Secondary Logo

Journal Logo

Thieves' Market

Discriminant Analysis of the Patient-Specific Functional Scale in Discerning Between Loss of Physical Function Categories in Community-Dwelling Older Adults

Taylor, J. David PhD, PT; Mathis, Ruth Ann PhD, PT

Author Information
Topics in Geriatric Rehabilitation: January/March 2020 - Volume 36 - Issue 1 - p 44-49
doi: 10.1097/TGR.0000000000000253
  • Free

Abstract

Adults who are 60 years or older have a relatively high incidence of physical impairments and experience a considerable decline in physical function over time.1 As a result, research suggests that older adults with physical disability experience an increased mortality rate.2 The financial burden of physical disabilities in older adults on health care is also considerable. According to the 2014 Medical Expenditure Panel Survey, financial expenditures are proportionally higher in older adults (65 years and older) than in children and younger adults, where older adults represented 39.8% of individuals in the high health care spending tier (US $11 110 US or more annually). In 2014, average health care expenditures for older adults were US $68 819.3 A significant contributor to the high health care expenditures in older adults is loss of physical function, where 42% of people in the United States who are 65 years or older (∼15.5 million) have reported having 1 or more limitations performing daily tasks (eg, walking 2-3 blocks).4 Therefore, accurate methods for assessing physical disability in older adults are important.

Patient-reported outcomes are common in clinical practice and epidemiology research. Several patient-reported outcome tools exist for measuring loss of physical function. Such outcome measurement tools consist of more generic assessment of physical function (eg, Medical Outcomes Study 36-item Short Form Survey),5 disease-specific (eg, Western Ontario and McMaster Universities Osteoarthritis Index),6 body region–specific (eg, Lower Extremity Functional Scale [LEFS]),7 and patient-specific (eg, Patient-Specific Functional Scale [PSFS]).8 All of these patient-reported outcome measures, except patient-specific assessments, contain standardized items (questions) and can be referred to fixed-item questionnaires.

Fixed-item instruments have strengths in that data can be compared between groups of patients and may have greater capabilities for detecting loss of physical function in patients with more severe physical disability.9 However, fixed-item outcome measures may have decreased abilities for identifying physical function deficits in patients with minor to moderate physical disability.9 Such assessments may also miss data regarding functional activities that are important to the individual. Thus, the ability of the instrument to evaluate patient-focused physical function is hindered.

The PSFS is a patient-reported outcome measure that is utilized to examine any physical function limitations that are specific to the patient. Administration of the PSFS enables a patient to report 3 to 5 functional activities that he or she is either unable to perform or has difficulty performing. The individual is instructed to rate the difficulty of performing each activity on a Likert-type scale of 0 to 10. A rating of “0” corresponds to being unable to perform the activity, whereas a rating of “10” corresponds to being able to perform the activity without difficulty.10 The PSFS has been validated as an index of physical function in patients with a variety of musculoskeletal disorders and older adults.8,11,12 However, after an extensive literature review, only one study was discovered that investigated the appropriateness of the PSFS in discriminating between different levels of physical function, which included only patients with musculoskeletal impairments.13

A problem exists in that limited data have been published on the ability of the PSFS in discriminating between different levels of loss of physical function and any such data are applicable to populations with only musculoskeletal conditions. An investigation of the ability of the PSFS in discriminating between loss of physical function categories in older adults would be important for the purposes of prevention, evaluation, screening, and diagnosis. The research hypothesis was that the PSFS can discriminate between levels of loss of physical function in older adults. The purpose of this study was to conduct a discriminant analysis of the PSFS to determine its ability to discern between categories of loss of physical function in community-dwelling older adults.

METHODS

Study design

This investigation was a substudy of a research project by Mathis et al,8 in which the reliability and validity of the PSFS as a measure of physical function in older adults were assessed. In brief, a cohort of community-dwelling older adults completed a set of physical function questionnaires, including the PSFS, and performed a series of physical performance tests. Correlations between the PSFS and the other measures of physical function were then evaluated. The Harding University and University of Central Arkansas institutional review boards approved the study by Mathis et al,8 and written informed consent was obtained from all participants.

This substudy was a discriminant analysis of the PSFS to determine its ability to discern between categories of loss of physical function in the cohort of community-dwelling older adults. Previous research found that the PSFS was a valid measure of physical function when compared with the LEFS, the Activities-specific Balance Confidence (ABC) scale, and the Short Physical Performance Battery (SPPB).8 Thus, PSFS, LEFS, ABC scale, and SPPB scores were divided into loss of physical function categories (mild, moderate, severe, and very severe) and discriminant analysis was used to determine the ability of the PSFS to discern between the different categories of loss of physical function.

Setting and participants

A cohort of community-dwelling older adults from 3 adult independent living facilities located in the central Arkansas US geographic area was recruited through the use of study recruitment flyers, word of mouth, and electronic mail. Participants were excluded from the study if any of the following criteria were met: (1) younger than 65 years of age, (2) acute or terminal illness, (3) myocardial infarction in the past 6 months, (4) unstable cardiovascular or metabolic disease, (5) neuromuscular or musculoskeletal disorders severely disrupting voluntary movement, (6) limb amputation, (7) upper or lower extremity fracture in the past 3 months, (8) cognitive impairment, or (9) inability to participate in this study due to a physical impairment. Each participant signed a written, informed consent prior to his or her participation in the study. Following informed consent, participants completed a detailed health history questionnaire.

Outcome measures

Patient-Specific Functional Scale

The PSFS is a questionnaire that collects data regarding patient perceptions of physical function limitations. Each participant identified 3 physical activities that he or she had difficulty performing and rated the degree of difficulty of performing each activity on an 11-point Likert scale (0 = unable to perform the activity; 10 = able to perform activity at the same level as before injury or problem). Deficits in physical function for each participant were expressed as the mean score of the 3 physical activities. As previously mentioned, the PSFS has been shown to be reliable and valid as an outcome measure of physical function in older adults.8

Lower Extremity Functional Scale

The LEFS is a region-specific, patient-reported measure of physical function that can be used to assess functional status in patients with a variety of lower extremity disorders. The LEFS consists of 20 questions designed to assess the degree of difficulty of specific functional tasks. Each question was scored using a 5-point Likert scale, where “0” was defined as “extreme difficulty” and “4” as “no difficulty.” The extent of physical function impairment was defined as the sum of all 20 questions, where higher scores represented higher levels of physical function. Previous research suggests that the LEFS is a reliable and valid measure of physical function in older adults.8,14

Activity-Specific Balance Confidence Scale

The ABC scale is a self-report 16-item questionnaire that asks people to score their perceived level of balance confidence when performing common indoor and outdoor activities of daily living. Each item was scored on an 11-level rating scale, from 0% (no confidence) to 100% (full confidence in performing the activity without losing balance). The total ABC score is the mean sum of the individual item scores. Reliability and validity of the ABC scale in older adults have been previously reported.8,15

Short Physical Performance Battery

The SPPB consists of 3 timed tests to measure balance, gait speed, and lower-limb force. Each test is scored from 0 to 4 points, with individuals not able to perform a test scored at 0 and 4 representing the best category of performance. The final score is the sum of the 3 assessments, ranging from a minimum of 0 to a maximum of 12 points. The findings of previous studies indicate that the SPPB is a reliable and valid assessment of physical function in community-dwelling older adults.8,16

Data analysis

Participants' characteristics data were reported as means and standard deviations for age (years), body mass (kilograms), height (centimeters), and body mass index. Means and standard deviations of PSFS, LEFS, ABC, and SPPB scores were also calculated.

The ability of the PSFS to discern between levels of loss of physical function was evaluated through discriminant analysis. Literature to support statistical methods of sample size estimation for discriminant analysis is somewhat conflicting. Also, because this investigation was a substudy of previous work,8 an a priori power analysis to estimate a sample size could not be conducted. Therefore, discriminant analysis was performed using the cohort of 31 participants from the previous study8 and the discussion of the findings was focused on statistical significance and clinical relevance.

Because of differences in scoring of the outcome measures, transformation of the data to z-scores was performed so that all outcome measures data would be on the same scale. The z-scores for each outcome measure were divided into categories, where z-scores of −1.0 or less were classified as “very severe loss of physical function,” z-scores of −0.999 to 0.0 were classified as “severe loss of physical function,” z-scores of 0.01 to 1.0 were classified as “moderate loss of physical function,” and z-scores of more than 1.0 were classified as “mild loss of physical function.” Discriminant analysis was then conducted to determine the accuracy of PSFS classifications, where the grouping variables were LEFS, ABC scale, and SPPB classifications and the independent variable was PSFS classification. The canonical correlation was used to evaluate the relationship between classification and the discriminant function (eigenvalue). A χ2 test was utilized to determine the statistical significance of the correlation. An α level of .05 was used for all statistical analyses. Statistical analyses were conducted using IBM SPSS 24 for Windows (SPSS, Inc, Chicago, Illinois).

RESULTS

Participant characteristics are outlined in Table 1. The classification tables (Tables 2–4) provide data regarding the accuracy of the PSFS to correctly classify the participants into very severe, severe, moderate, and mild loss of physical function categories, based on LEFS, ABC scale, and SPPB classifications.

TABLE 1 - Participant Characteristicsa
Gender (M:F) Age, y Body Mass, kg Height, cm Body Mass Index PSFS Score LEFS Score ABC Scale Score SPPB Score
11:20 81.1 (8.3) 70.6 (15.0) 164.5 (9.8) 26.0 (4.7) 4.5 (2.4) 49.9 (15.0) 71.3 (19.5) 8.1 (2.7)
A
bbreviations: ABC, Activities-specific Balance Confidence; LEFS, Lower Extremity Functional Scale; PSFS, Patient-Specific Functional Scale; SPPB, Short Physical Performance Battery.
a
Data are reported as mean (standard deviation) for continuous variables and counts for dichotomous variables.

TABLE 2 - Predicted PSFS Classification Versus LEFS Classificationa
LEFS Classification Predicted PSFS Classification
Very Severe Loss of Physical Function Severe Loss of Physical Function Moderate Loss of Physical Function Mild Loss of Physical Function
Very severe loss of physical function 5 1 0 0
Severe loss of physical function 0 8 0 3
Moderate loss of physical function 0 1 0 4
Mild loss of physical function 0 0 0 9
A
bbreviations: LEFS, Lower Extremity Functional Scale; PSFS, Patient-Specific Functional Scale.
a
Data are presented as counts of cases.

TABLE 3 - Predicted PSFS Classification Versus ABC Classificationa
ABC Classification Predicted PSFS Classification
Very Severe Loss of Physical Function Severe Loss of Physical Function Moderate Loss of Physical Function Mild Loss of Physical Function
Very severe loss of physical function 0 3 1 0
Severe loss of physical function 0 10 3 0
Moderate loss of physical function 0 1 7 0
Mild loss of physical function 0 1 5 0
A
bbreviations: ABC, Activities-specific Balance Confidence; PSFS, Patient-Specific Functional Scale.
a
Data are presented as counts of cases.

TABLE 4 - Predicted PSFS Classification Versus SPPB Classificationa
SPPB Classification Predicted PSFS Classification
Very Severe Loss of Physical Function Severe Loss of Physical Function Moderate Loss of Physical Function Mild Loss of Physical Function
Very severe loss of physical function 0 0 5 0
Severe loss of physical function 0 0 8 0
Moderate loss of physical function 0 0 14 0
Mild loss of physical function 0 0 4 0
A
bbreviations: PSFS, Patient-Specific Functional Scale; SPPB, Short Physical Performance Battery.
a
Data are presented as counts of cases.

For the discriminant analysis model using the LEFS as the grouping variable, 71.0% of original grouped cases were correctly classified by the PSFS. The PSFS correctly classified 83.3% of the participants as having very severe loss of physical function, 72.7% as having severe loss of physical function, and 100% as having mild loss of physical function, when LEFS classifications were used as the reference. The model indicated that the PSFS did not correctly classify any cases as having moderate loss of physical function, when considering the LEFS classification of moderate loss. The canonical correlation between the eigenvalue and classifications for the LEFS was 0.85 (P < .001).

When the ABC scale was used as the grouping variable, the PSFS correctly classified 54.8% of original grouped cases, where 76.9% were correctly classified as having severe loss of physical function and 87.5% as having moderate loss of physical function. The model indicated that the PSFS did not correctly classify any cases as having very severe or mild loss of physical function, when considering the ABC scale very severe and minor loss of physical function classifications. The canonical correlation between the eigenvalue and classifications for the ABC scale was 0.72 (P < .001).

When the SPPB was used as the grouping variable, the PSFS correctly classified 45.2% of original grouped cases, where 100% were correctly classified into the moderate loss of physical function category. The model indicated that the PSFS did not correctly classify any cases into the remaining loss of physical function categories, when using the SPPB classifications. The canonical correlation between the eigenvalue and classifications for the SPPB was 0.40 (P = .18). For the prediction of PSFS classification to be robust, an assumption of concern (Box's M) was not violated (P > .05 for all analyses).

The PSFS classifications of loss of physical function based on z-scores were transformed to original scores so that the reader can interpret the classifications from a clinical application perspective. Refer to Table 5 for PSFS classifications based on original scores.

TABLE 5 - PSFS Classifications Based on z-Scores and Associated PSFS Original Scoresa
PSFS Classification PSFS z-Scores PSFS Original Scores
Very severe loss of physical function −1.0 or less 0-1.33
Severe loss of physical function −0.999 to 0.0 2.33-4.33
moderate loss of physical function 0.01 to 1.0 5.00-6.66
Mild loss of physical function >1.0 7.33-7.75
A
bbreviation: PSFS, Patient-Specific Functional Scale.
a
Data are presented as ranges.

DISCUSSION

To our knowledge, this study is the first investigation of the use of the PSFS to classify community-dwelling older adults into very severe, severe, moderate, and mild categories of loss of physical function. The findings of this study suggest that the PSFS has the ability to discern between 3 different categories of loss of physical function, when categories of the LEFS are used as the reference (canonical correlation between the eigenvalue and classification for the LEFS was 0.85 [P < .001]). When the ABC scale was used as the grouping variable, the PSFS exhibited the ability to classify cases into severe and moderate categories of loss of physical function but not very severe or mild loss of physical function (canonical correlation between the eigenvalue and classification for the ABC scale was 0.72 [P < .001]). However, the PSFS was unable to accurately classify the study participants into loss of physical function categories, based on the SPPB categories (canonical correlation between the eigenvalue and classification for the SPPB was 0.40 [P = .18]).

The findings of the ability of the PSFS to be a stronger predictor of loss of physical function categories based on the LEFS classifications might be partially explained by the validity of the PSFS. Mathis et al8 found that the PSFS for measuring physical function in community-dwelling older adults exhibited the highest validity when compared with the LEFS (Pearson's r = 0.81, z-scores 95% limits of agreement = −1.2 to 1.2), whereas comparisons with the ABC scale (Pearson's r = 0.68, z-scores 95% limits of agreement = −1.6 to 1.6) and SPPB (Pearson's r = 0.37, z-scores 95% limits of agreement = −2.1 to 2.1) demonstrated less validity.

The results of this investigation are similar to the findings of previous studies. Abbott and Schmitt13 investigated the ability of the PSFS to discriminate between physical function severity states in patients with musculoskeletal disorders. Abbott and Schmitt13 divided LEFS data sets into quartiles (low, medium, and high) corresponding to different levels of physical function and found that mean PSFS scores differed by the quartile of LEFS scores. Our study was different from that of Abbott and Schmitt13 in that we divided PSFS and LEFS data sets into 4 categories of loss of physical function since 3 categories would have resulted in a wider range of scores and would make the findings more challenging to interpret from a clinical application perspective. Also, our study was specifically targeted to community-dwelling older adults, whereas Abbott and Schmitt13 included patients who were 18 years or older with musculoskeletal dysfunction.

The statistical significance and clinical relevance of the study results were considered. According to the study findings, the PSFS was a statistically stronger predictor of very severe, severe, and mild categories of loss of physical function, when the LEFS categories of loss of physical function were entered into the discriminant analysis. Classifications into different categories were based on z-scores. The z-scores of PSFS classifications for very severe, severe, and mild loss of physical function can be transformed into original scores. When PSFS z-scores were used for determining categories and z-scores were transformed to original scores, study participants who scored 0 to 1.33 were classified as having very severe loss of physical function, 2.33 to 4.33 as having severe loss of physical function, 5.00 to 6.66 as having moderate loss of physical function, and 7.33 to 7.75 as having mild loss of physical function. Since the statistical findings indicated that the PSFS had prediction ability to classify cases into very severe, severe, and mild categories of loss of physical function, these ranges of PSFS scores may be a clinical means of categorizing older community-dwelling adults into classifications of loss of physical function. The PSFS may provide clinicians with a method of discriminating between loss of physical function categories, which would be important for the purpose of measuring clinically important differences in loss of physical function. Clinicians should consider that the PSFS was a more robust predictor of loss of physical function classifications based on LEFS categories. Considering the LEFS is specific to measuring lower extremity dysfunction, the PSFS may have a greater ability to classify patients who have functional loss related to lower extremity disorders.

The findings of this investigation have strengths and limitations. One strength of this study was that the ability of the PSFS to discern between categories of loss of physical function was based on the classifications according to 3 validated measurement tools: LEFS, ABC scale, and SPPB. From a clinical relevance interpretation, another strength is that practitioners have an evidence-based method for classifying community-dwelling older adults into categories of loss of physical function using the PSFS, which is user-friendly and presently used in clinical practice.17–20 Limitations of this study include a relatively small sample size and lack of a subgroup analysis (eg, men vs women). Future research should address these limitations. Another area of future study would be the establishment of normative data for the PSFS in older adults.

In conclusion, the results of this investigation suggest that the PSFS can discern between different categories of loss of physical function in community-dwelling older adults. Since the PSFS is a relatively simple to use assessment of loss of physical function that is currently being utilized in clinical practice, clinicians can use the PSFS to also classify community-dwelling older adults into loss of physical function categories.

References

1. Seeman TE, Merkin SS, Crimmins EM, Karlamangla AS. Disability trends among older Americans: National Health and Nutrition Examination Surveys, 1988-1994 and 1999-2004. Am J Public Health. 2010;100(1):100–107.
2. Hirvensalo M, Rantanen T, Heikkinen E. Mobility difficulties and physical activity as predictors of mortality and loss of independence in the community-living older population. J Am Geriatr Soc. 2000;48(5):493–498.
3. Mitchell E. Concentration of Health Expenditures in the U.S. Noninstitutionalized Population. Rockville, MD: Agency for Healthcare Research and Quality; 2014. http://www.meps.ahrq.gov/mepsweb/data_files/publications/st497/stat497.pdf. Accessed January 12, 2018.
4. Federal Interagency Forum on Aging-Related Statistics. Older Americans 2008: key indicators of well-being. https://agingstats.gov/docs/PastReports/2008/OA2008.pdf. Accessed January 28, 2018.
5. Whitson HE, Sanders LL, Pieper CF, et alCorrelation between symptoms and function in older adults with comorbidity. J Am Geriatr Soc. 2009;57(4):676–682.
6. White DK, Master H. Patient-reported measures of physical function in knee osteoarthritis. Rheum Dis Clin North Am. 2016;42(2):239–252.
7. Dingemans SA, Kleipool SC, Mulders MAM, et alNormative data for the Lower Extremity Functional Scale (LEFS). Acta Orthop. 2017;88(4):422–426.
8. Mathis RA, Taylor JD, Odom BH, Lairamore C. Reliability and validity of the Patient-Specific Functional Scale in community-dwelling older adults. J Geriatr Phys Ther. 2019;42(3):E67–E72. doi:10.1519/JPT.0000000000000188.
9. Bieleman HJ, Reneman MF, van Ittersum MW, van der Schans CP, Groothoff JW, Oosterveld FG. Self-reported functional status as predictor of observed functional capacity in subjects with early osteoarthritis of the hip and knee: a diagnostic study in the CHECK cohort. J Occup Rehabil. 2009;19(4):345–353.
10. Stratford P, Gill C, Westaway M, Binkley J. Assessing disability and change on individual patients: a report of a patient specific measure. Physiother Can. 1995;47(4):258–263.
11. Westaway MD, Stratford PW, Binkley JM. The Patient-Specific Functional Scale: validation of its use in persons with neck dysfunction. J Orthop Sports Phys Ther. 1998;27(5):331–338.
12. Mannberg Bäckman S, Stråt S, Ahlström S, Brodin N. Validity and sensitivity to change of the Patient Specific Functional Scale used during rehabilitation following proximal humeral fracture. Disabil Rehabil. 2016;38(5):487–492.
13. Abbott JH, Schmitt JS. The Patient-Specific Functional Scale was valid for group-level change comparisons and between-group discrimination. J Clin Epidemiol. 2014;67(6):681–688.
14. Stasi S, Papathanasiou G, Korres N, et alValidation of the Lower Extremity Functional Scale in community-dwelling elderly people (LEFS-Greek); determination of functional status cut-off points using TUG test. Eur Geriatr Med. 2013;4(4):237–241.
15. Steffen TM, Mollinger LA. Age- and gender-related test performance in community-dwelling adults. J Neurol Phys Ther. 2005;29(4):181–188.
16. Freiberger E, de Vreede P, Schoene D, et alPerformance-based physical function in older community-dwelling persons: a systematic review of instruments. Age Ageing. 2012; 41(6):712–721.
17. Horn KK, Jennings S, Richardson G, Vliet DV, Hefford C, Abbott JH. The Patient-Specific Functional Scale: psychometrics, clinimetrics, and application as a clinical outcome measure. J Orthop Sports Phys Ther. 2012;42(1):30–42.
18. Jolles BM, Buchbinder R, Beaton DE. A study compared nine patient-specific indices for musculoskeletal disorders. J Clin Epidemiol. 2005;58(8):791–801.
19. Nicholas P, Hefford C, Tumilty S. The use of the Patient-Specific Functional Scale to measure rehabilitative progress in a physiotherapy setting. J Man Manip Ther. 2012;20(3):147–152.
20. Barten JA, Pisters MF, Huisman PA, Takken T, Veenhof C. Measurement properties of patient-specific instruments measuring physical function. J Clin Epidemiol. 2012;65(6):590–601.
Keywords:

classification; disability evaluation; geriatric assessment

© 2020 Wolters Kluwer Health, Inc. All rights reserved.