Interaction of multimorbidity and frailty on adverse health outcomes in elderly hospitalized patients

I study design

We conducted a retrospective cohort study using routinely collected administrative hospital and mortality data.

Setup and data

New South Wales (NSW) is Australia’s most populous state with 7.2 million residents in 2012.26. We used a NSW patient admissions data collection (hospital data) linked to mortality data for the period 1 January 2008 to 31 March 2013. Hospital data included records of all public and private hospital admissions ending with discharge, transfer , type change or death. Hospital admissions were coded using the International Statistical Classification of Diseases and Related Problems, Tenth Revision, Australian Modification (ICD-10-AM) and Australian Advanced Diagnostic Group (AR-DRG) codes.27. The Health Record Linkage Center linked the two data sets using probabilistic methods, with false positive and false negative rates of 0.5%28.

Construction of a study cohort

Our study replicated the inclusion criteria of the original HFRS publication9 with a cohort of NSW residents aged 75 and over having at least one unplanned admission to an emergency hospital between 1 January 2010 and 31 December 2012. For admissions ending in a change of type (eg from acute to subacute care) or transfer, consecutive periods of stay were constructed using admission dates and admission status from the first episode and separation dates and type of separation from the last episode of care. We selected one random hospital stay for each patient as their “index” admission.

Predictions and results

We classified the two primary analysis variables of interest, multimorbidity and risk of frailty, using ICD-10-AM diagnosis codes from the index admission and all hospitalizations during the previous two-year period.

Long-term conditions were ascertained from a list of 29 chronic conditions from the Charlson and Elixhauser indices29supplemented with major diseases from more recent systematic reviews3031 (Supplementary Table S4). Multimorbidity is defined as the presence of at least two conditions from this list.

We calculated continuous HFRS using 109 ICD-10 codes from Gilbert et al.9, adapted to the Australian modification (ICD-10-AM) (Supplementary Table S5). The HFRS captures frailty-related comorbidities as well as functional deficits and symptoms. The presence of each of the 109 ICD-10 codes was ascertained from the patients’ hospital records, assigned a weight, and the weights summed across all codes to obtain the HFRS9. We created dichotomous frailty groups of low frailty (HFRS < 5) and increased risk of frailty (HFRS ≥ 5, combining intermediate and high frailty) using the validated cut points from Gilbert et al.9.

We constructed a composite variable of risk of multimorbidity and frailty with four categories: neither multimorbidity nor at increased risk of frailty, only increased risk of frailty, only multimorbidity, and both multimorbidity and at increased risk of frailty.

Other covariates included age at index admission (in five-year age groups), sex, socio-economic status quantiles based on the Australian Bureau of Statistics’ Index of Relative Socio-Economic Advantage and Disadvantage (IRSAD) of Area Socio-Economic Indices (SEIFA) and number of hospitalizations in the previous two-year period (none, one, two or more).

Outcomes of interest included: mortality within 30 days of index admission; prolonged hospital stay (> 10 days in hospital); unplanned readmission within 30 days of discharge (for patients discharged alive), consistent with the original HFRS development study9. Outcomes were stratified by admission type, grouped into medical (no operating room procedure), surgical (involving significant operating room procedure), and other (involving non-operating room procedure) admissions based on AR-DRG procedures32.

Statistical analysis

We used descriptive statistics to compare demographic characteristics and crude outcome proportions between multimorbidity and frailty risk groups.

We constructed random intercept Poisson models to quantify the association of outcomes with multimorbidity and frailty, accounting for clustering within hospitals and adjusting for age, sex, socioeconomic status, and number of prior admissions. Effects are reported as relative risks (RRs) given the high frequency of outcomes33,34.

We calculated and presented the interaction analyzes as recommended by Knol and VanderWeele16. Interaction on an additive scale was assessed using the relative excess risk due to interaction (RERIRR), with adjustments for clustered data35. RERIRR= 0 implies no interaction (exact additivity), RERIRR> 0 means interaction more than additivity and RERIRR< 0 means an interaction less than additivity. The multiplicative scale interaction was assessed by including an interaction term in the adjusted Poisson model including both the main effects (multimorbidity and frailty) and an interaction term (multimorbidity*frailty). Significance of an interaction on the multiplicative scale is indicated when the relative risk of the interaction term is different from 1, and on the additive scale if the RERI is different from 0.

We used SAS version 9.4 (SAS Institute Inc., Cary, NC) for data management, analysis, and graphing.

Ethics approval

We obtained ethical approvals from the NSW Population and Health Services Research Ethics Committees (reference number 2009/03/141) and the Aboriginal Health and Medical Research Council Ethics Committee (reference number 684/09), with written informed consent waived. The study was conducted in accordance with the Australian National Health and Medical Research Council’s National Statement on Ethical Conduct in Human Research36.

Data availability

The datasets used in this article are available from the NSW Department of Health and the Registry of Births, Deaths and Marriages, NSW, Australia. The datasets were constructed with the permission of each of the source data custodians and with specific ethical approvals. The authors do not have permission to share data from individual unit records due to their highly confidential nature. Data are available to researchers on request and are subject to approval processes by data custodians and ethics committees as outlined on the NSW Health Records Center contact website (https://www.cherel.org.au/apply -for-linked -data).

Leave a Comment