The Association between Education and Cognitive Performance Varies at Different Levels of Cognitive Performance: A Quantile Regression Approach.

INTRODUCTION: Educational differences in cognitive performance among older adults are well documented. Studies that explore this association typically estimate a single average effect of education on cognitive performance. We argue that the processes that contribute to the association between education and cognitive performance are unlikely to have equal effects at all levels of cognitive performance. In this study, we employ an analytical approach that enables us to go beyond averages to examine the association between education and five measures of global and domain-specific cognitive performance across the outcome distributions. METHODS: This cross-sectional study included 1,780 older adults aged 58-68 years from the Longitudinal Aging Study Amsterdam. Conditional quantile regression was used to examine variation across the outcome distribution. Cognitive outcomes included Mini-Mental State Examination (MMSE) score, crystallized intelligence, information processing speed, episodic memory, and a composite score of global cognitive performance. RESULTS: The results showed that the associations between education and different cognitive measures varied across the outcome distributions. Specifically, we found that education had a stronger association with crystallized intelligence, MMSE, and a composite cognitive performance measure in the lower tail of performance distributions. The associations between education and information processing speed and episodic memory were uniform across the outcome distributions. CONCLUSION: Larger associations between education and some domains of cognitive performance in the lower tail of the performance distributions imply that inequalities are primarily generated among individuals with lower performance rather than among average and high performers. Additionally, the varying associations across some of the outcome distributions indicate that estimating a single average effect through standard regression methods may overlook variations in cognitive performance between educational groups. Future studies should consider heterogeneity across the outcome distribution.

Validated tests for language research with university students whose native language is English: Tests of vocabulary, general knowledge, author recognition, and reading comprehension.

We present five studies aimed at developing an L1 vocabulary test for English-speaking university students. Such a test is useful as an indicator of crystallized intelligence and because vocabulary size correlates well with reading comprehension. In the first study, we tested 100 written words with four answer alternatives, based on Nation's Vocabulary Size Test. Analysis suggested two factors, which we interpreted as the possible existence of two types of difficult words: unknown words for general knowledge and unknown words for specialized knowledge. In Study 2, we attempted to develop a vocabulary test for each type of word, and these tests were then validated in Study 3. Since the test for general words proved too easy for the target population, we improved it in a fourth study by creating and testing more difficult items. Finally, a fifth study was conducted to validate the new test. Unexpectedly, Study 5 found a high correlation (r = .82) between the general knowledge vocabulary test and the specialized knowledge vocabulary test, suggesting that they measure the same latent factor, contrary to our initial assumption. Both tests have high reliability (r > .85) and correlate well (r > .4) with general knowledge, author recognition, and reading comprehension. In addition, a collection of other language tests was used and improved to verify the validity of the vocabulary tests. An exploratory factor analysis of all tests identified three factors (text comprehension, crystallized intelligence, and reading speed), with the vocabulary tests loading on the factor crystallized intelligence, which in turn correlates with reading comprehension. Structural equation modeling confirmed the interpretation.

Link between fluid/crystallized intelligence and global/local visual abilities across adulthood.

Human visual processing involves the extraction of both global and local information from a visual stimulus. Such processing may be related to cognitive abilities, which is likely going to change over time as we age. We aimed to investigate the impact of healthy aging on the association between visual global vs local processing and intelligence. In this context, we collected behavioral data during a visual search task in 103 adults (50 younger/53 older). We extracted three metrics reflecting global advantage (faster global than local processing), and visual interference in detecting either local or global features (based on interfering visual distractors). We found that older, but not younger, adults with higher levels of fluid and crystallized intelligence showed stronger signs of global advantage and interference effects during local processing, respectively. The present findings also provide promising clues regarding how participants consider and process their visual world in healthy aging.

Crystallized and fluid intelligence are predicted by microstructure of specific white-matter tracts.

Studies of the neural basis of intelligence have focused on comparing brain imaging variables with global scales instead of the cognitive domains integrating these scales or quotients. Here, the relation between mean tract-based fractional anisotropy (mTBFA) and intelligence indices was explored. Deterministic tractography was performed using a regions of interest approach for 10 white-matter fascicles along which the mTBFA was calculated. The study sample included 83 healthy individuals from the second wave of the Cuban Human Brain Mapping Project, whose WAIS-III intelligence quotients and indices were obtained. Inspired by the "Watershed model" of intelligence, we employed a regularized hierarchical Multiple Indicator, Multiple Causes model (MIMIC), to assess the association of mTBFA with intelligence scores, as mediated by latent variables summarizing the indices. Regularized MIMIC, used due to the limited sample size, selected relevant mTBFA by means of an elastic net penalty and achieved good fits to the data. Two latent variables were necessary to describe the indices: Fluid intelligence (Perceptual Organization and Processing Speed indices) and Crystallized Intelligence (Verbal Comprehension and Working Memory indices). Regularized MIMIC revealed effects of the forceps minor tract on crystallized intelligence and of the superior longitudinal fasciculus on fluid intelligence. The model also detected the significant effect of age on both latent variables.

Cognitive reserve attenuates age-related cognitive decline in the context of putatively accelerated brain ageing in schizophrenia-spectrum disorders.

BACKGROUND: In schizophrenia, relative stability in the magnitude of cognitive deficits across age and illness duration is inconsistent with the evidence of accelerated deterioration in brain regions known to support these functions. These discrepant brain-cognition outcomes may be explained by variability in cognitive reserve (CR), which in neurological disorders has been shown to buffer against brain pathology and minimize its impact on cognitive or clinical indicators of illness. METHODS: Age-related change in fluid reasoning, working memory and frontal brain volume, area and thickness were mapped using regression analysis in 214 individuals with schizophrenia or schizoaffective disorder and 168 healthy controls. In patients, these changes were modelled as a function of CR. RESULTS: Patients showed exaggerated age-related decline in brain structure, but not fluid reasoning compared to controls. In the patient group, no moderation of age-related brain structural change by CR was evident. However, age-related cognitive change was moderated by CR, such that only patients with low CR showed evidence of exaggerated fluid reasoning decline that paralleled the exaggerated age-related deterioration of underpinning brain structures seen in all patients. CONCLUSIONS: In schizophrenia-spectrum illness, CR may negate ageing effects on fluid reasoning by buffering against pathologically exaggerated structural brain deterioration through some form of compensation. CR may represent an important modifier that could explain inconsistencies in brain structure - cognition outcomes in the extant literature.

Cognitive Predictors of Kindergarten Achievement in African American Children.

Three aspects of cognition (fluid intelligence, executive functioning, and crystallized intelligence) in pre-K were examined as predictors of math and reading achievement in kindergarten among an economically diverse sample of 198 African American children. From a variable-centered perspective, confirmatory factor analysis revealed that the three aspects of cognition can be distinguished. Subsequent regression analyses indicated that only executive functioning and crystallized intelligence predicted math and reading achievement in kindergarten. From a person-centered perspective, three profiles of cognition were identified: low fluid and crystallized intelligence with average executive functioning, average abilities in all three areas, and high abilities in all three areas, but particularly higher in executive functioning. Children with low fluid and crystallized intelligence during pre-K had the lowest math and reading skills in kindergarten, whereas children with the highest cognitive skills had the highest math and reading skills in kindergarten. Together, the variable-centered and person-centered results suggest that perhaps there should be increased focus on crystallized intelligence in early education programs, policies, and interventions in addition to a focus on executive functioning.

Wechsler Adult Intelligence Scale-III profile in the early stages of Alzheimer's disease: performance in subtests sensitive to and resistant to normal decline with ageing.

AIM: This study examined the significance of age-related subtest scores from the Japanese version of the Wechsler Adult Intelligence Scale-III in patients in the early stages of Alzheimer's disease (AD). METHODS: The subjects of this study included 58 elderly Japanese persons classified into two groups: AD group (n = 29) and control group (n = 29). These groups did not differ in age, years of education, gender ratio, Hasegawa's Dementia Scale-Revised score, or Full-Scale IQ score. No subject scored below the cut-off point on Hasegawa's Dementia Scale-Revised, a frequently used dementia screen test in Japan. RESULTS: At the index score level, General Ability Index scores were the only scores that differed significantly between the groups, with the AD group scoring significantly lower than the control group (P < 0.05, Hedges' g = 0.54). At the subtest level, information scores were the only scores that differed significantly between the groups, with the AD group significantly lower than the control group (P < 0.01, Hedges' g = 0.74). CONCLUSION: The General Ability Index is a composite score that deducts components of working memory and processing speed, which are sensitive to decline with normal ageing, from the Full-Scale IQ. It also served as a subtest measuring crystallized intelligence, especially of acquired knowledge of general and factual information. Therefore, the results of this study seem to suggest that Wechsler Adult Intelligence Scale-III profile of very early AD may be characterized by weak performance on subtests normally resistant to decline with ageing.

Subcortical regional morphology correlates with fluid and spatial intelligence.

Neuroimaging studies have revealed associations between intelligence and brain morphology. However, researchers have focused primarily on the anatomical features of the cerebral cortex, whereas subcortical structures, such as the basal ganglia (BG), have often been neglected despite extensive functional evidence on their relation with higher-order cognition. Here we performed shape analyses to understand how individual differences in BG local morphology account for variability in cognitive performance. Structural MRI was acquired in 104 young adults (45 men, 59 women, mean age = 19.83, SD = 1.64), and the outer surface of striatal structures (caudate, nucleus accumbens, and putamen), globus pallidus, and thalamus was estimated for each subject and hemisphere. Further, nine cognitive tests were used to measure fluid (Gf), crystallized (Gc), and spatial intelligence (Gv). Latent scores for these factors were computed by means of confirmatory factor analysis and regressed vertex-wise against subcortical shape (local displacements of vertex position), controlling for age, sex, and adjusted for brain size. Significant results (FDR < 5%) were found for Gf and Gv, but not Gc, for the right striatal structures and thalamus. The main results show a relative enlargement of the rostral putamen, which is functionally connected to the right dorsolateral prefrontal cortex and other intelligence-related prefrontal areas.

The influence of fluid and crystallized intelligence on the development of knowledge and skills.

BACKGROUND: Cattell's Gf-Gc distinction is quite generally recognized. However, the developmental part of the Gf-Gc theory of intelligence has not gained the same recognition. Results are inconsistent, but recent discussions emphasize the importance of homogeneity of samples with regard to education and language when investigating the developmental Investment theory. AIMS: The aim of this study was to investigate the influence of Gf and Gc on the development of knowledge and skills in a sample of children in compulsory school who are homogenous with regard to level of education, age, and cultural background. SAMPLE: Totally, 9,002 individuals from the evaluation through follow-up database born in 1972 and who left compulsory school in 1988 were included. These individuals were followed up in Grades 3, 6, and 9. METHODS: Structural equation modelling was used, and autoregressive path models were fitted. All modelling was performed using Mplus version 6.1. RESULTS: In the first step, a path model with a simplex structure was defined. However, a second model with direct relations of Gf on Gc in Grades 6 and 9 had better model fit, suggesting a continuous influence of Gf on Gc. However, no direct influence of Gf was found for the subject grades. CONCLUSION: Due to the continuous influence of Gf on the measures of Gc throughout compulsory school, support for Cattell's (1987) Investment theory was found.

Interaction between catechol-O-methyltransferase Val/Met polymorphism and cognitive reserve for negative symptoms in schizophrenia.

BACKGROUND: Cognitive reserve (CR) and the catechol-O-methyltransferase (COMT) Val/Met polymorphism are reportedly linked to negative symptoms in schizophrenia. However, the regulatory effect of the COMT genotype on the relationship between CR and negative symptoms is still unexamined. AIM: To investigate whether the relationship between CR and negative symptoms could be regulated by the COMT Val/Met polymorphism. METHODS: In a cross-sectional study, 54 clinically stable patients with schizophrenia underwent assessments for the COMT genotype, CR, and negative symptoms. CR was estimated using scores in the information and similarities subtests of a short form of the Chinese version of the Wechsler Adult Intelligence Scale. RESULTS: COMT Met-carriers exhibited fewer negative symptoms than Val homozygotes. In the total sample, significant negative correlations were found between negative symptoms and information, similarities. Associations between information, similarities and negative symptoms were observed in Val homozygotes only, with information and similarities showing interaction effects with the COMT genotype in relation to negative symptoms (information, ß = -0.282, 95%CI: -0.552 to -0.011, P = 0.042; similarities, ß = -0.250, 95%CI: -0.495 to -0.004, P = 0.046). CONCLUSION: This study provides initial evidence that the association between negative symptoms and CR is under the regulation of the COMT genotype in schizophrenia.

Regional brain structure mediates the association between sleep quality and intellectual abilities: the moderating role of socioeconomic status.

This study aims to reveal the association between sleep quality and crystallized intelligence (Gc), fluid intelligence (Gf), and the underlying brain structural basis. Using the data from the Human Connectome Project (N = 1087), we performed mediation analysis to explore whether regional brain structure related to sleep quality mediate the association between sleep quality and intellectual abilities, and further examined whether socioeconomic status (i.e., income and education level) moderate the mediation effect. Results showed that poorer sleep quality was associated with lower Gc rather than Gf, and worse sleep quality was associated with smaller volume and surface area in temporal lobe, including inferior temporal gyrus and middle temporal gyrus. Notably, temporal lobe structures mediated the association between sleep quality and Gc rather than Gf. Furthermore, socioeconomic status (i.e., income and education level) moderated the mediating effect, showing low socioeconomic status has a more significant mediating effect with stronger association between sleep quality and Gc as well as stronger association between temporal lobe structure and Gc in low socioeconomic status group. These findings suggest that individuals with higher socioeconomic status are less susceptible to the effect of sleep quality on Gc.

Chronotype, Time of Day, and Performance on Intelligence Tests in the School Setting.

Research suggests the existence of an association between chronotype and intellectual performance, but the nature of this link remains unclear. Studies conducted in a laboratory setting point to the synchrony effect (better performance at a person's preferred time of day) for fluid intelligence, but not for crystallized intelligence, whereas studies that have analyzed students' grades suggest that the effect exists for both. In the present study, we aimed to verify the synchrony effect by applying direct measures of crystallized intelligence, fluid intelligence, and subjective sleepiness-alertness in a sample of high school students during their morning or afternoon class. The results revealed a synchrony effect for crystallized, but not for fluid intelligence. During morning class, students with a morning chronotype performed better than evening chronotypes on a test of crystallized intelligence, whereas during afternoon class there was no difference between chronotypes. The association resulted from decreased performance during morning class in evening chronotypes that improved during afternoon class and constant performance in morning chronotypes. These effects were independent of sleepiness-alertness levels. The results suggest that individual differences between chronotypes may be important for tasks performed during morning classes, but not during afternoon ones, and that performance across school days may depend on time of day in evening chronotypes.

Effects of sleep-corrected social jetlag on measures of mental health, cognitive ability, and brain functional connectivity in early adolescence.

Approximately half of adolescents encounter a mismatch between their sleep patterns on school days and free days, also referred to as "social jetlag." This condition has been linked to various adverse outcomes, such as poor sleep, cognitive deficits, and mental disorders. However, prior research was unsuccessful in accounting for other variables that are correlated with social jetlag, including sleep duration and quality. To address this limitation, we applied a propensity score matching method on a sample of 6335 11-12-year-olds from the 2-year follow-up (FL2) data of the Adolescent Brain Cognitive Development study. We identified 2424 pairs of participants with high sleep-corrected social jetlag (SJLsc, over 1 hour) and low SJLsc (<= 1 hour) at FL2 (1728 pairs have neuroimaging data), as well as 1626 pairs at 3-year follow-up (FL3), after matching based on 11 covariates including socioeconomic status, demographics, and sleep duration and quality. Our results showed that high SJLsc, as measured by the Munich Chronotype Questionnaire, was linked to reduced crystallized intelligence (CI), lower school performance-grades, and decreased functional connectivity between cortical networks and subcortical regions, specifically between cingulo-opercular network and right hippocampus. Further mediation and longitudinal mediation analyses revealed that this connection mediated the associations between SJLsc and CI at FL2, and between SJLsc and grades at both FL2 and FL3. We validated these findings by replicating these results using objective SJLsc measurements obtained via Fitbit watches. Overall, our study highlights the negative association between social jetlag and CI during early adolescence.

The effect of sleep-corrected social jetlag on crystalized intelligence, school performance, and functional connectome in early adolescence.

Approximately half of adolescents encounter a mismatch between their sleep patterns on school days and free days, also referred to as "social jetlag". This condition has been linked to various adverse outcomes, such as poor sleep, cognitive deficits, and mental disorders. However, prior research was unsuccessful in accounting for other variables that are correlated with social jetlag, including sleep duration and quality. To address this limitation, we applied a propensity score matching method on a sample of 8853 11-12-year-olds from the two-year follow-up (FL2) data of the Adolescent Brain Cognitive Development (ABCD) study. We identified 3366 pairs of participants with high sleep-corrected social jetlag (SJLsc, over 1 hour) and low SJLsc (<= 1 hour) at FL2, as well as 1277 pairs at three-year follow-up (FL3), after matching based on 11 covariates including socioeconomic status, demographics, and sleep duration and quality. Our results showed that high SJLsc, as measured by the Munich Chronotype Questionnaire, was linked to reduced crystallized intelligence, lower school performance - grades, and decreased functional connectivity between cortical networks and subcortical regions, specifically between cingulo-opercular network and right hippocampus (cerc-hprh). Further mediation and longitudinal mediation analyses revealed that cerc-hprh connection mediated the associations between SJLsc and crystallized intelligence at FL2, and between SJLsc and grades at both FL2 and FL3. We validated these findings by replicating these results using objective SJLsc measurements obtained via Fitbit watches. Overall, our study highlights the negative association between social jetlag and crystallized intelligence during early adolescence.

Two sides of the same coin: distinct neuroanatomical patterns predict crystallized and fluid intelligence in adults.

Background: Crystallized intelligence (Gc) and fluid intelligence (Gf) are regarded as distinct intelligence components that statistically correlate with each other. However, the distinct neuroanatomical signatures of Gc and Gf in adults remain contentious. Methods: Machine learning cross-validated elastic net regression models were performed on the Human Connectome Project Young Adult dataset (N = 1089) to characterize the neuroanatomical patterns of structural magnetic resonance imaging variables that are associated with Gc and Gf. The observed relationships were further examined by linear mixed-effects models. Finally, intraclass correlations were computed to examine the similarity of the neuroanatomical correlates between Gc and Gf. Results: The results revealed distinct multi-region neuroanatomical patterns predicted Gc and Gf, respectively, which were robust in a held-out test set (R2 = 2.40, 1.97%, respectively). The relationship of these regions with Gc and Gf was further supported by the univariate linear mixed effects models. Besides that, Gc and Gf displayed poor neuroanatomical similarity. Conclusion: These findings provided evidence that distinct machine learning-derived neuroanatomical patterns could predict Gc and Gf in healthy adults, highlighting differential neuroanatomical signatures of different aspects of intelligence.

The Role of Cognition in Divergent Thinking: Implications for Successful Aging.

Promoting active and successful aging has become crucial to improve quality of life in later adulthood and reduce the impact of cognitive decline. Increasing evidence suggested that the ability to think creatively (e.g., via divergent thinking), similar to cognitive reserve, could represent a beneficial factor against the negative effects of aging. However, there is still little evidence investigating the relationships between divergent thinking, cognitive functions, and cognitive reserve in late adulthood. The present study explored these relationships in a sample of 98 individuals ranging from 61 to 88 years old (mean age: 72.44 ± 6.35). Results showed that visual, but not verbal, divergent thinking was affected by aging. Interestingly, visual divergent thinking performance was predicted by both the cognitive component of crystallized intelligence and cognitive reserve. Only the crystallized component of intelligence was found to mediate the aging effect on visual divergent thinking performance. These results suggest that in later adulthood a potential shift strategy to prior knowledge and semantic components over executive and control components of cognition could underlie a preserved ability to think divergently and, plausibly, creatively. Limitations of the study and implications for successful aging are discussed.

Environmental Harshness, Life History, and Crystallized Intelligence of Chinese Adolescents.

The present study examined longitudinal relations between adverse environment (at the community and family level), life history (LH) profile (conceptualized as a suite of behavioral and physical traits with converging adaptive functions), and crystalized intelligence (mathematics and vocabulary test scores) using data on 1,185 Chinese adolescents obtained from the China Family Panel Studies survey. Multilevel structural equation modeling indicates that early familial environmental harshness was negatively associated with slow LH profiles and crystalized intelligence, slow LH profiles were positively associated with crystallized intelligence, and early community-level environmental harshness strengthened the positive association between slow LH and crystalized cognitive abilities. The results underscore the importance of the childhood environment in fostering individual LH and cognitive development.

Stop Worrying about Multiple-Choice: Fact Knowledge Does Not Change with Response Format.

Declarative fact knowledge is a key component of crystallized intelligence. It is typically measured with multiple-choice (MC) items. Other response formats, such as open-ended formats are less frequently used, although these formats might be superior for measuring crystallized intelligence. Whereas MC formats presumably only require recognizing the correct response to a question, open-ended formats supposedly require cognitive processes such as searching for, retrieving, and actively deciding on a response from long-term memory. If the methods of inquiry alter the cognitive processes involved, mean-changes between methods for assessing declarative knowledge should come along with changes in the covariance structure. We tested these assumptions in two online studies administering declarative knowledge items in different response formats (MC, open-ended, and open-ended with cues). Item difficulty clearly increases in the open-ended methods although effects in logistic regression models vary slightly across items. Importantly, latent variable analyses suggest that the method of inquiry does not affect what is measured with different response formats. These findings clearly endorse the position that crystallized intelligence does not change as a function of the response format.

Discrepancy in Fluid and Crystallized Intelligence: An Early Cognitive Marker of Dementia from the LASI-DAD Cohort.

Background: Cognitive aging is a complex phenomenon, which comprises various cognitive skills, broadly categorized into fluid and crystallized intelligence. Crystallized intelligence (gc) tends to be maintained, as opposed to fluid intelligence (gf), which tends to decline rapidly with age. The association of the two with cognitive decline remains a matter of conjecture requiring further research. Aim: The aim of the study was to identify the variables of gc and gf from a population data of Longitudinal Aging Study in India-Diagnostic Assessment of Dementia (LASI-DAD) study and investigate its relationship with the onset of cognitive impairment using discrepancy analysis against neuropsychological tests. Methods: This analysis of data from LASI-DAD study was carried out on a sample of 3,223 participants. They were assessed on extensive thirteen cognitive tests and one subjective test of cognition. Standardized score was used for discrepancy analysis. Fluid ability minus crystallized ability was used to assess the cognitive impairment. Any statistical significance with the score difference >0.99 SD was defined as a presence of cognitive decline. Hindi Mental Status Examination (HMSE) and the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) were used as gold standard. Results: With increased discrepancy score, each cognitive parameter score declined which was found to be statistically significant. In HMSE (Normal = 25.81 ± 3.39; Impaired = 23.17 ± 3.54; p = <0.001), there was a drop of 2 point scores in identifying cognitive impairment in the population sample as per the gold standard. A similar trend was evident in other neurocognitive domains as well. Conclusion: Crystallized-fluid intelligence discrepancy analysis has a strong potential in predicting the onset of cognitive decline ahead of time, facilitating early intervention.

Does Socio-Economic Status Have Different Impact on Fluid and Crystallized Abilities? Comparing Scores on Raven's Progressive Matrices, Kaufman Assessment Battery for Children II Story Completion and Kilifi Naming Test Among Children in Ghana.

Current literature shows an association between intelligence and socio-cultural or socio-economic factors. The available evidence supports a stronger effect of exogenous factors on measures of crystalized intelligence than on fluid intelligence. Despite this, the sources of variability in fluid and crystalized intelligence have not been explored adequately in intelligence research. The purpose of this study was to compare performance on tests that measure fluid and crystallized intelligence among children selected from public and private schools in Ghana. We tested the assumption that socio-economic status (SES) will have a stronger effect on tests that measure crystallized intelligence than on fluid intelligence. We selected 185 children between 6 and 12 years from private and public schools, and used inclusion in a private or public school as a proxy for SES. We administered the Raven's Coloured Progressive Matrices (RCPM), a fluid intelligence test, the KABC II story completion subtest as a measure of inductive reasoning and crystallized intelligence and the Kilifi Naming Test, a verbal ability measure designed to minimize the effect of school on vocabulary. The results showed age-related improvement in scores on all three tests with effect sizes ranging from 0.42 to 0.52. We also found significant effect for type of school on all the tests with effect sizes ranging from 0.37 to 0.66. The results also showed an increasing disparity in performance on the tests favoring children selected from private schools. These suggest that fluid and crystalized intelligence are affected by socioeconomic factors. The results also showed that SES factors tend to affect crystallized ability more than it affects fluid ability. The results are discussed in the context of differences in socioeconomic resources available to children such as quality of education in low- and middle-income countries.