Mild cognitive impairment is a pre-cursor to dementia and Alzheimer's disease.1 Cognitive impairment is associated with increased rates of readmission, longer hospitalizations, and increased mortality in elderly patients.2–4

The Mini-Cog test is a quick and easy test that has been shown to have good accuracy at identifying cognitive impairment.4,5 Several cardiac populations with cognitive impairment, identified on Mini-Cog, have poor post-hospital discharge outcomes.6–10 While cognitive impairment is shown to be fairly common in heart failure patients, it is less established what the prevalence is in a broader cardiology cohort, and how significantly it impacts mortality and readmission rates in this broader population.8–12

Since April of 2015 the cardiopulmonary rehabilitation therapists at our institution have administered the Mini-Cog test to patients referred for inpatient (phase I) cardiopulmonary rehabilitation. Our study aims were to evaluate a large, consecutive cohort of patients referred for cardiopulmonary rehabilitation and determine the prevalence of cognitive impairment and compare readmission rates and mortality for those with and without cognitive impairment. We hypothesized that hospitalized patients receiving cardiopulmonary rehabilitation that perform poorly on the Mini-Cog would identify a group at high-risk for readmission and death.

The Mini-Cog exam was prospectively ordered on 1627 consecutive cardiopulmonary rehabilitation patient encounters from 1 March 2016 to 28 February 2017 (Fig. 1). Each patient's electronic medical record (EMR, Epic, United States) was retrospectively reviewed to collect and record study data. The institutional review board approved the study with a waiver of patient consent. Seven patients were excluded from the study cohort, as they died during their index admission. There were 152 patients that had more than one phase I cardiopulmonary rehabilitation order during the study period. Only the initial Mini-Cog score and associated data were included to remove duplicate patients from contributing to the outcome data.

Fig. 1.

Flow diagram of study subjects.

(0.28MB).

Of the 1440 patients, 986 (68%) completed the Mini-Cog (cohort A), with 46 (4.7%) positive Mini-Cog (cohort A1) and the remaining 940 (95.3%) having a negative Mini-Cog (cohort A2). The 95% confidence interval for the prevalence of a positive Mini-Cog was 3.4–6.1%. Cohort B includes 454 patients that did not perform the Mini-Cog. Reasons that the Mini-Cog was not completed are shown in Fig. 2. There were 22 patients in cohort B that completed the Mini-Cog but did so as an outpatient prior to admission. Of these 22 patients, 1 (4.5%) had a positive Mini-Cog. Cohort A1 was significantly older, more frequently female, with more chronic kidney disease, heart failure, and chronic obstructive pulmonary disease than the other cohorts. Baseline demographic data for the groups are shown in Table 1.

Fig. 2.

Reasons Mini-Cog not administered. OT, occupational therapy.

(0.11MB).

On univariate analysis cohort A1 had significantly higher rates of all-cause readmission compared with cohorts A2 and B (63% vs 44% and 47%; P=.02), and significantly higher mortality (28% vs 9% vs 15%; P<.001) (Table 2).

Before matching, there was a large imbalance of covariates between A1 and A2 ranging from 0.1 to 0.87 (absolute value of standardized mean differences) and empirical cumulative density functions max (eCDF max) mostly were ranging from 0.1 to 0.46 (Table 3). Optimal full matching yielded the best balance between two groups, as indicated in Table 4. Sample size remained intact, with the distribution of propensity score shown in Fig. 3. After matching all standardized mean differences and empirical cumulative density functions max (eCDF max) for covariates were all below 0.08 (except one value was 0.11), variance ratios were close to 1. Standardized mean difference plot is indicated in Fig. 4.

Fig. 3.

Distribution of propensity scores.

(0.16MB).

Fig. 4.

Standardized mean difference plot full matching. CAD, coronary artery disease; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; HTN, hypertension.

(0.19MB).

Survival curves analyzing freedom from the composite outcome of readmission plus death for matched cohorts are shown in Fig. 5. At the 3-month follow-up period in Cox proportional hazards model, cohort A1 had significantly more readmission and death compared to cohort A2 (P=.002). That is, at a given instant in time a patient with positive Mini-Cog is 93.7% times as likely to die or get readmitted to hospital as a patient with negative Mini-Cog adjusting for age, gender and 9 different co-morbidities (95% confidence interval, 1.262–2.974). The result for the entire follow-up period was not significant.

Fig. 5.

Survival at 3-months. A P-value≤.05 was considered statistically significant.

(0.16MB).

Assessing the Mini-Cog score as a continuous variable revealed progressive increases in death and the composite of death and readmission, along with progressive, significant increases in age, hypertension, chronic kidney disease, diabetes mellitus, and cerebrovascular accident (Tables 5 and 6).

Sub-group analysis found 9.5% of those ≥70 years of age had cognitive impairment. Patients≥70 years of age, with a positive Mini-Cog had higher rates of all-cause readmission and death compared to patients≥70 years of age with a negative Mini-Cog, but significant differences between the cohorts remained. Of those ≥80 years of age, 12.3% had cognitive impairment. The baseline characteristics were very similar between those with and without a positive Mini-Cog and ≥80 years of age, with no difference in readmission or death. Of the 590 patients<70 years of age, 8 (1.3%) had cognitive impairment.

Within cohort B were 35 (8%) patients that either had a history of dementia/cognitive impairment or had a Montreal Cognitive Assessment (MoCA) administered by an occupational therapist that identified cognitive impairment during their index admission. During the follow-up period 12 (34%) of these patients died and 21 (60%) met the combined end-point of readmission+death.

Finally, we conducted sensitivity analysis for optimal full matching. We used Huber–Maritz M-statistics with Γ=1 and P=.0069, which means the outcome and study conclusion was robust to hidden bias from an unobserved confounder.

In our phase I cardiac rehabilitation cohort, poor performance on the Mini-Cog identified an elderly group with significantly increased rates of readmission and death. This poorly performing Mini-Cog group was also found to be an older patient population with multiple co-morbidities.

A positive Mini-Cog was found to be associated with increased absolute levels of death, cardiac readmission and all-cause readmission, with the composite end-point of all-cause readmission+death reaching statistical significance (P=.05). Cox proportional hazards model with survival curves found cohort A1 to have significantly increased readmission+death at 3 months compared to cohort A2.

These results are consistent with previous publications that have found that cognitive impairment is associated with increased readmission rates and poorer survival in heart failure patients.8–12 A 2015 study that looked specifically at heart failure patients with a positive Mini-Cog found significantly increased rates of readmission and death.9 Interestingly, if the patients were discharged to a long-term care facility then their risk of readmission and death decreased.8 Another study found cognitive impairment to be an independent predictor of mortality in veterans with heart failure.14

The prevalence of 4.6% that we observed in our cohort is a relatively novel finding, as we are unaware of any other publication that evaluated the rate of cognitive impairment in a large, consecutive cardiopulmonary rehabilitation cohort. A 2017 publication from Germany found 36.7% of younger cardiac rehabilitation patients (average age 54.5 years), with a diagnosis of coronary artery disease, had a MoCA score below 26, which was used as the threshold for cognitive impairment.15 This rate of cognitive impairment is significantly higher than what we observed in our cohort, despite their cohort being significantly younger than ours. A potential explanation for the stark difference in the rate of cognitive impairment between our data and this publication is their use of MoCA to assess for cognitive impairment and our use of the Mini-Cog. The MoCA has been shown to have much lower specificity than the Mini-Mental State Examination (MMSE) for detecting cognitive impairment post-stroke.16 Interestingly, the MMSE was used as the gold-standard to establish the efficacy of the Mini-Cog for diagnosing cognitive impairment.5 It is also worth noting that 5% of the German cohort had more severe cognitive impairment, which could potentially be a group more similar to our cohort A1.15

The prevalence of cognitive impairment increased with age in our cohort, with 9.5% of those 70 or older and 12.3% of those 80 and older having a positive Mini-Cog test. This is consistent with previous publications, including a large retrospective series that found 11.6% of patients≥75 years of age to have undiagnosed cognitive impairment.2–5,8–11 Cognitive impairment was uncommon in patients<70 years of age in our cohort. With consideration of this finding, we have decided to no longer administer the Mini-Cog to phase I cardiopulmonary rehabilitation patients younger than 65 years of age.

The simple incorporation of the Mini-Cog into phase I cardiac rehabilitation provides prognostic information for patients that have a positive assessment. Identifying individuals with cognitive impairment provides an opportunity to intervene, which may slow the progression of their cognitive impairment and improve outcomes. The targets of interventions include reducing alcohol consumption, smoking cessation, treating vascular risk factors, cognitive activities (games, music, reading, etc.), cardiovascular exercise, strength training and even adopting a Mediterranean diet.17,18 Poor cardiorespiratory fitness is also associated with cognitive dysfunction, with cardiac rehabilitation being associated with improvements in cognition.19–21 Thus, referral to phase II cardiac rehabilitation will allow them to derive benefit that may improve their cognitive function, in addition to providing the typical benefits of cardiac rehabilitation – improved exercise capacity, improved quality of life, improved cardiovascular risk factor profile, and improved ability to perform activities of daily living.19–26 Even if these patients are unable to attend all prescribed sessions of cardiac rehab publications have identified the incremental benefit of individual sessions on readmission rates.27,28 Future retrospective studies could validate our findings in a larger database, while future prospective, randomized studies could evaluate interventions to reduce mortality and morbidity in patients identified as having cognitive impairment on the Mini-Cog assessment.

The Mini-Cog assessment can be incorporated into an inpatient cardiac rehabilitation program and will identify cognitive impairment in approximately 10% of those ≥70 years of age. Poor performance on the Mini-Cog identified an older group with multiple co-morbidities that had significantly increased rates of readmission and death at 3-months follow-up.

The authors received no financial support for their research, authorship, and/or publication of this article.

C. House collected data, analyzed data, and drafted the manuscript; H. Dang performed the statistical analysis; K. Moriarty critically revised the manuscript; W. Nelson conceptualized the study and critically revised the manuscript. All authors revised and accepted the manuscript.

The authors declare that there is no conflict of interest.