research thesis proposal written and oral.

1. Female Energy Deficiency Questionnaire – FED-Q: Development and Validation

2. ABSTRACT

Low energy availability in exercising women may cause menstrual dysfunction and growth impairment, which can result in low bone mineral density. A reliable a feasible tool for clinical assessment of energy deficiency is needed. This study aims to develop and validate the Female Energy Deficiency Questionnaire (FED-Q). It is anticipated to have approximately 20 items screening for menstrual dysfunction, bone health, exercise level, energy intake, eating disorders and disordered eating, as well as body mass index, and age. Feasibility will be assessed with a pilot study, followed by content validation with the analysis of experts in the field. Modifications will be made in the first draft for subsequent application in a larger cohort. Construct validity will be measured by correlation coefficients, sensitivity, and specificity in detecting energy deficiency, defined as low serum triiodothyronine and low resting metabolic rate ratio. Reliability will be measured in a test-retest method, which will be used to analyzed stability, and with correlation coefficients in a split-half method, for internal consistency. The FED-Q is expected to accurately predict the risk of energy deficiency in exercising women, allowing physicians, dietitians, nutritionists, athletic trainers, and coaches to easily screen their patients and athletes, hence avoiding persistent energy deficiency and related health impairment.

3. INTRODUCTION

Estimating energy deficiency is essential to assess the risk of developing the female athlete triad, which consists of menstrual dysfunction and impaired bone health caused by low energy availability (EA) (De Souza et al., 2014). A fast, feasible, and reliable tool for clinical use is yet to be designed. The aim of this study is to develop and validate the Female Energy Deficiency Questionnaire (FED-Q), a tool to assess energy deficiency (ED) in exercising women. We hypothesize that ED can be predicted with 90% sensitivity by the Female Energy Deficiency Questionnaire – FED-Q, a survey that screens for: the presence of menstrual dysfunction; impaired bone health; exercise level; dietary habits; and eating disorders and disordered eating. Secondary hypotheses are: 1. the total score in the FED-Q will correlate with resting metabolic rate ratio (RMRratio) and total triiodothyronine concentration; 2. the specific scores for menstrual dysfunction and bone health will be associated and correlated, respectively, with menstrual status and bone mineral density. Our null hypothesis is: a questionnaire that screens for the presence of menstrual dysfunction, impaired bone health, exercise level, dietary habits, and eating disorders and disordered eating is not an accurate predictor of energy deficiency in exercising women.

4. REVIEW OF THE LITERATURE

Low EA and ED can be caused by high exercise energy expenditure that is not sufficiently compensated with energy intake. EA is calculated as the difference between exercise energy expenditure and dietary energy intake per kilograms of fat-free mass per day ([EEE-DEI]/kg FFM/day). Low EA is defined as EA < 30 kcal/kg FFM/day. Persons with low EA are likely to be energy deficient, a common condition among female athletes and non-athlete exercising women. There is no gold-standard for the assessment of ED, but serum triiodothyronine (TT3 <80 ng/dL) and resting metabolic rate ratio (RMRratio = RMRmeasured/RMRpredicted <0.9) have been used as parameters of ED (Strock et al., 2020a; Strock et al., 2020b; De Souza et al., 2014).

When an individual is energy deficient, their organism does not have enough fuel for all basic physiological needs, and the systems that are most important to survival are prioritized. Reproduction is one of the first functions to be impaired because of low EA/ED (De Souza et al., 2014). Previous studies have reported high prevalence of menstrual dysfunction in exercising women, which was higher than that observed in sedentary women (De Souza et al., 2010; De Souza et al., 1998). One study comparing active and sedentary females found that more than 50% of the participants in the exercising group were anovulatory, condition not observed in any of the subjects in the sedentary group (De Souza et al., 2010). In addition, anovulatory exercising women have been reported to have lower energy intake compared to both active and sedentary eumenorrheic women, evidencing the relationship between inadequate fueling and menstrual dysfunction (De Souza et al., 1998).

Growth is also impaired in women with low EA, which, in more severe cases, may impact bone mineral density (BMD). Compared to women with normal BMD, females with low BMD were reported to exercise more and to have lower body fat percentage and fat mass. Age of menarche was found to be higher, which could be indicative of delayed menarche and menstrual dysfunction (Gibbs et al., 2014). Menstrual dysfunction and impaired bone health as consequences of low EA are the components of the Female Athlete Triad (Triad), which can include more severe cases with low BMD and eating disorders or disordered eating (De Souza et al., 2014).

As low EA and ED are the main cause for the impaired health outcomes, early and precise detection of low EA/ED risk is needed. Measurements of EEE and DEI – used to calculate EA – depend on self-reported data and depend on predictive equations, hence assessing EA is less accurate than measuring ED. In a study comparing exercising women with and without functional hypothalamic amenorrhea, the first group was found to have lower ED, but similar EA, indicating that the measures used to assess ED are more sensitive to altered metabolic function than EA (Strock et al., 2020c). However, for clinical assessments, examining TT3 and RMRratio may not be feasible methods, in terms of cost, time, and access. Blood serum analyses and the use of a metabolic analyzer are not rapid measures and may not be easily accessed by all physicians, dietitians, nutritionists, or athletic trainers.

Aiming to provide sports and health care professionals with an appropriate tool, studies presented different questionnaires that attempted to estimate low EA. While some of them were developed specifically for this purpose, other surveys that were tested are intended for screening different conditions that could be related to EA and the Triad. These are related to eating disorders, dietary habits, and body image. Of all available questionnaires, only a few have gone through a complete validation process and only one has been validated for use with adult female athletes (Sim & Burns, 2021).

To our knowledge, the Low Energy Availability in Females Questionnaire (LEAF-Q), proposed by Melin et al. (2014) is currently the only validated survey to estimate low energy availability in female athletes. Even though it seems to be a promising tool, the LEAF-Q has several limitations. A cut-off value (score ≥ 8) was used to determine if a participant was at risk of developing the Triad. The LEAF-Q was considered to have accurate sensitivity if it had detected any of the three Triad-related outcomes – low EA, menstrual dysfunction, and impaired bone health –, therefore neither ED nor low EA were considered strictly necessary to correctly identify a positive case. This raises concerns, because positive cases of low BMD or menstrual dysfunction do not necessarily mean that those were caused by low EA. More importantly, there were no significant differences in EA between groups above and below the cut-off point, meaning that the LEAF-Q can lead to concerning misinterpretations. In addition, the study of Rogers et al. (Rogers et al., 2021) showed that the LEAF-Q has a high sensitivity, but a very low specificity in the assessment of risk of Triad outcomes, therefore very low positive predictive values. The authors also found the score not to be related to RMRratio, indicating that it is likely inappropriate for assessing EA or ED.

Given the evidence in literature, it has been observed that there are no clinically reliable and valid tools to assess ED in exercising women. A questionnaire with the purpose of assessing low EA and/or ED must assess menstrual status, dietary habits, exercise level, and bone health. For better accuracy, validation should be assessed with correlations not only with EA, but mainly with RMRratio and serum TT3.

5. METHODS

a. Subject Characteristics Subjects will be women from 18 to 35 years old who have exercised at least 150 min per week in the

last 12 months. Both athletes and non-athletes will be included. Participants must be in good health and free of chronic diseases. Participants will be excluded if they are pregnant, lactating, taking hormonal contraceptives, taking medication that could alter calcium metabolism, or if they are recovering from bone injury at the time of the study. Participants will be recruited via flyers, emails, Study Finder, and the laboratory website.

b. General study design A questionnaire to assess ED in exercising women will be developed. An expert committee will be

formed to discuss the first draft of the survey. It will be pilot-tested in a small group of exercising women, when feasibility will be assessed. The expert committee will perform a second analysis, which will establish content validity. For reliability and construct validity, a larger group of exercising women will respond the questionnaire. The validation will be performed by correlating the final scores of the FED-Q with TT3, RMRratio,

menstrual status, BMD z-score, DEI, and EEE. Sensitivity and specificity of predicting low TT3 and RMRratio will be assessed. Reliability will be separated into stability and internal consistency analyses.

c. Specific study methods

The study will follow previously described steps of questionnaire development and validation (Boparai et al., 2018; Tsang et al., 2017) specified as follows. Approximately sixty questions will be formulated aiming to assess menstrual status, bone health, exercise level, dietary habits, and disordered eating/eating disorders, of which approximately 20 items are expected to remain in the final draft of the questionnaire. The FED-Q will be designed to be a self-administered tool containing close ended questions, with either Likert scale or yes/no answers. Demographics such as age, height, weight, and body mass index will be assessed as well. A committee of three to ten experts in the field of EA/ED/Triad will be formed to have a primary discussion about the suitability of the FED-Q in assessing ED. Following their comments on the first draft, modifications will be made in the items.

Subsequently, the FED-Q will be pilot-tested for feasibility in a group of thirty subjects. Immediately after the participant has responded the questionnaire, a member of the research team will read out the items to the subject, aiming to verify if the questions were clear and if they could be easily interpreted, as expected. Respondents’ opinion on specific questions may be asked by the interviewers. Based on the observations in the pilot testing, new adjustments in the FED-Q will be addressed. After modifications, content validity will be assessed. The expert committee will be again invited to analyze the extent to which the FED-Q comprises most of the dimensions of EA and ED in exercising women. In this step, each expert will evaluate the items without consulting others’ opinion. The Content Validity Index (CVI) will be used to measure the relevance of each item in the FED-Q (Boparai et al., 2018; Polit & Beck, 2006). This analysis is detailed in the Data Analytical Approach section. Other modifications may be addressed to the draft, after calculating the CVI.

Following the three rounds of modifications, scores will be attributed to each item. Responses that are not attributed to risk of ED will not add any points to the overall score, whereas those that might indicate risk of ED will add at least one point to the final score. Higher scores will be attributed to the items that, according to the literature and to expert opinion, are more critical indicators of ED.

The final draft of the FED-Q will be tested in a second group of exercising women for construct validity, followed by a reliability assessment, summarized in Figure 1. Two visits will be required. For construct validity, several measurements will be taken from the study group. In the first visit, RMR will be assessed with a ventilated hood system of indirect calorimetry (SensorMedics Vmax Series, Yorba Linda, CA) for subsequent estimation of the RMRratio (Cunningham 1991 equation and DXA ratio) (Strock et al., 2020a), body composition and BMD will be assessed with a Hologic QDR4500W DXA scanner (Hologic, Bedford, MA), and blood draws will be taken to assess TT3. On the first seven consecutive days between the first and the second visit, participants will complete nutrition logs and monitor exercise with a wearable device. Data will be used to calculate DEI (Nutrition Pro Software), EEE, energy balance (energy intake – energy expenditure), and EA ([EEE-DEI]/kg FFM/day). Starting on the first visit, urine samples of two consecutive menstrual cycles will be collected and provided by the participants to measure Ed1 and PdG excretion for an estimation of menstrual status. Construct validity will be assessed with statistical tests of correlation and association between the questions and the specific measurements that they are intended to measure, and mainly with the correlation of the total score with TT3 and RMRratio. A cut-off value will be set as predictor of ED (TT3 < 80 ng/dL and/or RMRratio < 0.9), and sensitivity and specificity will be assessed.

Eight to ten days following the first visit, subjects will come to the laboratory for the second visit, when they will provide the research staff with the exercise monitors and nutrition logs, and a second part of the urine samples. In this occasion, they will respond the FED-Q one more time. Reliability will be attested in terms of stability and internal consistency. Stability will be measured using a test-retest method. The responses given to the FED-Q will be compared. Little variability is expected between the two time points. Internal consistency will be measured with a split-half method. Details of this analysis are provided in the Data Analytical Approach section. The remaining urine samples will be delivered to the research team by the end of the first and second months of collection.

Figure 1. Summarized data collection for the assessment of construct validity and reliability of the Female Energy

Deficiency Questionnaire.

d. Data analytical approach, including details about your statistical model, assumptions

Validity and reliability assessments will require different statistical analyses, specified below. Content validity The expert committee will be instructed to use the Content Validity Index (CVI). They will be asked to

classify each item of the FED-Q from one to four points, in a Likert-type scale, where one means the item is irrelevant to assess ED; two, somewhat relevant; three, quite relevant; and four, highly relevant. The CVI for items (I-CVI) will be calculated as the percentage of experts who consider an item as relevant. i.e., if they award an item with minimum three points in the Likert scale. Items with an I-CVI of at least 80% will be considered acceptable. Inappropriate items will be reviewed or excluded (Boparai et al., 2018; Polit & Beck, 2006). The average CVI for scales (S-CVI/Ave) will be calculated as the average of all I-CVIs, which is also expected to result in a minimum of 80% (Polit & Beck, 2006).

Construct validity Data of TT3, RMRratio, BMD, body composition, EA, energy balance, and FED-Q score will be tested

for normal distribution. In case of not normally distributed, data will be log-transformed. If normal distributions are found, parametric tests will be used, otherwise non-parametric statistics will be applied. Pearson’s correlation coefficient or Spearman’s rho will be used to verify if the FED-Q score is correlated with TT3 and RMRratio, aiming to validate the questionnaire. Correlations will also be used to test if specific questions and items are related to the data that they are expected to screen for – BMD, menstrual dysfunction, exercise level, and DEI. Of the total score, a cut-off point will be determined as indicator of risk of ED, which will be used to classify participants as either at risk of having ED or not. Sensitivity and specificity in predicting ED (TT3 < 80 ng/dL and/or RMRratio < 0.9) will be assessed, and all variables will be compared between groups with Student’s T test or Mann-Whitney test.

First visit FED-Q, RMR, DXA, TT3, urine

Second visit FED-Q, exercise and nutrition

data, urine samples

Reliability

Data of first and second visits

Stability

Correlation coefficient of first and second visit

Internal consistency

Split-half method

Construct validity

FED-Q of first visit and physiological data

8-10 days interval

Wearable exercise monitor, nutrition logs, urine collection

Reliability

Stability of the FED-Q will be tested as the correlation coefficients between the scores of the first and the second visit. The split-half method will be used to measure internal consistency: the questionnaire will be divided into two halves with equivalent number of questions of each domain – DEI, menstrual dysfunction, exercise level, bone health, and eating disorders. A correlation coefficient will indicate if the questionnaire is consistent in screening for the same conditions (Tsang et al., 2017).

e. Detailed sample size calculations

Literature provides several different sample size requirements to validate a questionnaire. The lowest acceptable value seems to be 5:1, or 5 respondents for each item. Since the FED-Q is expected to have approximately 20 items, a minimum sample size would be of 100 exercising women (Tsang et al., 2017).

Another approach is to calculate the sample size based on the expected confidence level and margin of error, as follows (Ott & Longnecker, 2015):

𝑛 = 𝑍! " . 𝜎"

𝜀"

where n is the sample size, Z is the z-score for a confidence level of α/2, σ is the expected standard

deviation, and ε is the margin of error. Considering a 95% confidence interval with an average standard deviation of 11.6 ng/dL for TT3 in

similar populations (Strock et al., 2020a; 2020b; 2020c), and an acceptable margin of error of 2 ng/dL, the minimum sample size for the phase of validation is 132 exercising women. Applying a drop-out rate of 20%, the minimum sample size is 165. In addition to the 30 subjects in the pilot-testing phase, our sample size will be of 195 exercising women.

f. Timeline

Figure 2 indicates the timeline of all necessary phases for the completion of the study. Pilot testing is

expected to be finished by April 2022, allowing for analysis by the experts to be finished in the first week of June, and data collection to start in July. Data analysis will start after all subjects have completed the first visit and will be concluded by December 2022. A first draft of the research will be presented in February 2023, followed by a final version in March.

Figure 2. Research phases and timeline

6. Anticipated Results and Interpretation The FED-Q is expected to be a feasible, reliable, repeatable, and valid survey to assess ED in

exercising women. A minimum score, yet to be determined, is expected to predict low TT3 and/or low RMRratio with at least 90% sensitivity and 80% specificity. In addition, scores on menstrual status and bone health will be good indicators of menstrual dysfunction and low BMD. The FED-Q will be a tool that is accessible, rapid, simple to use, and easy to interpret. Its use will be intended for physicians, dietitians, nutritionists, coaches,

athletic trainers, and other health care professionals, as well as researchers, as a reliable measurement of ED.

7. Potential Pitfalls, Alternative Explanations, and Consideration of Alternate Approaches

There can be unexpected outcomes in any of the steps of development and validation of the FED-Q. If inconsistencies are observed during either the feasibility, reliability, or the validity steps, the tool will have to be re-evaluated. If the inconsistencies are specific to certain items, these will have to be thoroughly analyzed, and possibly replaced or removed. If major problems occur, such as the questionnaire itself not seeming reliable, going back to the first step of development (question framing) might be needed, which might even raise questions on whether it is possible to assess ED by means of a questionnaire. If it is concluded that the FED-Q is not a reliable, repeatable, and valid tool, other clinical approaches will have to be investigated and developed to be used for ED assessment in exercising women. REFERENCES Boparai, J. K., Singh, S., & Kathuria, P. (2018). How to design and validate a questionnaire: a guide. Current clinical pharmacology, 13(4), 210-215. De Souza, M. J., Miller, B. E., Loucks, A. B., Luciano, A. A., Pescatello, L. S., Campbell, C. G., & Lasley, B. L. (1998). High frequency of luteal phase deficiency and anovulation in recreational women runners: blunted elevation in follicle-stimulating hormone observed during luteal-follicular transition. The Journal of Clinical Endocrinology & Metabolism, 83(12), 4220-4232. De Souza, M. J., Nattiv, A., Joy, E., Misra, M., Williams, N. I., Mallinson, R. J., Gibbs, J.C., Olmsted, M., Goolsby, M., Matheson, G. (2014). 2014 Female Athlete Triad Coalition Consensus Statement on treatment and return to play of the female athlete triad: 1st International Conference held in San Francisco, California, May 2012 and 2nd International Conference held in Indianapolis, Indiana, May 2013. British journal of sports medicine, 48(4), 289-289. De Souza, M. J., Toombs, R. J., Scheid, J. L., O'Donnell, E., West, S. L., & Williams, N. I. (2010). High prevalence of subtle and severe menstrual disturbances in exercising women: confirmation using daily hormone measures. Human reproduction, 25(2), 491-503. Gibbs, J. C., Nattiv, A., Barrack, M. T., Williams, N. I., Rauh, M. J., Nichols, J. F., & De Souza, M. J. (2014). Low bone density risk is higher in exercising women with multiple triad risk factors. Medicine and science in sports and exercise, 46(1), 167-176. Melin, A., Tornberg, Å. B., Skouby, S., Faber, J., Ritz, C., Sjödin, A., & Sundgot-Borgen, J. (2014). The LEAF questionnaire: a screening tool for the identification of female athletes at risk for the female athlete triad. British journal of sports medicine, 48(7), 540-545. Ott, R. L., & Longnecker, M. T. (2015). An introduction to statistical methods and data analysis. Cengage Learning. Polit, D. F., & Beck, C. T. (2006). The content validity index: are you sure you know what's being reported? Critique and recommendations. Research in nursing & health, 29(5), 489-497. Rogers, M. A., Drew, M. K., Appaneal, R., Lovell, G., Lundy, B., Hughes, D., Vlahovich, N., Waddington, G., Burke, L. M. (2021). The Utility of the Low Energy Availability in Females Questionnaire to Detect Markers Consistent With Low Energy Availability-Related Conditions in a Mixed-Sport Cohort. International Journal of Sport Nutrition and Exercise Metabolism, 31(5), 427-437. Sim, A., & Burns, S. F. (2021). questionnaires as measures for low energy availability (LEA) and relative energy deficiency in sport (RED-S) in athletes. Journal of Eating Disorders, 9(1), 1-13.

Strock, N. C., De Souza, M. J., & Williams, N. I. (2020c). Eating behaviours related to psychological stress are associated with functional hypothalamic amenorrhoea in exercising women. Journal of Sports Sciences, 38(21), 2396-2406. Strock, N. C., Koltun, K. J., Mallinson, R. J., Williams, N. I., & De Souza, M. J. (2020b). Characterizing the resting metabolic rate ratio in ovulatory exercising women over 12 months. Scandinavian journal of medicine & science in sports, 30(8), 1337-1347. Strock, N. C., Koltun, K. J., Southmayd, E. A., Williams, N. I., & De Souza, M. J. (2020a). Indices of resting metabolic rate accurately reflect energy deficiency in exercising women. International journal of sport nutrition and exercise metabolism, 30(1), 14-24. Tsang, S., Royse, C. F., & Terkawi, A. S. (2017). Guidelines for developing, translating, and validating a questionnaire in perioperative and pain medicine. Saudi journal of anaesthesia, 11(Suppl 1), S80.