Introduction Background

In accordance with the Global Nursing Workforce and International Migration Report published by the International Council of Nurses in 2022, titled “Sustain and Retain in 2022 and Beyond”, the global nursing workforce is facing a shortage of approximately 13 million professionals due to the increased demand for healthcare services resulting from the COVID-19 pandemic.1 In fact, the shortage of nursing personnel on a global scale is not a unique phenomenon solely attributed to the COVID-19 pandemic.2 The insufficiency of nursing staff represents a pervasive challenge evident in diverse healthcare systems, including the National Health Service system (eg, the United Kingdom,3 Ireland,4 and the Nordic countries5), the social health insurance system (eg, Germany,6 Canada,7 Japan,8 and Taiwan9), the private health insurance system (eg, the United States10), and other contexts.11 The deficit in nursing staff not only compounds the challenges within the nursing profession’s working conditions but also exerts a detrimental influence on patient outcomes (or healthcare quality),12 and hospital financial performance.13 Prior research on nursing staffing shortages can be delineated into two primary themes. The first theme focuses on the physiological and psychological responses exhibited by nursing personnel in response to the deterioration of working conditions within the nursing practice, encompassing aspects such as burnout,14,15 depression,15 job satisfaction,14 intention to leave and turnover,5,12,14,16 and occupational injury risks.14,17 The second theme underscores the impact of nurse staffing shortages on various dimensions of care quality, including inpatient care quality,12,18–20 emergency care quality,21 ICU care quality,22 and long-term care quality.23,24 In order to enhance the nursing practice environment, mitigate nursing personnel shortages, and consequently ensure the sustained elevation of healthcare quality, most countries globally have instituted substantial regulations and controls governing nurse staffing levels within hospital settings.

According to a study conducted by Aiken et al, the average patient-to-nurse ratio (PNR, hereafter) was found to vary from approximately 3.4 to 17.9 patients per nurse across countries belonging to the Organization for Economic Cooperation and Development.25 Norway and Ireland exhibited the most favorable PNRs, with averages falling below seven patients per nurse. In close succession were the Netherlands, Sweden, and Switzerland, with PNRs ranging from seven to eight patients per nurse. The United Kingdom reported an average PNR of nearly 9, whereas Belgium and Spain demonstrated PNRs exceeding 10 patients per nurse, suggesting a comparatively higher care burden on nursing personnel. Moreover, recent studies suggest that the average PNRs in some advanced Asian countries are approximately 7.5 in South Korea, 10.5 in Japan, and 8.1 in Taiwan.26,27 The identified positive correlation between nurse staffing levels and care quality, as highlighted in the study by Aiken et al25 and many other studies,12,18–25,27 underscores the potential of regulating hospital nurse staffing as a crucial policy instrument in healthcare administration. This instrument has the capacity to serve as an effective means of monitoring and upholding inpatient care quality to elevated standards. Consequently, recent studies examining the effectiveness of nurse staffing policy have predominantly concentrated on the effect of changes in nurse staffing on inpatient care quality. The methodologies employed in these investigations frequently utilize intervention models within a natural experiment framework. For instance, de Cordova et al utilized interrupted time-series analysis to scrutinize the effects of reporting and disclosure measures on PNRs in the United States, and their findings indicated a noteworthy reduction in PNR subsequent to the implementation of reporting and disclosure measures, signifying an amelioration in the care burden on nursing personnel.28 Similarly, Van et al employed the same methodology to evaluate the impact of intervention measures on PNRs in the United States, revealing a positive influence of diminished PNRs on inpatient care quality.20 This positive correlation between nurse staffing levels and inpatient care quality is also evident in numerous other nursing studies.12,18–20,22,25,27

Demographic Change & Nurse Staffing

Notwithstanding the established positive influence of nurse staffing levels on inpatient care quality, the accelerated aging of the global population presents a substantial challenge to the effectiveness of hospitals’ nurse staffing policy from two distinct perspectives. First, the increasing elderly population amplifies the demand for healthcare, exacerbating the severity of nurse shortages.29 Second, the rise in unmarried or childless individuals contributes to a departure from familial care responsibilities,30 leading to fewer economically active people financing health and welfare systems, while the demands on these systems continue to increase.31,32 Hence, comprehending the potential impact of demographic structural changes on the effectiveness of hospitals’ nurse staffing policy emerges as a crucial concern within the realms of healthcare services. Previous studies on the influence of demographic structural changes on policy effectiveness have primarily included empirical studies within diverse economic markets, encompassing the monetary market,33,34 securities market,35 exchange market36 real estate market,37 and various commodity markets.37 Nevertheless, research examining the nexus between demographic changes and the effectiveness of healthcare policies is scarce within the literature on healthcare services. To the best of the authors’ knowledge, Lin et al conducted the initial study to scrutinize the impact of demographic changes on the effectiveness of the outpatient copayment policy in Taiwan, and their findings suggested that the effectiveness of the outpatient copayment policy would be enhanced by the aging population.38 In contrast to Lin et al results, Chen et al undertook a study on the influence of demographic structural changes on the effectiveness of emergency department copayment policies in Taiwan.39 Their results indicated that if the policy objective is to curtail the overall utilization of emergency services in medical centers, adjustments to copayment levels necessitated by an aging society would need to be significantly higher than the presently established standards.39

Purposes of the Study

In order to fill the research gap pertaining to the impact of demographic changes on nurse staffing policy effectiveness, this study endeavored to investigate the relationship between demographic changes and hospital nurse staffing policy effectiveness. Taiwan was specifically selected as the study region for three primary reasons: First, Taiwan stands as one of the most rapidly aging countries worldwide. The proportion of the elderly population (aged 65 and above) in Taiwan surged from 7% in 1993 to 14% in 2018, with projections anticipating a further increase to 20% by 2025.40 This transition from an aging society (7%) to an aged society (14%) occurred within a mere 25 years, and the subsequent shift from an aged society (14%) to a super-aged society (20%) is anticipated to transpire within a span of just 7 years, underscoring significant demographic changes.40 Second, while Taiwan’s National Health Insurance (NHI, hereafter) program provides comprehensive healthcare services, including inpatient, outpatient, and dental services, traditional Chinese medicine, and prescription drugs, the implementation of a global budgeting payment scheme aimed at controlling the rapid growth of healthcare expenditure may potentially restrict the allocation of nursing staff in hospitals.27,38,39 Third, despite regulatory efforts such as mandated minimum PNRs and the use of the PNR as a critical indicator for hospital accreditation,41 the nursing labor market in Taiwan is confronting a significant challenge, with approximately 40% of nurses expressing reluctance to enter the workforce.42 These circumstances collectively provide a compelling backdrop for the examination of the impact of demographic changes on hospitals’ nurse staffing policy.

In this study, we proceed with our investigation on the nexus between demographic changes and the effectiveness of hospitals’ nurse staffing policy in the following steps: First, we employ the time-varying parameter vector autoregressive (TVP-VAR, hereafter) model, as proposed by Nakajima,43 coupled with time-varying impulse-response analyses (TV-IRA, hereafter). This methodology allows us to simulate the impact of nurse staffing adjustments, specifically, the changes in PNRs, on inpatient care quality, as measured by the unplanned 14-day readmission rate after discharge (hereafter 14-day readmission rate). The cumulative response of the 14-day readmission rate to changes in PNR over a 12-month timespan serves as a metric for assessing the effectiveness of hospitals’ nurse staffing policy. This approach addresses a notable gap in the literature by examining the long-run impacts of nurse staffing on inpatient care quality.20,28,38,39 Second, we employ the autoregressive distributed lags (ARDL, hereafter) model, as introduced by Pesaran et al,44 to examine the long-run relationship among hospital nurse staffing policy effectiveness and age distribution.

Contribution of the Study

This study makes significant contributions to the literature on the effects of demographic changes on the effectiveness of hospitals’ nurse staffing policy based on two aspects: First, differing from prior research on the effectiveness of nurse staffing policies, which frequently concentrates on analyzing the association between nurse staffing levels and inpatient care quality or the short-run effects of nurse staffing policy changes based on one-shock interventions,20,28 this study introduces a novel approach by utilizing the TVP-VAR model in conjunction with the ARDL model. This innovative methodology enables the analysis of the impact of demographic changes on the effectiveness of hospitals’ nurse staffing policy in Taiwan. Second, in contrast to previous studies that relied on a single demographic indicator (such as the proportion of the elderly population, dependency ratio, or economic dependency ratio) to explore the association between demographic changes and policy effectiveness,34 this study conducts empirical analyses based on the age distribution of the entire population. The results of this study have the potential to serve as crucial references for the Taiwan government in formulating more comprehensive nurse staffing policies aimed at sustaining inpatient care quality within Taiwan’s NHI system.

Materials and Methods Research Design

The primary objective of this study was to investigate the relationship between demographic structural changes and the effectiveness of hospital nurse staffing policy. Demographic structural change refers to significant alterations in the composition of a population over time. To measure the effectiveness of hospital nurse staffing policies, we analyzed the impulse responses of inpatient care quality to changes in nurse staffing over a 12-month period. Therefore, the research design utilized in this study aligns with the observational time-series methodology, and several time-series models were used to analyze secondary data to investigate the primary objective of this study.

Data Collection and Samples

The data utilized for this research were sourced from multiple databases: the Open Database of Taiwan’s National Health Insurance (managed by the National Health Insurance Administrative), the Healthcare Quality Database of the Center of Quality Management (under the administration of the Joint Commission of Taiwan), and the Macroeconomics Statistics Database (administered by the Directorate General of Budget, Accounting, and Statistics). Specifically, data pertaining to inpatient care quality (quantified by 14-day readmission rates),27,45,46 labor input (measured through PNR denoting the average number of patients cared for by one nurse per shift),27 and length of stay per admission (utilized to control for the severity of illnesses)47 were collected for medical centers, regional hospitals, and district hospitals. These data were employed in the estimation of the hospital production function outlined in the data analysis section below. Subsequently, the effectiveness of hospitals’ nurse staffing policy, indicated by the cumulative response of the 14-day readmission rate to a standard deviation change in the PNR in acute care wards within a 12-month timespan, was derived from the estimated hospital production function. Additionally, demographic data encompassing the distribution of the population across seven age-specific groups (ie, under 15, aged 15‒24, aged 25‒34, aged 35‒44, aged 45‒54, aged 55‒64, and 65 or over), healthcare workers’ incomes (measured by the real wage level in the healthcare industry27), reimbursement payment per diem (utilized to control for hospital competition under Taiwan’s NHI global budgeting payment scheme27), and business cycles (indicated by the leading indicator of business cycles48) were collected. These data were employed to investigate the long-run effects of demographic change on the effectiveness of hospitals’ nurse staffing policy. Monthly data spanning from January 2015 to March 2023 were collected, resulting in a total of 99 monthly observations for our analyses. The determination of the observed study period relied upon the availability of the longest and consistently maintained time-series data accessible to academic researchers.

Data Analysis

The empirical procedure used to investigate the association between demographic structural change and the effectiveness of hospitals’ nurse staffing policy included two steps: First, we conducted the TV-IRA based on Nakajima’s method43 to obtain the responses of inpatient care quality to changes in nursing staffing within a 12-month timespan across our study period. The magnitude of these impulse responses served as a measure for the effectiveness of hospitals’ nurse staffing policy. The methodology utilized in this study to assess policy effectiveness has been extensively employed in prior policy evaluation studies, spanning analyses of monetary and fiscal policies,33 copayment policies for outpatient care services,38 and emergency department visits.39 Second, given that previous studies, including those by Lin et al38 and Chen et al,39 have delved into the impact of demographic change on healthcare policy effectiveness, we utilized the ARDL model proposed by Pesaran et al44 along with Fair and Dominguez’s method49 to delineate the nonlinear relationship between age distribution and the effectiveness of nurse staffing policy. The lag selection procedures and diagnostic assessments of the goodness-of-fit for the ARDL model drew upon methodologies established in prior studies, such as those conducted by Chang and Chen50 and Chen and Lin.51 For the sake of brevity, we omit the technical details of the specification and estimation process for the TPV-VAR and ARDL models. Interested readers are referred to the Supplementary Materials of this study for further information. The tabulated description of these two empirical models is presented in Table 1.

Table 1 Tabulated Description for Empirical Models

Ethical Considerations

This study uses secondary data, specifically monthly economic indicators, aggregate healthcare utilization, and demographic data for all residents in Taiwan. These data did not involve any human participants or tissue, and the data collection process was approved by the Research Ethics Committee of Taichung Tzu Chi Hospital with the Certificate of Exempt Review ID: REC110-23.

Results

Panel A of Table 2 provides descriptive statistics for all variables used in estimating the TVP-VAR model. As depicted in Panel A, the average 14-day readmission rates were approximately 1.276%, 1.704%, and 1.427% in medical centers, regional hospitals, and district hospitals, respectively. The mean PNR (ie, the average number of patients cared for by one nurse per shift) was approximately 7.508, 9.217, and 7.400 in medical centers, regional hospitals, and district hospitals, respectively. These results show that, on average, the number of patients that a nurse cares for during a shift varies within the range of 7.400 to 9.217. The length of stay per admission was 8.653, 9.264, and 14.011 days in medical centers, regional hospitals, and district hospitals, respectively. To ensure the comparability of our TV-IRA across different types of hospitals, standardized data were used to estimate the TVP-VAR model. Panel B of Table 2 presents the results of the break-point unit root test.52 As illustrated in Panel B, regardless of the chosen de-mean or de-trend specification, the null hypotheses of unit root time series were rejected, confirming the stationarity of the time series for all variables used in estimating the TVP-VAR model. The parameter stability tests, assessing the null hypothesis of a time-invariant parameter vector autoregressive model (hereafter, TIP-VAR) against the TVP-VAR model, are outlined in Panel C of Table 2. The Sup-F statistics proposed by Hansen53 yield p-values below the 10% significance level in eight out of nine equations within the three vector autoregressive systems for medical centers, regional hospitals, and district hospitals. Employing the Fisher χ2 method to obtain an aggregate Sup-F statistic54 for the entire vector autoregressive system in medical centers, regional hospitals, and district hospitals resulted in values of 41.834, 58.705, and 36.706, respectively. These generated χ2 values surpass the 1% critical value. Consequently, the null hypothesis of parameter stability in the vector autoregressive system was robustly rejected, validating the use of the TVP-VAR model for evaluating the responses of 14-day readmission rates to changes in nurse staffing levels in acute wards across various types of hospitals. The time plots illustrating all variables described in Table 2 are presented in Figure S1 in the Supplementary.

Table 2 Descriptive Statistics and Parameter Stability Tests for the TVP-VAR Model

The propagation mechanisms illustrating the impact of nurse staffing on inpatient care quality over the time scales of 1–12 months during our study period for medical centers, regional hospitals, and district hospitals are depicted in Figures 1A–C, respectively. As evident in these figures, the cumulative impulse responses of 14-day readmission rates to an increase in the PNR from the 3-month to 12-month time scales were consistently positive for all types of hospitals. These results suggest that a higher burden on nurses caring for patients is associated with poorer inpatient care quality, indicated by higher 14-day readmission rates. Given that Taiwan’s hospital nurse staffing policy aims to reduce PNRs to enhance inpatient care quality, the observed propagation mechanisms simulate the effectiveness of hospitals’ nurse staffing policy over different time scales. Specifically, they indicate the potential positive impact on inpatient care quality if a lower PNR was imposed at all time scales during our study period. It is crucial to note that the impact of the nurse staffing policy is most pronounced in regional hospitals, followed by medical centers, and is least significant in district hospitals, particularly at time scales of 6 months and above during our study period. In addition, the gaps in the response of inpatient care quality to a change in PNR reduced with increasing time scales, specifically from 3 months to 6 months, from 6 months to 9 months, and from 9 months to 12 months for all types of hospitals. These findings suggest that the effectiveness of hospitals’ nurse staffing policy diminishes over time. Furthermore, notable downward trends in the responses of inpatient care quality to a change in PNR were observed in medical centers and district hospitals. In contrast, the responses of inpatient care quality to a change in PNR in regional hospitals remained relatively stable throughout the study period from January 2015 to March 2023.

Figure 1 Time plots for the cumulative response of the 14-day readmission rate to a standard deviation change in the patient-to-nurse ratio at acute care wards within a 12-month timespan.

Notes: Sub-figures (A–C) illustrate the cumulative response of the 14-day readmission rate to a standard deviation change in the patient-to-nurse ratio at acute care wards within a 12-month timespan for medical centers, regional hospitals and district hospitals, respectively.

Panel A of Table 3 provides descriptive statistics for all variables used in estimating the ARDL model. As illustrated in Panel A, the means of the cumulative response of the 14-day readmission rate to a standardized deviation change in PNR in acute care wards within a 12-month timespan are 0.321, 0.626, and 0.288 for medical centers, regional hospitals, and district hospitals, respectively. These results indicate that, on average, a reduction of 0.321, 0.626, and 0.288 standard deviations in the 14-day readmission rate over a 12-month period for medical centers, regional hospitals, and district hospitals, respectively, can be anticipated with a corresponding reduction of one standard deviation in the PNR. The average shares of the population across seven age-specific groups during our observed period ranged from 11.6% (aged 15–24) to 16.1% (aged 35–44). The means of linear and quadratic transformations of age distribution were −4.521 and −17.714, respectively. Real inpatient care reimbursement payments per diem were approximately NTD 11,287, NTD 7,171, and NTD 4,076 for medical centers, regional hospitals, and district hospitals, respectively. The average real monthly regular earnings in the healthcare industry are around NTD 56,557, and the mean score of the composite leading indicator index is 84.087. The time plots illustrating all variables except for the effectiveness of hospitals’ nurse staffing policy described above are presented in Figure S2 in the Supplementary.

Table 3 Descriptive Statistics and Co-Integration Tests for ARDL Model

Panel B of Table 3 presents the results of the break-point unit root test. As illustrated, the cumulative response of the 14-day readmission rate to a standard deviation change in PNR at district hospitals and the composite leading indicator index is identified as a non-stationary (stationary) series when the de-mean (de-trend) specification is used to generate the testing statistics. Nevertheless, the first difference in these time series is found to be stationary, regardless of the chosen de-mean or de-trend specification for generating testing statistics, suggesting that these variables belong to the first-order stationary (I(1)) time series. Furthermore, irrespective of the chosen de-mean or de-trend specification, the null hypotheses of unit root time series are rejected for real inpatient care reimbursement payments per diem at all types of hospitals, linear and quadratic transformations of age distribution, average real monthly regular earnings in the healthcare industry, and the cumulative response of the 14-day readmission rate to a standard deviation change in PNR at medical centers and regional hospitals. Similar results rejecting the null hypothesis of the unit root were obtained when testing the differences in these time series for the unit root property. This confirmation suggests that these time series exhibit level stationary characteristics (I(1)). Hence, some of the variables employed for estimating the ARDL model exhibit level stationary characteristics (I(0)), while others fall into the first-order difference stationary category (I(1)). It is worth noting that the ARDL cointegration methodology can accommodate both I(0) and I(1) time series to test for long-run (ie, cointegrating) relationships.44 Consequently, we proceed with our cointegration analyses under the ARDL model.

Panel C of Table 3 indicates that our ARDL specifications satisfy the assumptions of goodness-of-fit, including independence, homogeneity, and normality in the residuals. Additionally, the upper and lower bounds of critical values for the cointegration relationship were derived from Pesaran et al44 by setting the number of explanatory variables to six . As indicated in Panel C of Table 2, the FPSS statistics used to test the null hypothesis of no cointegration against cointegrating (ie, long-run) relationships among inpatient care reimbursement payments per diem, linear and quadratic transformations of age distribution, average real monthly regular earnings in the healthcare industry, the composite leading indicator index, and the cumulative response of the 14-day readmission rate to a standard deviation change in the PNR at medical centers, regional hospitals, and district hospitals are 4.788, 4.901, and 3.668, respectively. These values exceed the upper bounds of critical values at the 5% (or more rigorous) significance level, strongly confirming the presence of cointegrating (ie, long-run) relationships among the examined time series.

Table 4 reports the ARDL estimates along with the respective estimated long-run coefficients and the effects of demographic changes on the effectiveness of hospitals’ nurse staffing policy (measured by the cumulative response of the 14-day readmission rate to a standard deviation change in PNR) retrieved using Fair and Dominguez’s methods.49 As illustrated in Table 4, the estimated coefficients for the adjustment parameters are significantly negative across the three types of hospitals, indicating that the ARDL models used to investigate cointegrating (ie, long-run) relationships among the examined variables are stably dynamic systems. Furthermore, R2 (adjusted R2) ranges from 95.39% to 97.57% (95.18% to 97.18%) were obtained across three types of hospitals, and all model specification tests for the null hypotheses of independence, homogeneity, and normality in residuals were not rejected (reported in Panel C of Table 3). These results collectively suggest a high level of goodness-of-fit for the data from the ARDL models. Furthermore, we found that the linear and quadratic transformations of age distribution are significantly associated with the effectiveness of hospitals’ nurse staffing policy at medical centers and regional hospitals, whereas this association is insignificant at district hospitals. Inpatient care reimbursement payments per diem are negatively associated with the effectiveness of hospitals’ nurse staffing policy at regional and district hospitals and have no significant impact at medical centers. Real monthly regular earnings in the healthcare industry generate no significant effect on the effectiveness of hospitals’ nurse staffing policy, but business cycles negatively impact the effectiveness of hospitals’ nurse staffing policy at all types of hospitals. The estimated coefficients for the shares of the population are only significant for the share of the population aged 45–54 at district hospitals, but the shares of the population in all age-specific groups have significant impacts on the effectiveness of hospitals’ nurse staffing policy at medical centers and regional hospitals.

Table 4 Effects of Demographical Transitions on Effectiveness of Nurse Staffing Policy

Figures 2A–C further illustrate downward trends in the effect of age distribution on the effectiveness of hospitals’ nurse staffing policy. Specifically, the shares of the population belonging to the two older age groups (the elderly (age ≥ 65) and those aged 55‒64) have negative effects on the effectiveness of hospitals’ nurse staffing policy at medical centers and regional hospitals, while the shares of the population in the other five younger groups are positively correlated with the effectiveness of hospitals’ nurse staffing policy at medical centers and regional hospitals. The downward trend of the effect of age distribution on the effectiveness of hospitals’ nurse staffing policy at district hospitals is found to be insignificant.

Figure 2 Association between demographic structural change and effectiveness of hospital nurse staffing policy.

Notes: Sub-figures (A–C) illustrate the relationship between demographic structural change and effectiveness of hospital nurse staffing policy at medical centers, regional hospitals and district hospitals, respectively.

Discussion Key Results & Policy Implications

The research aim to investigate the relationship between demographic structural changes and the effectiveness of hospital nurse staffing policy. Evidence from Figures 1, 2 and Table 4 indicate a positive association between PNR and 14-day readmission rate across various types of hospitals. Nevertheless, the effectiveness of hospitals’ nurse staffing policy is observed to diminish with population aging, particularly evident in medical centers and regional hospitals. Several policy implications generated from our results merit attention. First, it is noteworthy that the TV-IRA reveals a positive relationship between PNR and the 14-day readmission rate across the three types of hospitals (see Figures 1A–C). These findings suggest that a lower PNR is associated with higher inpatient care quality. Notably, these results not only justify the intention of the nurse staffing policy set by the Taiwan government to improve inpatient care quality through reducing the PNR, but they also align with those from previous studies investigating the association between nurse staffing and healthcare quality.12,18–20,22,25,27,47

Second, considering the significant and positive effects of the shares of the population in the childhood (age < 15) and four major working age groups (ie, aged 15‒24, aged 25‒34, aged 35‒44, and aged 45‒54 groups) on the effectiveness of hospitals’ nurse staffing policy (measured by the cumulative response of the 14-day readmission rate to a standard deviation change in PNR) at medical centers and regional hospitals, one may anticipate that the hospital nurse staffing policy (aiming to enhance inpatient care quality by reducing PNR) would be more effective at these hospitals when the proportions of children and those belonging to the major working age groups expand. Nevertheless, the shares of the population belonging to the two older age groups (55‒64 and the elderly (age ≥ 65)) yield significantly negative effects on the effectiveness of hospitals’ nurse staffing policy, suggesting that the hospital nurse staffing policy would be less effective when the share of those aged 55 years or above increases. These findings diverge from prior research examining the influence of structural age transitions on responses to hospital outpatient copayments. Previous investigations have suggested that population aging enhances the effectiveness of hospital outpatient copayment policies. For instance, Lin et al38 posit that population aging strengthens the effectiveness of outpatient copayment policies in Taiwan due to heightened price sensitivity among the elderly, who tend to utilize outpatient care services a substantial amount. In contrast, it is anticipated that population aging may attenuate the impact of nurse staffing policies. This expectation arises from the premise that an aging population creates increased demand for inpatient care services, thereby necessitating higher levels of nurse staffing in hospitals.

Third, the impact of population aging on diminishing the effectiveness of hospitals’ nurse staffing policy exhibits greater significance in medical centers and regional hospitals compared to district hospitals, as depicted in Figures 2A–C. This discrepancy can be attributed to the specialized roles of regional hospitals and medical centers, catering to tertiary care and handling complex illnesses, respectively.27,47 Consequently, a substantial proportion, approximately 79%, of inpatient care services are delivered by medical centers and regional hospitals, which also employ around 32% of the entire nursing workforce.55 This discernible effect of population aging on the effectiveness of healthcare policies is further corroborated in the context of copayment policies for emergency department visits within medical centers.39 Accordingly, targeted strategic interventions encompassing elements such as ensuring safety in practice, fostering mutual respect, providing adequate facilities, offering necessary support, facilitating continuing education, ensuring an ample supply of resources, maintaining appropriate staffing levels, and ensuring fair remuneration are imperative.56 These interventions should be specifically directed towards medical centers and regional hospitals to effectively address and mitigate the challenges arising from population aging, thereby enhancing effectiveness of hospitals’ nurse staffing policy.

Further Reflection for Policy Implications

It is important to address that our Discussion on the above policy implications should further reflect on the clinical settings of nurse staffing policy and the sensitivity of patient outcome measures. The former should be rooted in well-established evidence on the optimal safe nurse staffing levels per care setting, striking the golden balance between cost-effectiveness and cost-efficiency and satisfying many key decision-makers such as nurses, healthcare institution stakeholders, patients, and caregivers.31,32,57,58 The exclusion of these key decision-makers from nursing workforce policymaking is likely the main reason why the nursing shortage in clinical settings continues to worsen,57,58 despite the implementation of many targeted strategic interventions based on hospital nurse staffing literature.12,18–20,22,25,27,47 Regarding the sensitivity of patient outcome measures, it is recognized that utilizing the 14-day readmission rate as a gauge of inpatient care quality would be intricate due to its susceptibility to various patient characteristics. Factors such as the severity of illness, chronicity of the underlying condition, comorbidity levels, and the extent of social support significantly influence the likelihood of readmission.46

Given that individuals aged 55 years or older are prone to experiencing multiple health issues, thereby increasing the likelihood of readmission, merely increasing the number of nurses for this population may not substantially decrease their readmission rates. Moreover, while individuals in these age groups may frequently seek medical care at medical centers and regional hospitals for managing chronic conditions,55 they are most likely to be readmitted to district hospitals nearer to their residences when necessary. Hence, it is essential to consider a potential caveat in our discussion: population aging may diminish the effectiveness of nurse staffing policies.

Economic Incentives & Business Cycles

The inpatient care reimbursement payments per diem serve as a metric for hospital competition within Taiwan’s healthcare system, which operates under the NHI program. This system utilizes a global budgeting payment scheme to reimburse hospitals for their services.27 Notably, the total number of hospitals fluctuated between 480 and 494, while the total number of medical centers remained constant at 19 throughout the study period.55 Since each type of hospitals competes for its own share of the healthcare budget, this observation implies that competition among medical centers is considerably lower than that among regional and district hospitals. Consequently, the observed negative correlation between the inpatient care reimbursement payments per diem and the effectiveness of hospitals’ nurse staffing policy at regional and district hospitals indicates that heightened competition under the global budgeting payment scheme is likely to undermine the effectiveness of hospitals’ nurse staffing policy. Additionally, it is important to address the fact that the income level of healthcare workers, as measured by real monthly regular earnings in the healthcare industry, does not exert any discernible impact on the effectiveness of hospitals’ nurse staffing policy across all types of hospitals, as evidenced by the results presented in Table 4. Therefore, in the realm of policy instrument choices that significantly influence the effectiveness of hospitals’ nurse staffing policy, priority should be accorded to instruments that shape hospital behavior, such as augmenting reimbursement payments based on PNR, as opposed to individual economic incentives for nurses seeking employment at hospitals.

The markedly negative effects of business cycles (measured using the composite leading indicator) on the effectiveness of hospitals’ nurse staffing policy, as identified in Table 4, suggest that periods of economic prosperity are likely to undermine the effectiveness of hospitals’ nurse staffing policy. This negative association emanates from both the supply side of nurses and the demand side of healthcare services. On the supply side, during economic booms, there is an elevated likelihood of nursing staff transitioning to alternative professions.59 On the demand side, previous studies examining the relationship between business cycles and health conditions indicate that better economic conditions are correlated with poorer health conditions and an increased demand for healthcare services.39,60,61

Limitations of Study

It is imperative to delineate several research limitations to ensure comprehensive understanding of this study. The primary limitation inherent in our investigation pertains to the time-series methodology employed to examine the impact of demographic changes on the effectiveness of hospitals’ nurse staffing policy for inpatient care. It is crucial to acknowledge that the age distribution and inpatient care quality utilized in our analyses are based on aggregated data. Consequently, the empirical results derived from our models should not be extrapolated to individual behavior changes at specific ages in response to PNR adjustments in hospitals, in order to mitigate the risk of ecological fallacies.62 Second, despite the frequent use of the 14-day readmission rate as an indicator of inpatient care quality,27,45,46 it is important to recognize that the likelihood of 14-day readmission may be particularly sensitive within the older population, who are more likely to experience severe illness, multiple chronic conditions, and pronounced comorbidity levels.46 Consequently, we have advocated for prudence in our discussion regarding the interplay between demographic structural changes and the effectiveness of nurse staffing policy for inpatient care.

Thirdly, beyond demographic changes, numerous factors such as the physical and mental health status of nursing staff, working environment conditions, job-seeking behaviors or preferences and cost of inpatient care quality can influence the effectiveness of hospitals’ nurse staffing policy. Nevertheless, the time-series data for these variables are not available in the research databases in Taiwan, posing a constraint on the scope of this study. Fourth, similar to other studies in nursing workforce research, this study lacks the capacity to explore the concurrent dynamics of nurse staffing, quality of care, and costs across a continuum of change that determines the optimal safe nurse staffing level. Recent investigations conducted by Park et al have adopted Park’s sweet spot theory-driven Artificial Intelligence Algorithm to model the optimal safe nurse staffing zone, which aims to mitigate the limitations of traditional nursing workforce research.31,32,57,58 This research paradigm, by emphasizing patient outcomes or care quality alongside cost considerations, addresses the shortcomings that have hindered the effective implementation of knowledge into practice and policy.

Directions for Future Research

Based on the aforementioned discussions, it is recommended that future research endeavors focus on two main aspects. First, the application of Park’s sweet spot theory-driven Artificial Intelligence Algorithm30–32,57,58 should be employed to explore the dynamic interactions among nurse staffing, quality of care, and costs across a continuum of change within Taiwan’s healthcare system. Second, collecting individual-level data is essential to examine the intricate interactions among hospitals’ managerial actions affecting the quality of care, nurses’ decisions regarding employment at hospitals and patient outcomes in response to nurse staffing policies within Taiwan’s healthcare system.

Conclusion

The primary goal of the hospital nurse staffing policy in Taiwan is to enhance inpatient care quality by increasing the nursing staff level in hospitals. Nevertheless, Taiwan’s healthcare system is confronted with significant challenges due to the rapid aging of the population and a sustained decline in birth rates. This dual challenge results in an increasing demand for healthcare services along with a reduction in the nursing workforce, potentially impacting the effectiveness of hospitals’ nurse staffing policy. In response to these challenges, this study pioneers the use of TVP-VAR and ARDL models to investigate the impact of demographic changes on the effectiveness of hospitals’ nurse staffing policy in Taiwan. The time-varying impulse responses from the TVP-VAR model demonstrate a significantly positive association between PNR and 14-day readmission rate across diverse hospital sizes, encompassing medical centers, regional hospitals, and district hospitals. The ARDL model establishes long-term relationships among the effectiveness of hospitals’ nurse staffing policy (quantified by the cumulative response of inpatient care quality to adjustments in nurse staffing levels), age distribution, hospital competition, health worker income, and business cycles across all types of hospitals. An essential finding regarding these long-run relationships is the observed diminishing effectiveness of hospitals’ nurse staffing policy with population aging, particularly evident in medical centers and regional hospitals. To achieve the targeted objective of the hospital nurse staffing policy, aimed at improving inpatient care quality by reducing the PNR, it is strongly recommended, especially in an aging society, to institute mandated PNR in medical centers and regional hospitals that are significantly lower than those determined without considerations for demographic structural changes.

Ethics

The data collection process was approved by the Research Ethics Committee of Taichung Tzu Chi Hospital with the Certificate of Exempt Review ID: REC110-23.

Acknowledgments

This research was funded by the National Science and Technology Council in Taiwan for the research project entitled “Now-casting for the effect of nurse staffing on quality of inpatient care during the COVID-19 pandemic outbreak: Evidence from Taiwan” with Grant No. 110-2410-H-025-012.

Disclosure

The authors report no conflicts of interest in this work.

References

1. ICN. Influencing health policy to protect and respect nurses’ rights and improve health. International Council of Nurses; 2022. Available from: https://www.icn.ch/resources/publications-and-reports/icn-annual-report-2022-influencing-health-policy-protect-and. Accessed February28, 2024.

2. Drennan VM, Ross F. Global nurse shortages—the facts, the impact and action for change. Br Med Bul. 2019;130(1):25–37. doi:10.1093/bmb/ldz014

3. Finlayson B, Dixon J, Meadows S, Blair G. Mind the gap: the extent of the NHS nursing shortage. BMJ. 2002;325(7363):538–541. doi:10.1136/bmj.325.7363.538

4. Fleming P, Thomas S, Williams D, Kennedy J, Burke S. Implications for health system reform, workforce recovery and rebuilding in the context of the Great Recession and COVID-19: a case study of workforce trends in Ireland 2008-2021. Human Res Health. 2022;20(1):48. doi:10.1186/s12960-022-00747-8

5. Girardeau M. Nurses’ occupational turnover in the Nordics: a mixed method, systematized review; Oslo, Norway: Oslo Metropolitan University; 2022. Available from: https://oda.oslomet.no/oda-xmlui/bitstream/handle/11250/3018075/Girardeau_SIW_2022.pdf?sequence=1&isAllowed=y. Accessed February28, 2024.

6. Reiff E, Gade C, Böhlich S. Handling the shortage of nurses in Germany: opportunities and challenges of recruiting nursing staff from abroad, IUBH Discussion Papers; Human Resources (3/2020); 2020. Available from: https://www.econstor.eu/bitstream/10419/222921/1/1726039005.pdf. Accessed February28, 2024.

7. Boamah SA, Callen M, Cruz E. Nursing faculty shortage in Canada: a scoping review of contributing factors. Nursing Outlook. 2021;69(4):574–588. doi:10.1016/j.outlook.2021.01.018

8. Yoshioka-Maeda K, Iwasaki-Motegi R, Honda C. Resolution for the severe shortage of nurses for combating COVID-19 in Japan. Asia Pacif J Public Health. 2023;35(4):320–321. doi:10.1177/1010539520982489

9. Teng SW. Nursing manpower and solutions in Taiwan. Hu Li Za Zhi. 2014;61(2):5–12. doi:10.6224/JN.61.2.5

10. Haddad LM, Annamaraju P, Toney-Butler TJ. Nursing Shortage; Treasure Island (FL), USA: StatPearls Publishing; 2023. Available from: https://www.ncbi.nlm.nih.gov/books/NBK493175/. Accessed February28, 2024.

11. Nawagi F, Kneafsey R, Modber M, Mukeshimana M, Ndungu C, Bayliss-Pratt L. An overview of nursing and midwifery leadership, governance structures, and instruments in Africa. BMC Nurs. 2023;22(1):168. doi:10.1186/s12912-023-01336-3

12. Butler M, Schultz TJ, Halligan P, et al. Hospital nurse-staffing models and patient- and staff-related outcomes. Cochrane Database Syst Rev. 2019;4(4):CD007019. doi:10.1002/14651858.CD007019.pub3

13. Everhart D, Neff D, Al-Amin M, Nogle J, Weech-Maldonado R. The effects of nurse staffing on hospital financial performance: competitive versus less competitive markets. Health Care Manage Rev. 2013;38(2):146–155. doi:10.1097/HMR.0b013e318257292b

14. Shin S, Park JH, Bae SH. Nurse staffing and nurse outcomes: a systematic review and meta-analysis. Nursing Outlook. 2018;66(3):273–282. doi:10.1016/j.outlook.2017.12.002

15. Chen C, Meier ST. Burnout and depression in nurses: a systematic review and meta-analysis. Internat J Nurs Stud. 2021;124:104099. doi:10.1016/j.ijnurstu.2021.104099

16. Marufu TC, Collins A, Vargas L, Gillespie L, Almghairbi D. Factors influencing retention among hospital nurses: systematic review. Br J Nurs. 2021;30(5):302–308. doi:10.12968/bjon.2021.30.5.302

17. Du J, Zhang L, Xu C, Qiao J. Relationship between the exposure to occupation-related psychosocial and physical exertion and upper body musculoskeletal diseases in hospital nurses: a systematic review and meta-analysis. Asian Nurs Res. 2021;15(3):163–173. doi:10.1016/j.anr.2021.03.003

18. Twigg DE, Kutzer Y, Jacob E, Seaman K. A quantitative systematic review of the association between nurse skill mix and nursing-sensitive patient outcomes in the acute care setting. J Adv Nurs. 2019;75(12):3404–3423. doi:10.1111/jan.14194

19. Griffiths P, Recio-Saucedo A, Dall’Ora C, et al.; Missed Care Study Group. The association between nurse staffing and omissions in nursing care: a systematic review. J Adv Nurs. 2018;74(7):1474–1487. doi:10.1111/jan.13564

20. Van T, Annis AM, Yosef M, et al. Nurse staffing and healthcare-associated infections in a national healthcare system that implemented a nurse staffing directive: multi-level interrupted time series analyses. Internat J Nurs Stud. 2020;104:103531. doi:10.1016/j.ijnurstu.2020.103531

21. Drennan J, Murphy A, McCarthy VJC, et al. The association between nurse staffing and quality of care in emergency departments: a systematic review. Internat J Nurs Stud. 2024;153:104706. doi:10.1016/j.ijnurstu.2024.104706

22. Numata Y, Schulzer M, van der Wal R, et al. Nurse staffing levels and hospital mortality in critical care settings: literature review and meta-analysis. J Adv Nurs. 2006;55(4):435–448. doi:10.1111/j.1365-2648.2006.03941.x

23. Blatter C, Osińska M, Simon M, Zúñiga F. The relationship between nursing home staffing and resident safety outcomes: a systematic review of reviews. Internat J Nurs Stud. 2024;150:104641. doi:10.1016/j.ijnurstu.2023.104641

24. Clemens S, Wodchis W, McGilton K, McGrail K, McMahon M. The relationship between quality and staffing in long-term care: a systematic review of the literature 2008-2020. Internat J Nurs Stud. 2021;122:104036. doi:10.1016/j.ijnurstu.2021.104036

25. Aiken LH, Sloane DM, Bruyneel L, et al. Nurse staffing and education and hospital mortality in nine European countries: a retrospective observational study. Lancet. 2014;383(9931):1824–1830. doi:10.1016/S0140-6736(13)62631-8

26. Shin S, Park JD, Shin JH. Improvement Plan of Nurse Staffing Standards in Korea. Asian Nurs Res. 2020;14(2):57–65. PMID: 32305508. doi:10.1016/j.anr.2020.03.004

27. Chen WY. On the relationship among nurse staffing, inpatient care quality and hospital competition under the global budget payment scheme in Taiwan’s National Health Insurance system: mixed Frequency VAR analyses. Systems. 2022;10(5):187. doi:10.3390/systems10050187

28. de Cordova PB, Pogorzelska-Maziarz M, Eckenhoff ME, McHugh MD. Public reporting of nurse staffing in the United States. J Nurs Regul. 2019;10(3):14–20. doi:10.1016/S2155-8256(19)30143-7

29. Juraschek SP, Zhang X, Ranganathan V, Lin VW. United States registered nurse workforce report card and shortage forecast. Am J Med Qual. 2019;34(5):473–481. doi:10.1177/1062860619873217

30. JEC REPUBLICANS. The consequences of declining fertility for social capital. Joint Economic Committee Republicans; 2022. Available from: