Effects of Comprehensive Geriatric Care Models on Postoperative Outcomes in Geriatric Surgical Patients

A Systematic Review and Meta-analysis

Aparna Saripella; Sara Wasef; Mahesh Nagappa; Sheila Riazi; Marina Englesakis; Jean Wong; Frances Chung


BMC Anesthesiol. 2021;21(127) 

In This Article


We registered the protocol of this systematic review in the International Prospective Register of Systematic Reviews (PROSPERO) (registration number - CRD42020181779). The study was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.[25]

Definition of CGA: Comprehensive geriatric assessment (CGA) is a multi-dimensional, multi-disciplinary process which consists of medical, mental, social and functional needs of the elderly people, and an integrated and coordinated care plan that includes treatment and long term follow up.[26]

Study Selection Criteria

Inclusion criteria were: 1) randomized and non-randomized controlled studies, prospective and retrospective cohort trials, that enrolled patients aged over 60 years, undergoing elective non-cardiac high-risk surgery; 2) must have CGA as a component of the geriatric care model; 3) must have an intervention group using geriatric care model and a control group (standard care); 4) reported at least one of the following postoperative outcomes in both the intervention and control group: prevalence of delirium, LOS, 30-day readmission rates, 30-day mortality, and any other postoperative complications; and 5) limited to the English language.

Exclusion Criteria were: 1) emergency surgical procedures and non-geriatric population studies. 2) ERAS program without a CGA component.

Search Strategy

We searched Medline, Pubmed, Embase, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, Emcare Nursing, Web of Science, Scopus, CINAHL ClinicalTrials. Gov, and ICTRP (international Clinical Trials Registry Platform) for published and unpublished studies. The search strategy was developed with the help of information specialist (ME). We used both Medical Subject Headings (MeSH) and free text terms to identify relevant articles. Database searches were restricted from January 2009 to January 2020. The search strategy used controlled vocabulary terms and text word terms for each of the research topic components: care pathways and elderly and perioperative and study types. The full electronic search strategies used are shown in the Supplemental Digital Content (Appendix 1).

Study Process

The study authors prepared the pilot tested data collection form with the standard instruction for screening of the title, abstract, and full text, risk of bias assessment, data collection, and data analysis. Two reviewers (AS, SW) screened literature studies[27] (using Rayyan), assessed the risk of bias, collected data, and analysed independently. All conflicts were resolved by consensus and a third reviewer (FC).

Risk of Bias Assessment

For randomized controlled trials (RCT), we used a modified version of Cochrane Collaboration's tool to assess the risk of bias.[28] The Cochrane tool; studies received a "low", "high", or "unclear" rating for each risk category. We considered random sequence generation; allocation concealment; blinding of outcome assessment; incomplete outcome bias; and selecting outcome reporting. The domain "blinding of participants and personnel" was removed from the quality assessment, as it was difficult to blind the CGA group (S-Table 1). For non-randomized studies, we evaluated study quality in accordance with the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines[29] and Newcastle-Ottawa scale (NOS).[30] Quality Assessment is included in S-Table 2. The key points of study quality reviewed included: i) a clear identification of the study population, ii) a clear definition of the outcomes and outcome assessment, iii) no selective loss of patients during follow-up, and iv) important confounders and/or prognostic factors identified. We evaluated each point using Yes/No. If one of these key points was not clearly mentioned in a study, it was considered a 'No'. Each study was given a score using the Newcastle-Ottawa scale (S-Table 3).

GRADE - Quality of Evidence

We assessed the quality of the evidence by Grading of Recommendations Assessment, Development, and Evaluation (GRADE). GRADE system includes the risk of bias, inconsistency, indirectness, inaccuracy, and publication bias. For each outcome, GRADE starts with a baseline rating of high (4 points) for RCT and low (2 points) for observational studies. The outcome rating then can be adjusted (downgraded) after considering the 5 assessment criteria (S-Table 4).

Data Extraction

Two reviewers (AS, SW) independently extracted the data using a standardised data collection form. The study characteristics for instance, author, year of publication, country of origin, study design, total sample size, sample size in intervention group, and control group were collected. Patient characteristics including age, gender, co-morbidities, surgical procedure, details about the intervention, and perioperative care/multidisciplinary teams were extracted.

Comprehensive geriatric care models which used CGA as the intervention were compared to control groups which received standard care with no intervention. We collected the primary and secondary outcomes from surgery to 30 days post-discharge from hospital. The primary outcomes reported were prevalence of delirium and LOS. The secondary outcomes were 30-days readmissions rates, 30-days mortality rates, and a number of other postoperative complications.

Data Analysis

We reported the results according to meta-analysis of observational studies in epidemiology (MOOSE)[29] and preferred reporting items for systematic reviews and meta-analyses (PRISMA).[25] We pooled the data from RCTs and observational studies. Additionally, we explored the heterogeneity and pooled estimate based on the study type (RCTs vs. non-RCTs). The measure of association for postoperative outcomes was the weighted odds ratio (OR) with 95% confidence interval (CI) for dichotomous outcomes (delirium, readmission, mortality etc.), and the weighted mean difference (WMD) with 95% CI for the continuous outcome of LOS. The Mantel-Haenszel (M-H) method was used to combine dichotomous events, and the inverse variance method was used to combine continuous events. A high statistical heterogeneity was explored. In this analysis, the impact of each study on heterogeneity was explored by excluding studies one by one and recalculating the heterogeneity. We undertook sensitivity analyses for significant outcomes, by using alternative effect measures (odds ratio vs. risk ratio), pooling methods (Peto methods vs. Mantel-Haenszel method), and consideration on heterogeneity (random vs. fixed effect).