Effect of Multimodal Prehabilitation vs Postoperative Rehabilitation on 30-Day Postoperative Complications for Frail Patients Undergoing Resection of Colorectal Cancer

This randomized clinical trial assesses the extent to which a prehabilitation program vs postoperative rehabilitation affects 30-day postoperative complications in frail patients undergoing colorectal cancer resection.

Key Points

Question

To what extent does a prehabilitation program affect 30-day postoperative complications in frail patients scheduled for colorectal cancer resection?

Findings

In this randomized clinical trial including 110 frail patients undergoing colorectal surgery (87 [79.1%] minimally invasive) within an enhanced recovery pathway, prehabilitation did not affect postoperative complications (measured using the Comprehensive Complication Index) compared with postoperative rehabilitation. Other postoperative outcomes were also similar between groups.

Meaning

This study suggests that prehabilitation does not improve postoperative outcomes compared with postoperative rehabilitation in frail patients undergoing colorectal cancer resection.

Abstract

Importance

Research supports use of prehabilitation to optimize physical status before and after colorectal cancer resection, but its effect on postoperative complications remains unclear. Frail patients are a target for prehabilitation interventions owing to increased risk for poor postoperative outcomes.

Objective

To assess the extent to which a prehabilitation program affects 30-day postoperative complications in frail patients undergoing colorectal cancer resection compared with postoperative rehabilitation.

Design, Setting, and Participants

This single-blind, parallel-arm, superiority randomized clinical trial recruited patients undergoing colorectal cancer resection from September 7, 2015, through June 19, 2019. Patients were followed up for 4 weeks before surgery and 4 weeks after surgery at 2 university-affiliated tertiary hospitals. A total of 418 patients 65 years or older were assessed for eligibility. Of these, 298 patients were excluded (not frail [n = 290], unable to exercise [n = 3], and planned neoadjuvant treatment [n = 5]), and 120 frail patients (Fried Frailty Index,≥2) were randomized. Ten patients were excluded after randomization because they refused surgery (n = 3), died before surgery (n = 3), had no cancer (n = 1), had surgery without bowel resection (n = 1), or were switched to palliative care (n = 2). Hence, 110 patients were included in the intention-to-treat analysis (55 in the prehabilitation [Prehab] and 55 in the rehabilitation [Rehab] groups). Data were analyzed from July 25 through August 21, 2019.

Interventions

Multimodal program involving exercise, nutritional, and psychological interventions initiated before (Prehab group) or after (Rehab group) surgery. All patients were treated within a standardized enhanced recovery pathway.

Main Outcomes and Measures

The primary outcome included the Comprehensive Complications Index measured at 30 days after surgery. Secondary outcomes were 30-day overall and severe complications, primary and total length of hospital stay, 30-day emergency department visits and hospital readmissions, recovery of walking capacity, and patient-reported outcome measures.

Results

Of 110 patients randomized, mean (SD) age was 78 (7) years; 52 (47.3%) were men and 58 (52.7%) were women; 31 (28.2%) had rectal cancer; and 87 (79.1%) underwent minimally invasive surgery. There was no between-group difference in the primary outcome measure, 30-day Comprehensive Complications Index (adjusted mean difference, –3.2; 95% CI, –11.8 to 5.3; P = .45). Secondary outcome measures were also not different between groups.

Conclusions and Relevance

In frail patients undergoing colorectal cancer resection (predominantly minimally invasive) within an enhanced recovery pathway, a multimodal prehabilitation program did not affect postoperative outcomes. Alternative strategies should be considered to optimize treatment of frail patients preoperatively.

Trial Registration

ClinicalTrials.gov identifier: NCT02502760

Introduction

Colorectal cancer is the third most common cancer in North America and the second leading cause of cancer death in men and women.¹ The median age at diagnosis is 69 years, with patients older than 75 years being at higher risk of morbidity and mortality.² Although surgery remains a standard treatment of colorectal cancer,¹ the proportion of elderly patients undergoing surgery declines with age³ as a result of unfitness for surgery owing to frailty.^4,5 Frailty is characterized by age-related disturbances leading to decreased energy and muscle strength, weight loss, and sedentary activity levels, thus increasing an individual’s vulnerability to dependency and/or death.^5,6 Although preoperative frailty is associated with increased risk of postoperative complications and prolonged hospital length of stay (LOS),^7,8 limited evidence supports specific interventions to improve the outcomes of frail patients undergoing surgery.^9,10,11,12

Multimodal prehabilitation aims to enhance physiological reserve¹³ in anticipation of the predictable detrimental effects of surgery¹⁴ and facilitate postoperative recovery of functional capacity,^15,16 especially in patients with lower preoperative fitness.¹⁷ However, the effect of prehabilitation on postoperative complications remains unclear.¹¹ It is plausible to hypothesize that patients at greater risk for postoperative complications, such as the frail elderly, are more likely to benefit from prehabilitation.¹⁸ Hence, we conducted a randomized clinical trial to contribute evidence about the effects of prehabilitation on postoperative outcomes of frail patients undergoing colorectal cancer resection. Our primary hypothesis was that prehabilitation would reduce 30-day postoperative complications, as measured using the Comprehensive Complication Index (CCI; scores range from 0 to 100, with a score of 100 being death as a result of complications), compared with postoperative rehabilitation.

Methods

Trial Design

This study was a 2-site, single-blind, parallel-arm, superiority randomized clinical trial. The trial protocol is found in Supplement 1. Ethics approval was obtained from McGill University Health Center and the Sir Mortimer B. Davis Jewish General Hospital, Montreal, Quebec, Canada. All participating patients signed informed consent. The trial was registered a priori with ClinicalTrials.gov and followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.¹⁹

Study Participants

Consecutive patients older than 65 years and scheduled for surgical treatment of nonmetastatic colorectal cancer were considered for inclusion and screened for frailty using the validated Fried Frailty Index²⁰ (1 indicates no frailty; 2-3, intermediate frailty; and 4-5, frailty)²¹ (see eMethods 1 in Supplement 2 for details about frailty scoring). Patients were excluded if they (1) had a Fried Frailty Index of 1, (2) did not speak English or French, (3) had metastatic cancer, or (4) had premorbid conditions (ie, cardiorespiratory, musculoskeletal, and/or neurological) that contraindicated exercise and fitness assessments. All patients were treated within an enhanced recovery pathway (ERP) according to the Enhanced Recovery After Surgery guidelines.²² This ERP has been in place at the Montreal General Hospital since 2008²³ and at the Sir Mortimer B. Davis Jewish General Hospital since 2015.

Study Procedures

Eligible patients were randomly assigned 1:1 to receive a 4-week prehabilitation program (Prehab group) or a 4-week postoperative rehabilitation program (Rehab group). This design was chosen based on previous trial experience,²⁴ because it limits the risk of poor enrollment and attrition bias (ie, missing follow-up appointments by patients in the control group) by offering interventions to both groups. Randomization was achieved via computer-generated random numbers composed of 12 blocks of 10. Allocations were placed in sealed, opaque, consecutively numbered envelopes by an independent researcher (G.B.-D.). Allocation was concealed until the baseline assessment was completed, when envelopes were opened in numerical order. Outcome assessors (G.B.-D. and R.A.), surgeons (S.L., M.B., B.S., P.C., G.G., and N.M.), and statisticians (J.F.F.) were blinded to group assignment. Owing to the nature of the intervention, it was not possible to blind patients or intervention staff. To minimize performance bias, patients were informed that we were comparing 2 types of perioperative exercise programs, and one was not presented as potentially superior to the other.

Prehabilitation

After randomization, patients in the Prehab group received a personalized, supervised, and home-based multimodal program prescribed by a kinesiologist (R.A.), a nutritionist, and a psychology-trained nurse. The program started after the baseline visit and continued for 4 weeks before surgery, as recommended by the Canadian Oncological Society.²⁵ Participation in the Prehab program had no effect on surgical waiting time. The program did not continue after surgery. Details about our prehabilitation program have been reported elsewhere^16,26 and are summarized below.

Exercise Intervention

Patients were asked to attend training sessions supervised by a trained kinesiologist at the hospital prehabilitation unit once a week for 4 weeks. During these sessions, patients performed 30 minutes of moderate aerobic exercise (including a 5-minute warm-up) on a recumbent stepper (T5; NuStep, Inc), 25 minutes of resistance exercises using an elastic band, and 5 minutes of stretching. In addition to the supervised exercise sessions, participants were prescribed a personalized home-based program of aerobic activities (walk daily for a total of 30 minutes as moderate-intensity aerobic activity) and resistance training (elastic band routine 3 times per week) according to guidelines of the American College of Sports Medicine.²⁷

Nutrition Intervention

All participants had their nutritional status assessed by a registered dietitian. At baseline, participants were asked to complete a 3-day food diary from which macronutrients were estimated using food exchange lists and composition tables. Nutritional status was evaluated using the Subjective Global Assessment.²⁸ Macronutrient intake was evaluated based on dietary reference intake values, and food choices were compared to Eating Well With Canada’s Food Guide recommendations.²⁹ Patients were advised on how to improve their own daily dietary intake based on the balanced plate concept. Target protein intake was 1.5 g/kg of body weight (or adjusted body weight in obese patients) as per the European Society for Clinical Nutrition and Metabolism recommendation for patients with cancer.^30,31 If the patient did not meet the protein requirement by diet alone, they were provided with whey protein supplementation (Immunocal; Immunotec, Inc). Patients were instructed to use the supplements within 1 hour of their exercise training to maximize muscle protein synthesis.^32,33 Further nutritional counseling included caloric balance, bowel movement regularity, and glycemic control.

Psychological Intervention

The study participants had an assessment by a psychology-trained nurse. Potential causes of perioperative fatigue, anxiety, and depression were discussed. Study participants were provided with personalized coping strategies and received a compact disc with an audio track containing the instructions to help them perform the exercises at home 3 times a week. The intervention also included counseling regarding smoking and alcohol cessation. The use of nicotine replacement therapy was decided through shared decision-making.

Rehabilitation

Patients in the Rehab group received the same baseline assessment as patients in the Prehab group. The identical multimodal program was prescribed for patients in the Rehab group; however, the interventions started only after postoperative discharge from the hospital. The Rehab program was explained to patients a few days before surgery (at the preoperative assessment visit) and continued for 4 weeks after discharge.

Adherence to the Program

Adherence to the multimodal program was assessed accounting for attendance at the in-hospital exercise sessions. In addition, patients received a standardized instructional booklet with a diary to document their daily activities. Patients were contacted on a weekly basis by telephone to report adherence to the home-based component of the program.

Outcomes and Measures

Primary Outcome

Assessments were performed before randomization (baseline assessment), a few days before surgery (preoperative assessment), and at 4 weeks after surgery (4-week assessment). The primary outcome was the CCI, which integrates all complications with their respective severities on a continuous scale ranging from 0 (no burden due to complications) to 100 (death as a result of complications).^34,35 Postoperative complications were recorded to 30 days after surgery (definitions provided in eMethods 2 in Supplement 2) and scored by severity using the Clavien-Dindo classification.³⁶ The CCI was derived from these scores using the CCI calculator available online (http://www.assessurgery.com). Previous studies supported the validity of the CCI as a measure of postoperative morbidity and suggested that, compared with traditional morbidity measures (eg, overall rate of complications, rate of severe complications), it provides a more comprehensive and sensitive end point for surgical research.^34,35

Secondary Outcomes

Data regarding primary and total LOS, readmissions, and emergency department visits within 30 days after surgery were obtained from electronic medical records. Functional walking capacity was measured by the 6-minute walking test (6MWT) distance assessed at baseline, before surgery, and 4 weeks after surgery. This test has been previously validated as a measure of recovery after colorectal surgery,^37,38 and a change of at least 20 m is considered clinically meaningful in this patient population.³⁹ The effect of the intervention on patient-reported outcome measures (PROMs) targeted between-group differences in generic health status (36-Item Short Form Survey),⁴⁰ anxiety and depression (Hospital Anxiety and Depression Scale),⁴¹ and self-reported energy expenditure (Community Healthy Activities Model Program for Seniors questionnaire).⁴²

Statistical Analysis

Data were analyzed from July 25 through August 21, 2019. The sample size requirement for this trial was estimated for an α level of 0.05 and 90% power to detect a 10-point difference in the CCI between groups. This difference of 10 points is considered clinically relevant because it reflects the differential burden of one grade 1 complication in the Clavien-Dindo classification.^35,36 In a previous study,⁴³ the mean (SD) CCI in the participating institutions was 10.45 (15.8). Based on this estimate, a sample of 53 participants per group was considered sufficient for our analysis. We aimed to recruit a total of 120 participants to account for potential preoperative exclusions (eg, surgery not performed) and a possible increase in data variance owing to multiple imputation of missing data.

Data analysis was conducted using Stata, version 14 (StataCorp LLC), following a prespecified analysis plan available in eMethods 3 in Supplement 2. Thirty-day CCI was compared between groups using linear regression. Rates of overall complications and severe complications were compared using logistic regression. Between-group comparison of 6MWT data was conducted using linear regression (6MWT distance preoperatively and at 4 weeks) and logistic regression (preoperative improvement in 6MWT distance and recovery of 6MWT distance at 4 weeks). The PROM data (36-Item Short Form Survey, Hospital Anxiety and Depression Scale, and Community Healthy Activities Model Program for Seniors questionnaire scores) were compared using linear regression. Primary and total LOS were compared between groups using Cox proportional hazards regression and linear regression, respectively. Because we had a single prespecified confirmatory outcome, we did not adjust for multiple comparisons. To minimize potential bias arising from missing data, our primary analysis was conducted using multiple imputation by chained equations and predictive mean matching under the assumption that data were missing at random. Estimates from 20 imputed data sets were combined using Rubin rules.⁴⁴ To improve precision of estimates and compensate for potential between-group imbalances, estimates were adjusted for recruitment site and prognostic factors for complications and poor postoperative recovery, including age, sex, American Society of Anesthesiologists score, colon vs rectal surgery, minimally invasive vs open surgery, baseline 6MWT distance, body mass index, and Fried Frailty Index (2-3 vs 4-5).^{21,37,38,45,46,47} Both adjusted and unadjusted estimates were reported. Our primary analysis followed an intention-to-treat format, with participants being analyzed in the groups to which they were allocated. A secondary (post hoc), per-protocol analysis was performed to further characterize the effect of higher compliance with the Prehab program (ie, attendance at ≥75% of the in-hospital sessions) on postoperative outcomes. All statistical tests were 2 sided, and statistical significance was set at P < .05.

Results

A total of 418 patients were assessed for eligibility from September 7, 2015, through June 19, 2019. Of these, 298 patients were excluded (not frail [n = 290], unable to exercise [n = 3], and planned neoadjuvant treatment [n = 5]), and 120 frail patients were randomized to the Prehab or the Rehab group (Figure). Ten patients were excluded after randomization because they refused surgery (n = 3), died before surgery (n = 3), had no cancer (n = 1), had surgery without bowel resection (n = 1), or were switched to palliative care (n = 2). Hence, 110 patients were included in the intention-to-treat analysis (55 in the Prehab and 55 in the Rehab groups). Baseline and operative characteristics of the 2 groups are presented in Table 1.⁴⁸ The 2 groups were similar, except that patients in the Rehab group were generally older (≥75 years, 42 [76.3%] vs 32 [58.2%]) and had lower physical status (American Society of Anesthesiologists score of 3, 33 [60.0%] vs 43 [78.2%]). The prevalence of minimally invasive surgery (ie, laparoscopic surgery or transanal minimally invasive surgery) was high in both groups (42 [76.4%] in the Prehab group vs 45 [81.8%] in the Rehab group).

Figure. CONSORT Diagram.

Table 1. Baseline and Operative Characteristics in the Prehab vs Rehab Groups.

Characteristic	Prehab Group		Rehab Group
	No. of Patients	Data	No. of Patients	Data
Age
Median (IQR), y	55	78 (72-82)	55	82 (75-84)
≥75 y, No. (%)	55	32 (58.2)	55	42 (76.4)
Male sex, No. (%)	55	29 (52.7)	55	23 (41.8)
Body composition
Weight, median (IQR), kg	55	67.0 (60.6-82.1)	54	71.6 (57.0-79.7)
Lean body mass, mean (SD), kg	55	46.8 (9.1)	54	46.2 (10.2)
Fat, median (IQR), %	55	34.1 (26.1-38.5)	54	35.9 (25.4-41.1)
Body mass indexa
Median (IQR)	55	24.9 (23.0-30.1)	54	26.4 (23.8-30.6)
≥30.0, No. (%)	55	14 (25.5)	54	16 (29.6)
Baseline 6MWT distance
Mean (SD), m	55	325.3 (114.3)	54	304.0 (107.3)
Median (IQR), % predicted	55	57.9 (44.3-66.5)	54	52.7 (40.5-63.6)
<400 m, No. (%)	55	40 (72.7)	54	44 (81.5)
Grip strength, mean (SD), kg force	55	22.8 (8.3)	55	21.0 (6.7)
Fried Frailty Index, No. (%)b
2	55	25 (45.5)	55	17 (30.9)
3	55	16 (29.1)	55	22 (40.0)
4	55	7 (12.7)	55	10 (18.2)
5	55	7 (12.7)	55	6 (10.9)
Comorbidities, No. (%)
Diabetes type 2	55	19 (34.5)	55	21 (38.2)
Hypertension	55	29 (52.7)	55	42 (76.4)
Cardiovascular diseases	55	16 (29.1)	55	19 (34.5)
Atrial fibrillation	55	9 (16.4)	55	5 (9.1)
OSA	55	5 (9.1)	55	5 (9.1)
COPD	55	8 (14.5)	55	3 (5.5)
Arthritis/connective tissue disease	54	15 (27.8)	54	23 (42.6)
Dyslipidemia	55	27 (49.1)	54	26 (48.1)
Hypothyroidism	55	12 (21.8)	55	11 (20.0)
Asthma	55	7 (12.7)	55	1 (1.8)
Charlson comorbidity index, median (IQR)c	55	3 (2-4)	55	4 (3-5)
Smoker (current), No. (%)	55	6 (10.9)	55	5 (9.1)
Alcohol use (current), No. (%)d	55	5 (9.1)	55	3 (5.5)
ASA status, No. (%)
2	55	19 (34.5)	55	9 (16.4)
3	55	33 (60.0)	55	43 (78.2)
4	55	3 (5.5)	55	3 (5.5)
Metabolic status
CRP level, median (IQR), mg/L	50	6.2 (1.9-22.9)	47	4.5 (1.7-14.5)
Albumin level, mean (SD), g/dL	53	3.91 (0.47)	52	3.83 (0.40)
HbA1c level, mean (SD), %	47	6.2 (0.86)	49	6.3 (1.4)
Hemoglobin level, mean (SD), g/dL	54	11.66 (2.00)	51	11.16 (1.94)
Energy expenditure, median (IQR), kcal/kg/wk	52	36.4 (10.6-80.3)	53	23.0 (5-68.5)
Subjective global assessment, No. (%)e
A	54	16 (29.6)	45	20 (44.4)
B	54	27 (50.0)	45	12 (26.7)
C	54	11 (20.4)	45	13 (28.9)
Oncologic data, No. (%)
Rectal cancer site	55	18 (32.7)	55	13 (23.6)
Neoadjuvant chemotherapy	55	7 (12.7)	54	6 (11.1)
Tumor stage
0-I	55	18 (32.7)	53	15 (28.3)
II	55	15 (27.3)	53	18 (34.0)
III	55	19 (34.5)	53	16 (30.2)
IV	55	3 (5.5)	53	4 (7.5)
Surgical approach, No. (%)
Open	55	13 (23.6)	55	10 (18.2)
Minimally invasivef	55	42 (76.4)	55	45 (81.2)
Type of surgery, No. (%)
Ileocecal resection	55	2 (3.6)	55	1 (1.8)
Right hemicolectomy	55	23 (41.8)	55	23 (41.8)
Left hemicolectomy	55	5 (9.1)	55	8 (14.5)
Subtotal colectomy	55	2 (3.6)	55	1 (1.8)
Anterior/sigmoid resection	55	6 (10.9)	55	9 (16.4)
Transverse colectomy	55	1 (1.8)	55	1 (1.8)
Low anterior resection	55	10 (18.2)	55	7 (12.7)
Abdominoperineal resection	55	4 (7.3)	55	4 (7.3)
Other bowel surgery	55	2 (3.6)	55	1 (1.8)
Surgical data
Stoma creation, No. (%)	55	15 (27.3)	55	9 (16.4)
Duration of surgery, median (IQR), min	55	180 (110-280)	55	180 (130-300)
Intraoperative blood loss, median (IQR), mL	55	100 (50-200)	55	100 (50-200)

Abbreviations: ASA, American Society of Anesthesiologists; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; HbA_1c, glycated hemoglobin; IQR, interquartile range; 6MWT, 6-minute walk test; OSA, obstructive sleep apnea; Prehab, prehabilitation; Rehab, rehabilitation; TAMIS, transanal minimally invasive surgery.

SI conversion factors: To convert albumin and hemoglobin to grams per liter, multiply by 10; CRP to nanomoles per liter, multiply by 9.524; and HbA_1c to a proportion of total hemoglobin, multiply by 0.01.

^aCalculated as weight in kilograms divided by square of height in meters.

^bHigher scores indicate greater frailty.

^cScores range from 0 to 24, with higher scores indicating greater comorbidities.

^dAlcohol consumption above Canada’s Low-Risk Alcohol Drinking Guidelines (2017) for men (>15 standard drinks per week) and women (>10 standard drinks per week).⁴⁸

^eA indicates normal; B, nutrition risk; and C, malnutrition.

^fIncludes laparoscopic or TAMIS.

The median duration of time between the baseline assessment and surgery was 40 (interquartile range [IQR], 28-51) days in the Prehab group and 35 (IQR, 22-55) days in the Rehab group. Mean (SD) adherence with the in-hospital supervised exercise sessions was 68% (38%) in the Prehab group and 14% (27%) in the Rehab group. When accounting for overall adherence to the program (supervised and home-based; exercise and nutritional components), the Prehab group self-reported a mean (SD) of 80% (27%) compared with 30% (33%) in the Rehab group. No adverse events were reported.

No data were missing for the primary trial outcome (ie, CCI) or regarding complications, primary and total LOS, emergency department visits, and readmission rates. Rates of missing data for measures of walking capacity and PROMs depended on attendance at the trial visits in the hospital. Attendance at baseline was 100%. Attendance at the preoperative assessment visit was 83.6% in the Prehab group (n = 46) and 69.1% in the Rehab group (n = 38). Most of these nonattendances were not justified (7 in the Prehab group and 14 in the Rehab group), whereas some were attributed to medical reasons (2 in the Prehab group and 1 in the Rehab group). At 4 weeks after surgery, attendance at the postoperative assessment visit was 70.9% in the Prehab group (n = 39) and 56.4% in the Rehab group (n = 31). Nonattendance at the postoperative visits were not justified (8 in the Prehab group and 10 in the Rehab group) or attributed to medical reasons such as prolonged intensive care unit LOS or because patients felt too weak (7 in the Prehab group and 8 in the Rehab group).

Intention-to-Treat Analyses

Data regarding postoperative outcomes are presented in Table 2. No between-group difference was found in the primary outcome measure, 30-day CCI (adjusted mean difference, −3.2; 95% CI, −11.8 to 5.3; P = .45). A detailed description of postoperative complications is presented in eTable 1 in Supplement 2. We also did not identify between-group differences in overall complications, severe complications, primary and total LOS, emergency department visits, and readmissions. No between-group differences in preoperative or postoperative walking capacity were noted (Table 3). During the preoperative period, 26 of 47 patients (55.3%) in the Prehab group increased their 6MWT distance by 20 m or more (ie, minimal important difference) compared with 10 of 38 patients (26.3%) in the Rehab group (adjusted odds ratio [OR], 2.2; 95% CI, 0.9-5.8; P = .10). At 4 postoperative weeks, the proportion of patients returning to preoperative levels of 6MWT distance was 68.4% (26 of 38) in the Prehab group compared with 53.3% (16 of 30) in the Rehab group (adjusted OR, 1.9; 95% CI, 0.6-5.9; P = .26). Preoperative and postoperative PROMs are presented in Table 4; no between-group differences were identified in self-reported generic health status (36-Item Short Form Survey), anxiety and depression (Hospital Anxiety and Depression Scale), or energy expenditure (Community Healthy Activities Model Program for Seniors questionnaire). Results from unadjusted analyses (reported in eTables 2-4 in Supplement 2) were consistent with the primary analysis (ie, no between-group differences were observed for any of the outcome variables assessed in the trial).

Table 2. Postoperative Outcomes in the Prehab vs Rehab Groups.

Outcome	Prehab Group		Rehab Group		Adjusted Estimate (95% CI)a	P Value
	No. of Patients	Data	No. of Patients	Data
Comprehensive Complication Indexb
Mean (SD)	55	12.7 (21.5)	55	15.7 (25.3)	MD, –3.2 (–11.8 to 5.3)	.45
Median (IQR)	55	0 (0 to 12.2)	55	0 (0 to 29.6)
Complications, No. (%)b
Overall	55	25 (45.5)	55	25 (45.5)	OR, 0.9 (0.4 to 2.2)	.90
Severe	55	7 (12.7)	55	11 (20.0)	OR, 0.5 (0.1 to 1.6)	.23
Primary LOS, median (IQR), d	55	4 (3 to 8)	55	4 (3 to 8)	HR, 19.2 (–2.9 to 2.2)	.80
Total LOS, median (IQR), db	55	4 (3 to 8)	55	5 (3 to 9)	MD, –5.8 (–17.3 to 5.8)	.32
Emergency department visit, No. (%)b	55	3 (5.5)	55	6 (10.9)	OR, 0.3 (0.1 to 1.9)	.21
Hospital readmission, No. (%)b	55	2 (3.6)	55	5 (9.1)	OR, 0.3 (0.03 to 1.9)	.18

Abbreviations: HR, hazard ratio; IQR, interquartile range; LOS, length of stay; MD, mean difference; OR, odds ratio; Prehab, prehabilitation; Rehab, rehabilitation.

^aCoefficients were derived with imputation of missing data and adjustment for confounders, including recruitment site (fixed-effect), age, sex, American Society of Anesthesiologists score, colon vs rectal surgery, minimally invasive vs open surgery, baseline 6-minute walk test, body mass index, and Fried Frailty Index (2-3 vs 4-5). No data were missing for the variables described, so imputation was conducted only for adjustment variables. The multiple imputation model is described in eMethods 3 in Supplement 2.

^bMeasured at 30 days.

Table 3. Walking Capacity in the Prehab vs Rehab Groups.

6MWT Data	Prehab Group		Rehab Group		Adjusted Estimate (95% CI)a	P Value
	No.	Data	No.	Data
Baseline, mean (SD), m	55	325.3 (114.3)	54	304.0 (107.3)	NA	NA
Preoperative, mean (SD), m	47	346.1 (117.8)	38	315.8 (107.5)	MD, 11.2 (–13.7 to 36.1)	.37
Improved preoperatively, No. (%)b	47	26 (55.3)	38	10 (26.3)	OR, 2.2 (0.9 to 5.8)	.10
Postoperative (4 wk), mean (SD), m	38	336.4 (121.8)	30	286.1 (105.1)	MD, 18.5 (–20.2 to 57.3)	.34
Recovered at 4 wk, No. (%)c	38	26 (68.4)	30	16 (53.3)	OR, 1.9 (0.6 to 5.9)	.26

Abbreviations: MD, mean difference; 6MWT, 6-minute walk test; NA, not applicable; OR, odds ratio; Prehab, prehabilitation; Rehab, rehabilitation;.

^aCoefficients were derived with imputation of missing data and adjustment for confounders, including recruitment site (fixed-effect), age, sex, American Society of Anesthesiologists score, colon vs rectal surgery, minimally invasive vs open surgery, baseline 6MWT, body mass index, and Fried Frailty Index (2-3 vs 4-5). Details about the multiple imputation model are described in eMethods 3 in Supplement 2.

^bImprovement of 6MWT distance of 20 m or more compared with baseline.

^cRecovered to 6MWT distance within 20 m of the baseline value or above.

Table 4. Patient-Reported Outcome Measures in the Prehab vs Rehab Groups.

Measure	Prehab Group		Rehab Group		Adjusted Estimate (95% CI)a	P Value
	No. of Patients	Data	No. of Patients	Data
Total Physical SF-36 subscale, mean (SD)b
Baseline	53	49.4 (21.2)	52	52.9 (18.7)	NA	NA
Before surgery	42	57.1 (19.3)	33	58.7 (19.4)	MD, 0.16 (–6.7 to 7.0)	.96
4 wk After surgery	38	49.7 (20.2)	30	51.1 (14.7)	MD, –0.43 (–7.2 to 6.3)	.90
Total Mental SF-36 subscale, mean (SD)b
Baseline	53	54.2 (23.1)	52	58.8 (20.2)	NA	NA
Before surgery	42	59.2 (20.8)	33	66.0 (19.4)	MD, –2.8 (–10.7 to 5.0)	.47
4 wk After surgery	38	55.3 (22.6)	30	63.6 (15.3)	MD, –2.3 (–9.7 to 5.1)	.53
HADS Anxiety score, median (IQR)c
Baseline	52	6 (4 to 8)	53	5 (2 to 8)	NA	NA
Before surgery	41	6 (4 to 8)	33	4 (1 to 7)	MD, 0.88 (–0.59 to 2.3)	.23
4 wk After surgery	37	5 (2 to 7)	31	4 (1 to 8)	MD, –1.9 (–7.0 to 3.2)	.46
HADS Depression score, median (IQR)c
Baseline	52	5 (3 to 7.5)	53	4 (3 to 9)	NA	NA
Before surgery	41	4 (2 to 6)	33	4 (2 to 7)	MD, –0.48 (–1.7 to 0.70)	.42
4 wk After surgery	37	4 (2 to 5)	31	4 (1 to 7)	MD, –0.39 (–1.8 to 1.0)	.58
Energy expenditure, median (IQR), kcal/kg/wk
Baseline
Light	52	20.8 (9 to 51)	53	16 (1 to 65)	NA	NA
Moderate-vigorous	52	4 (0 to 21)	53	0 (0 to 3.5)	NA	NA
Before surgery
Light	44	23.8 (8.5 to 58.3)	33	15 (6.5 to 42)	NA	NA
Moderate-vigorous	44	17.6 (10.2 to 33)	33	3 (0 to 13)	MD, 5.2 (–11.3 to 21.8)	.53
4 wk
Light	38	20 (5 to 37.5)	30	16.9 (5 to 35)	NA	NA
Moderate-vigorous	38	12 (0 to 31.3)	30	13.5 (0 to 36)	MD, –4.2 (–22.5 to 14.1)	.65

Abbreviations: HADS, Hospital Anxiety and Depression Scale; MD, mean difference; NA, not applicable; SF-36, 36-Item Short Form Health Survey.

^aCoefficients derived with imputation of missing data and adjustment for confounders, including recruitment site (fixed-effect), age, sex, American Society of Anesthesiologists score, colon vs rectal surgery, minimally invasive vs open surgery, baseline 6-minute walk test, body mass index, Fried Frailty Index (2-3 vs 4-5), and baseline scores of the respective measures (in deciles). Details about the multiple imputation model are described in eMethods 3 in Supplement 2.

^bScores range from 0 to 100, with higher values indicating better health.

^cScores range from 0 to 21, with higher values indicating worse health.

Per-Protocol Analyses

A per-protocol analysis was conducted to compare the outcomes of patients adherent in the Prehab program (attended ≥75% of the in-hospital exercise sessions [n = 33]) with patients who did not undergo prehabilitation (Rehab group [n = 55]). Patients’ data and results are presented in eTables 5 through 8 in Supplement 2. Rates of preoperative meaningful improvement in 6MWT distance (≥20 m) were higher in the adherent Prehab group compared with the Rehab group (20 of 32 [62.5%] vs 10 of 38 [26.3%] of patients; adjusted OR, 3.8; 95% CI, 1.3-11.1; P = .02). Prehab patients who were adherent walked a mean of 29 m more than those in the Rehab group (adjusted mean difference, 29.1; 95% CI, 2.6-55.6; P = .03). At 4 postoperative weeks, 20 of 27 patients (74.1%) returned to preoperative levels of walking capacity in the Prehab group compared with 16 of 30 (53.3%) in the Rehab group (adjusted OR, 3.1; 95% CI, 0.8-11.7; P = .10). Despite this potential improvement in walking capacity when patients adhered to the Prehab program, no between-group differences in 30-day CCI (adjusted mean difference, −4.3; 95% CI, –14.5 to 6.0; P = .41) or in other outcome measures assessed in the trial were noted.

Discussion

In this randomized clinical trial, prehabilitation did not appear to reduce postoperative complications in frail patients undergoing colorectal cancer resection (predominantly minimally invasive) in the context of an ERP. We also did not observe between-group differences in primary and total LOS, rates of readmission and emergency department visits, postoperative walking capacity, and PROMs. Although our sample size was relatively small (n = 110), the lower bounds of the 95% CI observed for between-group differences in CCI (11.8) suggest that we were unlikely to miss a meaningful benefit of the intervention due to type II error (when accounting for the reported minimal important difference of 10).³⁵ The lack of difference between groups occurred even when excluding patients with low adherence to the program (per-protocol analysis), which further supports that the intervention was not efficacious in improving postoperative outcomes.

The present study is, to our knowledge, the first randomized clinical trial specifically focused on assessing the effect of prehabilitation on clinical outcomes of frail patients (Fried Frailty Index, ≥2). A major strength of our study was its methodological rigor in minimizing risk of bias. We used an elaborate statistical approach accounting for missing data and potential imbalances between groups, and sensitivity analyses (eg, with no adjustment for confounders and using a per-protocol approach) confirmed the robustness of our conclusions across different scenarios. The use of the CCI as a comprehensive and sensitive measure of postoperative morbidity is another major strength.^34,35 In light of these design considerations, we believe that our results contribute important new evidence about the role of prehabilitation in elderly frail patients who are candidates for colorectal cancer resection.

Preoperative frailty has been shown to be associated with increased risk of postoperative complications (OR, 2.54; 95% CI, 1.12-5.77) and increased LOS (OR, 1.69; 95% CI, 1.28-2.20) after abdominal surgery.²¹ However, evidence regarding the effect of exercise interventions before or after abdominal surgery in frail patients is limited.^11,49 To our knowledge, only 1 study had assessed the effect of prehabilitation on clinical outcomes of frail patients undergoing colorectal surgery. Chia et al⁵⁰ used an observational study design (preimplementation vs postimplementation) to show a reduction in primary LOS from 11 to 8 days in patients undergoing prehabilitation. In contrast, we did not observe between-group differences in primary LOS. In fact, the median LOS in our study was much lower (4 [IQR, 3-8] days in both groups) compared with that of the study by Chia et al,⁵⁰ which may reflect differences in contexts of care (ie, the use of an ERP).

Several reasons may explain the lack of effect of prehabilitation in our trial. First, our comparison group consisted of patients undergoing a postoperative rehabilitation program instead of usual care or a sham intervention. This design was chosen based on previous trial experience²⁴ to avoid the risk of poor enrollment and attrition bias leading to unbalanced group characteristics. The low adherence of patients to the rehabilitation program (14%), however, supports that this control intervention was closer to a sham condition than to an active comparator. Second, this trial was conducted in the context of a well-established ERP. In a recent large retrospective study,⁵¹ the complication rate in unselected patients undergoing colorectal surgery (38%) was similar to that observed in the frail patients who took part in the present trial (43%). This finding indicates that, even in this vulnerable population, the effects of prehabilitation may be limited when other aspects of perioperative care are already optimized. Third, a prehabilitation program lasting 4 to 5 weeks may not be sufficient to increase physiological reserve preoperatively and reduce postoperative complications. Significant increases in lean body mass, functional capacity, and muscle strength have been observed in elderly patients only after 12 weeks in programs involving more intense exercise and protein supplementation.⁵² Therefore, the limited time currently available to prepare frail patients for surgery might need to be revisited if sufficient anabolic stimulus is required to counteract surgery-related morbidity.

Limitations

The findings of this trial should be considered with some limitations. Risk of performance bias cannot be excluded, because patients and intervention staff were not blinded. Any bias attributed to lack of blinding, however, would likely have favored the Prehab group, and we found no between-group differences. Groups were not balanced at baseline for age and American Society of Anesthesiologists scores; however, our analyses were adjusted for these potential confounders. There was a considerable rate of missing data for secondary outcomes (6MWT and PROMs) as a result of missing follow-up appointments. These missing data were handled with multiple imputation to reduce the risk of attrition bias. Assessment of compliance with outpatient components of our program was based on self-reporting and therefore may be subject to reporting bias. Patients excluded from the trial after randomization (eg, those who did not have surgery) were not accounted for in our intention-to-treat analysis. Our sample predominantly involved cases with minimally invasive surgery (79.1%), so our results may not be generalizable to settings with greater prevalence of open surgery. Importantly, this trial was conducted in centers with well-established ERPs and, therefore, our results are only generalizable to similar contexts of care.

Conclusions

In frail patients undergoing resection of colorectal cancer within an ERP with high prevalence of minimally invasive surgery, a prehabilitation program involving exercise, nutritional, and psychological interventions did not appear to affect postoperative complications. A better understanding of which aspects of prehabilitation influence surgical outcomes in this context of care is needed. Alternative strategies should be considered to prepare frail patients for surgery.

Supplement 1.

Trial Protocol

Supplement 2.

eMethods 1. Fried Frailty Scoring Index

eMethods 2. Definition of Complications

eMethods 3. Statistical Analysis Plan

eTable 1. Detailed Description of Postoperative Complications in the Prehabilitation vs Rehabilitation Groups

eTable 2. Intention to Treat Unadjusted Postoperative Outcomes in the Prehabilitation vs Rehabilitation Groups

eTable 3. Intention to Treat Unadjusted Walking Capacity in the Prehabilitation vs Rehabilitation Groups

eTable 4. Intention to Treat Unadjusted Patient-Reported Outcome Measures in the Prehabilitation vs Rehabilitation Groups

eTable 5. Per-Protocol Baseline and Operative Characteristics in the Prehabilitation vs Rehabilitation Groups

eTable 6. Per-Protocol Adjusted postoperative outcomes in the Prehabilitation vs Rehabilitation Groups

eTable 7. Per-Protocol Adjusted Walking Capacity in the Prehabilitation vs Rehabilitation Groups

eTable 8. Per-Protocol Adjusted Patient-Reported Outcome Measures in the Prehabilitation vs Rehabilitation Groups

eReferences.

Supplement 3.

Data Sharing Statement