Experimental design and study population
It is a prospective cohort study to investigate the association between preoperative frailty and postoperative adverse outcomes after elective spine surgery. The trial was approved by the Ethics Committee of Shengjing Hospital under the ethical number 2021PS511K and registered with the China Clinical Trial Registry under the registration number ChiCTR2100049677.
The study population consisted of elderly patients undergoing elective spinal surgery at Shengjing Hospital. A comprehensive assessment was performed using the FRAIL scale, and clinically relevant data were obtained through a combination of preoperative visits, intraoperative monitoring, 30-day postoperative telephone follow-up, and a review of electronic medical records. Inclusion criteria included 1) elderly patients scheduled for elective spinal surgery, 2) age ≥ 65 years, 3) anesthesia was general anesthesia, and 4) patients and family members gave informed consent and were able to cooperate with the preoperative questionnaire and postoperative follow-up. Exclusion criteria included: 1) severe hearing, reading, or speech communication disorders that prevented normal communication; 2) dementia or severe cognitive dysfunction that prevented them from cooperating with the completion of the survey; 3) a history of severe mental illness that prevented them from refusing to or being able to cooperate with the study; and 4) refusal to participate by the patient or family members.
Frailty and outcome assessment
Patients were assessed for frailty on admission by an anesthetist (who was not involved in the follow-up study) using the FRAIL scale, a validated measurement tool recommended by the Clinical Practice Guidelines, which provides an understanding of the patient's level of frailty by asking simple questions [16]. It consists of five main questions with scores ranging from 0–5, with a maximum score of 5. A score of ≥ 3 is defined as frailty, a score of 1–2 is defined as prefrailty, and a score of 0 is considered healthy [15]. Higher scores often indicate a greater degree of frailty (more details in supplementary file 1).
The other two investigators ascertained the regression of the disease through postoperative visits, medical records, and telephone communication within 30 days after the operation. It is worth noting that these two investigators were not aware of the patient's preoperative frailty level. The primary outcome of this study was the occurrence of postoperative complications. Complications of interest included pulmonary infection, urinary tract infection, surgical incision infection, cardiovascular system complications, electrolyte disturbances, postoperative blood transfusion, and postoperative feeding difficulties.
Postoperative complications were also assessed using the Comprehensive Complication Index (CCI), a new method of evaluating postoperative complications based on the Clavien-Dindo classification (CDC) system [17, 18]. The CDC system was proposed in 2004 and divided into five grades according to the magnitude of the intervention to correct a given postoperative complication. The CDC system describes the likelihood of all complications, but in clinical studies, for ease of handling, only the highest grades of complications are usually reported. In other words, the CDC grading system focuses only on the most severe outcomes, whereas the CCI focuses on all postoperative outcomes and the number and severity of each complication. We obtained each participant's 30-day postoperative CCI through the online calculation form (available at www.assessurgery.com). Secondary outcome indicators for this study included prolonged hospital stay and adverse postoperative course. Prolonged hospital stay was defined as the length of hospital stay exceeding the 75th percentile of the cohort. The CCI being greater than the 75th percentile of the total population was seen as a deviation from the normal postoperative outcome of regression, and therefore was taken as a surrogate for adverse postoperative course.
Collection of other clinical data
All patients' spinal surgeries were performed at Shengjing Hospital, and patients were managed under anaesthesia using a standardised anaesthesia protocol when they entered the operating theatre. Besides the assessment of frailty and postoperative outcome indicators, other relevant data were routinely collected during the perioperative period in this study. The main indicators collected included: age, gender, BMI, ASA classification, Basic Activities of Daily Living (BADL), albumin level, duration of surgery, invasive size of the surgery, bleeding, fluid replacement, urine output, whether or not blood was transfused intraoperatively and the amount of blood transfused.
The BADL is a means of assessing the ability to perform activities necessary to maintain basic living and survival, and the main items include eating, bathing, grooming, dressing, controlling bowel movements, going to the toilet, bed, and chair mobility, walking, and walking up and down stairs. A score out of 100 is considered excellent for the ability to perform activities of daily living without dependence on others. A score greater than 60 is assessed as good, and a score less than or equal to 60 requires varying degrees of care from others. We graded the magnitude of invasiveness of spinal surgery using a rating system: grades 1 and 2 consisted mainly of microdiscectomy, lumbar laminectomy or anterior cervical surgery, and minimally invasive fusion; grades 3 and 4 consisted mainly of lumbar fusion, traumatic injury, or posterior cervical fusion, tumor, infection, deformity, or combined anterior and posterior cervical spine surgery. In the context of this study, levels 1 and 2 were defined as less invasive, and levels 3 and 4 as more invasive.
Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics for Macintosh (version 27.0). Categorical variables were expressed as numbers (percentage), continuous variables obeying normal distribution were expressed as mean and standard deviation (SD), and continuous variables disobeying normal distribution were expressed as the median and interquartile range (IQR). The Kruskal-Wallis test and chi-square test were used to compare the variability of different frailty subtypes. Further two-by-two comparisons were performed when the p-value was < 0.05, and p-values were corrected using the Bonferroni method.
In the analysis of factors influencing the postoperative-related outcome indicators, we used univariate analysis to screen out meaningful variables and then performed multifactorial logistic regression analysis to find independent risk factors associated with the occurrence of the study outcomes. The Hosmer-Lemeshow test was used to evaluate the superiority of fit of the model. When the p-value was > 0.05, it was judged that there was no significant difference between the predicted and true values. In addition, we used Graphpad Prism 8 to prepare forest plots for Logistic regression analysis.
Sample size calculation: this experiment was to investigate the relationship between 3 different levels: healthy, prefrailty, and frailty, and the occurrence of postoperative complications. By reviewing the previous literature [19], the complication rates of the healthy group, prefrailty group, and frailty group were 10.42%, 31.62%, and 47.17%, respectively. With an α of 0.05,1-β of 0.1, 173 patients were planned to be included, and the sample size was at least 204, taking into account a 15% loss-to-follow-up rate.