Comprehensive geriatric assessment predicts mortality and adverse outcomes in hospitalized older adults

Hospitalization is frequently required for the treatment of acute or uncontrolled illnesses and for invasive diagnostic procedures in older adults. Nonetheless, hospitalization is also considered a risk event for these patients [1‐4]. Elderly individuals suffer physiological changes typical of the aging process that make them more susceptible to adverse events during hospitalization, which may result in a series of complications unrelated to the initial cause of admission. These complications may lead to an increased length of hospital stay, functional decline and higher mortality [1]. Furthermore, one in every three hospitalized older adults loses the ability to perform activities of daily living (ADLs), and at least 20% develop delirium during their hospitalization [2, 5, 6]. Even so, evidence suggests that health care providers have low levels of awareness concerning the risks of hospitalization in this population [7].

The early identification of individuals at greatest risk for complications and unfavorable outcomes would enable a more adequate treatment plan and a better allocation of the resources available to the multidisciplinary team [8]. Moreover, while greater efficiency might be achieved in the proposed treatments, patients and families may be better prepared for the subsequent difficulties that follow hospital discharge. For this purpose, a systematic assessment upon hospital admission may be helpful [8].

The term “Comprehensive Geriatric Assessment” (CGA) was first used in the United Kingdom in the late 1930s. Later, its concept, parameters and indications motivated various scientific research studies [9]. The basic concepts and parameters used in CGAs have evolved over the years, including elements of traditional clinical examinations, evaluations conducted by social workers, functional evaluations performed by rehabilitation specialists, nutritional assessments and neuropsychological evaluation methods [10]. Such assessments are traditionally directed to the planning of interventions but have also been described as useful to determine prognoses and outcomes [9‐12].

The present study sought to develop a protocol for conducting a standardized and structured CGA at the time of hospital admission of older patients. We aimed to evaluate the applicability of the proposed model for thoroughly characterizing these patients and analyzed the impact of this strategy on the prediction of mortality and on adverse hospital outcomes.

In total, 746 cases were included in this study from an initial sample of 826 patients; 38 (4.6%) subjects admitted for end-of-life care were excluded; 42 (5.1%) cases were excluded from the analysis because of incomplete assessments. Reasons for inadequate completion of the protocols included the absence of informants accompanying patients with altered cognition (73.8%) and medical staff incompliance (26.2%). Regardless, adherence to the protocol exceeded 95%, and the evaluations took an average of 60 minutes to be performed.

The mean age of the population was 80.7 (±8.1) years, with 65.7% (490) of the participants identified as female and 38.1% (284) as married individuals. The mean years of education were 4.6 (± 3.6), and 37.9% (283) of the patients had low or very low socio-economic levels. At admission, 62.1% (463) of the patients were regularly using 5 or more medications. Further population characteristics and CGA component frequencies are outlined in Tables 1 and 2, respectively.

The overall mortality rate was 12.9% (96), and the leading cause of death was septic shock (46.7%) followed by cardiovascular complications (19.6%) and neoplastic disease complications (12.5%). Characteristics according to all-cause mortality can be found in Table 1. Compared with the patients who were discharged, those who died had a significantly higher number of impaired CGA components (Table 2; Figure 1), including functional dependency, cognitive decline and polypharmacy. Multivariate binary logistic regression indicated that IADL dependency, ADL dependency, malnutrition, poor social support, acute kidney injury and pressure ulcers at admission were all independently associated with in-hospital death (Table 3). The importance of malnutrition markers stood out, and the average score on the MNA was lower in patients who died (14.3 ± 5.9 vs. 18.2 ± 5.4; p < .001), with a good correlation between this score and albumin levels at admission (rho = 0.5; p < .001). Both the BISEP and CIRS-G scores served as predictors of in-hospital mortality in this population, though with a weak correlation between tests (rho = .14; p = .017). We also verified that neither mortality (p = .58), nor the frequency of nosocomial infections (p = .11), delirium (p = .32) or longer hospital stays (p = .11) significantly varied during the study extent.

The number of impairments among the 10 analyzed CGA domains was also significantly associated with the incidence of delirium and nosocomial infections (Figure 1). In total, 88 patients developed delirium while hospitalized, with a 26.1% mortality. These patients were more frequently male (40.9 vs. 28.8%; p < .001), demented (35.2 vs. 15.2%; p < .001), ADL and IADL dependent (75% vs. 40%, p < .001), malnourished (59.1 vs. 34.2%; p < .001), and infected at admission (40.9 vs. 29.6; p = .035). ADL and IADL dependency, malnutrition, and dementia were independently associated with this complication (Table 3). In total, 124 patients had nosocomial infections, of which 48.4% were respiratory infections and 27.4% were urinary tract infections; mortality reached 30.6% in this group. ADL dependency was frequent (72.6 vs. 53.7%; p < .001) and independently associated with nosocomial infections (OR = .89/95%CI = .85-.93/p < .001).

The mean and median lengths of stay were high (16.7 and 12.0 days, respectively), and the following CGA components were related to longer hospitalizations: IADL dependency, malnutrition and history of falls (Table 3). Only 52.0% (454) of the subjects were independent in all ADLs at admission and 32.7% (337) in all IADLs. Notably, 3 months before admission, the mean ADL and IADL scores were 8.9 ± 4.1 and 8.5 ± 7.1, respectively, while at admission, these scores had decreased to 7.3 ± 4.7 and 7.2 ± 6.2, indicating a significant functional decline (p < .001). Despite these findings, there were no significant changes in overall functional status during the hospital stay, nor were factors identified that could reliably predict the functional evolution throughout this period.

The importance of CGA emerges in environments such as the geriatric ward, recognizing that not only medical conditions but also social, neuropsychological, nutritional and environmental factors are crucial to the clinical evaluation [9, 10, 49, 50]. In such settings, we find a high frequency of individuals with cognitive impairment, functional dependence and malnutrition, as demonstrated in our results.

Recent studies have also investigated the use of CGA as a prognostic instrument and concluded that several of its components are cornerstones for clinical decision-making [12, 51, 52]. Our model proved valuable precisely in the detection of these key aspects and demonstrated that the functional, cognitive, nutritional and social components of CGA are predictors of in-hospital mortality. Various CGA domains also predicted other adverse outcomes, such as delirium incidence, nosocomial infections and longer hospital stays. Functional dependency was an especially important predictor of these events. Knowing this, early rehabilitation strategies are followed in our unit and possibly explain why no significant changes in overall functionality were observed throughout the hospitalization. Previous studies proposing early rehabilitation interventions have been able to prevent in-hospital functional decline, though not necessarily reverse it, indicating that post-discharge programs are essential to return patients to independence [53, 54]. Likewise, nutritional support and supplementation should be considered. The negative impact of social deprivation on prognosis is well established but poorly understood, and additional work is necessary to understand how to alleviate its effect on morbidity and mortality [55]. Finally, the importance of other indicators of clinical severity should not be forgotten, demonstrated herein by the impact of acute kidney injury on prognosis and by the usefulness of scales that reflect burden of illness, such as the BISEP and CIRS-G [56].

We verified an improvement in the detection of cognitive deficits, particularly the advance in the diagnoses of chronic conditions and the early identification of acute confusional states. That possibly 40% of cases of dementia had gone undetected in the outpatient setting is worrisome and should trigger a revision of follow-up strategies. Regarding the recognition of delirium at admission, previous data in the same setting, prior to the routine application of the CAM, indicated a prevalence of the condition of only 5.2% [6]. After the inclusion of the instrument in our CGA, the number increased almost four-fold. Though not independent predictors of mortality in this population, we confirmed the importance of prevalent delirium and dementia as associated factors to unfavorable outcomes and their accurate recognition is essential to potentially improve the quality of in-hospital care.

A limitation to this study is that we did not collect data to formally recognize frail individuals—a subset of patients for whom CGA can be particularly useful. The high frequency of multi-component CGA abnormalities that was found indicates that this was a high-risk group for the development of geriatric syndromes and that many subjects were likely to be frail individuals. Future studies on the association between in-hospital CGA and frailty characteristics are necessary to better understand the syndrome in the hospital setting.

A drawback that restricts the systematic implementation of CGA is that it is time consuming, as we observed in our results. However, we also found that hospitalization, by allowing more time to assess each patient, provided the possibility for a detailed and structured clinical evaluation. Regarding the results, the elevated presence of totally dependent and cognitively impaired patients associated to a floor effect in the functional measurements that were employed, might have played a part in the lack of functional variability that was described. Also, despite the subjects of this analysis having similar characteristics to those of previously reported studies [8, 57], this was a single-center study and our findings have limited generalizability.

Another limitation is that we did not address the long-term effects of using CGA in hospitalized older adults. Research focusing on its impact on post-discharge mortality, institutionalization and re-admissions should be pursued. Furthermore, controlled studies would be helpful to establish causality relations and to eliminate confusion factors. Homogeneous models of assessment must be further developed for the results to be comparable and for the best assessment strategies to be identified [58‐61].

The systematic incorporation of a standardized and scientifically based method of baseline assessment of hospitalized older patients aims to optimize patients’ clinical and functional outcomes and quality of life by increasing the overall detection of modifiable factors and implementing adequate care. The validity of CGA for identifying factors associated with the occurrence of death and other adverse outcomes in the setting of a geriatric in-patient unit was shown herein, as was the importance of thorough cognitive, functional, social and nutritional evaluations. Such care is critical to elucidate fundamental conditions for the therapeutic decision process.