Introduction

The coronavirus disease 2019 (COVID-19) pandemic began to have a major impact on society in 2019 [1] having unprecedented consequences on global health and economic systems.

In developed European countries with a very high older population the COVID-19 mortality was 83.7% for people > 70 years and 16.2% for people younger than 69 years in 2020 [2] with a higher prevalence of COVID-19 deaths in nursing homes (NH) [3]. Health problems and geriatric syndromes associated with ageing also determined the risk of mortality [4,5,6,7,8,9].

At the beginning of the pandemic Spain had a total of 326,613 people institutionalised in NHs [10] and a study conducted in Madrid reported a 14% mortality rate in older adults with COVID-19 in NHs [7]. The COVID-19 mortality has already been extensively studied in several countries, although most of these follow-ups have not exceeded 1 year [11]. Longer follow-ups would enable more accurate data and the identification of predictive factors that may be relevant to clinical practice.

The main aim of the study was to estimate overall and COVID-19 mortality parameters and analyse their predictive factors in older people living in NHs over a 2-year period.

Methodology

Study design and population

This is a 2-year multicentre observational cohort study. The study was conducted in five NH in Central Catalonia, Spain. It was designed following the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) standards for cohort studies [12, 13]. Residents aged 65 years and older permanently living in NH were included. Those in a coma or palliative care (short-term prognosis), those who refused (or their legal guardian) to participate in the study and those who left the NH during the 2‑year cohort period were excluded.

Sample size

For the calculation of the sample, the article by Burgaña Agoües et al. (2021) was taken as a reference due to its methodological similarity to the present article: Burgaña studied pandemic mortality due to COVID-19 in Spain, in people over 65 years of age resident in NH. Considering the findings of Burgaña Agoües et al. (2021) [14], i.e. the difference in proportions between individuals with severe functional impairment (23.0%) and deceased (11.1%), and with a confidence interval (CI) of 95% and a power of 80%, a sample size of 122 participants was estimated.

Study procedures

The primary outcomes were all-cause mortality and COVID-19 over the 2‑year follow-up period. The mortality registry included cause and date of death, deaths in total, deaths due to COVID-19 including confirmed cases, and deaths due to symptomatic suspicion of COVID-19 [15]. Additional COVID-19 data collected included the presence of the disease, whether they had symptoms [16], the performance of COVID-19 screening tests such as C‑reactive protein (CRP), serological tests, and/or rapid antigen tests for SARS-CoV‑2 (RAT) [17]. The information was collected through on-line interviews with NH professionals as due to COVID-19, they could not be accessed.

Sociodemographic and health information was obtained from health centre records and cross-checked with health professionals. All sociodemographic variables and those described in this article were collected at baseline (January 2020), just before the onset of the COVID-19 pandemic in Spain. Falls (number) during the last year were obtained from NH registers. Nutritional status was assessed using the mini nutritional assessment (MNA) [18]. The SARC‑F [19] was used to identify individuals at risk of developing sarcopenia. Functional capacity was measured using the modified Barthel index and results were classified according to the degree of dependency as: independent, slightly dependent, moderately dependent, severely or totally dependent [20]. Continence status was reported using section H of the minimum data set (MDS) version 3.024 [21]. Cognitive status was assessed using the Pfeiffer scale [22] and frailty using the clinical frailty scale (CFS) [23]. Sedentary behaviour (SB) and waking-time movement behaviours (WTMB) were assessed using the ActivPAL 3TM activity monitor (PAL Technologies Ltd., Glasgow, UK) [24].

The study was conducted over a 2-year period and ended in March 2022.

Statistical analysis

The nominal and ordinal quantitative variables were expressed according to frequency in percentages and the quantitative variables with mean and standard deviation (SD). Survival curves were formed using the Kaplan-Meier method and multivariate analysis was performed by Cox regression, using the hazard ratio (HR) as the measure of effect. The Statistical Package for the Social Sciences 27 (SPSS Inc., Chicago, IL, USA) was used for the analysis.

Results

We recruited 125 people, 67.6% of the total number of NH residents in the main study. Finally, 7 (3.8%) participants who left NHs to reside elsewhere were excluded (Fig. 1).

Fig. 1
figure 1

Flow chart of the sampling process

Full size image

The mean age of the participants was 85.1 years (SD = 7.3 years) and 104 (83.2%) were female. The mean number of months living in NH was 27.5 months (SD = 112.14 months). The analysis of health and sociodemographic variables is described in Table 1.

Table 1 Descriptive analysis of the sample of institutionalized older adults living in nursing homes of Central Catalonia, Spain (n = 125)
Full size table

In the 2‑year period from baseline to the end of the study, 59 participants (47.2%) died, of whom 25 (20.0%) died from COVID-19 and 34 (27.2%) from other causes. All COVID-19 deaths occurred in the first year of the study: 44 (74.5%) of the 59 individuals had already died within the first 90 days (at the peak of COVID-19).

Survival and associated factors according to the variable mortality

In the bivariate analysis, mortality was associated with functional impairment, urinary incontinence (UI), faecal continence, risk of sarcopenia, % of waking time in SB and with a p-value of less than 0.05. All other health and sociodemographic variables were not significant (Table 2).

Table 2 Association of variables to health, functional and sociodemographic1 with mortality (n = 125)
Full size table

The variables were tested for collinearity and none of them showed collinearity with each other. A multivariate analysis was performed with the model with adjusted values including the variables age with severe functional impairment and risk of sarcopenia. The result showed that functional impairment predicted mortality independently of age (which was not statistically significant) and sarcopenia risk (Table 3).

Table 3 Survival univariate and multivariate Cox analysis1 in older people living in NHs (n = 125)
Full size table

Survival and associated factors according to the variable COVID-19 or other-cause mortality

In the univariate analysis, functional impairment, living in a private NH, being older than 86 years, malnutrition and being female were risk factors for COVID-19 mortality. Functional impairment was associated with mortality from other health causes with a p-value of less than 0.050 (Tables 4 and 5; Fig. 2).

Table 4 Analysis of the association of health and sociodemographic variables1 NH residents with mortality (COVID-19 or other causes) (n = 59)
Full size table
Table 5 Univariate Cox analysis: association of COVID-19 mortality with other causes of death in relation to covariables1 in older people living in NHs (n = 59)
Full size table
Fig. 2
figure 2

Kaplan-Meier survival estimates

Full size image

We tested for collinearity between the variables and none of them were collinear with each other. The number of individuals in this variable is 59. Different combinations of significant variables, such as functionality, age and type of NH among others, were tested by multivariate analysis, but no significant results were found (Table 5).

Discussion

The main objective of this study was to examine the incidence of all-cause and COVID-19 mortality and to analyse the predictive factors in older NH residents over a 2-year period since the onset of the pandemic.

The results indicate that almost half of participants died with 20% being attributed to COVID-19. Most of the deaths (74.5%) were in the first 3 months of the study, coinciding with the outbreak of the COVID-19 pandemic in Spain. In the second year of the study, the survival curve became horizontal again, after the implementation of preventive measures. This study shows a higher incidence of mortality in the 1‑year period than other studies. Several articles on the pandemic phase report data on excess mortality [11]. A study in Barcelona reported a 3-month COVID-19 mortality rate of 11.1% in institutionalised older people [14]. For deaths from other causes, they reported excess mortality among institutionalised cases (34.8%) [14].

We also report the association of health, social and demographic variables with mortality: functional impairment, UI, sarcopenia risk and % of waking time in SB were found to be factors associated with mortality and functional impairment, type of NH and age were associated with COVID-19 mortality. The literature shows that UI and risk of sarcopenia are associated with mortality [25,26,27] and those who spent more time in SB had a higher risk of mortality [28].

Unlike other studies, our data show an increase of mortality in private NHs [29]. The NH type and size influenced mortality during the COVID-19 pandemic in other studies [29]. Braun et al. (2020) attributed these results to the lack of organisation and shortage of personal protective equipment (PPE) in private NHs [29].

This study has the limitation of the COVID-19 pandemic, which impeded access to NHs, increased deaths in older people and did not enable us to have a larger sample.

The strength of the study lies in the telematic data collection in NHs. This enabled us to extract real information on the health and social status of residents. The fact that we collected data prior to the start of the pandemic allowed us to make a comparison of the health and sociodemographic status of institutionalised older people. By having a cross-sectional analysis of the prepandemic sample, we provide data on the factors that predicted the risk of dying in a 2-year follow-up.

Conclusion

Almost half of this sample of NH residents died during the 2‑year observation period. One fifth of deaths were attributed to COVID-19 mostly in the first quarter, coinciding with the peak of the pandemic. Functional impairment was a risk factor for overall mortality and COVID-19 mortality, independent of age and risk of sarcopenia.