Multimorbidity and kidney health in old age : methodological considerations and longitudinal

<p dir="ltr"><b>Introduction</b>: Multimorbidity (the presence of 22 chronic conditions in one individual) and chronic kidney disease (CKD) are increasingly common in older adults, often occurring together and contributing to poor health outcomes. Both kidney function assessment and the operationalisation of multimorbidity pose important methodological challenges that influence how findings are interpreted and compared. Chronic conditions in older adults tend to cluster into multimorbidity patterns that may share common causes or mechanisms, yet the ways these patterns are measured and described vary widely, limiting comparability across studies. At the same time, assessing kidney function in this population is complicated by age-related physiological changes that affect the production and clearance of creatinine, a key biomarker for its estimation. Beyond these methodological issues, little is known about how specific combinations of chronic conditions relate to kidney outcomes, or how such insights might help identify individuals at higher risk of poor kidney health and shed light on underlying pathophysiological mechanisms.</p><p dir="ltr">The aim of this thesis was to explore the intersection of multimorbidity, kidney health and ageing, focusing on how kidney function is measured, how multimorbidity patterns are identified, and how these patterns relate to changes in kidney structure and function over time. This was addressed through one systematic review and three observational studies from two well-characterised cohorts.</p><p dir="ltr"><b>Study I</b>: Systematic review of 16 studies deriving multimorbidity patterns from primary care electronic health record data. Reported multimorbidity prevalence ranged from 14.0% to 93.9%. Mental health and cardiovascular patterns were identified in all studies, frequently alongside conditions from other organ systems, indicating that some patterns are highly replicable across populations. However, marked heterogeneity in analytical decisions related to disease ascertainment and coding, clustering methods, pattern labelling, and stratification limits comparability across studies and hampers evidence synthesis.</p><p dir="ltr"><b>Study II</b>: Cohort study of 3,094 adults aged 60 years and older from the Swedish National Study on Aging and Care in Kungsholmen (SNAC-K), comparing the Modification of Diet in Renal Disease (MDRD), Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI), Revised Lund-Malmö (R-LM), Berlin Initiative Study (BIS), and European Kidney Function Consortium (EKFC) creatinine-based eGFR equations. CKD classification agreement ranged from moderate to very high (Cohen's K 0.42-0.91), with MDRD and CKD-EPI generally providing higher GFR estimates than the other three equations. Agreement was poorest between MDRD and BIS, and highest between R-LM and EKFC. BIS showed the best discrimination for 15-year all-cause mortality (Harrell's C = 0.73), although prognostic accuracy declined in those aged over 78 years and with low calf circumference. Discrepancies between equations were not consistent across the spectrum of age, calf circumference, and BMI.</p><p dir="ltr"><b>Study III</b>: Cohort study of 3,094 adults from SNAC-K followed for 15 years, examining the number and patterns of chronic conditions in relation to kidney function decline. Multimorbidity was present in 87% of participants. The number of chronic conditions was independently associated with both absolute decline, measured as excess annual eGFR slope (ß -0.05, 95% CI: - 0.07; - 0.03), and relative decline, defined as >25% reduction from baseline (HR 1.23, 95% CI: 1.17; 1.29). Five multimorbidity patterns were identified, with the Unspecific, high burden and Cardiometabolic patterns showing the steepest absolute declines (ß -0.15, 95% CI: - 0.26; - 0.05 and ß -0.77, 95% CI: - 0.98; - 0.55, respectively) and relative declines (HR 1.45, 95% CI: 1.09; 1.92 and HR 3.45, 95% CI: 2.27; 5.23, respectively) compared with the least burdened pattern. The Cognitive and Sensory pattern was also associated with greater relative decline (HR 1.53, 95% CI: 1.02; 2.31), while the Psychiatric and Respiratory pattern showed no association.</p><p dir="ltr"><b>Study IV</b>: Cohort study of 205,449 adults aged 65 years and older undergoing outpatient albuminuria testing, examining eGFR-independent associations between multimorbidity patterns and incident albuminuria (>30 mg/g) and macroalbuminuria (>300 mg/g). Six patterns were identified in ages 65-74 and seven in >75 years. Patterns included both organ-specific (e.g. Cardiovascular, Eye) and more general patterns (Unspecific, Multisystem). For albuminuria, compared to those without multimorbidity, the highest risk in ages 65-74 was observed for the Cardiovascular pattern (HR 3.09, 95% CI: 2.85; 3.34), while in those >75 years, the Vascular pattern had the highest risk (HR 2.41, 95% CI: 2.08; 2.80). The Dementia pattern was not associated with albuminuria risk. High- burden patterns in older adults had elevated albuminuria risk, although this was attenuated after accounting for competing mortality. The 5-year cumulative incidence of albuminuria exceeded 10-14% in the highest-risk patterns, 5-9% higher than in those without multimorbidity.</p><p dir="ltr"><b>Conclusions</b>: This thesis synthesises evidence on multimorbidity patterns across primary care populations, showing that certain patterns are replicable and that methodological harmonisation is needed to improve comparability, strengthen evidence synthesis, and facilitate translation into clinical practice. It also shows that creatinine-based eGFR equations are not interchangeable when assessing kidney function in older adults, particularly in those with low muscle mass. They also differ in their predictive capacity for mortality risk, underscoring the importance of equation choice, clinical context, and confirmatory testing. Finally, the quantitative burden and qualitative composition of multimorbidity are important determinants of kidney health, with specific patterns differentially associated with the risk of functional decline and structural damage. Such patterns may serve as practical risk-stratification tools to identify older adults who could benefit from targeted monitoring, timely specialist referral, and earlier initiation of preventive interventions.</p><h3>List of scientific papers</h3><p dir="ltr">This doctoral thesis is based on the following four papers, which are referred to in the text as Study I, II, III and IV.</p><p dir="ltr">I. <b>Beridze G*</b>, Abbadi A*, Ars J, Remelli F, Vetrano DL, Trevisan C, Pérez LM, López-Rodríguez JA, Calderón-Larrañaga A. Patterns of multimorbidity in primary care electronic health records: A systematic review. J Multimorb Comorb. 2024;14. <a href="https://doi.org/10.1177/26335565231223350" target="_blank">https://doi.org/10.1177/26335565231223350</a> *Shared first authorship</p><p dir="ltr">II. <b>Beridze G,</b> Vetrano DL, Marengoni A, Dai L, Carrero JJ, Calderón- Larrañaga A. Concordance and Discrepancies Among 5 Creatinine- Based Equations for Assessing Estimated Glomerular Filtration Rate in Older Adults. JAMA Netw Open. 2023;6(3):e234211. <a href="https://doi.org/10.1001/jamanetworkopen.2023.4211" rel="noreferrer" target="_blank">https://doi.org/10.1001/jamanetworkopen.2023.4211</a></p><p dir="ltr">III. <b>Beridze G,</b> Dai L, Carrero JJ, Marengoni A, Vetrano DL, Calderón- Larrañaga A. Associations between multimorbidity and kidney function decline in old age: A population-based cohort study. J Am Geriatr Soc. 2025;73(3):837-848. <a href="https://doi.org/10.1111/jgs.19298" target="_blank">https://doi.org/10.1111/jgs.19298</a></p><p dir="ltr">IV. <b>Beridze G,</b> Mark PB, Sullivan MK, Walker H, Tanaka S, Faucon AL, Vetrano DL, Calderón-Larrañaga A, Carrero JJ. Multimorbidity patterns and the subsequent risk of albuminuria: Findings from the Stockholm Creatinine Measurements (SCREAM) project. [Submitted]</p>