A cross-sectional study of the relationship between community dwelling older adults’ self-perceived frailty and their Electronic Frailty Index score

A cross-sectional study of the relationship between community dwelling older adults’ self-perceived frailty and their Electronic Frailty Index score DATA & FILE OVERVIEW * Date of data collection: May-September 2023 * Geographic location of data collection: A single GP practice in Scotland File List: * Additional related data collected that was not included in the current data package: Data for non-respondents or those with invalid consent forms were securely destroyed (following participant-non respondent comparison). * Are there multiple versions of the dataset? No - this is the single file stored open share long term. METHODOLOGICAL INFORMATION Description of methods used for collection/generation of data: A postal survey (and reminder letter) asked community dwelling, older people (70 years and above), registered with a single GP practice in Scotland, to rate their own frailty by completing different self-report measures. The Scottish Primary Care Research Network (a Scottish Government funded network which facilitates research in primary care) helped identify potential participants at the practice using read codes reflective of the eligibility criteria. One thousand eligible potential participants were randomly sampled using the RAND () function in excel. The inclusion criteria were: registered with the medical practice, aged ≥70 years, able to engage with the documentation in English and able to give informed consent. The exclusion criteria were: significant cognitive impairment or considered to be in the last days of life. These were screened for using GP database Read-codes (reflective of a dementia diagnosis or end of life). The final list of potential participants was reviewed by a GP, who made no further exclusions. Potential participants were randomly sampled using the RAND () function in excel. Surveys were returned between May-September 2023. • Automatically generated eFI scores and their associated frailty severity group were extracted from the GP database (11 May 2023) using the data extraction tool, Scottish Primary Care Information Resource. eFI cut-points: fit = < 0.12, mild ≥ 0.12-0.24, moderate = 0.24-0.36, severe = ≥ 0.36. For methods see paper published. Methods for processing the data: * People involved with sample collection, processing, analysis and/or submission: The Scottish Primary Care Research Network (a Scottish Government funded network which facilitates research in primary care) helped identify potential participants at the practice using read codes reflective of the eligibility criteria. All data were inputted into the excel spreadsheet and analysis performed by Victoria Barber-Fleming. * Data inputting: data from the paper surveys were inserted manually into an excel spreadsheet. The paper surveys were then securely destroyed. * Missing data plan: Primary (agreement) analysis - missing data (self-rated frailty on an ordinal scale/'Q2') resulted in the participant being excluded from the primary analysis. Secondary (accuracy) analyses: missing self-report data resulted in exclusion from that specific accuracy test only. Regression analysis: missing data (self-rated frailty on binary scale/'Q1') resulted in exclusion from the regression calculation. Instrument- or software-specific information needed to interpret the data: * Python (3.12.7) was used as the base for all statistical analyses. The main packages and functions used: Pandas v2.2.2, Numpy v1.26.4, Statsmodels v0.14.2, sklearn v1.5.1, scipy v1.12.0 and v1.16.0, irrCAC package. Packages and functions used for specific statistical analysis: * Demographic calculations: SciPy v1.12.0: stats.shapiro ,stats.kruskal, fisher_exact (SciPy v1.16.0 using Python base 3.12.9), stats.levene, stats.mannwhitneyu, stats.brunnermunzel, , chi2_contingency * Agreement calculations: irrCAC (Gwet, Fergadis. 2021. irrCAC. Available at https://irrcac.readthedocs.io/en/latest/usage/usage_table_data.html, accessed 17/7/25). * Accuracy calculations: sklearn.metrics.confusion_matrix, sklearn.metrics.roc_curve, metrics.auc, confidenceinterval: roc_auc_score (Gildenblat, J. 2023. A python library for confidence intervals. Available at: https://github.com/jacobgil/confidenceinterval, accessed 17/7/25). Code for calculating optimal e FI cut point for self-perceived frailty adapted from (Stack overflow available at https://stackoverflow.com/questions/28719067/roc-curve-and-cut-off-point-python accessed 17/7/25). * Regression calculations: - statsmodels.formula.api: smf.logit model, statsmodels.stats.outliers_influence: variance_inflation_factor. Code for odds ratios and confidence intervals adapted from (Villazon, A. 2021. Logistic regression in Python with Statsmodels. Available at https://www.andrewvillazon.com/logistic-regression-python-statsmodels/ accessed 17/7/25). DATA-SPECIFIC INFORMATION FOR: [frailty_perceptions_data_with_variable_view] Variables/measurements - See excel spreadsheet 'Variable View' for a fuller description of each variable including missing data codes. * Number of variables: 14 * List of variables: Survey data: 'Q1', 'Q2', 'Q3', 'PRISMA-7', 'HADS', 'Help'. Demographic data: 'Sex', 'Age', 'Frailty group' (as defined by Electronic Frailty Index [e FI]), 'current e FI', 'KIS' (did the person have a Key Information Summary), 'ACP' (did the person have an Advanced Care Plan), 'Involved in QIP work' (had the person been involved in frailty quality improvement work at the practice), 'SIMD Qunitile' (Scottish Index of Multiple Deprivation). * Number of cases/rows: 375 * Specialized formats or other abbreviations used: SIMD Quintile - data is from 2020v2, each quintile represents 20% of the datazones in Scotland with quintile 1 containing the 20% most deprived data zones (Scottish Government, 2020). Participants completed the following self-report items: * a binary rating (variable label 'Q1') (‘Do you think you are living with frailty? – yes/no’). * an ordinal rating (variable label 'Q2') (‘Please rate yourself on the following scale in relation to frailty: fit/mild/moderate/severe frailty’). * Self-Reported Health ('Q3'): was rated in response to the question 'how would you rate your health status on a scale from 0 to 10?’. A score of ≤6 was adopted as the cut point for frailty. * PRISMA-7 (Raîche et al., 2008) is a seven-item, self-report questionnaire, designed originally to identify older adults with disability. The PRISMA-7 score is calculated by counting positive answers (marked as ‘yes’). A score of ≥ 3 was adopted as the cut point for frailty. * The Hospital Anxiety and Depression Scale (HADS) (Zigmond & Snaith, 1983) screened for potential anxiety and depression among participants. The scale comprises a total of 14 questions, each of which is scored 0-3. There are two sub scales (HADS-D, which asks questions to identify depression and HADS-A which asks questions to identify anxiety. Participants who scored ≥ 8 on either HADS-D or HADS-A were categorised as having depression or anxiety respectively. Missing HADS data was imputed using the individual participants’ sub-scale mean value. HADS summed scores for sub scales: Depression 'sum scores' were calculated by summing the values relating to the following questions ["HADS 2. I still enjoy the things I used to enjoy","HADS 4. I can laugh and see the funny side of things", "HADS 6. I feel cheerful", "HADS 8. I feel as if I am slowed down", "HADS 10. I have lost interest in my appearance","HADS 12. I look forward to things with enjoyment","HADS 14. I can still enjoy a good book or radio or TV programme"]. Anxiety 'sum scores' were calculated by summing the values relating to the following questions ["HADS 1. I feel tense or 'wound up'","HADS 3. I get a sort of frightened feeling as if something awful is about to happen", "HADS 5. Worrying thoughts go through my mind", "HADS 7. I can sit at ease and feel relaxed", "HADS 9. I get a sort of frightened feeling like 'butterflies' in the stomach", "HADS 11. I feel restless as if I have to be on the move", "HADS 13. I get sudden feelings of panic"] Participants were asked to indicate if they had help completing the survey. This is captured by the variable labeled 'Help'. 'Frailty group' represents categories of frailty as defined by the Electronic Frailty Index (e FI) (Clegg et al., 2016). e FI Cut points: fit = < 0.12, mild ≥ 0.12-0.24, moderate = 0.24-0.36, severe = ≥ 0.36 (Devereux et al., 2019) References Clegg, A., Bates, C., Young, J., Ryan, R., Nichols, L., Ann Teale, E., Mohammed, M. A., Parry, J., & Marshall, T. (2016). Development and validation of an electronic frailty index using routine primary care electronic health record data. Age and Ageing, 45(3), 353-360. https://doi.org/10.1093/ageing/afw039 Devereux N, Ellis G, Dobie L, Baughan P, Monaghan T. Testing a proactive approach to frailty identification: the electronic frailty index. BMJ Open Quality. 2019;8(3):e000682. https://doi.org/10.1136/bmjoq-2019-000682. Raîche, M., Hébert, R., & Dubois, M.-F. (2008). PRISMA-7: A case-finding tool to identify older adults with moderate to severe disabilities. Archives of Gerontology and Geriatrics, 47(1), 9-18. https://doi.org/https://doi.org/10.1016/j.archger.2007.06.004 Scottish Government. 2020. Scottish Index of Multiple Deprivation 2020v2 postcode lookup file. https://www.gov.scot/publications/scottish-index-of-multiple-deprivation-2020v2-postcode-look-up/ Zigmond, A. S., & Snaith, R. P. (1983). The Hospital Anxiety and Depression Scale. Acta Psychiatrica Scandinavica, 67(6), 361-370. https://doi.org/https://doi.org/10.1111/j.1600-0447.1983.tb09716.x * Information about funding sources that supported the collection of the data: Victoria Barber-Fleming’s role in this research was funded by the Legal & General Group (research grant to establish the independent Advanced Care Research Centre at University of Edinburgh). The funders had no role in conduct of the study, interpretation, or the decision to submit for publication. The views expressed are those of the authors and not necessarily those of Legal & General.