Hide Sidebar

Delirium: a persistent problem; a persistent risk

Delirium: a persistent problem; a persistent risk

Critique date: Fri, 05/01/2009 - 10:01am
Institution: 
Mount Sinai School of Medicine
   
Article Citation
Citation: 
Kiely DK, Marcantonio ER, Inouye SK, Shaffer ML, Bergmann MA, Yang FM, Fearing MA, Jones RN. Persistent Delirium Predicts Greater Mortality. JAGS 2009; 57: 55-61.
   
Clinical Bottom Lines
  1. Persistent delirium after admission to a post-acute care (PAC) facility is associated with a one-year mortality of 39%; a patient admitted to PAC with persistent delirium is 2.9 times as likely to die within the next year as a patient who clears their delirium.
  2. Resolution of delirium is associated with a concomitant reduction in mortality, so it is always worth attempting to clear delirium.
  3. This relationship with mortality persists when controlling for age, sex, comorbidities, function, and dementia.
   
Methods
Type of Study: 
Prognosis
Study Design: 
Cohort study
Follow-up Period: 
One year
Patient Population: 
N = 412; Mean Age 84 (+/- 7.4)
Female 64.8%
White 91%
Married 26.1%
Home residence before admission 93.2%
Charlson Comorbidity Index 2.3 +/- 2.4 (higher means more comorbidity)
ADL score 13.3 +/- 3.3 (0-16, high = more independent)
Dementia 37.9%; measurements of cognitive impairment 
Specific primary diagnosis for hospital admission (percentages given)
Overall, an older, frailer population

Inclusions:
  • Delirious at time of admission
  • aged 65 or older
  • admitted directly from acute medical or surgical hospitalization
  • English speaking
  • communicative before acute illness
  • lived within 25 miles of research site
  • must have survived 14 days and have at least one follow-up
 
Significant Exclusions: 
  • no significant hearing impairment
  • not admitted for terminal care
  • no end-stage dementia
  • not completely dependent for activities of daily living
Intervention/Exposure: 
Subjects were drawn from a prior randomized trial of a Delirium Abatement Program – a unit based intervention that enrolled subjects who met criteria for delirium within 5 days of admission to a post-acute care facility after a medical or surgical admission. Delirium was determined through use of the MMSE (30), Digit Span (5 forward, 4 back), and Delirium Symptom Interview (DSI). These were combined to complete the Memorial Delirium Assessment Scale to determine severity, and the Confusion Assessment Method (CAM). Persistent delirium was present if delirium was present at all follow up visits, or if it resolved and reoccurred. Multiple covariates were identified. Subjects were assessed using the above method at baseline, 2, 4, 12 and 26 weeks. Potential covariates were identified. Outcome measure was mortality.
 
For the original study, subjects were assigned to DAP intervention/non intervention in a subgroup with the other subjects at a given facility (the target intervention group was the staff/caregivers). This was a negative study; however, the authors of this paper still adjusted for DAP intervention status in case any small effect could be present.
Outcome Measures: 
Mortality, determined by:
  1. National Death Index (NDI) data in all except n=10 (2%)
  2. Trained research nurses obtaining medical record information (mortality within 30 days of admission to PAC)
  3. Telephone interview with proxy if subject died after 30 days/transfer to another facility or homes
   
Conclusion
Results: 
Follow-up: 412 patients who were evaluated (though 457 originally identified, 45 did not have at least one follow up assessment. These were not analyzed in final data. Final end point was at one year.
Total one year cumulative mortaliy: 39%
Died between 2 and 4 weeks: 4%
Died between 4 and 12 weeks: 11%
Died between 12 and 26 weeks: 13%
Died between 26 and 52 weeks: 12%
 
In survivors, delirium presence gradually decreased over time. 32% still delirious at 6 months (26 weeks).

Analysis: Multivariable analyses run for comorbidity analyses. Adjustments made for dementia presence and for DAP intervention group showed no differences. Mortality determined as described above; for few patients for whom there was disagreement between the NDI measurements and others, a secondary analysis was run which showed no difference when those data were removed.
 
Results: Association between delirium and one year mortality examined using survival models – specifically Cox proportional Hazard Model. Multivariate analyses used to adjust data for the 14 covariates identified. Survival curves were generated for 5 delirium resolution patterns (never resolved, plus resolution noted at each of the 4 follow up assessments).
Analyses for dementia, comorbidity, and DAP group were planned prior to study.
 
1) First set of results: One year mortality rate of those subjects with persistent/recurrent delirium compared with one-year mortality rate of subjects without. 
 
Model Hazard Ratio (HR) 95% Confidence Interval
Unadjusted; n=412 3.0 2.0-4.5
Adjusted for age, sex, comorbidity, dementia, function; n=393 2.9 1.9-4.4
Adjusted for age, sex, comorbidity, dementia, function, and DAP intervention status; n = 393 2.9 1.9-4.4
 
2) Second results; The one year mortality of subjects with persistent/recurrent delirium compared with one year mortality of subjects without; considered separately for whether underlying dementia was shown to be present.
 
Model Dementia (n = 156) No dementia (n = 256)
Unadjusted 3.1 (1.5-6.4) 3.1 (1.9-5.0)
Adjusted for age, sex, comorbidity, function 2.6 (1.2-5.4) 3.0 (1.8-5.0)

Survival Curves show that the subjects who clear more quickly have better survival estimates. In other words, the longer and more persistent a delirium is, the less likely an individual will survive.
Concerns Regarding Methodology, Applicability to Older Adults, etc.: 
  1. This is a strong study examining the relationship between delirium and mortality over repeated time assessments, and is the first study to do so; it is clinically relevant for discharging patients or for following post-acute care patients.
  2. Provides prognostic information to consider for providers and for family members.
  3. Validity in this specific population is quite good; tools have higher inter rater reliability and sensitivity/specificity. Outcome measure has multiple ways it is measured and correlated.
  4. Generalizability: few men and very few non-whites were included in the study. Good for considering largely home-to-PAC, frail, elderly patients.
Funding Source and Role: 
National Institute on Aging grants; R01, R03, K grants to authors. No role stated.
Created By: 
Kate Callahan, first-year fellow, Department of Geriatrics, Mount Sinai School of Medicine
This is a review of the validity of a single study; the ‘bottom lines’ do not reflect comparison with the rest of the literature on this subject.