Detection of Cognitive Impairment after Critical Illness with the Medicare Annual Wellness Visit: A Cohort Study

Abstract

Rationale: Cognitive impairment after critical illness is common in observational studies of older intensive care unit (ICU) survivors. The rate of screening for and diagnosis of cognitive impairment in ICU survivors in nonresearch settings is unknown.

Objectives: To determine how often cognitive impairment was detected in older adults in the year after critical illness at an academic medical center as part of 1) the Medicare Annual Wellness Visit (AWV) and 2) routine clinical care.

Methods: This study was a retrospective cohort study conducted at an urban academic medical center. The study included 696 patients aged 65 years and older admitted to the medical ICU between October 1, 2016, and October 1, 2018, and discharged alive. Patients were also required to have a health system–affiliated primary care provider. Patients were followed for 1 year. We defined cognitive impairment detected in the AWV as either an indicated diagnosis of cognitive impairment or dementia or patient, family, or provider indication of memory concerns during the AWV. We modeled the incidence of AWV completion and the detection of cognitive impairment using semiparametric additive models accounting for the competing risk of death.

Results: Over 1 year of follow-up, the cumulative incidence of mortality was 23.0% (95% confidence interval [CI], 19.9–26.1%), with 24.7% (95% CI, 21.5–27.9%) completing the AWV. The cumulative incidence of cognitive impairment first detected through the AWV was 3.4% (95% CI, 1.8–5.0%) at 1 year, with a higher cumulative incidence for diagnoses of cognitive impairment or dementia first indicated via encounter diagnosis codes or the electronic health record problem list (5.9%; 95% CI, 3.9–7.9%).

Conclusions: The results of our study suggest that the currently implemented AWV is unlikely to be an adequate mechanism for detecting cognitive impairment in a high-risk population such as those recovering from critical illness.

More than 1 million older adults survive an intensive care unit (ICU) stay in the United States each year, and this number will likely increase as the population ages (1, 2). In survivors, an ICU admission is often a life-altering event associated with new or worsening cognitive impairment (3–6). In cohort studies of ICU survivors, cognitive impairment is present in 40–100% of patients at the time of ICU discharge, with cognitive deficits often persisting in older patients (7–10). Despite the high burden of cognitive impairment reported in observational cohort studies, the rate of screening for and diagnosis of cognitive impairment in ICU survivors in nonresearch settings is unknown.

On January 1, 2011, the U.S. Centers for Medicare and Medicaid Services (CMS) instituted Medicare coverage for an Annual Wellness Visit (AWV) as part of the Patient Protection and Affordable Care Act (11, 12). As per CMS regulation, the AWV requires screening for cognitive impairment by “assessment of an individual’s cognitive function by direct observation, with due consideration of information obtained by way of patient report, concerns raised by family members, friends, caretakers, or others.” (12) The AWV provides a unique opportunity to detect cognitive impairment after critical illness because cognitive screening is not otherwise part of routine clinical care (11, 13, 14). The Society of Critical Care Medicine recently published a consensus statement recommending screening for cognitive impairment (together with other complications of critical illness) in all high-risk patients after discharge (15). The role of the AWV in detecting post-ICU cognitive impairment in older adults is unknown.

The primary objective of this study was to determine the utility of the currently implemented AWV as a mechanism for screening for post-ICU cognitive impairment in older adults at an academic health center. We also explored how often incident cognitive impairment was listed as a diagnosis, independent of the AWV, for older ICU survivors in the year after critical illness in our health system.

Methods

Study Design and Population

We conducted a retrospective cohort study of all patients aged 65 years and older (at the time of admission) discharged from the medical ICU at Wake Forest Baptist Health, a tertiary academic referral center. For this study, we included all patients with a medical ICU admission between October 1, 2016, and October 1, 2018. We followed patients for at least 1 year after ICU discharge via information contained in the electronic health record (EHR; Epic). Given that our primary objective was to determine the frequency of cognitive screening via the AWV, we restricted our study to patients with a primary care physician affiliated with our health system. Given our focus on detecting incident cognitive impairment after ICU discharge, we also tracked whether patients had a diagnosis of dementia or cognitive impairment before their ICU stay (defined below). This study was approved by the institutional review board at Wake Forest Baptist Health with a waiver of informed consent.

Outcome Definitions

For this study, we included patients with a primary care provider in any department affiliated with our institution. At our institution, the AWV is completed by physicians, advanced practice providers, or nurses depending on the particular outpatient primary care practice. The AWV is completed in our EHR with an AWV smartform template that permits tracking all components of the AWV as structured data elements. A validated cognitive assessment tool is not mandated as part of the AWV, but rather a subjective assessment is performed. Within our health system, the first component of the cognitive screening portion of the AWV asks whether the patient has been diagnosed with cognitive impairment or dementia (“Has a diagnosis of dementia or cognitive impairment?” yes or no). We assumed that AWV cognitive screening was complete if this question was answered as part of the AWV visit. This question is followed by the following structured question: “Are there any memory concerns by the patient, others or providers?” We therefore defined “detection” of cognitive impairment as part of the AWV if either a diagnosis of cognitive impairment or dementia was recorded or if the patient, family, or provider indicated memory concerns. We also examined the prevalence of diagnosis codes for cognitive impairment and delirium before the ICU admission, during the ICU admission, and in the year after critical illness. We considered diagnosis codes indicated as part of encounter diagnoses as well as those indicated on the EHR problem list. International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes for cognitive impairment included F00-F03, G20, G30.x, F06.8, G31.0x, G31.1, G31.83, G31.84, and R41.3 (16). ICD-10 codes for delirium were based on the union of codes used in two recent studies of delirium with administrative data (17). Finally, we extracted dates of death from the EHR. Mortality information in our EHR is supplemented through a monthly deterministic linkage (based on name, age, sex, date of birth, and race/ethnicity) to the North Carolina State Center for Health Statistics death index, which captures any individuals who die within North Carolina.

Other Patient Information

We also examined ICD-10 diagnosis codes (encounter diagnoses and those from the EHR problem list) in the 2 years before ICU admission for comorbidities within the Charlson comorbidity index (16). We similarly examined frailty through an EHR-derived frailty index, the electronic frailty index (eFI), based on the model of deficit accumulation (18). Briefly, the eFI is composed of 54 total items, integrating data from diagnosis codes, laboratory measurements, vital signs, smoking history and medications, and functional information from the Medicare AWV. We categorized frailty status on the basis of the eFI as fit (eFI ≤ 0.10), prefrail (0.10 < eFI ≤ 0.21), or frail (eFI > 0.21), as in prior work (18, 19).

Statistical Analyses

We estimated the incidence of AWV completion and the detection of cognitive impairment using semiparametric models accounting for the competing risk of death. We examined associations with AWV completion using the subdistribution hazard regression model of Fine and Gray with the following covariates: age, sex, White versus non-White race, frailty status, discharge disposition (discharged home vs. not discharged home), whether the patient had previously completed a Medicare AWV, and whether there were diagnosis codes for cognitive impairment and/or delirium in the 2 years before the ICU admission or during the ICU encounter (20). We used robust SEs to account for clustering of patients by primary care clinic. For the detection of cognitive impairment, we separately modeled whether indications for cognitive impairment first arose as part of AWV screening or whether cognitive impairment was first indicated by encounter diagnosis codes or was listed on the EHR problem list. All statistical analyses were performed using the timereg package for the R Statistical Computing Environment (21).

Results

During the study time period, 1,834 patients aged 65 years and older were discharged alive from the medical ICU (Figure 1). Of these, 1,138 patients did not have a primary care physician affiliated with our health system and were thus excluded from subsequent analyses, leaving a total of 696 patients in our analysis cohort. The mean age in the cohort was 75 years (SD 8.0). The cohort was 49% male, and 70% of patients were of White race/ethnicity (Table 1). A diagnosis code for cognitive impairment or delirium in the 2 years before admission was found in 110 patients (15.8%). The median electronic frailty index for this cohort was 0.22 (interquartile range [IQR], 0.16–0.28). A total of 134 patients (19.3%) required vasopressors during their ICU stay, and 136 patients (19.5%) received invasive mechanical ventilation. The median ICU length of stay was 2.6 days (IQR, 1.6–4.7 d). A total of 439 patients (63.1%) were discharged to the home.

Figure 1.

Cohort flow diagram describing entry into cohort and outcomes. All patients aged 65 years and older discharged from the medical ICU between October 1, 2016, and October 1, 2018, were eligible for inclusion into the cohort. Patients without a PCP in our health system or with a preexisting diagnosis of cognitive impairment were subsequently excluded. Patients were then followed for the following outcomes: death within 12 months, completion of Annual Wellness Visit (AWV), or diagnosis of cognitive impairment outside of the AWV. ICU = intensive care unit; PCP = primary care provider.

Table 1.

Patient characteristics of cohort with health system–affiliated primary care provider

Characteristics	Patients (N = 696)
Age, mean (SD), yr	74.9 (8.0)
Sex, F, n (%)	357 (51.3)
Race/ethnicity, n (%)
White	489 (70.3)
Black	187 (26.9)
Hispanic	9 (1.3)
Other	11 (1.6)
Charlson comorbidities, n (%)
Zero	73 (10.5)
One or two	354 (50.9)
Three or more	236 (33.9)
Unknown	33 (4.7)
eFI, median (IQR)	0.22 (0.16–0.28)
Frailty status based on eFI, n (%)
Fit (eFI ≤ 0.10)	40 (5.7)
Prefrail (0.10 < eFI ≤ 0.21)	232 (33.3)
Frail (eFI > 0.21)	346 (49.7)
Insufficient data to calculate eFI	78 (11.2)
Completed Medicare AWV visit in 2 yr before ICU admission, n (%)	266 (38.2)
Diagnosis code for cognitive impairment or delirium in 2 yr before ICU admission, n (%)	110 (15.8)
Diagnosis code for cognitive impairment or delirium during ICU admission, n (%)	81 (11.6)
ICU length of stay, median (IQR), d	2.6 (1.6–4.7)
Hospital length of stay, median (IQR), d	3.7 (1.8–7.0)
Vasopressor use in ICU, n (%)	134 (19.3)
Received mechanical ventilation, n (%)	136 (19.5)
Discharge disposition, n (%)
Acute care facility	243 (34.9)
Home or self-care	264 (37.9)
Home with home health	175 (25.1)
Hospital transfer	4 (0.6)
Other	10 (1.4)

Definition of abbreviations: AWV = Annual Wellness Visit; eFI = electronic frailty index; ICU = intensive care unit; IQR = interquartile range; SD = standard deviation.

The cumulative incidence of mortality at 1 year after ICU discharge was 23.0% (95% confidence interval [CI], 19.9–26.1%) (Figure 2). Accounting for the competing risk of death, the cumulative incidence of completion of the AWV was similar (24.7%; 95% CI, 21.5–27.9%) at 1 year of follow-up. In the cohort of 696 patients, 172 completed the AWV, and 160 died during follow-up. Of the 172 patients who completed the AWV, 156 did not have a preexisting code for cognitive impairment before ICU admission, and 23 had an indication for cognitive impairment noted during the AWV. When we examined factors associated with completion of the AWV, the strongest predictor was whether the individual had previously completed an AWV in the 2 years before the ICU admission (hazard ratio, 2.28; 95% CI, 1.75–2.97) (Table 2). Individuals with insufficient outpatient data in the EHR to calculate the eFI (which largely indicates limited contact with the health system before the ICU admission) were significantly less likely to complete the AWV in the year after ICU discharge (hazard ratio, 0.31; 95% CI, 0.13–0.75). Although almost one in four ICU survivors completed the AWV, the rate of detected cognitive impairment was much lower. The cumulative incidence of cognitive impairment first detected through the AWV was 3.4% (95% CI, 1.8–5.0%) at 1 year, with a higher cumulative incidence for diagnoses of cognitive impairment or dementia first indicated via encounter diagnosis codes or the EHR problem list (5.9%; 95% CI, 3.9–7.9%) (Figure 2).

Figure 2.

Completion of Medicare Annual Wellness Visit (AWV) accounting for competing risk of death (left) and detection of cognitive impairment indicated at AWV and diagnosis code for cognitive impairment accounting for competing risk of death (right).

Table 2.

Regression modeling for characteristics associated with completion of a Medicare AWV after critical illness accounting for the competing risk of death

Characteristic	Hazard Ratio (95% CI)	P Value
Age, per 5 yr increase	0.97 (0.82–1.16)	0.77
Sex, F	1.00 (0.68–1.47)	0.99
Non-White race	0.73 (0.56–0.95)	0.02
Completed Medicare AWV visit in 2 yr before ICU admission	2.28 (1.75–2.97)	<0.001
Frailty status (fit)
Prefrail	0.76 (0.32–1.83)	0.54
Frail	0.79 (0.46–1.36)	0.40
Insufficient data for eFI	0.31 (0.13–0.75)	0.001
Discharged home	1.21 (0.84–1.74)	0.31
Diagnosis code for cognitive impairment or delirium in 2 yr before ICU admission	0.89 (0.41–1.94)	0.77
Diagnosis code for cognitive impairment or delirium during ICU admission	0.63 (0.29–1.37)	0.24

Definition of abbreviations: AWV = Annual Wellness Visit; CI = confidence interval; eFI = electronic frailty index; ICU = intensive care unit.

Discussion

Persistent cognitive impairment is increasingly recognized as an important complication of critical illness, and expert groups now recommend screening for cognitive impairment in those patients at increased risk (15). In our study, we evaluated the role of the Medicare AWV in detecting post-ICU cognitive impairment in older adults in an academic health system. In the year after a medical ICU admission, the AWV was only performed in one of four patients with a primary care provider in our health system. In those patients completing the AWV, only 3% noted cognitive concerns or had cognitive impairment recognized as a diagnosis. Although the true burden of persistent cognitive impairment after illness in this population is unknown, given the high incidence of new or worsening cognitive impairment reported in observational cohort studies, this would suggest a potentially large burden of unrecognized cognitive impairment among ICU survivors in our health system.

Critical illness increases the risk of new or worsening cognitive impairment in older adults. In a well-characterized longitudinal cohort study (the Bringing to Light the Risk Factors and Incidence of Neuropsychological Dysfunction in ICU Survivors or BRAIN-ICU study), one of four patients had a degree of impairment similar in severity to patients with mild Alzheimer’s disease 1 year after ICU discharge, and one of three had impairment typically associated with moderate traumatic brain injury (4). In older survivors of severe sepsis, the most common cause of noncardiac critical illness, the prevalence of moderate or severe cognitive impairment increased by 10.6 percentage points and was associated with a high rate of new functional limitations (5). Cognitive declines of this magnitude are associated with increased likelihood of nursing home admission, increases in caregiver burden, and higher subsequent mortality (6, 22). The association between critical illness and cognitive impairment exists after adjusting for premorbid cognitive screening scores as well as comorbid conditions, suggesting that factors related to the acute illness may be causally related to cognitive decline (3, 23).

Recognizing cognitive impairment in patients after critical illness has important implications for older adults, their families, and the healthcare system. Among older adults, cognitive impairment is associated with institutionalization (24), subsequent hospitalization (25), and significant societal costs (22, 26). Cognitive impairment is a clinically dominant comorbidity, influencing the presentation and management of other conditions (27). People with cognitive impairment use more healthcare services and at higher cost than those without cognitive impairment, and they represent a subgroup at increased risk for emergency department visits and readmissions (28–30). The presence of cognitive impairment also has the potential to influence many aspects of patients’ care, from the effectiveness of doctor–patient communication to treatment adherence, selection of appropriate medications, and likelihood of follow-up (27). Furthermore, specific recognition of mild cognitive impairment is important because it increases the risk of dementia 10-fold and may be a subgroup important to target for closer monitoring as well as interventions shown to decrease the risk of progression (31, 32). The inability to routinely detect post-ICU cognitive impairment also limits our ability to identify modifiable risk factors contributing to the development of this complication of critical illness.

At present, the U.S. Preventative Task Force does not recommend routine dementia screening for the general population; however, experts in ICU recovery research recommend screening individuals with increased likelihood of developing cognitive impairment after critical illness (15, 33, 34). Guidelines for dementia screening in the general population also recommend clinicians assess cognitive function whenever cognitive impairment or deterioration is suspected (11, 27). The AWV, as currently implemented at our institution without a formal cognitive assessment tool, is unlikely to be an adequate mechanism for the widespread detection of cognitive impairment in the post-ICU population. There are multiple reasons why this may be the case. The AWV may not be completed frequently enough in this patient population because of lack of follow-up, readmissions, or institutionalization after an ICU stay. Although our AWV protocol (structured questions to elicit cognitive concerns without the use of a formal screening tool) is typical of many, the nature of cognitive impairment is such that patients may lack insight into their cognitive deficits and thus deny cognitive concerns when asked by their physicians. The results of this study suggest that a formal screening tool will be needed in this subgroup. The Society of Critical Care Medicine now recommends a formal assessment for post-ICU cognitive impairment, as well as other persistent complications of critical illness, at 2 to 4 weeks after ICU discharge (15). Given the high burden of cognitive impairment in older ICU survivors and the likely increase in this complication expected with the ongoing coronavirus disease (COVID-19) pandemic, health systems should consider adapting their AWV to include a formal cognitive assessment in patients recovering from critical illness, including those recovering from COVID-19.

There are several limitations to this study. It is a single-center study, which may limit the generalizability of our findings. We limited our study to patients with a primary care physician in our health system, but there may have been patients who received cognitive screening or diagnoses in different health systems, which would not be captured in our EHR. We also recognize that those patients without primary care access may be at an even greater risk for post-ICU cognitive impairment, and this group was not examined in this study. This study used EHR data to determine a diagnosis of cognitive impairment or dementia, which can be subject to misclassification and other errors. There may have been patients with cognitive impairment recognized or discussed by their physician without a diagnosis entered into the medical record. Future work is needed to examine the incidence of post-ICU cognitive impairment across health systems and to understand factors associated with a delayed or missed diagnoses.

In conclusion, we have found that post-ICU cognitive impairment is detected at a lower rate than expected in those recovering from critical illness at our academic center even with the implementation of cognitive screening questions imbedded in the AWV. Future work is needed to evaluate screening for post-ICU cognitive impairment in other health systems and to understand barriers to the recognition of this important complication of critical illness in older adults.