Construction and validation of content of one instrument to assess falls in the elderly

ABSTRACT Objective To develop and validate the content of the online Questionnaire for Fall Risk Assessment in the Elderly. Methods The instrument was developed based on the International Classification of Functioning, Disability and Health (ICF) of the World Health Organization. Initially, the set of items was submitted to evaluation of judges (healthcare professionals with experience in elderly health), who could suggest inclusion or exclusion of questions from the instrument; they were also asked to rate each question according to the expected scope. At this stage, clarity and relevance levels for each item were evaluated, generating a total of Content Validity Coefficient (CtVC). Results Content Validity Coefficient values were satisfactory for both clarity (CtVC=0.76) and relevance (CtVC=0.82) of the questions. Next, a group of elderly volunteers participating in a socializing group evaluated the questionnaire for comprehension. The level of comprehension for each item was identified on a Likert scale, ranging from 0 to 5. The questionnaire was considered easy to understand by most participants (95%), with a mean of 4.75 (±0.11) points for each item. Conclusion The instrument showed acceptable psychometric qualities for screening fall risk among the elderly population. Future studies shall investigate different validation aspects of construct for this measure.


❚ INTRODUCTION
There has been an exponential increase in the number of elderly individuals in the last decades, both in developed and developing countries. In Brazil, the number of elderly people is expected to triple by 2050. (1) In this context, geriatric healthcare is becoming more and more prominent, with an increasing number of scientific research in the field of Gerontology. Chronic diseases and their multidimensional impact on health and quality of life of elderly patients are a major concerns in this age group. Falls are very prevalent in this population and affect their functional capacity, autonomy, social life, and healthcare needs; therefore, falls deserve attention of healthcare professionals who assist the elderly. (2,3) Falls are considered a health indicator and a public healthcare issue among the elderly, especially older individuals. (4) According to the World Health Organization (WHO), the annual global fall rate ranges from 28 to 35% among individuals aged over 65 years, and this prevalence increases to 32 to 42% among those aged over 70 years. (1) According to the Brazilian Ministry of Health, the Unified Health System (SUS -Sistema Único de Saúde) spends over R$51 million in treatments related to falls, per year. (5) Falls have a multifactorial etiology, and are often associated to a decreased functional capacity, unstable gait and loss of balance resulting from sensorineural loss and osteoarticular deficits in the elderly. The main consequences of falls in the elderly include fractures; increased risk of death; fear of falling again, which leads to social isolation and restriction of Activities of Daily Life; general health decline; and increased institutionalization. (6,7) Considering the impact falls have in elderly health, it is important to identify risk factors and use validated instruments to assess fall risk in this population, especially when it comes to prevention. There are currently some instruments for fall risk assessment, but few have been prepared and validated in a way that could be self-applicable to the aging population in Brazil. (8,9) Questionnaires are important tools to obtain information, especially in population studies, and they must present psychometric qualities (validity and reproducibility) to ensure reliability of the indicators assessed. (3) To that end, instrument validity (including content, criteria and construction) is extremely relevant because it determines if the instrument is in fact able to evaluate the measures for which it was designed. (10) Validity of a questionnaire content -one type of validation used in this study, is subjectively judged by specialists in the field to determine if the questionnaire can represent the behavior of the sample being evaluated. This stage checks if the item representativeness adequately expresses the content to be evaluated. (11) It is an important stage in the development of a new instrument.
That means this type of evaluation determines if the instrument's contents effectively explore (with clear language and theoretical relevance) the items to measure a certain phenomenon to be investigated.
Considering the exponential growth of the elderly population, health problems caused by falls, and the lack of self-applicable questionnaires for fall risk assessment among the elderly, it is important to build instruments for this population that yield a more efficient and generalized understanding of older individuals. The objective of this tool is to improve safety and quality of health of elderly individuals, and to aid healthcare professionals in the evaluation and guidance in fall risk management.

❚ OBJECTIVE
To establish and test the metric principles of content validity of an instrument to assess falls in the aged.
❚ METHODS This is a methodological study developed in three stages: instrument construction, Content Validity Coefficient (CVC), and verbal comprehension.

Stage I -instrument construction
The first stage -construction of the online Assessment Instrument for Elderly Falls (IAQI -Instrumento de avaliação de quedas para idosos) -was based on the WHO International Classification of Functioning, Disability and Health (ICF). (12) The ICF is a useful tool to describe and understand an individual's health status and to identify contextual factors (environmental and personal) that favor their functionality. The ICF is widely used, including for population surveys on health and disabilities -it proposes a conceptual functionality and einstein (São Paulo). 2018;16(2):1-8 disability model, theoretically divided into two factors: (a) functionality and disability; and (b) contextual factors. Thus, in this study, each group of questions was divided into two components, in addition to domains, constructs, and positive and negative aspects proposed by the ICF (Appendix 1).
The IAQI was constructed in two blocks. The first focused only on identification and sociodemographic data, and the second block had eight questionsquestions 5, 7, and 8 had unfolding questions for positive answers, bringing us to a total of 14 questions about the outcome (fall risk).
Nine questions were related to functionality and disability factors and five were about contextual factors. For the former, we approached the domains and constructs that make up the components of body functions and structures, and activity and participation. For contextual factors, we addressed only domains and constructs inside the environmental factor component.
After preparing the first version, the instrument was submitted to content evaluation (clarity and relevance of questions), and the processes are described below:

Stage II -content validity
For this stage, we used the CVC procedure described by Hernandez-Nieto, in 2002. (13) The CVC was created to more adequately meet the needs of this kind of validity. It is recommended that item/question evaluators be three or five experts, with recognized theoretical and practical knowledge in the specific area. Content validity is calculated through the CVC because it is an ordinal scale.
Three evaluating judges participated in this stage. Inclusion criteria of the judges were: subject expertise and at least 10-year professional experience; Masters or Doctorate degree in the field of health; teaching or delivering care in public health. Two evaluators were lecturers and had worked in Public Health for 9 years; the third evaluator had a major in physical education and a Master's Degree in Human Aging (concluded four years before).
The material was sent electronically to the judges, with a request that it be returned within 30 days. The criteria used by the judges to evaluate the questions were clarity and relevance, presented on a Likert scale (graded from 1 to 5). The judges were asked the following questions: (14,15) regarding clarity (CVC) -"Do you believe this item/question is clear enough to be understood by the target population?"; and, regarding relevance (CVC) -" Do you believe this item is relevant to the study and the target population?" Regarding clarity, the questions could be scored from very unclear (1) to very clear (5). In the relevance scale, the questions could be scored from very irrelevant (1) to very relevant (5). In addition to the two scales, the judges could add a qualitative evaluation next to each question for further improvement. (15) Based on these answers, we were able to calculate the instrument CVC. This calculation is done in five stages. The first, based on the judge grades (1 to 5), is the calculation of the mean grade given to each item (M x ). The second stage, with base on the mean grade (M x ) is the calculation of the initial CVC for each item (CVC i ) divided by the maximum relevance or clarity grade (5). The third stage is the calculation of error (Pe i ) to discount possible biases of the judges for each question. In this case, one (1) is divided by the number of evaluating judges, to the power of the same number of evaluators. With that, the final CVC (fourth stage) of each item/question (CVC c ) can be calculated by subtracting; i.e, Pe i -CVC i = CVC c . The final stage is the calculation of the total CVC of the questionnaire (CVC t ) for each characteristic (language clarity, practical and theoretical relevance) by subtracting the mean Pe i (MPe i ) from the mean CVC i (MCVC i ). After calculations, the literature (16)(17)(18) considers acceptable questions with a CVC t between 0.7 and 0.8). (19)

Stage III -verbal comprehension
The final version of IAQI was used to test the clarity of questions and facility to answer in a group of elderly individuals, aiming to identify possible comprehension issues.
For this stage, we interviewed 24 elderly individuals (≥60 years), participants of a social group in the city of Porto Alegre (RS), Brazil. They were invited and agreed to participate in this stage of the study after its method and objective were clarified. While applying the questionnaire to the participants, the interviewer was in charge of observing and recording any comprehension difficulties for each question and the total interview time.
Each question received an answer using the Likert scale (0 to 5) to evaluate the level of comprehension. It was established that scores 0, 1, 2, and 3 would indicate insufficient comprehension. (20)

❚ RESULTS
Instrument to assess falls in the elderly was analysed based on two criteria: language clarity and relevance. Table 1 presents the questions referent to the second block (fall risk established based on the ICF) that is part of the initial version of the questionnaire.
In this version of the instrument, 28.5% of questions presented CVC c scores below 0.70 regarding clarity. Approximately 57.1% presented results between 0.70 and 0.80, and only 14% of questions had CVC c scores above 0.80. The items/questions with the lowest scores in the clarity evaluation were CVC c = q1=0.56; q5=0.63; q5.1=0.36; q8.4=0.69 (Table 2).
Considering CVC c values and observations made, a second round of evaluations was sent to the judges. In this case, a new form was sent containing only the criterion of clarity, because of its previous result of CVC t under 0.70. Table 3 presents the CVC values in the second evaluation after readjustment of the questionnaire. After the second evaluation of clarity, 21.4% of questions had CVC c scores under 0.70; about 28.1% had scores between 0.70 and 0.80; and 50% had CVC c results above 0.80.
After all suggestions proposed by the judges, it was possible to calculate the CVC t for clarity, with a result of 0.76, which is considered above the cut-off point suggested in the literature. (19) Regarding relevance, results were better -50% of questions received a CVC evaluation above 0.80. Total content validity coefficient for relevance was 0.82 (Table 3). Based on the CVC calculation (total score -CVC t ), the final version was elaborated and is presented in table 4.
After evaluation by the experts, IAQI verbal comprehension was evaluated by a group of volunteer elderly individuals. Instrument to assess falls in the elderly obtained a mean level of comprehension of 95% (mean±SD = 4.75±0.11). All 24 elderly individuals who participated voluntarily in the instrument pre-test, said the instrument was easy to understand. Only three participants gave four questions a score under 2 ("not very clear"), and the reasons they gave were they did not understand question 7.1 was related to the answer given to question 7, and they also did not understand some questions could have more than one option as an answer. If the answer for question 5 is "yes", the interviewee is asked how long he/she has been doing physical activities. If the answer to question 7 is "yes", the interviewee is asked how many kilos he/she has lost. If the answer to question 8 is "yes", the interviewee is asked four more questions about the fall. If the answer to question 8 is "no", the questionnaire is over.

❚ DISCUSSION
Healthcare services are facing a challenge to assist the elderly population, which is growing considerably and presents a profile marked by chronic diseases and morbidity. This study proposes the use of a selfapplicable, low-cost, quick-to-use instrument that is catered to the elderly population. It also offers educational guidelines to minimize public health issues, If the answer for question 5 is "yes", the interviewee is asked how long he/she has been exercising. If the answer to question 7 is "yes", the interviewee is asked how many kilos he/she has lost. If the answer to question 8 is "yes", the interviewee is asked four more questions about the fall. If the answer to question 8 is "no", the questionnaire is over. After evaluations done by experts, we were able to improve the questionnaire, and some questions were rewritten and/or sub-items were added to further adjust the tool to its target audience. These adjustments were made mainly to questions that were given low CVC c scores regarding clarity. For example, questions 5 and 5.1 about the regular practice of physical activities and how long the individual has been practicing them. This question was rewritten to distinguish between physical exercise and physical activity. Questions 1 and 8.4 were also adjusted. After the judges evaluated question 1, the answers included loss of balance (dizziness), decreased muscle strength (weakness), difficulty walking and fatigue (tiredness and exhaustion) noticed in the last month. All these answers are related to intrinsic factors related to falls in elderly individuals. (21) Delbaere et al., (22) established the psychometric properties of the simple test of choice stepping reaction time (CSRT), investigating its validity and reliability to predict falls in the elderly. CSRT accurately reflects the ability to advance quickly and appropriately to avoid obstacles at the end of a path, which is supported by neuropsychological, balance, and sensory systems that are paramount for balance control. The authors concluded that, for each increased SD in the test, elderly individuals presented 74% more chances of suffering multiple falls, which proved the study usefulness in the evaluation of individuals at risk of falling. However, clinical applicability depends on the training received by the professional who will apply the test. Moreover, elderly patients need to be evaluated by a healthcare professional to better predict the fall risk.
With the respective proposed alterations, question 1 got a better CVC c score for clarity (CVC 1st version =0.56; CVC 2nd version =0.83). After readjustment, question 8.4 received a score below the cut-off point suggested by the literature for clarity (CVC -0.63). Nevertheless, the question was included due to its relevance to the theme and adequate score for relevance (CVC=0.83).
In total, the instrument was finalized with 14 questions, six of which are multiple-choice, and the other eight have simple dichotomous responses ("yes" or "no").
Chang et al., (23) conducted the psychometric assessment of the psychological and social well-being indicators for elderly residents of Chicago, and observed that the use of short instruments with objective answers in population interviews is more adequate to detect health problems in elderly individuals, especially when the study population presents a low level of education. Since the IAQI, proposed in this study, is self-applicable, the questionnaire was carefully prepared with simple language that could be understood even by those with intellectual challenges. Moreover, the instrument's layout is catered to the elderly population. Pedreira et al., (3) evaluated the content validity of the Elderly Health Assessment Tool in a low schooling level population in the Southwest region of the Brazilian State of Bahia. The multidimensional questionnaire was composed of seven blocks (sociodemographic information, housing conditions, life habits, functional capacity, health status, mental health, and quality of life), with a total of 207 questions. After content validation by judges with experience in epidemiology of aging research, 50 questions were removed because the questionnaire einstein (São Paulo). 2018;16(2):1-8 was deemed too long. Elderly Health Assessment Tool showed a mean level of agreement score of 86% and a content validity index of 93.47%.
Despite presenting appropriate clarity and relevance scores, some of the IAQI questions were revised and altered based on the judges' comments. Question 2 (about reduced muscle strength) had satisfactory clarity and relevance scores in the first evaluation but, after the judges' observations, some examples of daily activities were added to aid the interviewee (i.e. climbing stairs, opening containers, carrying grocery bags, etc.).
Finally, after the judges' evaluation, a group of elderly volunteers evaluated the questionnaire verbal comprehension. These data were similar to the findings by Ulian et al., (24) who evaluated verbal comprehension after transcultural adaptation of the State and Trait Food-Cravings Questionnaire to Portuguese, and found a mean comprehension level of 95.4%.
This study has some limitations. Firstly, the evaluating judges were gerontology professionals from the Brazilian state of Rio Grande do Sul, which brings better knowledge about the local reality of these elderly individuals. Secondly, the study presented adequate clarity and relevance scores in the questions presented to evaluate risk of falls; however, other validations are necessary before it can be used in clinical settings.
For clinical uses, further analyses must be conducted, such as construct validity (confirmatory factorial analysis), precision (alpha, omega), and risk measure for each respondent (Receiver Operating Characteristic -ROC -curve). After these procedures, IAQI must be projected in a responsive platform to be adjusted to any kind of device. The unprecedented character of this instrument is the format in which it is being proposed -it is catered to the elderly population and was prepared to be self-applicable and comprehensible to these individuals. It can inform respondents about risk classification and guide them through measures of fall risk reduction.

❚ CONCLUSION
The total clarity and relevance Content Validity Coefficient was considered adequate. The items that presented clarity indices below literature recommendations were kept for being considered important for value measurement. The questionnaire for assessing fall risk in the elderly is a valid instrument from a content perspective (clarity and relevance items/ questions).