The Impact of Frailty on Spine Surgery: Systematic Review on 10 years Clinical Studies

Frailty is a condition characterized by a high vulnerability to low-power stressor. Frailty increases with age and is associated with higher complications and mortality. Several indexes have been used to quantify frailty. Spine diseases, both degenerative and oncologic, frequently require surgery which is related to complications and mortality. Aim of the present systematic review was to collect the most frequently used frailty indexes in clinics to predict surgical outcomes in patients affected by spine diseases, taking into account gender differences. Three databases were employed, and 29 retrospective clinical studies were included in this systematic review. The identified spine pathologies were primary and metastatic spine tumors, adult spine deformity (ASD), degenerative spine disease (DSD), cervical deformity (CD) and other pathologies that affected lumbar spine or multiple spine levels. Eleven indexes were identified: modified Frailty Index (mFI), Adult spinal deformity frailty index (ASD-FI), mFI-5, Metastatic Spinal Tumor Frailty Index (MSTFI), Fried criteria, Cervical deformity frailty index (CD-FI), Spinal tumor frailty index (STFI), Frailty Phenotype criteria (FP), Frailty Index (FI), FRAIL scale and Modified CD-FI (mCD-FI). All these indexes correlated well with minor and major postoperative complications, mortality and length of stay in hospital. Results on gender differences and frailty are still conflicting, although few studies show that women are more likely to develop frailty and more complications in the post-operative period than men. This systematic review could help the surgeon in the adoption of frailty indexes, before the operation, and in preventing complications in frail patients.

consensus on distinctive diagnostic criteria is still missing.
It is reported that the prevalence of frailty increases with age, from 4% for ages between 65 and 69 years to 26% for older than 85 [6] and it is more frequent in females than in males [6,7]. The worldwide occurrence of frailty varies extensively between 4% and 59% due to the heterogeneity of study populations and the use of different screening tools that consider different criteria [8,9].
The dramatic increase in old-aged population is one of the main concerns. According to the United Nations, the proportion of global population over 65 years of age is expected to rise from 9% in 2019 to 16% by 2050 [10]. With increase in life expectancy, chronic noncommunicable diseases have become prevalent together with a rising number of elderly patients affected by degenerative, traumatic, oncologic or infective pathologies.
These demographic and epidemiologic transitions have a deep impact on health care provision and economic burden. A recent study from Norway highlights that patients over 65 years represent only 15% of the population, but are responsible for almost half of the total healthcare cost [11]. Moreover, in a prospective cohort study from US, pre-frailty and frailty are associated with higher subsequent total healthcare costs in older community-dwelling men [12].
Thus, the preservation of independence in aged people and the prevention of disability are priority major challenges and frailty is becoming an increasingly important concept both for its deep impact on health outcomes and impaired quality of life.
Frailty is associated with increasing disability, hospitalization, adverse health outcomes and death [1]. A number of observational studies have also shown that frailty worsened postoperative outcomes as morbidity, mortality and length of stay (LOS) [13][14][15] and the severity of frailty syndrome has been reported to be directly correlated with post-surgical mortality rates and complications [16].
As frailty is correlated to general surgery outcome, it might also predict the outcome in patients undergoing spine surgery, who have been reported to record a high rate of intra-and post-operative complications [17][18][19][20]. Degenerative disorders of the spine are very frequent, with 90% of adults showing some degree of degeneration of the lumbar disk or spine by age 50 [21]. Degeneration of the spine includes a wide variety of clinical conditions, as disk degeneration, spinal stenosis, facet hypertrophy, osteophytosis, foraminal stenosis and instability, leading to back pain and/or associated neurological signs [21]. Back pain affects 15%-20% of adults each year [22] and about 17.000 new cases of spinal column injuries are reported every year in US [23]. In addition, vertebral fractures are the most frequent osteoporotic fractures among aged people, together with proximal femoral and wrist fractures [24]. Spinal metastases affect 30-70% of patients with primary tumors and can lead to spinal cord compression, pain, spinal instability and pathologic fractures [25].
Hypothesis of the present study is that frailty may play a key role in the outcome of spinal surgery and may provide a useful tool for risk prediction, facilitating the decision-making process and surgery planning in patients affected by spine disease. To the best of our knowledge, only one systematic review, on associations between frailty and spine disease, has yet been conducted [26].
The aim of this study was to systematically review 10-year clinical data regarding association between frailty and outcomes after surgery for spine disease, by identifying the most used frailty indices in spine surgery.

MATERIALS AND METHODS
Three databases were employed to individuate clinical studies included in the present systematic review: www.pubmed.com, www.webofknowledge.com and www.scopus.com (Fig. 1).
In the Pubmed database the search was performed with the following meshes: ((("Spine" In the Web of knowledge database, the search was performed with "(spine disease) AND (frailty)" keywords and the limits were English language, article document type and timespan 2010-2019, founding 25 studies.
In the Scopus database, "(spine disease) AND (frailty)" were also employed as keywords with the following limits: English language, article type and 2010-2019 years of publication and 70 studies were obtained. Relevant studies were firstly screened through title and abstract by one author (FV) and the studies that did not match the argument of the review were excluded. In the first database, 24 studies were included and 44 In the first database, 24 studies were included and 44 excluded because not inherent: they regarded other pathologies (17 studies), computational or surgery or rehabilitation methodology (5 studies), not regarded frailty (15 studies) or they were reviews (5 studies) and animal studies (2 studies). In the second database, 19 studies were excluded because regarded: 1) only sarcopenia (5 studies), no frailty (6 studies) or other orthopedic sites other than spine (8 studies). In the third database, 49 studies were excluded because they regarded only tumors (1 study), no frailty (18 studies), other orthopedic sites other than spine (28 studies) or they were reviews (2 studies).
Fifty-one studies were accepted and then submitted to www.mendeley.com to eliminate duplicates, finding 26 studies in common.
The full text of the remaining 25 studies were examined by two authors (FV, VB) and the studies characteristics and results were summarized in Table 1. Finally, an additional search was performed by reading the reference lists of the 25 studies, founding further 4 studies.
So, a total of 29 clinical studies were included in the present systematic review (Fig. 1).

Assessment of Methodological Quality
Risks of bias assessments for each study were indicated in Table 2. Most of the studies showed an overall risk of bias low or moderate (n = 23 studies, 79,3%). Only a fraction of the studies (n = 3 studies (10,3%) [29,30,52] had a high risk, due to the lack of information for at least one aspect of the study attrition item [29,30] or in analysis items, showing no statistical analysis [52]. All studies showed a low outcome measurement item because all studies had well described outcome measurement with a clear definition of the outcome, valid and reliable outcome measurements and the same method and setting of outcome measurement for all study participants. In addition, for all studies the Confounding Measurement and Account item was always moderate because the observed effect of the prognostic factors on outcome may be distorted by another factor related to the outcome.

Frailty assessment in spine tumor studies
Three different frailty indices were used for the identification and evaluation of frailty in patients affected by primary or metastatic spine tumors ( Table 3), all of them are based on the accumulation of deficit model suggested by Rockwood.  [29,31] were respectively used in benign or malignant primary spine tumors [28] and in spinal metastases with a primary tumor located in breast, lungs, thyroid, kidney or prostate [29,31]. Both indices grouped patients into 4 frailty categories: no frailty (0), mild frailty (1), moderate frailty (2) and severe frailty (≥3).
Modified Frailty Index (mFI) was employed in patients affected by spinal metastasis with a primary tumors in prostate, unknown sites, breast, lung, bladder, kidney, cervix, thyroid [30,31] or primary and metastatic spine tumors in extradural, intradural extramedullary and intramedullary locations [32]. The cut-off for not frailty is 0, that for pre-frailty is 0-0.21 and that for frailty is ≥ 0.27.

Results in spine tumor studies
Mild, moderate and severe frailty significantly increased all complications and LOS in hospital in 1589 patients with age between 28 and 61 years [28]. Similarly, moderate and severe frailty significantly increased mortality, while mild, moderate and severe frailty were associated with major complications and LOS in 4583 patients (age 54-70 years) [29].
In 41 [30], 108 [31] and 2170 [32] patients, with a mean age of 60 years, one group of authors did not find correlation between frailty status and mortality after surgery [30], while other authors showed that frail patients had higher mortality and LOS than not frail ones [32]. MSTFI was also compared with mFI, underling that mFI correlated with complications, while MSTFI with mortality [31].
Frailty significantly increased mortality rate, blood transfusion, pulmonary embolism/deep vein thrombosis (PE/DVT), any postoperative complications and reoperation rate than not frailty [39]. It was also observed that frailty significantly increased the absolute changes in postoperative Oswestry Disability Index (ODI), 36-Item Short Physical Component Summary (SF36 PCS), leg pain and the proportion of patients that reached substantial clinical benefit for ODI, SF-36 PCS, leg pain score also in comparison to frailty and severe frailty [33].
Prefrail and frail patients in good control group or poorly controlled group experienced more perioperative complications and postoperative C7 sagittal vertical axis (C7SVA) than frail patients [37]. The control group of frailty was defined as treatment following the appropriate guidelines for each mFI factor [37]. Making a comparison between the classic mFI and the truncated form mFI-5 items it was observed an excellent concordance, especially in the prediction of complications. The classic mFI was able to well correlate with total complications, perioperative complications, implant-related complications, while mFI-5 with severe complications [38].
In patients affected by DSD, frailty significantly increased risk of mortality, major complication, reoperation for postsurgical infection, LOS and discharge to a new facility [40,41], even if one study did not find association with the incidence of postoperative complications [40]. A correlation between frailty and body mass index (BMI) in the prediction of postoperative complications was founded [42]. Underweight, obesity, prefrailty and frailty separately increased postoperative complications and underweight significantly increased prefrailty and frailty, while obesity only frailty. In addition, underweight and normal weight associated with pre frailty or frailty, overweight associated with frailty and obese associated with not frailty or frailty increased postoperative complications [42]. The number of patients were 12 [40], 52671 [41] and 1970 [42] with a mean age of 62 years.
An amount of 61 [43] and 121 [44] severe frail patients with a mean age of 61 years, with CD, showed higher major complication, medical complications, overall comorbidity burden, depression and pulmonary disease in comparison to not frail ones [43,44] and cardiac arrest, mortality, deterioration in patient-reported measures of neck pain, neck disability and overall Health related quality of life (HRQoL) and LOS more than not frail and frail patients [44]. In addition, 121 [44] and 138 [45] frail patients with a mean age of 61 years significantly increased vascular complication, superficial surgical site infection, deterioration patient-reported measures of neck pain, neck disability, and overall HRQoL, LOS, minor and major complications than not frail ones [44,45].
In lumbar spine diseases, frailty was significantly associated with increase in any complications, pulmonary complications, mortality, reoperation, LOS, unplanned readmission, several postoperative complications, PE/DVT, sepsis, urinary tract infections (UTI), blood transfusion and wound complications than not frailty [47,48]. In comparison with American Society of Anesthesiologists (ASA) score, mFI was less predictive of postoperative comorbidities, even if both were associated with severe complications, LOS, infectious complications and discharge to higher-level care [46]. Frail patients increased any complications, disability, superficial and deep Surgical site infection (SSI), unplanned reoperation, medical complications (pneumonia, unplanned intubation, postoperative vent use, progressive renal insufficiency, acute renal failure, UTI, cerebrovascular accident (CVA)/stroke, myocardial infarction (MI), bleeding transfusions, sepsis, septic shock), 30-day readmissions, nonhome discharge, disability and poor HQoL more than not frail or pre-frail ones [49,50].
One hundred frail patients of 71 years showed significantly high reduction in cognitive recovery at 3 months after surgery than not frail and pre-frail ones, and pre-frail patients showed significant higher reduction in functional recovery 3 months after surgery not only in comparison to not frail, but also than frail patients [54].
Frailty index (FI) and frailty phenotype (FP) were compared, showing a moderate concordance because both indices observed that adverse outcomes significantly increased with frailty and pre-frailty, but FI was associated with increased discharge to postacute institutional care (PAC) and LOS in pre-frail and frail patients, while FP was associated with discharge to PAC and complications in pre-frailty and increased discharge to PAC, complications and LOS in frailty [55]. The patients were 122 with a mean age of 77 years.

Gender and frailty
Among the 29 studies, 3 addressed gender and its association with frailty or morbidity associated with some spine pathologies, such as ASD [33], DSD [41] and CD [44] (Table 4). Regarding female gender, 2 studies evaluated frailty through ASD-FI [33] and mCD [44], showing contrasting results. More precisely, in 1 study the percentage of not frail women was significantly higher than that of frail and severe frail (p = 0.028) [33], while in the second study the opposite was observed: the percentage of not frail women was significantly lower than frail and severe frail ones [44]. In women affected by ASD, the severity of frailty decreased [33], while in those affected by CD, the severity of frailty increased [44].
As for male gender, in men affected by DSD, the severity of frailty, evaluated with mFI, increased [41], because the percentage of frail men was significantly higher than not frail or pre frail ones (p < 0.0005). In addition, men showed higher major complications, LOS and discharge deposition than women (p < 0.0005) [40].
Frailty is defined as a reduction in physiological function, but it is separated from the concept of aging, leading to the conclusion that the physiological aging is distinct from the chronological one [57]. Frailty increases the health vulnerability and deterioration, especially in elderly and several different tolls are actually used to measure frailty.
Frailty prevalence increases with age and is correlated with disability, admission to hospital and mortality and it is observed an increase of its prevalence in patients undergoing surgery than the other patients (42%-50% Vs 4%-10%) [58]. Frailty is a predictor of morbidity and mortality, more than age in elderly patients undergoing general surgery. Before surgery, the measurement of frailty and the stratification of patients become important for predicting complications, even if no consensus is found on which is the best frailty tool [59].
As the population ages, spine surgery needs to grow, to improve neurologic adverse events and pain. Since spine surgery is associated with complications or even mortality, it is important to select patients at higher risk before surgery, also with a view to reducing the costs of the health system [60,61].
For these reasons, the present systematic review collects the most employed frailty indices able to evaluate the association of frailty and spine surgery outcomes for different spine diseases, to give an indication on which to use in the different cases before surgery.
Frailty indices are composed by items that comprise presence of some concomitant pathologies, the functional status, mood, cognitive capacity and health deficits measured by physician or by the patients. The cut-off that stratify the patients are obtained by dividing the number of the positive items for the total number of the items.
In this review one group of spine pathologies, requiring surgery, is primary or metastatic tumors. The spinal metastasis incidence is between 30 and 70% among patients with primary tumors and 10% of the metastases undergoes surgery [62]. Three frailty indices are used: STFI [45] in primary tumors and MSTFI [46,48] and mFI [47,48,51] in metastatic ones. STFI and MSTFI are correlated with perioperative complications and MSTFI with mortality. Both indices are composed of 9 items that regard the presence of cardiovascular, respiratory, urinary and musculoskeletal system disorders and malnutrition. "Pathologic fracture" and "congestive heart failure" items of STFI are replaced by "emergent/urgent case" and "anterior or combined surgical approach" items in MSTFI.
mFI is the most famous and the most frequently used frailty index in literature also in other pathologies and it consists of 11 variables, that regard non-independent functional status and the history of concomitant pathologies. In spine tumors it is observed that frailty, measured with mFI, is correlated with mortality and complications, even if only one study does not find a correlation between mFI and survival [30]. mFI and its truncated form mFI-5, characterized by 5 items, are also in common in other spine pathologies that required surgery, identified in this review, including ASD [37][38][39], DSD [40][41][42] or patients undergoing PLF [46,48,49], PLIF [46,48], ALIF [47], transforaminal lumbar fusion (TLF) [48], transforaminal lumbar interbody fusion (TLIF) [48], thoracic fractures [51], lumbar, cervical or thoracic procedures [52] and kyphoplasty [53].
ASD and other spine pathologies are usually associated with postoperative or perioperative complications, due to the invasiveness of surgical procedures, including large dissection, multilevel fusion or osteotomy [63].
In these cases, besides mFI, other indices are used as ASD-FI, CD-FI, Fried criteria, FRAIL scale, FI and FP criteria.
ASD-FI, composed of 40 items, is divided into health deficits documented by physician (14 items) and health deficits patient-reported ones (26 items) and is employed in patients affected by ASD [33][34][35][36]. Similar frailty index is CD-FI, employed in patients suffered of CD [43,45], composed by 40 items, health deficits documented by physician (20 items) and health deficits patient-reported ones (20 items). As mFI, also CD-FI possesses its truncated form, that comprises 15 items instead of 40 ones, that take into consideration the presence of diseases, BMI, weakness, anxiety and difficulty in sleeping or in walking [44].
FRIED criteria [50,56] and FRAIL scale [54] are easier than the other ones because they are composed by 5 items and regard prevalently subjective functional performances: weight loss, exhaustion, physical inactivity, slowness and handgrip strength. The differences between the two indices is that FRAIL scale substitutes physical inactivity, slowness and handgrip strength with ability to climb a flight of stairs unassisted, ability to walk two blocks unassisted and medical comorbidities. FP, employed only by one study [55], takes into consideration similar items to FRIED and FRAIL indices. The same study that employed FP, compared it with FI (40 items) [55], that regards the need for help in carrying out daily actions, weight loss, mood and presence of pathologies.
All these frailty indices correlate well with perioperative or postoperative complications, mortality or overall survival, LOS, discharge to a facility that is not home. The most evaluated complications regard the respiratory (acute respiratory distress syndrome, pleurisy, pneumothorax, pulmonary collapse, reintubation, pneumonia, pulmonary embolism), cardiac (cardiac arrest, myocardial infarction, iatrogenic stroke) and urinary (acute renal failure) systems, sepsis, shock, intraoperative vascular, visceral or neurological injury, deep infection, prolonged intubation, return to the operating room, unplanned re-intubation, venous thromboembolism, coma, perioperative blood transfusion, urinary tract infection, wound dehiscence, pseudoarthrosis incidence, excessive bleeding and delirium.
One of the most awaited discussions in the scientific community concerns the method by which complications are collected. Unfortunately, many groups use the most disparate and personalized methods to collect complications, making a comparison difficult and often underestimating the real percentage of the problem. Chen et al. show that another fundamental point in collecting complications lies in the people who collect them. Surgeons underestimate the problem, while a group of external observers, not involved in surgery, seems the best people to collect complications in the right measure [64]. The two most frequent grading methods for complications in vertebral and orthopedic surgery are the Spine Adverse Event Severity System (SAVES) and Clavein-Dindo one. The first allows systematic prospective collection of postoperative adverse events in spinal surgery and is divided into 14 intraoperative and 22 postoperative events, the second is based on the therapy used to correct a specific complication and is divided into 7 grades [65,66]. Only few studies employed these grading methods for complications, one study the SAVES [31], and 8 studies the Clavein-Dindo [37,38,41,42,49,[51][52][53].
The different definitions and classifications of complications by different investigators make difficult to compare studies, dividing complications into major (that lead to reoperation or permanent deficits) and minor ones. So, a limit of the present review could be the different methods used to record complications and the group that evaluates them. For this reason, frailty probably generates more complications than those published.
Most of the studies of the present review validate and evaluate one frailty index for each study. However, in 3 cases, the same study compared two frailty indices [31,38,55]. In patients with spinal metastases of primary tumors located in breast, lung and kidney, mFI and MSTFI were compared, showing that mFI correlated with postoperative complications, while MSTFI with mortality [31]. mFI and mFI-5 showed an excellent correlation across ASD surgery and were strong predictive for severe complications, but mFI correlated with total, perioperative and severe complications, while mFI-5 with severe complications [38]. Finally, a moderate accordance between FP and FI indices was observed. FP correlated with discharge to PAC and complications, while FI with discharge to PAC and LOS [55].
The patients included in the studies varied from a minimum of 41 [30] to a maximum of 52671 [41] and were both men and women, but a prevalence of studies enrolled more women than men [33-37, 39, 43-46, 48-51, 53, 55, 56]. This seems to presage that, between the two genders, there is a prevalence of women who are frail compared to men. Still now, few studies identify gender differences in frailty. Three frailty indices are able to discriminate between males and females, but the results are discordant because frailty severity seems to decrease [33,41] or increase [44] with female gender. More precisely, ASD-FI and mFI show that frailty severity decreases in women affected by ASD and DSD, respectively, and that men shows higher major complications, LOS and discharge disposition than women [33,41]. Conversely, mCD-FI indicates that frailty severity increases more in women than in men affected by CD [44].
However, since only 3 studies dealt with gender difference, with heterogeneity in the study design, study participants, and spine pathologies, it was difficult to draw any significant conclusion regarding this theme.
Frailty is a prevalent age condition, but in this review 14/29 studies considered also patients younger than 60 years [28, 29, 31, 32-39, 41, 42,49]. This reinforces the idea that physiological aging is distinct from the chronological one and that frailty indices can be applied at any age in pathologies of the spine.
Because it is one of the most complete indices, having 11 items that concern the functional status and the history of concomitant pathologies. At the same time, it is also less complex than other indices that may contain up to 40 items.
Clarity has not yet been made regarding the relationship between the frailty level and gender, even if a worsening of frailty is prevalently observed in women. Given the paucity of the studies regarding the comparison between different frailty indices in the same study and of the studies regarding the evaluation of gender in frailty, it will be mandatory to deepen these comparisons in future studies.