To the Editor:
We read with interest the study by Sonny et al.1 assessing frailty using the Hopkins phenotypic scale and of the health deficit–based modified frailty index in 1,042 patients undergoing noncardiac surgery. We are particularly interested in the finding that neither approach to diagnosing frailty was a clinically useful predictor of unexpected prolonged hospital stay, readmission, and serious complications. We agree that many previous studies of frailty have not adequately examined both odds ratios and predictive value of frailty assessments. However, we are concerned about the validity of the modified frailty index as a tool to measure frailty. Sonny et al. found only a weak correlation between the modified frailty index and Hopkins frailty scales: r = 0.28.
The modified frailty index was derived by identifying National Surgical Quality Improvement Program variables that aligned with the 70-item Canadian Study of Health and Aging Frailty Index.2 The resulting scale involves only 11 variables: 9 are medical comorbidities, and only 2 are alternative health deficits: “functional status–not independent” and “impaired sensorium.” This contrasts with the 2012 Frailty Consensus Conference, which defined six health domains affected by frailty that should be included in a frailty scale: physical performance, gait speed, mobility, nutritional status, mental health, and cognition.3 Few of these attributes are contained within the modified frailty index. Sonny et al. also reference the work of Searle et al.4 in describing the 10-variable threshold below which frailty estimates are unstable; this article, however, clearly states that at least 30 age-related deficits must be included across the spectrum of health when constructing an accurate cumulative deficit frailty index. It is also instructive that the majority of health deficits contained in the original Rockwood frailty index are nonmedical comorbidities (42 of 70 variables).5 We thus question whether the modified frailty index can be truly regarded as a measure of frailty or whether it is more an index of comorbidity. If this were the case, it is still unsurprising that the modified frailty index has been associated with adverse postoperative outcomes given the link between perioperative comorbidity and postoperative complications and mortality. In support of this proposition—that the modified frailty index is predominantly a comorbidity index—Sonny et al. reported higher Charlson comorbidity index values in patients diagnosed with frailty with the modified frailty index (median, 6; interquartile range, 4 to 8) compared with those diagnosed as frail with the Hopkins scale (median, 3; interquartile range, 2 to 6).
Although the modified frailty index is an understandably attractive measure to derive from the National Surgical Quality Improvement Program data sets with 11-point simplicity, we question its validity as a frailty tool because of the small number of items, the overrepresentation of comorbidity, and the under representation of other health domains. Caution must be exercised in its application as a proxy for frailty without validation research.
The authors declare no competing interests.