Measuring patient acuity and nursing care needs in South Korea: application of a new patient classification system

To lay the groundwork for securing the minimum number of nursing personnel to provide an appropriate level of nursing service, the Ministry of Health and Welfare in South Korea introduced and implemented a differential nursing management fee payment system in November 1999 for the patient-to-nurse placement ratio in general nursing units [8]. However, if the standard is converted to the number of patients per shift, which does not consider the number of patients cared for by nurses working in shifts, the number of patients that one nurse must take care of during the shift increases by fivefold or more [9].

It is unrealistic to regard any outcome as optimal considering that nursing service evaluation criteria such as the differential payment system for nursing care only cover the minimal nurse/patient ratio [8, 9]. Therefore, as an alternative to the minimum staffing standards based on the number of patients, such as the simple patient-to-nurse ratio, an approach to determining staffing requirements that considers the variations of individual patients according to the need for nursing or other factors that induce workload should be developed [10]. Among the main approaches for determining the needs of individual patients, PCS is used for grouping patients according to their nursing needs and assigning the required staffing level to each group based on diagnosis or levels of acuity and/or dependency. Because only KPCS-1 has been applied, the classification system should be investigated to determine whether it can calculate the nursing workforce by reflecting real-world practice [11].

The Korean Patient Classification System (KPCS) was developed in 2009 to measure nursing workload based on the nursing needs of inpatients [12]. It was then revised in 2010 to the Korean-type Patient Classification Tool-1 (KPCS-1) after verification of reliability and validity and is the most widely used system in clinical practice. Patient classification is the process of calculating the nursing requirement score for each inpatient and classifying patients accordingly, and the patient classification score refers to the sum of each added score according to the evaluation guidelines for each item of the patient classification tool. The KPCS-1 classifies patients by focusing on a direct nursing service consisting of 12 different nursing areas (measuring vital sign, monitoring, respiratory treatment, hygiene, diet, excretion, movement, examination, medication, treatment, special treatment, and education/emotional support), 50 nursing activities, and 73 items. To increase ease of use, the number of nursing activities of KPCS-1 was reduced to 11 areas, 34 nursing activities, and 57 items. KPCS-1 was revised to KPCS-GW (General Ward) [4]. According to the total score applied to the assigned score reflecting the workload weight of each nursing activity (Additional file 1), the patients were classified into four groups: Group 1 (mild; 1–10 points), Group 2 (moderate; 11–20 points), Group 3 (severe; 21–30 points), and Group 4 (critically ill; ≥ 31 points) [4].

The National Health Insurance Corporation, one of the health authorities in Korea, modified and utilized the KPCS-1 to investigate patients’ severity/need for nursing care when determining the appropriate level of nursing workforce, including registered nurses, nurse assistants, and formal care workers, for inpatients in integrated nursing care units [13]. Therefore, the KPCS-1 might be regarded as a reference standard in the Korean context. Thus, we selected KPCS-1 and KPCS-GW for comparison with APCS.

As a tertiary referral hospital, Asan Medical Center (Seoul, South Korea) aims to provide high-quality medical services for the efficient management of the nursing workforce while improving the calculation of nursing workload for cost-effective and appropriate nursing workforce management. Our research institute is the largest tertiary hospital in Korea with 2,700 beds and accounts for a high proportion of severe and emergent diseases such as stroke, acute myocardial infarction, and multiple major traumas. Therefore, since the KPCS and KPCS-1 do not sufficiently measure the amount of nursing work for nursing care for patients with severe diseases, we developed the APCS to measure the amount of nursing work by including a more precise scope of nursing work for direct nursing items for severely ill patients.

Accordingly, the Asan Medical Center developed the Asan Patient Classification System (APCS), which reflects the different nursing needs in each unit, and was designed and launched from 2010 to 2012. The APCS was developed by reviewing the workload management system for critical care nurses [14] and the KPCS tool [12] and establishing a consensus with the nursing management expert group and department for multiple items that are time-consuming for direct nursing. The APCS consists of 125 items across 14 areas and 101 nursing activities (Additional file 1). The development of APCS was completed by investigating whether computerized extraction is possible in targeting inpatients in internal and surgical procedures, excluding special departments such as psychiatry, emergency rooms, delivery rooms, neonatal rooms, and dialysis rooms [15]. Various types of nursing care to calculate nursing workload include direct nursing activities such as assessments, admissions and discharges, medications, medication information, respiratory, wounds, suctions, and drains. Additional activities such as nursing record documentation, identification of orders, diets, and patient/caregivers’ health education were entered into the electronic health record (EHR) of our hospital.

As for the range of patient classification scores, the range of total scores was 167–191 for APCS, 104–125 for KPCS-1, and 75–96 for KPCS-GW. The APCS reflects direct nursing activities performed in clinical practice. Further, in order to reflect the nursing needs of hospitalized patients with severe conditions, the range of patient classification scores was set to be higher than those of KPCS-1 and KPCS-GW. The nurses at Asan Medical Center regarded that dependency-acuity was increasing and that the staff number and skill mix should therefore reflect the increasing workload.

This study retrospectively analyzed the medical records of inpatients admitted to the general ward of a tertiary teaching hospital in Seoul, South Korea.

The study population comprised 50,314 patients admitted to 56 general medical-surgical units in March, June, September, and December 2019. Because surgery rates and hospitalizations were increased during the summer and winter seasons in Korea, PCS scores are likely to increase owing to the increase in nursing activities such as respiratory therapy and monitoring in the winter [16]. Therefore, we selected participants by quarter to exclude seasonal variations. Those who were treated in integrated care wards, intensive care units, and emergency rooms were excluded. Because clinical practice environment of them were apparently different from general wards [14].

Data on inpatient characteristics including department, ward, admission and discharge date, age, sex, and 125 nursing activities were collected from the electronic medical and nursing records within an existing EHR system. Patient-level patient acuity and nursing care need scores were automatically generated using the APCS, KPCS-1, and KPCS-GW by the hospital’s EHR.

The collected data were analyzed using IBM SPSS Statistics for Windows, version 21 (IBM Corp., Armonk, NY, USA). Spearman’s correlation and Kappa coefficients according to quartiles were calculated to determine the criterion validity of the APCS compared with the KPCS-1 and KPCS-GW. A Spearman coefficient was interpreted as having no association (ρ = 0) with a perfect monotonic relationship (ρ =  − 1 or + 1) [17]. Strength of kappa agreement was interpreted according to Landis and Koch's as follows:0–0.20 = slight, 0.21 0.40 = fair, 0.41 0.60 = moderate, 0.61 0.80 = substantial, and 0.81–1.00 = near-perfect agreement [18].

A histogram was generated to provide a visualization of the range and comparison of EHR-generated work intensity scores associated with KPCS-1 and KPCS-GW for age groups and departments. For the general characteristics of the subjects, the frequency of each item of the APCS, KPCS-1, and KPCS-GW, and the characteristics of the patient classification score (i.e., frequency, percentage, mean, and standard deviation) were calculated. A one-way analysis of variance (ANOVA) was applied to analyze the difference in the total score of the patient classification by age group, department, and season.

The study was approved by the Institutional Review Board of Asan Medical Center (approval #2021–0631). The need for informed consent was waived by the Institutional Review Board of Asan Medical Center, because of the retrospective nature of the study. Patient participation was anonymous, as names or identifying information were not collected, and the study was conducted in accordance with the Declaration of Helsinki.

A total of 50,314 study patients were included in the analysis, of whom 25,853 (51.4%) were male (mean age, 54.0 years; median age, 58 years). The mean length of hospital stay was 18.3 days, and the median length of hospital stay was 5 days. The average patient classification score using each PCS system was 28.3 points for APCS, 25.7 points for KPCS-1, and 21.6 points for KPCS-GW (Table 1).

When divided according to age groups, patients under 10 years of age had the highest average patient classification scores (APCS, 32.4 points; KPCS-1, 28.7 points; KPCS-GW, 23.9), while those aged 30–39 had the lowest scores (APCS, 25.4 points; KPCS-1, 22.4 points; KPCS-GW, 18.5 points). Patient classification scores were significantly different among the age groups (p < 0.001). In particular, the patient classification scores in the age group of 10 years and younger were significantly higher than those in the other age groups (p < 0.001; Fig. 1).

When divided according to the department (i.e., medical vs. surgical vs. pediatric vs. other), those admitted to pediatric departments had the highest patient classification scores (APCS, 33.1 points; KPCS-1, 29.2 points; KPCS-GW, 24.5 points), while those admitted to “Other” departments (i.e., Otorhinolaryngology, Family Medicine, International Healthcare Center, Anesthesiology and Pain Medicine, Urology, Obstetrics and Gynecology, Ophthalmology, Psychiatry, Dentistry, Dermatology) had the lowest scores (APCS, 23.4 points; KPCS-1, 21.3 points; KPCS-GW, 17.5 points). There was a statistically significant difference in the patient classification scores of the surgical departments (p < 0.001); Fig. 2).

The Spearman’s correlation coefficient for APCS, KPCS-1, and KPCS-GW were very strongly correlated (Spearman’s rho for APCS vs. KPCS-1; 0.98 (p = 0.001) and for APCS vs. KPCS-GW, Spearman’s rho 0.97 (p = 0.001) (Table 2).

The kappa value (Kc) between APCS and KPCS-1 was as high as 0.91, but 10.8% of the total patient values did not match, and KPCS-1 tended to classify patients into groups with lower nursing needs compared with APCS. For example, 1211 patients were classified in the third quartile (75%) according to APCS and in the second quartile (50%) according to KPCS-1, and 101 patients were classified in the fourth quartile (100%) according to APCS and in the second quartile according to KPCS-1 (Table 3). The kappa values (Kc) of the APCS and KPCS-GW were 0.88, and 14.8% of the kappa values did not match (Table 3).

The proposed APCS can be rated by data collected using EHR, whereas previous PCS tools were manually rated by staff nurses, unit managers, attending nurses, and clinical nurse experts [12, 25, 26]. The only studies applying RAFAELA PCS gathered data using medical records to classify the dependency/acuity of inpatients in general wards [27‐29]. Therefore, the new APCS has the advantage of being less labor-intensive than previous tools in terms of data collection. Therefore, the implementation of APCS is feasible for identifying and estimating the current workload of nursing workforce nurses regarding nursing care needs by efficiently classifying the dependency of inpatients in real time.

Spearman's correlation analysis with APCS, KPCS-1, and KPCS-GW showed that all three instruments had a very high correlation. The results may be explained by the fact that the APCS included some nursing activities and domains derived from the KPCS-1. Regarding agreement between APCS, KPCS-1 and KPCS-GW, the kappa value between the APCS and KPCS-1 was 0.91 and the kappa value (Kc) of the KPCS-GW was 0.88. Although there was a high level of agreement, the existing tools tended to underestimate the nursing needs of patients with high acuity compared with the APCS.

The APCS, which is based on 73 KPCS nursing activities, reflects the various tasks involved in 125 nursing activities by considering the opinions of the departments that can extract nursing tasks that are actually performed through a computational process rather than manual data entry. Notably, when analyzed according to age group, the group under 10 years of age showed the highest score among age groups. As pediatric patients are characterized by their developmental stage, their age and weight are considered in various medical devices as well as drug dosages [30]. In consideration of the specificity and complexity of pediatric patients, the APCS added pediatric nursing items including spoon feeding, infant/neonate bottle feed, preparation for pediatric sleep examination procedure, infant phototherapy/infant circumcision, first aid for febrile seizures in children, breastfeeding management, and newborns. Nursing, bed bathing, and treatment for newborns were also considered.

According to Song et al. (2018) and KPCS-1, the Pediatric Patient Classification System (PPS) was developed as a pediatric patient classification tool that considers the nursing characteristics of pediatric patients [30]. PPS has been shown to be a comprehensive tool that reflects the dependency and nursing care needs of pediatric and adult patients admitted to general wards. Importantly, a study on PPS showed that the patient classification score was significantly higher in the patient group aged less than 6 years [30], which is consistent with our current results.

In this study, the age group with the second highest patient classification score after the age group under 10 years of age was the 70 or older group. For the elderly, various types of nursing activities (e.g., assisting, changing positions, and helping patients stand upright) inevitably increase owing to the presence of chronic diseases, increase in the number of medications, blood sugar tests and intravenous administrations, use of complex catheters, tube feeding, and excretion. As the importance of intensive nursing care for preventing falls and risky behaviors in elderly patients continues to grow, nursing management fees have been expanded and revised from “applying and managing physical restraint” to “risk behavior management” [31]. The presence of comorbidities, decreased ADL, cognitive impairment, and depression in elderly inpatients is an important predictor of increased nursing care [32]. Elderly risk screening and assessment tools suggest that high-risk older adults have longer hospital stays, require more nursing care, and have an increased risk of falls [33]. Lee et al. (2014) explained the burden of nursing care for elderly patients in relation to complex health issues, lack of understanding, and the possibility of falls [34]. However, as there are insufficient data on the workload of nursing for elderly patients, research on developing a patient classification system is required to estimate the need for nursing care for elderly patients and the amount of nursing work. Accordingly, it is necessary to consider the establishment of a management fee for elderly patients and the arrangement of nursing staff in the ward.

A patient's nursing need/severity is related to the amount of nursing work, and the presence of an appropriate number of nurses affects the incidence of adverse reactions, including infection, pressure sores, falls, and prescription errors [35]. An optimal nurse-to-patient ratio is important to improve the quality of care and patient outcomes [36]. In Korea, the Ministry of Health and Welfare's 2019 medical delivery system improvement policy contains content such as improvement of the evaluation and compensation system for intensive care of critically ill patients in tertiary hospitals [37]. As the number of critically ill patients receiving care in tertiary hospitals increases, the need for nursing care and nursing workload will inevitably increase. In our study, 63.2% of patients were placed in groups 3 (severe) and 4 (critically ill) according to the KPCS-1, showing that our study group had a higher proportion of patients with more severe conditions compared to those in previous work. For example, the study of Song et al. [4] reported that 30% of the patients were classified into groups 3 and 4. Therefore, the APCS might be more suitable than the existing KPCS-1 and KPCS-GW because it reflects the real-world situation of tertiary hospitals in charge of severe patients. In other words, a comprehensive APCS to capture various nursing care services and workloads for severely ill patients could be applicable to upper-level general hospitals as well as general hospitals. APCS is feasible to collect data from the EHR compared to the manual methods of other tools, despite complexity of nursing activities.