UBC Faculty Research and Publications

The relationship between staff skill mix, costs and outcomes in intermediate care services Dixon, Simon; Kaambwa, Billingsley; Nancarrow, Susan; Martin, Graham P; Bryan, Stirling Jul 29, 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12913_2009_Article_1358.pdf [ 208.12kB ]
JSON: 52383-1.0223445.json
JSON-LD: 52383-1.0223445-ld.json
RDF/XML (Pretty): 52383-1.0223445-rdf.xml
RDF/JSON: 52383-1.0223445-rdf.json
Turtle: 52383-1.0223445-turtle.txt
N-Triples: 52383-1.0223445-rdf-ntriples.txt
Original Record: 52383-1.0223445-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessThe relationship between staff skill mix, costs andoutcomes in intermediate care servicesSimon Dixon1, Billingsley Kaambwa2, Susan Nancarrow3*, Graham P Martin4, Stirling Bryan5AbstractBackground: The purpose of this study was to assess the relationship between skill mix, patient outcomes, lengthof stay and service costs in older peoples’ intermediate care services in England.Methods: We undertook multivariate analysis of data collected as part of the National Evaluation of IntermediateCare Services. Data were analysed on between 337 and 403 older people admitted to 14 different intermediatecare teams. Independent variables were the numbers of different types of staff within a team and the ratio ofsupport staff to professionally qualified staff within teams. Outcome measures include the Barthel index, EQ-5D,length of service provision and costs of care.Results: Increased skill mix (raising the number of different types of staff by one) is associated with a 17%reduction in service costs (p = 0.011). There is weak evidence (p = 0.090) that a higher ratio of support staff toqualified staff leads to greater improvements in EQ-5D scores of patients.Conclusions: This study provides limited evidence on the relationship between multidisciplinary skill mix andoutcomes in intermediate care services.BackgroundThere has been growing international interest in ‘work-force engineering and redesign’ over recent years, whichhas resulted in an increase in research exploring theimpact of different approaches to staffing on patient andservice outcomes, particularly in the areas of medicineand nursing. There are several drivers for workforcechange including skills shortages; productivity improve-ments; cost containment; quality improvement; techno-logical innovation; and health sector reform. Themodernisation of the National Health Service has led tosubstantial changes to the numbers and types of staff,and their ways of working. For instance, workforceshortages and restructuring in the UK have createdopportunities for staff to perform roles that are outsidetheir traditional scope of practice[1].Intermediate care (IC) is a valuable setting in which toexplore new ways of working. Many IC services operateat the interface of numerous agencies, settings and pro-fessional groups, and require workforce structures thatcan reflect and respond to this complexity [2]. IC ser-vices tend to have non-hierarchical management struc-tures; and staff are often supervised by someone whoseprofessional background is different to their own. Medi-cal practitioners are sometimes the ‘gatekeeper’ to IC,however their level and mode of involvement varies [3]and non-medical practitioners often have a great deal ofautonomy [2]. Finally, IC services can be delivered in avariety of locations, including the patients’ own home,nursing homes, hospitals and community centres.Following the National Service Framework for OlderPeople (31), the number and type of community basedservices for older people have grown substantially andare set to expand further as acute care services are pro-gressively moved to primary and community care set-tings. Intermediate care services have diverse models ofstaffing, however typically intermediate care teams aremultidisciplinary [4-14] even in usual care settings, orwhen labelled ‘nurse led unit’, or ‘GP led unit’. They arelikely to include input from physiotherapy, occupationaltherapy and therapy assistants [5,10]. A wide range ofother staff may be involved in the delivery of intermedi-ate care, however this varies greatly across the differentservices [13]. There is no evidence about the ‘best way’* Correspondence: s.a.nancarrow@shu.ac.uk3Centre for Health and Social Care Research, Sheffield Hallam University,Montgomery House, 32 Collegiate Crescent, Collegiate Campus, SheffieldS10 2BP, UKDixon et al. BMC Health Services Research 2010, 10:221http://www.biomedcentral.com/1472-6963/10/221© 2010 Dixon et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.to staff an intermediate care service, and this is likely todepend on the setting and purpose of the service[10].Comparable studies are difficult to find, as most work-force studies explore the relationship between two dif-ferent practitioners rather than multidisciplinaryarrangements.Only one experimental study specifically examined theimpact of different models of staffing on costs and out-comes [8] by comparing hospital at home with care ona hospital ward. Staffing models were not attributed tooutcomes, however the research showed that cost effi-ciency of services was negatively influenced by employ-ing high grade nurses in roles with little direct clinicalinput. In contrast, the costs of the other members of themultidisciplinary team (eg therapists) constituted a rela-tively small component of the total cost. The authorssuggested that increasing the proportion of nursesinvolved in more direct nursing care could reduce thecosts of the service.There is evidence from a number of qualitative studiesthat intermediate care requires staff to work across pro-fessional boundaries, and that initially, this can createtensions, however generally this improves with time, andis perceived by staff to enhance patient and service out-comes[9,15,16].The literature demonstrates that patient satisfaction ispositively associated with well trained workers andrespectful staff, however is negatively associated withpoor recruitment and retention and delayed or absentworkers [17]. It is also evident that service user percep-tions of service quality are likely to be positively influ-enced by patient characteristics, such as age, andorganisational characteristics such as the intensity ofcare received, staffing organisation, employment condi-tions for staff, good recruitment and retention rates andgreater levels of staff experience and training [18]. Manyof the same factors have been found to significantlyinfluence patient functional gain [19]. Staff experienceand training such as competency of support workers indelivering rehabilitation and the presence of advancedpractice nurses in teams can improve patient functionalgains. Similarly patient functional outcomes can also beenhanced by greater intensity of care, greater therapyand general staffing levels and the use of agency staffhave also been found to improve functional gains andoutcomes.Teamwork, team order and organisation have alsobeen found to improve functional outcomes [20]. Sev-eral studies however have indicated that there are otherfactors that contribute to functional gain outside ofthese workforce variables. Patient characteristics such ashigher cognitive ability of patients [21], the patient mix[19] and a longer stay in a post-acute care facility [21]were all found to positively impact on functional gain.A systematic review of the ‘Evidence for the effective-ness of intermediate care’ [22] found that the evidencesupporting the development of specific intermediatecare services is quite heterogeneous, and still lacking.They reported that overall, intermediate care servicesare not associated with adverse consequences for recipi-ents, and intensive therapy can improve physical out-comes and patient satisfaction. Extrapolating from themain study findings, it appears that despite large varia-tions in staffing across services, there is little measurableeffect on the outcomes for service users. These findingssuggest that there may be potential for efficiency savingsin intermediate care services through the identificationof more effective models of interprofessional teamorganisation.There is a need for greater understanding and consul-tation around service user preferences for different typesof staffing (type, roles, numbers etc). For instance,Brown et al [23] found that home care workers were themost valued service provider in the health and socialcare team and it did not matter to the service userwhether or not the team was integrated as long as theirneeds were met.The aim of this study is to assess the relationshipbetween skill mix, patient outcome, length of stay andcost. This was part of a larger study exploring the rela-tionship between staffing and patient outcomes[24], andinvolved the reanalysis of data from a National Evalua-tion of Intermediate Care with the addition of datarelating to the skill mix of the teams included withinthe study.MethodsThe National Evaluation of Intermediate Care Services[25] was undertaken by the Universities of Birminghamand Leicester. It involved extensive qualitative and quan-titative data collection within five case-study sites inEngland between January 2003 and November 2004.The processes used for the collection and analysis ofquantitative data in the case-study sites are described indetail elsewhere [25,26].The case-studies were five primary care trusts selectedas to represent ‘whole systems’ (an area with a specificgeographical boundary) of intermediate care. By study-ing whole systems as opposed to individual service mod-els we aimed to achieve a more detailed understandingof the implementation of intermediate care and itsimpact upon system-level costs and outcomes.Quantitative data were collected by staff employed bythe intermediate care services according to protocolsestablished by the evaluation team. Staff completed astudy proforma with their patients, at the point of entryto the service, and then further questions were com-pleted on the day of discharge, transfer or following theDixon et al. BMC Health Services Research 2010, 10:221http://www.biomedcentral.com/1472-6963/10/221Page 2 of 7end of service provision. All intermediate care admis-sions over a defined period were included.Data were available on patient age, gender, Barthelscore at admission and discharge, EQ-5D at admissionand discharge, type of service defined in terms of admis-sion avoidance or other, and location of service in termsof residential or non-residential.The Barthel score is a measure of a patient’s ability toundertake a set of activities of daily living, such as feed-ing, bathing and grooming. It is typically completed bythe health professional, and is scored on a scale of zeroto twenty with zero indicating that the patient is fullydependent on others for all activities, and twenty indi-cating that the patient is independent [27,28]. The EQ-5D, formerly know as the EuroQol, is a generic measureused primarily by economists to calculate qualityadjusted life years (QALYs). It uses a single question toassess each of five health domains; mobility, self-care,usual activities, pain/discomfort and anxiety/depresssion.The EQ-5D has a complex scoring system, which rangesfrom 1 which indicates full health, through to -0.59 [29].Data on skill mix were collected as descriptive data,but not included in any of the analyses undertaken todate. These data recorded the types of health careworker included in each of the teams at the time of theevaluation, and the number of whole-time equivalents.These were summarised in terms of two skill mix vari-ables; ratio of support workers to qualified staff and thenumber of different professions included within theteam. For the purposes of these two measures, supportworkers included staff involved in the direct delivery ofpatient care but who do not have a professional qualifi-cation, and included assistant practitioners, therapyassistants, support workers, generic rehabilitation assis-tants, health care assistants and social care workers.Staff were categorised as ‘qualified’ if they had a recogni-sable professional title which is associated with tertiarytraining, and included nurses, doctors, allied healthpractitioners and social workers. The ‘number of differ-ent types of professions’ was simply a count of the num-bers of different types of practitioners (including supportworkers) involved in the delivery of patient care. Addi-tionally, the team data were used to calculate the totalnumber of WTEs employed, as a proxy for the size ofthe service.NHS ethical approval for the secondary analysis wasobtained in 2006 (06/Q1606/132).AnalysesData used in the National Evaluation, plus the additionalvariables defined from the team data were used toundertake a set of multivariate analyses. These were toassess:▪ The impact of skill mix on outcomes of care asmeasured by the change in the Barthel index.▪ The impact of skill mix on outcomes of care asmeasured by the change in the EQ-5D.▪ The impact of skill mix on length of care episode(or length of service provision).▪ The impact of skill mix on costs of care asmeasured.Based on previous analyses of costs and outcomes, therelationship with age was thought to be monotonic butnon-linear, and so age-squared was used as an indepen-dent variable. Likewise, based on economic theory, forthe analysis of costs total WTE squared was also definedto help identify possible economies of scale across theteams.Multivariate analyses were undertaken using individualpatient data, but taking into account the clustering ofcases within teams within STATA. Ordinary LeastSquares (OLS) regression was undertaken for the ana-lyses of outcomes (change in EQ-5D and Barthel) asdependent variables, whilst generalised linear modelswith a log link and gamma distribution were used forthe analyses of length of stay and cost per patient. Gen-eralised linear models (GLMs) are considered to bemore appropriate for the analysis of skewed and hetero-scedastic data while retaining the original scale of thedata [30]. To aid interpretation of GLM coefficients, theexponents of the coefficients were calculated. These canbe interpreted as the proportional change of the depen-dent variable because of a change of one unit in theindependent variable (32).When interpreting the statistical significance of themodels, we have adopted the approach of Bland [31]whereby p-values greater than 0.10 indicate little or noevidence of a relationship, values between 0.05 and 0.10indicate weak evidence of a relationship, values between0.01 and 0.05 indicate evidence of a difference or rela-tionship and values less than 0.01 indicate strong evi-dence of a difference or relationship.Additionally, the specification of the estimatedregression equations was assessed using the RamseyREST test [32]. This test performs auxiliary regres-sions that add in powers of the fitted values to theoriginal equations. Statistically significant coefficientson these new terms have been found to be indicativeof misspecification.ResultsAcross the four analyses, data were available on between337 and 403 patients, describing costs and outcomesacross 14 separate teams. Patient and team characteris-tics are summarised in Table 1.Dixon et al. BMC Health Services Research 2010, 10:221http://www.biomedcentral.com/1472-6963/10/221Page 3 of 7The relationship between skill mix and outcomes of careas measured by the Barthel indexThere is strong evidence that less independent patientson admission (as indicated by lower Barthel scores)were associated with greater improvements in Barthelover the period of care (Table 2). In particular, foreach one unit decrease in the baseline Barthel score,the change in Barthel score increased by 0.2854.None of the skills staffing parameters were statisticallysignificant. Whilst the overall explanatory power therelationship was significant, as evidenced by theblock F-test, there was also evidence of possiblemisspecification.The relationship between skill mix and outcomes of careas measured by the EQ-5DThere is strong evidence that lower EQ-5D scores onadmission are associated with greater improvements inEQ-5D over the period of care (Table 2). For each oneunit decrease in the baseline EQ-5D score, the changein EQ-5D score increased by 0.4363. There is also weakevidence that residential intermediate care services, andhigher support staff to qualified staff ratios are asso-ciated with greater improvements in EQ-5D scores. Thegain in EQ-5D for individuals in residential care was0.0582 units bigger than that of individuals in non-resi-dential care while a 1 unit increase in the ratio of sup-port staff to qualified staff increased the change in EQ-5D by 0.0464 units. Overall, the relationship has signifi-cant explanatory power, but misspecification issuggested.The relationship between skill mix and process of care asmeasured by length of care episodeAcute admission avoidance schemes are strongly asso-ciated with having shorter periods of intermediate care:the length of care for individuals in such schemes wasabout 18% shorter [exp(-0.2000) = 0.8187] compared tothat of individuals in other schemes. None of the skillsstaffing parameters were statistically significant.The relationship between skill mix and costs of careThere is strong evidence that older patients were asso-ciated with higher costs but these costs begin to fall aspatients become more elderly. For each one yearincrease in age, costs per case rose by 13.58% [exp(0.1273) = 1.1358] and further analysis indicates thatcosts begin to fall when individuals reach around 80years old. Residential services and longer periods of carewere strongly associated with higher costs. Costs forresidential services were almost five times bigger thanthose for non-residential services [exp(1.5892) = 4.8998]while an increase of 1 day in the length of care wasassociated with a 2.60% increase in costs [exp(0.0257) =1.0260). There was evidence that greater numbers of dif-ferent types of staff were associated with lower costs(Table 2). Having an extra category of staff decreasedcosts by about 17% [exp(-0.1827 = 0.8330]. The coeffi-cients on total staff numbers and total staff numberssquared suggest that cost per case initially increase by22.46% [exp(0.2026) = 1.2246] as teams grow by a factorof one individual, but after then begin to fall. Furtheranalysis indicates that the point at which cost per casebegins to fall is around 12 WTE staff which is 3 WTEstaff members larger than the largest team in the studyas shown in Table 1.Discussion and conclusionsThe analyses show that costs and outcomes of inter-mediate care are partly explained by differences inpatient and service characteristics, however, the impactof service skill mix is limited (Table 2). There is weakevidence (p = 0.090) that the ratio of support staff toqualified staff impact on health gains (measured by thechange in EQ-5D) seen during care, with higher propor-tions of support staff being associated with greaterimprovement. There is stronger evidence (p = 0.011)that higher numbers of different types of staff are asso-ciated with lower costs.There are several possible explanations for the greaterimprovements in EQ-5D in patients when who utilisemore support staff (SS) relative to qualified staff (QS).Qualitative feedback from the same study suggests thatsupport staff spend more time with patients than quali-fied staff, and perform more of the ‘hands on’ work,which may lead to better improvements in outcome.Table 1 Description of patient and team characteristicsPatient characteristics Median Min;MaxAge 82.14 62.34;100.63Baseline Barthel 15.00 3.00;20.00Change in Barthel (n = 398) 1.00 -5.00;14.00Baseline EQ5D 0.52 -0.59; 1.00Change in EQ5 D (n = 349) 0.07 -1.11; 1.16Length of care (days) 31.00 1.00; 232.00Cost per patient (£) 1241.68 40.07; 15,323.60Gender - n (%) for females 299 (74.19)Team characteristics Median Min-MaxRatio of support staff to qualified staff 0.67 0.00; 4.00Number of different types of staff 5.00 3.00; 9.00Total number of staff (WTEs) 7.75 1.82; 23.70IC function - n (%) for acute admissionavoidance215 (53.35)IC setting - n (%) for residential IC 102 (25.31)Note: n = 403 unless otherwise stated. 403 observations were used as thissample represents the set of patients on which all four sets of regressionanalyses were run.Dixon et al. BMC Health Services Research 2010, 10:221http://www.biomedcentral.com/1472-6963/10/221Page 4 of 7Alternatively, it could mean that additional SS allow abetter service to be delivered, for example, increasingthe number of SS staff may allow for service develop-ment. This second interpretation is in line with findingsseen in general practice [33].This second interpretation is less plausible as someaspects of service expansion will be controlled for by the‘total number of staff’ variable within the regression. Inother words, increasing SS staff without reducing QSstaff is not responsible for the better outcomes associatedwith the higher support staff to qualified staff ratios.Other possible explanations are that intermediate carepatients may not require the intensive or specialisedtreatment of support staff, thus a higher ratio of SS toQS may be the optimum combination that will lead tobetter outcomes. Similarly, it may be that those patientswho do require more specialised input are directed toservices that provide that input.The impact of greater numbers of different types ofstaff on costs could reflect economies produced by spe-cialisation. Understanding how costs were calculatedwithin the National Evaluation is important before con-sidering this issue further. Cost per patient was calcu-lated based on a cost per day for the entire servicebased on budgets and an individual patient’s length ofcare. So, cost per patient is driven either by the servicebudget or length of stay. As the relationship betweennumber of different types of staff and length of care issmall and statistically insignificant, it appears that theeffect is through the size of the service budget. Themechanism by which service budgets are reduced isopen to speculation. Two possible processes are reducednumber of visits and/or the use of smaller numbers ofstaff.The results also show a potential conflict betweenpatient outcomes and costs; increasing support staffTable 2 Regression resultsChange in Barthel score1n = 398Change in EQ5D score2n = 337Length of care (days)n = 403Cost (£s)n = 403Age 0.3085(0.2272)0.0336(0.0366)-0.0490(0.0890)0.1273***(0.0363)Age squared -0.0022(0.0014)-0.0002(0.0002)-0.0002(0.0005)-0.0008***(0.0002)Gender 0.2852(0.2469)0.0670(0.0365)-0.0225(0.0754)0.0352(0.0379)Baseline Barthel score -0.2854***(0.0794)-0.0136(0.0111)0.0039(0.0083)Baseline EQ5 D score -0.4363***(0.0651)0.1067(0.1026)0.0450(0.1075)Admission avoidance 0.5030(0.3478)0.0044(0.0400)-0.2000**(0.0832)0.0565(0.0618)Residential care 0.6126(0.6976)0.0582*(0.0327)0.0835(0.2587)1.5892***(0.3578)Length of care -0.0003(0.0057)-0.0003(0.0005)- 0.0257***(0.0030)Ratio of support to qualified staff 0.2277(0.4819)0.0464*(0.0254)-0.0564(0.1042)0.0600(0.1076)Number of different staff types -0.0529(0.1532)0.0161(0.0103)0.0470(0.0608)-0.1827**(0.0715)Total number of staff -0.0064(0.0258)-0.0005(0.0011)0.0010(0.0104)0.2026***(0.0646)Total number of staff squared -0.0085***(0.0022)Constant -4.9479(8.8322)1.1658(1.4717)6.0765(3.7505)0.8401(1.535)R-squared 0.1932 0.2505 0.0318 0.2730Block F-test3 <0.0001 <0.0001 - -RESET test 0.0002 0.0441 - -1 Positive changes reflect gains in a patient’s level of independence.2 Positive changes reflect improvements in a patient’s health related quality of life.3 Tests the hypothesis that all parameters are equal to zero.* 0.10 > p-value > 0.05** 0.05 > p-value > 0.01*** p-value <0.01Dixon et al. BMC Health Services Research 2010, 10:221http://www.biomedcentral.com/1472-6963/10/221Page 5 of 7numbers relative to qualified staff appears to improvehealth outcomes (as measured by the EQ-5D), but ifthis is achieved at the expense of multidisciplinarity (asmeasured by numbers of different types of staff) thencosts will increase.This study is important because it uses existing data toexplore the potential for a relationship between differentstaffing models and patient outcomes in services that arean important showcase of the NHS modernisationagenda with the introduction of new roles, integratedhealth and social care services, and interdisciplinaryworking. Accurate data on staffing and patient outcomesin community based services can be costly and difficultto capture. The relationships demonstrated within thisstudy indicate that at the very least, further research iswarranted into the relationship between outcomes andstaffing to support more efficient and effective ways todeliver patient care.LimitationsThe regressions have reasonable explanatory power,however there is evidence from the RESET test thatthere is misspecification. Possible causes of this could bethe choice of regression technique or the omission ofrelevant variables. The Barthel and EQ-5D scores topossess some characteristics that are similar to trun-cated data, with minimum and maximum permittedscores (and hence changes in scores). Consequently,some studies that have analysed quality of life data ofthis kind have used truncated regressions and censoredleast absolute deviations (CLAD) regressions [34,35].These were undertaken, however, they did not affect theresults appreciably.Likewise, the possibility of omitted variables wasinvestigated by analysing other specifications thatincluded interaction terms between the staff mix vari-ables. These additional regressions led to problems withinterpretation probably caused by using so many clus-ter-based independent variables in the face of so fewclusters. The RESET tests also indicated that misspecifi-cation problems persist even the presence of these morecomplex specifications.Two individuals recorded gains in EQ-5D of 1.59,implying that they moved from the worst possible stateat admission to the best state at discharge. These‘extreme’ changes might reflect misunderstanding on thepart of the respondents. Excluding these two individualsfrom the analysis however did not alter the results above.Whilst the presence of clustering was taken intoaccount in the analysis, it should be noted that thesmall number of clusters will limit our ability to detectany associations that are present. This is exacerbated bythe limited variability seen between the clusters in termsof skill mix (Table 1).Interpretation of the results is also limited by the factthat we do not know the number of visits and type oftherapy/care provided at the visits. So, for example, wedo not know whether the improved outcomes associatedwith support staff is due to the type of input (’x’ ratherthan ‘y’) or more frequent input (’more of x’).It is feasible that the relationship between staffingnumbers and outcomes is due to the staff identifyingpatients with greater potential to improve and allocatingmore staffing resources to those patients. However, ifthis were the case, the mechanisms by which this wasperformed was not clear or systematic.In conclusion, this study provides limited evidence ofthe role of skill mix on the costs and outcomes of inter-mediate care services. The work is based around anobservational dataset and the use of skill mix variablesat the service level, which together may limit our abilityto identify possible relationships. A controlled studywith clearly defined packages of inputs being providedto patients, would provide a clearer picture of how skillmix can impact on costs and outcome of intermediatecare services. Until such work is done, services will con-tinue to develop in a largely piecemeal way, with theconsequences of this being largely hidden.AcknowledgementsThis project was funded by the National Institute for Health Research ServiceDelivery and Organisation programme (project number 08/1519/95). Theviews and opinions expressed therein are those of the authors and do notnecessarily reflect those of the NIHR SDO programme or the Department ofHealth.We are grateful to colleagues from the Universities of Birmingham andLeicester (the ICNET team) who participated in the National Evaluation ofIntermediate Care Services from which data used in this study wereobtained. We are also thankful to the intermediate care-coordinators andthe staff from the case-study sites that provided the quantitative data andclarified follow-up questions. The National Evaluation was funded by theDepartment of Health (Policy Research Programme) and the MedicalResearch Council. The funders were not involved in the study design, in thewriting of the manuscript or in the decision to submit the manuscript forpublication.Author details1School of Health and Related Research, University of Sheffield, RegentCourt, 30 Regent Street, Sheffield S1 4DA, UK. 2Health Economics Unit,Health and Population Sciences, University of Birmingham, Edgbaston,Birmingham B15 2TT, UK. 3Centre for Health and Social Care Research,Sheffield Hallam University, Montgomery House, 32 Collegiate Crescent,Collegiate Campus, Sheffield S10 2BP, UK. 4Department of Health Sciences,University of Leicester, Adrian Building, University Road, Leicester LE1 7RH,UK. 5Centre for Clinical Epidemiology and Evaluation, University of BritishColumbia, Seventh Floor, 828 West 10th Avenue, Research Pavilion,Vancouver BC V5Z 1M9, Canada.Authors’ contributionsSN and GPM helped link staffing to other quantitative data from thenational evaluation. SB, SD and BK developed the analysis plan for theeconometric analysis. SB and BK performed the econometric analysis, andinterpreted and discussed the results. SD, SN and GPM wrote the first draftof the manuscript. All authors critically revised the manuscript, and read andapproved the final manuscript. SN obtained the NIHR funding for theproject.Dixon et al. BMC Health Services Research 2010, 10:221http://www.biomedcentral.com/1472-6963/10/221Page 6 of 7Competing interestsThe authors declare that they have no competing interests.Received: 25 October 2009 Accepted: 29 July 2010Published: 29 July 2010References1. Nancarrow S, Borthwick A: Dynamic professional boundaries in the healthcare workforce. Sociology of Health and Illness 2005, 27(7):897-919.2. Abbott A: The System of Professions: An Essay on the Division of ExpertLabour. Chicago: Chicago University Press 1988.3. Anderson RL, Lyons JS: Needs-based planning for persons with seriousmental illness residing in intermediate care facilities. Journal of BehavioralHealth Services & Research 2001, 28(1):104-110.4. Cohen M, et al: Workload as a determinant of staff injury in intermediatecare. International Journal of Occupational and Environmental Health 2004,10(4):375-383.5. Enderby P, Wade D: Community rehabilitation in the United Kingdom.Clinical Rehabilitation 2001, 15:577-581.6. Griffiths J, Austin L, Luker K: Interdisciplinary teamwork in the communityrehabilitation of older adults: an example of flexible working in primarycare. Primary Health Care Research and Development 2004, 5:228-239.7. Griffiths P: Nursing-led in-patient units for intermediate care: a survey ofmultidisciplinary discharge planning practice. Journal of Clinical Nursing2002, 11(3):322-330.8. Jones J, et al: Economic evaluation of hospital at home versus hospitalcare: cost minimisation analysis of data from a randomised controlledtrial. British Medical Journal 1999, 319:1574-1550.9. Nancarrow S: Dynamic role boundaries in intermediate care services.Journal of Interprofessional Care 2004, 18(2):141-151.10. Parker S: A survey of day hospital and home based rehabilitationservices in England. Draft 2006.11. Rudd AG, et al: Randomised controlled trial to evaluate early dischargescheme for acute stroke. British Medical Journal 1997, 315:1039-1044.12. Shield F: Developing a therapy-led community rehabilitation team.Managing Community Care 1998, 6(4):160-168.13. Vaughan B, Lathlean J: Intermediate care models in practice. King’s Fund:London 1999.14. Wiles R, et al: Nurse-led intermediate care: patients’ perceptions.International Journal of Nursing Studies 2003, 40(1):61-71.15. Booth J, Hewison A: Role overlap between occupational therapy andphysiotherapy during in-patient stroke rehabilitation: an exploratorystudy. Journal of Interprofessional Care 2002, 16(1):31-40.16. Nancarrow S: Improving intermediate care: giving practitioners a voice.Journal of Integrated Care 2004, 12(1):33-41.17. Anderson WL, Wiener JM, Khatutsky G: Workforce issues and consumersatisfaction in Medicaid personal assistance services. Health CareFinancing Review 2006, 28(1):87-101.18. Netten A, Jones K, Sandhu S: Provider and care workforce influences onquality of home-care services in England. Journal of Aging & Social Policy2007, 19(3):81-97.19. Nelson A, et al: Nurse staffing and patient outcomes in inpatientrehabilitation settings. Rehabilitation Nursing 2007, 32(5):179-202.20. Strasser D, et al: Team functioning and patient outcomes in strokerehabilitation. Archives of Physical Medicine & Rehabilitation 2005, , 86:403-408.21. Gindin J, et al: Predictors of rehabilitation outcomes: a comparison ofIsraeli and Italian geriatric post-acute care (PAC) facilities using theminimum data set (MDS). Journal of the American Medical DirectorsAssociation 2007, 8(4):233-42.22. Parker S, et al: Systematic review: Evidence of the Effectiveness ofIntermediate care..23. Brown L, Tucker C, Domokos T: Evaluating the impact of integratedhealth and social care teams on older people living in the community.Health & Social Care in the Community 2003, 11(2):58-94.24. Nancarrow SA, et al: The relationship between workforce flexibility andthe costs and outcomes of older peoples’ services. NIHR SDO:Southampton 2010, 296.25. Barton P, et al: A National Evaluation of the Costs and Outcomes ofIntermediate Care for Older People. The Unversity of Birmingham and TheUniversity of Leicester: Birmingham & Leicester 2005.26. Kaambwa B, et al: Costs and health outcomes of intermediate care:results from five UK case study sites. Health and Social Care in theCommunity 2008, 16(6):573-81.27. Collin C, et al: The Barthel ADL index: a reliability study. InternationalDisability Studies 1988, 10:61-63.28. Mahoney F, Barthel D: Functional evaluation: The Barthel Index. MarylandMedical Journal 1965, 14:61-65.29. Dolan P: Modelling valuations for EuroQol health states. Medical Care1997, 35:1095-108.30. Blough D, Ramsey S: Using generalized linear models to access medicalcare costs. Health Services and Outcomes Research Methodology 2000,1(2):185-202.31. Bland M: An introduction to medical statistics. Oxford: Oxford UniversityPress, 3 2000.32. Ramsey JB: Tests for Specification Errors in Classical Linear Least SquaresRegression Analysis. Journal of the Royal Statistical Society: Series B(Statistical Methodology) 1969, 31(2):350-371.33. Richardson G, et al: Skill mix changes: substitution or servicedevelopment? Health Policy (Amsterdam, Netherlands) 1998, 45(2):119-132.34. Clarke P, Gray A, Holman R: Estimating utility values for health states oftype 2 diabetic patients using the EQ-5D (UKPDS 62). Medical DecisionMaking 2002, 22:340-349.35. Saarni S, et al: The impact of 29 chronic conditions on health-relatedquality of life: a general population survey in Finland using 15 D andEQ-5D. Quality of Life Research 2006, 15:1403-1414.Pre-publication historyThe pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1472-6963/10/221/prepubdoi:10.1186/1472-6963-10-221Cite this article as: Dixon et al.: The relationship between staff skill mix,costs and outcomes in intermediate care services. BMC Health ServicesResearch 2010 10:221.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitDixon et al. BMC Health Services Research 2010, 10:221http://www.biomedcentral.com/1472-6963/10/221Page 7 of 7


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items