The value of routine screening for cognitive impairment in a primary care setting: a retrospective cross-sectional study


 BACKGROUND

Cognitive impairment is not uncommon among older individuals but is often underdiagnosed in the primary care settings. Our objective is to identify the prevalence of varying degrees of cognitive impairment in older adults in urban primary care clinics.
METHODS

This cross-sectional study was undertaken from May to July 2019. The Montreal Cognitive Assessment (MoCA) was administered to participants 65-year and older who were seen during a routine visit to our primary care clinic. The participants were recruited on a sequential basis. The primary outcomes of the study were the MoCA scores, and the level of cognitive impairment, if any, indicated by the scores.
RESULTS

Out of the 133 participants, 46 (34.6%) scored below the cut-off of 23 out of 30, indicating certain level of cognitive impairment. The mean MoCA score was 23.24. The average age was 75.19 years. Average years of education was 12.6. In our cohort, higher MoCA scores were associated with increased years of education. MoCA scores were not inversely correlated with age. Language proficiency in the MoCA test version administered had significant impact on the MoCA scores.
CONCLUSIONS

A relatively high prevalence of cognitive impairment was found in our cohort. Further study is required to accurately assess the prevalence of cognitive impairment in general population. However, the findings attest the value of routine screening for cognitive impairment in primary care settings and warrant routine screening of older individuals to help in early detection of mild cognitive impairment.

demonstrated to be effective as a screening tool for cognitive impairment in various clinical settings, including in older patients with cancer 13 , in amnesic patients 14 , and in the elderly at the primary health care level 15 . One drawback in using the MoCA is that the test scores could be affected by years of education 16 , and despite the one-point correction for less than grade 12 education, the suggested cut-off score of 26 to diagnose MCI may over-diagnose MCI in patients with less formal education [17][18] .
Our objective is to analyze the use of MoCA testing in a primary care clinic to identify the prevalence of various levels of cognitive impairment in participants to determine if routine screening of MCI is warranted in all primary care clinics.

Study aim and design
In this cross-sectional study, the MoCA was administered to the patients individually and the scores were recorded to evaluate cognitive impairment in patients. The demographic data including age, gender, years of education, and rst language were recorded. Data on comorbidities including hypertension, diabetes (DM), dyslipidemia, other cardiovascular risk factor, family history of dementia, depression, and brain surgery/injury/stroke were also analyzed.
The English version of the MoCA was the default version for assessment. As the clinic has a sizable Mandarin-speaking patient population, the participants who only speak Mandarin were administered the Mandarin version of the MoCA.

Setting and patients
Our cohort included participants 65 years and older who were patients visiting an urban primary care clinic in the Greater Toronto Area (GTA). The participants were recruited sequentially from the patients visiting the clinic who met the age criteria. A total of 144 patients were invited for assessment from May to Jul 2019. Six patients (4.2%) declined to participate in the study. Five patients (3.5%) were excluded from the nal cohort because of complete language barrier, hearing impairment, or visual impairment that precluded completion of the MoCA without assistance from a third party. The nal cohort included 133 patients (92.3%).

Statistical analysis
The independent t-test was used to examine the impact on the MoCA scores by gender, as well as potential confounders including hypertension, diabetes, hypercholesterolemia, depression or other psychiatric diagnoses, and family history of dementia. Level of signi cance is p < 0.05.
The subjects were categorized into 3 groups based on age in years, 65-74, 75-84, and ≥ 85, respectively. The subjects were also strati ed by years of education in years, ≤ 12, 13-15, and ≥ 16, to analyze the potential effect of years of education on MoCA scores. These strati cations were set to correspond to the level of education up to high school, college/university, and postgraduate degrees, respectively.
The subjects were also categorized based on whether the MoCA was administered in their rst language (only English and Mandarin versions available) to analyze the potential impact of language barrier. We de ned rst language based on patient indication as either their mother tongue or language of greatest uency, if other than English or Mandarin.
The association of MoCA scores and age subgroups, years of education subgroups, and language pro ciency in the MoCA version used was analyzed by ANOVA test. We used IBM® SPSS Statistics 26 and JASP 0.12.2.0 for statistical analysis.

Results
A total of 144 patients aged 65 years and older were recruited. Of these patients, 6 (4.2%) declined to participate in the study and 5 (3.5%) were excluded from the nal cohort. Of the 133 patients included in the study, 46 (34.6%) had a MoCA score below the cut-off of 23, indicating certain level of cognitive impairment (Table 1). Thirty-three participants (24.8%) were in the MCI category. The mean MoCA score was 23.24 (min: 9, max: 30).  Using two-way ANOVA, we saw no signi cant difference among the different age groups strati ed by education level, and no interaction between the age and education level. Signi cant difference was between different language groups (p < 0.001), and there was signi cant interaction of the language factor on the age groups (p = 0.021), mostly with the ≥ 85 years age group. There was signi cant difference in MoCA scores among different education groups (p = 0.011) and language factors (p = 0.007). There was no signi cant interaction of language with education level (p = 0.211).

Interpretations
In our cohort, 46 participants (34.6%) had a MoCA score lower than the cut-off of ≥ 23 indicating cognitive impairment. Of these, 29 (72.5% of the cognitive impaired participants, or 24.2% of all participants in this study) had mild cognitive impairment. The percentage of MCI among our patients was unexpectedly high despite the more conservative cut-off and education level adjusted scores, although not higher than prior studies (up to 36.7%) [5][6] . In addition, we found that MoCA scores were signi cantly lower when the test language of MoCA is not the test-takers' mother tongue or the most uent language. Our clinic is an urban primary care clinic in the Greater Toronto Area (GTA) serving a diverse community with many immigrants. English is not the rst language in a signi cant proportion (at least 30%) of our cohort. This makes our study cohort different from prior published studies which usually had relatively homogeneous cohorts. However, our multi-ethnocultural community has become quite typical for many large Canadian or American cities, and the ndings from our cohort may re ect similar results in many Canadian or American cities.
Our results concur with another study that education could affect the MoCA scores 16 . The participants with a total of education less than 12 years scored lower in MoCA. There is a signi cant difference in the MoCA scores when comparing the group with ≤ 12 years of education (high school graduate or fewer years of education) with the group with ≥ 16 years of education (university graduate or higher degree). The difference is also signi cant when the education level groups were strati ed by the MoCA version concordant or not with the rst language (Table 3). It is unclear if more years of education could slow down the progression of cognitive impairment or just mask the ability of current screening tools to identify cognitive impairment. DM (p = 0.057) and hypertension (p = 0.047) were also found to be borderline signi cant towards the participants' MoCA score. As our study had a relatively small sample size, a clearer statistical association may be elucidated with an increased sample size. Several studies had previously shown a signi cant association between DM and dementia [22][23] or cognitive impairment 24 . However, the association of DM and cognitive impairment is not statistically conclusive in our study due to borderline p value. Few studies had examined the relationship between hypertension or cardiovascular disease and cognitive impairment rather than vascular dementia. Our study revealed a marginal statistically signi cant association between hypertension and cognitive impairment. As hypertension is a common comorbid condition to diabetes, we suggest that the potential effects of DM and hypertension/cardiovascular disease on dementia and cognitive impairment be further investigated together in the future. Moreover, our study showed borderline statistically signi cant association between depression and cognitive impairment which merits further investigation.
In our study, we use the MoCA instead of MMSE to screen for cognitive function. Both tools rely heavily on reading, writing, and verbal response. Only English and Mandarin versions were used in our study due to administrators' personal language pro ciency. In our study, language barrier demonstrates a signi cant impact on MoCA score, the difference is most signi cant in the ≥ 85 years group (Table 4). It is unclear whether participants with a hearing or visual impairment, or other communication disorders perform less well in the MoCA, yet not representing their real cognitive functions. We may have to consider provide more choices in the language versions of MoCA, including sign language and Braille assistance, to offset any potential confounding effect. Two participants of interest (screened at their own requests) were not included in this study. The rst participant was a 29-year-old gentleman who had brain trauma and scored 18 in the MoCA. The second participant was a 55-year-old gentleman with strong family history of dementia scored 21 in the MoCA. They both fell in the mild cognitive impairment category at a very young age. It is unclear if prior brain injury and positive family history could affect the cognitive functions. The potential impact from these factors could also be investigated in future study.

Limitations
The MoCA and other cognitive tests may be susceptible to repeat practice effects and should be taken into consideration when repeatedly employing the cognitive tests in older adults 25 . In Ontario, individuals 80 years and older must complete the drawing of a clock as part of the license renewal process. It is unclear whether there are positive practice effects that could potentially improve the performance when these individuals subsequently take any cognitive tests.
Various risk factors have been reported to be associated with an increased risk of dementia, including school district and nancial status 26 , anticholinergic drugs 27 , and exposure to certain air particulates such as PM2.5 and ozone (O 3 ) 28 . Our study did not investigate these potential factors.

Conclusion
More than 30 percent of the participants in our study have certain level of cognitive impairment using the MoCA as a simple screening test. The overall higher percentage of MCI compared to other studies is unexpected. Most of the participants with cognitive dysfunction had mild cognitive impairment, which may not be easily identi ed in a routine primary care encounter. It is important to raise awareness so that similar screening tests can be applied in all primary care settings to early identify patients with cognitive impairment. For clinics with high turn-over and unable to conduct a standard MoCA, other screening tests such as the Mini-Cog may be an alternative for rapid screening.
In our study, we found that participants older than 85 years and with education less than 12 years scored lower in the MoCA when the test version is not concordant with their rst language. These factors impact both screening process and compliance with treatment. With these risk factors in mind, physicians can be more vigilant about early screening and recognition of potential cognitive impairment. Ultimately, we hope that screening for cognitive impairment in appropriate patients will become routine practice in primary care settings. Cognitive function test is a part of standard care in routine assessment of certain patients. No additional or unnecessary testing was involved in this study. Informed consent was obtained from all participants. The data was obtained by a retrospective review of patient charts in the electronic medical record. There has been no inappropriate disclosure of patient information and no breach of con dentiality in any way.

List Of Abbreviations
At no time was there any risk of harm to the patients in terms of physical injury, adverse drug effects or interactions, or mental anguish, nor of unauthorized disclosure of information.
The research relies on secondary use of anonymous information, and the process of data linkage, analysis, or dissemination of results does not generate identi able information. REB approval is not required according to the national regulations, Article 2.4, https://ethics.gc.ca/eng/tcps2-eptc2_2018_chapter2-chapitre2.html#a.

Consent for publication
Not applicable.
Availability of data and materials