This is a cross-sectional study conducted on data extracted using International Classification of Diseases – 9 (ICD − 9-CM) and International Classification of Diseases – 10 (ICD − 10) coding over 5 years (2015–2020) at the Aga Khan University Hospital, Karachi, Pakistan. It includes all patients aged 18 and above who were admitted to internal medicine service through emergency room or electively. Ethical clearance (2021-7023-20010) was taken from Aga Khan University Ethics Review Committee before conducting the study. Admission list of all patients admitted during the study period 2015–2020 was obtained through the hospital information management system of Aga Khan University.
Primary outcome variable is readmission to hospital within 30 days (of discharge or LAMA [Leave against medical advice]) .
Thirty-day readmission is defined as ‘unplanned readmissions within 30 days of discharge from the index (i.e., initial) admission’ (5).
Independent variables were LACE index and HOSPITAL score.
LACE Index: LACE refers to Length of stay, Acuity of the admission, Comorbidity of the patient (measured with the Charlson comorbidity index score), and Emergency department visits (measured as the number of visits in the six months before admission). LACE indexes ranges from 0–19, with an expected probability for readmission of 0 to 43.7% based on the initial validation study of the LACE score. It was classified as a score of 0–4 = Low; 5–9 = Moderate; and a score of ≥ 10 = High risk of readmission (14). Charlson Comorbidity Index (CCI): CCI predicts the mortality for a patient who may have a range of concurrent conditions, such as heart disease, AIDS, or cancer. A score of zero means that no comorbidities were found; the higher the score, the higher the predicted mortality rate is (17, 18)
HOSPITAL score
refers to Haemoglobin at discharge, discharge from an Oncology service, Sodium level at discharge, Procedure during the index admission, Index Type of admission, number of Admissions during the last 12 months, and Length of stay. HOSPITAL scores of 0–4 points is classified as low risk for readmission (5%), 5–6 points intermediate risk (10%), and 7 or more points as high risk (20%) based on the initial validation study of the HOSPITAL score (15).
Other variables included patient demographics, health coverage status, acuity of care, ward type, number of hospital visits in the past 12 months, laboratory investigations including haemoglobin, platelets, creatinine, sodium, potassium, bicarbonate etc.
Type of admissions:
Elective Admission
The admission of patients to the hospital from clinic or through direct ward visit.
Emergent Admission
The admission of patients from the emergency department.
30-day readmission
The admission of patient again in internal medicine department within 30-days of leaving the hospital from index admission (discharged or LAMA).Categories of Bed according to Cost
The categories of bed according to the costs are described below (19):
High-cost beds
High-end hospital beds come with sophisticated features and capabilities like integrated patient monitoring, computerised adjustments, customised mattress systems, and improved mobility options. These beds are intended for use in specialist medical facilities, acute care settings, and critical care units.
Intermediate-cost beds
A balance between expensive and inexpensive hospital beds is provided by intermediate-cost beds. They cost less than high-end beds, but they have fewer features and functions than low-priced beds. Long-term care homes, rehabilitation centres, and general medical and surgical wards are the intended uses for these beds. Generally, they provide conventional adjustments, simple mattress systems, and necessary mobility alternatives.
Low-cost beds
In terms of features and functionality, inexpensive hospital beds are less sophisticated. Standard adjustments, simple bedding systems, and necessary mobility alternatives could be provided by them. These beds are the most frequently utilised for hospital admissions. They are made to be economical, practical, and simple to use.
Types of Bed according to Acuity Level:
-
General: General wards have 5:1 ratio of patient to nurse care. Intensive Care Unit (ICU): These units have 1:1 patient to nurse ratio.
-
Special Care Unit (SCU): These units have 2:1 ratio of patient to nurse care.
Data Acquisition
Electronic data was extracted from the following sources (i) Hospital Information Management Systems (HIMS) for diagnosis, demographics, health coverage plan, length of stay, procedures, and mortality (ii) Patient Business and Services Department (PBSD) for health coverage plan, cost of bed, acuity level. (iii) Laboratory data acquired from information technology and through MyPatient application (internal servers of AKU health record system).
Data Extraction and Processing
Data Integration:
The different datasets acquired from the various departments (internal medicine, infectious diseases, endocrinology) were consolidated into one master datasheet using the patient admission number as the primary key, and deduplication was performed to remove any erroneous rows. Furthermore, any patients who did not meet the specified inclusion/exclusion criteria, as outlined earlier, were filtered out at this stage.
Disease categorization:
Patient co-morbid and primary diagnosis for each admission were documented in the original dataset using International Classification of Diseases – 9 (ICD − 9-CM) coding up to 2020, and International Classification of Diseases – 10 (ICD − 10) coding thereafter. We organized related ICD codes into broader disease categories based on ICD coding which were used in the subsequent analyses. A list of the ICD codes and their corresponding categories can be found in the supplement.
CCI calculations:
Moreover, the CCI includes a set of 19 disease categories in its scoring criteria. Thus, to incorporate the CCI into the analysis, the ICD Codes were further categorized to create separate columns for the 19 comorbidities included in the CCI. The Enhanced ICD-9-CM codes were used instead of Deyo’s ICD-9-CM (20) for CCI calculation was used. If a patient had a particular CCI comorbid condition, a score was assigned to the corresponding column based on the CCI's recommended calculation weights. To determine the final CCI score for each patient, the individual column scores were summed up to produce an aggregate CCI score.
Laboratory values:
The laboratory dataset included all investigations performed during the hospital stay. Wherever possible, the most recent value, on the day of discharge, for each parameter were included in the analysis. If a particular lab parameter was not available for the discharge date, a value, up to a maximum of 5 days before the discharge date, was used instead in the subsequent analysis. Patients with missing laboratory data while calculating only the HOSPITAL score (Haemoglobin and Sodium) (about 21%) were excluded from the analyses, and no imputation process was applied in such cases.
Data Analysis
Python scripts were written to perform the above-mentioned data aggregation, transformations, filtering, and groupings needed to prepare the dataset for the analysis. All subsequent descriptive and inferential statistics were then performed using R Statistical Software (v4.3.2; R Core Team 2023).
The normality of the continuous variables was checked with Anderson-Darling test as the total dataset was beyond the limitation of Shapiro-Wilk test (5000 cases). Most of the variables in our dataset were normally distributed (p > 0.05)
Mean and standard deviation is reported for quantitative variables. Frequency and percentages are for qualitative variables. T-test and chi-square was conducted to compare continuous and categorical variables, respectively, in readmitted vs not readmitted patients within 30 days.
Patients with HOSPITAL score > 5 and LACE score > 10 were considered high risk for readmission (15, 21). Area under the curve (AUC) was calculated using the ROC method and Brier score was also calculated. A p-value < 0.05 was considered statistically significant for all analyses.
Patient and Public Involvement:
None