A systematic review of the intervention components, adherence and outcomes of enhanced recovery programmes in older patients undergoing elective colorectal surgery

Background Enhanced recovery programmes (ERPs) aim to attenuate the surgical stress response and accelerate recovery after surgery, but are not specifically designed for older patients. The objective of this study was to review the components, adherence and outcomes of ERPs in older patients (≥65 years) undergoing elective colorectal surgery. Methods Pubmed, Embase and Cinahl were searched between 2000 and 2017 for randomised and non-randomised controlled trials, before-after studies, and observational studies. The methodological quality of the studies was evaluated using the MINORS quality assessment. The review was performed and reported according to the PRISMA guidelines. Results Twenty-one studies, including 3495 ERP patients aged ≥65 years, were identified. The ERPs consisted of a median of 13 intervention components. Adherence rates were reported in 9 studies and were the highest (≥80%) for pre-admission counselling, no bowel preparation, limited pre-operative fasting, antithrombotic and antimicrobial prophylaxis, no nasogastric tube, active warming, and limited intra-operative fluids. The median post-operative length of stay was 6 days. The median post-operative morbidity rate (Clavien-Dindo I-IV) was 23.5% in-hospital and 29.8% at 30 days. The in-hospital post-operative mortality rate was 0% in most studies and amounted to a median of 1.4% at 30 days. The median 30-day readmission rate was 4.9% and the median reoperation rate was 5.0%. Conclusions ERPs in older patients were in accordance with the ERP consensus guidelines. Although the number of intervention components applied increased over time, outcomes in earlier and later studies remained comparable. Adherence rates were under-reported. Future studies should explore adherence and age-related factors, such as frailty profile, that could influence adherence. Trial registration PROSPERO 2018 CRD42018084756. Electronic supplementary material The online version of this article (10.1186/s12877-019-1158-3) contains supplementary material, which is available to authorized users.


Background
Fast Track protocols, also known as Enhanced Recovery After Surgery (ERAS®), or Enhanced Recovery Programmes (ERPs), have been developed by surgeons and anaesthesiologists to reduce the surgical stress response, accelerate recovery, and improve overall post-operative outcomes [1]. They were initially introduced in the early nineties by Kehlet and colleagues as standard of care for colorectal surgery, and have spread to other surgical specialties [1,2]. ERPs generally include about 20 evidence-based intervention components during the peri-operative period, and require the active participation of a multidisciplinary team and the patient [3][4][5].
Due to demographic aging and advances in surgical and anaesthetic techniques, the demand for surgical procedures in older persons is rapidly increasing [6,7], but the ability of older patients to actively participate in ERPs and to achieve the same results as younger patients has been debated [8,9]. After all, ERPs were not specifically designed for older patients. On the other hand, older patients might actually benefit more than younger patients, because they are more susceptible to adverse post-operative outcomes and longer hospital stays [9,10].
In 2014, Bagnall et al. published a first review about the safety, feasibility and efficacy of ERPs in patients aged 65 and over undergoing colorectal surgery, including 16 studies published before February 2014 [11]. Launay-Savary et al. performed a new search until 2015 and included two extra studies: a meeting abstract and a study later retracted from literature [12]. With new literature emerging since the latest reviews, and ERPs being further standardised, a new systematic literature review was deemed useful.
The primary objective of this review was to map intervention components of ERPs in patients aged 65 years and older undergoing elective colorectal surgery. In addition, we analysed adherence to individual ERP components and outcomes of the ERPs in patients aged 65 years and older.

Methods
This review was performed in the framework of the PRISMA guidelines (www.prisma-statement.org) and was registered in PROSPERO, the international prospective register of systematic reviews (registration number CRD42018084756).

Search strategy
An electronic bibliographic database search through PUBMED, EMBASE, and CINAHL was performed.
The following search string was used for PubMed and adapted for the other databases (Additional file 1):  [tiab] OR resection[tiab]))) AND (ERAS [tiab] OR "enhanced recovery" [tiab] OR "accelerated recovery" [tiab] OR "expedited recovery" [tiab] OR "fast track" [tiab] OR multimodal [tiab] OR multi-modal[tiab])). The search was limited to English, Dutch, French, German and Spanish articles published between January 1, 2000 and November 17, 2017. Reference lists and PubMed-citations of the included articles, as well as former systematic review articles related to the topic, were cross-referenced to retrieve additional relevant studies.

Selection of relevant papers
All studies including adults aged 65 and over, undergoing elective colorectal surgery, were eligible for inclusion. The individual components of the applied ERP had to be described in detail, including at least one of the following outcomes for the (subgroup of ) older patients: length of stay (LOS) or post-operative morbidity. Study designs included were randomised and non-randomised controlled trials, before-after studies and prospective and retrospective observational studies. Studies were excluded if they included patients undergoing emergency surgery, if they referred to general guidelines instead of giving a detailed description of their ERP protocol, if they focused on limited (< 5) intervention components, or if no full text was available. Two reviewers (KF, MP) independently screened the titles and abstracts of the articles retrieved by the search and removed duplicate articles, using Endnote. Full texts of relevant abstracts and articles obtained by cross-referencing were assessed for inclusion (KF, MM). Any discrepancies were discussed with a fourth reviewer (MD).

Data extraction and synthesis
Two reviewers (MP, MM) extracted data, each from half of the included studies, using standardised data extraction forms. A third reviewer (KF) doublechecked the correctness and completeness of the extracted data. In case of disagreement, a fourth reviewer (MD) was consulted. The following characteristics of the included studies were extracted: first author, year of publication, country and setting, study design, study population, in-and exclusion criteria, sample and age distribution, and the subsample of older people in an ERP included in this review. The reported ERP intervention components were grouped into 20 key components, in line with the 2012 ERAS® Society guidelines for peri-operative care in elective colonic and rectal/pelvic surgery [3,4]. The total number of studies including a certain ERP intervention component as well as the total number of ERP intervention components per study were calculated. Apart from LOS and post-operative morbidity, the following outcomes, if available, were reported: post-operative mortality, time to reach discharge criteria, adherence to ERP components, 30-day readmission rate and reoperation rate. Post-operative morbidity and mortality were considered in-hospital and within 30 days of surgery. Post-operative morbidity was reported as the percentage of patients experiencing post-operative complications (severity grades I to IV according to the Clavien-Dindo classification) [13]. If the study reported the percentage of older patients in which an intended ERP component was actually applied, this percentage was considered as the adherence to that ERP component. The most frequently reported discharge criteria, also referred to as ERP recovery goals, were summarised by reporting the time (in days or hours) to reach each recovery goal (i.e. no morbidity evidence, ambulation, first flatus, first stool, oral intake, pain control with oral drugs) [1,14].

Risk of bias (quality) assessment
The methodological quality of the included studies was assessed independently by three reviewers (KF, MP and MM) through the Methodological Index for Non-Randomised Studies (MINORS), containing eight items (the maximum item score per item is 2, the ideal global score 16) [15]. Any discrepancies were discussed and agreed upon with a fourth reviewer (MD).

Data synthesis and analysis
Results from individual studies and study groups are provided in the tables as percentages and averages (means or medians). To summarize the data in the manuscript, medians and ranges across study groups were calculated. Due to many differences among studies regarding study population, age groups, ERPs, risk of bias, definitions of outcomes and other methodological aspects, a formal meta-analysis was not performed.

Article selection
The search generated 4562 articles (Fig. 1). After removing 1429 duplicates and excluding 2810 articles based on title and abstract, 323 full texts plus 2 additional articles found by cross-referencing, were evaluated for inclusion. Finally, 21 articles were included . Five of the included articles did not provide ERP details, but were included because the applied ERP was described in detail in another publication [18,21,22,25,26,33]. Although inclusion periods of the two studies published by Braga et al. partially overlapped, we reported the results as two individual studies [19,20].

Study and patient characteristics
Three studies compared ERPs with conventional care in an older age group [16,24,28] (Table 1). Ten studies compared ERPs in old versus young patient groups [17,18,20,22,23,25,27,29,31,35], while 2 studies compared ERPs in different older age groups [19,32]. Three studies described ERPs in a single older age group [21,34,36]. Three studies investigated patient factors that influence ERP outcomes, including age [26,30,33]. A total of 7610 patients were included in all studies, of whom 3495 (46%) patients (those ≥65 years in whom an ERP was applied) were considered in this review. Four studies divided the older patients in different age groups [17][18][19]32]. As a consequence, the 21 studies included in this review comprised 26 study groups.

Risk of bias (quality) assessment
The scores on the MINORS quality and risk of bias assessment ranged from 8 to 14 out of 16 (Additional file 2). Only 8 studies had a clearly stated aim, including the population, the intervention and at least one primary outcome. Although not always divided into primary and secondary endpoints, all studies had endpoints appropriate to the aim of the study. Lower total scores were due to non-consecutive inclusion of patients, non-prospective collection of data, absence of prospective study size calculation, or unblinded assessment of the study endpoints.

Components of the ERP
The number of ERP intervention components in the described ERP protocols varied between 7 and 16, with a median of 13 ( Table 2). All programmes described early post-operative mobilisation, early post-operative oral intake, opioid sparing multimodal post-operative analgesia, early urinary catheter removal, and avoidance of nasogastric tubes as part of the intervention. Prevention of post-operative ileus (by chewing gum, laxatives, or Alvimopan), post-operative nausea and vomiting (PONV) prophylaxis or treatment, avoidance of sedative premedication, and pre-operative optimisation were not mentioned in over half of the studies. None of the ERPs mentioned peri-operative glycaemic control. Anaesthetic protocol information was often very limited. Therefore the scoring for 'standard anaesthesia protocol' was based on the described regional anaesthesia technique, taking into account an evolution in the recommendations over time: In the past epidural anaesthesia was recommended in all patients. In the 2012 ERAS® Society guidelines epidural anaesthesia remains the standard in open surgery, but for laparoscopic surgery spinal analgesia or PCIA are recommended as an alternative [3,4].

Adherence to the ERP
Nine studies reported adherence rates to a minimum of 1 and a maximum of 15 ERP components [19-23, 26, 29, 32, 34] (Table 3). Adherence was the highest for preand intra-operative ERP components, and lower for post-operative components. ERP components with adherence rates ≥80% were: pre-admission counselling, no bowel preparation, limited pre-operative fasting, antithrombotic and antimicrobial prophylaxis, no nasogastric tube, active warming, and limited intra-operative fluids. Other ERP components with adherence rates ≥60% included carbohydrate loading, PONV prophylaxis, and opioid sparing analgesia. 'Early intake of oral liquids' ranged between 49 and 84% on post-operative day (POD) 0 and 'early intake of solid foods' ranged between 51 and 86% on POD 1. Early mobilisation ranged between 55 and 90% on POD 0. Three studies reported 'global compliance' and used the percentage of patients that fully adhered to a number of selected components: global compliance ranged between 56 and 85% [20][21][22].

Outcomes of the ERP
Results from individual studies per study group are given in Table 4. The median post-operative morbidity rate during the hospital stay was 23.5%, ranging from 5 to 37.8% (9 study groups) and was 29.8% at 30 post-operative days, ranging from 18.8 to 52.6% (13 study groups). The in-hospital mortality rate was 0% in 8 study groups, and 1.2 and 1.6% in two others. The median mortality at 30 days was 1.4%, ranging from 0 to 16.2% (17 study groups).
LOS was not reported in a uniform way: four studies reported mean post-operative hospital stay (median among the studies 5.5 days, ranging from 4.7 to 7.3 days). Eleven studies reported median post-operative hospital stay (median among the studies 6.0 days, ranging from 2.5 to 11 days). Two studies reported mean total hospital stay (5.0 and 9.0 days, respectively). Three studies reported the median total hospital stay (6.0, 8.0 and 12.0     18. Early oral intake      Table 4 Outcomes of the ERP in older patients Bardram 2000 [36] Age (years) days, respectively). Two studies reported LOS including readmission days [21,33], and one study reported mean LOS for open and laparoscopic surgery separately (9.3 and 4.2 days, respectively) [35], which explains the absence of LOS-results for these studies in the table.
Among the 16 studies reporting the 30-day readmission rate, the median reported rate was 4.9%, ranging from 1.1 to 25.0% (21 study groups). Among the 12 studies reporting reoperation rate, the median reported rate was 5.0%, ranging from 0.6 to 10.5% (16 study groups). Additional file 3 summarises the attainment of ERP recovery goals. Time to readiness for discharge (TRD or achievement of all recovery goals) was reported in 3 studies with a mean of 6.4, a median of 5.0, and a median of 5.5 days, respectively [19,20,32].

Discussion
This review analysed intervention components, adherence and outcomes of ERPs in 21 studies in older elective colorectal surgical patients.
The median number of ERP intervention components per study was 13 (range 7-16), compared to 10 (range [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] in the earlier review by Bagnall et al. (2014) [11]. When comparing the studies included in our review, those published before 2014 implemented a median of 13 (range 7-15), and those after 2014 a median of 15.5 (range [11][12][13][14][15][16] components. This demonstrates that more recent ERPs implement more ERP intervention components than older programmes do. To facilitate comparison of studies, the ERP interventions were grouped into 20 key components based on the 2012 ERAS® Society guidelines [3,4]. This explains minor differences in the reported number of ERP components per study between this review and the review by Bagnall et al., as well as minor differences with what authors of the included studies report. Although only 4 studies were conducted after the publication of the 2012 ERAS® Society guidelines, the ERP components in these guidelines do not differ substantially from previous consensus recommendations [37,38], nor do they differ from more recent guidelines by other societies (e.g. from the United States, France) [39,40]. It is worth mentioning that some components remain somewhat vague in the guidelines: for example, no exact time frame is given for 'early feeding' and 'early mobilisation' , nor is the kind of food and the intensity of physical exercise specified. The included studies therefore construct their own protocols, which leads to heterogeneity.
Some ERP components, e.g. pre-operative optimisation, antithrombotic and antimicrobial prophylaxis, PONV prophylaxis and treatment, were less often reported than expected. Possibly, these components are considered peri-operative routine and are therefore not systematically mentioned by the authors in the ERP protocol. Tight peri-operative glycaemic control with insulin was not reported in any of the ERPs. We hypothesize this is due to the fact that it only appeared as an ERP component in the latest 1 guideline in 2012, with a warning for the risk of hypoglycaemia in a ward setting [3,37,38].
To allow for a correct interpretation of the effectiveness of the programmes, adherence to individual ERP components, i.e. which ERP components the patients actually receive (dependent on the providers) and which ones they are able to carry out or tolerate (dependent on the patients), needs to be reported as well [18,41,42]. Nine studies (43%) reported adherence, although the difference between interventions offered and tolerated was often unclear, and adherence was only given for a selection of ERP components. Former studies demonstrated that ERP adherence is lower in the post-operative period compared to the pre-and intra-operative period, and this seems equally true for the older patients in our review [43][44][45]. This should not automatically be considered as an implementation failure: declining adherence in the post-operative phase might reflect the development of complications, for which additional measures can be taken [44]. Five studies compared adherence in older versus younger patients [20,22,23,26,29], two in different older age groups [19,32], and two studies described adherence in a single older age group [21,34]. As the age limits of the groups varied and the ERP components assessed were too heterogeneous or limited in number, it was not possible to draw firm conclusions about adherence in relation to age. It was not possible either to draw conclusions about the factors that affect adherence among different age groups, such as location [19,21,26] and invasiveness [21] of the surgery, placement of a stoma [21,26], and cancer diagnosis [19,26]. Only a higher ASA class seems consistently associated with lower adherence [20,26]. Possibly, ability to adhere to the programme in older patients may also be influenced by age-related factors: co-morbidity, frailty, functional impairments, cognitive impairments, ageism, or lack of personnel or materials (e.g. physiotherapists or walking aids) to take care of the more challenging older patients. Nevertheless, none of the presumed age-related factors above were studied.
A secondary objective of this review was to summarise outcomes of ERPs in older patients, without comparison to younger patients or conventional care: the two earlier reviews studied the safety and feasibility of ERPs (by analysing cohort studies comparing different age groups), and their efficacy (by analysing the only two existing RCTs of ERP versus conventional care in older patients [24,28]). They concluded that ERPs are safe and feasible, with a comparable post-operative morbidity in the younger and older patient population in the majority of the studies, and that ERPs had significantly better outcomes than conventional care [11,12]. In this novel review we included 8 new studies (we excluded emergency surgery, which explains why two studies from the previous reviews were not included). The outcomes in the new studies are comparable to those in the older studies. Unfortunately, differences in age ranges of recruited patients, differences in in-and exclusion criteria of the studies, differences in ERP pathways, and differences in definitions of outcomes or poorly defined ERP components and outcomes, and incomplete reporting of adherence preclude further causative analysis. Since the details of individual complications were very heterogeneously reported in studies, we decided to report overall complication rates. Only one study reported the incidence of postoperative delirium [24].
Some methodological aspects need to be taken into account while interpreting the review results. First, only six studies included older patients or subgroups of older patients prospectively [23,24,27,28,31,34]. The other studies were secondary analyses, database analyses or retrospective studies. Second, often only subgroups of patients from the original studies were included. For this reason, it was not possible to give more details about the included older population in terms of ratios for colon versus rectal surgery, laparoscopic versus open surgery, cancer versus benign disease, or ASA class. Third, the many differences among studies precluded meta-analysis of the results. Better quality research and standardised reporting is needed to draw conclusions on the optimal composition and the outcomes of the programme in older patients. Fourth, the included studies only report the chronological and not the biological age of their patients in terms of co-morbidity, frailty, mental capacity and functional dependency. The older patients in this review may be apt to selection bias, as mainly physically and mentally fit patients tend to be recruited in studies, and there might also be a referral bias [46]. A strength of this review is the comprehensive search string with selection of papers in five different languages, without limitations on study design, and without searching for 'older patients'. This ensures a comprehensive overview of publications from the last 17 years describing the components, the adherence and the outcomes in (subgroups of ) patients aged over 65 undergoing elective colorectal surgery.
The data provided by this review will be of added value to guide clinical decision-making and patient counselling. In addition, clinicians and researchers can use this overview as a reference to evaluate their own data. For future studies, there is a need to standardise and further fine-tune definitions of ERP intervention components and outcomes, and to provide guidance for standardised reporting [42,47]. This will facilitate comparison among studies and allow meta-analysis. Adherence for all of the ERP components should be reported in a uniform way, and should reflect which ERP components patients actually receive, as well as which interventions they are able to carry out or tolerate. Large studies should describe patient characteristics, adherence and outcomes for their older patient groups in detail, to enable secondary analysis of the older patient population. Future studies and (inter)national audit initiatives will have to be specifically designed to study older patients in relation to their frailty profile, preferably incorporating a comprehensive geriatric assessment for the evaluation of the older individual's functional, cognitive and psychosocial status, comorbidities and polypharmacy [10,48,49]. Based on these findings, it will be possible to thoroughly describe the older population studied. This will allow to determine whether frail older patients can follow a standard ERP or whether it should be tailored to this specific population, and to interpret outcomes in relation to older patients' profiles.