results from the small prevalence of cancer in an unselected population.
The large heterogeneity of the results was a surprising fact. This may be due to the different ways in which the studies were conducted, to differences in the interpretation of DRE or to different methods of inviting a population for screening. Even the five studies considered to be of ‘good-quality’ showed substantial heterogeneity.
From this study the following two conclusions can be formed.
- (i) The evidence from general-practice-based studies for the diagnostic value of DRE for the diagnosis of prostate cancer is based on a large number of low-quality studies and five good-quality studies. Reporting can be improved, e.g. by systematically refering to the Cochrane criteria list.
(ii) The DRE may have a place as an initial test when screening for prostate cancer. A negative test result of DRE has a high predictive value. The sensitivity being only moderate, however, should prevent the GP from drawing conclusions on the sole basis of such a result. Owing to its very low predictive value, a positive test result cannot be advocated as the basis for any important diagnosis without further confirmation. It therefore is very welcome that in recent studies, the DRE's diagnostic value has been studied in combination with other tests, e.g. ultrasonography or blood tests.
|
Appendix: criteria for evaluating the quality of diagnostic studies |
|---|
|
TopAbstractIntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
Criteria for study validity
- Was the test compared with a valid reference standard?
- Were the test and reference standard measured independently (blind) of each other.
- Was the choice of patients who were assessed by the reference standard independent of the test results?
- Was the test measured independently of all other clinical information?
Criteria relevant to the applicability of the results
- Spectrum of disease (e.g. cancer stage distribution if reference standard positive).
- Spectrum of non-disease.
- Setting.
- Duration of illness before testing.
- Previous tests/referral filter.
- Co-morbid conditions.
- Demographic information.
Test procedures
- Description of how the test was done.
- The explicit threshold used.
- Percentage excluded because test was unfeasible or result was indeterminate.
- Test reproducibility.
Indirect measurements of quality and applicability
- Year of publication.
- Disease prevalence.
- Sample size.
- Prospective or retrospective design.
- Published as a paper or as an abstract.
|
|
References |
|---|
|
TopAbstractIntroductionMethodsAnalysisResultsDiscussionAppendix: criteria for...References |
|---|
1 Irwig L, Macaskill P, Glasziou P, Fahey M. Meta-analytic methods for diagnostic test accuracy. J Clin Epidemiol 1995; 48(1): 19–130.[Web of Science][Medline]
2 Begg CB. Biases in the assessment of diagnostic tests. Statistics Med 1987; 6: 411–423.
3 Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology. A basic science for clinical medicine. 2nd edn. Boston: Little, Brown, 1991.
4 Mulrow C, Linn WD, Gaul MK, Pugh JA. Assessing quality of a diagnostic test evaluation. J Gen Internl Med 1989; 4: 288–295.[Web of Science][Medline]
5 Midgette A, Stukel T, Littenberg B. A meta-analytic method for summarizing diagnostic test performances. Med Decision Making 1993; 13: 253–257.
6 Dersimonian R, Laird N. Meta-analysis in clinical trials. Contr Clin Trials 1986; 7: 177–188.
7 Laird N, Mosteller F. Some statistical methods for combining experimental results. Int J Technol Assess Health Care 1990; 6: 5–30.[Medline]
8 Berlin J, Land NM, Sacks HS, Chalmers TC. A comparison of statistical methods for combining event rates from clinical trials. Stat Med 1989; 8: 141–151.[Web of Science][Medline]
9 Eddy D, Hasselblad V, Shachter R. Fast-Pro: software for meta-analysis by the confidence profile method. Boston: Acad Press, 1992.
10 Kirby RS, Kirby MG, Feneley MR, McNicholas T, McLean A, Webb JAW. Screening for carcinoma of the prostate: a GP based study. Br J Urol 1994; 74: 64–71.[Web of Science][Medline]
11 Vihko P, Kontturi M, Lukkarinen O, Ervasti J, Vihko R. Screening for carcinoma of the prostate. Cancer 1985; 56: 173–177.[Web of Science][Medline]
12 Chodak GW, Keller P, Schoenberg HW. Assessment of screening for prostate cancer using the digital rectal examination. J Urol 1989; 141: 1136–1138.[Web of Science][Medline]
13 Ciatto S, Bonardi R, Mazzotta A et al. Confronto tra due modalita di screening per il carcinoma prostatico. La Radiologia Medica 1994; 88: 453–457.[Medline]
14
Lee F, Torp-Pedersen ST, Siders DB. The role of transrectal ultrasound in the early detection of prostate cancer. Cancer J Clinicians 1989; 39: 337–360.
15 Terris MK, Stameg TA. Utilization of polyclonal serum prostate specific antigen levels in screening for prostate cancer: a comparison with corresponding monoclonal values. Br J Urol 1994; 73: 61–64.[Web of Science][Medline]
16 Pode D, Shapiro A, Lebensart P, Meretyk S, Katz G, Barak V. Screening for prostate cancer. Israelian J Med Sci 1995; 31: 125–128.
17 Moon TD, Clejan S. Prostate cancer screening in younger men: prostate-specific antigen and public awareness. Urology 1991; 38: 216–219.[Web of Science][Medline]
18 Dalkin BL, Ahmann FR, Kopp JB. Prostate specific antigen levels in men older than 50 years without clinical evidence of prostatic carcinoma. Urology 1993; 150: 1837–1839.
19 Palken M, Cobb OE, Simons E, Warren BH, Aldape HC. Prostate cancer: comparison of digital rectal examination and transrectal ultrasound for screening. Urology 1991; 145: 86–92.
20 Deliveliotis CH, Alivizatos G, Karayiannis A et al. The value of prostatic specific antigen in the early diagnosis of prostatic cancer: a Greek view. Urology 1995; 75: 637–641.
21 Teillac P, Bron J, Tobolski F et al. Dépistage du cancer de la prostate. Etude de 600 cas. Annales d'Urologie 1990; 24: 37–41.
22 Babaian RJ, Mettlin C, Kane R et al. The relationship of prostate-specific antigen to digital rectal examination and transrectal ultrasonography. Cancer 1992; 69: 1195–1200.[Web of Science][Medline]
23 Catalona WJ, Richie JP, Ahmann FR et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. Urology 1994; 151: 1283–1290.
24 Brawer MK, Chetner MP, Beatie J, Buchner DM, Vessella RL, Lange PH. Screening for prostatic carcinoma with prostate specific antigen. J Urol 1992; 147: 841–845.[Web of Science][Medline]
25 Menor FA, Orgaz RE, Gimeno MAN, Arbej JAP, Perez EM, Esteban FA. Ecografia transrectal y cancer de prostata. II. Comparacion con el tacto rectal. Urologia Oncologica 1990; 43: 261–263.
26 Richie JP, Kavoussi LR, Ho GT et al. Prostate cancer screening: role of digital rectal examination and prostate-specific antigen. Ann Surg Oncol 1994; 1: 117–120.[Web of Science][Medline]
27 Dillman RO, Williams T. Value of prostate cancer screening in a comprehensive community cancer center. Cancer Investig 1994; 12: 456–462.[Web of Science][Medline]
28 Catalona WJ, Smith DS, Ratliff TL et al. Measurement of prostate-specific antigen in serum as a screening test for prostate cancer. N Engl J Med 1991; 324: 1156–1161.[Abstract]
29 Gustafsson O, Norming U, Armgard LE et al. Diagnostic methods in the detection of prostate cancer: a study of a randomly selected population of 2,400 men. J Urol 1992; 148: 1827–1831.[Web of Science][Medline]
30 Littrup PJ, Kane RA, Murphy GP et al. Cost-effective prostate cancer detection. Cancer 1994; 74: 3146–3158.[Web of Science][Medline]
31 O'Rourke K, Detsky A. Meta-analysis in medical research: strong encouragement for higher quality in individual research efforts. J Clin Epidemiol 1989; 42: 1021–1024.[Web of Science][Medline]
32 Knottnerus J, Leffers P. The influence of referral patterns on the characteristics of diagnostic tests. J Clin Epidemiol 1992; 45: 1143–1154.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. W.S. Rutjes, J. B. Reitsma, M. Di Nisio, N. Smidt, J. C. van Rijn, and P. M.M. Bossuyt Evidence of bias and variation in diagnostic accuracy studies. Can. Med. Assoc. J., February 14, 2006; 174(4): 469 - 476. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Harris and K. N. Lohr Screening for Prostate Cancer: An Update of the Evidence for the U.S. Preventive Services Task Force Ann Intern Med, December 3, 2002; 137(11): 917 - 929. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||