Medicine & Science in Sports & Exercise:
Clinical Sciences: Clinical Investigations
Treadmill versus Shuttle Walk Tests of Walking Ability in Intermittent Claudication
ZWIERSKA, IRENA1; NAWAZ, SHAH2; WALKER, RICHARD D.2; WOOD, RICHARD F. M.2; POCKLEY, A GRAHAM2; SAXTON, JOHN M.1
1The Centre for Sport and Exercise Science, Sheffield Hallam University, Sheffield, UNITED KINGDOM; and 2Division of Clinical Sciences (North), University of Sheffield, Sheffield, UNITED KINGDOM
Address for correspondence: Dr. John M. Saxton, The Centre for Sport and Exercise Science, Sheffield Hallam University, Collegiate Crescent Campus, Sheffield, S10 2BP, United Kingdom; E-mail: firstname.lastname@example.org.
Submitted for publication March 2004.
Accepted for publication July 2004.
This study was supported by the British Heart Foundation (Project Grant PG-99130).
Purpose: To compare treadmill and shuttle walk tests for assessing functional capacity in patients with intermittent claudication, with respect to test-retest reliability, cardiovascular responses, and patient preferences.
Methods: Patients with stable intermittent claudication (N = 55, ages 52–85 yr, median age 68 yr) were recruited from the Sheffield Vascular Institute at the Northern General Hospital, Sheffield, UK. Each patient performed an incremental shuttle walk test, a constant-pace shuttle walk test, and a standardized treadmill test (3.2 km·h−1, 12% gradient), each on three occasions. The incremental shuttle walk began at 3 km·h−1 and increased by 0.5 km·h−1 every minute, whereas the constant-pace shuttle walk was performed at the fixed pace of 4 km·h−1. Claudication distance (CD), maximum walking distance (MWD), heart rate (HR), and blood pressure were assessed in each testing session. The patients also completed a test preference questionnaire.
Results: CD and MWD for both shuttle walks were greater than the corresponding walking distances achieved in the treadmill test (P < 0.001). Average coefficients of variation for repeated incremental shuttle walk, constant-pace shuttle walk, and treadmill tests were 15.9%, 21.1%, and 18.7%, respectively, for MWD, corresponding to average intraclass correlation coefficients of 0.87, 0.82, and 0.87. Treadmill walking evoked greater increases in HR and blood pressure (P < 0.001), and fewer patients expressed a preference for it (24 vs 43% for shuttle walking).
Conclusion: These findings indicated that shuttle walk testing exhibits similar test-retest reliability as treadmill testing, but that it evoked a lower level of cardiovascular stress and is preferred to treadmill testing by a large proportion of patients.
Intermittent claudication is the most common symptomatic manifestation of mild to moderate atherosclerotic peripheral arterial disease. It is a frequent complaint among the elderly (22), with a prevalence estimated at 3–7% in the general population (10). The condition is associated with a moderate to severe limitation in walking ability (18,24), which can adversely affect social, leisure, and occupational activities (10).
The functional capacity of patients with intermittent claudication is usually assessed using a standardized treadmill walking test, which provides an objective measure of walking performance (11). However, treadmills can be expensive and require the presence of trained personnel (4), which means that they might not be available in all clinical and rehabilitation settings. Furthermore, walking capacity assessed via treadmill testing might not accurately reflect the influence of claudication on everyday functional ability (6,19), and some elderly patients find treadmill assessments stressful (1) or impossible to perform due to restricting factors other than claudication pain (21).
One alternative or complementary exercise testing modality to treadmill walking for assessing the effect of the disease or treatment intervention on the functional capacity of a patient is a shuttle walk test, such as that developed by Singh et al. (26) for patients with chronic airways obstruction. In shuttle walk tests, patients walk back and forth between two cones placed a set distance apart on flat ground at a pace that is controlled by audiotape bleeps. Walking speed can be increased incrementally, which gradually stresses the cardiorespiratory system to a symptom-limited maximum (26), potentially making it safer for patients with cardiac and respiratory conditions.
Although incremental (26) and constant-pace (25) shuttle walk tests have been validated for the assessment of functional capacity in patients with chronic airways obstruction, this method of assessing functional ability has not been systematically investigated in patients with stable intermittent claudication. We hypothesized that shuttle walk testing would elicit less cardiovascular stress and yet be as reproducible as standardized treadmill testing for the assessment of walking performance in patients with mild to moderate intermittent claudication. Second, because shuttle walking is easy to perform, and requires only minimal resources and less patient familiarization, we expected that patients would prefer the shuttle walk assessment.
Fifty-five patients (52–85 yr, median age 68 yr) with stable intermittent claudication were recruited from the Sheffield Vascular Institute in the Sheffield Teaching Hospitals NHS Trust, Northern General Hospital, Sheffield, UK. Ethical approval was obtained from the North Sheffield local research ethics committee, and all patients provided informed consent before entering the study. Patients were excluded from the study if they had experienced symptoms of intermittent claudication for less than 12 months, had undergone a revascularization process within the previous 12 months, or if the initial assessment established that they suffered from regular shortness of breath, exercise-limiting angina, or severe arthritis. Demographic data and the resting ankle to brachial artery blood pressure index (ABPI) for the most symptomatic leg are shown in Table 1.
An incremental and a constant-pace shuttle walk test were compared with an internationally accepted, standardized treadmill protocol (15). The incremental shuttle walk test was a modified version of the protocol developed by Singh et al. (26). The speed of the constant-pace shuttle walk test was considered to be suitable for a large proportion of patients on the basis of previous walking assessments performed in our laboratory. Patients repeated the two shuttle walk protocols and the treadmill protocol on three occasions in a random order on separate days, thereby performing nine tests in total.
Shuttle walk tests.
Both the incremental and constant-pace shuttle walk tests required patients to walk back and forth between two cones placed 10 m apart on a flat floor. Walking speed was controlled by bleeps recorded onto an audiotape, and the accuracy of the timed signal was ensured by the inclusion on the tape of a calibration period of 1 min. Patients began walking on hearing the first bleep and aimed to reach the opposite cone by the next. If patients arrived before the bleep, they were required to wait for it before walking back. Patients arriving after the bleep were given verbal instructions to increase their walking speed so as to make it to the opposite cone before the next bleep. Patients achieved the correct walking pace within two or three repetitions (20–30 m).
The initial walking speed for the incremental shuttle walk was 3 km·h−1. At the end of each minute, the time interval between audible bleeps decreased, resulting in a step-increase in walking speed of 0.5 km·h−1. For the constant-pace shuttle walk test, patients walked at a fixed speed of 4 km·h−1 throughout the test.
An internationally accepted, standardized treadmill protocol (walking speed 3.2 km·h−1 at 12% gradient) (15) was used. Patients were instructed to use the handrail for balance only and not for supporting their weight.
Before each test, patients rested in a recumbent position before heart rate (HR) and systolic and diastolic blood pressure (at the brachial artery) were measured using a noninvasive ambulatory blood pressure recorder (Reynolds Medical, UK). The rate–pressure product (RPP), an index of myocardial workload (14,20), was calculated as the product of peak systolic blood pressure and HR.
During each testing protocol, the distance to the onset of claudication pain (claudication distance, CD) was determined, at which point patients were encouraged to continue walking until they could no longer tolerate the claudication pain (maximum walking distance, MWD). All tests concluded when patients (i) reached their MWD or (ii) could no longer maintain the required walking pace due to breathlessness or reasons other than intolerable claudication pain. For practical reasons, tests were also stopped if patients were able to walk comfortably at the prescribed pace for >10 min.
Systolic and diastolic blood pressure and HR were reassessed with the patient in a recumbent position as quickly as possible after the cessation of walking. As the treadmill and the couch were positioned centrally on either side of the 10-m walkway, the time taken to reach the couch for measurements was similar between test protocols. At the end of the study, patients completed a test preference questionnaire.
Walking performances and physiological variables between the test protocols were analyzed using repeated-measures ANOVA. The Student paired t-test was used to identify group mean differences, in which significance was indicated. The coefficient of variation is the main outcome measure used to describe the test-retest reliability of walking performance measures in the literature (3,8,13,16,17,22) and was used as the main outcome measure of test-retest reliability in the present study. Use of the coefficient of variation is recommended when the data are heteroscedastic or display nonuniform error variance, that is, when test-retest variation in a performance measure depends on the magnitude of the mean test score (2) as was generally the case for the walking protocols used in this study. The coefficient of variation (defined as the typical error, or within-subject standard deviation between trials expressed as a percentage of the mean score) between the first and second, and second and third trials of each test protocol was determined according to the method of Hopkins (http://www.sportsci.org/resource/stats/index.html). The intraclass correlation coefficient and Pearson product moment correlation coefficient were also determined for the repeated trials. As the exclusion of outliers had no effect on the results, all outliers were included in the data analysis. Data were analyzed using the SPSS (SPSS UK Ltd, Woking, UK) and sportsci.org (http://www.sportsci.org/resource/stats/index.html) statistical packages, and are expressed as means ± SEM. The level of statistical significance was set at P < 0.05.
Patient attendance and compliance.
Demographic data (sex, age, diabetic, and smoking status) and resting ABPI for the most symptomatic leg are shown in Table 1. Of the 55 patients recruited for this study, 51 patients completed all nine tests (459 tests in total). Of the 459 tests completed, some were terminated due to patients not being able to maintain the required walking pace for reasons other than intolerable claudication pain, or because they were able to walk comfortably at the prescribed pace for >10 min. Data from these assessments were excluded from the analysis. The number of tests that were terminated due to reasons other than attaining MWD were 7, 17, and 11 for the incremental shuttle walk, constant-pace shuttle walk, and treadmill, respectively. The percentage of tests eliciting CD were 86%, 86%, and 83%, and the percentage of tests eliciting MWD were 86%, 67%, and 80% for the incremental shuttle walk, constant-pace shuttle walk, and treadmill test, respectively. The reduced percentage of constant-pace shuttle walk tests eliciting MWD was mainly due to a higher proportion of patients (22%) being able to walk for >10 min.
The average CD and MWD were similar between the two shuttle walk tests. However, the average CD and MWD achieved in the incremental and constant-pace shuttle walk tests were approximately twice the distances achieved in the treadmill test (P < 0.001). The average CD for the incremental shuttle walk, constant-pace shuttle walk, and treadmill tests were 149 ± 11, 147 ± 16, and 75 ± 10 m, respectively, corresponding to 263 ± 17, 217 ± 17, and 115 ± 13 m for average MWD on the three tests (Fig. 1). Walking abilities varied widely, with the ranges in MWD for the incremental and constant-pace shuttle walks and treadmill test being 99–552, 92–497, and 67–557 m, respectively.
FIGURE 1Claudication...Image Tools
The CD and MWD achieved on the three occasions on which the incremental and constant-pace shuttle walk tests were undertaken were not statistically different. In the case of the treadmill protocol, there was no difference in the MWD achieved; however, the CD recorded for the second and third trials was greater than that recorded for the first trial (P < 0.05). The CD recorded for the second and third trials were not significantly different (Fig. 1).
The test-retest reliability of the incremental shuttle walk was slightly better than the constant-pace shuttle walk and treadmill protocol for MWD (Table 2), with average test-retest coefficients of variation of 15.9%, 21.1%, and 18.7%, respectively. However, the average test-retest coefficient of variation in the measured CD for the treadmill protocol (21.1%) was lower than that for the incremental (27.5%) and constant-pace (28.7%) shuttle walks. Intraclass and Pearson product moment correlation coefficients for repeated trials of the respective protocols for CD and MWD were generally above 0.80 (Table 2).
Preassessment blood pressures and HR were similar for all tests (Fig. 2). Systolic blood pressure, HR, and the RPP were elevated above resting values after all test protocols (P < 0.001), whereas diastolic blood pressure was only elevated after the incremental shuttle walk (P < 0.05) and treadmill tests (P < 0.001). The treadmill test elicited a higher level of cardiovascular stress than the incremental and constant-pace shuttle walks (Fig. 2), as indicated by greater increases in systolic blood pressure (P < 0.001), HR (P < 0.001), and RPP (P < 0.001). The increase in diastolic blood pressure after the treadmill test was greater than that observed after the incremental shuttle walk test (P < 0.05). Although the treadmill test elicited higher blood pressure and HR responses, these parameters remained within accepted safe limits.
FIGURE 2Pre- and pos...Image Tools
When asked if they had any preferences for the shuttle walk and treadmill protocols, 22 patients (43%) indicated that they preferred shuttle walking, whereas only 12 (24%) expressed a preference for the treadmill test. Seventeen of the patients (33%) did not express a preference for either test. When asked to indicate which of the two shuttle walks they preferred, nine (18%) and 11 (21%) patients preferred the incremental and constant-pace shuttle walk tests, respectively. The majority of patients (61%) expressed equal preference for both shuttle walk tests.
This study compared an incremental and a constant-pace shuttle walk test with an internationally accepted, standardized treadmill protocol for assessing walking performance in patients with stable intermittent claudication. A prerequisite for any test that is to be used to quantify walking capacity in this patient group is that it displays acceptable test-retest reliability (17), as defined in terms of the coefficient of variation and intraclass and Pearson product moment correlation coefficients of repeated tests, and that there is no significant difference in the readings obtained from repeated tests.
This study demonstrates that the intraclass and Pearson product moment correlation coefficients for MWD were high (≥0.81) and comparable for all protocols. The lowest coefficient of variation was observed for the incremental shuttle walk test, thereby indicating a superior test-retest reliability for this protocol. In the case of the CD, the lowest coefficient of variation was observed for the treadmill test, although a learning effect was apparent between the first two assessments. This is an important finding, as previous treadmill test-retest reliability studies have assessed patients on only two occasions (5,17), and a more thorough familiarization of patients under test conditions might be required before a reproducible constant-pace treadmill CD can be determined. The need for such a familiarization appears to be reduced for shuttle walk protocols, as no learning effect was observed between the repeated tests. This may also be true for incremental treadmill protocols, which have been shown to have higher test-retest reliability than constant-pace treadmill tests (8,23). Nevertheless, in the clinical setting, it is good practice to ensure that all patients are familiarized with the testing procedures before assessment.
The test-retest reliability data for treadmill and shuttle walk testing obtained in this study are comparable to those reported by others in the literature. Most treadmill studies have reported coefficients of variation for CD and MWD in the range 15–40% (3,8,16,17,22), and two studies have reported good test-retest reliability coefficients (>0.90) for incremental (26) and constant-pace (25) shuttle walk tests in patients with chronic airways obstruction. The variability of test-retest reliability coefficients reported from treadmill studies can probably be explained by differences in the testing protocols used and/or the way in which the test was administered, as well as the level of prior patient familiarization and characteristics of the sample population. It is of interest to note that test-retest reliability for the MWD was better than that for CD in both the shuttle walk and treadmill protocols, which is in agreement with previous treadmill studies and the accepted view that the CD is a less reliable measure of walking ability in this patient group (8,9,12,13).
The CD and MWD achieved in the incremental shuttle walk test were approximately twice the distances achieved in the treadmill test. These data support the view that treadmill protocols using inclines do not reflect the influence of claudication on everyday functional activities (19), nor do they measure the true functional capacity (6) of patients walking on level ground. In the clinical setting, treadmill testing is widely used to exceed the capacity of lower-limb collateral circulation, thereby helping to establish the diagnosis of exercise leg pain in patients with a normal resting ABPI. However, the ability to define a change in functional capacity is an important component of clinical trials that are evaluating the efficacy of treatment interventions, such as exercise rehabilitation (13). Thus, when the primary aim of an exercise test is to estimate the effect of lower-limb peripheral arterial disease on daily ability (either before or after treatment), our results suggest that shuttle walking could be considered as a useful alternative or complementary testing modality.
During exercise testing, patients with lower-limb peripheral arterial disease can experience cardiovascular problems such as hypertension (7). The treadmill test elicited a higher level of cardiovascular stress than the incremental and constant-pace shuttle walks, as indicated by its induction of greater increases in systolic blood pressure, HR and RPP. The increase in diastolic blood pressure after the treadmill test was also greater than that observed after the incremental shuttle walk test. Incremental shuttle walking stresses the cardiovascular system more progressively to a symptom-limited maximum (26), potentially making it safer than the incline treadmill test used in this study for patients with cardiac and respiratory conditions. The reduction in cardiovascular stress associated with incremental shuttle walking tests could be particularly beneficial in instances when patients with more severe cardiovascular comorbidity are being evaluated by helping to reduce the risk of an exaggerated blood pressure response to exercise and the probability of cardiovascular events. Similar benefits can be gained from standardized progressive treadmill tests.
The majority of patients who expressed a preference said that they would prefer to undertake a shuttle walk test rather than a treadmill test. A number of reasons were given, including a fear of falling over on the treadmill, and finding treadmill walking too fast and difficult to perform. Of those patients expressing a preference for shuttle walking, patients preferring the incremental shuttle walk found the increase in speed to be more interesting and challenging than the constant-pace shuttle walk test. However, some patients considered the constant-pace shuttle walk to be more representative of day-to-day activities.
For patients with mild claudication symptoms, walking speed in the constant-pace shuttle walk was too slow to assess MWD, as 11 patients were able to walk for longer than 10 min despite experiencing claudication pain. A similar problem has previously been highlighted for constant-pace treadmill protocols (23). The use of constant-pace tests can also be problematic in intervention studies, as patients limited by claudication on entry to a program might fail to exhibit a claudication response after an intervention from which they experience a large improvement (23). Conversely, the constant speed was too fast for six patients, and the test was terminated because the pace could not be maintained or the patient became breathless. A constant-pace walking test based on individual exercise capacity (25) would be a better approach, although the time commitment required to derive individualized walking speeds that are based on a percentage of peak oxygen consumption would be inappropriate for most clinical and rehabilitation settings.
In contrast to the limitations associated with constant-pace tests, our results show that the incremental shuttle walk can be used to assess CD and MWD in patients with a wide range of walking abilities, as previously observed for incremental treadmill tests (16). Patients can therefore be successfully reassessed after an intervention, irrespective of the level of improvement. However, the main potential weakness of the incremental shuttle walk is that changes in the ability to walk for long periods at constant submaximal intensities after an intervention might not be discernible from the results of this test. This was true for some patients with chronic airways obstruction (25) but has yet to be confirmed for patients with peripheral arterial disease.
In summary, we propose that the incremental shuttle walk test can be considered as an alternative or complementary testing modality to treadmill walking when there is a need to estimate the effect of lower-extremity peripheral arterial disease on daily functional ability. Shuttle walk testing has similar test-retest reliability as treadmill testing, but evokes a lower level of cardiovascular stress and is preferred to incline treadmill testing in a large proportion of patients. Shuttle walking is also cheaper and easier to perform than treadmill testing, requiring less patient familiarization and operator training. A potential advantage of the incremental shuttle walk over the constant-pace shuttle walk is the possibility that it could be used to assess CD and MWD in patients with wide-ranging disease severity before and after treatment interventions where an improvement in walking ability is expected. This needs to be confirmed in future studies.
1. Amirhamzeh, M. M., H. J. Chant, J. L. Rees, L. J. Hands, R. J. Powell, and W. B. Campbell. A comparative study of treadmill tests and heel raising exercise for peripheral arterial disease. Eur. J. Vasc. Endovasc. Surg
. 13:301–305, 1997.
2. Atkinson, G., and A. M. Nevill. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Med
. 26:217–238, 1998.
3. Cachovan, M., W. Rogatti, F. Woltering, A. Creutzig, C. Diehm, H. Heidrich, and P. Scheffler. Randomized reliability study evaluating constant-load and graded-exercise treadmill test for intermittent claudication. Angiology
4. Cameron, A. E. P., A. Porter, S. Rosser, A. E. C. F. Da Silva, and L. M. De Cossart. The Stresst'er Ergometer as an alternative to treadmill testing in patients with claudication. Eur. J. Endovasc. Surg
. 14:433–438, 1997.
5. Chaudhry, H., A. Holland, and J. Dormandy. Comparison of graded versus constant treadmill test protocols for quantifying intermittent claudication. Vasc. Med
. 2:93–97, 1997.
6. Coughlin, P. A., P. J. Kent, E. P. Turton, et al. A new device for the measurement of disease severity in patients with intermittent claudication. Eur. J. Vasc. Endovasc. Surg
. 22:516–522, 2001.
7. Gardner, A. W., J. S. Skinner, C. X. Bryant, and L. K. Smith. Stair climbing elicits a lower cardiovascular demand than walking in claudication patients. J. Cardiopulm. Rehabil
. 15:134–142, 1995.
8. Gardner, A. W., J. S. Skinner, B. W. Cantwell, and L. K. Smith. Progressive vs single-stage treadmill tests for evaluation of claudication. Med. Sci. Sports Exerc
. 23:402–408, 1991.
9. Gardner, A. W., J. S. Skinner, and L. K. Smith. Effects of handrail support on claudication and hemodynamic responses to single-stage and progressive treadmill protocols in peripheral vascular occlusive disease. Am. J. Cardiol
. 68:99–105, 1991.
10. Hiatt, W. R. Quality of life assessment in peripheral arterial disease. Atherosclerosis
11. Hiatt, W. R. Functional assessment of the claudicant. Importance of treatment and follow-up strategies. Minerva Cardioangiologica
12. Hiatt, W. R., A. T. Hirsch, J. G. Regensteiner, and E. P. Brass. Clinical trials for claudication: assessment of exercise performance, functional status, and clinical end points: vascular clinical trialists. Circulation
13. Hiatt, W. R., S. Nawaz, J. G. Regensteiner, and K. F. Hossack. The evaluation of exercise performance in patients with peripheral vascular disease. J. Cardiopulm. Rehabil
. 12:525–532, 1988.
14. Kitamura, K., C. R. Jorgensen, F. L. Gobel, H. L. Taylor, and Y. Wang. Hemodynamic correlates of myocardial oxygen consumption during upright exercise. J. Appl. Physiol
. 32:516–522, 1972.
15. Labs, K. H., J. A. Dormandy, K. A. Jaeger, C. S. Stuerzebecher, and W. R. Hiatt. Transatlantic conference on clinical trial guidelines in peripheral arterial disease: clinical trial methodology: Basel PAD Clinical Trial Methodology Group. Circulation
16. Labs, K. H., M. R. Nehler, M. Roessner, K. A. Jaeger, and W. R. Hiatt. Reliability of treadmill testing in peripheral arterial disease: a comparison of a constant load with a graded load treadmill protocol. Vasc. Med
. 4:239–246, 1999.
17. Labs, K. H., M. Roessner, M. Aschwanden, C. Jeanneret, A. Gehrig, and K. A. Jaeger. Reproducibility of fixed load treadmill testing. J. Vasc. Invest
. 4:55–60, 1998.
18. McDermott, M. M., K. Lui, J. M. Guralnik, et al. The ankle brachial index independently predicts walking velocity and walking endurance in peripheral arterial disease. J. Am. Geriatr. Soc
. 46:1355–1362, 1998.
19. Montgomery, P. S., and A. W. Gardner. The clinical utility of a six-minute walk test in peripheral arterial occlusive disease patients. J. Am. Geriatr. Soc
. 46:706–711, 1998.
20. Nelson, R. R., F. L. Gobel, C. R. Jorgensen, K. Wang, Y. Wang, and H. L. Taylor. Hemodynamic predictors of myocardial oxygen consumption during static and dynamic exercise. Circulation
21. Perakyla, T., H. Lindholm, and M. Lepantalo. Assessment of intermittent claudication: a comparison of questionnaire, visual analogue scale and subjective estimate information with post-exercise ankle-brachial pressure index. Clin. Physiol
. 19:445–449, 1999.
22. Perakyla, T., H. Tikkanen, J. von Knorring, and M. Lepantalo. Poor reproducibility of exercise test in assessment of claudication. Clin. Physiol
. 18:187–193, 1998.
23. Regensteiner, J. G., and W. R. Hiatt. Exercise rehabilitation for patients with peripheral arterial disease. Exerc. Sport Sci. Rev
. 23:1–24, 1995.
24. Regensteiner, J. G., J. F. Steiner, R. J. Panzer, and W. R. Hiatt. Evaluation of walking impairment by questionnaire in patients with peripheral arterial-disease. J. Vasc. Med. Biol
. 2:142–152, 1990.
25. Revill, S. M., M. D. Morgan, S. J. Singh, J. Williams, and A. E. Hardman. The endurance shuttle walk: a new field test for the assessment of endurance capacity in chronic obstructive pulmonary disease. Thorax
26. Singh, S. J., M. D. L. Morgan, S. Scott, D. Walters, and A. E. Hardman. Development of a shuttle walking test of disability in patients with chronic airways obstruction. Thorax
PERIPHERAL ARTERIAL DISEASE; EXERCISE TESTING; TEST-RETEST RELIABILITY; CARDIOVASCULAR RESPONSES
©2004The American College of Sports Medicine
Highlight selected keywords in the article text.