PMID: 39165228 1 | IntroductionRobust evidence has consistently demonstrated that physical fit-ness is a strong predictor of survival in adults, with poor fitnesslevels associated with worse prognosis [1]. While this is undis-putable for aerobic fitness (commonly termed cardiorespiratoryfitness) [1, 2], recent studies have suggested that muscle strengthand balance, some of the non-aerobic components of physicalfitness may also be predictors of all-cause mortality in middle-aged and older subjects [3–6].We previously found a strong association between the perfor-mance on the sitting-rising test (SRT), a simple functional taskassessing the ability to sit and rise from the floor using a 0–10scale [4], and survival. Since we have also found a significantassociation between the SRT results and body flexibility [7], itcould be relevant to specifically investigate the relationship be-tween flexibility and mortality.Flexibility can be defined as the maximal physiological passiverange of a given joint movement [8]. Range of motion (ROM)considerably varies for the different joint movements, and it iscommon to observe relatively high and low ranges of motion inthe same individual [9]. Although measurements of joint motionand flexibility have been obtained for decades, utilizing linear(e.g., sit-and-reach, back scratch and hand-behind-back reach)[10, 11], angular (e.g., goniometry) [12, 13] and dimensionlessscales (e.g., Beighton) [14], or estimated by questionnaires [15],in comparison with aerobic and muscle strength fitness, thereare fewer studies on flexibility. Indeed, as recently pointed out inan editorial [16], flexibility exercises are often recommended butflexibility is rarely evaluated. In this context, few studies haveexplored the associations between joint mobility and selectedhealth outcomes [17, 18]. To the best of our knowledge, no pub-lished evidence currently exists on the association between bodyflexibility, as assessed in several joint movements, and mortal-ity risk.As part of our ongoing research in the CLINIMEX (ExerciseMedicine Clinic), Rio de Janeiro, Brazil, we investigated the rela-tionship between body flexibility and mortality. This study rep-resents a natural progression of our research focus on the impactof fitness in health outcomes and aims to contribute valuableinsights into the potential impact of body flexibility on longevity.2 | Materials and Methods2.1 | SampleThis study utilized data from the CLINIMEX Exercise open co-hort. Between March 1994 and February 2023, a total of 9055non-athlete individuals [19], voluntarily underwent a medical-functional evaluation to assess their health and levels of physi-cal fitness and to obtain advice regarding exercise and/or sportspractice. Most individuals were Caucasians and most belongedto upper Brazilian's socioeconomic and education strata. Mostwere referred by their attending physicians, while some wereself-referred for the evaluations. For individuals who underwentmultiple evaluations, only data from their initial (baseline) eval-uation were utilized for this study.After excluding 5916 individuals for several reasons (see datasetflowchart in Figure S1 of Appendix S1), the final study popu-lation reflected a convenience sample [20] of 3139 individuals(2087 men/1052 women) aged between 46 and 65 years, withavailable data and/or results for censored vital status, levels ofbody flexibility, body mass index (BMI) and, health status infor-mation—major relevant diseases and/or regular use of relevantdrugs. The final sample included individuals (90% of the initialage-selected sample) who agreed to be evaluated and withoutlocomotor or musculoskeletal complaints or clinical restrictionsfor maximal passive ROM that could preclude or impair thebody flexibility evaluation.2.2 | Flexibility Assessment (Flexitest)All evaluations throughout the nearly 29-year period were con-ducted consistently by a team of five highly skilled sports andexercise physicians adequately trained to obtain valid results inusing the assessment tool. The CLINIMEX evaluation protocolencompassed various components as detailed elsewhere [21].The Flexitest was formally proposed in the mid-80s [22] as acomprehensive body flexibility assessment tool encompass-ing 20 body joint movements, measuring ROM in seven setsof joints (number of movements): ankle (two), knee (two), hip(four), trunk (three), wrist (two), elbow (two), and shoulder (five)(see Table 1 for a kinesiological description of the 20 joint move-ments). A complete description of the Flexitest is available in apreviously published book [8].Briefly, for all movements, there are five scores, graded from0 to 4 in a crescent mobility order, to visually rate flexibilitymeasurements. The measure is taken passively, through a slowand gradual performance of the movement until the maximumrange is attained. Once the maximum amplitude is reached, it iscompared with standardized evaluation maps (Figure 1 in thetext and Figures S2–S21 in Appendix S1). A numerical score isascribed when the amplitude reached equals the correspondingone in the standardized evaluation map. For example, when theamplitude reaches position 2 on the map, score 2 is given untilthe maximum ROM reaches the level of score 3. There are nofractional or intermediate values, nor does the scoring considerthe closest level of the scale on the comparison between the ROMobtained by the evaluator and standardized evaluation maps.Therefore, only upon reaching the specified level for a score isthat score given; otherwise, even if the ROM in the movementis closer to the following score, it is rated according to the lowerscore (Figure S22 in Appendix S1 for an example of evaluationin the movement XVII of Flexitest).The distribution of the 0–4 scores tends to follow a bell-shapeddistribution with a central tendency to a score 2 in the 20 move-ments evaluated when considering individuals from childhoodto seniority, with scores 1 and 3 being less common and the ex-treme scores, 0 and 4, being quite rare and often considered assigns of hypo- or hypermobility, respectively. As the same score'sdistribution tends to be observed in all the 20 movements, it isvalid to add the scores of all 20 movements to obtain a body flex-ibility or overall joint mobility index, called the Flexindex, thatranges from 0 to 80.3 of 11As a routine, the Flexitest is administered exclusively on theright side of the body. To ensure standardization, no warm-upactivities are taken before evaluation. Adjustments were madeto the sequence and body positions of the measurements to ac-commodate the administration of the Flexitest in various set-tings, such as an examining room or bed and for individuals ofall ages and fitness levels. The duration of the Flexitest applica-tion can vary but it typically takes 3–4 min.2.3 | Additional CovariatesAdditional covariates considered for this study were: age (years),sex, height (measured in 0.1 cm), weight (measured in 0.1 kg),BMI (kg/m 2), and a simplified health status indicator. Forthe latter, individuals were classified as healthy or unhealthy.Unhealthy individuals fulfilled one or more of the followingcriteria: (a) history or diagnosis of any major cardiovascularor respiratory disease or cancer or diabetes, (b) the presence ofabnormal cardiovascular or respiratory symptoms and signs,(c) the presence of dyslipidemia or obesity, (d) to be in regularuse of medications for major chronic disease; however, the useof medications for anxiety, depression, insomnia, erectile dys-function, hormone replacement therapy and vitamins and/orsupplements were deemed acceptable and not included in theunhealthy group. On the other hand, for example, individualswith chronic obstructive pulmonary disease or cardiac valvediseases or cancer, were classified as unhealthy, even if someclassical risk factors such as hypertension, dyslipidemia or obe-sity were not present.2.4 | Mortality OutcomesVital status—alive or died and main cause of death accord-ing to ICD-10, was last censored on October 31st, 2022, fromthe official records of the Health Secretary of Rio de JaneiroState. Rather than analyzing all-cause mortality, we chose torestrict vital data analysis to natural and non- COVID-19 mor-tality. This approach was chosen to eliminate the potential biasin analyzing deaths that were either atypical or unrelated tophysical fitness. Indeed, the COVID-19 pandemic severely im-pacted the Brazilian population, particularly those residing inthe Copacabana area of Rio de Janeiro city, which includes mostindividuals in the CLINIMEX Exercise cohort [21]. During thepeak months of the pandemic, COVID-19 was the main cause ofdeath within our cohort. Additionally, the detailed examinationTABLE 1 | Flexitest: Kinesiological description of the 20 jointmovements.Movementnumber Kinesiological descriptionI Ankle dorsiflexionII Ankle plantar flexionIII Knee flexionIV Knee extensionV Hip flexionVI Hip extensionVII Hip adductionVIII Hip abductionIX Trunk flexionX Trunk extensionXI Trunk lateral flexionXII Wrist flexionXIII Wrist extensionXIV Elbow flexionXV Elbow extensionXVI Shoulder posterior adductionat 180° of abductionXVII Shoulder extension andposterior adductionXVIII Shoulder posterior extensionXIX Shoulder lateral rotationat 90° of abduction aXX Shoulder medial rotationat 90° of abduction aa At 90° elbow flexion.FIGURE 1 | Example of Flexitest's evaluation map—movement XX.4 of 11 Scandinavian Journal of Medicine & Science in Sports, 2024of individual causes of death revealed a relatively high numberof deaths attributed to external causes such as homicides, trafficaccidents, and suicides, which were likely unrelated to physicalfitness.Causes of death, after excluding COVID-19 (ICD-10 codeB342) and external causes (ICD-10 codes V/W/X/Y), werecategorized into four major groups: cardiovascular, cancer,respiratory diseases, and others. The duration of follow-upwas expressed in days, calculated as the time between theevaluation and last censored dates. All individuals read andsigned a consent form allowing the evaluation to be under-taken and the results obtained to be used in de-identified formfor research purposes. Both the evaluation and research pro-tocols were formally registered with relevant governmentalbodies and approved by an external research ethics commit-tee—CONEP/Brazil CAAE: 40122320.8.0000.9433; Report #5.721.476.2.5 | Statistical AnalysisSex-specific results were reported to account for differences inflexibility between men and women, as well as between survi-vors and non-survivors. Baseline characteristics were presentedas mean ± standard deviation (SD) with minimum and maxi-mum values for continuous variables and count (percentage) forcategorical variables. Depending on the type of variable beinganalyzed, chi-square or t-tests were used to compare sex andsurvival status between men and women. A two-way ANOVA(sex and age-group) with multiple comparisons was conductedto assess the differences in the Flexindex values between menand women within four distinct 5-year age intervals: 46–50,51–55, 56–60, and 61–65 years.Results were divided into three groups, using percentiles 10 and90 as cutoffs. Kaplan–Meier curves were constructed for theproposed range of percentiles and log-rank tests were used toanalyze survival times according to Flexindex. Cox proportionalhazard regression models were then used to estimate hazard ra-tios (HRs) with 95% confidence intervals (CIs) for the associa-tion between Flexindex and mortality risk, separately by sex. Toassess the independence of the association in men and women,HRs were calculated and progressively adjusted using four mod-els: unadjusted (model 1), age (model 2), plus BMI (model 3), andplus health status (model 4). The covariates were selected basedon the observed associations with the Flexindex using the avail-able data. All statistical analyses were conducted using Stataversion MP 16 (Stata Corp, United States) and/or Prism 10.2.3.(Graphpad, United States), and a p < 0.05 as a criterion for statis-tical significance.3 | ResultsTable 2 displays six columns of results: men (all), men (survi-vors), men (non-survivors), women (all), women (survivors),and women (non-survivors). Of the 3139 study participants(mean age 55 years)—2087 (66.5%) were men and 1052 (33.5%)were women—with no age difference by sex (p = 0.079). Duringfollow-up time (median [interquartile range]) (men: 4457[2888–6620] and women: 4366 [2768–6927] days; p = 0.581),302 deaths (9.6%) occurred, with a higher death rate in men thanwomen, 10.7% versus 7.4% (p = 0.003). After excluding 14 and 26deaths, respectively, due to external causes and COVID-19, wefound that causes of death varied according to sex (p = 0.001).Cardiovascular diseases and cancer were the most common un-derlying causes of death in men (43.3%) and women (37.2%), re-spectively. As is common in a clinical convenience sample, mostof men and women were classified as unhealthy, with a slightlyhigher percentage of healthy women compared to men (19.2 vs.15.7%, p = 0.013). Clinical variables were reported separately bysex, indicating those presenting with diagnosed coronary arterydisease, arterial hypertension, dyslipidemia, obesity, and diabe-tes mellitus, with higher percentages for all of these variables inmen (p < 0.01).Regarding anthropometric data, there were significant differ-ences between men and women. On average, men were 13 cmtaller than women (p < 0.001). Mean weight was also differ-ent, 85.5 kg for men and 68.6 kg for women (p < 0.001). MeanBMI values were slightly higher in men compared to women,27.9 and 26.3 kg/m 2 , respectively (p < 0.001). Among men,those who died tended to be older (3 years) and slightly shorter(p = 0.011), with a minimal difference of 1.4 kg in weight(p = 0.194) and a modestly higher BMI of 0.8 kg/m 2 (p = 0.007)compared to the men who survived. For women, significantdifferences were observed in age and all anthropometric com-parisons between survivors and non-survivors. Particularlynoteworthy were 6.2 kg heavier (p < 0.001) and a 2.9 kg/m 2higher BMIs among women who did not survive (p < 0.001)(Table 2).In the assessment of maximal passive ROM for all 20 jointmovements measured by the Flexitest, women demonstratedsignificantly higher scores compared to men (p < 0.001).Considering the total of 62 780 movements evaluated bythe Flexitest, only 627 (1.0%) received a score of 4, how-ever, there was a greater prevalence in women, comprising531 (0.8%) versus men—96 (0.2%) (p < 0.001). A 35% highervalue for Flexindex (mean ± SD) was observed in women—41.1 ± 9.4—as compared with men—30.5 ± 8.7—(p < 0.001).Figure 2 illustrates the Flexindex for each sex across four 5-year age intervals, consistently indicating a higher Flexindexfor women (p < 0.001) and a similar decline with age in bothmen and women (p < 0.001). When comparing survivors andnon- survivors within each sex, Flexindex was higher amongthose who survived, with men scoring 30.9 ± 8.6 compared to27.7 ± 9.1 (p < 0.001) and women scoring 41.3 ± 9.4 comparedto 37.7 ± 9.6 (p = 0.001) (Figure 3).By determining the deciles of Flexindex and calculating thecorresponding death rates in men and women, a clear gradedand inverse relationship between body flexibility and mortal-ity risk was observed. This allowed for an appropriate datamodeling approach in the survival analysis. Examining theupper extremes of the Flexindex distribution, specificallythe 2.3% (48 individuals) and 5.2% (52 individuals) with thehighest values, corresponding to Flexindex cutoff values of49 in men and 56 in women, only one death was recorded inthese 100 men and women during a mean follow-up period of≥12 years.5 of 11TABLE 2 | Main characteristics of the 3139 middle-aged men and women according to sex, anthropometric, Flexindex, vital, and clinical data. aCharacteristicsMen WomenAll(N = 2087)Survivors(N = 1863)Non-survivors(N = 224) pc All (N = 1052) pbSurvivors(N = 974)Non-survivors(N = 78) pcAge (years) 55.5 ± 5.7 55.1 ± 5.7 58.2 ± 5.4 <0.001 55.1 ± 5.6 0.079 54.9 ± 5.6 57.4 ± 5.3 <0.001Height (cm) 174.8 ± 6.8 175.0 ± 6.8 173.8 ± 6.4 0.011 161.6 ± 6.5 <0.001 161.7 ± 6.5 161.2 ± 6.7 0.044Weight (kg) 85.5 ± 14.9 85.3 ± 14.8 86.7 ± 16.2 0.194 68.6 ± 13.9 <0.001 68.1 ± 13.4 74.3 ± 18.0 0.0002BMI (kg m−2) 27.9 ± 4.3 27.8 ± 4.3 28.6 ± 4.7 0.007 26.3 ± 5.2 <0.001 26.1 ± 5.0 29.0 ± 6.7 <0.001Flexindex 30.5 ± 8.7 30.9 ± 8.6 27.7 ± 9.1 <0.001 41.1 ± 9.4 <0.001 41.3 ± 9.4 37.7 ± 9.6 0.001Follow-up duration (days) 4457(2888–6620)4366(2768–6927)0.581Deaths: yes (%) 10.7 — — — 7.4 0.003 — — —Status indicator: healthy (%) 15.7 19.2 0.013Coronary artery disease: yes (%) 32.7 10.9 <0.001Arterial hypertension: yes (%) 40.8 26.0 <0.001Dyslipidemia: yes (%) 43.1 34.7 <0.001Obesity: yes (%) 25.6 21.3 0.007Diabetes mellitus: yes (%) 11.6 6.4 <0.001a Mean ± standard deviation for all continuous variables, except follow-up duration that was expressed as median [interquartile range].b Comparing all men versus all women.c Comparing survivors versus non-survivors in both sexes.6 of 11 Scandinavian Journal of Medicine & Science in Sports, 2024Subsequently, Kaplan–Meier survival curves and log-ranktests (Figure 4) were calculated for men: [percentile range](death rate%): 6–19 [P1–10] (21.2%), 20–42 [P11–90] (9.6%), and43–57 [P91–99] (7.8%) (x2 = 56.2; p < 0.001—x2 = 40.7 for trend;p < 0.001), and for women: 9–29 [P1–10] (15.4%), 30–52 [P11–90](7.0%), 53–68 [P91–99] (2.0%) (x2 = 28.2; p < 0.001 and for trend;x2 = 22.9; p < 0.001). These findings confirmed a graded trendand a clear survival demarcation among the chosen Flexindex'spercentile ranges for both sexes.Detailed results of Cox hazard regression modeling are pre-sented in Table 3. In brief, for men, in the analysis adjusted forage, the HR (95% CI) for mortality comparing the extreme groupsof Flexindex was 1.91 (1.55–2.36; p < 0.001); which was min-imally decreased to 1.87 (1.50–2.32; p < 0.001) when BMI wasadded to the model and not attenuated with further adjustmentfor the health status (healthy or unhealthy)—to 1.87 (1.50–2.33;p < 0.001). The corresponding adjusted HRs (95% CIs) in womenwere age-adjusted (model 2)—8.23 (2.32–52.27; p = 0.005)—age& BMI-adjusted (model 3)—4.87 (1.25–32.29; p = 0.047) —age,BMI, and health status-adjusted (model 4)—4.78 (1.23–31.71;p = 0.047) (Table 3).4 | DiscussionTo the best of our knowledge, this is the first cohort study toshow that a reduced level of body flexibility—as assessed bythe Flexindex, a composite score of maximal passive ROM of 20movements in seven major joints—is related to higher mortalityin a large middle-aged cohort of men and women. By analyzingdeath rates according Flexindex's data distribution, a consistentpattern emerged among middle-aged men and women, indicat-ing a gradual and inverse relationship between Flexindex andboth natural and non- COVID-19 mortality during an averagefollow-up of nearly 13 years.The association was most pronounced at the extremes of theFlexindex score distribution [death rate%]—men: [P91–99] 7.8%versus [P1–10] 21.2% and women: [P91–99] 2.0% versus [P1–10]15.4%.—and remained significant when further sequentially ad-justed for age, BMI and health status. It is noteworthy that theaddition of a health status indicator in the multivariate modeldid not appreciably modify the HR values either in men orFIGURE 2 | Flexindex results (mean ± standard error of the mean),sex-separated at 5-year of age intervals; values inside the columns areindicating the death rate (%) for this specific sex and 5-year of ageinterval. p < 0.001 for all sex and age-group comparisons by ANOVA2-way (sex and age-group), except by the 46–50 versus 51–55 years'comparison in women.FIGURE 3 | Boxplots (5%–95%) of Flexindex results of survivorsand non-survivors' men and women; arrows indicated a relevant cutoffvalue of Flexindex—49 (men) and 56 (women)—for very low probabilityof mortality in 13 years (1%).FIGURE 4 | Survival curves for Flexindex results in men [N = 2087]and women [N = 1052] aged 46–65-year- old.7 of 11women, indicating that the observed association between flex-ibility and survival is not significantly modulated by commonchronic diseases and/or by the regularly use of medications.The HRs for men and women were nearly 2 and 5, respectively,for lower Flexindex when comparing the selected lowest andhighest percentile ranges, indicating a significantly higher riskTABLE 3 | Associations of Flexindex with natural and non- COVID-19 deaths in 3139 men and women aged 46–65 years- old.Flexindex Events/individualsNatural & non- COVID-19 deathspaHazard ratio (95% CI)MenModel 1—unadjusted6–19 49/231 2.02 (1.64–2.48) <0.00120–42 159/1651 1.55 (1.33–1.81) <0.00143–57 16/205 1 (Reference)Model 2—adjusted by age6–19 49/231 1.91 (1.55–2.36) <0.00120–42 159/1651 1.51 (1.30–1.77) <0.00143–57 16/205 1 (Reference)Model 3—adjusted by age and body mass index6–19 49/231 1.87 (1.50–2.32) <0.00120–42 159/1651 1.49 (1.28–1.75) <0.00143–57 16/205 1 (Reference)Model 4—adjusted by age, body mass index and health status indicator6–19 49/231 1.87 (1.50–2.33) <0.00120–42 159/1651 1.50 (1.28–1.76) <0.00143–57 16/205 1 (Reference)WomenModel 1—unadjusted9–29 18/117 11.69 (3.36–73.59) 0.00130–52 58/831 3.65 (1.14–22.28) 0.07153–68 2/104 1 (Reference)Model 2—adjusted by age9–29 18/117 8.23 (2.32–52.27) 0.00530–52 58/831 2.95 (0.72–12.18) 0.14053–68 2/104 1 (Reference)Model 3—adjusted by age and body mass index9–29 18/117 4.87 (1.25–32.29) 0.04730–52 58/831 2.30 (0.70–14.20) 0.25553–68 2/104 1 (Reference)Model 4—adjusted by age, body mass index and health status indicator b9–29 18/117 4.78 (1.23–31.71) 0.04730–52 58/831 2.30 (0.69–14.22) 0.25453–68 2/104 1 (Reference)Abbreviation: CI, confidence interval.a Cox proportional hazards modeling.b See the text for an explanation of health status indicator.8 of 11 Scandinavian Journal of Medicine & Science in Sports, 2024of mortality with reduced body flexibility. Notably, the Kaplan–Meier curves for men and women revealed that survival dif-ferences between percentile ranges emerged early after theevaluation and were already discernible at 5–6-year.4.1 | Flexitest as the Assessment Toolfor Evaluating FlexibilityIn flexibility assessment there are several issues to consider.The three most important are: (a) the ROM can be measured byactive—the individual itself performing the movement—or bypassive techniques, in which external forces are applied, usuallyby an evaluator. Due to anatomical and/or physiological rea-sons, in general, the maximal ROM is greater when measured inpassive than in active mode. Thus, when considering flexibilityassessment, it is crucial to differentiate methods that evaluateactive and passive range of joint motion; (b) the concept of spec-ificity [9], that is, the ROM is specific for joints and movements.In this sense, it is possible to find in a single individual withan above normal ROM for elbow flexion and a below averagemotion for the elbow extension and knee flexion. Thus, ideally,when attempting to evaluate body flexibility in a broader sense,it is more appropriate to use passive methods that will assess theROM in various movements in several different joints; and (c) itis advantageous to consolidate all measurements of ROM in asingle composite and statistically validated score.Linear flexibility assessment methods, such as the classical1950s sit-and-reach test [10] tend to rely on active movements en-compassing several joint movements, represented by a single nu-merical results expressed in centimeters or inches. Such tests areintended to reflect, respectively, back or leg or upper body flexi-bility, missing the chance to provide a complete body flexibilityscore. Angular methods such as goniometry, in which resultsare expressed in angles between body parts [13, 23], could beapplied either in active or passive modes, but fail in generatinga unique or composite score that reflects overall body flexibility.On the other hand, the Flexitest [7, 24–26] adequately contem-plates these three important aspects of flexibility assessment, inpassively evaluating maximal ROM in 20 movements in sevendifferent joints, and in generating the Flexindex, a compositescore of body flexibility.4.2 | Comparison With Other StudiesEven though ROM or flexibility have been investigated fordecades [9, 10, 13, 14, 27, 28] and flexibility exercises havebeen widely recommended by leading health institutions[27, 29, 30], there are scarce data relating body flexibility torelevant health-related and mortality outcomes. Most of theavailable information is on risk of injuries [27, 31] or poor ar-terial stiffness [32, 33], and virtually no previous study has as-sessed the association between levels of a joint mobility indexor body flexibility and mortality. This knowledge gap maybe attributed to the inherent limitations of traditional meth-ods used to assess flexibility, such as the sit-and-reach test[34], goniometry and the Beighton laxity scale [35]. In 2002,Katzmarzyk & Craig [36] studied musculoskeletal fitness ina large sample of Canadians aged 20–69 years over a 13 yearperiod and failed to find an association between trunk flex-ion mobility and mortality. Comparing the earlier study to thecurrent report is difficult given their use of a 1950s era sit-and-reach test, a limited method to assess level of body flexibility,along with their use of a wider age range and a lower totaldeath rate (2.9%) during follow-up. On the other hand, thereis an abundance of observational and intervention studies onflexibility [25, 26, 37], often reporting group distinctions and/or favorable results in ROM after specific flexibility trainingand even a reduction in systolic and diastolic blood pressurelevels among Japanese older adults who engaged in flexibilitytraining three times/week [38]. However, none of these stud-ies has specifically investigated associations between levels offlexibility and all- or cause- specific mortality.4.3 | Data InterpretationAs previously reported in many flexibility studies [24, 26, 39]using various assessment tools, women tend to demonstrate con-siderably higher levels of joint mobility as compared to men. Inour study, this was also observed in all 20 joint movements, withsignificantly higher scores for women. In addition, the averageFlexitest score, a joint mobility index, was 35% higher for womenand, interestingly, women aged 61–65 years were, on average,10% more flexible than men aged 46–50 years – (37 ± 9 vs. 34 ± 8),that is, 15 years younger. It is important to acknowledge that gen-eralized joint hypermobility, now commonly referred to as hy-permobility spectrum disorders (HSD), is a clinically significantcondition associated with connective tissue abnormalities andrelated comorbidities [40]. However, generalized hypermobilityis more common in children, adolescents and young adults. In acomprehensive review of all subject's medical history and accord-ing to the results of our Flexitest assessment (by the number of in-dividually attributed scores of 4), it is highly unlikely that >0.5%of our individuals could be classified as having generalized hy-permobility at middle-age. Therefore, in practical terms, we con-sider that all men and women in the upper extremes of Flexindexwere not exhibiting HSD and thus, our main findings should notbe extrapolated directly to individuals diagnosed with HSD.Although we found a significant and graded association betweenFlexindex and natural mortality risk, the differences in Flexindexassessment results between survivors and non-survivors wererelatively modest, much smaller than the differences in Flexindexbetween men and women at similar ages (Figure 3). The most im-pressive differences in the death rates were seen at the extremesof the distributions, especially in women. According to our data,very high levels of Flexindex were associated with very low mor-tality, especially in women. Comparing the percentile ranges formen and women permitted us to highlight this interesting find-ing between the genders.4.4 | Limitations of the StudyThe CLINIMEX Exercise cohort has general features that shouldbe considered before extrapolating these findings to other popula-tions or different age-groups. The cohort primarily included indi-viduals of Caucasian ethnicity belonging to higher socioeconomicand educational levels. Furthermore, these individuals voluntarily9 of 11sought evaluation at a private clinic located in Copacabana, Riode Janeiro, Brazil, indicating a high level of motivation and inter-est in undergoing a medical-functional assessment.There were significant differences in sex regarding Flexindex,clinical variables and death rates. However, in performing sep-arate survival analyses by sex and in adding a health status in-dicator as a covariate in a fully adjusted model, we were able toshow that similar trends for higher mortalities occurred withcomparable reduced body flexibility levels in men and womenaged 46–65 years. While data on various clinical variableswere reported, no formal data were available regarding cur-rent smoking status for all individuals in the study. However,a separate analysis with data from 1559 adults of CLINIMEXExercise cohort indicated that current smokers were only 5%with roughly 2% smoking >20 cigarettes/day, figures unlikelyto influence the association between flexibility and naturaland non- COVID mortality. In addition, we were unable tocontrol for time of the day of the flexibility measurements, dietor psychological variables or to analyze several other variablesthat could potentially influence our results. These includepast and current exercise or physical activity or sports habitsand amount of sedentary time or behavior and/or experiencewith activities known to favorably influence flexibility (i.e.,Tai Chi Chuan, yoga or dance practice), level of other physi-cal fitness components or a more detailed analysis of clinicalvariables. Nevertheless, one of the most universally acceptedtheories in exercise and sports medicine is the specificity oftraining; meaning, in a simplified manner, that only physi-ological attributes or physical fitness components that aretrained are maintained or improved. Thus, the most effectivestrategy for improving flexibility is to regularly practice activ-ities that require maximal passive ROM, most often includingspecific stretching or flexibility exercises. On the other hand,there is no biological plausibility to expect that individualswho regularly walk, run or cycle would significantly improvetheir maximal passive ROM in shoulder or trunk movementsor even in lower limbs joints.The observational design of the study does not establish cau-sality. Accordingly, we are unable to show a definitive mech-anism to explain the association between low levels of bodyflexibility and premature mortality. Other studies have sug-gested that there is an association between poor trunk flex-ibility and selected vascular abnormalities [33], incidence ofhypertension [41] and premature atherosclerosis [42]. It is rea-sonable to consider that reduced flexibility at middle-age maybe a consequence of poor overall lifestyle habits rather thana causal risk factor for mortality. Nevertheless, our data indi-cate that there is a possibility to assess middle-aged men andwomen who may have an increased risk of premature deathusing a simple, low- cost, and non-invasive assessment toollike the Flexitest. Finally, our inability to control for physicalactivity patterns or smoking may have influenced the relation-ship between flexibility and survival.4.5 | Strengths of the StudyNotable strengths of the study were: (a) the availability of flexi-bility exposure and mortality data from a large open prospectivecohort including men and women; (b) the use of unique Flexitestthat assessed ROM across 20 different movements per individ-ual that allowed calculation of the pooled Flexindex, an indi-cator of the level of body flexibility; (c) the fact that Flexitestfacilitated the measurement of maximal passive ROM across asubstantial number of joint movements (62 780 movements intotal), conducted in a standardized manner by a team of fivephysicians over three decades; and (d) the availability of severalvalid anthropometric, health, clinical, and vital status data thatwere used to characterize the individuals, to account for poten-tial confounding variables and in evaluating outcomes over along-term follow-up period.4.6 | Future StudiesOur results represent an intriguing research area for the influ-ence of non-aerobic components of physical fitness on health.After having identified a significant impact of flexibility in deter-mining survival, future studies should be performed to addresstwo major questions: (a) whether assessing flexibility in severaljoint movements provides additional predictive value beyondthat of aerobic fitness alone, muscle strength and power, and bal-ance in determining survival outcomes; and (b) whether specificstretching or flexibility training, such as Tai Chi Chuan or yogapractice, or detraining-induced changes in body flexibility levelsimpact survival.5 | ConclusionsOur findings support the significance of flexibility as an integralcomponent of health-related physical fitness. An assessment ofbody flexibility using the Flexitest may serve as a useful com-plement to routine evaluation protocols for providing exerciserecommendations and prescriptions to healthy and unhealthymiddle-aged individuals. Reduced levels of body flexibility wereassociated with shortened survival in a convenience sample ofmen and women aged 46–65 years.5.1 | PerspectiveIt is widely recognized that physical fitness is a strong predictorof survival in adults [1]. While this is unquestionable for aero-bic or cardiorespiratory fitness [1, 2], recent studies have indi-cated that muscle strength and balance are also predictors ofall- cause mortality in middle-aged and older subjects [3–5]. Sofar, it is virtually unknown is there is an association betweenflexibility—another of the non-aerobic physical fitness com-ponents—and mortality. This study shows that reduced bodyflexibility, as assessed by the Flexitest—20 movements in sevendifferent joints -, is related to a reduced survival in middle-agedmen and women followed by a median of 12–13 years. This as-sociation lends further support for considering flexibility as acomponent of health-related physical fitness. Routine assess-ment of body flexibility may be considered as part of exercise-related evaluation protocols favoring a more individualizedprescription of flexibility exercises. Since body flexibility levelscould be improved by specific training regimens, future stud-ies should explore whether training-induced gains in flexibility10 of 11 Scandinavian Journal of Medicine & Science in Sports, 2024will provide independent and/or additive survival benefits inmiddle-aged and older individuals.