The implementation of a CGA in clinical practice and the number of publications on this subject has greatly increased during the last few decades. As “usual care” evolves with new evidence, thwe researchers limited the researchersour research to research published since 2010 to limit heterogeneity in control groups. the researchers chWe chose to develop research algorithms for the researchersour narrative review, even though algorithms are not commonly used in narrative reviews, to ensure the better objectivity of the researchersour work. In the researchersour research algorithm, the researchers rewe required the presence of the term “geriatric assessment”. This means that the researchers we did not consider mono-domain interventions in the absence of an initial global geriatric assessment in order to avoid confusing results. This choice may have led us to reject potentially relevant articles, such as the study by Gilbert et al., which showed that perioperative nutritional management overseen by a geriatrician and a dietician (without a CGA) improves adherence to international guidelines for older patients scheduled for colorectal cancer surgery, with no significant impact on adverse events ^[12][30]. A study by Ommundsen et al., which was a monocentric trial regarding preoperative geriatric assessment and global intervention, was also not selected, as no specific geriatric domains were cited in the title and abstract ^[13][31]. GIn this study, geriatric intervention failed to reduce grade II–V complications in frail, older patients scheduled for colorectal cancer surgery. This may be linked with the absence of teamwork, with a unique “medical doctor specializing in geriatric medicine” who carried out one session evaluation and intervention based on comorbidity management, a drugs review, and recommendations to prevent delirium and malnutrition. the researchers We focused on patients over 65 years old, even though 60-year-old patients are sometimes considered as “older patients” in the oncology literature ^[14][32]. As shown here, the specific impacts of CGA-based interventions on cancer patients have not yet been demonstrated, although some studies have shown positive results for older people in geriatric care settings. Therefore, the use of the CGA in a different population may be questionable. Symptoms such as fatigue, weight loss, or depression may be associated with various physiopathological processes and prognoses, and thus require different interventions according to age, comorbidities, and frailty. Some domains assessed in the CGA are more commonly used to select patients and adapt treatment than to develop interventions. For instance, the efficacity of interventions dedicated to cognitive impairment or fatigue are more difficult to achieve than those concerning other CGA domains, as shown in the ouresearchers results. Nevertheless, the efficacities of intervention protocols such as the HELP program in the prevention of delirium in hospitalized older patients have been demonstrated ^[15][33]. Therefore, interventional studies remain necessary so that their impact on older patients with cancer and on a treatment’s cognitive toxicity or chemo brain can be evaluated. Other interventions could be designed to reduce adverse events and optimize the adherence of patients with cognitive impairment (e.g., optimizing home intervention professionals, organizing relief periods to prevent helpers’ exhaustion, daily or multi-daily home nurses’ interventions, home calls, and electronic, customized devices to help individuals to remember to take drugs and evaluate the observance (e.g., a communicating pill box)).

Multi-domain interventions seem promising, especially when they are based on global assessments. Three recent, good-quality clinical trials assessed multi-domain interventions on chemotherapy toxicity: GAIN ^[16][27], GERICO ^[17][28], and GAP70+ ^[18][29]. GAIN was the most convincing trial, with a CGA-based intervention characterized by predetermined assessment thresholds. A multi-disciplinary team trained in geriatrics, which included a geriatric nurse practitioner in charge of follow-ups, delivered it. Here This study was strongly focused on the management section of the CGA. In the control group, the oncologist had access to the results of the CGA (as in the intervention arm) but implemented fewer interventions (12.5% of identified interventions were implemented, versus 76.8% in the intervention arm).  The present review highlighted the heterogeneity of possible interventions, control groups, and endpoints. Considering the variability of patients included in these studies, a meta-analysis cannot be undertaken.

As shown in Table 1, both the intervention and control groups were quite different in these three trials. The intervention was either delivered by a team (GAIN), a geriatrician alone (GERICO), or an oncologist informed by recommendations derived from the CGA (GAP70+). The control groups were also different, with no CGA in the GERICO trial and a complete CGA sent to the oncologist in the GAIN trial. So, the control group in the GAIN trial seemed similar to the intervention group in the GAP70+ trial. The authors tried to base the interventions on guides ^[18][29] or thresholds ^[16][27], but standardization seemed difficult to implement considering the lack of evidence for each domain. The ongoing study “PREPARE”, which aims to evaluate, in a randomized controlled trial, the impact of CGA intervention programs in older adults with cancer, addressed this issue with a geriatric care protocol written by an independent committee and a mandatory centralized nurses’ training program to standardize intervention programs ^[19][34].

Table 1.

Comparison of the three randomized trials on global CGA-based interventions.

References	Population (Age, Type of Cancer)	Number of Patients	Type of Study	CGA Intervention	Results
					Primary Outcome	Secondary Outcomes
Li et al. [16][27] GAIN	65+, solid cancer	613	RCT Single-center	Intervention and referrals, based on predetermined thresholds, delivered by a multi-disciplinary team (oncologist, nurse practitioner, social worker, physical/occupation therapist, nutritionist, and pharmacist) Follow-up by the geriatric nurse practitioner Control group: CGA is sent to the oncologist	AE grade 3–5: reduction in CGA group of 10.1% (95% CI 1.5–18.2; p = 0.02)	More completion of chemotherapy treatment plan in the CGA group: 28.4% vs. 13.3%, p < 0.01
Lund et al. [17][28] GERICO	70+, colorectal cancer	142	RCT Single-center	Intervention and referrals, tailored by a geriatrician Follow-up after two months or more frequently Control group: no CGA is performed; no change in chemotherapy treatment	Completion of chemotherapy treatment plan: 45% vs. 28%, p = 0.00366.	Less AE grade 3–5 in the CGA group: 28% vs. 39%, p = 0.156
Mohile et al. [18][29] GAP70+	70+, incurable solid tumors or lymphoma	718	Cluster-randomized trial	Geriatric assessment summary and management recommendations (including dose reduction) sent to the oncologist Control group: oncologists received alerts for impaired depression or cognitive score	Relative risk of AE grade 3–5 in CGA group of 0.74 (95% CI 0.64–0.86; p = 0.0001)

thWe researchers note that some endpoints are “oncologic” (e.g., treatment completion, toxicity, and recurrence time) and others are “geriatric” (e.g., functional or nutritional status and deprescription). The CGA can be used to tailor interventions, but its components are sometimes valuable endpoints (e.g., functional or cognitive status). the researchers We believe that global endpoints, including the impact of interventions on cancer evolution or patient-centered goals, should be preferred. Overall mortality rates, quality of life, or time spent in hospitals seem interesting outcomes. Concerning ongoing studies, the authors of the PREPARE study ^[19][34] choose a co-primary endpoint encompassing overall survival rates and health-related quality of life, whereas Røyset et al. preferred physical function, with quality of life as a secondary objective ^[20][35].

These points underline the issues that can arise when designing CGA-based intervention trials. Older patients with cancer usually present several altered domains, limiting the clinical impact of monodomain compared to multi-domain interventions. Monodomain interventions do not seem to be efficient enough, unless the population is highly selected, which decreases the external validity. In this context, two ongoing randomized controlled trials opted for a selected population of patients with colorectal cancer to evaluate a monodomain intervention for patients addressed for surgery ^[14][32] or a multi-domain intervention for patients addressed for radiotherapy ^[20][35].

The management section of a CGA is a complex intervention that cannot be completely standardized, unlike the assessment section. Randomized controlled trials have been designed to evaluate new drugs, but they may not be the best way to evaluate complex interventions that are highly variable in clinical settings, depending on the provider and the context ^[21][36]. In The researchers our review, we chose to include observational studies that are frequently more important sources of knowledge in gerontology than randomized controlled trials. Randomization, if chosen, should be clustered as much as possible to avoid contamination bias. Researchers may thus have to make compromises between internal and external validity. Internal validity refers to the reliability of the results and requires a strong standardization of interventions, as well as so-called “usual care”. External validity refers to the generalizability of results and implies flexible real-life practices and the accurate description of contexts ^[22][37]. Stepped wedge trial designs seem promising to achieve both internal and external validity. This design allows for the progressive implementation of innovation in different settings, together with its evaluation ^[23][38], whereas randomization can be ethically questionable, as the comprehensive geriatric assessment is currently largely recommended. This ethical concern could lead to a recruitment bias in randomized trials if clinicians have access to the CGA outside clinical trials and prefer not to randomize the frailest patients. Stepped wedge trials remain difficult to implement and costly, as they require the collaboration of multiple centers of inclusion. However, it is not impossible in the geriatric field, as demonstrated in the study by Gilbert et al. ^[12][30].