1 s2.0 0895435693900534 Main
1 s2.0 0895435693900534 Main
1 s2.0 0895435693900534 Main
00
Printed in Great Britain. All rights reserved Copyright 0 1993 Pergamon Press Ltd
Abstract-A Physical Activity Scale for the Elderly (PASE) was evaluated in a sample
of community-dwelling, older adults. Respondents were randomly assigned to complete
the PASE by mail or telephone before or after a home visit assessment. Item weights
for the PASE were derived by regressing a physical activity principal component score
on responses to the PASE. The component score was based on 3-day motion sensor
counts, a 3-day physical activity diary and a global activity self-assessment. Test-retest
reliability, assessed over a 3-7 week interval, was 0.75 (95% CI = 0.69-0.80). Reliability
for mail administration (r = 0.84) was higher than for telephone administration
(r = 0.68). Construct validity was established by correlating PASE scores with health
status and physiologic measures. As hypothesized, PASE scores were positively associated
with grip strength (r = 0.37), static balance (r = +0.33), leg strength (r = 0.25) and
negatively correlated with resting heart rate (r = - 0.13), age (r = - 0.34) and perceived
health status (r = -0.34); and overall Sickness Impact Profile score (r = -0.42). The
PASE is a brief, easily scored, reliable and valid instrument for the assessment of
physical activity in epidemiologic studies of older people.
median 1980 household income (less than or analysis. Height and weight were measured
more than $20,000). Half of the towns in each using standardized procedures patterned after
stratum were randomly selected. Due to its the Pawtucket Heart Study protocol [25], with
size, the city of Springfield was a separate self- respondents in stockinged feet and indoor
representing stratum. In the second stage, 1989 clothing. Height was rounded up to the nearest
Massachusetts street lists (a state-mandated eighth of an inch and weight was rounded down
census of persons of voting age) in the selected to the nearest pound.
towns were used to identify eligible older adults. Grip strength of the dominant hand was
Persons born in 1924 or earlier were systematic- assessed with respondents in a standing position.
ally sampled from these lists at a rate propor- Static balance of the dominant leg (same side
tional to the total number of adults aged 65 as dominant arm) was assessed by the one leg
or older in each town. Unequal selection prob- stance test with eyes closed [26]. Respondents in
abilities in the first stage were offset by sampling stockinged feet were instructed to close their
eligible adults in Springfield at half the rate used eyes and raise their non-dominant foot from the
in the other 11 towns to produce a self-weighting floor. Balance time (to the nearest 0.1 second)
sample of individuals in the target area. was assessed with a stopwatch from the time the
non-dominant foot left the floor until either the
Survey protocol dominant foot was displaced, the non-dominant
Each adult in the sample was randomly leg touched the dominant leg, or the non-
assigned to one of four groups based on the type dominant leg touched the floor.
of PASE administration (telephone or mail) and Isometric knee extensor strength at 60” knee
the timing of data collection (home visits con- flexion was measured with a portable Isokinetic,
ducted before or after the PASE questionnaire). Inc. (Grand Rapids, MI) knee unit [271. This
Half of the sample received home visits first and unit consists of a padded seat and a bracket
then completed either the mail or telephone that holds a spring gauge and a cuff assembly
questionnaire; the other half were administered for attachment to the respondent’s leg directly
the physical activity questionnaire prior to the above the lateral malleolus. Respondents sat on
home visit. Using the addresses appearing in the padded seat with their popliteal fossa placed
the street lists, each eligible person was sent against the front of the padded surface. For
an introductory letter explaining the purposes stabilization the thigh of the leg being tested was
of the study. Subjects were then contacted by strapped to the seat. Respondents sat with a
telephone to schedule a home visit. A minimum straight back with hands grasping the side of the
of 10 calls was made to each household to locate padded seat and were asked to exert maximal
respondents. Persons who had died, lived in nurs- force against the ankle cuff. Testing of the
ing homes, had serious cognitive impairments, dominant leg always preceded testing of the
or could not speak or read English were not non-dominant leg.
eligible. Persons assigned to groups in which The results of three separate trials were
interviews were to be conducted prior to visits recorded for grip strength, static balance and leg
were asked to complete the PASE even if strength. The mean of these three trials was used
they refused to permit a home visit. Baseline in statistical analyses.
respondents were recontacted 3 weeks later and Health status was assessed by the Sickness
asked to complete a second questionnaire to Impact Profile (SIP), a measure of the impact
assess test-retest reliability. of disease on daily activities and behaviors in
12 functional areas [28, 291. Demographic
In -home protocol characteristics were reported using standard
All home visits were conducted by trained field items from national surveys. At the conclusion
technicians between January 1990 and February of the home visit, field technicians explained the
1991. Written informed consent was obtained use of the movement counter and an activity
from all respondents. Home visit measures were diary. Respondents were asked to wear the
collected in the following order. Blood pressure movement counter and record their activity
(BP) was measured three times using a standard patterns for the next 3 days.
mercury sphygmomanometer with the respond- Activity monitor. Physical activity was
ent seated for at least 5 minutes prior to monitored using a Caltrac Personal Activity
measurement with legs uncrossed at the time Computer (Hemokinetics Inc., Madison, WI).
of readings. The last BP reading was used in the Details regarding development and construction
156 R~CHARLI
A. WASHBURNet al.
of the Caltrac as well as the validity of the ity diary, and the global self-report of physical
Caltrac for older people are available elsewhere activity. This approach, which is rooted in
[30, 311. The Caltrac is a small, lightweight (9.5 classical test theory [33] and confirmatory factor
cm x 7.0 cm x 1.25 cm; weight = 75 g) device analysis [34], treats these three measures as
designed to measure acceleration via a piezo- fallible indicators of an unobserved physical
electric bender element. A numerical score (kcal) activity construct. A principal component score
is provided by a liquid crystal display. The total for each subject was computed from the respect-
kcal score is a function of the respondent’s basal ive item loadings. These component scores,
metabolic rate calculated by a computer chip which represent our most refined estimate of the
programmed with the respondents age, height, underlying physical activity construct, were then
weight and gender, plus additional caloric regressed on responses to the questionnaire to
expenditure resulting from body movement. derive the optimal item weights for the PASE.
Since our purpose was to use the Caltrac only Total PASE scores were computed by multi-
as a movement counter, we by-passed the meta- plying the amount of time spent in each activity
bolic program as instructed by the manufacturer (hours per day over a 7-day period) by the re-
and used daily Caltrac counts in the analysis. spective weights and summing over all activities.
Respondents were instructed to wear the Caltrac
on a belt over the dominant hip and record Validation and reliability assessment
Caltrac readings and the time of day both in The stability of the PASE over time was
the morning and on retiring for the evening on assessed by the test-retest reliability correlation
a chart attached to an activity diary. between baseline scores and follow-up scores
Activity diary. For each waking hour during reported 3-7 weeks later. To validate PASE
the 3 day observation period, respondents were scores, Pearson correlations were computed
asked to maintain an activity diary of the amount between these scores and measurements taken
of time spent in eight activity categories: lying during home visits. Validation measures included
down, sitting, standing, standing light work physiologic characteristics known to be affected
(dishes, dusting), standing moderate/heavy work by activity levels (heart rate, body mass index,
(carpentry, gardening, lifting), walking, light balance, grip and leg strength) [9, 10,35-371 as
sport and recreation (golf, bowling, ball games), well as aspects of health status that influence the
and heavy sport and recreation (running, ability to perform physical activities (total SIP
cycling). Daily energy expenditures (METS) score, self-assessed health status, and selected
were calculated by multiplying the amount acute and chronic health conditions). Correla-
of time spent in an activity by a MET value tions with the validation measures were also
reflecting the intensity of that activity. MET computed for six respondent subgroups (based
values ranged from 1.0 for lying down to 6.0 on mode of questionnaire administration, gender
for heavy sport and recreation [32]. and age) to determine the consistency of these
After the third day, respondents also com- associations. In addition, we examined seasonal
pleted a 5-point scale assessing their level of trends and respondent characteristics associated
physical activity. Scale values ranged from with PASE scores.
1 = not active at all to 5 = extremely active.
Caltracs, diaries and self-report scale scores
RESULTS
were returned to the investigators by mail. Daily
averages for the Caltrac counts and diary METS Response rates
were determined for the 3-day monitoring Dispositions for the 1288 names sampled
period. Data were not included in the averages from the street lists are shown in Table 1. Two
if the reporting periods for the diary and Caltrac hundred twenty-four persons (19.8%) were
differed by more than 2 hours on a given day. ineligible for the study. Contact was not made
with another 159 whose eligibility status could
PASE scoring not be determined. Of those known to be
To devise a set of weights for the PASE items eligible, 36.0% consented to a home visit and
that would provide the best overall estimate to complete the PASE. An additional 15.5%
of an older person’s physical activity level, a completed the PASE but refused a home visit.
criterion measure of physical activity was Table 2 compares the background characteristics
created from a principal components analysis of participants with non-participants. Non-
of Caltrac counts, METS totals from the activ- participants were on average 2 years older than
Physical Activity Scale for the Elderly 157
Table 1. Disposition of street list names Table 2. Background characteristics of participants and
non-participants
NllmbeC
of Participants
Cases Disposition
Homevisit
159 No contact (moved, telephone disconnected, unlisted Non- and PASE
telephone number, no answer) participants PASE only
224 Ineligible (deceased, nursing home resident, mental/ Characteristics (n = 668)8 (n = 277) (n = 119)
physical impairment, younger than 65 years)
Age (yr) 75.0** 73.0 73.4
136 Refused telephone screener 71.4
Percent female 61.7* 57.0
251 Refused home visit, not asked to complete PASE
Percent living with spouse 40.9 48.7 40.3
122 Refused home visit, and failed to complete PASE
Town income (median
119 Refused home visit, but completed PASE 18.5
dollars in thousands) 18.9 18.9
277 Completed home visit and PASE 18.8
Percent employed 17.2 17.7
1288 Total names sampled from street lists Perceived health 2.89b 2.73 2.86
(1 = excellent to 5 = poor)
Activity level 2.98b 2.89 2.98
(1 = very high to
participants. Women were more likely than 5 = very low)
men to refuse a home visit, However, those who Worry about health 2.32b 2.26 2.34
(1 = not at all to
completed the telephone screener but refused to 4 = most of the time)
participate in all other aspects of the study were
‘Includes no contact cases, screener refusals, and those who
similar to participants with respect to perceived failed to complete a PASE.
health, physical activity levels and perceived bn = 378 non-participants who refused home visit and PASE
worry about their health. but completed telephone screening.
*p < 0.05; **p < 0.01.
PASE score descriptive statistics
Figure 1 shows the results of the principal to 3.54 with higher scores indicating greater
components analysis for the 193 subjects with physical activity.
complete data for the Caltrac, activity diary and Weights for individual activities were esti-
global self-report item. The inter-item correla- mated by regressing component scores on the
tions among the three physical activity measures complete set of items in the original version of
were moderately high and in the expected direc- the PASE. Twelve types of activity accounted
tion. The three measures had similar factor for 41.4% of the variation in component scores.
loadings on a single underlying component Seven low expenditure activities (sleeping,
(eigenvalue = 1.87). The internal consistency napping, quiet activities, flexibility exercises,
of these items as measured by Cronbach’s alpha stair climbing, shopping or errands, and jobs
was 0.69. The resulting component scores (mean involving sitting with slight arm movements)
= 0, standard deviation = 1) ranged from -2.44 that were not significantly associated with
Cronbach’s (I = ,694
Fig. 1. Relationships between physical activity component and indicators of physical activity (n = 193).
Curved arrows signify zero-order correlations; straight arrows indicate component loadings.
158 hXARD A. WASHBURNet al.
(144.3)
\.;,,
j loo (llGj------..__
---._
?! $+...\.
‘.._
; ‘.._
s” .‘v WOMEN
50 (62.3)
I I I
Aee@otJP
Table 4. Validity correlations for mail and telephone versions of PASE by mode, gender and age group
Mode Gender Age group
All Mail Telephone
Validation measures subjects Questionnaire Questionnaire Female Male 65-70 71-99
Perceived health -0.34** -0.26* -0.37*+ -0.41** -0.29** -0.39** -0.24**
(1 = excellent,
5 = poor)
Anv restricted -0.12 0.03 -0.21** -0.23** 0.07 -0.16 -0.09
activity days
(1 = yes, 0 = no)
Sick Impact Profile
Total Score -0.42** -0.42** -0.46** -0.37.’ -0.40** -0.36.’ -0.42..
Heart rate -0.13* -0.32** -0.01 -0.15 -0.03 -0.16 -0.12
Systolic BP -0.09 -0.03 -0.14’ -0.19* 0.08 -0.09 0.01
Diastolic BP -0.07 0.12 0.06 0.05 0.04 0.05 0.09
Body mass (kg/m2) 0.01 -0.05 0.03 0.08 -0.04 -0.10 0.01
Grip strength 0.37** 0.34** 0.37** 0.40** 0.32** 0.26** 0.38**
Balance 0.33** 0.39** 0.33** 0.33** 0.29** 0.14 0.42**
Dominant leg 0.25** 0.24’ 0.26+* 0.32** 0.06 0.12 0.25;’
strength
Non-dominant leg 0.28** 0.23* 0.30** 0.33” 0.10 0.09 0.33**
strong correlations with Sickness Impact Profile nificant correlations across a variety of health
scores and perceived health status, but a much status and physiologic measures provides strong
weaker relationship with restricted activity days evidence for the convergent validity of the
in the previous week. PASE scores were also PASE scoring algorithm.
positively correlated with grip strength, static PASE scores exhibited seasonal variations
balance, and leg strength in both the dominant (Fig. 4). As one would expect in New England,
and non-dominant legs. Activity levels measured the highest levels of physical activity are
by PASE were not associated with body mass reported during the summer months while the
index or blood pressure readings in this sample. lowest levels occurred during the coldest months
With few exceptions, these correlations were of winter. The correlation between average
consistent by mode of administration, gender, monthly temperatures and monthly PASE
and age group. The pattern of statistically sig- means in this sample was 0.83 (n = 12 months).
PASE Scorellkmperaturc
160
60 -
Jan Feb Mar Apr May Jun Jul Aug Sep Ott Nov Dee
Interview Month
(N-314)
estimates is beyond the scope of a brief physical version. The high reliability coefficient for mail
activity assessment instrument. administration indicates that random error is
Comparisons with data collected during home not a major problem with the mail version of the
visits provide strong evidence for the convergent instrument.
validity of the PASE. The correlations between Because of these concerns, we recommend
PASE scores and health status, strength and that the telephone version of PASE be the
balance were all in the hypothesized direction method of first choice and suggest that the
and of moderate strength (range r = 0.25 to mail questionnaire be used in a modified form.
r = 0.42). Larger correlations would not be Additional respondent instructions have been
expected given the influence of factors other added to the mail version clarifying the proper
than physical activity on health-related variables. categorization of activities. These clarifications
Only one other study has assessed the validity should reduce reporting error in the mail
of physical activity questionnaire with older version. The revised mail version of PASE,
respondents. However, this analysis was based however, should be field tested to confirm that
on a small, non-representative sample using the recording problem has been corrected.
only 24-hour activity recalls and pedometer Although we did not specifically evaluate the
counts as validation criteria [38]. reliability and validity of a face-to-face version
The PASE test-retest reliability coefficient of PASE, our experiences with telephone
(0.75) exceeds those reported for other physical interview administration suggest this mode of
activity surveys. Sallis et al. [20], for example, administration should provide reliable and valid
reported a 2 week test-retest correlation of 0.67 physical activity assessments.
for the Five-Cities Activity survey in 53 men and This investigation was unique in applying an
women of a mean age of 41. In a random empirical approach to constructing a physical
population sample of 633 men and women, activity instrument and in evaluating it in a
ages 25-65 years, Washburn et al. [39] reported probability sample of community-dwelling older
a 7-12 week test-retest correlation for the adults. Although our analysis reveals that
Harvard Alumni Physical Activity survey of participants as compared with non-participants
0.58. Like the PASE, both the Five-Cities and were slightly younger and more likely to be male,
Harvard Alumni surveys are based on 7 day there were no differences between participants
recalls. In the current study, discordance between and non-participants in employment, perceived
physical activity estimates over the 3-7 week health, concern about health or level of physical
follow-up period can reflect actual changes in activity. This implies that the validation and
physical activity as well as unreliable reporting. reliability results may be generalized to the popu-
In this context, the PASE test-retest correla- lation of community-dwelling older persons.
tions of 0.68 for telephone and 0.84 for mail The substantive finding that physical activity of
administration are comparatively large for a older persons, as measured by PASE, is related
physical activity assessment instrument. to age, gender, employment status, and chronic
The observed effect of mode of administra- respiratory disease factors in this sample were
tion on physical activity estimates is of concern. consistent with the literature [40,41]. It is not
Our results indicated that PASE scores were clear, however, why PASE scores for those
nearly 18 points higher, on average, when the individuals reporting cancer were higher than
PASE was administered by mail compared to average unless many of these cancers were in
telephone. The direction of this effect suggests remission.
that the observed difference was most likely due Future administration of PASE in larger
to respondents’ double reporting activities in the samples of older persons will be needed to
mail version of PASE. For example, field staff develop normative values of physical activity
indicated that subjects reported walking in in older persons. In addition, it is important to
response to the question specifically on walking determine the sensitivity of PASE in detecting
and again under light or moderate recreational change in physical activity to assess its utility as
activity. This did not occur in the telephone an evaluation instrument.
version where the interviewer could probe in
response to questionable information. However,
interviewer probing may introduce inconsistency REFERENCES
in the responses and be responsible for the lower 1. Heckler MM. Health promotion for the elderly. Public
test-retest reliability seen with the telephone Health Rep 1985; 100: 225-230.
162 RICHARDA. WASHBURNet al.
2. Kaplan GA, Seeman TE, Cohen RP et al. Mortality 23. Gentry EM, Kalsbeer WD, Hogelin GC et al. The
among the elderly in the Alameda County study: behavioral risk factor surveys: II. Design, methods,
behavioral and demographic risk factors. Am J Public and estimates from combined state data. Am J Prov
Health 1987; 77: 307-312. Med 1985; 6: 9-14.
3. Mor V, Murphy J, Masterson-Allen S et al. Risk of 24. Washburn RA, Jette AM, Janney CA. Using age-
functional decline among well elders. J Clln Epidemiol neutral physical activity questionnaires in research
1989; 42: 895-904. with the-elderly. J Aging dealth 1990; 2: 341-356.
4. Sorock GS, Bush TL, Golden AL et al. Physical 25. McKinlav_ SM. Kinn _. DM. Johnson P et al. A field
I
activity and fracture risk in a free-living elderly cohort. approach for obtaining physiological measures in
J Gerontol 1986; 43: M134-139. surveys of genera1 populations: response rates. reli-
5. Campbell AJ, Borrie MJ, Spears GF. Risk factors for ability, and costs. Health Survey Research Methods:
falls in a community-based prospective study of people Proc Fourth Conf on Health Survev Reseach Methods.
70 years and older. J Gerontol 1989; 44: Ml12-117. Washington, DC: 1984. w
6. Aloia JF, Vaswani AN, Yeh JK et al. Pre-menopausal 26. Briggs RC, Gossman MR, Birch R et al. Balance per-
bone mass is related to physical activity. Arch Intern formance among non-institutionalized elderly women.
Mad 1988; 148: 121-123. Phys Ther 1989; 69: 748-756.
I. Snow-Harter C, Marcus R. Exercise, bone mineral 27. Finucane SD, Walker ML, Rothstein JM ef al.
density, and osteoporosis. Exer Sport Sci Rev 1991: 19: Reliability of isometric muscle testing of knee flexor
351-388. - and extensor muscles in patients with connective tissue
8. Hagberg JM. Effect of training on the decline of VO, disease. Phys Ther 1988; 68: 338-343.
max with aging. Federation Proc 1987: 46: 1830-1833. 28. Bergner M, Bobbitt RA, Carter WB et al. The Sickness
9. Cauley JA,>e&ni AM, LaPorte RE et al. The decline Impact Profile: development and final revision of a
of grip strength in the menopause: relationship to health status measure. Med Care 1981; 19: 787-805.
physical activity, estrogen use and anthropometric 29. Bergner M, Bobbitt RA, Kressel S et al. The Sickness
factors. J Chron Dis 1987; 40: 1115-1120. Impact Profile: conceptual formulation and method-
IO. Cooper C, Barker DJP, Wilkham C. Physical activity, ology for the development of a health status measure.
muscle strength, and calcium intake in fracture of the Int J Health Serv 1976; 6: 393-415.
nroximal femur in Britain. Br Med J 1988: 297: 30. Wong TC, Webster JG, Montoye HJ et al. A port-
i443-1446. able accelerometer device for measuring human energy
Il. LaPorte RE, Kuller LH, Kupfer DJ et al. An objective expenditure. IEEE Trans Biomed Eng 1981; 28:
measure of physical activity for epidemiologic research. 467-47 1.
Am J Eoidemiol 1978: 108: 161-175. 31. Washburn RA, Janney CA, Fenster J. The validity
12. Klesges-RC, Klesges’ LM, Swenson AM et al. A of objective physical activity monitoring in older
validation of two motion sensors in the prediction of individuals. Res Q Exer Sport 1990; 61: 114-117.
child and adult physical activity. Am J Epidemiol 1985; 32. Reiff GG, Montoye HJ, Remington RD er al.
122: 400-410. Assessment of physical activity by questionnaire and
13. Mattiasson-Nilo I, Sonn U, Johannesson K et al. interview. J Sports Med Phys Fitness 1967; 7: 135-142.
Domestic activities and walking in the elderly: evalu- 33. Bohrnstedt GW. Measurement. In: Rossi PA,
ation from a 30-hour heart rate recording. Aging 1990; Anderson A, Eds. Handhook of Survey Research.
2: 191-198. New York: Academic Press; 1983.
14. Carp FM, Carp A. The validity, reliability and 34. Long JS. Confirmatory Factor Analysis. Beverly Hills,
generalizability of diary data. Exp Aging Res 1981; 7: CA: Sage Publications; 1983.
281-297. 35. Frontera WR, Meredith CN, O’Reilly KP et al.
15. Sidney KH, Shephard RJ. Activity patterns of elderly Strength conditioning in older men: skeletal muscle
men and women. J Gerontol 1977; 32: 25-32. hypertrophy and improved function. J Appl Physiol
16. Washburn RA, Montoye HJ. Assessment of physical 1988; 64: 1038-1044.
activity by questionnaire: a review. Am J Epidemiol 36. Lichtenstein MJ, Shields SL, Shiavi RG ef al. Exercise
1986; 123: 563-516. and balance in aged women: a pilot controlled clinical
17. Dalsky GP, Stocke KS, Ehsani AA et al. Weight- trial. Arch Phvs Med Rehabil 1989: 70: 138-143.
bearing exercise training and lumbar bone mineral 37. Bray GA. Exercise and obesity. In:’ Bouchard C,
content in post-menopausal women. AM Intern Med Shephard RJ, Stephens T et al., Eds. Exercise, Fit-
1988; 108: 824-828. ness, and Health: a Consensus of Current Knowledge.
18. Isaacs B. Clinical and laboratory studies of falls in Champaign, IL: Human Kinetics Books; 1990:
old people: prospects for prevention. Clin Gerintr Mad 497-510.
1985; 1: 513-524. 38. Voorrips LE, Ravelli ACJ, Dongelmans PCA et al.
19. Marti B, Pekkanen J, Nissinen A et al. Association of A physical activity questionnaire for the elderly.
physical activity with coronary risk factors and physical Med Sci Sports Exer 1991; 23: 974-979.
ability: twenty year follow-up of a cohort of Finnish 39. Washburn RA, Smith KW, Goldfield SRW et a/.
men. Age Aging 1989; 18: 103-109. Reliability and physiologic correlates of the Harvard
20. Sallis JF, Haskell WL, Wood PD et al. Physical Alumni Activity Survey. J Clin Epidemiol 1991; 12:
activity assessment methodology in the five-city project. 1319-1326.
Am J Epidemiol 1985; 121: 91-106. 40. Dannenberg AL, Keller JB, Wilson PWF et al. Leisure
21. Paffenbarger RS Jr, Wing AL, Hyde RT. Physical time physical activity in the Framingham Offspring
activity as an index of heart attack risk in college study: description, seasonal variation, and risk factor
alumni. Am J Epidemiol 1978; 108: 161-17.5. correlates. Am J Epidemiol 1989; 129: 76-88.
22. Taylor HL, Jacobs DR, Schucker B et al. A question- 41. Stephens T. Jacobs DR Jr, White CC. A descriptive
naire for assessment of leisure time physical activities. epidemiology of leisure-time physical activity. Public
J Chron Dis 1978; 31: 741-755. Health Rep 1985; 100: 147-158.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: