Background
The importance of skilled nurses in the healthcare profession has been highlighted due to societal changes involving an aging population and global outbreaks of infectious diseases, and interest in the nursing labor force has continued to increase [1]. However, recent negative experiences related to COVID-19 and the increased workload have resulted in an even greater intention among nurses to leave their jobs [2, 3]. Since skilled and experienced nurses are crucial for nursing care quality [4], it is important to address the issue of job turnover.
South Korea has a nurse-to-population ratio of 4.4 clinical nurses per 1,000 people, which is only 45.4% of the average of 9.7 in the Organization for Economic Co-operation and Development (OECD) [5]. In comparison with other OECD countries, the staffing level of nurses in medical institutions in South Korea is low. Contrarily, the annual number of new nursing graduates per 100,000 population is 42.4, which is higher than the OECD average of 31.4 [5]. This suggests that registered nurses show little interest in working at hospitals. As a result, the South Korean government introduced “Measures to Improve Working Environment and Treatment for Nurses” in 2018 to boost nurses’ interest in working in medical institutions [6]. However, the rate of turnover for clinical nurses has increased from 12.0% in 2016 [7] to 15.4% in 2019 [8], and it remained 14.2% in 2021 [9]. Therefore, it is important to establish more fundamental measures to encourage nurses to work in hospitals. This study aimed to identify the combined effects of nurse staffing level and turnover rate on patient outcomes in South Korea and present related evidence.
Anzeige
Previous studies that have examined the association between patient-to-nurse ratios and patient outcomes have shown that hospitals with larger numbers of patients to care for, coupled with poor nursing practice environments, have generally experienced increased rates of in-hospital mortality and poor patient outcomes like resuscitation failure and infection [10‐17]. Turnover among nurses has been linked to deficiencies in quality of care, higher incidents of falls and pressure ulcers, and increased patient deaths [18, 19]. However, turnover was not found to be associated with readmission, and consistent results on various patient outcomes were insufficient [4]. Additionally, previous studies on turnover mainly focused on the determinants and intention of nurse turnover [3, 20], and the association of turnover with patient outcomes has been understudied [19].
The reality of South Korean nurses involves two problems simultaneously: low nurse staffing levels and a high turnover rate. Therefore, analyzing the combined effects of low nurse staffing levels and a high turnover rate on patient outcomes is an important task. There is a correlation between low clinical nurse staffing levels and high turnover rates, and a cyclic relationship arising from low staffing levels causing a deterioration of the work environment, such as excessive workload and burnout, leading to high turnover rates. This, in turn, leads to the hiring of new nurses, causing poor work environments due to the time required for training and workload for the remaining nurses [19, 21]. A Korean study also found that a higher bed-to-nurse ratio was closely associated with higher turnover [22].
Therefore, this study aimed to investigate the combined effect of the nurse staffing level and actual turnover rate on patient outcomes, particularly in-hospital mortality, using Korean administrative data, as a basis for explaining the necessity of strategies for fundamentally resolving nurse shortage in South Korea.
Methods
Study design
This cross-sectional study utilized national administrative (National Health Insurance Service; NHIS) data from January and December 2016 to investigate the combined effect of the bed-to-nurse ratio and nurse turnover rate on in-hospital mortality.
Anzeige
Setting and participants
This retrospective, cross-sectional study utilized NHIS data from 459,113 patients. In South Korea, there is a national health insurance system overseen by the government that covers all citizens, and fee-for-service claim data containing medical records provided to patients by all medical institutions are managed by NHIS. The study population was extracted from this big data. The inclusion criteria for this study were adults aged 19 to 85 who were admitted to acute care hospitals (tertiary, secondary, and primary hospitals) with 35 Korean Diagnosis-Related Group (KDRG) categories (Table 1). To select the study population, 22 frequent non-cancer Korean diagnostic groups (KDRGs) with high annual hospitalization rates of over 5,000 patients and a hospital mortality rate of approximately 1% were identified in the preliminary analysis. Thirteen KDRGs with high hospital mortality rates reported in previous studies were also added. This resulted in a total of 35 KDRG patient groups as the study population [14]. The exclusion criteria were admission for medical treatment/care due to official duties or injuries caused by a third party, length of stay of 2 days or shorter or over 365 days, and transfer to another hospital during hospitalization.
Table 1
Patients distribution among the 35 KDRG categories (N = 459,113)
KDRG | Patients N (%) |
|---|---|
B01 Major craniotomy except trauma | 11,014 (2.40) |
B02 Other craniotomy except trauma | 3,165 (0.69) |
B03 Craniotomy for trauma | 3,920 (0.85) |
B63 Degenerative nervous system disorders | 15,492 (3.37) |
B66 Stroke | 59,141 (12.88) |
B72 Sequelae of neurologic disease | 1,570 (0.34) |
E61 Respiratory infection and inflammation | 78,816 (17.17) |
E65 Pleural disorders | 4,142 (0.90) |
E66 Pulmonary edema and respiratory failure | 2,948 (0.64) |
E72 Other respiratory system diagnoses | 7,428 (1.62) |
F02 Cardiac valve procedures with cardiac cath | 932 (0.20) |
F03 Cardiac valve procedures without cardiac cath | 1,248 (0.27) |
F04 Coronary bypass | 1,489 (0.32) |
F06 Major reconstructive vascular procedures | 1,253 (0.27) |
F07 Percutaneous cardiovascular procedures | 44,777 (9.75) |
F63 Heart failure and shock | 10,556 (2.30) |
F66 Other vascular disorders | 7,345 (1.60) |
F71 Cardiac arrhythmia and conduction disorders | 9,641 (2.10) |
F75 Other circulatory system diagnoses | 2,883 (0.63) |
G01 Esophageal procedures | 808 (0.18) |
G02 Rectal resection | 4,368 (0.95) |
G03 Major small and large bowel procedures | 12,832 (2.79) |
G04 Stomach and duodenal procedures | 13,905 (3.03) |
G61 Gastrointestinal hemorrhage | 9,132 (1.99) |
G65 Gastrointestinal obstruction | 8,251 (1.80) |
G68 Other digestive system diagnoses | 10,867 (2.37) |
H01 Pancreas, liver and shunt procedures | 6,159 (1.34) |
H60 Cirrhosis and alcoholic hepatitis | 21,747 (4.74) |
H65 Diseases of the biliary tract | 9,877 (2.15) |
I66 Connective tissue diseases | 8,142 (1.77) |
L60 Renal failure | 25,532 (5.56) |
L63 Kidney and urinary tract infection | 42,151 (9.18) |
T60 Sepsis | 5,052 (1.10) |
T64 Other infectious and parasitic diseases | 5,909 (1.29) |
X62 Poisoning/toxic effects of drugs and others | 6,621 (1.44) |
Next, to obtain accurate estimates from the statistical analysis, hospitals below the minimum patient volume per hospital (i.e., 19 or fewer inpatients and 30 or fewer nurses) were excluded [16, 23]. Hospitals with a submission rate of less than 90% for individual nurses’ joining and leaving, which is essential for turnover calculation, were also excluded, along with patients admitted to these hospitals. Eventually, a total of 459,113 patients admitted to 403 hospitals were analyzed (Fig. 1), as well as 111,342 nurses affiliated with those hospitals during the study period.
For multivariate logistic regression and proportional hazards analysis in medical studies where less than half of the subjects experience the outcome (e.g., death), a sample size is that at least ten outcomes are required for every independent variable [24]. Since the number of deaths in this study was 13,675 and 10 independent variables were included in this model, the sample size in this study met this requirement.
Fig. 1
Flow chart of patient selection based on the National Health Insurance Service data
×
![]()
Ethics approval and consent to participate
Approval from the Institutional Review Board of the Eulji university (EUIRB–2023–005) was received for the study and also approval was received for waiving of informed consent as a secondary analysis before the study began. In addition, approval to use the original data was obtained through the NHIS. The NHIS data contained patients’ personal information and hospital data, including nurse staffing information; however, personal information that could identify individuals was not collected. This study was conducted in accordance with the Ethical Principles for Human Subjects, as defined by the Helsinki Declaration.
Variables
The dependent variable was in-hospital mortality in acute care hospitals, defined as cases where the discharge date on the final treatment bill matched the patient’s reported date of death to Statistics Korea. The independent variables included the bed-to-nurse ratio in general wards and the nurse turnover rate. Using the differentiated inpatient nursing fees by staffing grades in South Korea, the bed-to-nurse ratio was categorized as less than 2.5, 2.5–3.4, 3.5–4.4, and 4.5 or greater [14, 17]. Since November 1999, differentiated inpatient nursing fees by staffing grades have been implemented in South Korea to encourage adequate nurse staffing levels through financial incentives. Every three months, each hospital must report the bed-to-nurse ratio, calculated as the average number of beds over three months divided by the average number of nurses in a ward during the same period, for both intensive care units and general wards to the Health Insurance Review and Assessment Service (HIRA).
Nurse turnover referred to a nurse leaving a hospital during the study period, and the turnover rate was calculated as the number of nurses who terminated their positions at the hospital in 2016 divided by the total number of nurses during the same period [25]. The data on nurse turnover were obtained from the information submitted separately to NHIS by medical institutions. Of the 111,342 nurses in 403 hospitals, 13,349 nurses left their jobs. The mean turnover rate was 12%, which is in line with the average clinical nurses’ turnover rate reported by the Korean Hospital Nurses Association in 2016 [7]. Therefore, for this study, hospitals were categorized based on whether their turnover rate was less than 12% or 12% or more.
Anzeige
The control variables included the characteristics of hospitals and patients. Hospital characteristics included hospital type (tertiary hospital, secondary hospital, and primary hospital) based on medical law, hospital ownership (public sector, educational foundation, and private), hospital location (Seoul, metropolitan cities, and small and medium-sized cities), and enrolled patient volume (20–999, 1,000–1,999, 2,000–2,999, 3,000 or greater). Bed-to-physician ratio was also divided into four groups (less than 3.3, 3.3–4.9, 5.0–9.9, 10 or greater) based on the number of licensed beds. Patient characteristics included age, sex, Charlson comorbidity index (CCI) (0, 1–5, 6 or greater), and admission route (emergency room and outpatient department).
Statistics
The statistical analyses used the χ2 test to assess differences between patients with died or survived and to assess differences in nurse turnover according to hospital characteristics. Generalized estimating equation (GEE) multilevel multivariate logistic regression was used to calculate the adjusted odds ratios (ORs) for in-hospital mortality, controlling for confounding variables (hospital type, hospital ownership, hospital location, beds per physician, age, gender, CCI, and admission route). Due to the possibility of an unknown correlation resulting from a two-tiered structure, with patients nested within hospitals, GEE multilevel logistic regression was used for valid parameter estimation regarding the impact of both individual and cluster characteristics [26]. In addition, the analysis included 403 medical institutions. Thus, the Huber-White sandwich estimator was used to conservatively estimate the regression coefficients. This approach is typically applied in GEE regression analysis for clusters larger than 40 [27]. Data were analyzed using SAS 9.4 (SAS Institute, Cary, NC, USA). Significance was set at P < .05.
Results
Hospital characteristics
The characteristics of the hospitals included in this study are presented in Table 2. The largest group comprised 106 hospitals (26.3%) with a bed-to-nurse ratio of ≥ 4.5 and a nurse turnover rate of ≥ 12%, followed by 96 hospitals (23.8%) with a bed-to-nurse ratio of 2.5–3.4 and a nurse turnover rate of ≥ 12%. Among hospitals with a bed-to-nurse ratio of < 2.5, there were more hospitals with a turnover rate of < 12% (7.7%) than with a turnover rate of ≥ 12% (0.7%). However, among the hospitals with a bed-to-nurse ratio of ≥ 2.5, more hospitals had a turnover rate of ≥ 12% than a turnover rate of < 12%. For instance, 55 hospitals (13.7%) had a bed-to-nurse ratio of 2.5–3.4 and a turnover rate of < 12%, and 96 hospitals (23.8%) had a bed-to-nurse ratio of 2.5–3.4 and a turnover rate of ≥ 12%.
Table 2
Hospitals characteristics (N = 403)
Variable | Categories | Hospitals n (%) |
|---|---|---|
Hospital type | Tertiary hospital | 42 (10.4) |
Secondary hospital | 251 (62.3) | |
Primary hospital | 110 (27.3) | |
Hospital ownership | Public sector | 74 (18.4) |
Educational foundation | 64 (15.9) | |
Private | 265 (65.8) | |
Hospital location | Seoul | 73 (18.1) |
Metropolitan cities | 117 (29.0) | |
Small and medium-sized cities | 213 (52.9) | |
Enrolled patient volume | 20–999 | 267 (66.3) |
1,000–1,999 | 60 (14.9) | |
2,000–2,999 | 35 (8.7) | |
3,000 or greater | 41 (10.2) | |
Bed-to-physician ratio | Less than 3.3 | 77 (19.1) |
3.3–4.9 | 34 (8.4) | |
5.0–9.9 | 143 (35.5) | |
10 or greater | 149 (37.0) | |
Bed-to-nurse ratio and nurse turnover rate | < 2.5 and < 12% | 31 (7.7) |
< 2.5 and ≥ 12% | 3 (0.7) | |
2.5–3.4 and < 12% | 55 (13.7) | |
2.5–3.4 and ≥ 12% | 96 (23.8) | |
3.5–4.4 and < 12% | 22 (5.5) | |
3.5–4.4 and ≥ 12% | 54 (13.4) | |
≥ 4.5 and < 12% | 36 (8.9) | |
≥ 4.5 and ≥ 12% | 106 (26.3) |
Anzeige
Nurse turnover
The study included 111,342 nurses, and 13,349 of them left the hospital in 2016, resulting in a nurse turnover rate of 12.0%. Among the hospital characteristics examined, hospital type, ownership, location, enrolled patient volume, and bed-to-physician ratio were found to be significantly different regarding turnover (P < .001 for all), as shown in Table 3. The actual turnover rate in hospitals with turnover rate of ≥ 12% was 20.4%, which is about three times higher than that 7.0% in hospitals with turnover rate of < 12% (P < .001). Hospitals with a bed-to-nurse ratio of ≥ 4.5 and a turnover rate of ≥ 12% had the highest turnover rate (23.4%), followed by hospitals with a bed-to-nurse ratio of 3.5–4.4 and a turnover rate of ≥ 12% (21.4%) (P < .001).
Table 3
Associations of nurse turnover with hospital characteristics (N = 111,342)
Variable | Categories | Total n (%) | Turnover (n = 13,349, 12.0%) n (%) | Retention (n = 97,993, 88.0%) n (%) | χ2 | p |
|---|---|---|---|---|---|---|
Hospital type | Tertiary hospital | 44,153 (39.7) | 3,001 (6.8) | 41,152 (93.2) | 2176.24 | < 0.001 |
Secondary hospital | 60,602 (54.4) | 8,895 (14.7) | 51,707 (85.3) | |||
Primary hospital | 6,587 (5.9) | 1,453 (22.1) | 5,134 (77.9) | |||
Hospital ownership | Public sector | 22,069 (19.8) | 1,937 (8.8) | 20,132 (91.2) | 1510.53 | < 0.001 |
Educational foundation | 42,452 (38.1) | 3,719 (8.8) | 38,733 (91.2) | |||
Private | 46,821 (42.1) | 7,693 (16.4) | 39,128 (83.6) | |||
Hospital location | Seoul | 29,696 (26.7) | 2,748 (9.3) | 26,948 (90.7) | 294.34 | < 0.001 |
Metropolitan cities | 32,987 (29.6) | 4,405 (13.4) | 28,582 (86.6) | |||
Small and medium- sized cities | 48,659 (43.7) | 6,196 (12.7) | 42,463 (87.3) | |||
Enrolled patient volume | 20–999 | 29,295 (26.3) | 5,227 (17.8) | 24,068 (82.2) | 2059.98 | < 0.001 |
1,000–1,999 | 17,836 (16.0) | 2,700 (15.1) | 15,136 (84.9) | |||
2,000–2,999 | 21,386 (19.2) | 2,321 (10.9) | 19,065 (89.1) | |||
3,000 or greater | 42,825 (38.5) | 3,101 (7.2) | 39,724 (92.8) | |||
Bed-to-physician ratio | Less than 3.3 | 62,424 (56.1) | 5,215 (8.4) | 57,209 (91.6) | 2131.39 | < 0.001 |
3.3–4.9 | 11,752 (10.6) | 1,388 (11.8) | 10,364 (88.2) | |||
5.0–9.9 | 23,823 (21.4) | 4,220 (17.7) | 19,603 (82.3) | |||
10 or greater | 13,343 (12.0) | 2,526 (18.9) | 10,817 (81.1) | |||
Nurse Turnover rate | < 12% | 69,729 (62.6) | 4,862 (7.0) | 64,867 (93.0) | 4449.54 | < 0.001 |
≥ 12% | 41,613 (37.4) | 8,487 (20.4) | 33,126 (79.6) | |||
Bed-to-nurse ratios and nurse turnover rate | < 2.5 and < 12% | 33,620 (30.2) | 2,181 (6.5) | 31,439 (93.5) | 4606.49 | < 0.001 |
< 2.5 and ≥ 12% | 2,310 (2.1) | 358 (15.5) | 1,952 (84.5) | |||
2.5–3.4 and < 12% | 29,521 (26.5) | 2,228 (7.6) | 27,293 (92.4) | |||
2.5–3.4 and ≥ 12% | 24,442 (22.0) | 4,792 (19.6) | 19,650 (80.4) | |||
3.5–4.4 and < 12% | 3,304 (3.0) | 225 (6.8) | 3,079 (93.2) | |||
3.5–4.4 and ≥ 12% | 6,900 (6.2) | 1,477 (21.4) | 5,423 (78.6) | |||
≥ 4.5 and < 12% | 3,284 (3.0) | 228 (6.9) | 3,056 (93.1) | |||
≥ 4.5 and ≥ 12% | 7,961 (7.2) | 1,860 (23.4) | 6,101 (76.6) |
Patients’ characteristics and differences in in-hospital mortality
The overall patient characteristics and differences in frequency between patients who died and survived are depicted in Table 4. Out of 459,113 patients, 13,675 died during their hospital stay, resulting in a mortality rate of 3.0%. The in-hospital mortality was highest in secondary hospitals (3.2%) and lowest in primary hospitals (2.3%). Among the hospital characteristics, hospital type, hospital ownership, hospital location, enrolled patient volume, and bed-to-physician ratio (P < .001 for all) showed statistically significant differences according to in-hospital mortality. There was no significant difference in the turnover rate and in-hospital mortality, but the combined variable of the bed-to-nurse ratio and turnover rate presented a statistically significant difference according to in-hospital mortality (P < .001). The largest patient volume (n = 124,882, 27.2%) was found in hospitals with a bed-to-nurse ratio of < 2.5 and a turnover rate of < 12%. The in-hospital mortality rate (4.3%) was highest in hospitals with bed-to-nurse ratio of 3.5–4.4 and a turnover rate of < 12%, followed by hospitals (3.7%) with a bed-to-nurse ratio of ≥ 4.5 and a turnover rate of ≥ 12%. In addition, among the patient characteristics, in-hospital mortality was significantly different according to age, sex, CCI, and admission route (P < .001 for all) (Table 4).
Table 4
Differences in outcome (death or survival) by hospital and patient characteristics (n = 459,113)
Variable | Categories | Total n (%) | Died (n = 13,675, 3.0%) n (%) | Survived (n = 445,438, 97.0%) n (%) | χ2 | p | ||
|---|---|---|---|---|---|---|---|---|
Hospital characteristics | ||||||||
Hospital type | Tertiary hospital | 170,721 (37.2) | 4,708 (2.8) | 166,013 (97.2) | 124.29 | < 0.001 | ||
Secondary hospital | 260,768 (56.8) | 8,346 (3.2) | 252,422 (96.8) | |||||
Primary hospital | 27,624 (6.0) | 621 (2.3) | 27,003 (97.7) | |||||
Hospital ownership | Public sector | 80,030 (17.4) | 2,928 (3.7) | 77,102 (96.3) | 196.61 | < 0.001 | ||
Educational foundation | 166,329 (36.2) | 5,050 (3.0) | 161,279 (97.0) | |||||
Private | 212,754 (46.3) | 5,697 (2.7) | 207,057 (97.3) | |||||
Hospital location | Seoul | 117,221 (25.5) | 2,954 (2.5) | 114,267 (97.5) | 161.77 | < 0.001 | ||
Metropolitan city | 123,580 (26.9) | 3,547 (2.9) | 120,033 (97.1) | |||||
Small and medium- sized city | 218,312 (47.6) | 7,174 (3.3) | 211,138 (96.7) | |||||
Enrolled patient volume | 20–999 | 115,915 (25.3) | 3,406 (2.9) | 112,509 (97.1) | 44.53 | < 0.001 | ||
1,000–1,999 | 80,581 (17.6) | 2,412 (3.0) | 78,169 (97.0) | |||||
2,000–2,999 | 87,039 (19.0) | 2,875 (3.3) | 84,164 (96.7) | |||||
3,000 or greater | 175,578 (38.2) | 4,982 (2.8) | 170,596 (97.2) | |||||
Bed-to-physician ratio | Less than 3.3 | 241,682 (52.6) | 7,008 (2.9) | 234,674 (97.1) | 27.65 | < 0.001 | ||
3.3–4.9 | 46,278 (10.1) | 1,483 (3.2) | 44,795 (96.8) | |||||
5.0–9.9 | 107,579 (23.4) | 3,137 (2.9) | 104,442 (97.1) | |||||
10 or greater | 63,574 (13.9) | 2,047 (3.2) | 61,527 (96.8) | |||||
Turnover rate | < 12% | 273,070 (59.5) | 8,051 (3.0) | 265,019 (97.0) | 2.13 | 0.144 | ||
≥ 12% | 186,043 (40.5) | 5,624 (3.0) | 180,419 (97.0) | |||||
Bed-to-nurse ratio and turnover rate | < 2.5 and < 12% | 124,882 (27.2) | 3,089 (2.5) | 121,793 (97.5) | 317.02 | < 0.001 | ||
< 2.5 and ≥ 12% | 6,692 (1.5) | 169 (2.5) | 6,523 (97.5) | |||||
2.5–3.4 and < 12% | 121,866 (26.5) | 4,001 (3.3) | 117,865 (96.7) | |||||
2.5–3.4 and ≥ 12% | 97,858 (21.3) | 2,759 (2.8) | 95,099 (97.2) | |||||
3.5–4.4 and < 12% | 11,372 (2.5) | 484 (4.3) | 10,888 (95.7) | |||||
3.5–4.4 and ≥ 12% | 30,046 (6.5) | 817 (2.7) | 29,229 (97.3) | |||||
≥ 4.5 and < 12% | 14,950 (3.3) | 477 (3.2) | 14,473 (96.8) | |||||
≥ 4.5 and ≥ 12% | 51,447 (11.2) | 1,879 (3.7) | 49,568 (96.3) | |||||
Patient characteristics | ||||||||
Age, years | Less than 19 | 1,460 (0.3) | 2 (0.1) | 1,458 (99.9) | 5809.32 | < 0.001 | ||
20–29 | 14,826 (3.2) | 57 (0.4) | 14,769 (99.6) | |||||
30–39 | 22,621 (4.9) | 172 (0.8) | 22,449 (99.2) | |||||
40–49 | 48,171 (10.5) | 670 (1.4) | 47,501 (98.6) | |||||
50–59 | 92,868 (20.2) | 1,502 (1.6) | 91,366 (98.4) | |||||
60–69 | 96,920 (21.1) | 2,105 (2.2) | 94,815 (97.8) | |||||
70–79 | 120,770 (26.3) | 4,911 (4.1) | 115,859 (95.9) | |||||
80 or greater | 61,477 (13.4) | 4,256 (6.9) | 57,221 (93.1) | |||||
Gender | Male | 239,230 (52.1) | 8,032 (3.4) | 231,198 (96.6) | 248.11 | < 0.001 | ||
Female | 219,883 (47.9) | 5,643 (2.6) | 214,240 (97.4) | |||||
Charlson comorbidity index | 0 | 55,299 (12.0) | 608 (1.1) | 54,691 (98.9) | 4124.06 | < 0.001 | ||
1–5 | 299,345 (65.2) | 6,946 (2.3) | 292,399 (97.7) | |||||
6 or greater | 104,469 (22.8) | 6,121 (5.9) | 98,348 (94.1) | |||||
Admission route | Outpatient department | 255,154 (55.6) | 3,263 (1.3) | 251,891 (98.7) | 5742.03 | < 0.001 | ||
Emergency room | 203,959 (44.4) | 10,412 (5.1) | 193,547 (94.9) | |||||
Anzeige
Combined effect between bed-to-nurse ratio and turnover rate on in-hospital mortality
The combined effect of the bed-to-nurse ratio and turnover rate on in-hospital mortality is presented in Table 5. Compared with hospitals with bed-to-nurse ratio of ≥ 4.5 and a turnover rate of ≥ 12%, the in-hospital mortality odds ratio was significantly lower in hospitals with bed-to-nurse ratios of 2.5–3.4 and < 2.5. With the same bed-to-nurse ratio of 2.5–3.4, the risk of in-hospital death was 3%p lower in hospitals with a turnover rate of < 12% (OR = 0.63; 95% CI, 0.50–0.79) than in hospitals with a turnover rate of ≥ 12% (OR = 0.66; 95% CI, 0.54–0.80). With a bed-to-nurse ratio of < 2.5, the risk of in-hospital death was 4%p lower in hospitals with a turnover rate of < 12% (OR = 0.59; 95% CI, 0.44–0.79) than in hospitals with a turnover rate of ≥ 12% (OR = 0.63; 95% CI, 0.48–0.82).
Table 5
Associations between bed-to-nurse ratio and turnover rate and hospital mortality (n = 459,113)
Variable | Categories | Odds ratioa | 95% confidence interval | p |
|---|---|---|---|---|
Bed-to-nurse ratio and turnover rate combined | < 2.5 and < 12% | 0.59 | 0.44–0.79 | < 0.001 |
< 2.5 and ≥ 12% | 0.63 | 0.48–0.82 | < 0.001 | |
2.5–3.4 and < 12% | 0.63 | 0.50–0.79 | < 0.001 | |
2.5–3.4 and ≥ 12% | 0.66 | 0.54–0.80 | < 0.001 | |
3.5–4.4 and < 12% | 0.96 | 0.66–1.38 | 0.809 | |
3.5–4.4 and ≥ 12% | 0.84 | 0.66–1.08 | 0.170 | |
≥ 4.5 and < 12% | 0.86 | 0.65–1.13 | 0.271 | |
≥ 4.5 and ≥ 12% | 1.00 |
Discussion
Using data from the NHIS, this study examined the correlation between nurse labor force and in-hospital mortality, specifically looking at the combined impact of the bed-to-nurse ratio and turnover rate. The findings suggest that improving bed-to-nurse ratio of 2.5–3.4 (compared to ≥ 4.5 beds in general wards) resulted in a significant 34% reduction in the odds of mortality, and this risk further decreased to 37% when combined with a turnover rate of < 12%. With a bed-to-nurse ratio of < 2.5, the reduction of mortality was even more significant, and the results also showed a decrease in mortality risk with a lower turnover rate at the same staffing level. These findings have an important implication—namely, that reducing turnover rates can have a significant positive impact on reducing in-hospital mortality.
In this study, the statistically significant decrease in mortality odds ratio for bed-to-nurse ratios of 2.5–3.4 highlights the importance of nurse staffing levels in South Korea. Studies utilizing the same criteria as this study have shown statistically significant reductions in inpatient mortality risk [28] and postdischarge mortality in patients who underwent surgery at hospitals [14] with a bed-to-nurse ratio of less than 3.5. However, previous studies on the correlation between nurse staffing and patient outcomes, especially mortality, in other countries used different variables to measure nurse staffing; therefore, it would be difficult to directly compare significant findings regarding the association between nurse labor force and patient mortality [12, 13, 16]. Specifically, previous studies used nursing hours per patient day (NHPPD) or the nursing skill mix as nurse labor force variables, but this study used the bed-to-nurse ratio based on data submitted to calculate nursing management fees through the health insurance system as a measurement of the nurse labor force in South Korea. The bed-to-nurse ratio is the mean number of beds over three months divided by the mean number of nurses and differs from the number of patients actually cared for by one nurse during a shift. A study that estimated the number of patients cared for by one nurse during a shift found that a bed-to-nurse ratio of 2.5 equaled 11.9 patients [29]. Among the hospitals included in this study, only 8.4% had a bed-to-nurse ratio of less than 2.5, while only 37.5% had a bed-to-nurse ratio of 2.5–3.4. This indicates that the staffing levels in Korea are insufficient, as this is very low compared to a previous study, according to which the patient-to-RN ratio (i.e., the ratio of the total number of patients to the total number of registered nurses who provided direct nursing care during a shift) was 6.7 [30]. A previous study on nurse staffing levels [10] reported that the average number of patients cared for by a nurse during a shift was 8.3. In addition, reducing the average number of patients cared for by a nurse from 8 to 6 and increasing the proportion of nurses having bachelor’s degrees from 30 to 60%, addressing these two factors combined could decrease patient mortality risk by about 30%. Therefore, it is necessary to improve nurse staffing levels in South Korea.
It was difficult to find previous studies to contextualize the combined effect of job turnover and staffing levels as observed in this study. Previous research has shown that nurse turnover in nursing homes has a significant correlation with patient mortality rates, indicated by an 8.3–17.4% increase in discharge death rates for every 10%p increase in nurse turnover [19]. However, another study showed that nurse turnover was not significantly correlated with readmission rates in nursing homes [4]. These previous studies noted that the effect of the turnover rate on patient mortality or quality of care had been underestimated [19] and pointed out the need to analyze both turnover and retention rates due to the importance of maintaining experienced nurses who contribute to the quality of care provided by medical institutions [4]. Based on these suggestions, our study aimed to identify the joint association of nurse staffing levels and the turnover rate with patient outcomes, and we found a statistically significant association with in-hospital mortality. A systematic review also confirmed that studies on nurse turnover mainly focused on economic impact, and studies on patient outcomes had small sample sizes, making it challenging to determine the noneconomic impacts [18]. Excessive nurse turnover negatively affects the collaborative work process between nurses and medical teams, which can harm patient outcomes [18, 31]. Therefore, future research is necessary to identify the association between nurse turnover and patient results.
In this study, the hospitals with the worst work environment (i.e., a bed-to-nurse ratio of ≥ 4.5, and a turnover rate of ≥ 12%) constituted the largest proportion (26.3%). In contrast, the percentage of hospitals with the same bed-to-nurse ratio but a turnover rate of < 12% was 8.9%. This result shows that there were more hospitals with higher turnover rates when the staffing level was poor. Meanwhile, hospitals with the best work environment (i.e., a bed-to-nurse ratio of < 2.5) and a turnover rate of ≥ 12% accounted for only 0.7% of the total, while hospitals with the same bed-to-nurse ratio but a turnover rate of < 12% comprised 7.7% of all hospitals, indicating that there were more hospitals with a low turnover rate when the staffing level was good. This study aimed to generate highly valid results by through a multilevel analysis of the actual turnover rate and nurse staffing levels for all hospitals that met the research criteria [26]. These findings align with a study that found a poorer staffing level to be associated with a higher risk of new nurses’ turnover and a lower proportion of experienced nurses [22]. Nurse workload can be important predictors of nurse turnover and nurse staffing buffers the workload-turnover relationship as a first-stage moderator [32, 33]. Thus, these results demonstrate the importance of nurse staffing levels, and it can be expected that improving the staffing level will reduce the turnover rate and improve patient outcomes.
Strengths and limitations
This study is significant as it analyzed national data and made a distinctive contribution by demonstrating the association of nurse staffing levels and turnover rates with in-hospital mortality. However, there are some limitations to the study. Firstly, this was a cross-sectional study and may not be sufficient to infer causality. Secondly, poorer staffing levels were linked to higher turnover rates. The study only included hospitals that submitted 90% or more of nurse-related data and excluded hospitals that did not submit such data; thus, it cannot be ruled out that nurse staffing levels were overestimated. Nevertheless, the uniqueness of this study lies in its confirmation of the association of the bed-to-nurse ratio in general wards and turnover rates with in-hospital mortality using national big data.
Conclusions
The present study aimed to establish the basis for addressing the nurse shortage in South Korea by determining that more fundamental measures need to be implemented. Specifically, this study investigated the association of the combination of the bed-to-nurse ratio, as a variable showing nurse staffing levels, and the nurse turnover rate in Korean hospitals with in-hospital mortality risk among patients who were admitted to acute care hospitals. This study made a significant contribution by determining that lower bed-to-nurse ratios and improved nurse staffing levels significantly reduced the risk of in-hospital mortality. Furthermore, hospitals with low turnover rates, despite having the same bed-to-nurse ratio as hospitals with high turnover rates, had significantly lower in-hospital mortality risks. Since a low nurse staffing level means a high workload and is associated with high nurse turnover, fundamental measures to improve nurse staffing levels should be established to reduce the workload. In other words, a package policy that can simultaneously decrease the turnover rate of nurses and increase the nurse staffing level is essential for reducing the risk of in-hospital mortality.
Acknowledgements
The authors would like to acknowledge the NHIS for authorizing the first author to use the NHIS data.
Declarations
Ethics approval and consent to participate
Approval from the Institutional Review Board of the Eulji university (EUIRB–2023–005) was received for the study and also approval was received for waiving of informed consent as a secondary analysis before the study began. In addition, approval to use the original data was obtained through the NHIS. The NHIS data contained patients’ personal information and hospital data, including nurse staffing information; however, personal information that could identify individuals was not collected. This study was conducted in accordance with the Ethical Principles for Human Subjects, as defined by the Helsinki Declaration.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
