grzybowksi

H

Pc

Sa

b

c

a

KRMM

1

rihatmas

WT

0d

ARTICLE IN PRESSG ModelEAP-2332; No.of Pages9Health Policy xxx (2009) xxxxxx

Contents lists available at ScienceDirect

Health Policy

journa l homepage: www.e lsev ier .com/ locate /hea l thpol

lanning the optimal level of local maternity service for small ruralommunities: A systems study in British Columbia

tefan Grzybowskia,b,, Jude Kornelsena,b, Nadine Schuurmanc

Centre for Rural Health Research, Vancouver Coastal Health Research Institute, CanadaDept of Family Practice, University of British Columbia, CanadaDept of Geography, Simon Fraser University, Canada

r t i c l e i n f o

eywords:ural health servicesaternal health servicesodels

a b s t r a c t

Objectives: To develop and apply a population isolation model to define the appropriatelevel of maternity service for rural communities in British Columbia, Canada.Methods: Iterative, mathematical model development supported by extensive multi-methods research in 23 rural and isolated communities in British Columbia, Canada, whichwere selected for representative variance in population demographics and isolation. Mainoutcome measure was the Rural Birth Index (RBI) score for 42 communities in rural BritishColumbia.Results: In rural communities with 1h catchment populations of under 25,000 the RBI scorematched the existing level of service in 33 of 42 (79%) communities. Inappropriate service

for the rural population was postulated and supported by qualitative data available on 6 ofthe remaining 9 communities.Conclusions: TheRBI is a potentially pragmatic tool in British Columbia to help policymakersdefine the appropriate level ofmaternity service for a given rural population. The conceptualstructure of the model has broad applicability to health service planning problems in otherjurisdictions.. Introduction

Rural maternity health services across Canada are cur-ently in flux, as evidenced by the closure of small andsolated services and the migration of women from theirome communities to give birth [15]. Health plannersre tasked with the challenge of making resource alloca-Please cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

ion decisions that are economically viable and meet theaternity health care needs of rural populations withincontext of competing priorities [6,7]. Additional pres-

ures arise out of the nature of health care delivery systems

Corresponding author at: Centre for Rural Health Research, 530-1501est Broadway, Vancouver, British Columbia, Canada V6J4Z6.

el.: +1 604 742 1794; fax: +1 604 742 1798.E-mail address: [email protected] (S. Grzybowski).

168-8510/$ see front matter 2009 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.healthpol.2009.03.007 2009 Elsevier Ireland Ltd. All rights reserved.

themselves,whichare characterizedby their dynamic com-plexity and lack of stasis. This is further complicated by thelack of a systematic approach to rural health services plan-ning and the absence of a robust evidence base to informsuch planning [710]. A review of policy documents fromBritish Columbia covering the past decade provides littleevidence of systematic planning formaternity care servicesin general and for rural maternity care services in particu-lar. This lack of direct policy attention to rural maternitycare means much of the decision making with respect toservices has occurred in an ad hoc manner in response toa local or regional sense of crisis [9]. This paper presentshe optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

a case study of systems modeling related to the assess-ment of the appropriate level of maternity services forcommunities in rural British Columbia, Canada with pop-ulations of less than 25,000 and addresses the question:What level of service should be provided to the community

dx.doi.org/10.1016/j.healthpol.2009.03.007http://www.sciencedirect.com/science/journal/01688510http://www.elsevier.com/locate/healthpolmailto:[email protected]/10.1016/j.healthpol.2009.03.007

INalth PolARTICLEG ModelHEAP-2332; No.of Pages92 S. Grzybowski et al. / He

based on population need in order to optimize service sus-tainability?

2. Background

British Columbia is Canadas western-most provincewith a population of 4 million, of which 25% lives outsideofmajor urban centres. The provinces geography is charac-terized by mountainous valleys and coastal communities,which present significant transportation challenges, espe-cially in inclementweather. In thepast 10years,many smallrural maternity services have closed in British Columbiaand across Canada [2,5,11], with 20 closures in BritishColumbia alone since 2000 [8,12]. These closures haveoccurred for a variety of reasons including the centraliza-tion of serviceswithin ahealth authority [4,13,14], concernsabout the safety of a small unit in the face of bad out-comes [5], and difficulties in recruiting providers to staffsmall rural maternity units [4,1518]. These closures havetaken place against a backdrop of policy recommendationsthat speak to the importance of supporting women to beable to give birth closer to home [19,20], and a relativelythin and inconclusive body of literature about the safety ofsmall rural maternity services with and without cesareansection capacity [2127].

A review of the literature on rural health servicesplanning demonstrates a number of approaches to theapplication of predictivemodeling to service delivery chal-lenges [2830]. However, most of these models attemptto include determining factors (geography and populationdemographics)withamultiplicityof feasibility characteris-tics (e.g. analysis of existing facilities and human resourcesissues) without prioritizing between the two types ofcharacteristicspopulation need and feasibility. For exam-ple, the British Columbia Standards of Accessibility identifypopulation characteristics and distance/geography along-side professional competence and critical mass in theirdetermination of minimal requirements of accessibility toservices [7]. Likewise, The Rural Birthing Services Frameworkfrom Australia identifies the importance of determiningpopulation need based on demographics and isolation asa means of elucidating the level of service desired througha consultative development process [31]. The frameworkdoes not, however, distinguish between core populationcharacteristics and associated feasibility characteristicssuchas the complementand trainingof the clinical staff andsupport services. While The Rural Birthing Services frame-work is conceptually useful, the criteria of how to apply itas a predictive tool are not explicit. A similar critique maybe levelled against Battye andMcTaggarts development ofa model for sustainable delivery of outreach allied healthservices to remote north-west Queensland, Australia [29].

Other research has investigated the relationshipbetween community characteristics and the presence ofmaternity services in small rural hospitals. For example,Lambrew and Ricketts (1993) applied logistic regression toPlease cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

understand the association between demographics, geo-graphic location, and socioeconomic status and the abilityof rural hospitals to sustain local obstetrical services. Theauthors concluded that Community conditions, such aslevels of unemployment, racial composition or proximity PRESSicy xxx (2009) xxxxxx

to other hospitals or metropolitan areas, may be primaryrather than secondary determinants of the viability of localobstetrical care systems [32, p. 284]. These findings sug-gest the importance of disaggregating the characteristicsof the population, such as population birth numbers, socialvulnerability, and isolation, from other social characteris-tics when planning services.

3. Methods

This model development set out to answer the ques-tion, Can we predict the appropriate level of sustainablematernity service for a rural community based on popula-tion need?Weused John vonNeumanns operationalizationof the term model to refer to a mathematical constructwhich, with the addition of certain verbal interpretation,describes observed phenomenon. The justification of sucha mathematical construct is solely and precisely that it isexpected to work [cited in 33, p. 273]. More recent workinforming our approach comes from thefield of operationalresearch and identifies stages of knowledge acquisition andmodel abstraction and highlights the advantages of keep-ing a model simple [3537]. The approach undertaken wasinformed by the recognition that systems are dynamic andmulti-faceted and must be studied using multiple meth-ods and an iterative process to ensure the responsivenessof the model [34]. Furthermore, we privileged a transdisci-plinary approach to model development that incorporatedin-depth understanding of the clinical and social reali-ties of rural maternity care. The former was informed byone of the authors (S.G.), who worked as a rural familyphysician providing maternity care in an isolated commu-nity for 11 years and the latter by a sociologist (J.K.), whohas studied the social aspects of childbirth for the past10 years. The research support team included individu-als with backgrounds in anthropology, geography, healthpolicy and administration, psychology, and epidemiology.The model development was underscored by a recognitionthat in the special circumstance of rural and isolated com-munities, three dominant characteristics are predictive ofrural service stability: population birth numbers, the socialvulnerability of the local population, and the degree of iso-lation of the community. Identification of the importance ofthese characteristics arose out of the authors comprehen-sive understanding of rural maternity care service deliveryissues based on prolonged research engagement with arange of rural study communities and key informants.

Prolonged engagement occurred directly through ninestudieswhich used both quantitative and qualitativemeth-ods. The first three studies have been completed anda number of publications have resulted [4,22,38]. Theremaining studies are currently in progress and includean investigation of: newborn outcomes by rural maternityservice level, 19941999; parturient women and fami-lies experience of birth in and away from their homecommunities [39,40]; care providers and administratorshe optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

experiences of providingmaternity care [41,42]; the effectsof small hospital maternity service closures on referralhospitals [43]; the experience of general practitioner sur-geons providing surgical obstetrics to rural communities;and the application of a logic model approach to the

dx.doi.org/10.1016/j.healthpol.2009.03.007

IN PRESSHalth Policy xxx (2009) xxxxxx 3

pt1iccCmtBtdcrEptba

3

mgwtsmom2iyibus

R

wbb

hsbB

Table 1Isolation factor scale.

Surface travel time Isolation factor (IF)

4h 4

(2) Defining the population birth score (PBS): We calcu-lated the average number of births in the 1h catchmentarea of a hospital over five years, and divided this numberby 10 to reduce the value of the variable to a more concisenumber.Hadwe left thevalue as a rawnumberof births, theresults of our formula would have run from 0 to 250 ratherthan from 0 to 25. We weighted the other components ofthe formula to reflect the values in the population birthscore, and applied this uniformly across all communities.

(3) Calculating the adjustment for population vulnerabil-ity (APV): Each community catchmentwas assigned a scoreof social vulnerability derived from the previously vali-dated British Columbia Statistics Socio-Economic Indices,which measure the social vulnerability of a Local HealthArea population over a range of1 (socially advantaged) to+1 (socially disadvantaged) based on the following factors:economichardship, crime, healthproblems, education con-cerns, children at risk, and youth at risk [46]. We initiallyapplied this variable to each communitys birth score byadjusting the score by a factor of 0.81.2 depending on theBC Stats score (where 0.8 correlated with a 1 score and1.2 correlated with a+1 score). Further sensitivity analysisdeterminedbest fitwith the existingqualitativedata for the23 communities in our program of research over a 0.81.4range. Asymmetry in the adjustments reflects an aware-ness that, in British Columbia, social vulnerability has amore significant influence on access to health services, and,ultimately, health outcomes than social advantage [39,40].

(4) Measuring proximity to nearest cesarean section ser-vice and attributing an isolation factor (IF): Surface traveltime to the servicewas categoricallyweighted as illustratedin Table 1. Even if a community currently had cesareansection services, in order to accurately assess a given com-munitys needs, the isolation factor was defined based onthe distance to the next closest service with cesarean sec-tion capability. Further, for populations that were less than1h from services, we recognized that proximity can detractfrom the sustainability of a maternity service and conse-quently weighted the factor accordingly (see Table 1). Adistance of 1h travel time is recognized as an importantthreshold for appropriate access to emergency services,including intrapartum care [7]. In our model, travel timesof greater than 1h were weighted in a stepwise fashionand reflected our qualitative findings which emphasizedthe important influence of isolation on sustainability ofhe optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

very small maternity services. We purposefully dealt withisolation as a summative component in the formula asour qualitative research suggested that it is only in smallbirthing populations (3090 births per year) that isolationis a critical factor.

dx.doi.org/10.1016/j.healthpol.2009.03.007

ARTICLE IN PRESSG ModelHEAP-2332; No.of Pages94 S. Grzybowski et al. / Health Policy xxx (2009) xxxxxx

maternFig. 2. 1h surface travel time catchments and

This formula results in a Rural Birthing Index (RBI)score for each rural British Columbia community in oursample. Each score correlates to a recommended servicelevel (see Table 2). We parameterized the model againstexisting stable rural maternity services in British Columbia(exhibiting these organizational properties) to calibrate theRBI scores to existing levels of service.We also scored com-munities that have changed service levels since 2000. Wedefined rural maternity service levels according to degreeof local cesarean section capability as, in British Columbia,Please cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

local surgical capability is themost important determinantof the proportion of women who will be able to deliverlocally [47,48]. Furthermore it is important to recognizethat cesarean section capacity in a community directly

Table 2Application of the RBI Score to community service levels.

Rural birth index(RBI) score

Maternity service level

07.0 ANo local intrapartum services7.09.0 BLocal intrapartum services without

operative delivery9.014 CLocal GP surgical services1427 DMixed model of specialists and GP surgeons>27 ESpecialist only modelsity service levels of rural BC hospitals, 2007.

depends on the presence of a broad range of ancillaryobstetric services, such as pediatric, anaesthetic, and nurs-ing support, an operating room, and neonatal resuscitationservices.

4. Results

The Rural Birth Index scores accurately reflect the exist-ing level of maternity service in 33 of 42 British Columbiacommunities (79%) (see Table 3). Of the remaining 9communities, we postulate that most are inappropriatelyserviced based on qualitative data gathered in 6 of the7 communities that we have visited to date, while theremaining two communities have not been included inqualitative data collection. Based on these results, theRBI tool has identified the appropriate level of ruralmaternity service for 38 of 42 communities studied. Forcommunities with an inappropriate or unsustainable levelof service the formula flags them as under- or over-served.he optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

The RBI tool identifies British Columbia communitiesfor which there is a difference between the optimal servicelevel predicted and the existing service. Table 3 categorizesthe 42 communities in our study by existing service level,and identifies the communities that do not currently pro-

dx.doi.org/10.1016/j.healthpol.2009.03.007

ARTICLE IN PRESSG ModelHEAP-2332; No.of Pages9S. Grzybowski et al. / Health Policy xxx (2009) xxxxxx 5

Table 3RBI scores applied to British Columbia rural maternity services.

Hospital catchmentpopulation

PBS APV Nearest C-sectionservice

Travel time Isolationfactor (IF)

Rural birthindex (RBI)

Predicted levelof service

Communities where intrapartum services have closed since 1996 (Existing service A)a

Kimberley 9,049 5.5 0.88 Cranbrook 32m 2 2.8 ASummerland 11,891 7.1 0.84 Penticton 20m 3 3.0 ASparwood 6,643 5.0 1.35 Fernie 30m 3 3.7 AEnderby 7,724 5.7 1.08 Vernon 43m 2 4.1 AAshcroft 2,900 1.8 1.20 Kamloops 1h 38m 2 4.2 APrinceton 4,899 2.4 1.24 Penticton 1h 57m 2 5.0 AAlert Bay 1,200 1.4 1.14 Comox 5h 44m 4 5.6 ALytton 5,000 3.2 1.20 Kamloops 1h 38m 2 5.8 ANakusp 5,194 3.2 1.00 Nelson 2h 32m 3 6.2 ACastlegar 13,614 9.0 0.94 Trail 33m 2 6.5 AClearwater 5,340 4.6 1.00 Kamloops 1h 32m 2 6.6 ABella Bella 1,200 2.1 1.27 Vancouver >4h 4 6.7 AHope 8,891 6.9 1.32 Chilliwack 45m 2 7.1c BMasset 2,700 3.0 1.12 Prince Rupert >4h 4 7.4 BPort Hardy 5,000 5.7 1.14 Campbell River 2h 47m 3 9.5 COliver 19,521 12.1 0.98 Penticton 45m 2 9.9 CMerritt 11,749 10.5 1.35 Kamloops 54m 1 13.2 CCommunities with no local surgical service (Existing service B)Queen Charlotte City 2,700 3.0 1.12 Prince Rupert >4h 4 7.4 BPort McNeill 3,500 4.0 1.14 Campbell River 2h 23m 3 7.6 BInvermere 10,559 6.1 0.93 Cranbrook 1h 41m 2 7.7 BSalt Spring Island 10,000 5.7 0.91 Victoria 2h 51m 3 8.2 BTofino 4,873 4.3 1.3 Port Alberni 2h 06m 3 8.4 BFort St. James 7,000 8.9 1.20 Vanderhoof 45m 2 8.7 BHazelton 5,756 7.3 1.34 Smithers 1h 11m 1 10.8 C

Communities with surgical services provided by General Practitioner Surgeons (Existing service C)Revelstokeb 8,593 7.9 0.90 Salmon Arm 1h 23m 1 8.1 BGoldenb 7,914 6.4 0.93 Cranbrook 3h 04m 3 9.0 CGrand Forks 10,992 7.2 0.99 Trail 1h 44m 2 9.1 CLillooet 4,800 5.4 1.21 Kamloops 2h 53m 3 9.6 CBella Coola 3,394 4.6 1.27 Williams Lake >4h 4 9.9 CKitimat 11,721 10.4 0.97 Terrace 1h 09m 1 11.1 CVanderhoof 8,000 10.1 1.20 Fort St. James 53m 1 11.2 C100 Mile House 14,945 10.2 1.05 Williams Lake 1h 23m 1 11.7 CBurns Lake 7,889 8.9 1.16 Vanderhoof 1h 32m 2 12.3 CCreston 12,961 12.0 0.99 Cranbrook 1h 34m 2 13.8 CFort Nelson 6,742 9.3 1.06 Fort St. John >4h 4 13.9 CSmithers 18,085 22.4 0.97 Hazelton 1h 11m 1 22.8 D

Communities with surgical services provided by General Practitioner Surgeons and Specialists (Existing service D)Fernie 15,894 12.0 0.92 Cranbrook 1h 26m 1 12.0 CTrail 20,325 13.3 0.98 Nelson 1h 07m 1 14.1 DPowell River 20,720 14.8 1.09 Comox 1h 38m 2 18.2 DNelson 24,930 21.0 0.96 Trail 1h 07m 1 21.1 DPrince Rupert 16,625 19.9 1.22 Terrace 1h 39m 2 26.3 D

Communities with surgical services provided by Specialists without General Practitioner Surgeons (Existing Service E)T

N olumbiat values dt

vh

ntiiMoLvv

errace 22,396 26.2 1.20 Kitimat

otes: (a) These are closed maternity services as defined by the British Cime alone due to frequent seasonal mountain pass road closures. (c) Boldhe predicted and existing service level.

ide their predicted level of service, the majority of whichave no local maternity care (Level A).For communities whose predicted level of service does

ot match the existing level of care, our program of quali-ative research indicates that consequences include servicenstability and a heightened sensitivity to factors influenc-ng unsustainable service. For example, the community ofPlease cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

asset on the Queen Charlotte Islands which scores 7.4n the RBI scale (scores of 7.09.0 correlate with serviceevel B, local maternity care without cesarean section) hasacillated between providing and not providing local ser-ices during the past 10 years. In this community, the safety0h 55m 1 30.3 EReproductive Care Program [8]. (b) Isolation is underestimated by travelenote RBI scores of communities for which there is a difference between

of providing local services without access to cesarean sec-tion is weighed against the reality of women refusing totravel away to access services and birthing outside thesystem. Each time the local service closes, community pres-sure mounts on the hospital and local providers to reopenservices. This culminated in 2008 in an innovative localcommunity initiative leading to the completion of a newhe optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

hospital facility that will provide enhanced local birthingservices.

The RBImodel indicates that the study communitywiththe greatest discrepancy between existing and predictedlevel of maternity service is Merritt, a community in the

dx.doi.org/10.1016/j.healthpol.2009.03.007

ARTICLE IN PRESSG ModelHEAP-2332; No.of Pages96 S. Grzybowski et al. / Health Policy xxx (2009) xxxxxx

anningFig. 3. A theoretical framework of pl

interior of British Columbia. The community lost Level Bmaternity services in 2001 andhas not provided local intra-partumcare since.Merritt has on average 105 annual births(PBS=10.5), a highly vulnerable population of women oflow socioeconomic status (APV=1.35), and is 54min fromthe nearest cesarean section service across a high moun-tain pass highway (IF =1). Consequently, Merritt has anRBI score of 13.2 and not only should have local maternityservices, but also local cesarean section when compared tocommunities in British Columbia with similar populationcharacteristics.

When a communitys RBI score is close to a servicetransition point (scores near 7, 9, and 14), the potentialfor influencing factors to dramatically shift levels of ser-vice is much higher than when the score is closer tothe midpoint of the range for a given service level (seeTable 2). In such instances, local system and resourcechallenges can outweigh population need. For instance,the small, isolated community of Bella Bella on BritishColumbias central coast (RBI score 6.7), lost local mater-nity services in 2001 due to a confluence of factorsPlease cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

including difficulty recruiting and retaining a general prac-titioner with enhanced skills (GP Surgeon) to provide localaccess to cesarean section, and the reluctance of physi-cians to offer maternity care without the availability ofsuch services [49]. The community has expressed a deter-and evaluating rural health services.

mination to have local birth return to Bella Bella, but careprovider recruitment and retention challenges continue toundermine the reestablishment of sustainable local ser-vices.

5. Discussion

This is a systems study of modeling a rural healthcare service based on community population size, socialvulnerability, and degree of isolation. The model hasbeen parameterized and tested for rural maternity healthservices within British Columbia. The Rural Birth Indexformula thatwe have derived andmodified based on sensi-tivity analysis and fit with our qualitative understanding ofrural community maternity service stability and appropri-ateness works well for the majority of the rural servicesin the province. This demonstrates the potential utilityof privileging population need and isolation as definingcharacteristics of sustainable services. Most importantly itdramatically highlights communities that have a servicelevel outof syncwith themajorityof communities inBritishhe optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

Columbia (e.g. Merritt; see Table 3) and provides policymakersandprogrammerswithobjectivedata toplanhealthservices to meet population need.

The consequences of providing an inappropriate levelof service for a rural community include negative med-

dx.doi.org/10.1016/j.healthpol.2009.03.007

INHalth Pol

icrreaotnhmoma[srtwaRttseds

modslatfFganab

trttpbsrctpeaftsrs

ARTICLEG ModelEAP-2332; No.of Pages9S. Grzybowski et al. / He

cal and psychosocial outcomes for patients as well asare providers [4042]. For example, within the context ofural birthing services in British Columbia, our exploratoryesearch suggests that if no local intrapartum servicesxist in a community where local services are appropriateccording to the RBI score, clinically significant numbersf womenwill choose sub-optimal alternatives to travelingo access maternity care at a referral hospital. These alter-atives include the 10 cm strategy (arriving at the localospital fully dilated to preclude transfer out of the com-unity), and even unassisted home birth [38,39]. Theseptions are associatedwith significantmorbidity and occurore frequently in communities that are underservednd where barriers to accessing maternity services exist3840].Where these sub-optimal solutions are being cho-en by birthing women, it indicates a need for local serviceeview on the part of administrators and planners. Further,he frequency of such solutions to diminished access to careill increase as the size of the birthingpopulation increasesnd as the social resources of the women decrease. TheBI score can be used in such circumstances to definehe optimal level of service for the community. Wherehe appropriate level of service is no local maternity care,trategies for efficient management of emergency deliv-ries and transfer can be developed, such as emergencyelivery skills upgrading for local medical and ambulancetaff.

The RBI model was designed to be applied to rural com-unities with populations under 25,000 and is focusedn access to maternity services. For larger maternity careelivery services, such as those found in urban centres, theystem is more complex and isolation is not a factor asocal surgical services are assumed.However, this formulaicpproach and the conceptual characteristics underpinninghe RBI may be adaptable to strategizing delivery modelsor other health services in rural and isolated communities.or instance, services such as cancer, palliative, or emer-ency care can be planned using an appropriate frameworknd adaptation of the RBImodel that focuses on populationeed, isolation, andvulnerability. Themodelmaybe furtherpplied to plan services to meet the needs of communitiesased on population projections [50].Adaptation of the formula for other rural jurisdic-

ions and health services requires parameterization of theelationships between the variables to reflect the charac-eristics of a given environment or service. For instance,he population birth score variable can be adapted to aopulation case volume score to measure the annual num-er of patients in a community seeking a particular healthervice. The adjustment for population vulnerability (APV)equires minor adaptation depending on the vulnerabilityharacteristics of populations within different health sys-ems and social contexts, and depends on the availability ofopulation-based social vulnerability data, such as socio-conomic status scores. The results of our APV sensitivitynalysis are relevant to the context of a Canadian publiclyPlease cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

unded system of medicine and the mitigating influenceshat universal access to health care provides. The effect ofocial vulnerability in other jurisdictions would need to beecalibrated based on a location-specific sensitivity analy-is. For example, adaptation of the formula for jurisdictions PRESSicy xxx (2009) xxxxxx 7

without publicly funded health care will have to accountfor lack of universal coverage and the consequent chal-lenges to access for the uninsured. In such jurisdictions,we would predict that the formula will require a greaterrange of adjustment for population vulnerability depend-ing on the proportion of a given rural population that iswithout medical insurance. We believe that the isolationfactor is transferable to other rural jurisdictions and ser-vices. In addition, the output of the adapted formula willhave to correspond with a model of optimal service levelsthat reflects the current health care system of a given juris-diction: existing levels of service, types of care providers,and service level transition points (see Table 2).

This case study provides rural health services plannersin British Columbia with a recommended appropriate andsustainable level of service for a given small rural commu-nity compared to other sustainable rural services in theprovince. This potentially provides an objective founda-tion for the planning process and a rationale for healthservice decisions. This is stage one of what we envisionas a three-stage planning process (see Fig. 3). Stage one,the deterministic stage, is the objectivemeasurement of thecharacteristics of population need related to the serviceunder study: population size, vulnerability, and isolation(i.e. determining the RBI). Stage two, the feasibility stage,addresses the question: What are the pragmatic issues thatneed to be considered in locating a particular health service ina given rural community? These issuesmay involve a reviewof existing facilities, availability of health human resources,history of the service in the community, and considera-tion of transport and economic issues. A potential approachto this stage can be a decision analysis framework. Stagethree, the prioritizing stage, is addressed at the senior plan-ning table for thehealth administrative agencyand involvesestablishing the importance of the given service within acontext of competing service issues. The methodology forstages two and three will be the subject of future researcharticles. In addition, population-based service catchments(see Fig. 2) canprovidehealth plannerswith thedata to sys-tematically monitor a broad range of health outcome andutilization indicators and form the basis for a comprehen-sive quality improvement framework [51].

6. Conclusion

The development and application of the Rural BirthIndex in the province of British Columbia, Canada, demon-strates the potential utility of systems thinking in creatinga pragmatic tool to help policymakers define the appropri-ate level of maternity service for a given rural communitybased on population need and degree of isolation. Programplanning can be facilitated by disaggregating the processinto three stages: determining need, assessing feasibility,and prioritizing service demands (Fig. 3). The efficacy ofthis modeling exercise was catalyzed by a multi-methodstransdisciplinary approach and prolonged immersion inhe optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

the phenomenon. Further study is needed to explicatethe application of the conceptual model to other ruraljurisdictions and health services. Specifically, further studymay address the adaptation of the RBI for rural maternitycare service in other jurisdictions, explore the formulaic

dx.doi.org/10.1016/j.healthpol.2009.03.007

INalth Pol

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

ARTICLEG ModelHEAP-2332; No.of Pages98 S. Grzybowski et al. / He

fit with care provider models in other health care sys-tems, and investigate the conceptual approach of planningother health services based on population need and isola-tion.

Conflict of interest

The authors have no potential conflicts of interest.

Acknowledgments

We would like to thank the Rural Maternity researchteam of Andrew Birse, Shelagh Levangie, Kate Leblanc,Cynthia Lin Hseih, and Sarah Munro for the tremendoussupport that they provided us in researching and prepar-ing this manuscript, and thank Janusz Kaczorowski for hisvaluable suggestions.Wewould also like to thank the Cana-dian Institutes of Health Research for funding this research,in addition to the Vancouver Coastal Health Research Insti-tute and theMichael Smith Foundation for Health Researchfor their ongoing support.

References

[1] Chamberlain M, Barclay K. Psychosocial costs of transferringindigenous women from their community for birth. Midwifery2000;16:11622.

[2] Hutten-Czapski PA. Decline of obstetric services in northern Ontario.Canadian Journal of Rural Medicine 1999;4:726.

[3] Jasen P. Race, culture, and the colonization of childbirth in northernCanada. Social History of Medicine 1997;10:383400.

[4] Kornelsen J, Grzybowski S. Is local maternity care an optional servicein rural communities? Journal of Obstetrics and Gynecology Canada2005;27:3279.

[5] Rourke JTB. Trends in small hospital obstetrical services in Ontario.Canadian Family Physician 1998;44:211724.

[6] Ministries of Health Services and Health Planning. Enhancing healthservices in remote and rural communities of British Columbia. Victo-ria, BC: Government of British Columbia; 1999.

[7] Ministries ofHealthServices andHealthPlanning. Standardsof acces-sibility and guidelines for provision of sustainable acute care servicesby health authorities. Victoria, BC: Government of British Columbia;2002.

[8] British Columbia Reproductive Care Program. Report on the find-ings of a consensus conference on obstetrical services in rural andremote communities. Vancouver, BC: British Columbia ReproductiveCare Program; 2002.

[9] Kornelsen J, Grzybowski S. Rural womens experiences of mater-nity care: implications for policy and practice. Ottawa, ON: Statusof Women Canada; 2005.

10] Lyons RF, Gardner P. Strategic initiative in rural and northernhealth research. Ottawa, ON: Canadian Institutes of Health Research,; 2001 [accessed 05.02.07].

[11] Hutten-Czapski PA. The state of rural health care. Presentation to theStandingSenateCommitteeonSocialAffairs, ScienceandTechnology,; 2001 [accessed05.02.07].

12] BC Perinatal Database Registry. British Columbia deliveries bymater-nal residence and delivery hospital highest level of service/care.Vancouver, BC: BC Reproductive Care Program; 20002007.

13] Benoit C, Carroll D, Millar A. But is it good for non-urban womenshealth? Regionalizing maternity care services in British Columbia.Canadian Review of Sociology and Anthropology 2002;39:37396.

14] Klein M, Johnston S, Christilaw J, Carty E. Mothers, babies, and com-munities: centralizing maternity care exposes mothers and babiesPlease cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

to complications and endangers community sustainability. CanadianFamily Physician 2002;48, 117779, 118385.

15] Chance G. Family practice obstetrics: time of opportunity. CanadianFamily Physician 1995;41:5489.

16] Larsen-Soles T. Physician numbers in rural British Columbia. Cana-dian Journal of Rural Medicine 2001;6(1):2430.

[

[

PRESSicy xxx (2009) xxxxxx

[17] Thommasen HV, Berkowitz J, Grzybowski S. Community factorsassociated with long-term physician retention. BC Medical Journal2000;42:4269.

18] Thommasen HV, Lavanchy M, Connelly I, Berkowitz J, GrzybowskiS. Mental health, job satisfaction, and intention to relocate amongphysicians in rural British Columbia. Canadian Family Physician2001;47:73744.

19] First Nations Leadership Council and the Government of BritishColumbia. The transformative change accord. Victoria, BC: FirstNations Health Plan; 2005.

20] First Nations and Inuit Health Branch. Maternal child health programfor First Nations communities. Presented at: First Nations and InuitHealth BranchMaternal andChildHealth ProgramMeeting; February1, 2006; Vancouver, BC.

21] Cameron B, Cameron S. Outcomes in rural obstetrics, Atherton Hos-pital 19912000. Australian Journal of Rural Health 2001;1: S3942.

22] Grzybowski S, Cadesky AS, Hogg WE. Rural obstetrics: a 5-year prospective study of the outcomes of all pregnancies in aremote northern community. Canadian Medical Association Journal1991;144:98794.

23] Heller G, Richardson DK, Schnell R, Misselwitz B, Knzel W, SchmidtS. Are we regionalized enough? Early-neonatal deaths in low-riskbirths by the size of delivery units in Hesse Germany 19901999.International Journal of Epidemiology 2002;31:10618.

24] Leeman L, Leeman R. Do all hospitals need cesarean deliverycapability? An outcomes study of maternity care in a rural hospi-tal without on-site cesarean capability. Journal of Family Practice2002;51:12934.

25] Moster D, Terje Lie R, Markestad T. Neonatal mortality rates in com-munities with small maternity units compared with those havinglarger maternity units. British Journal of Obstetrics and Gynecology2001;108:9049.

26] Rosenblatt RA, Reinken J, Shoemack P. Is obstetrics safe in small hos-pitals? Evidence from New Zealands regionalised perinatal system.Lancet 1985;2(8452):42932.

27] Viisainen K, Gissler M, Hemminki E. Birth outcomes by level ofobstetric care in Finland: a catchment area based analysis. Journalof Epidemiology and Community Health 1994;48:4005.

28] Allen D. Relationship of infant mortality to the availability of obstet-rical care in Indiana. Journal of Family Practice 1991;33(6): 60913.

29] Battye K, McTaggart K. Development of a model for sustainabledelivery of outreach allied health services to remote north-westQueensland, Australia. Rural and Remote Health 2003;3(3):194.

30] Horev T, Pesis-Katz I, Mukamel D. Trends in geographic dispari-ties in allocation of health care resources in the US. Health Policy2004;68:22332.

31] Rural and Regional Health Services Branch, Rural and Regional Healthand Aged Care Services. Rural birthing services: a capability basedplanning framework. Melbourne, Victoria: Victorian GovernmentDepartment of Human Services; 2004.

32] Lambrew JM, Ricketts TC. Patterns of obstetrical care in single hospi-tal, rural communities. Medical Care 1993;31:82233.

33] Gleick J. Chaos:makinganewscience.Middlesex,UK:PenguinBooks;1987.

34] Begun JW, Zimmerman B, Dooley KJ. Health care organizations ascomplex adaptive systems. In: Mick SS, Wyttenback ME, editors.Advances in health care organization theory. San Francisco, CA:Jossey-Bass; 2003. p. 25388.

35] Kotiadis K, Robinson S. Conceptualmodelling: knowledge acquisitionand model abstraction. In: Mason SJ, Hill RR, Mnch L, Rose O, Jeffer-son T, Fowler JW, editors. Proceedings of the 2008 winter simulationconference; December 710, 2008; Miami, Florida. California: IEEE:Computer Society Press; 2008. p. 9518.

36] Robinson S. Simulation: the practice of model development and use.Chichester: John Wiley and Sons; 2004.

37] Robinson S. Issues in conceptual modelling for simulation: setting aresearch agenda. In: Robinson S, Taylor S, Brailsford S, Garnett J, edi-tors. Proceedings of the 2006 operational research society simulationworkshop. Birmingham, UK: Operational Research Society; 2006. p.16574.

38] Kornelsen J, Grzybowski S. The reality of resistance: the experiencesof rural parturientwomen. Journal ofMidwifery andWomensHealth2006;51:2605.he optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

39] Kornelsen J, Grzybowski S. Safety and community: thematernity careneeds of rural parturient women. Journal of Obstetrics and Gynecol-ogy Canada 2005;27:55461.

40] Kornelsen J, Grzybowski S. The costs of separation: the birth experi-ences ofwomen in isolated remote communities in British Columbia.Canadian Womens Studies 2005;24(1):7580.

dx.doi.org/10.1016/j.healthpol.2009.03.007http://www.irsc.gc.ca/e/27537.htmlhttp://www.srpc.ca/librarydocs/Kirby_SRPC.htm

INHalth Pol

[

[

[

[

[

[

[

[

[

[

ARTICLEG ModelEAP-2332; No.of Pages9S. Grzybowski et al. / He

41] Grzybowski S, Kornelsen J, Cooper E. Rural maternity care servicesunder stress: the experiences of providers. Canadian Journal of RuralMedicine 2007;12(2):8994.

42] Kornelsen J, Grzybowski S. Obstetric services in small rural commu-nities: what are the risks to care providers? Rural and Remote Health2008;8:943.

43] Kornelsen J, Moola S, Grzybowski S. Does distance matter? Increasedinduction rates for rural women who have to travel for intrapartumcare. Journal of Obstetrics and Gynecology Canada 2009;31(1):217.

44] Saltelli A, RattoM, Andres T, Campolongo F, Cariboni J, Gatelli D, et al.Global sensitivity analysis: the primer. Toronto: Wiley-Interscience;2008.Please cite this article in press as: Grzybowski S, et al. Planning tcommunities: A systems study in British Columbia. Health Polic

45] Schuurman N, Fiedler R, Grzybowski S, Grund D. Defining rationalhospital catchments for non-urban areas based on travel-time. Inter-national Journal of Health Geographics 2006;5(43):111.

46] Ministry of Labour and Citizens Services, British Columbia. BC Stats:socio-economic indices by local health area, [accessed 05.02.07].

[

PRESSicy xxx (2009) xxxxxx 9

47] Lynch N, Thommasen H, Anderson N, Grzybowski S. Does havingcesarean section capability make a difference to a small rural mater-nity service? Canadian Family Physician 2005;51(9):1239.

48] Kornelsen J, Grzybowski S, Iglesias. S. Is local maternity care sus-tainable without general practitioner surgeons? Canadian Journal ofRural Medicine 2006;11(3):21820.

49] Kornelsen J, Kotaska A, Waterfall P, Willie L, Wilson D. The geogra-phy of belonging: the experience of birthing at home for Aboriginalwomen. Unpublished data.

50] Ministry of Labour and Citizens Services, British Columbia.BC Stats: British Columbia and regional population projec-tions, [accessed 05.02.07].he optimal level of local maternity service for small ruraly (2009), doi:10.1016/j.healthpol.2009.03.007

51] Centre for Rural Health Research, editor. Proceedings from theBest Practices in Rural Maternity Care Service Outcomes Report-ing Conference; 2008 May 2; Richmond, BC. Vancouver: Centrefor Rural Health Research, ; 2008 [accessed 18.02.09].

dx.doi.org/10.1016/j.healthpol.2009.03.007http://www.bcstats.gov.bc.ca/data/sep/i_lha/lha_main.asphttp://www.bcstats.gov.bc.ca/data/pop/pop/popproj.asphttp://www.ruralmatresearch.net/documents/BestPracticesinOutcomesReportingProceedings-CRHRMay2008.pdf

Planning the optimal level of local maternity service for small rural communities: A systems study in British ColumbiaIntroductionBackgroundMethodsThe formula

ResultsDiscussionConclusionConflict of interestAcknowledgmentsReferences