Economic evaluation of nurse staffing and nurse substitution in health care: A scoping review

Economic evaluation of nurse staffing and nurse substitution in health care: A scoping review

International Journal of Nursing Studies 48 (2011) 501–512 Contents lists available at ScienceDirect International Journal of Nursing Studies journa...

185KB Sizes 0 Downloads 15 Views

International Journal of Nursing Studies 48 (2011) 501–512

Contents lists available at ScienceDirect

International Journal of Nursing Studies journal homepage: www.elsevier.com/ijns

Review

Economic evaluation of nurse staffing and nurse substitution in health care: A scoping review Yevgeniy Goryakin, Peter Griffiths *, Jill Maben King’s College London, National Nursing Research Unit, United Kingdom

A R T I C L E I N F O

A B S T R A C T

Article history: Received 6 March 2010 Received in revised form 15 July 2010 Accepted 26 July 2010

Objective: Several systematic reviews have suggested that greater nurse staffing as well as a greater proportion of registered nurses in the health workforce is associated with better patient outcomes. Others have found that nurses can substitute for doctors safely and effectively in a variety of settings. However, these reviews do not generally consider the effect of nurse staff on both patient outcomes and costs of care, and therefore say little about the cost-effectiveness of nurse-provided care. Therefore, we conducted a scoping literature review of economic evaluation studies which consider the link between nurse staffing, skill mix within the nursing team and between nurses and other medical staff to determine the nature of the available economic evidence. Design: Scoping literature review. Data sources: English-language manuscripts, published between 1989 and 2009, focussing on the relationship between costs and effects of care and the level of registered nurse staffing or nurse–physician substitution/nursing skill mix in the clinical team, using cost-effectiveness, cost-utility, or cost–benefit analysis. Articles selected for the review were identified through Medline, CINAHL, Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects and Google Scholar database searches. Review methods: After selecting 17 articles representing 16 unique studies for review, we summarized their main findings, and assessed their methodological quality using criteria derived from recommendations from the guidelines proposed by the Panel on CostEffectiveness in Health Care. Results: In general, it was found that nurses can provide cost effective care, compared to other health professionals. On the other hand, more intensive nurse staffing was associated with both better outcomes and more expensive care, and therefore cost effectiveness was not easy to assess. Conclusions: Although considerable progress in economic evaluation studies has been reached in recent years, a number of methodological issues remain. In the future, nurse researchers should be more actively engaged in the design and implementation of economic evaluation studies of the services they provide. ß 2010 Elsevier Ltd. All rights reserved.

Keywords: Workforce Nurse staffing Skill mix Substitution Economic evaluation Cost-effectiveness Nurse practitioners Nurse-led care

What is already known about the topic?

* Corresponding author at: Professor of Health Services Research, School of Health Sciences, University of Southampton, United Kingdom. Tel.: +44 020 7848 3012. E-mail address: peter.griffi[email protected] (P. Griffiths). 0020-7489/$ – see front matter ß 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijnurstu.2010.07.018

 Most existing literature reviews of studies on the nurse provided-care have focused on the effect of nurses on patient outcomes.  The general finding has been that greater nurse staffing as well as greater proportion of registered nurses in the health workforce is associated with better patient outcomes.

502

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

 However, such studies generally do not consider the effect of nurse staff on both patient outcomes and costs of care, and therefore say little about the cost-effectiveness of nurse-provided care. What this paper adds  In general, registered nurses can provide cost effective care, compared to other health professionals.  On the other hand, more intensive nurse staffing was associated with both better outcomes and more expensive care, and therefore cost effectiveness was not easy to assess.  There is a need for further rigorous economic evaluation of the effect of variation and change in nurse staffing levels and nurse–physician substitution/changes in nursing skill mix.

1. Introduction Nurses comprise the largest single group in the health care workforce in many countries, and account for a significant proportion of all health expenditure. In the late 1990s it was estimated that spending on nursing and midwifery staff contributed almost half of all spending on healthcare in the UK (Jenkins-Clarke, 1999; Richardson, 1999). In recent years, there has been an increasing interest in demonstrating the impact of this crucial part of the health care workforce. Studies have examined the impact on patient outcomes of variation in the numbers of registered nurses, skill mix changes within the nursing workforce and skill mix changes in the wider clinical team involving substitution between doctors and registered nurses. A number of reviews have examined the extensive literature which explores the relationship between nurse staffing levels and patient outcomes (e.g. Kane et al., 2007; Lang et al., 2004; Lankshear et al., 2005). There is an increasing amount of evidence that higher levels of registered nurse staffing and a higher proportion of RNs in the workforce may be associated with reduced mortality and other improvements in patient outcome, although results are not entirely consistent and significant uncertainty remains (Griffiths, 2009). This work has generally been undertaken in acute care settings, although other settings, including nursing homes and primary care have also been studied, with positive associations shown between higher levels of nurse registered staffing and outcomes, including pressure sore incidence and control of diabetes (e.g. Griffiths et al., 2010; Konetzka et al., 2008). There is also an increasing number of studies examining changes in the skill mix of the workforce involving nurse for doctor substitution. These studies generally involve nurses managing an aspect of patient care or a group of patients where medicine (qualified doctors) has hitherto been the lead profession. Most of this research involves the use of nurse practitioners in primary care, but studies have also investigated nurse for doctor substitution in other settings including inpatient care. These too have previously been reviewed (Griffiths et al., 2007; Laurant et al.,

2004) and, broadly, suggest that appropriately trained nurses can deliver quality and outcomes equivalent to doctors. The motivation for exploring nurse for doctor substitution has been in large part economic and has been supported by policy in many countries including the UK. Shifting care from doctors to nurses is believed to provide an answer to physician scarcity and to act as a mechanism for cost containment (Laurant et al., 2004). On the other hand, there is also a widely acknowledged global shortage of registered nurses (Buchan and Calman, 2004; ICN, 2006; WHO, 2006) and thus the opportunity costs of using nurses to deliver what has previously been medical care might be high. Despite the economic underpinnings of decision making in this area, there are few economic evaluations, even though without such evidence, it is hard to make an informed choice about allocating money to alternative healthcare programmes (Vincent, 2002). There have been some reviews (e.g. Allred et al., 1998; La¨ma˚s et al., 2009; Siegel, 1998; Thungjaroenkul et al., 2007), but most are dated or focus on nursing intervention in general (i.e., without considering skill mix or workforce changes), or have little focus on the methodological quality of the economic component of the studies. 2. Economic issues arising While lower patient-to-nurse ratios may contribute to better health care and thus lead to better patient outcomes (Aiken, 2008), the costs of this strategy are unclear. The picture is complex. More nurse hours per patient may contribute to higher per patient expenditures, although this may be offset by a reduction in hospital costs through a reduction in the length of stay or adverse outcomes experienced by patients (Dall et al., 2009). As a result, costs per patient may fall or rise, and while the patient outcomes are likely to improve, the cost effectiveness (i.e., the ratio of the difference between costs and effects of two alternatives) can be unclear. In addition, better nurse staffing may improve nurse retention and thus lead to less nurse stress and better treatment outcomes because of more attention to the patients, fewer medical errors and lower recruitment and training expenditures by hospitals (Aiken, 2008; Greenberg, 2006), implying a lower cost effectiveness ratio, and therefore greater value for money. From a wider, societal perspective, additional benefit from better patient outcomes may be higher productivity from the (former) patients, higher wages, and therefore additional cost offsets. Moreover, better nurse care may lead to lower future medical expenditure per patient (for example, in primary care after discharge), as well as lower indirect costs of care such as the travel costs and lost productivity while attending appointments, or less time spent by family members helping their sick relatives. Such effects, usually long-term, may be unaccounted for in traditional cost-effectiveness analysis. Some reviews have found that at greater nurse staffing levels, the positive contribution of additional nurse staffing on patient outcomes appears to become smaller (Kane et al., 2007; Lang et al., 2004), which suggests the existence of optimal level of nurse staffing, not necessarily the one

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

that leads to the best treatment outcomes (Spetz, 2005). Such levels may depend on various patient, hospital and staff characteristics, and therefore may be context-specific (Curtin, 2003). For example, almost all studies suffer from a failure to consider levels of medical staff as a variable even though these may be highly correlated with levels of nurse staffing (Griffiths, 2009). Furthermore, some researchers have found that increases in nurse–patient ratios may be linked with little change in patient outcomes, as described in a study exploring the effect of mandated ratios in California (Hall and Buch, 2009). This does not prove that such interventions are not cost effective, since it is possible that the costs per patient for greater nurse staffing could in fact be lower if there were certain economies of scale and scope in service provision. Moreover, the studies where the main focus of interest is patient health outcomes may miss the indirect cost offsets mentioned above. Therefore, information on both costs and effects should be collected in appropriately conducted economic evaluations. As far as skill mix and nursing staff substitution is concerned, again the link with cost-effectiveness is ambiguous. On the one hand, it is possible that experienced and properly trained nurse practitioners can provide services similar in quality to those provided by physicians, particularly for less complicated medical conditions, while also costing less per unit of time worked, especially taking into account their training costs and lower salary (McGrath, 1990). If this is true, it would be economically rational to delegate a certain proportion of physicianprovided services to nurse practitioners. In some countries, for example the United Kingdom, there has been significant support for such changes at a policy level, with enabling measures such as nurse prescribing implemented over recent years (Latter et al., 2007). However, cost savings depend on the salary differentials between nurses and doctors, and those may be offset by lower productivity of nurses (Laurant et al., 2004). For example, nurses may take longer to treat patients than physicians and use more resources to provide similar quality and quantity of care. Nurse-provided consultations may take longer, nurse practitioners may require additional supervision from doctors (especially in more difficult cases), and they may order more tests and investigations, all of which would increase total costs per patient treated. While some inferences can be drawn from existing research on the link between nurse staffing models and patient outcomes, direct economic evaluations in which both costs and consequences of alternative workforce models are studied simultaneously are required. This evidence may be of use in understanding which medical staffing models offer the greatest value for money. Therefore we undertook a scoping review of economic evaluations of nurse staffing, including comparisons between different levels of nurse staffing and the substitution of nurses for other healthcare workers. 3. Method We systematically searched for studies that explored the relationship between costs and effects of care and the

503

level of registered nurse staffing or nursing skill mix in the clinical team (including nurse for doctor substitution) using cost-effectiveness, cost-utility (i.e., cost effectiveness analysis, with the compared alternatives having a common health outcome variable-quality adjusted life years), or cost–benefit analysis (i.e., comparison between costs and benefits of a given intervention, expressed in a common monetary term). In those cases where the analytical framework was not explicitly stated, we nevertheless included a study when it simultaneously considered both costs and effects of care as outcome variables. We excluded studies that focused on association between nurse staffing variables and patient outcomes only, without considering costs of care. We also excluded studies that looked at nursing costs only. We did not consider interventions performed by nurses when compared to no intervention or interventions performed by nurses in different settings (e.g. at leg ulcer clinics vs. care at home) because our main interest is in comparing the nurses with other workforce, and not in comparing the effect of specific interventions or settings where nurses work. We did not include economic evaluations of care using different service configurations which included changes in the composition of the clinical team in addition to other variation in care delivery (e.g. in clinical nurse specialist care vs. inpatient multidisciplinary team care; on site vs. off-site care; nursing home vs. hospital care), because in such cases it is difficult to tease out the contribution of the health workforce from the effect of different settings and services. We searched Medline, CINAHL, Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects and Google Scholar using combinations of the following freetext and MeSH terms: nursing staff, medical staff; economic evaluation, cost–benefit, cost utility, cost effectiveness, staff mix, and skill mix in articles’ titles and abstracts. Where appropriate, we used truncation symbols, e.g. nurs*, effective*, staff*. We limited our search to articles published in English, in 1989–2009. As this is a scoping review, we made no a priori selection for quality. We used established criteria for judging the quality of economic studies derived from recommendations from the guidelines proposed by the Panel on Cost-Effectiveness in Health Care (Drummond et al., 2005). Specifically, the match of the articles against the criteria listed in Table 1 was considered.

4. Results 4.1. Inclusion criteria and study types 1050 citations were retrieved from Medline. CINAHL searches returned 189 articles. Similar searches were conducted at Google Scholar1 and Cochrane Reviews databases, as well as at the Database of Abstracts of Reviews of Effects. Search at the last two databases did not return any additional articles not already found through

1 Since Google Scholar returned a vast number of search results, only the first 100 links for any specific search term were checked.

504

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

Table 1 Quality criteria. Was a societal perspective adopted by authors? Were least biased research design methods used? Did effectiveness estimates incorporate both benefits and harms? Was mortality and morbidity combined using QALYs? Was detailed description of costs given? Was comparison with existing practice/viable low-cost alternative given?  Was discounting undertaken at appropriate rates?  Was sensitivity analysis for the main estimated parameters of interest (preferably multiway) performed?  Was incremental cost-effectiveness ratio (ICER) presented, and compared with other appropriate interventions?      

Source: adapted from ‘‘Methods for the Economic Evaluation of Health Care Programmes’’ by Drummond et al., Third Edition, p. 46.

the first three database searches. Seventeen articles reporting sixteen studies met the criteria for inclusion into our review. Two papers were based on the same underlying study (Cowan et al., 2006; Ettner et al., 2006), although the analysis performed was different. Ten studies were based on studies conducted in USA, four in the UK, one in Canada and one Australia. The majority of studies (11) focused on nurse–physician substitution/nursing skill mix variables, while the remaining five considered nurse staffing levels. Most studies (seven) were conducted in acute/critical/intermediate care. They were followed by general hospital care (3); by neonatal/prenatal care (3); primary care (1) and other sectors (A&E departments and anticoagulant care). See Tables 2 and 3 for full summaries. In all skill mix/staff substitution studies, the main approach was to compare the outcomes between treatment and control groups in an experimental/quasiexperimental framework where skill mix of staff varied between groups. Primarily skill mix was considered in terms of formal qualifications or employment grades (e.g. nurse/doctor). In the staffing studies, nurse labour inputs were variously measured as RN hours per patient day; median and recommended staffing levels for nurses; proportion and number of nurse hours worked and nurse to patient ratios. 4.2. Perspective In fifteen articles, the perspective chosen for economic analysis was not reported, so we had to deduce it from the information given. The recurring weakness in most studies was the lack of societal perspective. Only in two cases (Dall et al., 2009; Mitchell-DiCenso et al., 1996) was a societal perspective the methodological approach of choice; in others, the perspective was either provider’s, or payer’s. Since various treatment-related costs and benefits may accrue outside of health care system, ignoring them may distort the picture on the overall societal costs from a given healthcare intervention. 4.3. Methodological issues with study designs Six of sixteen studies employed randomized control trials (RCTs) as the design of choice (Cowan et al., 2006; Ettner et al., 2006; Griffiths et al., 2001; Mitchell-DiCenso

et al., 1996; Pioro et al., 2001; Richardson et al., 2001; Venning et al., 2000), although in two of them (Cowan et al., 2006; Ettner et al., 2006), staff rather than patients were randomized between wards. Although there are inherent strengths in the causal inferences that can be drawn from this research design, it was impossible for RCTs of this type to be blinded, for obvious reasons. Also, even in randomized trials, there may still be remaining factors besides pure differences in staff composition (for example, nurses may still be supervised by GPs or medical director to a certain extent), which might have been the case in all studies reviewed here. It should also be noted that no study on the cost effectiveness of different nurse staffing levels relied on randomized trial data. The studies with non-RCT-based design relied mostly on observational, retrospective data, where bias was controlled through the use of regression analysis. However, it is challenging to adjust for a differential patient case mix between sites, let alone for a host of hospital and other patient-level unobservable factors that may confound the observed relationship between the variables of interest. For example, teaching hospitals that use advanced technology may employ more RNs (Curtin, 2003), and a ‘‘toxic’’ work environment may be an underlying issue in hospitals where there are fewer nurses per patient (Curtin, 2003). Geographical settings may also differ considerably. In three studies (Dall et al., 2009; Rothberg et al., 2005; Rothschild et al., 2009), the link between nurse staffing levels and patient outcomes/total costs was measured indirectly, by estimating the probability of adverse events/ length of stay (LOS) based on patient-to-nurse ratios, and then linking this with cost savings from avoided adverse events/shorter LOS. In the future, to avoid making unnecessary assumptions, it may be better to model the relationship between staff variables and patient outcomes and costs directly. One of the quality criteria chosen for this review (see Table 1) is that effectiveness estimates should incorporate both benefits and harms to health arising from a given intervention. None of the studies reviewed here explicitly did that. Moreover, of all studies reviewed, only one (Ganz et al., 2005) followed the recommendation that changes to mortality and morbidity should be measured by the changes in quality adjusted life years (QALYs), and three (Bissinger et al., 1997; Mitchell-DiCenso et al., 1996; Pioro et al., 2001) considered both morbidity and mortality without combining them into a single indicator. It should be noted, however, that QALYs may not always be correlated with nurse-sensitive indicators such as patient satisfaction with care, especially in supportive and palliative care (Douglas and Normand, 2005), therefore using other measures is probably justified. For example, several studies paid attention to quality of care indicators (including patient satisfaction with care) as an outcome variable (Mitchell-DiCenso et al., 1996; Pioro et al., 2001; Pratt et al., 1993; Taylor et al., 1997; Venning et al., 2000). Some other authors (Pratt et al., 1993; Sakr et al., 2003; Taylor et al., 1997) used intermediate, or process-related measures of care (e.g. process errors for minor injury units; proportion of time patients spent in therapeutic range), because significantly adverse events

Table 2 Nurse-physician substitution and nursing skill mix studies. Authors

Location and medical area

Inpatient settings Bissinger 36 bed neonatal et al., 1997 intensive care unit, USA

Intervention/control

Outcomes (costs and effectiveness)

Perspectivea

Results

Retrospective study of 2 matched groups of 70 critically ill, 24 months old infants.

Quality of care, LOS, morbidity, mortality, composite quality of care index. Charges (proxy for cost of care)

Hospital

No differences in LOS, morbidity, mortality. Per patient NNP charges usually lower for all categories of charges.

Effectiveness: LOS, mortality, and readmission 4 months after discharge Costs: hospital costs

Hospital

Average LOS lower for patients in the experimental group. No significant differences in mortality or readmissions. Significantly lower costs in treatment group per patient. Intervention costs were $1187 per patient and associated with a significant $3331 reduction in non-intervention costs. Because health outcomes were comparable for the 2 groups, the intervention was considered cost-effective No effect on health outcome measures, but LOS longer for treatment group. Even though cost per patient day was lower for the treatment group, since LOS was longer for them, greater total patient costs in the treatment group. Fewer deaths in CNS/NP team; very little difference in costs and other outcomes. Overall, inconclusive evidence.

USA, 610 bed tertiary hospital

Comparative, 2-group, quasi-experimental design. 1207 patients.

Ettner et al., 2006

USA, 610 bed tertiary hospital

Comparative, 2-group, quasi-experimental design. 1207 patients.

C: 626 patients in usual care management group. I: 581 patients in NP group

Effectiveness: as measured in Cowan et al. (2006) study; quality of care. Costs: Intervention costs were compared to the difference in non-intervention costs.

Hospital

Griffiths et al., 2001

Intermediate care patients in acute hospital wards (UK)

RCT, 175 patients.

I: nursing led inpatient units (89 patients) with no routine medical intervention. C: usual (consultant-managed) units (87 patients)

Effectiveness: LOS, Barthel index Costs: direct healthcare costs (bottom-up; discounted bottom-up; cost per bed day)

Hospital

Mitchell et al., 1997

Neonatal intensive care, Canada

RCT, 821 patients.

I: 414 randomized to care by clinical nurse specialist/ neonatal practitioner team. C: 407 care by pediatric residence team

Societal

Pioro et al., 2001

General medical ward in an academic teaching hospital in Midwestern USA

RCT, 381 patients.

I: 193 patients treated by NPs and medical director. C: 188 patients treated by medical house staff

Effectiveness: mortality; neonatal complications; length of stay; quality of care; long-term outcomes. Costs: hospital cost per infant, plus family costs. Effectiveness: LOS; complications; 30-day mortality; patient assessments of care; changes in activities of daily living; symptom severity. Costs: hospital charges; costs (for specific departments only).

Pratt et al., 1993

Acute medical and acute surgical care in Australian hospital

Retrospective study on the difference in costs and outcomes in 2 hospital wards. 393 acute medical and acute surgical care patients in Australian hospital.

Comparison of staffing patterns in two hospital wards staffed as follows: - all-RN staff - mix between 80% RN and 20% EN

Effectiveness: subjective judgment of quality of nursing care by patients; correctness and accuracy of records; independent observation on quality of care. Costs: daily/hourly costs of care per patient

Hospital

Hospital

Similar LOS, charges, in-hospital complications. Similar health outcomes. Results similar in both intention to treat, and actual treatment analysis. No clear evidence of what is more cost-effective. Little difference in patient outcomes. However, in the acute medical ward, RN staff was more expensive; in the acute surgical ward the staff mix was more costly. Overall, mixed evidence.

505

Cowan et al., 2006

I: neonatal care received by 35 infants from neonatal nurse practitioners. C: neonatal care received by 35 infants from medical house staff C: 626 patients in usual care management group I: 581 patients in NP group

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

Research design; participants

506

Table 2 (Continued ) Location and medical area

Research design; participants

Intervention/control

Outcomes (costs and effectiveness)

Perspectivea

Results

Richardson et al., 2001

Intermediate care patients in acute hospital wards (UK)

RCT, 177 patients.

I: 97 patients in nursing led inpatient unit with no routine medical intervention. C: 80 patients in usual (consultant-managed) care units.

Effectiveness: Goldberg General Health Questionnaire (GHQ); Barthel index. Costs: consultations, drugs, cost per bed day. Post-discharge cost data was included, but for 1 month only.

Payer

No effect on health outcome measures, but LOS longer for treatment group. Greater total costs in the experimental group. However, the post-discharge costs were significantly lower for the treatment group. If maintained for 5–10 months, it could cancel out greater hospital costs for this group.

Non-experimental, retrospective data; interviews. 156 participant women.

Comparison between three staffing models explored: physician based; mixed staffing; clinical nurse specialists.

Effectiveness: several neonatal-maternal physiological variables; Kessner index. Cost: total cost per patient visit.

Hospital

Before after cohort study: nurse-led minor injury unit replacing A&E department. 2762 patients. Sequential retrospective study, 2 hospital sites in the UK. Patients divided into 2 groups: A (newly referred) and B (randomly identified). 430 patients altogether. RCT, 1303 patients.

Comparison of care provided to 1447 patients in A&E department, and to 1315 nurse-led injury unit patients. In groups A and B, further comparison between patients treated in nurse specialist care and consultant care was made

Effectiveness: process errors; waiting times. Costs: Per minor injury case. For estimating total costs, both workload unit (i.e., patient costs), and follow-up costs of treatment were included. Effectiveness: mean proportion of time patients spent on therapeutic range; number of adverse dug interactions; patient satisfaction with service provision. Cost: total treatment-related costs.

Provider

No difference in health outcomes. Nurse-staffed clinic had the greatest satisfaction and lowest cost per visit and thus highest cost effectiveness (on this dimension). Significantly fewer process errors and better waiting times at nurse-led units. However, costs were greater at nurse units. No formal ICER ratios derived.

Comparison between care provided by nurse practitioners (n = 641 patients) and by General Practitioners (n = 651 patients).

Effectiveness: patient satisfaction, health status. Costs: visit costs

Outpatient settings Gravely Three prenatal et al., 1992 care clinics, USA

Sakr et al., 2004

Accident and Emergency department, UK

Taylor et al., 1997

Anticoagulant service, UK

Venning et al., 2000

Primary care centers, UK

Provider

No difference in total costs between 2 clinics. No difference in time patients spent in therapeutic range, but greater satisfaction of patients in the treatment group.

Provider

No difference in health outcomes and in health service costs; greater patient satisfaction for nurses. Costs were similar because of longer consultation times, tests and return times for nurse-provided care.

Abbreviations: A&E: accident and emergency; CNS: clinical nurse specialist; GP: general practitioner; EN: enrolled nurse; ICER: incremental cost effectiveness ratio; LOS: length of stay; NNP: neonatal nurse practitioner; NP: nurse practitioner; QALYs: quality adjusted life years; RCT: randomized control trial; RN: registered nurse. a Here, perspective refers to the viewpoint on which costs and benefits of compared alternatives should be considered relevant.

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

Authors

Table 3 Nurse staffing studies. Location and medical area

Research design/ participants

Intervention/control

Outcomes (costs and effectiveness)

Perspective

Results

Dall et al., 2009

USA, nonfederal acute care hospitals

Staffing variable: RN hours per patient day.

Effects: Savings from avoided, nosocomial complications, shorter LOS, greater national productivity. Costs: annual cost of employing a nurse

Societal

As nurse staffing levels increase, patient risk of nosocomial complications and hospital length of stay decrease, resulting in medical cost savings. Diminishing effect at higher levels.

Ganz et al., 2005

Skilled nursing facilities for post-acute care, USA

Regression analysis and literature review of hospital discharge data from the 2005 Nationwide Inpatient Sample. Markov cohort simulation of recommended versus median staffing post-acute care patients.

Effectiveness: life expectancy, quality-adjusted life expectancy. Costs: hospital costs. Nursing costs aggregated per patient day.

Insurer

ICER of recommended versus median staffing was $321,000 per discounted quality-adjusted life year gained.

Needleman et al., 2006

Acute general care hospitals, USA

Regression analysis of observational data from 799 hospitals, with 5,075,969 medical and 1,104,659 surgical discharges.

Effectiveness: avoided death; avoided adverse outcomes; LOS. Costs: both short and long term hospital, per patient costs modeled.

Provider

Most cost-effective option appears is to raise the proportion of RNs without changing licensed hours. Decreases in urinary tract infections, pneumonia, and shock or cardiac arrest are associated the most with increasing the proportion of RNs.

Rothberg et al., 2005

General medical and surgical care, USA

Analysis of observational data from literature, and from 2 large hospital studies General medical and surgical patients.

Comparison between median staffing level for registered nurses, licensed staff and nurse assistants and recommended staffing level for them. 3 options compared: - raise proportion of RN to 75 percentile, without change in licensed hours; - raise number of licensed hours to 75 percentile, without change in proportion of RN staff; - raise both to 75 percentile. Comparison of ICERs for patient-to-nurse ratios ranging from 8:1 to 4:1.

Effects: mortality; LOS. Costs: Nursing cost per patient.

Institutional

Rothschild et al., 2009

Two critical care coronary units in New England, USA

Regression analysis of observational data Critical care patients.

Comparison of the savings from averting adverse events for various critical care nurse staffing levels (1:1, 1:2, and 1:3)

Effects: cost savings from avoiding adverse events. Costs: Nurse staffing costs

Hospital

8:1 ratio was the least expensive, but associated with the highest patient mortality. Smaller ratios improved mortality and increased costs, becoming progressively less cost-effective. However, ICER never exceeded $136,000/life saved. Marginal nurse staffing costs under higher staff levels were smaller than the expected cost savings from prevented adverse events.

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

Authors

507

508

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

may be rare, or true state of health may be hard to measure. In addition, treatment by nurses may involve symptomfree progression of disease (especially in an on-demand model of follow-up care), which may be very difficult to measure with any indicator (Helgesen et al., 2000). 4.4. Possible selection bias in some studies In some studies (Bissinger et al., 1997; Richardson et al., 2001), there were substantial differences in mortality between comparison groups, and as a result, differentials in the rate of missing outcome data. Even though the authors argued these differences were not statistically significant, this may have been due to small sample sizes, rather than true lack of statistically significant differences in mortality. Certainly the randomized controlled trials were not powered to detect differences in this significant outcome. There was another possibility for selection bias in some studies, for example when a considerable number of patients (around 47 per cent) initially randomized to receive care from nurses, eventually received ‘‘usual care’’, because of nurse practitioners or medical staff request (Pioro et al., 2001). The authors did not find significant baseline differences between this group and those who stayed in the nurse-treated group, but even in this case, some unobservable confounding factors may have been unaccounted for. 4.5. Issues with costing methodology As far as costs are concerned, in studies like this they may be hard to estimate, especially when resources are not traded for money (Douglas and Normand, 2005). Having said that, six authors employed a bottom-up/micro costing approach (i.e., the approach where resource use and prices are collected for separate items consumed) (Bissinger et al., 1997; Graveley and Littlefield, 1992; Mitchell-DiCenso et al., 1996; Pioro et al., 2001; Taylor et al., 1997; Venning et al., 2000), while the rest used a less detailed top-down, case-mix or disease-specific per diem design. When some resources were shared, it was not usually explained by the authors what the basis for allocating costs was. In three studies (Bissinger et al., 1997; Dall et al., 2009; Pioro et al., 2001), charges rather than payments were used for estimation. Cost of pharmaceuticals was not always included (the exceptions were Bissinger et al., 1997; Mitchell-DiCenso et al., 1996; Pioro et al., 2001; Richardson et al., 2001; Sakr et al., 2003; Venning et al., 2000), although their consumption may in part depend on whether care is managed by nurses or other medical staff. In all studies, there was also a lack of information on possible co-payments and fees that patients may have paid under different arrangements. No authors adequately accounted for costs accruing outside of the healthcare system, for example for the travel and time costs for patients and their families, or for productivity losses or gains from specific interventions, although some authors explicitly considered post-discharge costs and outcomes (Cowan et al., 2006; Ettner et al., 2006; Mitchell-DiCenso et al., 1996; Pioro et al., 2001; Richardson et al., 2001). One of them (Richardson

et al., 2001) found that the initially higher costs of nursebased treatment in the hospital setting may be partly offset with less expensive community care, and therefore postdischarge costs should be considered if cost substitution between settings is suspected. Almost all studies took a relatively short-term perspective on estimating costs and outcomes (although a few considered longer term outcomes, but not costs (Ganz et al., 2005; Mitchell-DiCenso et al., 1996)). While an obvious explanation for this is that it may be hard to justify expenses on a long-term trial, some studies (e.g. Richardson et al., 2001) showed that the conclusions may be affected by the length of follow-up. As mentioned above, this may be especially true if cost-shifting between hospital and primary care in the community is potentially sizeable. Another possibility is that in the longer run, training and supervision costs may decrease for nurses, thus improving their cost-effectiveness relative to other, more skilled and expensive staff (Pratt et al., 1993). In no cases were the opportunity costs explicitly included. Considering such costs may be important when it is suspected that some services may be replaced as a result of an intervention. For example, it may be important to know how the doctors use their freed up time as a result of the transfer of some of the services from them to nurses (Richardson, 1999). Also, one study (Pratt et al., 1993) reported significant stress arising from the obligation of registered nurses to supervise enrolled nurses, which may be the case in other set-ups (e.g. physicians supervising nurses), and therefore including the cost of stress in the analysis may be important. Finally, in no study were the training costs of nurses and other medical personnel considered. 4.6. Issues with choice of alternatives In almost all staff mix studies, a comparison of the proposed intervention with an existing and/or viable lowcost alternative was made. In a few staff substitution studies, however, although an alternative was specified, it was not clear if the list was exhaustive (e.g. see Bissinger et al., 1997; Cowan et al., 2006; Ettner et al., 2006; Pratt et al., 1993). In addition, some authors (e.g. Ganz et al., 2005; Needleman et al., 2006) did not consider a range of possible staffing levels, but rather examined how a change from a status quo to a new level (such as recommended nursing levels) may affect costs and outcomes of care. Such a perspective may be limiting, since there can be other, more cost effective staffing levels. 4.7. Discounting Another quality criterion from Table 1 is that the discounting for both costs and effects should be undertaken at appropriate rates (usually assumed to be 3–5 per cent a year). However, in most cases the outcomes were measured over a relatively short period of time, so this recommendation may be of little relevance here. However, on two occasions discounting was attempted (Dall et al., 2009; Ganz et al., 2005).

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

4.8. Sensitivity analysis and extrapolation of findings in the future In less than half of all articles was a sensitivity analysis to test the robustness of findings to different assumptions performed (Ettner et al., 2006; Ganz et al., 2005; Griffiths et al., 2001; Needleman et al., 2006; Richardson et al., 2001; Rothberg et al., 2005; Rothschild et al., 2009; Taylor et al., 1997). In almost all cases, only a one-way sensitivity analysis was undertaken. Only two studies reported incremental cost effectiveness ratios (ICER) (Ganz et al., 2005; Rothberg et al., 2005). One explanation for this is that in many cases (especially in nurse–physician substitution/skill mix studies), one strategy clearly dominated another, so ICER was not necessary for demonstrating costeffectiveness results. Also, where it was presented, there was little, if any, discussion of acceptable threshold levels. Explicit decision trees were usually not used nor were Markov-chain modelling techniques to extrapolate shortrun findings into the future. The latter could be due to the fact mentioned earlier-almost all studies took a rather short-term outlook. 4.9. Cost effectiveness findings The majority (eight) of staff substitution studies were conducted in inpatient settings, while the rest (four)—were in the community facilities. Although most reviewed studies have found that nurses can provide cost effective care compared to other, more expensive health professionals (Bissinger et al., 1997; Cowan et al., 2006; Ettner et al., 2006; Graveley and Littlefield, 1992; Taylor et al., 1997; Venning et al., 2000), this conclusion was not uniform. Thus some studies have found that nurses substituting for doctors in post/sub-acute nursing-led intermediate care units (i.e., inpatient wards where care management responsibilities usually discharged by doctors (referral, admission and discharge decisions) are given to nurses) were providing more expensive care similar in quality to other health professionals (Griffiths et al., 2001; Richardson et al., 2001), or that the outcomes of care were better for nurse-treated patients, but at higher expense (Sakr et al., 2003). In three articles, the evidence was inconclusive (Mitchell-DiCenso et al., 1996; Pioro et al., 2001; Pratt et al., 1993). The drivers of cost effectiveness appear to depend on the settings where the studies were conducted. For example, in outpatient settings, the greatest benefit of nurse-provided care appears to be greater patient satisfaction with care (Graveley and Littlefield, 1992; Taylor et al., 1997; Venning et al., 2000), although no difference was usually found in other health outcomes, nor in costs. On the other hand, in inpatient settings, several studies (Bissinger et al., 1997; Cowan et al., 2006; Ettner et al., 2006) found that nurses were providing less expensive care, with similar health outcomes. In the nurse staffing studies, it was generally not easy to assess the cost effectiveness of more intensive nursing care in acute care hospitals, as no strategy was clearly dominant over others. For example, one study found that in the USA, increasing staffing from median to recommended levels

509

was hard to justify on cost effectiveness grounds alone (Ganz et al., 2005). On the other hand, another study (Rothberg et al., 2005) reported that even with the lowest patient-to-nurse ratio, the incremental cost effectiveness ratio did not exceed USD 136,000 per life saved, a level comparable with many existing interventions. 4.10. Generalizability issues A weakness of the staff mix articles was that most of them (eight) were based on studies conducted on a single site, which may make it hard to generalize their findings to other settings. On the other hand, although all nurse staffing studies were based on data collected across multiple sites, the data was observational, rather than experimental. A related problem is a predominance of studies with small sample sizes in the staff mix group, and possible lack of power to detect statistically significant relationships for rare outcomes. Finally, most studies (twelve) were conducted in specialized care setting (e.g. in neonatal, anticoagulation or critical care), with fewer in general care settings. Cost-effectiveness analysis is but one approach to economic evaluation in healthcare. A major limitation of this method is its inability to compare interventions with different outcomes, or to confirm whether a given intervention is worth undertaking on its own merit. For example, one can be interested to know whether overall monetary benefits from a given intervention are larger than the overall costs. While this goal may be very difficult to achieve, given the challenge of estimating health and some other benefits in monetary terms, one can be interested in a less ambitious, but still useful goal of finding out whether cost savings and/or revenue gains from improving quality may offset the costs of interventions. Three reviewed studies employed an approach similar to the cost–benefit analysis, although it was somewhat restricted, in that outcomes measured only parts of the monetary gains from nurse staff interventions, such as costs savings from reduced length of stay and from avoiding adverse events (Dall et al., 2009; Needleman et al., 2006; Rothschild et al., 2009).2 The general finding was that the monetary benefit from increasing nursing staff levels was greater than the corresponding costs. 5. Discussion Previous reviews of studies on nursing and patient outcomes highlighted a number of methodological problems: small sample sizes, studies conducted on single sites, lack of randomization and resulting potential patient, staff and hospital-level confounding, and variations in cost measurements across sites. Indirect costs were rarely considered, and there was also usually lack of information on post-discharge outcomes of care (Griffiths et al., 2007; Kane et al., 2007; Lang et al., 2004). Our review has

2 Needleman et al also more explicitly employed cost-effectiveness approach, since they also considered avoided death as another outcome variable.

510

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

highlighted similar methodological issues and has further outlined specific methodological weaknesses related to cost-effectiveness analysis. Specifically, we have found that although most studies on economic evaluation of skill mix employed a randomized control design, this was not the case when the main variable of interest was nurse staffing levels. For stronger causal inference and confidence that effects seen are not confounded, ideally more of these studies should rely on random assignment of patients or units (e.g. wards), although this may be technically challenging and expensive. On the other hand, most nurse-physician substitution/nursing skill mix studies, where random assignment was employed more frequently, had small sample sizes, and they were usually conducted on a single site, in a small geographic area, and/or specific healthcare sector. Although randomized controlled trials have very important advantages compared to alternative methods, there are also important drawbacks to their use in economic evaluation. For example, most RCT studies reviewed here had a rather limited time horizon, and by not taking the future into account, they can give a distorted picture on the longer-term flow of costs and benefits associated with a particular intervention (Drummond et al., 2005). For such purposes, special decision modelling approaches (e.g. Markov modelling), where both costs and effects of an intervention are projected into the future, based on assumptions about transition probabilities between various health states, costs, benefits, as well as future discount rates, have been developed (Drummond et al., 2005). Similarly, short-term observational studies can also benefit from correctly implemented extrapolation techniques. Among the articles we reviewed, only one (Ganz et al., 2005) utilized Markov cohort simulation approach. All studies included in the review were conducted in high-income, English-speaking countries, which may have altogether different healthcare systems, demographic trends and income levels from the rest of the world. Thus, even with the most rigorous design, external validity of findings may be questionable. The solution for this appears to be conducting considerably more multisite studies that go beyond examining single disease areas. In addition, more use of decision modelling may help the available evidence from other settings be adapted to local conditions (Drummond et al., 2005). Likewise, a preferred solution is always to conduct properly designed longer term studies. However, when this is not feasible, Markov modelling techniques may be a step in the right direction. One of the major ‘Panel on Cost Effectiveness’ recommendations, not followed in most studies reviewed here, is the use of societal perspective. Given the huge social investment in nursing this seems an appropriate perspective. Having said that, some researchers have argued that provider perspective may be more appropriate when hospitals themselves pay for interventions. In addition, in some countries (e.g. the US) healthcare costs are mainly funded through private channels, in which case the application of societal perspective may not be accepted by the main institutional participants. However, this

qualification is of less relevance in a state-funded system similar to the one in the United Kingdom. On a related note, with rare exceptions, post-discharge costs were not included in the studies reviewed. One common explanation for this is that cost consideration is limited by the perspective chosen. However, as mentioned above, wherever possible, the societal perspective should be adopted. When properly conducted, it will give the most complete picture on the costs and benefits to society from a given intervention. Even without the wider societal perspective, hospital stays make up increasingly small components of the episode of care even for simple elective surgical procedures, and so costs of care occurring elsewhere in the system need to be considered. Only one study employed QALYs as the outcome variable designed to measure effectiveness of interventions. Although it is possible that nurse-provided care may not always lead to a significant change in QALYs, the question of finding outcomes that can be used commonly across studies remains. Without them, making comparisons and pooling results between studies may be very difficult. The same logic applies when one advocates a more common system for measuring costs across studies. This review has two potentially significant limitations. Because of the time and resource constraints, we did not consider the grey literature of unpublished manuscripts, dissertations and articles. In addition, all our articles were written in English, which implies that we may have missed knowledge from non-English speaking countries. However the predominance of English language scientific publication means that it is unlikely that there is a significant volume of literature in proportion to that considered here. While there are limitations in this literature, there are encouraging signs. First, the very fact of an increasing number of economic evaluation studies of nurse staffing and nurse-physician substitution/nursing skill mix in recent years is a promising development in itself. Moreover, as a greater number of authors have started to employ robust designs, more methodologically sound conclusions are made. The researchers were also utilizing a range of health outcome variables, which may help understand cost effectiveness of interventions in various dimensions. In the future, it is hoped that these trends will be supported with more research funding supporting methodologically strong studies. What are the policy implications of the main findings in the literature? Importantly, if policymakers were to follow the main message of some articles that greater nurse staffing, or that nurse/other staff substitution may be cost effective, they would also need to carefully consider any general equilibrium implications of such a policy move. For example, mandating lower patient-to-nurse ratios may lead to a greater demand for nurses and therefore higher nurse wages in a new equilibrium. This may require new assumptions on the wage level for nurses, and thus lead to revised cost-effectiveness estimations. In addition, decision making may be constrained by legislative restrictions on nursing practice in some countries (including their prescribing authority), as well as by general nurse

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

shortages. In the US in particular, a significant nurse shortage constraint does appear to exist (Greenberg, 2006). On the provider side, hospitals may have little incentive to provide high quality nursing care, since they do not generally get compensated for it (Aiken, 2008). Therefore, they may provide less than optimal amount of nursing care, unless they are mandated to do so. This is because the cost savings from the reduction of costly adverse events from higher quality care may accrue differently to hospitals versus payers depending on the payment arrangement (Rothschild et al., 2009). Mandated staffing ratios may incentivise employing less qualified and lower grade registered nursing staff in order to meet minimum numbers which again may change the underlying economic and clinical basis of the policy. In conclusion, nurses have, in general, to date treated economic evaluation of their services with considerable suspicion (Jenkins-Clarke, 1999), although this may recently have begun to change. One explanation for this may be that many believe the holistic process of nursing to be no less important than the treatment outcomes, and they feel that this is often ignored in such evaluations (Jenkins-Clarke, 1999). This is unfortunate, since answering the significant economic questions that arise from nursing work requires the perspective of those who best understand nursing work–nurses themselves. An important goal of further research would be bringing more nurses to participate in economic evaluation of their services, together with greater attention to patient experience of nursing services, and the value placed on this by patients and by society.

Acknowledgements ‘‘This is an independent report commissioned and funded by the Policy Research Programme in the Department of Health. The views expressed are not necessarily those of the department.’’ Conflict of interest: One of the authors (PG) was primary investigator on two of the studies reviewed. Primary appraisal of these studies was undertaken by a researcher who was not involved in the original studies and we see no conflict of interest arising. There are no other potential conflicts of interest. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ijnurstu. 2010.07.018. References Aiken, L.H., 2008. Economics of nursing. Policy, Politics & Nursing Practice 9 (2), 73. Allred, C.A., Arford, P.H., Mauldin, P.D., Goodwin, L.K., 1998. Cost-effectiveness analysis in the nursing literature, 1992–1996. Journal of Nursing Scholarship 30 (3), 235–242. Bissinger, R.L., Allred, C.A., Arford, P.H., Bellig, L.L., 1997. A cost-effectiveness analysis of neonatal nurse practitioners. Nursing Economic 15 (2), 92.

511

Buchan, J., Calman, L., 2004. The Global Shortage of Registered Nurses: An Overview of Issues and Actions. International Council of Nurses, Geneva. Cowan, M.J., Shapiro, M., Hays, R.D., Afifi, A., Vazirani, S., Ward, C.R., Ettner, S.L., 2006. The effect of a multidisciplinary hospitalist/physician and advanced practice nurse collaboration on hospital costs. Journal of Nursing Administration 36 (2), 79. Curtin, L.L., 2003. An integrated analysis of nurse staffing and related variables: effects on patient outcomes. Online Journal of Issues in Nursing 8 (3), 118. Dall, T.M., Chen, Y.J., Seifert, R.F., Maddox, P.J., Hogan, P.F., 2009. The economic value of professional nursing. Medical Care 47 (1), 97. Douglas, H.R., Normand, C., 2005. Economic evaluation: what does a nurse manager need to know? Journal of Nursing Management 13 (5), 419– 427. Drummond, M.F., Sculpher, M.J., Torrance, G.W., O’Brien, B.J., Stoddart, G.L., 2005. Methods for the Economic Evaluation of Health Care Programmes. Oxford University Press, USA. Ettner, S.L., Kotlerman, J., Afifi, A., Vazirani, S., Hays, R.D., Shapiro, M., Cowan, M., 2006. An alternative approach to reducing the costs of patient care? A controlled trial of the Multi-Disciplinary Doctor– Nurse Practitioner (MDNP) model. Medical Decision Making 26 (1), 9. Ganz, D.A., Simmons, S.F., Schnelle, J.F., 2005. Cost-effectiveness of recommended nurse staffing levels for short-stay skilled nursing facility patients. BMC Health Services Research 5 (1), 35. Graveley, E.A., Littlefield, J.H., 1992. A cost-effectiveness analysis of three staffing models for the delivery of low-risk prenatal care. American Journal of Public Health 82 (2), 180. Greenberg, P.B., 2006. Nurse-to-patient ratios: what do we know? Policy Politics and Nursing Practice 7 (1), 14. Griffiths, P., 2009. RN + RN = better care? What do we know about the association between the number of nurses and patient outcomes?. International Journal of Nursing Studies 46 (10), 1289–1290. Griffiths, P., Edwards, M., Forbes, A., Harris, R., Ritchie, G., 2007. Effectiveness of intermediate care in nursing-led in-patient units. Cochrane Database of Systematic Reviews 2, CD002214. Griffiths, P., Harris, R., Richardson, G., Hallett, N., Heard, S., WilsonBarnett, J., 2001. Substitution of a nursing-led inpatient unit for acute services: randomised controlled trial of outcomes and cost of nursingled intermediate care. Age and Ageing 2001 (30), 483–488. Griffiths, P., Murrells, T., Maben, J., Jones, S., Ashworth, M., 2010. Nurse staffing and quality of care in UK general practice: cross-sectional study using routinely collected data. British Journal of General Practice 60 (1), e36–e48. Hall, L.M.G., Buch, E., 2009. Skill mix decision-making for nursing. International Centre for Human Resources in Nursing, Geneva. Helgesen, F., Andersson, S.O., Gustafsson, O., Varenhorst, E., Gobe´n, B., Carnock, S., Sehlstedt, L., Carlsson, P., Holmberg, L., Johansson, J.E., 2000. Follow-up of prostate cancer patients by on-demand contacts with a specialist nurse: a randomized study. Scandinavian Journal of Urology and Nephrology 34 (1), 55–61. ICN, 2006. ICN and FNIF: The Global Nursing Shortage: Priority Areas for Intervention. ICN. Jenkins-Clarke, S., 1999. Does nursing needthe dismal science’? The case for economic evaluations in nursing. Journal of Research in Nursing 4 (6), 448. Kane, R.L., Shamliyan, T.A., Mueller, C., Duval, S., Wilt, T.J., 2007. The association of registered nurse staffing levels and patient outcomes: systematic review and meta-analysis. Medical Care 45 (12), 1195. Konetzka, R.T., Stearns, S.C., Park, J., 2008. The staffing: outcomes relationship in nursing homes. Health Services Research 43 (3), 1025–1042. La¨ma˚s, K., Willman, A., Lindholm, L., Jacobsson, C., 2009. Economic evaluation of nursing practices: a review of literature. International Nursing Review 56 (1), 13–20. Lang, T.A., Hodge, M., Olson, V., Romano, P.S., Kravitz, R.L., 2004. Nurse– patient ratios: a systematic review on the effects of nurse staffing on patient, nurse employee, and hospital outcomes. Journal of Nursing Administration 34 (7–8), 326. Lankshear, A.J, Sheldon, T.A., Maynard, A., 2005. Nurse staffing and healthcare outcomes: a systematic review of the international research evidence. Advances in Nursing Science 28 (2), 163. Latter, S., Maben, J., Myall, M., Young, A., 2007. Evaluating the clinical appropriateness of nurses’ prescribing practice: method development and findings from an expert panel analysis. Quality and Safety in Health Care 16 (6), 415–421. Laurant, M., Reeves, D., Hermens, R., Braspenning, J., Grol, R., Sibbald, B., 2004. Substitution of doctors by nurses in primary care. Cochrane Database of Systematic Reviews 4 DOI:10.1002/14651858. CD14001271.pub14651852.

512

Y. Goryakin et al. / International Journal of Nursing Studies 48 (2011) 501–512

McGrath, S., 1990. The cost-effectiveness of nurse practitioners. The Nurse Practitioner 15 (7), 40. Mitchell-DiCenso, A., Guyatt, G., Marrin, M., Goeree, R., Willan, A., Southwell, D., Hewson, S., Paes, B., Rosenbaum, P., Hunsberger, M., 1996. A controlled trial of nurse practitioners in neonatal intensive care. Pediatrics 98 (6 Pt 1), 1143. Needleman, J., Buerhaus, P.I., Stewart, M., Zelevinsky, K., Mattke, S., 2006. Nurse staffing in hospitals: is there a business case for quality? Health Affairs 25 (1), 204. Pioro, M.H., Landefeld, C.S., Brennan, P.F., Daly, B., Fortinsky, R.H., Kim, U., Rosenthal, G.E., 2001. Outcomes-based trial of an inpatient nurse practitioner service for general medical patients. Journal of Evaluation in Clinical Practice 7 (1), 21–33. Pratt, R., Burr, G., Leelarthaepin, B., Blizard, P., Walsh, S., 1993. The effects of All-RN and RN-EN staffing on the quality and cost of patient care. The Australian Journal of Advanced Nursing: A Quarterly Publication of the Royal Australian Nursing Federation 10 (3), 27. Richardson, G., 1999. Identifying, evaluating and implementing costeffective skill mix. Journal of Nursing Management 7 (5), 265–270. Richardson, G, Griffiths, P., Spilsbury, P., Wilson-Barnett, J., Batehup, L., 2001. Economic evaluation of a nursing-led intermediate care unit. International Journal of Technology Assessment in Health Care 17 (3), 442–450. Rothberg, M.B., Abraham, I., Lindenauer, P.K., Rose, D.N., 2005. Improving nurse-to-patient staffing ratios as a cost-effective safety intervention. Medical Care 43 (8), 785.

Rothschild, J.M., Bates, D.W., Franz, C., Soukup, J.R., Kaushal, R., 2009. The costs and savings associated with prevention of adverse events by critical care nurses. Journal of Critical Care 24 (3), 471–477. Sakr, M., Kendall, R., Angus, J., Saunders, A., Nicholl, J., Wardrope, J., 2003. Emergency nurse practitioners: a three part study in clinical and cost effectiveness. British Medical Journal 20 (2), 158. Siegel, J.E., 1998. Cost-effectiveness analysis and nursing research—is there a fit? Image. Journal of Nursing Scholarship 30 (3), 221– 222. Spetz, J., 2005. The cost and cost-effectiveness of nursing services in health care. Nursing Outlook 53 (6), 305–309. Taylor, F.C., Gray, A., Cohen, H., Gaminara, L., Ramsay, M., Miller, D., 1997. Costs and effectiveness of a nurse specialist anticoagulant service. British Medical Journal 50 (10), 823. Thungjaroenkul, P., Cummings, G.G., Embleton, A., 2007. The impact of nurse staffing on hospital costs and patient length of stay: a systematic review. Nursing Economics 25 (5), 255. Venning, P., Durie, A., Roland, M., Roberts, C., Leese, B., 2000. Randomised controlled trial comparing cost effectiveness of general practitioners and nurse practitioners in primary care. British Medical Journal 320 (7241), 1048. Vincent, D., 2002. Using cost-analysis techniques to measure the value of nurse practitioner care. International Nursing Review 49 (4), 243– 249. WHO, 2006. The World Health Report 2006: Working Together for Health. Geneva, World Health Organization.