Abstract
Context
Nursing home (NH) residents with advanced dementia are commonly suspected of having infections. Most episodes are treated with antimicrobials, although evidence supporting bacterial infections is often lacking. The extent to which other interventions are employed in managing suspected infections is unknown.
Objectives
To describe interventions used to manage suspected infections in advanced dementia, and identify factors associated with greater intervention use.
Methods
Residents with advanced dementia who experienced suspected infections in 35 Boston NHs were followed for 12 months. Data describing interventions used in managing each episode were ascertained, including: blood draws, chest radiographs, procurement of urine samples, and hospital transfers. Resident and episode characteristics associated with greater intervention use were identified using mixed model regression.
Results
A total of 240 residents experienced 496 suspected infections involving the following interventions: any, N=360 (72.6%); hospital transfer, N=51 (10.3%); blood draw, N=215 (43.3%); chest radiograph, N=120 (24.2%); and urine sample, N=222 (44.8%). Factors associated with greater intervention use included: black race (adjusted odds ratio [AOR] 3.19, 95% confidence interval [CI], 1.37 – 7.44); no do not hospitalize order (AOR, 1.83, 95% CI, 1.16–2.90); not on hospice (AOR, 5.41, 95% CI, 2.14–13.70); and suspected source being respiratory (AOR, 10.67, 95% CI, 4.99 – 22.80), urine (AOR, 15.79, 95% CI, 7.41–33.66) or fever of unknown source (AOR, 20.26, 95% CI, 8.42 – 48.73) versus skin/soft tissue.
Conclusion
NH residents with advanced dementia frequently experience potentially burdensome interventions when suspected of having an infection. Advance directives to limit such interventions may be appropriate for residents whose goal of care is comfort.
Keywords: Nursing home, dementia, palliative care, infection
Introduction
In 2013, Alzheimer’s disease affected more than five million Americans, and was the sixth leading cause of death in the U.S. (1). Currently, the aggregate cost of long-term care for Americans with dementia totals $203 billion annually, with Medicare and Medicaid covering 70% of the total cost (2). Given our rapidly aging population and the economic and medical burden of dementia, determining how to best care for these patients has become a clinical and research priority (3).
Advanced dementia is characterized by severe cognitive impairment, inability to communicate verbally, inability to ambulate and dependence on others for all activities of daily living. Because more than 70% of such patients die in nursing homes (NHs) (4), current research has been directed at optimizing end-of-life care in these facilities. Although previous work has found that the majority of health care proxies for residents with advanced dementia prefer an approach based on comfort care, residents frequently receive interventions that can be considered burdensome (5).
Suspected infections are common among NH residents with advanced dementia (5,6), and there exists a growing body of literature addressing concerns regarding the approach to their management (7,8). The vast majority of suspected infections are treated with antimicrobials, although benefit in terms of either symptomatic relief or life-prolongation remains unclear in this population (8–10). Chief among the potential reasons for lack of benefit is the fact that minimal clinical criteria to support the presence of a bacterial infection are often not present for treated episodes (11,12). Antimicrobial misuse, whether a result of inaccurate diagnoses or misalignment with patient preferences, has sequelae with respect to both individual patient burden near the end of life and the emergence of multidrug-resistant organisms (6,8,13).
Prior research regarding suspected infections among NH residents with advanced dementia has largely focused on the impact of potential antimicrobial overuse (6,11,13). However, the use of burdensome interventions in the clinical assessment of suspected infections is less well described. These assessments may involve a variety of interventions that may be potentially uncomfortable or disturbing in these very frail residents, such as bladder catheterization or hospital transfers. Even seemingly innocuous procedures in more robust patients, such as a blood draw or obtaining a chest radiograph, can be traumatic for residents with advanced dementia. Given the fact that the majority of suspected infections do not meet clinical criteria for treatment, we may be subjecting residents to burdensome interventions unlikely to yield useful information or beneficial treatments.
The present study seeks to examine the burdensome medical interventions experienced by NH residents with advanced dementia using data collected during the Study of Pathogen Resistance and Exposure to Antimicrobials in Dementia (SPREAD), a prospective study of 362 NH residents with advanced dementia in 35 NHs throughout the Boston area. Our objectives were to describe and quantify the burdensome interventions as well as to identify factors associated with greater use of interventions in response to suspected infections in this cohort.
Methods
Data were collected during the National Institutes of Health-funded SPREAD study, the methodology of which is described in detail elsewhere (14). The health care proxies of residents with advanced dementia provided informed consent for their own participation, as well as for that of the residents. The Hebrew SeniorLife Institutional Review Board approved the study’s conduct.
All data were collected between September 2009 and November 2012 from 362 NH residents with advanced dementia and their health care proxies in 35 Boson-area NHs. Eligibility criteria required that the NH resident was 65 years or older, had dementia with a score of 7 on the Global Deterioration Scale (GDS) (15), had an available English-speaking health care proxy, and had resided in the NH for at least 30 days. A GDS score of 7 is characterized by severe memory deficits, verbal ability of less than five words, incontinence and inability to walk independently (15).
Resident and Proxy Variables
All residents enrolled in SPREAD were followed for 12 months or until death. Data were collected during baseline and quarterly assessments as well as monthly chart reviews. During the baseline resident assessment, data on demographics (age, gender, race) and common comorbidities (diabetes, congestive heart failure, chronic obstructive pulmonary disease) were extracted from the medical record. A nurse who cared for the resident quantified their functional ability using the Bedford Alzheimer’s Nursing Severity Subscale (BANS-S) (16) (range 7–28; higher scores indicate greater disability). Cognitive status was assessed by direct examination of the resident at baseline by a member of the research team using the Test for Severe Impairment (TSI) (17) (range 0 – 24; lower scores indicate greater impairment). At baseline and quarterly assessments, data on the presence or absence of a “do not hospitalize” (DNH) order and hospice enrollment were collected from the chart. Whether the resident was cared for on a dementia special care unit was determined at baseline. The residents’ health care proxies were interviewed by telephone at baseline by a trained research assistant. Proxies were asked the frequency with which they visited the NH, and their perception of the resident’s primary goal of care (life prolongation vs. comfort).
Infectious Episodes
Each month, trained research nurses reviewed the resident’s medical record to determine whether a suspected infectious episode had occurred during the previous month, as indicated by any of the following: 1) documentation of a suspected infection by a physician, physician’s assistant, or nurse; 2) antimicrobial use (non-prophylactic); or 3) temperature above 37.9 degrees Celsius. The source of the suspected infection was categorized as one of the following: respiratory tract, urinary tract, skin/soft tissue, or fever of unknown origin based on chart documentation.
For each suspected infection, the following data were collected: 1) documentation of a clinical examination by a nurse, physician or physician’s assistant within 72 hours of the episode; 2) occurrence on a weekend (vs. weekday); 3) occurrence within 30 days of resident death; 4) antimicrobial use; 5) documentation of a hospital transfer (admission or emergency room visit); 6) whether a blood draw was done for any reason (i.e., complete blood count, electrolytes, renal function, culture); 7) whether a chest radiograph was obtained; and 8) procurement of a urine sample. Based on chart documentation of signs and symptom, each suspected infectious episode was classified as either meeting or not meeting the minimal clinical criteria for initiation of antimicrobials based on the 2001 Society for Healthcare Epidemiology of America consensus guidelines (12). These criteria were adapted slightly for urinary tract infections as advanced dementia patients cannot reliably express many of the classic symptoms associated with these infections (14, 18).
Burdensome Interventions
In the analyses, hospital transfers, blood draws, chest radiographs, and urine samples were classified as burdensome interventions. Our eligibility criteria required the residents were incontinent of urine, therefore obtaining a urine specimen was considered burdensome because this procedure likely involved bladder catheterization. Chest radiographs may be viewed as a relatively simple, non-invasive test in robust patients. However, we considered them as potentially burdensome in these bedbound and profoundly cognitively impaired residents, as obtaining a chest radiograph in a NH setting sometimes requires a transfer to a radiology facility, or at minimum for a bedside test, entails repositioning and patient handling.
Statistical Analyses
Analyses were conducted at two levels, both the resident level and the episode level. To characterize the burden of medical interventions at the resident level, we calculated the proportion of residents who experienced 0, 1, 2, 3 or ≥ 4 interventions in response to a suspected infection over the entire study period. Given that the residents’ observation time varied, (i.e., some residents died during the study), we also calculated the average rate of interventions per person-year. Rates were calculated for all interventions as well as each specific intervention. At the episode level, we calculated the proportion of suspected infections for which any and each burdensome interventions occurred. Proportions were calculated for all suspected infections as well as stratified by suspected source of infection.
To identify factors associated with greater use of burdensome interventions at the episode level, we constructed a “burden index” score (outcome) for each suspected episode. The “burden index” was calculated by creating a sum of the interventions, with hospital transfers assigned a value of 2, and all the other interventions assigned a value of 1, because of the relative greater burden of a hospital transfer compared to the other interventions (19). Thus, the index had a possible range of 0 (no interventions) to 5 (all interventions including hospital transfer). Because the burden associated with a hospitalization depends on several factors such as length of stay and the treatments and tests performed during the admission, we chose a relatively conservative approach by weighting hospitalizations as twice as burdensome as other interventions. While this is likely an underestimation, a conservative approach would only bias our results toward the null. Based on its distribution, the burden index for each episode was analyzed as an ordinal outcome categorized as 0, 1, 2 and > 3 interventions per episode.
In the analyses examining factors associated with a higher burden index (outcome) independent variables included resident and episode characteristics. Resident characteristics were: demographics such as age (dichotomized at the median), gender and race (white vs. black), comorbidities, BANS-S score (dichotomized at the median), TSI score (dichotomized as > 0 vs. 0), the absence of a DNH order, lack of hospice enrollment, special care unit, proxy visit frequency (dichotomized at <8 hours per week vs. greater) and whether the stated goal of care was comfort. Episode features included: the suspected source, clinical examination within 72 hours, weekend occurrence, occurrence within 30 days of death, and whether clinical criteria for antimicrobial treatment were present. Most factors were static and carried forward from the baseline assessment (i.e., age, gender), but dynamic variables (BANS-S score, DNH orders, hospice status and proxy variables) were chosen from the quarterly assessments reflecting the resident’s status at the time of the episode. Skin/soft tissue source was the referent group for the independent variable describing the source of the suspected infection.
Multinomial logistic regression was used to examine the bivariate association between each independent variable and the burden index (outcome), adjusting for clustering at the facility and resident level using mixed models. The proportional odds assumption for each model was tested and met. Therefore, one odds ratio (OR) is presented for the multinomial outcome. Independent variables associated with the burden index (outcome) at P < 0.10 in bivariate analyses were entered into an adjusted multinomial logistic regression model, where the proportional odds assumption was tested and met. Again, mixed model analysis accounted for the random effects of clustering at the facility and resident levels (because of repeated measurements). ORs with 95% confident intervals (CIs) were computed. To ensure that the burden index structure did not bias our results, we also ran sensitivity analyses where we weighted hospital transfer as one point.
To determine whether the lack of a DNH order was the driving force behind the association between DNH and our burden index given that hospital transfers were more highly weighted, we also examined the bivariate associations between the lack of a DNH order (independent variable) and the use of each type of intervention: hospital transfer, blood draw, chest radiograph, and urine sample. These bivariate associations also were adjusted for clustering at the individual and facility levels. All analyses were conducted using STATA, versions 12.0 and 13.1 (StataCorp LP, College Station, TX).
Results
Resident and Proxy Characteristics
Among 951 eligible NH residents with advanced dementia, 362 (38.1%) were recruited into the SPREAD study. Non-participation was a result of refusal by proxies (n = 587) or physicians (n = 2). Non-participating and participating eligible NH residents did not differ by age, sex or race.
Of the 362 enrolled residents, the present analyses are limited to the 240 residents (66.3%) who experienced at least one suspected infection during the study follow-up period. There were no statistically significant differences for resident demographic characteristics (age, gender, race) between participants and non-participants Additionally, there were no statistically significant differences in demographic characteristics, BANS-S or TSI scores between participating residents with and without suspected infections. A total of 104 of the residents with suspected infections (43.3%) died during the study and three residents were lost to follow-up. The median resident follow-up time was 280 days (range 2–384 days).
Baseline resident characteristics are presented in Table 1. The median age was 86 years, 16.3% were male, and 93.3% were white. The presence of comorbid conditions was: diabetes (21.7%), congestive heart failure (16.3%), and chronic obstructive pulmonary disease (13.3%). The median BANS-S score was 22 and 40.4% had a TSI score greater than 0. At baseline, 59.6% of residents did not have a DNH order and 90.8% were not enrolled in hospice. Among proxies, 88.3% visited fewer than eight hours per week, and 95.0% reported that comfort was the residents’ primary goal of care.
Table 1.
Baseline Characteristics of NH Residents with Advanced Dementia Who Experienced at Least One Suspected Infection1
| Residents | N=240 |
|---|---|
| Age ≤ 86 (median), n (%), range | 107 (44.6%), 68–105 |
| Male | 39 (16.3%) |
| White (vs. Black) | 224 (93.3%) |
| Comorbid Conditions | |
| Diabetes | 52 (21.7%) |
| Congestive heart failure | 39 (16.3%) |
| Chronic obstructive pulmonary disease | 32 (13.3%) |
| BANS-S Scorea ≥ 22, (median), n (%), range | 123 (51.3) 14–27 |
| Test of Severe Impairmentb > 0 | 97 (40.4%) |
| Lack of Do Not Hospitalize Order | 143 (59.6%) |
| Not on Hospice care | 218 (90.8%) |
| Proxy visits resident < 8 hours/week | 212 (88.3%) |
| Proxy states comfort is primary goal of care c | 227 (95.0%) |
Bedford Alzheimer’s Nursing Severity Scale, range 7–28, higher scores represent more severe functional impairment.
Range 0–24, lower scores represent greater impairment. Scores dichotomized at 0 vs. > 0.
N = 239 because of missing data.
Characteristics of Suspected Infections
The mean number of suspected infections among the 240 residents was 2.1, ± 1.4 SD (range, 1–9). The cohort experienced a total of 496 episodes, for which the suspected sources were distributed as follows: respiratory tract, 29.8%; urinary tract, 39.5%; skin/soft tissue, 13.9%; and fever of unknown origin, 16.7%. Only 198 of 496 (39.9%) suspected infections met clinical criteria for initial antimicrobial treatment.
Intervention Use at the Resident Level
Among the 240 residents, the mean number of interventions per resident during the entire observation period (i.e., up to 12 months or death) was 2.5 ± 2.6 SD (range 0 to 14 interventions). The number of interventions was distributed among these residents as follows: none, n=47 (19.6%); one n=49 (20.4%); two, n=55 (22.9%); three, n=34 (14.2%), and ≥ four, n=55 (22.9%). The mean rate of intervention use among the residents was 3.69 interventions per person-year (Table 2). The mean rates of specific intervention use in person-years were: 0.36 hospital transfers, 1.35 blood draws, 0.73 chest radiographs, and 1.24 urine samples.
Table 2.
Rates of Intervention Associated with Infection Work-Up per Person-Year among 496 Suspected Infections Occurring in 240 NH Residents with Advanced Dementia
| Intervention Rate (per person-year) | ||
|---|---|---|
| Intervention | Rate (SD) | Range |
| Hospital transfers | 0.36 (1.27) | (0, 11.52) |
| Blood draws | 1.35 (2.14) | (0, 14.55) |
| Chest radiographs | 0.73 (1.34) | (0, 11.25) |
| Urine samples | 1.24 (1.95) | (0, 13.14) |
| Total | 3.69 (5.22) | (0, 45) |
Intervention Use at the Episode Level
Among the 496 suspected infections experienced by the residents in our analyses (N=240), at least one intervention was used in 72.6% of episodes (n=360) (Table 3). Intervention use was highest for suspected urinary tract infections (95.9%), followed by respiratory infections (67.6%), fevers of unknown source (66.3%) and skin/soft tissue infections (24.6%). Overall, 10.3% of suspected infections resulted in hospital transfers, 43.3% in blood draws, 24.2% in chest radiographs, and 44.8% in urine samples.
Table 3.
Interventions per Suspected Infectious Episode among 496 Episodes Occurring in 240 NH Residents with Advanced Dementia
| Intervention, No. (%) | ||||||
|---|---|---|---|---|---|---|
| Suspected source of infection | N | Hospital transfer | Blood draws | Chest radiographs | Urine sample | At least one intervention |
| Respiratory | 148 | 16 (10.8) | 69 (46.6) | 87 (58.8) | - | 100 (67.6) |
| Urinary Tract | 196 | 17 (8.7) | 85 (43.4) | - | 184 (93.9) | 188 (95.9) |
| Skin/Soft Tissue | 69 | 4 (5.8) | 16 (23.2) | - | - | 17 (24.6) |
| Fever (unknown source) | 83 | 14 (17.3) | 45 (54.2) | 33 (39.8) | 38 (45.8) | 55 (66.3) |
| Total | 496 | 51 (10.3) | 215 (43.3) | 120 (24.2) | 222 (44.8) | 360 (72.6) |
Factors Associated with Greater Use of Interventions
Using the burden index analyzed as an ordinal outcome categorized as 0, 1, 2, or ≥ 3, the following covariates were found to be associated with higher burden in unadjusted analyses (P < 0.10): younger resident age, black race, TSI score > 0, lack of a DNH order, not enrolled in hospice, care goal was not comfort, weekend occurrence, and suspected source not being skin/soft tissue (Table 4). These variables were included in a multivariable model with the same outcome. In this model, the following variables remained independently associated with greater burden (P <0.05): black race (adjusted OR [AOR] 3.19, 95% CI, 1.37 – 7.44); lack of DNH order (AOR, 1.83, 95% CI, 1.16–2.90); not enrolled in hospice (AOR, 5.41, 95% CI, 2.14–13.70); and suspected source being respiratory tract (AOR, 10.67, 95% CI, 4.99 – 22.80), urinary tract (AOR, 15.79, 95% CI, 7.41–33.66) or fever of unknown source (AOR, 20.26, 95% CI, 8.42 – 48.73) versus skin/soft tissue. The proportional odds assumption held for the burden index outcome in the adjusted model. Sensitivity analysis revealed the same pattern of results when using the burden index outcome with equally weighted element (hospitalizations given a value of 1, rather than 2).
Table 4.
Association of Medical Interventions Burdena with Resident, Proxy and Episode Factors among 496 Suspected Infections Occurring in 240 NH Residents with Advanced Dementia
| Characteristicb | Total (N=496) n (%) |
Unadjusted OR (95% CI) |
Adjusted OR (95% CI) |
|---|---|---|---|
| Resident | |||
| Age ≤ 86 (median) | 228 (46) | 1.69 (1.09 – 2.63) | 1.43 (0.93 – 2.19) |
| Male | 83 (16.7) | 1.50 (0.84 – 2.68) | -- |
| Black race | 40 (8.1) | 4.14 (1.74 – 9.90) f | 3.19 (1.37 – 7.44) g |
| Diabetes | 124 (25.0) | 1.05 (0.63 – 1.77) | -- |
| Congestive heart failure | 90 (18.1) | 1.12 (0.63 – 1.99) | -- |
| Chronic obstructive pulmonary disease | 76 (15.3) | 0.86 (0.46 – 1.60) | -- |
| BANS-S Score c > 22 (median) | 264 (53.2) | 0.76 (0.49 – 1.19) | -- |
| Test for Severe Impairment4 > 0 | 200 (40.3) | 1.47 (0.94 – 2.29) f | 1.31 (0.85 – 2.01) |
| Lack of Do Not Hospitalize Order | 347 (70.0) | 2.98 (1.90 – 4.66) f | 1.83 (1.16 – 2.90) g |
| Not enrolled in Hospice | 460 (92.7) | 6.26 (2.56 – 15.33) f | 5.41 (2.14 – 13.70) g |
| Special Care Unit | 210 (42.3) | 0.90 (0.57 – 1.43) | -- |
| Proxy | |||
| Visit frequency < 8 hours/week | 433 (87.3) | 1.38 (0.72 – 2.67) | -- |
| Resident care goal is comforte | 460 (93.5) | 0.42 (0.16 – 1.07) f | 0.63 (0.26 – 1.55) |
| Infection Episode | |||
| Weekend occurrence | 99 (20.0) | 1.48 (0.94 – 2.33) f | 1.35 (0.85 – 2.14) |
| Clinical exam within prior 72 hours | 281 (56.7) | 1.19 (0.80 – 1.76) | -- |
| More than 30 days from death | 434 (87.5) | 1.37 (0.77 – 2.44) | -- |
| Meets clinical criteria for treatment | 198 (39.9) | 0.73 (0.50 – 1.08) | -- |
| Source | |||
| Skin/soft tissue | 69 (13.9) | 1 [referent] | 1 [referent] |
| Respiratory tract | 148 (29.8) | 12.55 (6.00 – 26.25) f | 10.67 (4.99 – 22.80) g |
| Urinary tract | 196 (39.5) | 22.22 (10.69 – 46.19) f | 15.79 (7.41 – 33.66) g |
| Fever of unknown source | 83 (16.7) | 25.60 (10.99 – 59.61) f | 20.26 (8.42 – 48.73) g |
Measured by the “burden index.” The “burden index” is a numerical representation of the total number of medical interventions per infectious episode. Blood draws, urine samples and chest radiographs are counted as “1” each and hospitalizations as “2” each.
All variables were assessed at baseline except for BANS-S, do not hospitalize orders, hospice status and proxies’ variables, which correspond to the time of the suspected infection.
Bedford Alzheimer’s Nursing Severity Scale. Range 7–28; higher scores represent more severe functional impairment.
Range 0–24, lower scores represent greater impairment. Scores dichotomized at 0 vs. > 0.
N = 492 because of missing data.
P < 0.10.
P < 0.05.
Do Not Hospitalize Orders
We wanted to further investigate whether DNH orders were solely associated with the hospital transfer component of the burden index. Therefore, we performed separate bivariate analyses of the association between the lack of a DNH order and the risk of each type of burdensome intervention, using mixed models to account for grouping at the facility and resident levels. In addition to hospital transfers, the lack of a DNH order was independently associated (P < 0.05) with higher rates of blood draws and chest radiographs during the assessment of infections.
Discussion
In this prospective study of NH residents with advanced dementia, nearly three-quarters of the suspected infections resulted in a hospital transfer, blood draw, chest radiograph or procurement of a urine sample. On average, residents with suspected infections experienced 3.7 such interventions per year. The use of these interventions was associated with the lack of a do not hospitalize order, as well as lack of hospice enrollment, indicating that prior documentation of advance directives aligned with a palliative approach results in lower likelihood of burden during the management of these episodes. Given that the majority of suspected infections in our cohort did not meet criteria for antimicrobial treatment, and that comfort was the goal of care for the vast majority of residents, our results indicate that there may be an overuse of diagnostic interventions (including hospital transfers) associated with suspected infections. Thus, our findings demonstrate that the potential burden associated with the management of suspected infections in residents with advanced dementia goes beyond the misuse of antimicrobials (6,8,20) and extends to other interventions used to treat and work-up these episodes.
Among residents who experienced a suspected infection, a roughly equal proportion received 0, 1, 2, 3 or ≥ 4 interventions, with individual person-year rates being highest for blood draws and urine samples. Given that the urinary tract was the most common suspected source of infection, and that these suspected infections had the highest proportion of interventions (95.9%), efforts to limit the work-up of urinary tract episodes is likely to have the highest yield in lessening the overall intervention burden. In addition, prior reports from this cohort have documented that suspected urinary tract infections are the most likely to be mistreated with antimicrobials (11) and, therefore, offer the best opportunity to reduce both inappropriate treatment and intervention burden in this population.
We found that among the resident factors associated with an increased intervention burden was black race, despite the relatively low representation of blacks in this cohort. This finding aligns with previous research demonstrating a trend toward more aggressive end-of-life care among blacks as compared to whites across a variety of medical diagnoses. (21–23) The reasons for this association remain unclear, in particular, whether they represent differences in counseling or true differences in patient or family preferences. Thus, ensuring that all patients and families are given adequate education about the disease course is an important precursor to establishment of advance directives and approaches to care.
In this study, both hospice status and DNH orders appeared to translate to less aggressive care for suspected infections. Because the core definition of hospice is a shift from life-prolonging to comfort care, this association is appropriate and demonstrates the utility of hospice as a mechanism for guiding care for acute illness. Similarly, although a DNH order only specifically precludes hospital transfers, in our study it was also associated with lower rates of blood draws and chest radiographs. Thus, the DNH order seems to be a marker for less aggressive care overall. Whether this was the intention of the DNH order remains unclear, and dialogue between proxies and providers regarding the intended scope of care should ideally be documented with the use of more specific advance care directives such as “no blood draws” or “no antibiotics” in order to clarify intentions.
Consistent with prior research in similar populations (5), in this study an overwhelming majority of health care proxies (95%) stated that the resident’s treatment goal was comfort care. However, in our adjusted analysis, this stated care goal did not have a significant effect on the outcome of burden. This may indicate a lack of communication or follow through between proxies and health care providers regarding how to translate an overall goal of care into specific directives for the management of acute illness. Such a barrier to communication regarding care goals has been demonstrated in previous work (24,25) and argues for continued efforts to engage proxies and providers in ongoing conversations regarding specific treatments in addition to overall goals.
This study has certain limitations that deserve acknowledgement. Because SPREAD was conducted in the Boston area, its generalizability is uncertain. However, the characteristics of the facilities and residents reflect those of NHs nationwide (14). Much of the data was collected from the medical record, and may be subject to information bias because of inaccurate documentation. However, we suspect that under-documentation rather than over-documentation of interventions is more likely, which would bias our results towards under-reporting the actual burden of interventions. Inadequate power may account for the lack of significance between some of the covariates in our model and the burden index. Our recruitment rate was 38%, although this is similar to other studies involving NH residents (26). Lastly, we are making an assumption that the interventions explored in this study are burdensome to residents, based on prior research (19). However, it is difficult to discern whether residents with advanced dementia find particular interventions painful, and although measures exist to assess discomfort in advanced dementia based on caregiver observation (27), we did not directly employ them in this study.
Prior to this study, previous work has suggested that the burden associated with the assessment of infections among end-stage dementia patients may be high, but none had directly examined this issue. Our results demonstrate the need for care goals to guide treatment choices prior to the potential initiation of antimicrobial therapy during the work-up itself. Certainly, providers should be aware of the general overuse of interventions and antimicrobials in this population, but proxies and families themselves can take some proactive steps to ensure that treatment decisions align with their goals of care by using specific advance directives.
Discussions between providers and proxies regarding the potential burden associated with work-up of infections could help to decrease the overall likelihood of patients experiencing these interventions, especially if wishes are documented as specific advance directives. Several states now have advance directive forms including Physician Orders for Life Sustaining Treatment (POLST) or similar documents that could be used for this purpose. A useful next research step would employ qualitative interviews with proxies and providers to provide a more in-depth understanding of how they perceive the degree of burden associated the interventions used to assess suspected infections among patients with advanced dementia. Ultimately, a more judicious approach to infection management on the part of the provider and better counseling and advance care planning between the provider and the proxy may result in end-of-life care for these frail patients that better aligns with their goals.
Acknowledgments
This work was supported by funding from grants R01AG032982, K24AG033640 (SLM), 1 T35 AG038027-04 MSTAR (EY), and K23AG034967 (JLG) from the National Institute on Aging.
Footnotes
Disclosures
The authors declare no conflicts of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Hebert LE, Weuve J, Scherr PA, Evans DA. Alzheimer disease in the United States (2010–2050) estimated using the 2010 census. Neurology. 2013;80:1778–1783. doi: 10.1212/WNL.0b013e31828726f5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Thies W, Bleiler L Alzheimer’s Association. 2013 Alzheimer’s disease facts and figures. Alzheimers Dement. 2013;9:208–245. doi: 10.1016/j.jalz.2013.02.003. [DOI] [PubMed] [Google Scholar]
- 3.Mitchell SL, Black BS, Ersek M, et al. Advanced dementia: state of the art and priorities for the next decade. Ann Intern Med. 2012;156:45–51. doi: 10.1059/0003-4819-156-1-201201030-00008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Mitchell SL, Teno JM, Miller SC, Mor V. A national study of the location of death for older persons with dementia. J Am Geriatr Soc. 2005;53:299–305. doi: 10.1111/j.1532-5415.2005.53118.x. [DOI] [PubMed] [Google Scholar]
- 5.Mitchell SL, Teno JM, Kiely DK, et al. The clinical course of advanced dementia. N Engl J Med. 2009;361:1529–1538. doi: 10.1056/NEJMoa0902234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.D’Agata E, Mitchell SL. Patterns of antimicrobial use among nursing home residents with advanced dementia. Arch Intern Med. 2008;168:357–362. doi: 10.1001/archinternmed.2007.104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Catic AG, Berg AI, Moran JA, et al. Preliminary data from an advanced dementia consult service: integrating research, education, and clinical expertise. J Am Geriatr Soc. 2013;61:2008–2012. doi: 10.1111/jgs.12530. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Givens JL, Jones RN, Shaffer ML, Kiely DK, Mitchell SL. Survival and comfort after treatment of pneumonia in advanced dementia. Arch Intern Med. 2010;170:1102–1107. doi: 10.1001/archinternmed.2010.181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Goldfeld KS, Hamel MB, Mitchell SL. The cost-effectiveness of the decision to hospitalize nursing home residents with advanced dementia. J Pain Symptom Manage. 2013;46:640–651. doi: 10.1016/j.jpainsymman.2012.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.van der Steen JT, Mehr DR, Kruse RL, Ribbe MW, van der Wal G. Treatment strategy and risk of functional decline and mortality after nursing-home acquired lower respiratory tract infection: two prospective studies in residents with dementia. Int J Geriatr Psychiatry. 2007;22:1013–1019. doi: 10.1002/gps.1782. [DOI] [PubMed] [Google Scholar]
- 11.Mitchell SL, Shaffer ML, Loeb MB, et al. Infection management and multidrug-resistant organisms in nursing home residents with advanced dementia. JAMA Intern Med. 2014;174:1660–1667. doi: 10.1001/jamainternmed.2014.3918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Loeb M, Bentley DW, Bradley S, et al. Development of minimum criteria for the initiation of antibiotics in residents of long-term-care facilities: results of a consensus conference. Infect Control Hosp Epidemiol. 2001;22:120–124. doi: 10.1086/501875. [DOI] [PubMed] [Google Scholar]
- 13.van der Steen JT, Ooms ME, van der Wal G, Ribbe MW. Withholding or starting antibiotic treatment in patients with dementia and pneumonia: prediction of mortality with physicians’ judgment of illness severity and with specific prognostic models. Med Decis Making. 2005;25:210–221. doi: 10.1177/0272989X05275400. [DOI] [PubMed] [Google Scholar]
- 14.Mitchell SL, Shaffer ML, Kiely DK, Givens JL, D’Agata E. The study of pathogen resistance and antimicrobial use in dementia: study design and methodology. Arch Gerontol Geriatr. 2013;56:16–22. doi: 10.1016/j.archger.2012.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Reisberg B, Ferris SH, de Leon MJ, Crook T. The Global Deterioration Scale for assessment of primary degenerative dementia. Am J Psychiatry. 1982;139:1136–1139. doi: 10.1176/ajp.139.9.1136. [DOI] [PubMed] [Google Scholar]
- 16.Volicer L, Hurley AC, Lathi DC, Kowall NW. Measurement of severity in advanced Alzheimer’s disease. J Gerontol. 1994;49:M223–226. doi: 10.1093/geronj/49.5.m223. [DOI] [PubMed] [Google Scholar]
- 17.Albert M, Cohen C. The Test for Severe Impairment: an instrument for the assessment of patients with severe cognitive dysfunction. J Am Geriatr Soc. 1992;40:449–453. doi: 10.1111/j.1532-5415.1992.tb02009.x. [DOI] [PubMed] [Google Scholar]
- 18.D’Agata E, Loeb MB, Mitchell SL. Challenges in assessing nursing home residents with advanced dementia for suspected urinary tract infections. J Am Geriatr Soc. 2013;61:62–66. doi: 10.1111/jgs.12070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Morrison RS, Ahronheim JC, Morrison GR, et al. Pain and discomfort associated with common hospital procedures and experiences. J Pain Symptom Manage. 1998;15:91–101. [PubMed] [Google Scholar]
- 20.Van Der Steen JT, Pasman HR, Ribbe MW, Van Der Wal G, Onwuteaka-Philipsen BD. Discomfort in dementia patients dying from pneumonia and its relief by antibiotics. Scand J Infect Dis. 2009;41:143–151. doi: 10.1080/00365540802616726. [DOI] [PubMed] [Google Scholar]
- 21.Frahm KA, Brown LM, Hyer K. Racial disparities in end-of-life planning and services for deceased nursing home residents. J Am Med Dir Assoc. 2012;13:819.e817–811. doi: 10.1016/j.jamda.2012.07.021. [DOI] [PubMed] [Google Scholar]
- 22.Hopp FP, Duffy SA. Racial variations in end-of-life care. J Am Geriatr Soc. 2000;48:658–663. doi: 10.1111/j.1532-5415.2000.tb04724.x. [DOI] [PubMed] [Google Scholar]
- 23.Kwak J, Haley WE. Current research findings on end-of-life decision making among racially or ethnically diverse groups. Gerontologist. 2005;45:634–641. doi: 10.1093/geront/45.5.634. [DOI] [PubMed] [Google Scholar]
- 24.Mitchell SL, Kiely DK, Hamel MB. Dying with advanced dementia in the nursing home. Arch Intern Med. 2004;164:321–326. doi: 10.1001/archinte.164.3.321. [DOI] [PubMed] [Google Scholar]
- 25.Givens JL, Selby K, Goldfeld KS, Mitchell SL. Hospital transfers of nursing home residents with advanced dementia. J Am Geriatr Soc. 2012;60:905–909. doi: 10.1111/j.1532-5415.2012.03919.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Fisch J, Lansing B, Wang L, et al. New acquisition of antibiotic-resistant organisms in skilled nursing facilities. J Clin Microbiol. 2012;50:1698–1703. doi: 10.1128/JCM.06469-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Warden V, Hurley AC, Volicer L. Development and psychometric evaluation of the Pain Assessment in Advanced Dementia (PAINAD) scale. J Am Med Dir Assoc. 2003;4:9–15. doi: 10.1097/01.JAM.0000043422.31640.F7. [DOI] [PubMed] [Google Scholar]