International Journal for Quality in Health Care 15:309-318 (2003)
© 2003 International Society for Quality in Health Care
Paper |
Development and initial validation of a measure of coordination of health care
National Centre for Epidemiology and Population Health, Australian National University, Canberra, Australian Capital Territory 0200, Australia
Objective. To describe the development and initial validation of the self-administered Client Perceptions of Coordination Questionnaire.
Design. The instrument was developed between 1996 and 1997 through iterative item generation; within a framework of six domains of coordination, addressed across four sectors of health care provision.
Setting. 1193 individuals with complex and chronic health care needs as judged by their general practitioners (GPs), who were participants in a 2-year randomized controlled trial of a coordinated care intervention in Australia. Other samples were collected in one general practice (98) and from attendees of a chronic pain management course (29).
Main measures. Face and content validity, completion rates, transferability, internal consistency, and construct validity of the 32-item instrument.
Results. Most items achieved excellent completion and comprehension rates. The instrument was transferable to another chronically unwell population. Cronbach's alpha of the entire instrument was 0.92, and for six individual scales scores ranged from 0.31 to 0.86. The six scales based on principal components analysis were acceptability, received care, GP, nominated provider, client comprehension, and client capacity. The first four scales were satisfactory, but the client scales were inadequate with poor internal consistency, and convergent and discriminant validity. People with chronic pain syndromes had significantly worse experiences for almost all items, supporting construct validity.
Conclusion. This instrument is one of the first to attempt to measure coordination of health care. Its strengths include ease of completion, transferability, and promising psychometric properties and construct validity. Problems capturing data about the patient's contribution to coordination highlight a lack of theoretical development in this area. A valid measure of coordination should be useful in needs assessment, program evaluation, and individual provider/practice audit, and would contribute to research into the experience and measurement of patient-focused care.
Keywords: coordinated care, health services research, patient evaluation of care
This paper reports the development and initial validity testing of the Client Perceptions of Coordination Questionnaire (CPCQ) during a trial of care coordination in Australia in the late 1990s. It was designed to address two facets of health care reform at this time—patient-centred care, and coordination of care between providers and over time. Selecting patients to evaluate coordination acknowledged the unique position they have at the centre of all their various services requiring coordination.
Three broad types of validity are assessed through the following sequential questions:
- Do items capture relevant perceptions about coordination, and do respondents understand terms as expected (i.e. content and face validity)?
- Are the instructions and items interpreted consistently by target populations, resulting in consistent profiles of responses between similar groups (i.e. transferability, an aspect of reliability)?
- Does the instrument have basic psychometric properties, particularly inter-item correlation, meaningful clustering of concepts and clinical plausibility for our a priori assumptions about coordination of care (i.e. construct validity)?
Methods
Design
The CPCQ was developed during 1996–1997 using standard techniques [1–3]. We addressed its theoretical framework, item content and format, response options, terminology, and length through a process of literature review, trial document review, focus groups, and field and pilot testing. Eighteen experts on community care evaluation reviewed a final draft in early 1997, and items were further tested in several field tests and a multi-centre pilot study.
We had judged that important dimensions of coordinated care were absent from existing patient assessment tools. These included: care from multiple providers for different conditions, interplay between health and social care, and comprehensive care over time (rather than discrete episodes). Furthermore, most instruments devised prior to the 1990s did not include patient involvement in decisions [4], active participation in managing chronic health conditions [5–7], or perceived need for care [8]. Evaluation has recently paid attention to these, with new instruments addressing experiences across health care interfaces and customer focus [9–11].
Lacking a theoretical framework for coordination of health care, we drew some items for the CPCQ from policy directives for the Australian coordinated care trials [12,13] that we expressed exclusively from a patient perspective. Further items were identified from the literatures on patient satisfaction [14–18] and organization of care [19,20], and from national consumer consultations about the proposed coordinated care trials [21,22]. We conceived of coordination as a complex construct, incorporating both overall impacts of care as well as discrete key processes. Questionnaire items were designed to capture aspects of coordination that were grouped into six domains: identification of need; access to care (drugs, tests or imaging, and services); patient participation, including empowerment; patient–provider communication; inter-provider communication; and global assessment of care.
Multiple items were developed for each domain relating to the patient's overall experience of all health and social care services. In keeping with common practice, a single item on coordination was included in the global assessment of care domain. We were also interested in experiences with a provider who may, from the patient's perspective, be playing a key role in care coordination. For the coordinated care trial, the patient's usual general practitioner (GP) was required to assume this role, but some patients also had other key providers (e.g. case managers). To accommodate both, we developed matching sets of items relating to care from both the GP and another key provider (nominated by the patient) about identification of needs, patient participation, and inter-provider communication. Several items also addressed the satisfaction and involvement of carers. Thus the six domains were addressed across four structurally distinct sections within the instrument: overall care; GP care; nominated provider care; and carers.
The items were expressed as questions about how often aspects of care were experienced—i.e. ‘How often did you...?’—in order to capture information averaged across providers and over time. Likert-type response options were chosen because they were more reliable than visual analogue scales in the pilot study, where the average age of respondents was 74 years. Most items had five frequency options: never, rarely, sometimes, mostly, and always. Questions about experiences of care were confined to the previous 3 months to standardize responses within a reasonable period of memory, yet allowing a long enough period in which to have experienced care from multiple providers.
Since the CPCQ was to be used predominantly with elderly patients with chronic complex health problems, we were unable to test a large bank of potential items. There was also a time pressure due to the trials' beginning at government direction. Final items were thus chosen from a limited set of 44 items tested with 110 participants in the multi-centred pilot study. The pilot included test–retest analysis of responses given 1–2 weeks later for 99 valid sets of data, with only reliable items retained. The final CPCQ was a 32-item instrument completed by either the patient or their carer. Of this total, 25 items were perceptions of care, two were preferences for care (see Appendix), and five identified the respondent and the type of key provider or carer being addressed (not further described).
Setting and intervention
The eight Australian coordinated care trials were a major health reform initiative aiming to overcome inequity of access to services and perceived problems of duplication and inefficient care for people with complex and chronic health conditions [13]. The majority of these people were envisaged to be old, frail, lacking strong social supports, and to have multiple health problems. They were thought to be the heavy users of health and social care services in Australia. These assumptions are widespread in developed countries and efforts to improve their care use similar approaches [23–25].
The trial of care coordination for which the CPCQ was developed was in the relatively affluent Australian Capital Territory (ACT). Patients were eligible if, in the opinion of their GP, they had chronic complex health problems and could benefit from improved coordination of their health and social care. The intervention was the completion of a multi-disciplinary ‘bio-psycho-social’ assessment and care plan coordinated by the GP, and purchase of subsequent services from pooled national and state medical and social care funds [26]. Because the CPCQ was to be used for both intervention and control (usual care) patients, it made no mention of the intervention. Additional data to test the CPCQ were collected from (a) another community-based sample drawn from one general practice in the ACT, using the same selection criteria as the trial (general practice validation sample); and (b) attendees of a chronic pain program. People with chronic pain experience failure of health services to ‘cure’ their pain and a stigma associated with complaining of chronic pain. The literature on chronic pain acknowledges problems managing this condition relating to its complexity, use of multiple providers, and over-reliance on investigations and new therapies, and endorses the need for coordination [27,28]. We therefore hypothesized that people with chronic pain would experience poor coordination.
Main measures
We considered the appropriateness of the content and structure of the CPCQ by examining item completion, comprehensibility, and transferability. Transferability, that is the extent to which similar groups respond to items similarly, was examined using the general practice validation sample. A number of psychometric statistics were used to examine relationships between items in the instrument [29,30]. A principal components factor analysis explored underlying components of responses to the instrument, and subsequently these were used to form six scales. Scales were desirable to reduce the data to several meaningful scores, and were examined for psychometric robustness as well as conceptual value. Correlation statistics included Cronbach's alpha as a test of internal consistency of both the entire instrument and subsequent derived scales, and Spearman correlation coefficients, firstly for each item with the global coordination item as a test of homogeneity, and secondly to examine discriminant scale validity of items with other scales. Construct validity was tested using a ‘known groups’ approach, a necessary early test in the absence of other measures of coordination [1,31]. As explained earlier, we assumed chronic pain sufferers to be a ‘known group’ at risk for poor coordination of care. Responses to items by people with chronic pain were compared to those who did not have chronic pain using the 
2 statistic.
The trial evaluation assumed that patients enrolled in a trial with chronic complex health conditions would have seen at least two providers in the previous 3 months, and the CPCQ was developed accordingly. However, in practice about one-quarter of respondents had not seen multiple providers, and therefore could not complete all items. Consequently, analyses have been conducted on two sample sets. Completion rates per item have been calculated using the full sample, while analyses relating to construct validity use only those respondents in two studies that had seen multiple providers (888 in the coordinated care trial and 80 in the general practice validation study).
Results
Of 1271 patients recruited to the coordinated care trial, 1193 (94%) completed the baseline CPCQ. Ninety-eight (78%) of 126 patients in the general practice validation study completed the CPCQ and SF-36. Patients with chronic pain were identified through a community-based chronic pain management course (29 people) and from the coordinated care trial sample (60 people); however, the pain group did not complete the SF-36. Trial patients were predominantly female, elderly, and were heterogeneous in terms of number and type of health conditions and severity of illness. They had a mean number of 5.6 chronic conditions, reflected in their SF-36 scores, which were well below Australian norms. Patients in the general practice validation sample were comparable to the trial group in terms of age, gender, and health status, while patients in the chronic pain group were on average younger (Table 1).
|
Completion and comprehension rates
Respondents should have completed all items in the section on overall care. The rate of missing responses for most items in this section was less than 1% with 93% (n = 1112) of respondents completing all 15 items. However, many more respondents than we anticipated had had no medicines (3.5%), no tests or assessments (11.1%), or no change in needs (34.0%) in the previous 3 months. Furthermore, for the 25% who had seen only one provider, responses about ‘conflicting advice’ and ‘confusion about roles of providers’ were sometimes inconsistent. Table 2 shows the proportions of missing responses, use of ‘out’ (or ‘not-applicable’) options and valid skips, and our assessment of ‘inconsistent’ responses for the first three sections of the CPCQ. The total number of ‘problem’ responses for an item is the sum of missing, out option, and inconsistent responses.
|
Respondents should also have completed all items in the GP section. However, only 81% (n = 966) did so. The largest proportion of missing responses in this section was for communication between providers (5.6%), and this also occurred with the nominated provider section (10.3%). A surprisingly high proportion (12.0%) of respondents felt that no decisions about care had been made by their GP in the previous 3 months.
Only 74% of respondents (n = 888) had seen a provider other than their GP in the past 3 months, and could therefore answer the questions in the nominated provider section. Of these, up to 10% of all five items had missing responses. There were problems with terminology for about 5% of respondents, who misinterpreted the terms ‘case manager’ and ‘service provider’. Misunderstanding was apparent from question marks and notes such as ‘I have no providers’ written alongside otherwise valid responses about a specialist or nurse. Some community providers, such as home help and respite workers, were not understood to be service providers and occasionally were included as carers. Respondents also occasionally nominated their GP or specialist physician as a case manager.
Transferability
The pattern of responses to the CPCQ for the trial and general practice validation groups was very similar (Figure 1) and no item had significantly different responses between the two populations. Furthermore, the ranking of items by mean score in each group was very similar (Table 3). The five most frequently experienced aspects of care were ‘get medicines’, ‘GP—involved in decisions’, ‘GP—agree about needs’, ‘satisfied with care’, and either ‘felt like complaining’ (general practice validation study) or ‘repeated tests unnecessarily’ (trial). See Appendix for details of descriptors.
|
|
Psychometric properties of the instrument
The following analyses test whether items in the instrument are linked to an underlying construct (in this case coordination), yet are significantly different in that they identify discrete ‘conceptual’ entities (i.e. scales). Several tests suggested the items were relevant to coordination. Cronbach's alpha of the 25 perceptual items was 0.92 (high internal consistency). Each perceptual item was statistically significantly correlated to the global coordination item. The correlation coefficients exceeded 0.40 for 10 of the remaining 14 items in the overall care section (range 0.26–0.72), for three of the five items in the GP section (range 0.29–0.48), and for three of the five items in the nominated provider section (range 0.39–0.49). However, it is noteworthy that the two items asking about preferences for involvement in decision-making (that were not perceptual items) were not correlated with the coordination item (–0.005 and –0.017). The magnitude of the unrotated factor loadings on the first principal component exceeded 0.55 for most items, and may be used as a test of unidimensionality [30].
A series of principal components factor analyses were conducted on three data sets from among those respondents who had seen more than one provider and had completed all items in any given section. These datasets consisted of 279 complete responses to 25 perceptual items across four sections of the questionnaire, 498 complete responses to the 19 items in the overall care and GP sections, and 543 complete responses to the 15 items in the overall care section only. A consistent patterning of components was found. Most items from the overall care section formed the first component, combining both global items (such as perceived quality) and specific aspects of care (such as waiting too long). However, the strength of the factor loadings for global and specific items was clearly distinguishable, so we divided the first component into two parts for subsequent scale development. Items about the two key providers always formed discrete components. Finally, items about the patients themselves, rather than about what providers had or had not done, clustered into two components. Eigen values for each of the five components and the strongest loading of each item using all 25 perceptual items are shown in Table 3. Each analysis explained a little over 50% of the variance between respondents, suggesting the need to further define what elements of coordinated care should be collected.
Six scale scores were calculated from the five components: global and specific scales (from the first component) and one scale each from the remaining four components. Cronbach's alphas indicated good internal consistency for four scales: ‘acceptability’ 0.86 (five items); ‘received care’ 0.75 (five items); ‘GP’ 0.72 (four items); and ‘nominated provider’ 0.80 (four items) [1]. Less internally consistent scales were ‘client comprehension’ 0.49 (three items), ‘client capacity’ 0.31 (four items), and a composite ‘client item’ 0.53 (seven items). Cronbach's alphas were recalculated with each single item removed in turn. The resulting minimum correlations for the scales of ‘acceptability’, ‘received care’, and the two provider scales were still high, exceeding 0.62 [32]. However, the adjusted client scale correlations ranged from only 0.23 to 0.60 (comprehension) and 0.13 to 0.27 (capacity). The poor internal consistency of these scales throws doubt on the reliability of any resulting scale score. Spearman correlation coefficients for all possible pairs of scales were lower than the internal consistency correlations, varying from 0.30 (‘GP’ and ‘client comprehension’) to 0.70 (‘acceptability’ and ‘received care’), supporting internal consistency of each scale.
Construct validity
Respondents with chronic pain reported significantly less favourable care in all but three items (i.e. 18 items). These items were ‘confused about providers’, ‘GP—future care’, and ‘understand conditions’. Unfavourable responses of never, rarely, and sometimes are combined for simplicity in Table 4, although the reported P values relate to all five response options. Overall, those with chronic pain felt they received fewer services, had lower quality services that were less responsive to their needs, and had more duplication of tests and assessments. They were also more likely to feel their care was not well coordinated and to feel like complaining. Only 65.5% of the chronic pain group felt their care was mostly or always well coordinated, compared with 83.9% of those without chronic pain.
|
Discussion
In order for the CPCQ to be useful we need to demonstrate that it could be understood and completed by the target population of predominantly elderly patients with complex and chronic care needs; was transferable between similar populations; had robust psychometric properties; and had construct validity.
In terms of comprehension, it was important that patients could understand the items, see the relevance to coordination, and choose responses from the available options [3,32]. High completion rates and a high degree of transferability suggest these were achieved. However, we experienced problems with comprehensibility relating to use of ‘out’ options and terminology. Allowing ‘not applicable’ or ‘out’ options is problematic when aiming to devise a generic instrument. Using them may be accurate (increasing respondents' ability to find an appropriate response option), but their use may lessen generalizability of findings and will impact on scoring. An alternative approach may be to determine that respondents have some need for coordination, for example by specifying a minimum number of different providers seen before completing the instrument. Even so, ‘out’ options may still be needed for some items, such as those about medicines. Terminology also remains a challenge when names change and patients confuse the roles and profession of providers.
In spite of these difficulties, we argue that items in the overall care section and the content of items in the provider sections are relevant to an instrument on coordination. Those in the overall care section tended to cluster into the first principal component and could be placed in scales with robust psychometric properties of internal consistency, and convergent and discriminant validity. The internal consistency of four scale scores was very good, with Cronbach's alphas between 0.70 and 0.90. While the Spearman correlation coefficients for each item with the global coordination item are likely to be statistically significant with such large sample sizes [33], coefficients exceeding 0.40 (a total of 16 items) could be considered adequate.
Our experience with focusing on single providers may not be a useful way to proceed in a coordination measure. Assuming that the coordinator may have a significant impact on successful coordination, we created opportunities to capture perceptions of two key providers. Not surprisingly the principal component analysis placed these items into separate components that might be expected from both the structure of the instrument, and that patients might separate experiences with one provider from all other considerations. The use of such scales to interpret coordinated care is not appealing, since coordinated care is a model seeking to break down barriers between providers. The question of whether or not there is value in including items about specific providers because of their assumed crucial role is open to debate, as is how to present and use such data.
For a measure of coordination to be useful we must have confidence in our capacity to reduce data to meaningful scale scores. Although four scales had good scale properties, further development was not justified due to the poor reliability of the client scales. Two of those four scales were also provider specific scales and we were already reluctant to further refine that approach.
The problem of poor validity of the client scales highlighted an important developmental issue—should an attempt be made to include items about patients themselves in the instrument? Are patients instrumental in coordination, as political rhetoric and consumer literature suggest? We regarded the role of patients actively participating in their care as an ideological, if not yet actual, feature of chronic illness care. Patient-centred approaches to chronic illness care, and the likely protective impact of strong social networks and personal commitment to maintaining good health, mandate that we consider the role of patients in coordination of their care. In practical terms we conceptualized the patient playing a role in iterative decision-making and communication—key problems for coordination [23]. We included several items in the CPCQ to capture this dimension. It is clear that we need to better conceptualize the role of the patient as an agent of coordination in order to improve the instrument.
The reliability and construct validity tests were promising but require further testing. Differences in responses between those with chronic pain and those without a formal diagnosis of chronic pain were clinically plausible and raise questions about the ease with which some conditions might be coordinated. However, the challenge for validating a completely new type of measure is to find data appropriate for analysis, for example of criterion validity. Larger scale test–retest studies would be needed on any revised instrument prior to further tests of construct validity as we have attempted.
A pressing need to empirically evaluate coordination of care for people with chronic conditions is a worldwide issue and not confined to the Australian setting. Although health systems vary, patients with multiple chronic health conditions everywhere are exposed to multiple providers with ensuing information and communication challenges. These patients must negotiate all manner of agency criteria limiting access to services and methods that prioritize patient needs. As providers increasingly attempt to adopt patient-centred approaches, it is timely and appropriate to consider patient perceptions as a measure of coordination.
The CPCQ's weaknesses point to gaps in our current thinking about coordination and the use of patient-derived data in evaluating health care. To measure patient perceptions of a very diverse range of needs, service types and providers over time is clearly very complex, and there is a need for work on both the theory of care coordination and who requires it. Any attempt to measure coordination must somehow incorporate the highly individualized health and social care needs of chronically unwell and aging populations. It must also address differences between people such as empowerment and self-management, since these skills may contribute to actual coordination as well as to how patients evaluate their experience of what is done for them. Although the client scales we derived had poor psychometric properties and require considerable development, we believe this dimension is important and should be included.
Conclusions
The CPCQ was designed to encompass a complex construct about multiple services delivered by different agencies, over time, and for all health problems. It is one of the first instruments attempting to collect this type of data in a short, self-administered instrument. Its strengths, weaknesses, and areas for future work can be identified.
Strengths include brevity, ease of administration, perceived relevance, and comprehension. The instrument was transferable and its psychometric properties promising in parts. Construct validity was demonstrated by significantly less favourable care for those with chronic pain syndromes than the remaining trial sample for most items.
Despite problems with terminology and the structure of the CPCQ, we can begin to understand what elements of coordination should be addressed and how these can be captured. A valid measure of coordination should be useful in needs assessment, program evaluation, and individual provider/practice audit, and contribute to research into the experience and measurement of patient-focused care. This study provides a starting point for further development of the theoretical dimensions, types of items, ways of expressing items, and ways of combining data needed to develop a measure of patient perceptions of coordination.
Appendix Items from the CPCQ
|
Address reprint requests to Dr Clare McGuiness, National Centre for Epidemiology and Population Health, Australian National University, Australian Capital Territory 0200, Australia. E-mail: clare.mcguiness{at}calvary-act.com.au 
Accepted for publication March 26, 2003.
References
- Streiner DL, Norman GR. Health Measurement Scales. A Practical Guide to their Development and Use. 2nd edn. Oxford: Oxford University Press, 1989.
- Tudiver F, Ferris LE. Creating an original measure. In Stewart M, Bass M, Dunn E, Norton PG, Tudiver F, eds, Tools for Primary Care Research. Newbury Park: Sage Publications, 1992: 86–96.
- Moser SC, Kalton G. Survey Methods in Social Investigation. 2nd edn. London: Heinemann Educational Books, 1969.
- Woolf SH. Shared decision-making: the case for letting patients decide which choice is best. J Fam Prac 1997; 45: 205–208.[Web of Science][Medline]
- MacStravic S. Optimising the patient's role. An essential component of disease management. Dis Manage Health Outcomes 1999; 6: 1–7.
- Gruman J, VonKorff M. Self-management services. Their role in disease management. Dis Manage Health Outcomes 1999; 6: 151–158.[CrossRef]
- Lang F. The evolving roles of patient and physician. Arch Fam Med 2000; 9: 65–67.
[Free Full Text] - Laerum E, Steine S, Finset A et al. Complex health problems in general practice: do we need an instrument for consultation improvement and patient involvement? Theoretical foundation, development and user evaluation of the Patient Perspective Survey (PPS). Fam Pract 1998; 15: 172–181.
[Abstract/Free Full Text] - Baker R, Preston C, Cheater F et al. Measuring patients' attitudes to care across the primary/secondary interface: the development of the patient career diary. Qual Health Care 1999; 8: 154–160.[Abstract]
- Safran DG, Taira DA, Rogers WH et al. Linking primary care performance to outcomes of care. J Fam Pract 1998; 47: 213–220.[Web of Science][Medline]
- Jung HP, Wensing M, Grol R. What makes a good general practitioner: do patient and doctors have different views? Br J Gen Pract 1997; 47: 805–809.[Web of Science][Medline]
- Department of Health and Human Services. Proposals for Trials in Coordinated Care: A Background Paper. Canberra: Commonwealth Department of Human Services and Health, 1995.
- Council of Australian Government. Health and Community Services: Meeting People's Need Better. Canberra: COAG, 1995. [Available on-line.] http://www.health.gov.au/archive/1999/coag1_95/coag1_95.htm (last accessed 21 February 2003).
- Baker R. Development of a questionnaire to assess patients' satisfaction with consultations in general practice. Br J Gen Pract 1990; 40: 487–490.[Web of Science][Medline]
- Baker R. The reliability and criterion validity of a measure of patients' satisfaction with their general practice. Fam Pract 1991; 8: 171–177.
[Abstract/Free Full Text] - Howie JG, Heaney DJ, Maxwell M, Walker JJ. A comparison of a Patient Enablement Instrument (PEI) against two established satisfaction scales as an outcome measure of primary care consultations. Fam Pract 1998; 15: 165–171.
[Abstract/Free Full Text] - Hall JA, Dornan MC. What patients like about their medical care and how often they are asked: a meta-analysis of the satisfaction literature. Soc Sci Med 1988; 27: 935–939.[CrossRef][Web of Science][Medline]
- Williams SJ, Calnan M. Key determinants of consumer satisfaction with general practice. Fam Pract 1991; 8: 237–242.
[Abstract/Free Full Text] - Wensing M, Grol R, Smits A. Quality judgements by patients on general practice care: a literature analysis. Soc Sci Med 1994; 38: 45–53.[CrossRef][Web of Science][Medline]
- Flocke S. Measuring attributes of primary care: development of a new instrument. J Fam Pract 1997; 45: 64–74.[Web of Science][Medline]
- Consumers Health Forum. Consumer Views on Coordinated Care. Canberra: Consumers Health Forum, 1995.
- Health Issues Centre. Developing a Consumer Focus for Coordinated Care. Health Issues Centre: Melbourne, 1996. [Available on-line.] http://home.vicnet.net.au/~hissues/resources.htm (last accessed 21 February 2003).
- Chen A, Brown R, Archibald N, Aliotta S, Fox PD. Best Practices in Coordinated Care. Princeton, NJ: Mathematica, Policy Research, Inc., 2000.
- Connell CM, Kole SL, Benedict CJ, Holmes SB, Gilman S, Beane GE. Increasing coordination of the dementia service delivery network: planning for the Community Outreach Education Program. Gerontologist 1994; 34: 700–706.[Abstract]
- Fine M. Coordinating Health, Extended Care and Community Support Services: Issues for Policy Makers and Service Providers in Australia. Sydney: Social Policy Research Centre, 1997. [Available on-line.] http://www.sprc.unsw.edu.au/dp/dp080_2up.pdf (last accessed 21 February 2003).
- Dugdale P, Richards B, Sibthorpe B, Fletcher M, Alexandric V, McGuiness C. Care Plus The Australian Capital Territory Coordinated Care Trial. In Leigh J, ed., The Australian Coordinated Care Trials. Background and Trial Descriptions. Canberra: Commonwealth Department of Health and Aged Care, 1999: 57–70.
- Becker N, Sjogren P, Bech P, Olsen AK, Eriksen J. Treatment outcome of chronic non-malignant pain patients managed in a danish multidisciplinary pain centre compared to general practice: a randomised controlled trial. Pain 2000; 84: 203–211.[CrossRef][Web of Science][Medline]
- Weisberg WB, Clavell AL Jr. Why is chronic pain so difficult to treat? Psychological considerations from simple to complex care. Postgrad Med 1999; 106: 141–160.[Medline]
- Dawson-Saunders B, Trapp RG. Basic and Clinical Biostatistics. East Newark: Appleton & Lange, 1994.
- Kline P. An Easy Guide to Factor Analysis. London: Routledge, 1994.
- Ferris LE, Norton P. Basic concepts in reliability and validity. In Stewart M, Bass M, Dunn E, Norton PG, Tudiver F, eds, Tools for Primary Care Research. Newbury Park: Sage Publications, 1992: 64–76.
- Foddy W. Constructing Questions for Interviews and Questionnaires. Cambridge: Cambridge University Press, 1993.
- Armitage P, Berry G. Statistical Methods in Medical Research. Oxford: Blackwell Scientific Publishing Co., 2001.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. F. Van Vorst, R. Araya-Guerra, M. Felzien, D. Fernald, N. Elder, C. Duclos, and J. M. Westfall Rural Community Members' Perceptions of Harm from Medical Mistakes: A High Plains Research Network (HPRN) Study J Am Board Fam Med, March 1, 2007; 20(2): 135 - 143. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||