Natural course of care dependency in residents of long-term care facilities: prospective follow-up study

The present prospective follow-up study was conducted within the framework of the CRANBERRY trial. The Cranberry trial is a double-blind randomized placebo-controlled multi-center trial, in which a total of 21 LTCFs from the University Nursing Home Research Network in South-Holland, the Netherlands, participated (trial registration NTR1266). The CRANBERRY study assesses the effectiveness of cranberry capsules to prevent urinary tract infections in vulnerable older persons living in intramural care settings in which care for the most vulnerable older persons is provided by a multidisciplinary team including elderly-care physicians, nursing assistants, licensed practical nurses, registered nurses and paramedical professionals. Residents aged 65 years and over were included. Excluded were residents with a life expectancy shorter than 1 month or using coumarin. For detailed information on the study design and outcomes we refer to the publication of the original trial [20].

The Medical Ethics Committee of the Leiden University Medical Center approved the study. Written informed consent was obtained from all participants. For participants incapable of giving informed consent due to cognitive impairment, a guardian provided written consent.

At baseline, and at 6 and 12 months follow-up, an assessment was made of the care dependency status by interviewing the responsible nurses who care for the participants. For this the Care Dependency Scale (CDS) was used, which is a tool completed by nursing staff for assessment of the care dependency status of institutionalized residents [21]. The CDS has satisfactory reliability and validity [22‐24], and consists of 15 items, measuring basic care needs on a 5-point scale. The total CDS score ranges from 15 (completely dependent on care) to 75 (almost independent of care). The CDS 15 items are eating and drinking, continence, body posture, mobility, day and night pattern, getting (un)dressed, body temperature, hygiene, avoidance of danger, communication, contact with others, sense of rules and values, daily activities, recreational activities and learning ability.

Since women and men differ in their baseline care dependency status and the CDS scores were not normally distributed, women and men were separately ranked into gender-specific 33% groups according to their baseline CDS score. Thereafter, we combined the lowest, middle and highest 33% for women and men, to generate three gender-specific CDS groups. The ‘low score’ CDS group indicates participants most dependent on care and the ‘high score’ CDS group indicates participants the most independent of care.

The difference in the cumulative incidence of mortality between the CDS groups was explored with Kaplan-Meier curves, with corresponding log-rank test. Cox proportional hazards models, adjusted for age, gender, cranberry use, LTCF, length of stay on the ward, somatic comorbidity (myocardial infarction, stroke, cancer, diabetes mellitus, COPD, urine incontinence, and urinary tract infection in the preceding year) and dementia were used to present mortality risks based on continuous CDS score at baseline.

The change in care dependency for survivors between 0–6 months and 7–12 months is presented by the number of participants in the three CDS groups who improved, stayed stable, or degraded to another CDS group during the two 6-month periods. For the analysis of CDS change in the subsequent 7–12 months, participants were newly classified in gender-specific 33% groups at the 6-month CDS assessment.

Analyses were performed with IBM SPSS Statistics for Windows, version 20.0.

Table 1 presents the baseline characteristics of the total population and of the three CDS groups. Overall, almost 75% of the study population was female and the median age was 84 (IQR 79–88) years. The median CDS score was 44 (IQR 30–56). At baseline, women had a lower CDS score compared with men: 43 (33rd percentile 34, 66th percentile 51) vs. 47.5 (33rd percentile 37, 66th percentile 55); Mann–Whitney U-test; P = 0.013.

There was no significant difference in age between the CDS groups (Kruskal-Wallis test; P = 0.180). The CDS score was negatively correlated with the length of stay on the ward: participants who stayed the longest had the lowest CDS scores (Kruskal-Wallis test; P < 0.001). There were no significant differences between the CDS groups for cranberry use, myocardial infarction, stroke, cancer, diabetes mellitus, COPD and urinary tract infection in the preceding year. However, urine incontinence and dementia were more frequently present in the low CDS group compared with the other groups (Table 1).

Figure 1 presents the mortality rate for the three CDS groups; the highest mortality rate was in the group with the lowest CDS score (log rank 32.2; df = 2; P for trend <0.001).

The mortality risk at 12-month follow-up, based on continuous CDS scores at baseline are presented in Table 2. The crude analysis shows, that per point decrease in CDS score, the mortality risk increased with 2% (HR 1.02; 95% CI 1.01-1.03). The adjusted models showed similar results.

Additional analysis showed that a one point decrease in CDS score between 0 and 6 months was related to an increased mortality risk of 4% during the subsequent 6 months follow-up, adjusted for baseline CDS score (HR 1.04; 95% CI 1.03-1.06).

Table 3 shows the variation in care dependency among survivors for the CDS groups at 0–6 months (n = 714) and 7–12 months (n = 564) follow-up, based on the 33% gender-specific CDS score at baseline for the first 6 months and 33% gender-specific CDS score at 6-month follow-up. The pattern of ‘improvement’, ‘being stable’ and ‘degradation’ of care dependency was almost similar over the two 6-month periods.

Table 4 presents the results of the crude and adjusted multivariate linear regression analysis at 6 months. The adjusted model at 6 months showed that gender, age, baseline CDS score, cancer, urine incontinence and dementia predicted an accelerated decrease of dependency scores at 6 months. Cranberry use, LTCF, myocardial infarction, stroke, diabetes mellitus, COPD, and urinary tract infection in the preceding year, were not associated with the CDS score at 6 months. We did not find co-linearity between the dependent variable (CDS-score at 6 months) and the independent variables in both the crude and adjusted model. The Variance Inflation Factors ranges between 1.0 and 1.4. The multivariate linear regression model for men and women separately showed similar results (data not shown).

In studying the natural course of care dependency, the relation between care dependency and mortality is important. It can be hypothesized that higher care dependency leads to higher mortality risk. There are a few studies unraveling this relation. The study of Marengoni et al. showed that baseline disability was a strong predictor for mortality, independent of number of diseases [25]. Also Chen et al. showed that the sum of care problems, independent of comorbidity, is a predictor of 12-month mortality in LTCF residents [26], and Ferrucci et al. concluded that mortality after severe disability onset was high [27]. Within our study, we found similar results. A one point decrease in baseline CDS score was related to a 2% higher mortality risk in the forthcoming 12-months, also when adjusting for age, gender, cranberry use, LTCF, length of stay on the ward, comorbidity and dementia.

A recent Swiss study among 10,199 nursing home residents (70% women, 74% aged 80 years and above) observed a decrease in activities of daily living (ADL) of 35% and an increase in ADL of almost 14% among residents, within a period of median 6 months (SD 3 months) [12]. They used the Minimum Data Set Activities of Daily Living (MDS-ADL) and looked at ADL performance as primary outcome. Another study in low ADL-dependent LTCF residents in the USA, found that 69% of these LTCF residents with higher physical function remained stable in their ADL performance during 6 months of follow-up [13]. Our study shows a similar trend for care dependency. The majority of the LTCF residents remained stable in their care dependency status, only 10% improved and 25% deteriorated. The variability in the pattern of ‘improvement’, ‘being stable’ and ‘degradation’ of care dependency varies in a similar pattern over two subsequent 6-month periods.

It is known that LTCF residents with cognitive impairment experience a deterioration in their ADL performance [12, 13, 15], which make them increasingly dependent on nursing care.

As mentioned earlier, nutritional status [12, 13], cognitive impairment [12, 13, 15], absence of daily contact with proxies [12], depression [12, 16], neuropsychological deficits [17], incontinence [12, 13, 18] and infections [19] were mentioned as predictors for deterioration in ADL performance. The study of Dijkstra et al. showed that the degree of care dependency at entry to the study was one of the strongest predictors of follow-up CDS ratings [14]. Our study confirms that the baseline CDS score is predictive, but showed also that gender, age, urine incontinence, dementia and cancer; predict an increase in care dependency over time.

The present study included a large sample of 890 residents residing in 21 Dutch LTCFs. Our study participants represent a vulnerable population; with a median age of 84 years and a high dependency on nursing care (median CDS score of 44 points). This median baseline CDS score is comparable with that of other studies in nursing homes [11, 28]. Because a recent international comparison of the CDS demonstrated its usefulness for comparative research across countries [28], the results of the present study might be generalizable to LTCFs worldwide.

Daily nursing care in LTCF focuses on residents care dependency as a process in which the residents’ self-care decreases, and in which care demands makes a person increasingly dependent on nursing care [7]. Although other instruments to assess care dependency are available (e.g. the MDS-ADL, Barthel index [29], or Katz [30]) we decided to use the Care Dependency Scale. The CDS comprises all domains of nursing care; it is not limited to basic ADL, but also includes the individual’s capacity for social contacts, recreational activities, and learning abilities. The CDS is easy to administer, the responsible nurse could assess the CDS usually in less than five minutes and has shown satisfactory reliability and validity [22‐24].

Our study was nested in the CRANBERRY trial. Since the CRANBERRY study is a randomized-controlled trial and half of the participants underwent treatment with cranberry, this could have influenced the course of care dependency. However, there was no cranberry effect on care dependency over time. Therefore the CRANBERRY trial gives us the possibility to explore whether there are predictive factors of changes in care dependency. However, this means that not all earlier mentioned predictors of change in ADL were included in the dataset.

Within this study we were particularly interested to explore the personal characteristics of LTCF residents on the natural course of care dependency. An institutional effect on mortality and care dependency was not found. Other factors dependent on organizational characteristics of the long-term care facilities would be of interest for further research, since these characteristics could influence the care dependency status of LTCF residents as well. However, this was outside the scope of our study.

Another possible limitation of the present study is that we studied a selected period of 12 months. Classification of the participants into the three CDS groups was based on the prevalent CDS score at baseline, and we have no data on the CDS score of the residents at admission to the LTCF. Because care dependency is a dynamic process, the change in CDS score (and therefore the results) might be different if we had known the care dependency status when the residents were first admitted.

A regular and simple assessment of care dependency can be valuable, since this allows nursing staff to become more aware of the variability in the care dependency status of their residents, manage care, and provide better tailored care for individual residents. In daily nursing care, they are the first professionals who might observe subtle changes in the care dependency status of residents and therefore can better anticipate residents’ care needs. The present study shows that residents can increase or as well as decrease in their level of care dependency. Care dependency may be influenced by individually tailored interventions and this needs further exploration in research. In addition, in view of the association between the CDS score and mortality, it seems relevant to train staff in providing palliative care as well as restorative care [31].