Influence of multimorbidity on cognition in a normal aging population: a 12-year follow-up in the Maastricht Aging Study
Abstract
Objective
The prevalence of multimorbidity has risen considerably because of the increase in longevity and the rapidly growing number of older individuals. Today, only little is known about the influence of multimorbidity on cognition in a normal healthy aging population. The primary aim of the present study was to investigate the effect of multimorbidity on cognition over a 12-year period in an adult population with a large age range.
Methods
Data were collected as part of the Maastricht Aging Study (MAAS), a prospective study into the determinants of cognitive aging. Eligible MAAS participants (N = 1763), 24–81 years older, were recruited from the Registration Network Family Practices (RNH) which enabled the use of medical records. The association between 96 chronic diseases, grouped into 23 disease clusters, and cognition on baseline, at 6 and 12 years of follow-up, were analyzed. Cognitive performance was measured in two main domains: verbal memory and psychomotor speed. A multilevel statistical analysis, a method that respects the hierarchical data structure, was used.
Results
Multiple disease clusters were associated with cognition during a 12-year follow-up period in a healthy adult population. The disease combination malignancies and movement disorders multimorbidity also appeared to significantly affect cognition.
Conclusions
The current results indicate that a variety of medical conditions adversely affects cognition. However, these effects appear to be small in a normal healthy aging population. Copyright © 2010 John Wiley & Sons, Ltd.
Introduction
In today's society, multimorbidity, the occurrence of multiple medical conditions within one person (van den Akker et al., 2001), has risen considerably because of the increase in longevity and the rapidly growing number of older individuals. In contrast to comorbidity, which is defined as multiple medical conditions additional to an index condition, multimorbidity views all medical conditions from an overall perspective. Prevalences of multimorbidity ranging from 30% (van den Akker et al., 1998) to over 68% (Fortin et al., 2005) of the population under study have been reported. Multimorbidity is associated with several negative health outcomes such as activity limitations (Perruccio et al., 2007), decreased health related quality of life (Fortin et al., 2006), and impaired cognitive functioning (Bergman et al., 2007). Several studies have suggested that impaired health status may enhance cognitive decline in normal and pathological cognitive aging (Vermeer et al., 2003; Bergman et al., 2007).
Various diseases, such as diabetes mellitus type I and II (van Boxtel et al., 1998), stroke (Dik et al., 2000), and peripheral artherosclerosis (Comijs et al., 2009), have already been related to a decrease in cognitive performance. Several medical conditions, such as brain infarcts, are also known to enhance the development of dementia (Vermeer et al., 2003).
Several studies focused on the relationship between a specific medical condition and cognition, which most often implies the exclusion of persons with comorbid diseases (Comijs et al., 2009). Since research suggests that up to 68% of persons in a hospital setting suffer from multimorbidity (Patrick et al., 2002), studies limited to persons with only one single disease may therefore be less representative. Moreover, in a geriatric population cumulative illness is associated with more enhanced cognitive decline compared to older with a single disease (Patrick et al., 2002).
The majority of studies that focused on the relationship between medical conditions and cognition have been cross-sectional (van Boxtel et al., 1998; Patrick et al., 2002). Since cognitive decrements due to health status are likely to develop over the course of many years, the underlying process should be investigated longitudinally. In most cases, longitudinal studies included only a small number of participants (Bergman et al., 2007) or were limited to specific population samples such as older persons (Comijs et al., 2009) or patients admitted for clinical rehabilitation (Patrick et al., 2002). However, it is not ruled out that being diagnosed with one or more medical conditions could also affect cognition in younger people. Younger people could be preoccupied with their impaired health status which could result in worsening cognitive performance. Hence, it is important to evaluate the effect of multimorbidity on cognition in a normal aging adult population.
Research reporting on the association between morbidity and cognition was mainly restricted to evaluate only a limited number of chronic medical conditions. In a longitudinal study with 6 years of follow-up, the influence of seven somatic chronic diseases on changes in cognitive functioning was studied in persons 62 to 85 years older (Comijs et al., 2009). However, in order to specify the differential effects of various medical conditions on cognitive functioning, it is necessary to evaluate a greater variety of medical conditions. Furthermore, most studies were based on self-reported morbidity status (Comijs et al., 2009). As the agreement between self-reports and medical records tends to weaken in older age groups (Goebeler et al., 2007), this is highly problematic when studying the effects of multimorbidity in older persons. In patients suffering from multiple medical conditions, the agreement between self-reports and medical records also tends to be lower (Kriegsman et al., 1996). Hence, medical records from general practitioners (GPs) may provide more valid data on morbidity, especially in countries such as the Netherlands, where GPs are the gatekeepers to other health facilities.
The aim of the present study was to evaluate the effect of morbidity and multimorbidity on cognitive functioning, measured by established neurocognitive tests, in an adult population with a wide age range. Results could indicate which specific medical conditions (i.e., morbidity) and which specific combinations of two medical conditions (i.e., multimorbidity) are associated with cognitive performance.
Methods
The present study was conducted on data gathered in two longitudinal projects: the Registration Network Family Practices (RNH) and the Maastricht Aging Study (MAAS). MAAS is a prospective study of determinants of normal cognitive aging (Jolles et al., 1995). Participants of MAAS were recruited from the RNH (RegistratieNetwerk Huisartspraktijken, RNH), which contains demographic and health information of over 120 000 patients inhabited in the South of Limburg, the Netherlands (Metsemakers et al., 1992).
Maastricht Aging Study (MAAS)
In 1992, 10 396 people were recruited from the RNH database of whom 4490 were willing to participate (43.2%), 3531 refused participation (34%), and 2375 did not respond (22.8%). Since MAAS' primary aim was to investigate normal aging, participants were excluded from baseline participation when medical records from the RNH database contained medical conditions known to interfere with cognition: history of coma, cerebrovasculair disorder, tumor of the nervous system, congenital malformation of the nervous system, multiple sclerosis, Parkinsonism, epilepsy, dementia, organic psychosis, schizophrenia, affective psychosis, or mental retardation. The 4490 eligible participants were screened by telephone for medical conditions not documented in the RNH database (history of transient ischemic attack, brain surgery, hemodialysis for renal failure, electroconvulsive therapy, or daily psychotropic drug use). Of the remaining 4189 participants (301 were excluded based on the telephone screening), 1823 (24 to 81 years older at baseline) were randomly selected, stratified for age (12 discrete groups between 24 and 81 years of age at baseline), sex, and occupational achievement (2 levels).
Registration Network Family Practices (RNH)
The morbidity status of all MAAS participants at baseline, 6- and 12-year follow-up was retrieved from the RNH database. The RNH is a continuously updated database, which contains the medical records of patients from 21 family practices in the south of the Netherlands. It includes all relevant current and past health problems. A health problem was defined as ‘anything that has required, does or may require health care management and has affected or could significantly affect a person's physical or emotional well-being (Metsemakers et al., 1996). The RNH database contains health problems which are only coded by the GP when they are permanent (no recovery expected), chronic (duration longer than 6 months), or recurrent (more than three recurrences within 6 months), or when they have lasting consequences for the functional status or prognosis of the patient. Problems are coded in a standardized fashion, according to the International Classification of Primary Care (ICPC) (WONCA, 1987), using the criteria of the International Classification of Health in Primary Care (ICHPPC-2) and other more current guidelines of the Dutch College of GPs. A diagnosis is made by the GP or by a medical specialist to whom a patient may be referred. Membership of the RNH registry ends by migration or death. All patients included in the RNH database have been informed about the potential anonymous use of their health information. If a patient does not agree, the inclusion of this patient in the RNH database is stopped. Sociodemographical characteristics of samples in other studies, which made use of the RNH database, have been shown to be comparable to the Dutch population (Metsemakers et al., 1996).
Outcome measures
Two representative neuropsychological tests were used to assess the cognitive domains of memory and processing speed at all three measurements.
Visual verbal learning test
The visual verbal learning test (VVLT) is the Dutch version of the Rey VVLT, which measures the ability to learn new verbal information and retrieve information from memory (van der Elst et al., 2005). Fifteen monosyllabic words were visually presented one at a time on five consecutive trials. Participants were asked to recall as many words as possible, with no restriction concerning the order of recall (immediate recall). After a 20 minutes delay, participants were again asked to recall as many correct words as possible (delayed recall). The latter score (ranging from 0 to 15) was used as outcome measure. Parallel versions of the memory tasks were used for each assessment.
Letter digit substitution task
The letter digit substitution task (LDST) is used to measure complex information processing under time pressure (Lezak et al., 2004). A sheet with letters, each matched to its own number (1–9) was presented to the participant. Participants were asked to match as much letters and numbers as possible in 90 s, by filling in the right numbers belonging to the right letters. The total number of correctly filled in numbers was used as outcome measure.
Independent variables
Based on either previously conducted studies or from clinical experience, 96 chronic medical conditions with high prevalence and/or potential to impact brain functioning were included in the analyses (see Appendix I). The medical conditions on which exclusion for MAAS participation was based were also included in order to monitor possible occurrences after baseline.
Beforehand, single morbidity codes were combined into clusters of medical conditions by three experts (MvdA, MvB, JM), including two medical doctors, to limit the number of independent variables (see Appendix I). Some morbidity codes were not clustered because they were highly prevalent and/or because their effect on cognition has been well documented (diabetes mellitus, dementia, and Parkinsonism). Note that one person can belong to two or more clusters when diagnosed with multiple medical conditions. Moreover, a person can be diagnosed with more than one disease in one cluster.
Time in years since diagnosis was incorporated in order to take disease duration into account. If a person was diagnosed with more than one disease within one cluster, the cluster was coded with the first date of a diagnosis. The reference group consisted of all participants without any of the 96 diseases. All medical conditions of all MAAS participants were assessed by looking at the medical records at baseline, at 6- and 12-year follow-up.
Potential confounders/covariates
Potential confounders were taken into account: gender, age, education level measured on a 8-point ordinal scale (ranging from primary education to university degree) (de Bie, 1987), smoking behavior at baseline (yes/no), alcohol intake at baseline (number of alcoholic units per week), and living arrangements in three categories (living with family, living alone, living in a home for the older). Self-reported depression as assessed by the SCL-90 was inserted as a possible confounding factor (Derogatis and Cleary, 1977). Total morbidity score (in seven categories, ranging from no disease present to six or more diseases present) was defined as the total number of medical conditions per individual. Total cluster score (in seven categories, ranging from no cluster of diseases present to six or more clusters present) was defined as the total number of disease clusters in which an individual had at least one disease.
Statistical analyses
As the independent variables and the outcome measures were measured three times, data were analyzed with Linear Mixed Models. Since measurements within subjects of this study are more alike than measurements between subjects, there is correlation between the repeated measurements. This within-subjects variation that accounts for changes within each subject through repeated measurements in time and the between-subjects variation accounting for differences between the subjects performance can be modeled by a multilevel approach (Tan, 2008). Moreover, missing data may depend on the observed covariates and outcome measures (missing at random). As previously stated the dependent variables were measured three times. At similar times, the independent variables (i.e., medical conditions) were retrieved from the RNH database. However, since reporting the results of the 6-year follow-up will decrease the comprehension of the present study, only the results at 12-year follow-up will be reported.
Because serial correlation could be present between the repeated measurements, a marginal model with an unstructured covariance structure was used. When evaluating the plots of the residuals and when taking the maximum likelihood ratio and bayesian information criterion (BIC) score into account, this model provided the best fit. The multilevel models included fixed terms for follow-up (time of measurement), age, gender, education, smoking behavior, alcohol intake, living arrangements, total morbidity score, and total cluster score. Since the factors of total morbidity score and total cluster score were highly correlated (r = 0.95; p < 0.00), they were always tested in separate models. A covariate was included in the model if it significantly influenced the cognitive domain under study (r < 0.05). To limit the number of variables in a multilevel model, each disease cluster (and the three separate medical conditions) was evaluated separately using univariate analysis.
Those disease clusters that appeared highly prevalent and that had a significant effect on cognition in the univariate analyses were selected in order to evaluate their effect in a multivariate analysis using a backward method. All other clusters were excluded from all the following analyses.
Clusters that still appeared significant in the multivariate analysis were selected for analyzing the influence of multimorbidity on cognition. In order to determine whether specific combinations of disease clusters (i.e., multimorbidity) were associated with cognitive performance, longitudinal effects were tested using ‘disease cluster × disease cluster’ interactions.
All analyses were conducted with the SPSS statistical software package version 16.0 for Windows (SPSS Inc., Chicago, IL, USA). p-values of 0.05 or less were considered statistically significant.
Results
Descriptive characteristics of the study sample are reported in Table 1. The data set contained 4196 measurements among 1763 participants. The sample size differed between the three measurements which is due to dropout.
| Characteristics (baseline) | |
|---|---|
| Follow-up (no. of participants) | |
| Baseline | 1763 |
| 6 years | 1310 |
| 12 years | 1123 |
| Mean age in years (SD)a | 55.4 (15.9) |
| Age groups (%) | |
| Age 24–40 | 596 (33.8) |
| Age 41–60 | 567 (32.2) |
| Age 61–81 | 600 (34.0) |
| Gender (% females) | 886 (50.3) |
| Total morbidity score (%) | |
| 0 | 404 (22.9) |
| 1 | 374 (21.2) |
| ≥2 | 985 (55.9) |
| Total cluster score (%) | |
| 0 | 404 (22.9) |
| 1 | 427 (24.2) |
| ≥2 | 932 (52.9) |
| Days per week of alcohol intake (%)b | |
| 1 | 172 (9.8) |
| 2 | 78 (4.4) |
| 3 | 180 (10.2) |
| 4 | 543 (30.8) |
| 5–7 | 467 (26.5) |
| Educational levelc | |
| Low (%) | 650 (36.9) |
| Medium (%) | 712 (40.4) |
| High (%) | 399 (22.6) |
| Living arrangement | |
| Living with family (%) | 1490 (84.5) |
| Living alone (%) | 235 (13.3) |
| Home for the older/commune (%) | 8 (0.5) |
| Unknown/others | 30 (1.7) |
| SCL-90 Depressive symptom score (SD)d | 23.1 (8.6) |
| Smoking (% NO) | 1250 (70.9) |
- a Mean age on 15 June 1993.
- b For 323 (18.3%) people this information was missing.
- c For two people this information was missing; Low is defined as primary school and/or lower vocational education, medium as secondary school and/or medium level vocational education, high as higher vocational education or university degree.
- d Mean score on the SCL-90 subscale depressive complaints.
Higher education was associated with a significant better performance of memory (p < 0.001) and processing speed. Higher age negatively influenced memory and processing speed (p < 0.001) and women scored significantly higher than men in both cognitive domains (respectively, p < 0.001 for memory and p < 0.002 for processing speed). Depressive complaints were associated with impaired memory (p = 0.013) and speed (p < 0.001). The remaining confounders, including total morbidity score and total cluster score, appeared not to affect cognition significantly, and were therefore dropped from all analyses.
Multiple clusters appeared to be significantly related to cognition (Table 2). Seven clusters were selected in order to investigate their influence on cognition in the multivariate analysis (i.e., cerebrovasculair diseases, cardiac diseases, malignancies, movement disorders, asthma/COPD/ bronchitis, endocrine diseases, and diabetes mellitus). Since the medical conditions in the cluster ischemic diseases and in the cluster other cardiovascular diseases are rather comparable, these two clusters were regrouped into one cardiac disease cluster.
| Verbal memory | Processing speed | Na | |
|---|---|---|---|
| All malignancies | −0.047** | −0.341** | 135 |
| Peptic ulcers | −0.233 | −0.210** | 35 |
| Other gastro-intestinal diseases | −0.037* | −0.226** | 94 |
| Diseases of the eye | −0.067* | −0.330** | 34 |
| Diseases of the ear | −0.024 | −0.123* | 65 |
| Ischemic diseases | −0.061** | −0.370** | 130 |
| Pulmonary embolism and phlebitis | −0.036 | −0.246** | 43 |
| Cerebrovascular diseases | −0.074* | −0.467** | 66 |
| Arrhythmias & heart failure | −0.052** | −0.261** | 108 |
| Other cardiovascular disease | −0.054 | −0.227* | 96 |
| Movement disorders | −0.017 | −0.087** | 318 |
| Parkinsonism | −0.536* | −0.698 | 5 |
| Migraine & headache | 0.004 | −0.065 | 41 |
| Other diseases of nervous system | −0.021 | −0.078 | 12 |
| Mood disorders | −0.220 | −0.234** | 57 |
| Alzheimer | −6.220** | −7.021** | 5 |
| Other mental disorders | −0.005 | −0.144** | 53 |
| Asthma, COPD, and bronchitis | −0.020* | −0.078** | 114 |
| Other chronic respiratory diseases | −0.010 | −0.052 | 125 |
| Eczema, psoriasis chronic skin ulcer | −0.028** | −0.065 | 82 |
| Endocrine diseases | −0.018 | −0.160** | 169 |
| Diabetes mellitus (Type I and II) | −0.052** | −0.341** | 101 |
| Diseases of the urinary tract | −0.230 | −0.117* | 44 |
- * p ≤ 0.01;
- ** p ≤ 0.05.
- a Number of participants at 12-year follow-up.
In the multivariate analyses, only the cardiac cluster and malignancies were associated with a significant lower verbal memory performance (see Table 3). Hence, only one interaction, cardiac diseases and malignancies, was evaluated for the effect of multimorbidity on verbal memory in a univariate model. The interaction of these two disease clusters appeared not to be significantly associated with verbal memory.
| Verbal memory | Processing speed | Na | |
|---|---|---|---|
| Cerebrovascular diseases | NS | −0.262** | 66 |
| Cardiac diseases | −0.047** | −0.183** | 180 |
| Malignancies | −0.038** | −0.213** | 135 |
| Movement disorders | NS | −0.062* | 318 |
| Asthma/COPD/bronchitis | NS | NS | 114 |
| Endocrine diseases | NS | NS | 169 |
| Diabetes Mellitus | NS | NS | 101 |
- NS = not significant.
- * p ≤ 0.01;
- ** p ≤ 0.05.
- a Number of participants at 12-year follow-up.
Cerebrovascular diseases, cardiac diseases, malignancies, and movement disorders were all significantly related to a decrease in processing speed in the multivariate analysis (see Table 3). Therefore, these four disease clusters were selected in order to evaluate the effect of their interactions, and thus of multimorbidity. Since four disease clusters appeared to significantly affect processing speed in the multivariate model, six interaction effects were possible (see Table 4). These interactions were investigated in six separate univariate analyses. Only the combination of malignancies and movement disorders was related to a faster decline in processing speed.
| Verbal memory | Processing speed | Na | |
|---|---|---|---|
| Cardiac diseases* malignancies | NS | NS | 40 |
| Malignancies* movement disorders | NA | −0.012* | 53 |
| Malignancies* cerebrovascular diseases | NA | NS | 16 |
| Cerebrovascular diseases* movement disorders | NA | NS | 32 |
| Cerebrovascular diseases* cardiac diseases | NA | NS | 29 |
| Movement disorders* cardiac diseases | NA | NS | 84 |
- NA = non-applicable, NS = not significant.
- * p ≤ 0.05.
- a Number of participants at 12-year follow-up.
Discussion
This study was conducted to evaluate the differential longitudinal effects of 96 chronic diseases, grouped into clusters, on cognition in a healthy aging population. The present results indicate that specific disease clusters indeed were associated with a one or both tested domains of cognitive functioning during a 12-year follow-up period. Several disease clusters appear to be related to cognitive decline.
Since a wide variety of disease clusters appear to be related to cognitive functioning, there might be (unexpected) common pathogenetic mechanisms, such as genetics, inflammation, immunological mechanisms, or mood, underlying these associations. The present findings might indicate that an exclusive disease-specific approach is insufficient in explaining these associations. The current study may, therefore, provide a conceptual basis for moving away from organ and disease specific approaches toward a health-based, integrative approach.
Multimorbidity is common in general practice as evidenced through a prevalence of 55.2% in our study population. Investigating the influence of disease cluster interactions (i.e., disease cluster × disease cluster) on cognition showed that only the combination malignancies and movement disorders was significantly related to cognitive decline. Hence, persons suffering from a movement disorder as well as a malignancy perceive more cognitive decline on 12-year follow-up than persons who are suffering from only one of these two disease clusters. Multimorbidity appeared to be less related to cognition than expected based on the negative influence of multimorbidity on other health-related outcomes. The present results indicate that specific disease clusters have a much more profound effect on cognition than combinations of disease clusters (i.e., multimorbidity).
All presented results were controlled for age, sex, depressive complaints, and education which all appeared to be significantly related to cognitive functioning. In contrast to previous research, which suggested that the total number of morbid conditions was significantly related to memory, but not to speed of processing (Nguyen et al., 2007), no association between total morbidity score and cognition was found in the present study. This dissimilarity might be the result of the inclusion of as much as 96 medical conditions. Since so many medical conditions were included, not all conditions will have the same effect on cognition. For example, a person diagnosed with only one medical condition (e.g., dementia) can encounter much more cognitive decrements than a person diagnosed with multiple medical conditions. Total cluster score was also considered as a possible confounder. The number of disease clusters appeared not to be significantly related to memory or processing speed.
Since two neuropsychological tests were used as distinct outcome measures, it might be argued that compound scores (several raw test scores being grouped into one score) would improve the robustness of the underlying cognitive construct. For reasons of parsimony, the authors decided to, only use the abovementioned tests as a selection of all possible cognitive tests. Both tests are relatively robust, sensitive to age effects, and are among the most widely used tests in clinical practice as well as in cognitive research (Lezak et al., 2004).
The current study has several advantages over previous studies. First of all, this is the first study that incorporated 96 chronic medical conditions grouped into different disease clusters. By doing so, a more complete inventory of the relation between medical conditions and cognition could be made. Secondly, this study provided data with a follow-up of 12 years in a large population-based sample. Since cognitive decrements due to morbidity status are expected to develop in the course of many years, a realistic picture of the influence of morbidity status on cognitive decline emerges from the present study. Thirdly, we have evaluated the causal relationship between morbidity and cognition in a general practice based setting. Consequently, the results are more representative for the general population than results of studies conducted in smaller and less healthy homogeneous samples. Moreover, in contrast to previous studies, the current study assessed medical status by using a general practice database instead of using self-reported medical status. Fourthly, although several studies already considered the relationship between morbidity and cognition, most of them, in contrast to the present study, lacked data on disease duration.
Despite the previous mentioned strengths, our findings must be interpreted in light of some possible limitations. Firstly, since the present study evaluated the effects of 96 chronic medical conditions on cognition, clustering was needed in order to conduct statistical analyses. Clustering was based on the medical knowledge and consensus of three of the co-authors, which remains a rather subjective method for clustering. Furthermore, by evaluating the effects of disease clusters on cognition, a potential strong association between a single disease and cognitive function may become diluted when mixed with the effects of other diseases within a cluster. However, medical conditions that are known for their profound effect on cognition were grouped into separate clusters.
Secondly, defining multimorbidity by a simple disease count might be overly simplistic for studies such as the present one. Evidence suggests that disease severity, which could not be evaluated in this study, might yield a more accurate relationship between morbidity and cognitive functioning (Rozzini et al., 2002). Thirdly, the total number of conditions registered in the RNH database reflects the GPs perspective of the relevant health problems. Health problems may still be missing because the patient did not report them to the GP or because the GP did not judge them to be clinically significant. However, the number of undocumented health problems appears to be rather small (Metsemakers et al., 1996).
Key Points
-
Multimorbidity is highly prevalent in a normal aging population
-
Chronic medical conditions can act as a mediating factor in explaining cognitive performance
-
Total morbidity score appears to be unrelated to cognitive functioning
-
More research is necessary in order to expand our knowledge regarding the influence of chronic diseases on cognitive functioning.
Conflict of interest
None declared.
Appendix I
ICPC codes and the corresponding clusters.
| Cluster description | ICPC code |
|---|---|
| All malignancies | A79, B72, B73, D74, D75, D76, D77, F74, N74, R84, R85, S77, T71, U75, U76, U77, W72, X75, X76, X77, Y77, Y78 |
| Peptic ulcers | D85, D86 |
| Other chronic gastro-intestinal diseases | D92, D93, D94, D97 |
| Diseases of the eye | F83, F84, F94 |
| Diseases of the ear | H83, H86, |
| Ischemic diseases | K74, K75, K76 |
| Pulmonary embolism and phlebitis | K93, K94 |
| Cerebrovascular diseases | K89, K90, K91 |
| Arrhythmias & heart failure | K78, K79, K80, K77, K82, K83 |
| Other cardiovascular diseases | K87, K92 |
| Movement disorders | L84, L85, L88, L89, L90, L91, L95, L98 |
| Parkinsonism | N87 |
| Migraine & Headache | N89, N90 |
| Other diseases of the nervous system | N92, N70, N86, N88 |
| Mood disorders | P76, P73 |
| Alzheimer | P70 |
| Other mental disorders | P28, P71, P72, P74, P75, P77, P79, P80, P98 |
| Asthma, COPD and bronchitis | R91, R95, R96, |
| Other chronic respiratory diseases | R70, R75, R97 |
| Eczema, psoriasis and chronic skin ulcer | S87, S91, S97 |
| Endocrine diseases | T85, T86, T92, T93, T99 |
| Diabetes mellitus (Type I and II) | T90 |
| Diseases of the urinary tract | U04, U85, U88 |
