Abstract
Background Measurement of function usually involves the use of both performance-based and self-report instruments. However, the relationship between both types of measures is not yet completely understood, in particular for older adults attending primary care.
Objective The main objective of the study was to investigate the association between the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0) and the Short Physical Performance Battery (SPPB) for older adults at primary care. A secondary objective was to determine the influence of sociodemographic and health-related variables on this relationship.
Design This was a cross-sectional study.
Methods A total of 504 participants aged 60 years and older from 18 different primary care centers underwent a one-session assessment including: sociodemographic variables, comorbidities, performance, self-reported disability, pain, depressive symptoms, and physical activity. Performance was assessed using the SPPB, and self-reported disability was assessed using the WHODAS 2.0.
Results The correlation between WHODAS 2.0 and SPPB scores was strong (r=.65). Regression analysis showed that the SPPB total score explained 41.7% of the variance in WHODAS 2.0 scores (adjusted R2=41.6%). A second model including the SPPB subtests (balance, gait, and sit-to-stand), depressive symptoms, number of pain sites, pain intensity, and level of physical activity explained 61.7% of the variance in WHODAS 2.0 scores (adjusted R2=60.4%). No model improvement was found when considering the 6 WHODAS 2.0 individual domains.
Limitations The cross-sectional nature of the study does not allow inferences on causal relationships.
Conclusions This study's findings confirm that self-report and performance-based measures relate to different aspects of functioning. Further study is needed to determine if primary care interventions targeting lower extremity performance and depressive symptoms improve self-reported disability.
Between the years of 2000 and 2050, the proportion of the world's population over 60 years of age will double from about 11% to 22%.1 The absolute number of people aged 60 years and older is expected to increase from 605 million to 2 billion over the same period.1 Aging is associated with an increase in a number of health conditions and unrecognized comorbidities that increase the risk of functional decline.1,2 Early identification of changes in function can lead to recommendations for appropriate interventions aimed at preventing or slowing the disablement process.
Measurement of function is complex and usually involves the use of both performance-based and self-report instruments. Performance-based instruments capture how well an individual can perform a task and usually involve the completion or timing of strength, balance, or mobility tasks by an assessor.3 Self-report instruments capture people's perception of their capability to perform a range of activities in their day-to-day life.3 Increasing evidence suggests that self-report and performance-based measures capture different but complementary constructs.4,5 One of the most commonly mentioned advantages of self-report measures is that they evaluate multiple aspects of function in one test.6,7 Mentioned disadvantages are that self-report measures are influenced by expectations and beliefs of the patients, impaired cognition or memory, and inability to answer accurately.8–10 A commonly reported disadvantage of performance-based instruments is that activities do not necessarily reflect day-to-day relevant aspects of functioning, as they comprise isolated tasks performed in an artificial environment.6,7 However, performance-based instruments appear to be more sensitive to early deterioration of function and to be able to predict falls, loss of ability to walk, increased reliance on others, higher risk of institutionalization, and an increased likelihood of death.11–16 These characteristics of performance-based instruments are of major relevance in the primary care setting, where the identification of people at risk of further decline of function is essential. There is consensus that the combined use of both self-report and performance-based disability measures provides the clinician with a more comprehensive view of a patient's functional status.17,18 However, routine utilization of both self-report and performance-based tests in clinical practice is likely to be time-consuming. Lack of time appears to be the main barrier to the use of standardized outcome measures.19
Several studies have compared self-report and performance-based measures, and results indicate poor-to-moderate correlation,8,20 suggesting that they measure different constructs. However, the association between self-report and performance-based disability measures is likely to differ according to the type of measures used, type of patients, and clinical setting. In this study, we used the 12-item World Health Organization Disability Assessment Schedule (WHODAS 2.0) as a self-report measure of disability and the Short Physical Performance Battery (SPPB) as a performance-based functional measure shown to be associated with disability. The WHODAS 2.0 is a valid and reliable generic instrument covering 6 domains of functioning: cognition, mobility, self-care, getting along, life activities, and participation.21 The SPPB is one of the most often used performance-based instruments both in research and clinical practice and is easy to administer without the use of specialized equipment.12,22 Furthermore, it has been shown to be reliable and valid among several elderly populations that differed in terms of culture, language, and education.23 To our knowledge, how SPPB scores relate to WHODAS 2.0 scores has not been investigated. The aims of this study were: (1) to investigate the association between the WHODAS 2.0 as a measure of self-reported disability and the SPPB as a measure of performance-based function for older adults in primary care and (2) to determine the influence of sociodemographic variables (age, sex, education) and health-related variables (number of comorbidities, pain, depressive symptoms, and physical activity) on this relationship.
Method
Participants
Participants were recruited at 18 primary health care practices located across the municipalities of Aveiro, Portugal. They were either referred by health care practitioners or directly invited by the researchers. Participants could be enrolled in the study if they were ≥60 years of age and were able to give written informed consent, which was ascertained by asking participants to explain in their own words what the study involved. The number of participants from each municipality and health care practice was proportional to the population served. This proportionality was achieved by performing an a priori sample size calculation considering the total number of inhabitants from the 3 municipalities aged 18 years and older, a confidence level of 95%, and a confidence interval width of 4%. The total sample size calculated for the study (N=504) was subdivided according to the percent contribution of each municipality, resulting in 259 participants from the municipality of Aveiro, 147 participants from the municipality of Ihavo, and 98 participants from the municipality of Vagos. The number of participants assessed at each primary care practice within the same municipality was calculated based on the percentage of inhabitants served at each practice by sex and age group. All participants signed an informed consent statement prior to their participation.
A total of 504 participants (338 women and 166 men) with a mean age of 70.9 years (SD=7.5) entered the study. A detailed characterization of the sample is presented in Table 1.
Sample Characteristics (N=504)
Procedure
Data were collected in a one-session interview between February 2012 and March 2014 at the primary health care practice that each participant usually attended. All researchers involved in data collection were previously trained by 2 of the authors (A.G.S. and A.Q.). Training included discussion of procedures and application of instruments to participants not included in the study. In addition to self-report and performance-based measures, demographic and health characteristics, pain, depressive symptoms, and physical activity were assessed, as they are important determinants of disability that could influence the association between performance and self-reported disability.20,24 The procedures and instruments used are further explained below.
Self-reported disability.
Self-reported disability was assessed using the Portuguese version of the 12-item, interview-administered version of WHODAS 2.0, which is valid and reliable.25 The WHODAS 2.0 is a disability assessment instrument based on the conceptual framework of the International Classification of Functioning, Disability and Health with a recall period of 30 days.21 It has 2 questions from each of 6 domains: (1) cognition, (2) mobility, (3) self-care, (4) getting along, (5) life activities, and (6) participation. The WHODAS 2.0 total score was calculated as the sum of the individual scores assigned to the 12 items (simple score method): 1 (none), 2 (mild), 3 (moderate), 4 (severe), and 5 (extreme/cannot do).21 The sum score for self-reported disability, therefore, ranged from 0 (no disability) to 60 (complete disability), with higher scores indicating higher levels of disability. Additionally, and for the regression analysis, domain scores were calculated as the sum of the scores of the 2 items from each domain.
Performance-based function.
Performance-based function was assessed using the SPPB.13 The SPPB total score is a composite score based on the individual scores of 3 timed subtests: (1) the ability to stand with the feet side by side, semitandem, or tandem for 10 seconds (balance), (2) usual walking speed (calculated over 3 m), and (3) the ability to rise from a chair as quickly as possible 5 consecutive times. Each of the 3 performance measures was assigned a score ranging from 0 to 4, with 0 indicating the inability to complete the test and 4 indicating the highest level of performance. A summary score (range=0–12) was subsequently calculated by adding the scores of the individual tests. Higher scores reflect higher levels of function.13
Demographics and chronic conditions.
Demographics included age, sex, and education. The presence of chronic conditions was ascertained by asking participants whether they had any of the following conditions: (1) hypertension, (2) diabetes, (3) cardiovascular disorders, (4) respiratory disorders, (5) cancer, (6) osteoarthritis (eg, back, hip, or knee), (7) other known medical condition, or (8) any medical condition for which the nature or medical diagnosis was not known. The total number of reported chronic conditions was counted, a procedure already used to categorize comorbidities.26,27
Pain, depression, and physical activity.
Pain, depression, and physical activity were assessed, as they have been shown to be associated with disability.20,28,29 Details of assessment are presented below.
In the present study, global pain intensity (pain intensity considering all of the pain sites) on the day of data collection was measured using a 10-cm vertical numeric rating scale, anchored with 0 (“no pain”) and 10 (“most severe pain imaginable”). Participants also were asked to mark on a body chart where they felt pain in the week preceding data collection. The number of painful body sites was counted and categorized as: (1) a single pain site, (2) 2 pain sites, (3) 3 or more pain sites but not meeting the criteria for widespread pain, or (4) widespread pain. Widespread pain was defined as pain on the left and right sides of the body, pain above and below the waist, and axial skeletal pain.30 Both the numeric rating scale and the body chart have been shown to be valid and reliable in older people.31,32 Pain frequency during the week before the interview was categorized as: (1) seldom (once a week) or occasionally (2–3 times a week) or (2) often (more than 3 times a week) or always (all days). Pain duration was categorized as: (1) <6 months or (2) ≥6 months.
Depressive symptoms were assessed using the short version of the Portuguese translation of the Geriatric Depression Scale (GDS).33 This is a valid and reliable 15-item self-report scale of depression initially developed for adults aged 65 years or older.34 However, Weintraub et al35 showed that its level of sensitivity and specificity for patients aged less than 65 years is comparable to that of patients aged more than 65 years. Participants were asked to respond by answering “yes” or “no” to each of the 15 items: 10 items indicate the presence of depression when answered positively, and 5 items indicate depression when answered negatively. A score of 1 is given when the answer is indicative of depression, and the total score ranges from 0 to 15. For the regression analysis, we dichotomized the results as: (1) no depressive symptoms (≤4 points) and (2) depressive symptoms (≥5 points).35
Physical activity was assessed using the Portuguese version of the Rapid Assessment of Physical Activity (RAPA) questionnaire.36,37 This questionnaire was specifically designed for use with older people and has 9 items (7+2) with response options of “yes” or “no.” The total score of the first 7 items ranges from 1 to 7 points, with the respondent's score categorized into 1 of 5 levels of physical activity: (1) sedentary; (2) underactive; (3) regular underactive, light activities; (4) regular underactive; and (5) regular active. The last 2 items relate to strength training and flexibility and are scored separately for individuals reaching item 7 only.36
Data Analysis
Predictive Analytics Software version 20 (PAWS software [formerly SPSS Statistics]) (IBM Inc, Armonk, New York) was used for statistical analysis. Descriptive statistics were used to characterize the sample in terms of age, sex, years of formal education, chronic conditions, depressive symptoms, WHODAS 2.0 scores, SPPB scores, RAPA scores, and pain characteristics. Mean and standard deviation were reported for continuous variables, and count and proportion were reported for categorical variables. Multiple linear regression models were used to predict WHODAS total score and domain scores as a function of: (1) SPPB total score and SPPB subtest (sit-to-stand, balance, and gait) scores and (2) SPPB subtest (sit to stand, balance, and gait) scores, education, number of chronic conditions, RAPA score, GDS score, pain intensity, number of pain sites, pain duration, and pain frequency. The variables of age and sex were considered covariates in all models, and a forced entry method was used for the regression parameters estimation (enter method). The strength of the correlation between the SPPB and the WHODAS 2.0 was interpreted as low if the coefficient was <.3, moderate if it was between .3 and .5, and strong if it was >.5.38 A correlation matrix for all independent variables was performed to assess multicollinearity (tolerance values <0.2 and variation inflation factor >10), and the residuals' normality was confirmed. Level of significance was set at P<.05.
Role of the Funding Source
Dr Sa-Couto's work was supported by Portuguese funds through the Center for Research and Development in Mathematics and Applications (CIDMA), University of Aveiro, and the Portuguese Foundation for Science and Technology (Fundação para a Ciência e a Tecnologia [FCT]) within project PEst-OE/MAT/UI4106/2014.
Results
Pain, Depressive Symptoms, and Physical Activity
A total of 376 participants (74.6%) reported pain in at least one body site during the week preceding data collection. Mean global pain intensity at the time of data collection was 5.8 (SD=2.3). Most participants had multisite or widespread pain (n=224; 59.6%), pain for 6 months or longer (n=313; 83.2%), and pain that was very often or always present (n=286; 76.5%) (Tab. 1).
The GDS and RAPA data were obtained from 501 participants (data were missing from 3 participants). Of these participants, 184 (36.7%) had a GDS score of 5 or more points, indicating depressive symptoms. On the basis of the RAPA scores, 113 participants (22.6%) were categorized as sedentary; 52 (10.4%) as underactive; 174 (34.7%) as regular underactive, light activities; 105 (21.0%) as regular underactive; and 57 (11.4%) as regular active.
Performance-Based Function and Self-Reported Disability
The mean SPPB score was 8.2 (SD=2.6). The percentage of participants unable to complete the 3 subtests varied between 0.4% and 10.3%, and the percentage of participants reaching the highest performance varied between 10.3% and 76.4% (Tab. 2). The mean WHODAS 2.0 score was 19.6 (SD=7.9). More than 50% of the participants reported no difficulty in items 3 (learning a new task), 4 (joining in community activities), 5 (been emotionally affected), 6 (concentrating), 8 (washing), 9 (getting dressed), 10 (dealing with people), 11 (maintaining a friendship), and 12 (day-to-day work/school). The percentage of participants who reported at least moderate difficulty varied between 2.4% (item 10 [dealing with people]) and 42.3% (item 1 [standing for long periods]). A detailed characterization of self-reported disability is presented in Table 3.
Short Physical Performance Battery Subtests and Number (Percentage) of Participants in Each Scoring Category
WHODAS 2.0 Individual Items and Number (Percentage) of Participants in Each Scoring Categorya
Regression Models of Association Between Performance-Based Function and Self-Reported Disability
The correlation between the WHODAS and SPPB total scores was considered strong (r=.65). When plotting SPPB scores against WHODAS scores (Figure), the data indicated that the variability in WHODAS scores was higher for participants with lower SPPB scores, suggesting that the association between WHODAS and SPPB scores is weaker for people with lower levels of performance. The total SPPB score explained 41.7% of the variance in WHODAS 2.0 scores (R2=41.7%, adjusted R2=41.6%).
Relationship between the Short Physical Performance Battery (SPPB) and the 12-item World Health Organization Disability Assessment Schedule (WHODAS 2.0), both presented as total scores. The linear equation is given by: WHODAS 2.0=35.8 – 2.0 × SPPB; 95% confidence interval for SPPB=−2.2, −1.8; R2=41.7%.
Correlation coefficients between the SPPB subtest scores and WHODAS total score ranged from −.51 (gait subtest) to −.55 (for both sit-to-stand and balance subtests). Entering the individual scores of each of the 3 subtests, the multivariate model explained a similar proportion of the variance in WHODAS 2.0 scores (R2=44.6%, adjusted R2=44.0%) (Tab. 4) as for the total SPPB score (see previous paragraph).
Multiple Linear Regression Results for Prediction of WHODAS 2.0 Scoresa
Table 5 presents results for regression analysis for total WHODAS scores when other variables were included in the model. Only the SPPB subtest scores, GDS score, number of pain sites, pain intensity, and RAPA score were considered predictors of the WHODAS 2.0 scores in the model. Together, these predictors explained 60.5% of the variance in WHODAS scores. The variables of age, sex, education, acute pain, and pain frequency were not significant. The SPPB sit-to-stand subtest was the most important predictor, explaining 34.4% of the WHODAS variance alone. The second most important predictor was GDS score, followed by SPPB balance subtest score and number of pain sites. The SPPB gait subtest score was the fifth most important predictor, adding only 3.6% of explained variance to the previous model.
Multiple Linear Regression Results for Prediction of WHODAS 2.0 Total Score (R2=61.7%, Adjusted R2=60.5%)a
The WHODAS 2.0 total score results from adding the individual scores of the 6 distinctly different domains. Therefore, we explored the association between each of the 6 WHODAS domain scores and the SPPB subtests. The resulting regression models explained less variability than the regression model for the total WHODAS 2.0 score. Explained variance varies between a minimum of 8.1% for domain 4 (getting along) and a maximum of 39.8% for domain 5 (life activities). Similarly, in a second multivariate model, the percentage of explained variance for WHODAS 2.0 domain scores was lower than for the WHODAS 2.0 total score. Maximum explained variance was 54.9% (adjusted R2=53.4%) for domain 5 (variables in the model in order of relevance: SPPB sit-to-stand subtest score, GDS score, SPPB balance subtest score, number of pain sites, pain intensity, and SPPB gait subtest score). For the other WHODAS 2.0 domains, the explained variance results were: domain 1 (cognition: R2=29.1%, adjusted R2=26.8%), domain 2 (mobility: R2=46.5%, adjusted R2=44.7%), domain 3 (self-care: R2=43.6%, adjusted R2=41.2%), domain 4 (getting along: R2=17.3%, adjusted R2=14.6%), and domain 6 (participation: R2=49.2%, adjusted R2=47.5%). Across the 6 domains, GDS and SPPB (balance subtest) scores remained in the final model for all domains, SPPB (sit-to-stand subtest) scores remained in the final model for 4 domains and were the most important predictor for all of them (domains 2, 3, 5, and 6), and SPPB (gait subtest) scores remained in the final model for 3 domains (2, 5, and 6). Pain intensity remained in the final model for 5 domains (all except cognition).
Discussion
The primary aim of the present study was to investigate the association between the WHODAS 2.0 (self-reported disability) and the SPPB (performance-based measure of function) for older adults in primary care. Additionally, the study aimed to determine the influence of a set of other variables known to be associated with disability (age, sex, education, number of comorbidities, pain, depressive symptoms, and physical activity) on this relationship.
This study showed a strong relationship between SPPB and WHODAS 2.0 scores (r=.65). However, in the regression model, the SPPB scores explained only 41.7% of the variance in WHODAS 2.0 scores. When sociodemographic and health-related variables and the individual scores of the 3 SPPB subtests were included in the model, the sit-to-stand subtest was the most important predictor of WHODAS 2.0 scores, explaining 34.4% of its variance. Despite the fact that all SPPB subtests remained in the model, depressive symptoms, pain intensity, physical activity, and number of pain sites also emerged as important predictors of WHODAS 2.0 scores, explaining together with the 3 SPPB subtests 60.5% of the variance in WHODAS 2.0 scores. The moderate percentage of variance explained by this model was likely due to the different constructs measured by the 2 instruments. The WHODAS 2.0 is a comprehensive measure of self-reported disability covering 6 domains of functioning, whereas the SPPB is believed to assess mainly lower limb function.
In general, previous studies have shown lower correlations between self-report and performance-based instruments. Ferrer et al39 found moderate-to-strong agreement (.41–.55) between gait speed and the Five-Times-Sit-to-Stand Test and self-reported functional capacity assessed through a set of 9 questions in a sample of community-dwelling older adults (N=626). Terwee et al20 reported correlation coefficients of .34 between the DynaPort Knee Test (DPKT) and the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) physical functioning subscale score and of .50 between the DPKT and the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36) physical functioning subscale score in a sample of 163 patients with knee osteoarthritis. Gandhi et al40 found low-to-moderate correlations between the WOMAC and SF-36 and the Timed “Up & Go” Test (TUG) in a sample of patients with hip and knee joint replacement (r=.29 between TUG and WOMAC preoperatively; r=.43 between TUG and WOMAC at 12-week follow-up postoperatively; r=−.26 between TUG and SF-36 preoperatively; r=−.34 between TUG and SF-36 at 12-week follow-up postoperatively).
Maly et al28 found that pain was the most important determinant of the WOMAC physical functioning subscale score, explaining 69% of its variance in a sample of participants with knee osteoarthritis. The addition of quadriceps muscle strength to the model increased the percentage of explained variance to only 73%. In contrast, pain explained 38% of the variance in SF-36 scores, and adding hamstring muscle strength to the model increased the percentage of explained variance to 55%. Bean et al,5 in a sample of community-dwelling older adults with mobility limitations (N=137), found that a multivariate regression model including leg velocity, exercise tolerance, chronic conditions, sex, and falls efficacy explained 42% of the variance of the Late-Life Function and Disability Instrument. The different instruments used, the different recall period, and the different types of participants might explain the different results.
Our results showed that the Five-Times-Sit-to-Stand Test was the most important predictor of the WHODAS 2.0 total score as well as of the scores of domains 2 (mobility), 3 (self-care), 5 (life activities), and 6 (participation). Previous studies have shown this test to be associated with the risk of falling in community-dwelling older adults (duration of >15 seconds),15,41 highlighting the relevance of its use in the primary care setting. It also has been included as 1 of the 3 components of a frailty index (the Study of Osteoporotic Fractures Index).42 Lower scores in the sit-to-stand test may reflect muscular weakness of the lower limbs or postural balance deficits,43 which are essential to an independent life. Additionally, the Five-Times-Sit-to-Stand Test is probably the most demanding test of the SPPB battery (as illustrated by the higher percentage of participants unable to perform it compared with the other subtests). Therefore, it is likely to be more discriminative. The inclusion of variables other than the SPPB subtests in the final model (in particular, depressive symptoms and pain) highlight the relevant contribution of these variables to the perception that individuals have regarding their capability of performing day-to-day activities. Furthermore, it suggests that these variables may constitute important targets for primary care interventions aiming to prevent self-reported disability.
Interestingly, when plotting SPPB scores against WHODAS 2.0 scores, the data suggest that the association between these 2 instruments is likely to be lower for patients with worse performance. This finding might suggest that self-reported disability might be associated with different factors or predictors for subgroups of participants with different performance levels. This finding should be further investigated in future studies. In general, the wide variation of WHODAS scores for participants achieving the same level of performance seems to support a previous claim that performance measures might be more appropriate to identify individuals at risk for functional decline.16 It also might be related to the wide scope of domains of functioning covered by the WHODAS 2.0, which requires that several domains of functioning be affected to achieve high scores.
The results suggest that responses to the 12 WHODAS 2.0 items differ based on the domain assessed. For example, the 2 items from the mobility domain had the lowest ceiling effects (standing=39.1%, walking=44.8%). In contrast, the 2 items from domain 4 (getting along) showed the highest ceiling effects (dealing with people you do not know=92.5%; maintaining a friendship=91.3%). These findings suggest that mobility scores are better able to identify individuals with disability in this population of older adults seen in primary care settings. The higher similarity of the mobility items (standing and walking) with the SPPB tests also would suggest that the mobility domain would show the highest association with the SPPB. However, the proportion of variance explained by the SPPB scores was higher for total WHODAS 2.0 scores compared with the percentage of variance explained for any of the 6 domain scores. These results appear to further support the idea that more information is gained with more than one WHODAS domain.
The mean SPPB score (X̅=8.2, SD=2.6) and the mean WHODAS 2.0 score (X̅=19.6, SD=7.9) found in this study are similar to those reported in previous studies for adults aged 65 years and older. Cecchi et al,44 in a sample of 120 community-dwelling older adults with hip pain and 886 community-dwelling older adults without hip pain reported mean SPPB scores of 8.7 (SD=3.3) and 9.8 (SD=3.1), respectively. Bean et al5 assessed 137 older adults with limited mobility and reported a mean SPPB score of 8.7 (SD=1.5). Sousa et al45 assessed older people living in 7 low- and middle-income countries, and mean WHODAS 2.0 scores varied between 15.7 (SD=15.4) and 33.2 (SD=28.9).
Study Limitations
There were several limitations to the study. The cross-sectional nature of the study does not allow inferences on causal relationships between performance and self-reported disability. The WHODAS 2.0 assesses domains that require intact functioning of the upper limbs as well as lower limbs. In addition, cognitive ability is likely to influence WHODAS scores. Therefore, including upper limb performance and cognitive measures might improve the model. Future studies might explore whether combinations of upper and lower limb performance measures increase the ability to predict self-reported disability.
The results of this study give support to previous studies suggesting that disability- and performance-based instruments measure different constructs. Although a performance-based test of function remained in the final model as the most important predictor of WHODAS 2.0 total score and domain scores, other variables (in particular, depressive symptoms and pain) also contributed to explaining WHODAS scores. The results provide support for studies examining interventions aimed at improving lower extremity performance and depressive symptoms to determine if these interventions improve self-reported disability. In addition, our results appear to give some support to the claim that performance-based measures of function might be better at identifying early signs of functional decline.
Footnotes
Dr Silva, Dr Queirós, and Dr Rocha provided concept/idea/research design. All authors provided writing and data analysis. Dr Queirós provided data collection. Dr Silva provided project management, facilities/equipment, and institutional liaisons. Dr Silva and Dr Sa-Couto provided consultation (including review of manuscript before submission).
The authors thank all people from the Agrupamento de Centros de Saúde Baixo Vouga who directly or indirectly contributed to this work and the students who contributed to data collection.
The study received ethical approval from the Regional Health Administration Commission, Coimbra, Portugal.
Dr Sa-Couto's work was supported by Portuguese funds through the Center for Research and Development in Mathematics and Applications (CIDMA), University of Aveiro, and the Portuguese Foundation for Science and Technology (Fundação para a Ciência e a Tecnologia [FCT]) within project PEst-OE/MAT/UI4106/2014.
- Received July 29, 2014.
- Accepted May 19, 2015.
- © 2015 American Physical Therapy Association
References
Issue highlights
