Original Research

Impact of an inpatient geriatric consultative service on outcomes for cognitively impaired patients



Impact of geriatric consultative services (GCS) on hospital readmission and mortality outcomes for cognitively impaired (CI) patients is not known.


Evaluate impact of GCS on hospital readmission and mortality among CI inpatients.


Secondary data analysis of a prospective trial of a computerized decision support system between July 1, 2006 and May 30, 2008.


Study conducted at Eskenazi hospital, Indianapolis, Indiana, a 340‐bed, public hospital with over 2300 yearly admissions of patients ages 65 years or older.


There were 415 inpatients aged 65 years and older with CI enrolled from July 2006 to March 2008.


Thirty‐day and 1‐year mortality and hospital readmission following the index admission. Cox proportional hazard models were used to determine the association between receiving GCS, readmission, or mortality while adjusting for demographics, discharge destination, delirium, Charlson Comorbidity Index, and prior hospitalizations. The propensity score method was used to adjust for the nonrandom assignment of GCS.


Patients receiving GCS were older (79 years old, 8.1 standard deviation [SD] vs 76 years old, 7.8 SD; P < 0.001) with higher incidence of delirium (49% vs 29%; P < 0.001). No significant differences were found between the groups for hospital readmission (hazard ratio [HR] = 1.19; 95% confidence interval = 0.89‐1.59) and mortality at 12 months of index admission (HR = 0.91; 95% confidence interval = 0.59‐1.40). However, a significant increase in readmissions was observed for the GCS group (HR = 1.75; 95% confidence interval = 1.06‐2.88) at 30 days postdischarge.


One‐year postdischarge outcomes of CI patients who received GCS were not different from patients who did not receive the service. New models of care are needed to improve postdischarge readmission and mortality among hospitalized patients with CI. Journal of Hospital Medicine 2015;10:275–280. © 2015 Society of Hospital Medicine

© 2015 Society of Hospital Medicine

Under the Patient Protection and Affordable Care Act of 2010, commonly referred to as the Affordable Care Act, hospitals face up to a 3% penalty in Medicare reimbursements for patients readmitted within 30 days of initial discharge, and measures have been proposed for modifying payments to hospitals based on their performance on this metric.[1] Cognitive impairment (CI) is considered a major risk factor for poor postdischarge outcomes including mortality and hospital readmission.[2, 3] Hospitals are seeking strategies to reduce postdischarge mortality and rehospitalization among patients with and without CI.[4] Such strategies include use of transitional care coaches, patient and caregiver education, postdischarge follow‐up, and provision of geriatric consultative services (GCS) for the care of complex patients in the hospital setting.[5, 6, 7]

GCS utilize comprehensive geriatric assessments and multidisciplinary processes to recognize and modify risk factors that may lead to poor outcomes among hospitalized patients.[8, 9, 10, 11] Implementation of GCS models including Acute Care for Elders and, recently, the Mobile Acute Care of the Elderly services have shown many benefits among older patients including a reduction in the hospital length of stay and readmission rates.[12, 13] The benefits of such services among hospitalized elders suffering from CI, however, are not well established. The objective of this article was to evaluate the impact of GCS on the readmission and mortality rates of older adults with CI within 12 months of their hospitalization to an urban, public hospital. We hypothesized that GCS will reduce both 12‐month hospital readmissions and mortality rates among this vulnerable group of older adults.


The study was approved by the Indiana University institutional review board, and informed consent for identifiable chart review was obtained from subjects or their legally authorized representatives.


The study was conducted at Eskenazi hospital, Indianapolis, Indiana, a 340‐bed, university‐affiliated, public hospital with over 2300 admissions of patients aged 65 years or older every year.


Four hundred fifteen hospitalized patients aged 65 years or older suffering from CI were enrolled into an original, randomized, controlled trial that evaluated the effect of a computerized decision support system on their quality and outcome of care between July 1, 2006 and May 30, 2008.[14] The computerized decision support included reminders for physicians to reduce the prescription of 18 anticholinergics, minimize physical restraints and Foley catheterization, and increase referral to the local GCS.[15] That previous trial neither showed an impact on quality of care nor health utilization among older patients, including mortality and hospital readmission rates. The current study uses the data from the clinical trial cohort to evaluate the effect of GCS on the 12‐month mortality and hospital readmission rates for hospitalized elders with CI (Figure 1).

Figure 1

Flowchart depicting patients who received intervention.

Inclusion and Exclusion Criteria

Individuals were eligible for enrollment if they were aged 65 years or older, hospitalized on a medical ward, able to speak English, and had evidence of CI within 48 hours of hospital admission. Individuals were excluded if they were previously enrolled, were aphasic, or unresponsive. The presence of CI was based on the Short Portable Mental Status Questionnaire (SPMSQ),[16] a brief 10‐item screening test with a sensitivity of 86% and specificity of 99% for dementia using a score of 7 or less (maximum possible score of 10).[16] The SPMSQ scoring process adjusts for participant educational and racial status, which was a benefit to its use given the urban setting of our hospital serving a large proportion of minority and low‐education patients. A physician‐trained research assistant administered the SPMSQ within 48 hours of hospital admission.

Geriatric Consultative Services

GCS is an interdisciplinary team of a geriatrician, a geriatric pharmacist, a case manager nurse, a social worker, a medical assistant, physical therapists, and a representative of the local Area Agency on Aging. There may be a geriatric fellow and/or medicine resident available to the team based on their rotation structure. Team‐based bedside rounds are performed on new consults only, but all patients are seen individually by the team clinicians. The team emphasizes prevention of functional decline and polypharmacy, recognition and treatment of geriatric syndromes including dementia and delirium, and early discharge/transition planning. Consensus recommendations are prepared and documented in the consult notes section of the electronic medical records. Recommendations deemed critical are discussed directly with the primary teams, but no orders are placed by the GCS team. The GCS team is available on all weekdays but not on weekends or major holidays.

Study Outcomes

For this secondary analysis, we used the Regenstrief Medical Record System (RMRS) to measure 2 outcomes: hospital readmission and mortality rates up to 1 year from discharge following index hospitalization, defined as the first admission in the original clinical trial. The RMRS is the primary instrument for processing data and monitoring patient and physician activity for the hospital.[17, 18] The RMRS is linked with a state‐wide health information exchange to capture data on hospitalization outside the hospital. The RMRS also contains death certificate information for all registered patients who die in or outside the Eskenazi hospital.

Other Data Collections

Delirium was assessed at screening and then every weekday using the Confusion Assessment Method (CAM) by a trained research assistant.[19] CAM evaluates 10 symptoms of delirium specified in the Diagnostic and Statistical Manual of Mental Disorders‐III‐Revision: acute onset, fluctuating course, inattention, disorganized thinking, altered level of consciousness, disorientation, memory impairment, perceptual disturbances, psychomotor agitation or retardation, and sleep/wake disturbance. Participant demographic characteristics, including age, sex, ethnicity, and years of education, were collected from the RMRS and from interviews performed at the time of cognitive screening. Information on length of hospital stay and discharge destination (eg, home vs facility, including skilled nursing and acute rehabilitation facilities) was also obtained from the RMRS. Charlson Comorbidity Index score was calculated using International Classification of Diseases, Ninth Revision codes gathered from 1 year before admission until the time of each participant's discharge from the hospital.[20] The Acute Physiology Score (APS) from the Acute Physiology and Chronic Health Evaluation (APACHE) III was derived from data available in the RMRS to measure the severity of illness.[21] Although the APACHE III was developed in the intensive care unit using data from the first 24 hours after admission, for our study we used the worst laboratory test value during the entire hospital stay to calculate the APS.[22]

Statistical Analysis

Baseline variables are presented as means and standard deviations for continuous variables, and percentages for binary categorical variables. Comparisons between patients receiving GCS and those who did not were performed using 2 tests for categorical variables and Kruskal‐Wallis test for continuous variables. Cox proportional hazard models were used to determine the association between receiving GCS and time to hospital readmission or mortality within 30 days or 1‐year postindex admission while adjusting for other covariates. For the models using time to readmission, patients without readmission were censored either at the endpoint (30 days or 1 year) or at time of death for those who died within the time frame in each model. Because GCS was not randomly assigned, we also conducted a propensity score analysis.[23] A logistic model for the probability of receiving GCS was conducted using patient demographic variables and information collected before and at the time of GCS. Stratified Cox proportional models using quintiles of predicted probability of receiving GCS were used in a propensity‐adjusted Cox model. All data analyses were performed using SAS version 9.3 (SAS Institute, Inc., Cary, NC).


Between July 1, 2006 and May 30, 2008, 415 CI patients were enrolled in the original trial, with 176 receiving the GCS. As shown in Table 1, the GCS and non‐GCS groups differed significantly. The GCS group was older (79.2 years old, 8.1 standard deviation [SD] vs 75.8 years old, 7.8 SD; P0.001), scored lower on the SPMSQ (4.7, 2.7 SD vs 5.5, 2.7 SD; P=0.002), had fewer chronic conditions with a lower mean Charlson Comorbidity Index Score (2.1, 1.86 SD vs 2.8, 2.6 SD; P=0.023), but a higher percentage of delirium (48.9% vs 29.3%), a lower percentage of being discharged home (37.5% vs 56.1%), and a higher mean length of stay (6.4 days, 6.4 SD vs 5.6 days, 5.9 SD; P=0.004). They also had a lower malignancy rate (6.2% vs 14.6%; P=0.007) and a lower number of hospitalizations in the previous year (0.5 admissions, 0.9 SD vs 0.7 admissions, 1.1 SD; P=0.035). No differences were observed in regard to gender, ethnicity, history of myocardial infarctions, chronic obstructive pulmonary disease, cerebrovascular disease, peripheral vascular disease, diabetes, and use of anticholinergic medicines.

Baseline Characteristics and Summary Outcomes Within 1 Year of Hospital Discharge Based on Exposure to Geriatric Consult Services
No GCS, n=239GCS, n=176P Value*
  • NOTE: Abbreviations: ACB, Anticholinergic Burden Scale; APS, Acute Physiology Score; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; GCS, geriatric consult services; MI, myocardial infarction; PVD, peripheral vascular disease; SD, standard deviation; SPMSQ, Short Portable Mental Status Questionnaire. *P value was calculated using 2 tests for categorical variables and Kruskal‐Wallis test for continuous variables.

Baseline characteristics
Mean age (SD)75.8 (7.8)79.2 (8.1)<0.001
% Female66.1 [n=158]68.2 [n=120]0.657
% African American54.8 [n=131]63.6 [n=112]0.071
Mean SPMSQ score (SD)5.5 (2.7)4.7 (2.7)0.002
Admission diagnoses
MI15.5 [n=37]13.6 [n=24]0.675
CHF38.1 [n=91]34.7 [n=61]0.475
PVD7.1 [n=17]9.7 [n=17]0.370
Cerebrovascular13.8 [n=33]19.3 [n=34]0.140
COPD41.0 [n=98]33.0 [n=58]0.094
Diabetes47.7 [n=114]40.9 [n=72]0.169
Malignancy14.6 [n=35]6.2 [n=11]0.007
Metastatic cancer8.8 [n=21]1.7 [n=3]0.002
Mean Charlson Comorbidity (SD)2.8 (2.6)2.1 (1.8)0.023
Mean APS (SD)24.5 (13.8)25.9 (13.5)0.231
Definite ACB Use35.2 [n=84]27.8 [n=49]0.136
Length of stay5.6 (5.9)6.4 (6.4)0.004
% Any delirium29.3 [n=70]48.9 [n=156]<0.001
% Discharged home56.1 [n=134]37.5 [n=66]<0.001
No. of inpatient stays prior year0.7 (1.1)0.5 (0.9)0.035
Follow‐up outcomes
% Readmission within 30 days15.1 [n=36]22.7 [n=40]0.054
% Readmission within 1 year54.4 [n=130]56.3 [n=99]0.765
% Death within 30 days4.2 [n=10]1.7 [n=3]0.253
% Death within 1 year26.8 [n=64]23.9 [n=42]0.569
% Readmission or death within 30 days18.0 [n=43]24.4 [n=43]0.113
% Readmission or death within 1 year64.8 [n=155]63.1 [n=111]0.708

Table 2 describes the association of various factors with receiving GCS. Patients who were positive for delirium (odds ratio [OR]=1.65; 95% confidence interval=0.98‐2.77) and were older (OR=1.04; 95% confidence interval=1.01‐1.08) had a higher propensity to receive GCS, whereas, the presence of metastatic cancer resulted in a lower propensity (OR=0.15; 95% confidence interval=0.02‐1.16) of receiving GCS. The logistic model estimated area under the receiver operating characteristic curve was 0.707.

Results of Logistic Regression Model for Receiving a Geriatric Consult Service Consult Based on Patients' Demographic and Hospital Variables
Adjusted OR (95% CI)P Value
  • NOTE: Abbreviations: CI, confidence interval; OR, odds ratio. *Based on the Anticholinergic Burden Scale.

Age1.04 (1.011.08)0.006
Female1.02 (0.641.63)0.942
African American1.11 (0.711.72)0.657
Short Portable Mental Status Questionnaire score1.00 (0.911.10)0.990
Acute Physiology Score1.00 (0.981.02)0.769
Charlson Comorbidity Score1.11 (0.841.46)0.471
Length of hospital stay1.02 (0.981.07)0.299
Definite anticholinergic use*0.74 (0.461.20)0.219
Any delirium during hospital stay1.65 (0.982.77)0.061
Diabetes mellitus0.72 (0.411.26)0.253
Myocardial infarction0.83 (0.411.66)0.593
Congestive heart failure0.83 (0.471.47)0.524
Peripheral vascular disease1.39 (0.613.18)0.433
Cerebrovascular disease1.30 (0.652.59)0.464
Malignancy0.45 (0.171.21)0.113
Metastatic cancer0.15 (0.021.16)0.069
Chronic obstructive pulmonary disease0.91 (0.531.55)0.727

Table 3 provides results from the Cox models for receiving GCS on readmission and mortality outcomes adjusting for various sets of covariates and with the propensity score adjustment. Model 1 presents unadjusted hazard ratio (HR). Model 2 presents HRs adjusting for a common set of covariates that were significantly associated with at least 1 of the outcomes, whereas model 3 presents the results adjusting for all covariates. All 4 models yielded similar results. As evident from this table, propensity‐adjusted HR for 30‐day readmission was still significantly higher among patients receiving GCS (HR=1.75; 95% confidence interval=1.06‐2.88) but not at 1 year (HR=1.19; 95% confidence interval=0.89‐1.59). There was a trend for decreased mortality for the GCS group at 30 days (HR=0.35; 95% confidence interval=0.09‐1.35), but it disappeared at 1 year (HR=0.91; 95% confidence interval=0.59‐1.40). A composite outcome of readmissions and mortality did not show any difference between the GCS and no‐GCS groups.

Adjusted Hazard Ratios for Geriatric Consult Service Consult Estimated Using Cox Proportional Hazard Models and Propensity‐Adjusted Models
Outcome VariablesModel 1Model 2Model 3Propensity Adjusted
HR (95% CI)P ValueHR (95% CI)P ValueHR (95% CI)P ValueHR (95% CI)P Value
  • NOTE: Abbreviations: CI, confidence interval; HR, hazard ratio. Model 1: unadjusted; Model 2: adjusting for age, Charlson Comorbidity Score, Acute Physiology Scale, disposition status, congestive heart failure, and peripheral vascular disease; Model 3: adjusting for all demographic and hospital variables.

Readmission within 30 days1.65 (1.05, 2.59)0.0301.73 (1.08, 2.78)0.0241.84 (1.133.00)0.0151.75 (1.062.88)0.029
Readmission within 1 year1.13 (0.87, 1.46)0.3731.24 (0.94, 1.63)0.1251.26 (0.941.68)0.1171.19 (0.891.59)0.245
Death within 30 days0.43 (0.12, 1.56)0.1990.34 (0.09, 1.28)0.1100.25 (0.061.02)0.0530.35 (0.091.35)0.126
Death within 1 year0.95 (0.65, 1.45)0.8060.87 (0.58 1.31)0.5060.93 (0.601.42)0.7240.91 (0.591.40)0.669
Readmission or Death within 30 days1.48 (0.97, 2.26)0.0701.49 (0.96, 2.33)0.0781.56 (0.982.47)0.0611.55 (0.972.48)0.069
Readmission or death within 1 year1.05 (0.82, 1.34)0.6991.11 (0.86, 1.43)0.4121.15 (0.881.50)0.3181.08 (0.831.42)0.569


To our knowledge, this is the first study to analyze the impact of GCS on hospital readmission and mortality rates of CI patients. Our results did not show any short‐term or long‐term benefits of GCS for CI patients. Recent studies exploring cost benefits of the GCS have found trends toward lower readmission, but none focused on patients with CI.[6, 24, 25] It is important to note that our study did not use random allocation to assigning the patient into the GCS or control group, thus raising the possibility that patients who received GCS were sicker and were medically and socially more complex than those who did not receive the consult. Moreover, GCS consultation is preferentially sought for and completed for patients with CI and functional limitations, consistent with our finding that GCS patients more often have delirium and are less‐often discharged home.

The nature of the GCS team is another important consideration. Our GCS model did not include unit cohorting of patients, an important component of other proposed GCS models.[26] A recent meta‐analysis found that the GCS models without unit cohorting of patients did not have an impact on 1‐ or 12‐month readmission rates.[27] Low adherence to consultant recommendations (less than 33%) was thought to be a reason for such results. Importance of cohorting with regard to accomplishing recommendations by primary teams, importance of unit staff expertise in geriatric principles, and impact of a unit model on teamwork has also been highlighted by another review.[28] These findings lend to the hypothesis that unit cohorting and direct order placement by the GCS team may improve outcomes among CI patients, including a reduction in readmission rates.

Although readmissions rates were not statistically different between GCS and control groups at 1‐year postdischarge, 30‐day readmission rates were higher among the GCS group. Previous research among older heart failure patients found that a comprehensive transitional care intervention at the time of hospital discharge significantly shortened the time to readmission in the intervention group (P=0.026).[29] The factors identified by the study authors included enhanced supervision by the transitional healthcare teams along with improved awareness and education among treated patients that may have facilitated early recognition of clinical deterioration.[29] A recent study with intensive outpatient care that resulted in increased admissions among chronically ill adults provided a similar conclusion.[30]

GCS patients showed a trend toward decreased mortality as did patients enrolled in previous studies evaluating GCS models in the inpatient setting, as suggested by a recent review.[27] A caveat to note is that these trends favored ward‐styled GCS services as compared to our open GCS model,[27, 28] although the factors cited in these dedicated units affecting mortality included prompt attention to early rehabilitation, delirium management, and prevention of pressure ulcers and are also frequently implemented for patients in our GCS service model and therefore may have produced similar results.

Our neutral results in regard to the readmissions need to be interpreted with caution. First, this study was conducted in a hospital that supports expert geriatric and palliative care teams, both in the inpatient and the ambulatory settings, that provide consultative services and train medicine teams and hospital nursing staff. On the outpatient side, the presence of a robust geriatrics house‐calls program and the Geriatric Resources for Assessment and Care of Elders team results in above‐average care for the control group, and thus may also impact apparent outcomes.[31, 32] Second, 30‐day readmissions represent a complex outcome. Two recent reviews of hospital‐initiated interventions have shown that evidence regarding best strategies to decrease 30‐day readmissions is unclear.[33] Neither review included studies that targeted patients with CI only. The 2 programs that reduced 30‐day readmissions were multifaceted and included personnel who provide bridging between the hospital and the outpatient setting.[34] The GCS does include a focus on postdischarge resources, but does that on a case‐by‐case basis and no formal posthospital follow‐ups are provided. Moreover, the value of 30‐day readmission rates as a marker of quality, even though used by policymakers as an indicator of hospital quality, remains controversial.[35, 36] Broadening the outcomes of interest to include patient‐centered outcomes including satisfaction with care, that have shown to impact other health outcomes, may help improve understanding the benefits of GCS in hospitals.[37] Other comprehensive transitional care models that failed to show a benefit on 30‐day readmissions in older patients still resulted in higher satisfaction among patients.[38] Unfortunately, our evaluation did not include an assessment of patient satisfaction and quality of transitions.

Since the study period, GCS at our hospital now has incorporated a more robust focus on advance care planning (ACP) and execution of Physician Orders for Scope of Treatment that were legislated in the state in July 2013. The GCS team members are expert in carrying out complex ACP discussions and also partner with the inpatient palliative care team. It is quite possible that a study of more recent outcomes will yield more positive results for the selected outcomes. Thus, for future trials that aim to study the impact of GCS in the inpatient settings, it may be advisable to include important quality markers such as implementation of ACP and patient satisfaction along with the health utilization outcomes.


As mentioned previously, it is possible that our risk adjustment was insufficient to account for all the medical and psychosocial differences among groups. For example, the overall anticholinergic impact of various medications such as antipsychotic medications and histamine‐2 blockers was assessed via the Anticholinergic Burden Scale on admission, but we did not have information on medication prescribing during the stay. We were further limited by lack of baseline functional status and socioeconomic details, both of which are related to 30‐day readmissions. For example, living alone, prior use of assist devices, and belonging to lower socioeconomic status are correlated with higher readmission rates.[39, 40] Patients with available social support may receive more intense supervision and may seek medical attention sooner. On the other hand, worsening health among CI patients without any approximate social support may be unnoticed for days. Absence of details of inpatient interventions may also have resulted in unmeasurable confounders that could have impacted our study outcomes. Finally, lack of information on the uptake of GCS recommendations by the primary teams is another limitation of this analysis. Future trials should include strategies to address these information gaps.


Our results comparing inpatient geriatrics consultative services with usual care in hospitalized elders having cognitive impairment failed to demonstrate an impact on readmissions and mortality. A clinical lesson learned, however, is that much work is still required to reduce readmission and mortality rates in this especially vulnerable patient population.


Disclosures: This work was supported by grants from a Geriatric Academic Career Award (K01HP20517) through Health Resources and Services Administration, R01AG034205 and K23‐AG043476 from the National Institute on Aging, and the John A. Hartford Foundation Center for Excellence in Geriatric Medicine. The sponsors had no role in the study design, evaluation, or manuscript development. The authors report no conflicts of interest.


Online-Only Materials

PDF icon Supporting Information (1)80.99 KB
   Comments ()