Prevalence of concordant and discordant comorbidity in chronic kidney disease: a large cross-sectional study

Abstract Background: Chronic kidney disease (CKD) is commonly comorbid with hypertension, diabetes and cardiovascular disease (CVD). However, the extent of comorbidity in CKD across a wider range of conditions is not well documented. Aim: To ascertain the prevalence of comorbidity (across 39 physical and mental health comorbidities) in adults aged 25 years and over with CKD in a large nationally representative primary care population. Design and Setting: Cross-sectional analysis of a primary care dataset representing 1,274,374 adults in Scotland. Method: Secondary analysis of general practice electronic medical record data. Comparison was made with the entire population without CKD, standardised for age, sex, and socioeconomic status (SES). Results: A total of 98.2% of adults with CKD had at least one comorbidity (vs. 51.8% in controls). After adjustment for age, sex, and deprivation, people with CKD were more likely to have 1-3 conditions (OR 11.2, 95% CI 10.3-12.2), 4-6 conditions (OR 24.9, 95% CI 22.9-27.0), and ≥7 conditions (OR 38.9, 95% CI 35.6-42.6. Furthermore, all concordant (7 out of 7) and the majority of discordant physical health conditions (17 out of 24) and mental health conditions (6 out of 8) had significantly positive associations with CKD after adjustment. Conclusion: CKD is associated with extreme comorbidity, across a wide range of mental and physical conditions. Routine care for people with CKD should include recognition and management of comorbidities and clinical guidelines should support clinicians to do this.


Introduction
Chronic kidney disease (CKD) is a leading cause of mortality and morbidity 1, 2 and commonly occurs in people with coexistent comorbidity 3,4 which is associated with adverse clinical outcomes. 5 Risk of death in CKD rises exponentially as kidney function deteriorates, largely attributable to cardiovascular disease (CVD). 6,7 Diabetes and hypertension are the leading causes of CKD 6,8 and risk factors for the progression of CKD and CVD overlap. Targeting modifiable risk factors can therefore improve survival and quality of life by reducing CVD in those with CKD, 9 and progression of CKD to end-stage renal dysfunction (ESRD). 1,[7][8][9] CKD is associated with complications that affect all body systems, 1 and people with CKD experience significantly lower health-related quality of life compared with the general population. 6 The coexistence of CKD in the context of both concordant comorbidities (those with shared pathophysiology and/or pharmacological treatments), and non-concordant conditions (where pathophysiology is unrelated and/or where treatments of different conditions are complicating or contradictory) are associated with increased healthcare utilization, length of in-patient hospital stay, and mortality. 3,10,11 Comorbidity and polypharmacy are common in CKD, even in its early stages, and are associated with use of potentially hazardous nephrotoxic medications 12 and significant treatment burden affecting a person's ability to cope with treatment. 4,13 There is a marked social gradient in the prevalence of CKD, 14 where low socioeconomic status (SES) is associated with worse CVD and mortality outcomes. 15,16 People of lower SES are over-represented in those who develop CKD. 17 Although CKD is known to be associated with concordant comorbidity, in particular potentially causal diseases such as hypertension and diabetes, 11 the majority of existing literature examines a limited number of comorbid conditions and without reference to SES. 3,11 Thus, there is a gap in evidence regarding the broader range of conditions that are typically managed in primary care, which includes concordant and discordant physical conditions and mental health conditions. This study examines prevalence of a wide range of comorbid conditions in people with and without CKD, using nationally representative primary care data from a large dataset in Scotland. 18

Methods
This study used data obtained from the Primary Care Clinical Informatics Unit at the University of Aberdeen of 1,274,374 adults aged ≥25 years, who were alive and permanently registered with 314 general practices on the 31 st March 2007 (31% of all practices in Scotland). These practices had A c c e p t e d M a n u s c r i p t -B J G P -b j g p 2 0 X 7 1 4 1 2 5 recorded routine electronic clinical data as part of the Scottish Programme for Improving Clinical Effectiveness in Primary Care (SPICE-PC), a voluntary scheme run by the Scottish Government. The dataset is a nationally representative sample in terms of patients' age, sex, and SES, 19 and was created for a previous study examining multimorbidity. 18 SES was measured using the Carstairs index, 20 which assigns SES based on postcode of residence and was grouped into deciles (equal tenths of the population ranked by Carstairs Index). This score uses four indicators judged to represent material disadvantage in the population taken from census data (lack of car ownership, low occupational social class, overcrowded households and male unemployment).
The dataset contained information on age, sex, SES, and 40 long-term conditions, made up of 32 physical health (including CKD), and 8 mental health conditions. Of physical health conditions, those categorised as concordant with CKD were hypertension, stroke/Transient Ischaemic Attack (TIA), peripheral vascular disease (PVD), diabetes, heart failure, atrial fibrillation (AF), 11 and coronary heart disease (CHD), 3 according to classification from previous studies. Remaining physical health conditions and mental health conditions were classified as discordant (Table 1).
Individuals were identified as having CKD if their primary care electronic medical record (EMR) contained a code for CKD as part of the CKD disease register for the Quality and Outcomes Framework (QOF -the UK national pay-for performance system which incentivised recording and care for CKD stages [3][4][5]. No individuals had missing data for variables included in this analysis and therefore none were excluded from the study. The control group was defined as the entire population aged ≥25 years without a code for CKD in their EMR.
The overall number of physical and mental health conditions each individual had was documented.
We used binary logistic regression models with presence of CKD as the outcome variable and report odds-ratios and 95% confidence intervals (95% CI) for explanatory variables. Unadjusted and models adjusted for age, sex, and SES, were calculated. A P-value of <0.05 was considered statistically significant. All statistical analyses were performed in R version 4.0.2. 21

Demographics
A c c e p t e d M a n u s c r i p t -B J G P -b j g p 2 0 X 7 1 4 1 2 5 A total of 1,274,374 individuals were included in the study, of which 33,567 (prevalence=2.6%) had a code for CKD in their EMR. The control group included all individuals without a CKD code (n=1,240,807). Table 2 and Figure 1 show the characteristics of people with and without CKD. Men accounted for 36.4% of the CKD group, compared with 49.2% in the control group. People with CKD were older than those without (mean 74.9 years vs. 50.6 years). Prevalence of CKD differed significantly between the youngest and oldest. Using 45-54 years as the reference age group, the adjusted odds-ratio (aOR) for CKD presence for those aged 25-34 years was 0.12 (95% CI 0.10-0.15) whilst the aOR for those aged ≥ 85 years was 45.7 (95% CI 42.7-48.8). People with CKD were more likely to live in the most deprived areas (aOR vs most affluent 1.39, 95% CI 1.32-1.47), with aORs for deprivation and CKD significant from deprivation decile 7 and above ( Figure 1).

CKD and total comorbidity
Markedly higher levels of comorbidity were found in people with CKD compared with controls in both unadjusted analysis and age, sex, and deprivation adjusted comparisons. Strikingly, only 1.8% of people with CKD had no comorbidities, compared with 48.2% in the control group (Table 3)
All seven concordant conditions and almost all discordant physical health conditions (17 out of 24) were significantly more common in people with CKD after adjustment for age, sex, and deprivation

Summary
In this analysis of a large, nationally representative primary care dataset, we have found that almost all (98%) of patients with CKD had comorbidity, including concordant and discordant physical conditions, and mental health conditions. Extreme levels of comorbidity were common in CKD. For example, after adjusting for age, sex, and deprivation, having ≥7 conditions was over 40 times more common in the CKD group compared with the rest of the population. Positive associations with CKD were found in all seven concordant physical conditions, 17 of the 24 discordant conditions, and six out of eight mental health conditions after adjustment. Concordant diseases such as hypertension, heart failure, diabetes, CHD, and PVD were diseases with the highest aORs in the CKD group. This is unsurprising given than hypertension, diabetes, and PVD are known to cause CKD, and CVD processes are accelerated in the context of CKD. 11 The majority of discordant diseases, with less research evidence base to explain concurrent prevalence in people with CKD, also showed significant associations with the strongest aORs seen in rheumatological disease, IBD, and pain. Of the mental A c c e p t e d M a n u s c r i p t -B J G P -b j g p 2 0 X 7 1 4 1 2 5 health conditions, schizophrenia and associated conditions, depression, and learning disability had the strongest associations with CKD after adjustment.
Our study found that fewer people in the CKD group were male, which could be explained by the higher mean age of the CKD group in the context of lower life expectancy of males compared to females in Scotland. 22 We also found that people of lower SES are more likely to have CKD after age and sex adjustment. This is an important finding given that individuals of lower SES have greater mortality due to comorbid factors which then predict mortality on an independent basis, 23 including obesity and associated cardiovascular risk factors. 17 Adjusting for age, sex, and deprivation reversed the direction of association for PD and dementia so both had odds-ratios <1. In the case of PD, a possible explanation might be the protective effects of cigarette smoking, 24 where those who smoke might be more likely to develop CKD as a result of CVD while being protected from PD. With regard to dementia, this finding could relate to survivor effects given that people without CVD and diabetes comorbidity are more likely to survive to an age where Alzheimer's Disease is very common.

Strengths and limitations
The study has a number of strengths, notably the use of a large nationally representative primary care dataset, including almost one-third of the Scottish population. Like all studies using routine data, identification of CKD relied on how well GPs recorded the disease in the EMR and some underascertainment is likely. However, CKD was part of the QOF at the time of data extraction which means GPs were financially incentivised to keep accurate registers. The prevalence of CKD in this study was 2.6% compared to the Global Burden of Disease (GBD) study estimate of 5.2%. 25 This discrepancy can be explained by the GBC estimated including CKD stages 1 and 2 (accounting for around half of cases) compared to this study examining CKD 3-5. Data was collected in 2007 and prevalence rates of at least some of the diseases are likely to have changed given the 13-year interim period, although patterns of disease relationships are likely to be similar. Clinical detail, such as the stratification of CKD severity by estimated glomerular filtration rate and proteinuria were not available. However, our study included of a wide range of concordant and discordant physical and mental health conditions, allowing an in-depth analysis of the existence of comorbidities by number and type. Due to the cross-sectional nature of the dataset, causal pathways between variables could not be identified. Comparison made with number of comorbidities between those with CKD and those A c c e p t e d M a n u s c r i p t -B J G P -b j g p 2 0 X 7 1 4 1 2 5 without may not be a direct comparison because having any one condition already may increase the risk of having other comorbidities. However, in a comparison between those with CKD against those with any other condition it would be difficult to apply any statistical test because both groups are overlapping populations and we recognise this as a limitation, but it is remains clear that people with CKD have very high levels of comorbidity.  5 The study reported that comorbidity was seen in 96% of those with CKD and 40% of those with CKD had two or more comorbidities.

Comparison with existing literature
Several studies have examined the relationship between CKD and concordant conditions including CVD and diabetes. [26][27][28][29] Few have examined associations with discordant physical and mental health conditions which are addressed in detail in our study. Bowling et al studied the association between number of chronic conditions stratified by the presence of one or more discordant conditions. 4 They found that at least one discordant comorbidity was associated with higher risks of emergency department (ED) visits, and hospitalisation. A large Canadian study by Tonelli et al studied half a million people with CKD and found that adverse clinical outcomes including mortality, hospitalisation and myocardial infarction were common in all types of comorbidity including concordant, discordant and mental health conditions. 12 A recent systematic review and meta-analysis of multimorbidity and adverse clinical outcomes concluded that there is an association between multimorbidity and increased mortality and morbidity in people with CKD. 6 A key recommendation for future research was describing the prevalence of a wide range of comorbidities associated with CKD, which is addressed in our study.
High levels of comorbidity are commonplace in people with CKD, which means that complex and discordant treatment regimens are likely to be common. These regimens can be challenging to adhere to in the context of poor coping mechanisms, and a lack of knowledge about self-management strategies and social support. 30 Clinicians can also find the management of CKD in the context of multimorbidity challenging, given that the majority of clinical guidance relates to single-disease models often with conflicting advice between conditions. Such guidance can drive cumulative polypharmacy, without providing direction on how best to prioritise recommendations for individuals in whom treatment burden will sometimes be overwhelming. 31

Conclusions
The findings of this study show that not only concordant, but discordant physical and mental health conditions were significantly more common in individuals with CKD. It is clear that integrated guidance and combined generalist and specialist expertise is required to ensure that appropriate care is delivered to people with CKD, recognising the complexity of their health needs. Further prospective cohort studies are required to determine the sequence in which different types of comorbidities and individual conditions emerge over the life course in people with CKD and evaluate generalist approaches to CKD care. A c c e p t e d M a n u s c r i p t -B J G P -b j g p 2 0 X 7 1 4 1 2 5