Materials and methods  

Population selection

The LiTherIM (Limitation of Therapeutic effort inInternal Medicine) study is a retrospective study conducted in the internal medicine department of Lausanne University Hospital (CHUV) in Switzerland. The department has 200 beds with some periodic variations, including IMCU beds (14 to October 2018, 16 since November 2018), with extensions during the SARS-CoV-2 pandemic.

Categories

Most patients are under the responsibility of the internal medicine medical staff, and some are under the responsibility of one of the specialties geriatrics, pneumology, gastroenterology, immunology, endocrinology or nephrology. The intensive care unit service of the hospital has 35 beds, with extensions during the SARS-CoV-2 pandemic.

The hospital guidelines require the physician in charge of the patient to document a resuscitation code (full-code or Do Not Resuscitate in case of cardiac arrest) in a dedicated folder of the electronic medical record.

Therapies

The resuscitation code and other limitations of therapeutic efforts must be filled in after medical evaluation and discussion with the patient. Relatives or medical representatives are involved in case of loss of capacity of discernment. Within the first 48 hours after admission, a “to redefine” code can be filled in, for example, if discussion with the patient was not possible. This code would be considered as “full-code” in case of cardiac arrest during this period. A Do Not Resuscitate code can be unilaterally decided based on medical evaluation of futility of cardiopulmonary resuscitation, but this decision must be integrated in a discussion with the patient or relatives on the goals of care. Additional emergency instructions can be added if relevant, in particular ICU and IMCU admission or not, by selecting a corresponding checkbox. Until July 2018, only Do Not Admit to ICU and Do Not Admit to IMCU options existed. By the end of July 2018, the options ICU admission “Yes” or “No” and IMCU admission “Yes” or “No” were introduced. Capacity for monitoring and organ support therapy in IMCU versus ICU is provided in appendix table S1.

Inclusion criteria

All hospital stays of adult patients admitted to the internal medicine department between 1 January 2013 and 31 July 2023. Exclusion criteria were explicit refusal of reuse of personal data and absence of a documented resuscitation code.

The analysis was carried out at the level of hospital stays not patients, as a given patient might have multiple hospitalisations over the study period, with clinical and personal evolution such as new diagnoses, leading to a decision of new limitation of therapeutic efforts in subsequent stays. Nevertheless, we used the patient ID as a cluster, as patients present specific characteristics that might influence the decisions.

Since January 2013, all patients hospitalised at CHUV are asked for general consent that allows future use of medical records and blood tests performed during their hospitalisation. For patients who did not give explicit consent, reuse of the data was authorised if there was no clear refusal, according to article 34 of the Swiss law on human research, considering that a large proportion of these participants was deceased at the time of the start of the study protocol, which would imply disproportionate difficulties to contact every participant or relatives, and that the results would be of better reliability, without any detrimental effect for the participants. The study was approved by the ethics committee (CER-VD, 22 January 2024, ref 2023-01828). Data were extracted from the electronic records by a dedicated team and anonymised for the analysis.

As a primary outcome, we calculated the yearly prevalence of the presence of the following limitations of therapeutic efforts: Do Not Resuscitate code, Do Not Admit to ICU order or Do Not Admit to IMCU order. As secondary outcomes, we sought the factors associated with limitations of therapeutic efforts, focusing on age, sex, medical diagnoses, religion, length of stay, ICU or IMCU admission; incidence of new Do Not Resuscitate codes during the hospital stay; factors associated with new Do Not Resuscitate codes; impact of the SARS-CoV-2 pandemic on prevalence of limitations of therapeutic efforts.

As a resuscitation code is mandatory for all hospitalised patients, patients were dichotomised into Do Not Resuscitate or “full-code”. “To redefine” codes were considered as “full-code” for the analysis. As Do Not Admit to ICU and Do Not Admit to IMCU are limitations of therapeutic efforts which can be added, any patient without this limitation documented in the medical record was considered free of this limitation of therapeutic effort and labelled “Yes” for the analysis (Do Not Admit to ICU versus “ICU-yes and not documented”; Do Not Admit to IMCU versus “IMCU-yes and not documented”).

In case of multiple documentation of limitation of therapeutic efforts through a single hospital stay, we used the latest documentation to calculate the prevalence and the related factors. The annual prevalence was calculated as a percentage of hospital stays based on the year of admission. To calculate the incidence of new Do Not Resuscitate codes, only patients whose resuscitation code was “full-code” on admission were included, and patients with “to redefine” initial codes were excluded. The incidence was calculated as the percentage of hospital stays during which a change from an admission “full-code” to a Do Not Resuscitate code at the end of the stay was documented.

Covariates

Demographic and administrative data collected were: age at admission (continuous), sex (male, female), religion (Christian Protestant, Christian Catholic, Muslim, Other, unknown), admission date, hospital length of stay, admission to ICU or IMCU (yes, no) and death during hospital stay (yes, no).

We sought the following main or co-diagnoses based on the coding of the International Classification of Diseases 10th revision (complete list available in table S2): chronic obstructive pulmonary disease (COPD), heart failure, cirrhosis, or((Language Editor: Delete “or”?)) chronic hepatic dysfunction (alcohol- or non-alcohol-related), stage IV or V chronic kidney disease, malignant neoplasm, metastatic cancer and dementia. As frailty was not coded as such, we sought diagnoses related to age-acquired frailty: delirium, difficulty in walking and falls, and protein-energy malnutrition.

Statistical analysis

Analysis was performed using Stata® software version 18.0 (Stata Corp, College Station, TX, USA). Results were expressed as number of hospitalisations (percentage) for categorical variables and as median [interquartile range] for continuous variables. Bivariate analysis of the factors associated with limitations of therapeutic efforts was carried out using the chi-squared test for categorical variables and the Kruskal-Wallis test for continuous variables. Multivariable analysis was conducted using a mixed logistic regression model clustering on patient ID and adjusting for year (continuous), sex (male, female), comorbidities (present, absent), religion (Christian Protestant, Christian catholic, Muslim, Other, without/unknown), length of stay (one day increase) and ICU/IMCU stay (yes, no). Results were expressed as odds ratios with 95% confidence intervals. Statistical significance was set at a two-tailed p <0.05.