Sedentary Behaviors and Health Outcomes Among Adults

C

Snbpdtd

FIA

itA

1

Sedentary Behaviors and Health OutcomesAmong Adults

A Systematic Review of Prospective Studies

Karin I. Proper, PhD, Amika S. Singh, PhD, Willem van Mechelen, MD, PhD,Mai J.M. Chinapaw, PhD

Context: Nowadays, people spend a substantial amount of time per day on sedentary behaviors andit is likely that the time spent sedentary will continue to rise. To date, there is no review of prospectivestudies that systematically examined the relationship between diverse sedentary behaviors andvarious health outcomes among adults.

Purpose: This review aimed to systematically review the literature as to the relationship betweensedentary behaviors and health outcomes considering the methodologic quality of the studies.

Evidence acquisition: In February 2010, a search for prospective studieswas performed in diverseelectronic databases. After inclusion, in 2010, themethodologic quality of each study was assessed. Abest-evidence synthesis was applied to draw conclusions.

Evidence synthesis: 19 studies were included, of which 14 were of high methodologic quality.Based on inconsistency in fındings among the studies and lack of high-quality prospective studies,insuffıcient evidence was concluded for body weight–related measures, CVD risk, and endometrialcancer. Further, moderate evidence for a positive relationship between the time spent sitting and therisk for type 2 diabetes was concluded. Based on three high-quality studies, there was no evidence fora relationship between sedentary behavior and mortality from cancer, but strong evidence forall-cause and CVD mortality.

Conclusions: Given the trend toward increased time in sedentary behaviors, additional prospectivestudies of high methodologic quality are recommended to clarify the causal relationships betweensedentary behavior and health outcomes. Meanwhile, evidence to date suggests that interventionsaimed at reducing sedentary behavior are needed.(Am J Prev Med 2011;40(2):174–182) © 2011 American Journal of Preventive Medicine

tiisf

iobetw(jUhs

ontextedentary behavior is a different construct thanphysical (in)activity, with both behaviors havingdifferent determinants.1 In an editorial2 of this jour-

al, in 2007, Biddlemade clear that the study of sedentaryehavior was becoming popular and he encouraged toerformmore research on sedentary behavior. In order toistinguish between light-intensity activities and seden-ary behavior, in 2008, Pate and colleagues3 provided aefınition of sedentary behavior. Following that defıni-

rom the Department of Public and Occupational Health and the EMGOnstitute for Health and Care Research, VU University Medical Center,msterdam, The NetherlandsAddress correspondence to: Karin I. Proper, PhD, VU University Med-

cal Center, Department of Public and Occupational Health, EMGO Insti-ute for Health and Care Research, Van der Boechorststraat 7, 1081 BTmsterdam, The Netherlands. E-mail: [email protected].

w0749-3797/$17.00doi: 10.1016/j.amepre.2010.10.015

74 Am J Prev Med 2011;40(2):174–182 © 2011 Amer

ion, sedentary behavior refers to activities that do notncrease energy expenditure substantially above the rest-ng level (1.0–1.5METs) and it includes activities such asleeping, sitting, lying down, watching TV, and otherorms of screen-based entertainment.Considering these examples of sedentary behaviors, it

s clear that given the increasing availability of technol-gy, people currently spend a lot of time on sedentaryehaviors and it is likely that time spent sedentary willven continue to rise. To illustrate, based on results fromhe American Time Use Survey (ATUS), it appeared thatatching TV accounted for about half of leisure timewww.bls.gov/news.release/pdf/atus.pdf). Also, manyobs nowadays involve much time spent sitting at work.sing data from a representative sample of Dutch house-olds, Jans et al.4 found that on average Dutch workerspent sitting about 7 hours per day, of which one third

ican Journal of Preventive Medicine • Published by Elsevier Inc.

sm

ldvttallsoaHst

rHbormupistecssoolsts

E

IAbPucs1sbw

itptvcsstav

sact

1

2

3

DDtmmsoitTinnttnstig

edomctwr

Proper et al / Am J Prev Med 2011;40(2):174–182 175

F

howed sedentary time,measured by accelerometer, to beade up most of the time, namely, 509 minutes per day.Becausemost adults are working, and thus aremore or

ess forced to sit continuously during a large part of theay, it is important to get insight into the potential ad-erse health effects of sedentary behaviors. In their sys-ematic review, Van Uffelen and colleagues6 examinedhe evidence on associations between occupational sittingnd health risks and concluded that there currently isimited evidence for a positive relationship. Only in theast decade has the literature increased on the role ofedentary behavior in the development of adverse healthutcomes, with studies5,7,8 suggesting an independentdverse health effect of sedentary behavior. Very recently,amer et al.9 found an independent association betweenedentary behavior, indexed by TV- and screen-basedime, and poorer health scores among adults.Few reviews10–12 have summarized the literature with

espect to the health implications of sedentary behavior.owever, those reviews focused on one type of sedentaryehavior or one health outcome, with overweight andbesity being most frequently examined. Except for theeview by Marshall and colleagues,11 which performed aeta-analysis regarding the relationship between mediase and body fatness among children and youth, therevious reviews did not perform a systematic approachn that they assessed the methodologic quality of thetudies or used a best-evidence synthesis ormeta-analysiso draw conclusions. Finally, in a review of studies on theffect of sedentary behavior on health, Pate et al.3 con-luded that most of the identifıed studies did rather mea-ure insuffıcient levels of physical activity instead of mea-uring sedentary behavior. Thus to date, the health effectsf different sedentary behaviors are still unclear. The aimf the present study was to systematically review theiterature with respect to the relationship between diverseedentary behaviors and health outcomes among adultsaking into account the methodologic quality of thetudies.

vidence Acquisition

dentification and Selection of the Literatureliterature search was conducted in several electronic

ibliographic databases, namely, PubMed, Embase,sycINFO, and the Cochrane Library. The keywordssed referred to the exposure (sedentary behavior); out-ome (health-related); and study design (longitudinal de-igns). English written publications published between989 and February 25, 2010, were applied as a limit. As aystematic review to the relationship between sedentaryehaviors and health among children and adolescents

ebruary 2011

ted to the adult population. However, for the purpose ofhis specifıc review, only studies that involved the adultopulation were included. Next to the search in elec-ronic databases, the authors’ personal databases and pre-ious reviews and references of key publications werehecked. Studies that examined the relationship betweenitting and (low) back pain were excluded for two rea-ons. First, the type of sedentary behavior was assumedo be specifıc, namely, sitting at work. Second, Chennd colleagues13 recently performed a systematic re-iew on this relationship.The titles and abstracts of all citations derived from the

earchwere screened independently by two of the currentuthors. In case of uncertainty to either include or ex-lude the study, the full paper was read. To be included inhe review, the study had to meet the following criteria:

. The study had to have a longitudinal design, eitherprospective or retrospective.. The study had to involve an adult, nonpatient popula-tion (i.e., average age �18 years).. The study had to measure sedentary behavior as de-fıned by Pate et al.3

ata Extraction and Quality Assessmentata were extracted from all studies selected with regardo the study population, follow-up duration, type andeasurement of sedentary behavior, type and measure-ent of the health outcome, statistical analysis, and re-ults. The selected studies were evaluated on their meth-dologic quality. In doing so, two of the authorsndependently scored the quality of each study accordingo a standardized set of predefıned criteria (Table 1).14–16

he list consists of 15 items, which distinguished betweennformativeness (I, n�6) and validity/precision (V/P,�9). Each quality criterion was rated as positive (1),egative (0), or unknown (?). A positive score was given ifhe publication provided an informative description ofhe criterion at issue and met the quality criterion. Aegative score was given in case of an informative de-cription, but an inadequate execution or lack of descrip-ion of the item concerned. In case of an unclear orncomplete description of the item, a question mark wasiven.The results of the scorings were compared and differ-

nceswere discussed during a consensusmeeting. If, afteriscussion, agreement could not be reached, one of thether authors was consulted to achieve the fınal judg-ent. If necessary, the fırst author of the publication wasontacted by e-mail to provide additional information aso the item that had a question mark in the scoring. Theyere given a time frame of 2 weeks, and received oneeminder. For each study, a total quality score was as-

osl

LAtga

taghhstt

T

a

b

c

d

e

I n on

176 Proper et al / Am J Prev Med 2011;40(2):174–182

n the validity/precision (V/P) criteria. A study was con-idered of high quality if the methodologic score was ateast 5 of 9.

evels of Scientific Evidencefter summarizing the included studies, it appeared thathe studies were very heterogeneous, especially with re-ard to the type and measurement of sedentary behavior

able 1. Criteria list for assessment of the methodologic

Criteriaa

Study population and participation (baseline)

1. Adequate description of source populationb

2. Adequate description of sampling frame, recruitment merecruitment, and place of recruitment (setting and geogr

3. Participation rate at baseline at least 80%, or if the nonselective (show that baseline study sample does not sigpopulation of eligible subjects)

4. Adequate description of baseline study sample (i.e., indstudy) for key characteristics (n, age, gender, SB, and h

Study attrition

5. Provision of the exact n at each follow-up measurement

6. Provision of exact information on follow-up duration

7. Response at short-term follow-up (up to 12 months) wasat baseline and response at long-term follow-up was atbaseline

8. Information on not selective nonresponse during follow-u

Data collection

9. Adequate measurement of SB: done by objective measuaccelerometry, heart rate monitoring, observation) and nreport � –; no/insufficient information � ?)

10. SB was assessed at a time prior to the measurement o

11. Adequate measurement of the health outcome: objectivhealth outcome done by trained personnel by means ofprotocol(s) of acceptable quality and not by self-report (sinsufficient information � ?)

Data analyses

12. The statistical model used was appropriatee

13. The number of cases was at least 10 times the numbervariables

14. Presentation of point estimates and measures of variab

15. No selective reporting of results

Rating of criteria: � � yes; – � no; ? � unknownAdequate � sufficient information to be able to repeat the study� is given only if adequate information is given on all items.� is given only if nonselective dropout on key characteristics (age,tables.� is given if a multivariate regression model was used., criterion on informativeness; SB, sedentary behavior; V/P, criterio

nd the health outcome. Additionally, of those studies t

hat examined the same health outcome, the statisticalnalysis varied among the studies, including the cate-orization of the independent variable (sedentary be-avior) resulting in different types of effect sizes (e.g.,azard ratio [HR]; OR; risk ratio [RR]; or beta) makingtatistical pooling impossible. Therefore, to synthesizehe methodologic quality of the studies and to be ableo draw conclusions regarding the relationship be-

lity of prospective studies14–16

I, V/P% studies meeting

the item

I 84

s, period oflocation)b

I 58

onse was notantly differ from

V/P 16

als entering theoutcome)c

I 79

I 42

I 89

east 80% of the n70% of the n at

V/P 58

easurement(s)d V/P 26

i.e.,self-report (self-

V/P 11

health outcome V/P 95

asurement of thedardizedeport � –; no/

V/P 53

V/P 100

e independent V/P 100

CI or SE) I 95

V/P 100

er, sedentary behavior, health outcomes) is reported in the text or

validity/precision

qua

thodaphic

respnific

ividuealth

at lleast

p m

res (ot by

f the

e mestanelf-r

of th

ility (

gend

ween sedentary behavior and the health outcome, a

www.ajpm-online.net

bt

ecisqw

ESTfadmptamipa

MTmhi1m

S

Swoshtsiesa

tfasmwosbfit

Spireawf2fwsioab9rtsTbwitf

Sqtbswgch

Sibd

Proper et al / Am J Prev Med 2011;40(2):174–182 177

F

est-evidence synthesis was applied,14,16 consisting ofhe following three levels:

1. Strong evidence: consistent fındings inmultiple (�2)high-quality studies;

2. Moderate evidence: consistent fındings in one high-quality study and at least one low-quality study, orconsistent fındings in multiple low-quality studies;

3. Insuffıcient evidence: only one study available or in-consistent fındings in multiple (�2) studies.

Similar to previous reviews that applied this best-vidence synthesis,14,16–18 results were considered to beonsistent when at least 75% of the studies showed resultsn the same direction, which was defıned according toignifıcance (p�0.05). If there were two or more high-uality studies, the studies of low methodologic qualityere disregarded in the evidence synthesis.

vidence Synthesisearch and Selectionhe search resulted in a total of 10,555 records (4994rom PubMed, 4564 from Embase, 457 from PsycINFO,nd 540 from Cochrane Library). After removing theuplicate publications, a total of 8424 publications re-ained. After screening the titles and abstracts, 137 fullapers were read. Of those, most were excluded becausehe study applied a cross-sectional design or the studypplied a defınition of sedentary behavior that did noteet the criterion. Finally, 19 prospective studies were

ncluded.19–37 The characteristics of these studies areresented in Appendix A (available online at www.jpm-online.net).

ethodologic Quality Assessmenthe scoring of the 19 prospective studies led to a disagree-ent of 17%. The majority of the studies (n�14) was ofigh quality. The proportion of studies meeting the qual-ty items varied considerably per item,with only 11% (2 of9 studies) scoring positive on the objective measure-ent of sedentary behavior.

edentary Behavior—Health Outcomes

edentary behavior—body weight/BMI gain. Thereere three prospective studies,20,21,37 of which two weref high quality, investigating the relationship betweenedentary behavior and body weight gain. Further, twoigh-quality studies22,31 investigated the relationship be-ween sedentary behavior and BMI gain. Of these fıvetudies, four21,22,31,37 assessed the time spent onTVview-ng, showing inconsistent results. For example, Coakleyt al.21 found a signifıcant relationship between the timepent on TV/VCR and self-reported body weight gain

ebruary 2011

ionship was found among those aged �55 years. Theseındings were supported by the studies of Jeffery et al.31

nd Crawford et al.,22 who found no signifıcant relation-hip between the time spent TV viewing and objectivelyeasured body weight/BMI gain among men andomen, respectively. In contrast, the more recent studyf Raynor et al.37 found TV viewing time to be related toelf-reported body weight gain over a 1-year period,eta(t)�0.12 (3.89), p�0.001. Based on the inconsistentındings among the prospective studies identifıed, there isnsuffıcient evidence for a longitudinal relationship be-ween sedentary behavior and body weight/BMI gain.

edentary behavior—overweight or obesity. Fourrospective studies26,30,34,35 were identifıed that exam-ned the relationship between sedentary behavior and theisk for overweight or obesity. The study of Graff-Iversent al.26 found increased ORs for overweight (BMI�27)mong women performing light, moderate, or heavyork compared to those with sedentary work; ORs variedrom 1.18 (95% CI�1.04, 1.34) to 1.67 (95% CI�1.38,.03). However, no signifıcant relationships were foundor men. Although a different cut-off point for over-eight was used (BMI�25), Meyer et al.34 did not fınd aignifıcant relationship between the time spent TV view-ng and the risk for developing overweight over a periodf 6 years. In contrast, Novak et al.35 found for both mennd women viewing more programs on TV per week toe at increased risk for developing overweight (OR�1.51,5% CI�1.05, 2.18, and OR�1.73, 95% CI�1.09, 2.76,espectively) over 14 years. The study of Hu et al.30 washe only study that investigated the relationship withelf-reported obesity using a cut-off point of BMI�30.hey found different results for the different sedentaryehaviormeasured used (Appendix A, available online atww.ajpm-online.net). Based on the inconsistent fınd-ngs among the studies, there is insuffıcient evidence forhe relationship between sedentary behavior and the riskor overweight or obesity.

edentary behavior—waist gain. Only one low-uality prospective study33 examined the relationship be-ween sedentary behavior and waist gain, both measuredy self-report. Over a period of 6 years, they found aignifıcant relationship with an increase of 20 hours pereek of viewing TV to be related to 0.30 (0.12) cm waistain (p�0.02). Based on this single study, there is insuffı-ient evidence for the relationship between sedentary be-avior and waist gain.

edentary behavior—type 2 diabetes. Two low-qual-ty prospective studies29,30 investigated the relationshipetween sedentary behavior and the incidence of type 2iabetes in men and women, respectively. The study of

tm2pdstshdplss

StaCtfhstesfh(oimtt

bafIfhtcosv

Shtntrt

cedlTbit

SiatItrplivtofor

fmtrttsfaif

ttCt

DTtscs1Tp

178 Proper et al / Am J Prev Med 2011;40(2):174–182

ime spent TV viewing and type 2 diabetes among 37,918ale health professionals (p-value for trend 0.02). In003, Hu et al.30 reported the relationship over a 6-yeareriod between various sedentary behaviors and inci-ence of diabetes type 2 in 68,497 women. They foundignifıcant positive relationships for TV viewing, sit-ing at work or away from home or driving, and otheritting at home with those viewing TV for at least 6ours per week being at increased risk of developingiabetes type 2 compared to those viewing 0–1 hourser week. Based on the consistent fındings of these twoow-quality studies, there is moderate evidence for aignifıcant positive relationship between the timepent sitting and the risk for type 2 diabetes.

edentary behavior—cardiovascular disease risk fac-ors. Four high-quality prospective studies19,24,25,28 ex-mined the relationship between sedentary behavior andVD risk factors. Beunza et al.19 studied different seden-ary behaviors, TV viewing, PC use, and driving, andound no signifıcant relationship for any sedentary be-avior and self-reported incidence of hypertension. Thetudy of Fung et al.25 investigated the relationship be-ween the weekly time spent TV/VCR viewing and sev-ral biomarkers of CVD risk, namely, cholesterol mea-ures (e.g., total, LDL, HDL, triglycerides); leptin;ıbrinogen; insulin; C-peptide; and HbA1c among maleealth professionals. With the exception of leptinp�0.05), no signifıcant relationships were found for anyf the CVD biomarkers. Finally, the remaining two stud-es24,28 examined the relationship between objectivelyeasured sedentary behavior and insulin resistance, with

he study of Ekelund et al.24 also investigating the rela-ionship with self-reported TV/video viewing.Although a trend was observed for the relationshipetween sedentary time and fasting insulin (p�0.07) over1-year follow-up period, no signifıcant relationship wasound with the homeostatis model assessment (HOMA-R) score, nor for the time spent TV/video viewing andasting insulin orHOMA-IR score.24 Helmerhorst et al.28

ad a longer follow-up time, namely, 5.6 years, and foundhe objectively measured time spent sedentary signifı-antly related to insulin resistance. Based on the fındingsf the studies identifıed, there is insuffıcient evidence for aignifıcant relationship between sedentary behavior andarious CVD risk factors.

edentary behavior—endometrial cancer. Twoigh-quality studies27,36 investigated the relationship be-ween sedentary behavior and a specifıc type of cancer,amely, endometrial cancer. Patel et al.36 measured theime spent sitting outside work and found no signifıcantelationship with endometrial cancer. This was in con-

ant trend for the time spent sitting and the incidence ofndometrial cancer, with those sitting at least 5 hours peray being at increased risk. However, this signifıcant re-ationship was not observed for the time spent viewingV/videos (p�0.26). Based on the inconsistencies foundetween and within the two studies identifıed, there isnsuffıcient evidence for the relationship between seden-ary behavior and endometrial cancer.

edentary behavior—mortality. Three studies23,26,32

nvestigated the relationship between sedentary behaviorndmortality. All three studies applied a different seden-ary behavior measure. The low-quality study of Graff-versen et al.26 assessed the relationship between seden-ary work and mortality and did not fınd a signifıcantelationship with the exception of heavy occupationalhysical activity, which was related signifıcantly to aower risk of mortality. The second prospective studydentifıed, which was of high methodologic quality,32 in-estigated the relationship between sitting time andmor-ality among 17,013 Canadians aged 18–90 years. Theutcomes included mortality from all causes, mortalityrom CVD, mortality from cancer, and mortality fromther diseases based on the Canadian mortalityegistrations.No signifıcant relationship was found for mortality

rom cancer. However, for all-cause, CVD, and otherortality, there were signifıcant relationships between

he time spent sitting and mortality, with an increasedisk of those who reported sitting for at least three quar-ers of their time compared to those sitting for none of theime.32 These fındings were confırmed by themost recenttudy,23 which also was rated as of high quality. Theyound that each 1-hour increment in TV/videos viewingppeared to be signifıcantly related with all-causemortal-ty and CVD mortality; no signifıcant relationship wasound for mortality from cancer.23

Based on the fındings of the two high-quality studies,here is strong evidence for a relationship between seden-ary behavior and mortality from all causes and fromVD, but no evidence for the relationship between seden-ary behavior and mortality from cancer.

iscussionhe present review aimed to systematically summarizehe literature with regard to the relationship betweenedentary behavior and health outcomes, taking into ac-ount themethodologic quality of the studies. Despite thetart date of the literature being 1989, the majority (12 of9) of the studies included were published after 2005.his indicates that the topic of sedentary behavior as arobable independent predictor for certain health out-

www.ajpm-online.net

edtfsm

lpwftUcboptfts

tabrwwsstfrmstto

CTtlhhtpipo

lp

otopntirwrts

MAqw

srwianamipsmratir

dfotsttiftArdo

Proper et al / Am J Prev Med 2011;40(2):174–182 179

F

rature. Based on the studies identifıed, moderate evi-ence for a signifıcant positive relationship between theime spent sitting and incidence of type 2 diabetes wasound. Further, strong evidence was found for a relation-hip between sedentary behavior and all-cause and CVDortality, but not for mortality from cancer.However, in contrast to some suggestions of previous

iterature,10,38 the conclusion of strong evidence for aositive relationship between sedentary behavior andeight outcomes could not be confırmed. Despite a dif-erent sitting-time measure (i.e., occupational sitting),his review’s conclusions were in line with those of Vanffelen et al.6 They also found mixed results among theross-sectional studies, with the prospective studies noteing able to confırm a positive relationship betweenccupational sitting and BMI. The conclusion of theresent review of insuffıcient evidence was partly due tohe lack of high-quality prospective studies. In particular,or waist gain, there was only one prospective study. Forhe other body weight outcomes, there were multipletudies, but fındings from those studies were mixed.These inconsistencies may be caused by differences in

he methodology used in the studies. For example, itppeared that the two studies that objectively measuredody weight (and height) did not detect a signifıcantelationship between TV viewing and weight/BMI gain,hereas the remaining three studies that measured bodyeight by self-report did fınd a positive relationship withedentary behavior. This pattern was also true for thetudies that examined the risk for overweight and obesity:hose measuring overweight or obesity by self-reportound signifıcant positive relationships. These results areemarkable and emphasize the importance of objectivelyeasured body weight and height, as it is known thatelf-reportedweight andheight are less accurate and validhan objectively measured body weight and height, andherefore theymay lead tomisclassifıcation of overweightr obesity.39

omparison Findings with Previous Reviewso the best of our knowledge, this review is the fırst onehat attempted to summarize the literature regarding theongitudinal relationship between diverse sedentary be-aviors, excluding occupational sitting, and diverseealth outcomes. Despite differences inmethodology be-ween this review and previous ones, results can be com-ared. Based on the fındings of the cross-sectional stud-es, the review of Williams et al.38 concluded generallyositive associations between TV viewing and indicatorsf overweight.However, inconsistent results were found based on the

ongitudinal studies included, thereby supporting the

ebruary 2011

utcomes, they also found similar results based on longi-udinal studies. For example, the positive associationsbserved between TV viewing and lipoproteins or bloodressure/hypertension from cross-sectional studies wereot confırmed by the few longitudinal studies in their andhe present review. It thus seems that the evidence to dates insuffıcient to conclude that there is a longitudinalelationship between sedentary behavior and bodyeight and certain other health outcomes, such as CVDisk factors. In this context, it is worthwhile to emphasizehat the insuffıcient evidence is partly due to lack oftudies with a high level of methodologic quality.

ethodologic Issueslthough most studies were considered to be of highuality, there are a few methodologic issues that areorth mentioning.First, the large majority of the studies (17 of 19) mea-

ured sedentary behavior by self-report, with differentecall periods used (e.g., 12 months, a typical day, or 1eek). From a study40 that tested the reliability and valid-ty of the instruments measuring sedentary behavior, itppeared that the validity of self-report instruments foron-occupational sedentary behaviors varies consider-bly. The authors therefore recommended the develop-ent of a reliable and valid self-report instrument cover-

ng diverse sedentary behaviors. Because a substantialroportion of the adult population is working, and thuspends considerable (sitting) time at work, the develop-ent of a self-report instrument that measures the entireange of sedentary activities, including time spent sittingt work, seems necessary. In this context, it is worthwhileo consider the possible influence on this review’s fınd-ngs, which mainly reflect the relationship between self-eported sedentary behavior and health outcomes.Another methodologic issue that needs to be ad-ressed, concerns the follow-up period. The length of theollow-upmeasurement was not included in themethod-logic quality criteria list, but may explain the inconsis-encies in the results among studies. For example, twotudies24,28 investigated the relationship between objec-ively measured sedentary behavior and insulin resis-ance. The study of Ekelund et al.24 did not fınd a signif-cant relationship after 1 year, but with a medianollow-up of 5.6 years, Helmerhorst et al.28 found seden-ary time to be signifıcantly related to insulin resistance.lthough this was not valid to all health outcomes, it isecommended for future studies to apply a follow-upuration of at least a few years, especially for those healthutcomes for which it is plausible that effects will occur

MTtptorcFitvsid

bitdbotisiorBbb

pofdmpbsiapmulpf

STTl

osoorbasocfAd

trbpqnqtwtt

ebiaodidacactpoas

CTdtfaBi

180 Proper et al / Am J Prev Med 2011;40(2):174–182

echanisms for the Observed Relationshipshere are several theories explaining the relationship be-ween sedentary behavior and health outcomes. In theirublication, Williams et al.38 provided a framework forhe potential relationship betweenTV viewing and healthutcomes. According to their framework, TV viewing iselated indirectly to overweight and obesity through in-reased energy intake and decreased energy expenditure.urther, these two energy-balance–related behaviors cannfluence health outcomes either directly or indirectly,hrough overweight and obesity. Findings from this re-iew showed various results for the independent relation-hip between sedentary behaviors and weight outcomes,ncluding overweight and obesity, and moderate evi-ence for a relationship with type 2 diabetes.These results suggest that there is a direct relationshipetween sedentary time and type 2 diabetes. It is thereforenteresting to conduct a mediation analysis investigatinghe relationship between sedentary behavior and type 2iabetes and to examine the contribution of energy-alance–related behaviors (diet and physical activity) andverweight and obesity in this relationship. Consideringhe role of body fatness on inflammatory factors,41 it alsos recommended to include body fatness as an outcome intudies regarding the relationship with sedentary behav-or. Next to its possible independent effect on healthutcomes, fatness may be an important mediator in theelationship between sedentary behavior and health.ased on studies of the role of fatness in the relationshipetween fıtness and inflammatory pathways, it would alsoe interesting to examine eventual gender differences.Further, a mediation analysis may be valuable to ex-lain other relationships. For example, how can the lackf evidence for the relationship with some CVD riskactors be explained, in combination with the strong evi-ence for a relationship between sedentary behavior andortality from CVD? Another mechanism that has beenroposed to explain the relationship between sedentaryehavior andhealth outcomes, include changes in cardiactroke volume and output.32,42 In addition, based on an-mal studies,12,43 a physiologic link with sedentary timend lipoprotein lipase activity has been found. Theresent review’s conclusion of strong evidence for CVDortality supports those fındings.However, there are stillncertainties in the underlying behaviors and (physio-ogic) mechanisms that likely explain the health effects ofrolonged sedentary time. It is thus a topic that needsurther research.

trengths and Limitationshere are several strengths andweaknesses of this review.he strengths include the systematic approach in the

logic quality of each study, and the use of a best-evidenceystem based on which conclusions were drawn. More-ver, a search in diverse databases was performed andnly prospective studies were included. Because the cur-ent review was limited to prospective studies, it is possi-le to report on longitudinal relationships and not onssociations only. Although the studies applied differenttatistical models, they all analyzed the change in healthutcome in particular. The studies either examined thehange in health by the difference between baseline andollow-up values (delta scores) or by applying anNCOVA inwhich the value at follow-upwas the depen-ent value and adjusting for the baseline value.Aweakness of this review however is the subjectivity of

he rating. Although the quality rating was done by twoeviewers independently, the scoring may be subject toias. To minimize this, the authors were contacted torovide information about the question marks of theuality criterion of concern. However, some authors didot reply, resulting in the study being scored as low-uality. To illustrate, if the author(s) had replied posi-ively on one validity and precision item, the study qualityould change from low to high and consequently lead tohe conclusion of strong evidence for a relationship withype 2 diabetes (instead of moderate evidence).This example thus highlights the sensitivity of the best-

vidence system used. Thus, a meta-analysis would haveeen preferred. However, as described earlier, the studiesdentifıed were too heterogeneous for such a quantitativenalysis. It is thus worthwhile to consider that the heter-geneity inmethods among the studies, such as the use ofifferent measurement instruments of sedentary behav-or or health outcome and the variation in follow-upuration, may have led to the inconsistent fındingsmong the studies. Nevertheless, we believe the studiesan be compared in a best-evidence synthesis, especiallys conclusions were drawn for each specifıc health out-ome. Also, except for two studies, all measured seden-ary behavior by self-reports and generally showed com-arable fındings. Finally, by presenting all characteristicsf each study and its results (AppendixA, available onlinet www.ajpm-online.net), readers may interpret the re-ults themselves.

onclusionhis review of prospective studies showed moderate evi-ence for an independent relationship between sedentaryime and type 2 diabetes. In addition, strong evidencewasound for sedentary behavior to be related to all-causend CVD mortality, but not for mortality from cancer.ecause ofmixed results and the lack of prospective stud-es, there was insuffıcient evidence for a relationship be-

www.ajpm-online.net

fiatclrso

Np

R

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

Proper et al / Am J Prev Med 2011;40(2):174–182 181

F

actors. Given the increasing modern technology with anncreasing availability of computers, TV,DVD, and otherutomatic devices, the trend of spending time in seden-ary behaviors involving prolonged sitting is likely toontinue. Considering this trend, the increasing preva-ence of obesity and type 2 diabetes, and the presentesults, interventions aimed at the reduction of the timepent on sedentary behavior together with the promotionf physical activity should be developed.

o fınancial disclosures were reported by the authors of thisaper.

eferences1. Owen N, Leslie E, Salmon J, Fotheringham MJ. Environmental deter-

minants of physical activity and sedentary behaviour. Exerc Sport SciRev 2000;28:153–8.

2. Biddle SJH. Sedentary behavior. Am J Prev Med 2007;33(6):502–4.3. Pate RR, O’Neill JR, Lobelo F. The evolving defınition of “sedentary.”

Exerc Sport Sci Rev 2008;36(4):173–8.4. Jans MP, Proper KI, Hildebrandt VH. Sedentary behavior in Dutch

workers. Differences between occupations and business sectors. Am JPrev Med 2007;33(6):450–4.

5. Peters T,Moore SC, Xiang YB, et al. Accelerometer-measured physicalactivity in Chinese adults. Am J Prev Med 2010;38(6):583–91.

6. Van Uffelen JGZ, Wong J, Chau JY, et al. Occupational sitting andhealth risks. A systematic review. Am J Prev Med 2010;39(4):379–88.

7. DunstanDW, Salmon J, OwenN, et al. Associations of TV viewing andphysical activity with the metabolic syndrome in Australian adults.Diabetologia 2005;25:2254–61.

8. Mota J, Ribeiro J, Santos MP, Gomes H. Obesity, physical activity,computer use, and TV viewing in Portuguese adolescents. PediatrExerc Sci 2006;18(1):113–21.

9. Hamer M, Stamatakis E, Mishra GD. Television- and screen-basedactivity and mental well-being in adults. Am J Prev Med2010;38(4):375–80.

0. Foster JA, Gore SA, Smith West D. Altering TV viewing habits: anunexplored strategy for adult obesity intervention? Am J Health Behav2006;30(1):3–14.

1. Marshall SJ, Biddle SJ, Gorely T, Cameron N, Murdey I. Relationshipsbetween media use, body fatness and physical activity in children andyouth: a meta-analysis. Int J Obes Relat Metab Disord 2004;28(10):1238–46.

2. HamiltonMT, Hamilton DG, Zderic TW. Role of low energy expendi-ture and sitting in obesity, metabolic syndrome, type 2 diabetes, andcardiovascular disease. Diabetes 2007;56:2655–67.

3. Chen S-M, Lu M-F, Cook J, Bass S, Lo SK. Sedentary lifestyle as a riskfactor for low back pain: a systematic review. Int Arch Occup EnvironHealth 2009;82(7):797–806.

4. Hoogendoorn WE, van Poppel MNM, Bongers PM, Koes BW, BouterLM. Systematic review of psychosocial factors at work and private lifeas risk factors for back pain. Spine 2000;25:2114–25.

5. Hayden JA, Cote P, Bombardier C. Evaluation of the quality of prog-nosis studies in systematic reviews. Ann Intern Med 2006;144:427–37.

6. Singh AS, Mulder C, Twisk JW, van Mechelen W, Chinapaw MJ.Tracking of childhood overweight into adulthood: a systematic reviewof the literature. Obes Rev 2008;9(5):474–88.

7. Proper KI, Koning M, Van der Beek AJ, Hildebrandt VH, Bosscher R,

ebruary 2011

grams on physical activity, physical fıtness and health. Clin J SportMed2003;13(2):106–17.

8. Proper KI, Staal JB, Hildebrandt VH, Van der Beek AJ, Van MechelenW. Effectiveness of physical activity programs at worksites with respectto work-related outcomes. Scand J Work Environ Health 2002;28(2):75–84.

9. Beunza JJ, Martínez-González MA, Ebrahim S, et al. Sedentary behav-iors and the risk of incident hypertension: the SUN Cohort. Am JHypertens 2007;20(11):1156–62.

0. Blanck HM, McCullough ML, Patel AV, et al. Sedentary behavior,recreational physical activity, and 7-year weight gain among post-menopausal U.S. women. Obesity (Silver Spring) 2007;15(6):1578–88.

1. Coakley EH, Rimm EB, Colditz G, Kawachi I, Willett W. Predictors ofweight change inmen: results from theHealth Professionals Follow-upStudy. Int J Obes Relat Metab Disord 1998;22(2):89–96.

2. Crawford DA, Jeffery RW, French SA. Television viewing, physicalinactivity and obesity. Int J Obes Relat Metab Disord 1999;23(4):437–40.

3. Dunstan DW, Barr ELM, Healy GN, et al. Television viewing andmortality: The Australian diabetes, obesity and lifestyle study (Aus-diab). Circulation 2010;3:384–91.

4. Ekelund U, Brage S, Griffın SJ, Wareham NJ; ProActive UK ResearchGroup. Objectively measured moderate- and vigorous-intensity phys-ical activity but not sedentary time predicts insulin resistance in high-risk individuals. Diabetes Care 2009;32(6):1081–6.

5. Fung TT, Hu FB, Yu J, et al. Leisure-time physical activity, televisionwatching, and plasma biomarkers of obesity and cardiovascular diseaserisk. Am J Epidemiol 2000;152(12):1171–8.

6. Graff-Iversen S, Selmer R, Sørensen M, Skurtveit S. Occupationalphysical activity, overweight, and mortality: a follow-up study of47,405 Norwegian women and men. Res Q Exerc Sport 2007;78(3):151–61.

7. Gierach GL, Chang SC, Brinton LA, et al. Physical activity, sedentarybehavior, and endometrial cancer risk in the NIH-AARP Diet andHealth Study. Int J Cancer 2009;124(9):2139–47.

8. Helmerhorst HJF, Wijndaele K, Brage S, Wareham NJ, Ekelund U.Objectively measured sedentary time may predict insulin resistanceindependent of moderate- and vigorous-intensity physical activity.Diabetes 2009;58(8):1776–9.

9. Hu FB, Leitzmann MF, Stampfer MJ, Colditz GA, Willett WC, RimmEB. Physical activity and television watching in relation to risk for type2 diabetes mellitus in men. Arch Intern Med 2001;161(12):1542–8.

0. Hu FB, Li TY, Colditz GA, Willett WC, Manson JE. Television watch-ing and other sedentary behaviors in relation to risk of obesity and type2 diabetes mellitus in women. JAMA 2003;289(14):1785–91.

1. Jeffery RW, French SA. Epidemic obesity in the U.S.: are fast foodsand television viewing contributing? Am J Public Health 1998;88(2):277–80.

2. Katzmarzyk PT, Church TS, Craig CL, Bouchard C. Sitting time andmortality from all causes, cardiovascular disease, and cancer. Med SciSports Exerc 2009;41(5):998–1005.

3. Koh-Banerjee P, Chu NF, Spiegelman D, et al. Prospective study of theassociation of changes in dietary intake, physical activity, alcohol con-sumption, and smokingwith 9-y gain inwaist circumference among 16587 U.S. men. Am J Clin Nutr 2003;78(4):719–27.

4. Meyer AM, Evenson KR, Couper DJ, Stevens J, Pereria MA, Heiss G.Television, physical activity, diet, and body weight status: the ARICcohort. Int J Behav Nutr Phys Act 2008;17(5):68.

5. Novak M, Ahlgren C, Hammarström A. A life-course approach inexplaining social inequity in obesity among young adult men andwomen. Int J Obes (Lond) 2006;30(1):191–200.

6. Patel AV, Feigelson HS, Talbot JT, et al. The role of body weight inthe relationship between physical activity and endometrial cancer:results from a large cohort of U.S. women. Int J Cancer 2008;

3

3

3

4

4

4

4

A

S

S

182 Proper et al / Am J Prev Med 2011;40(2):174–182

7. Raynor DA, Phelan S, Hill JO, Wing RR. Television viewing and long-term weight maintenance: results from the National Weight ControlRegistry. Obesity (Silver Spring) 2006;14(10):1816–24.

8. Williams DM, Raynor HA, Ciccolo JT. A review of TV viewing and itsassociation with health outcomes in adults. Am J Lifestyle Med2008;2:250–9.

9. Nyholm M, Gullberg B, Merlo J, et al. The validity of obesity based onself-reported weight and height: Implications for population studies.Obesity 2007;15(1):197–208.

0. Clark BK, Sugiyama T, Healy GN, Salmon J, Dunstan DW, Owen N.Validity and reliability ofmeasures of television viewing time and othernon-occupational sedentary behaviour of adults: a review. Obes Rev2009;10(1):7–16.

1. HamerM. The relative influence of fıtness and fatness on inflammatory

2. Saltin B, Blomqvist G, Mitchell JH, Johnson RL Jr, Wildenthal K,Chapman CB. Response to exercise after bed rest and after training.Circulation 1968;38:71–8.

3. Hamilton MT, Healy GN, Dunstan DW, Zderic TW, Owen N. Toolittle exercise and too much sitting: inactivity physiology and the needfor new recommendations on sedentary behaviour. Curr CardiovascRisk Rep 2008;2:292–8.

ppendix

upplementary data

upplementary data associatedwith this article can be found, in the

Did you know?The AJPM online archive includes issues back to 1998.

Visit www.ajpm-online.net today!

www.ajpm-online.net