Sears, Jeanne M. PhD, MS, RN; Wickizer, Thomas M. PhD, MPH; Franklin, Gary M. MD, MPH; Cheadle, Allen D. PhD; Berkowitz, Bobbie PhD
The quality of occupational health care has been linked to outcomes such as duration of work-related disability and is therefore increasingly receiving attention in efforts to improve outcomes and reduce costs in workers’ compensation systems.1,2 Workers’ compensation-related health care is in particular need of quality improvement; outcomes, medical costs, and patient satisfaction are generally worse than for medical care in general.2,3 Nevertheless, monitoring and improving quality of care is challenging, and there is much variation between jurisdictions in terms of both the specifics of how workers’ compensation benefits are provided and the data available to conduct system-level research.2
Administrative data offer an efficient way of tracking quality improvement efforts, but have been underutilized to date in the workers’ compensation arena.4 The use of administrative databases to assess quality of care requires clarity regarding how the quality and process of care might be represented in available data, a plausible link with outcomes, and careful consideration of the constraints of the data structure.5 Nevertheless, administrative data can be used to evaluate both system-level and provider-level processes that affect work-related disability.6
In Washington State, most health care for injured workers is supervised by a single public payer, the Washington State Department of Labor and Industries (L&I), and therefore substantial leverage can be brought to the development, implementation, and evaluation of innovative community-wide health delivery interventions aimed at improving quality and preventing disability.7 L&I has previously conducted a formal evaluation of a statewide managed care pilot project and is in the process of evaluating an innovative quality improvement pilot project, the Washington State Centers of Occupational Health and Education (COHE) project, which offers financial incentives for occupational health best practices in two large regions of the state.2,7 In 2004, the Washington State legislature enacted a 3-year pilot program that authorized nurse practitioners (NPs) to function as attending providers for injured workers. The formal evaluation of this program offered the opportunity to examine the utility of several potential quality and process of care measures available from L&I’s administrative databases.
A primary goal of this study was to identify quality and process of care indicators available in administrative workers’ compensation data and document their association with work disability outcomes. A second goal was to demonstrate the application of these indicators by using them to assess whether NPs performed differently than did primary care physicians (PCPs) in serving as attending providers for workers with uncomplicated back injuries.
Materials and Methods
Approximately 70% of non-federal employees in Washington are covered by the State Fund (the state-administered workers’ compensation insurance fund). Self-insured employers account for the balance. Injured workers can select the health care provider of their choice, but the provider must enroll with L&I before billing for workers’ compensation-related services. Providers of workers’ compensation-related care in Washington are legally required to file an accident report within 5 days of identifying a work-related injury or illness, and the provider that files the accident report is recorded as the first attending provider. Workers have the right to change attending providers with notice to L&I. The attending provider is responsible for ensuring adequate monitoring and care for the injured worker, including both health care and disability management. In Washington State, a claimant becomes eligible for time loss compensation after losing >3 days from work. (Time loss refers to compensated time away from work for a work-related injury or illness.)
Substitute House Bill (SHB) 1691 (Washington State Laws of 2004, chapter 65) took effect July 1, 2004 and authorized NPs to independently perform those functions of an attending physician within their scope of practice, with the exception of rating permanent impairment. This program was implemented because of concerns about timely access to health care for injured workers in rural areas, delays in work-related injury reporting, and efforts by NPs to expand their scope. Before SHB 1691, NPs could be reimbursed for providing health care to injured workers, but were restricted from performing some functions related to disability management (such as signing accident report forms and certifying time loss benefits). Many studies have documented that NPs provide safe and cost-effective care, generally equivalent in quality to that of physicians,8–15 but little information is available regarding care provided by NPs specifically to injured workers.16
Study Population and Data Sources
The study population included workers 18 to 70 years of age who were injured on or after July 1, 2004 and had both 1) a first medical visit at a primary care facility (eg, office, clinic, urgent care), and 2) an accepted State Fund workers’ compensation claim filed by an NP or PCP, between July 1, 2004 and June 30, 2005 (the first year after implementation of SHB 1691). PCPs were defined as allopathic and osteopathic physicians (MDs and DOs) specializing in general practice, family practice, or internal medicine. (L&I does not record specialty for NPs.) Claims were included only if the injury occurred in Washington and the first attending provider was located in Washington. Providers reporting >500 claims in the year after implementation were excluded, because such high claims activity likely reflected a more specialized practice of occupational health rather than primary care. Providers participating in the COHE project were also excluded, because of the financial incentives offered for occupational health best practices.
To appropriately assess differences in process indicators across provider type, the identification of a relatively homogeneous sample of claimants with a prevalent disorder was desirable.5,17 The injury category of back sprains or strains was selected for this study (using American National Standards Institute Z16.2 coding18) because it was the largest in absolute numbers and accounted for the largest proportion of medical costs and time loss days. To increase homogeneity, the sample was further restricted by excluding those claims with evidence of high baseline severity or complexity. To this end, fatalities and claims with total permanent disability were excluded (based on claims data), as were claims with inpatient, ambulance, or emergency department services at the first medical visit (based on medical billing data). Claimants presenting with baseline radiculopathy (based on medical billing data), coexisting conditions that might delay recovery, or who had received prior treatment for the same or similar condition were also excluded. Information on prior treatment and coexisting conditions was obtained from the provider via the accident report form.
L&I provided claimant, provider, and medical billing data through June 30, 2006, resulting in 12 to 24 months of follow-up data for each claim. Data was extracted on October 2, 2006 to allow for 3 additional months of bill processing time. The University of Washington Human Subjects Division reviewed and approved this study.
Study Design and Data Analysis
This study used administrative data to compare the performance of NPs with that of PCPs to assess the potential impact of SHB 1691’s expansion of the definition of attending provider on quality of care for injured workers.19 Claims were assigned to the NP or PCP group based on the initial attending provider type, and the two groups were compared using a two-sample test of proportions for each process indicator.
Standardization has been recommended when assessing quality in a diverse population.17 Although we controlled for injury type and severity via exclusion and there was little difference in employment-related variables across provider type, NPs in this sample were more likely to be responsible for the care of claimants who were younger and female (see Results). We used direct standardization on the basis of age (categorized as 18 to 24, 25 to 34, 35 to 44, 45 to 54, and 55 to 70 years) and sex to generate the proportion of claims for which each process of care indicator was met, using the entire population-based sample as the reference population. Therefore, the standardized results reflect theoretical proportions that would hold if NPs and PCPs were responsible for the care of injured workers with the same age and sex distribution (that of the overall sample). Direct standardization is a reasonable alternative to logistic regression when the number of comparison groups is small and a straightforward presentation of proportions is preferable to odds ratios.20 Findings based on direct standardization were compared with findings from logistic regression models in which robust variance estimates were used to account for heteroskedasticity, and standard errors were adjusted to account for any correlation of outcomes within a specific provider’s practice (which could be related to unmeasured factors such as office procedures and training in occupational health). All statistical analyses were conducted using Stata 8.2 for Windows (StataCorp, College Station, TX).
A number of claim characteristics were compared across attending provider type. These included 1) whether a claimant received any time loss compensation, 2) whether there was a permanent partial disability payment, 3) whether the claim remained open at 6 and 12 months after report filing, 4) whether a closed claim had been reopened, 5) whether there were any transfers of attending provider over the life of the claim, 6) whether the claimant had attorney representation, and 7) whether any claim disputes (protests or appeals) had been filed by the claimant, provider, or employer. These comparisons are presented to give a picture of possible differences in case mix and administrative friction between NP and PCP claims, but these characteristics are only partially under the control of the attending provider. They could also be affected by unmeasured worker, injury, and employment characteristics, and are intertwined with claim costs and disability in complex ways. For each of these indicators, a lower percentage would be considered “better” (theoretically associated with lower potential friction cost, dissatisfaction, disability, etc).21,22
Quality and Process of Care Indicators
The following quality and process of care indicators were identified for use in this study based on meeting the criteria of availability in administrative data, relevance to assessing differences in quality and process of care across provider type, and plausible association with work-related disability: 1) whether the first visit occurred within 1 day of injury, 2) whether the accident report was filed within 7 days of the first visit, 3) whether there was early telephone communication about care coordination or treatment, 4) whether work-relatedness was specified on the accident report, and 5) regularity of care while on time loss. Variants of these indicators were evaluated for use in the COHE project via a systematic review of the best available data and the development of formal consensus among clinical experts,2,7,23 and were among the quality and performance measures for process of care suggested by Pransky et al.1
Each process indicator was framed as a binary variable, so that each claim either met or did not meet the criteria for each indicator. The association of the first four process indicators with disability was described in two ways: 1) the percent of claims involving any time loss compensation, and 2) the median number of compensated time loss days for those claims with any time loss compensation. These were baseline indicators, in the sense that they were met (or not) very early in the natural history of a claim, so that associations with the presence or duration of time loss can be at least tentatively interpreted as causal. The association of the more downstream regularity of care indicator with time loss compensation was described using only median number of time loss days, because the denominator included only claims with any time loss compensation.
As a step toward further validation of these indicators, association with claim volume was assessed, because volume has often been associated with quality and performance in clinical settings.24,25 For each provider, the number of claims for which that provider filed an accident report or served as an attending provider at any point between July 1, 2004 and June 30, 2005 (the year after implementation) was derived using all State Fund claims (not solely those claims eligible for inclusion in this study). As a proxy for familiarity with the L&I system, providers who were responsible for at least the median provider-level claim volume of 24 claims during the year after implementation of SHB 1691 were considered higher volume providers (because volume might affect a provider’s efficiency or knowledge of best practices).
Each of the five selected quality and process of care indicators is described in more detail below. This includes the rationale for selecting the indicator and the criteria used to determine whether each indicator was met.
First Visit Within 1 Day of Injury.
Reducing delays in diagnosis and treatment may shorten work absence.26,27 L&I’s Attending Doctor’s Return to Work Desk Reference lists same-day scheduling for work-related injuries or illnesses as a best practice.28 We considered this indicator to be met if the first medical visit occurred within 1 calendar day of the injury, because we had no information on the time of day that the injury occurred.
Report Filed Within 7 Days of First Visit.
Delays in accident report filing are associated with longer work absence.27,29,30 Providers of workers’ compensation-related care in Washington State are legally required to file the accident report within 5 days of identifying a work-related injury or illness; however, compliance is inconsistent. Time from the first visit to filing of the accident report was identified in the COHE quality indicator report as a useful performance indicator, and the COHE project uses 2 days as a benchmark for incentives.23,27 We considered this indicator to be met if the accident report was received by L&I within 7 calendar days of the first medical visit, reflecting the 5-day requirement and a 2-day allowance for weekends.
Telephone Communication About Case at the First Visit.
Return to work can be facilitated by improving communication between health care providers, employers, and injured workers, and communication failures are linked with poor outcomes.26,31,32 We considered this indicator to be met if there was a submitted and allowed bill for discussing or coordinating care or treatment with the injured worker, department staff, nurse case manager, department medical consultant, or employer by telephone on the first medical visit date. (Telephone calls for authorization, resolution of billing issues, or ordering prescriptions are not payable and were not included.) The COHE quality indicator report recommended the use of two-way communication with the employer at the first visit when the worker is off or expected to be off work.23 Unfortunately, the special billing codes created for the COHE project and used to identify such calls were unavailable to the non-COHE providers in this study. We did not limit the denominator to claims resulting in compensated time loss, because it was feasible that early communication could prevent time loss of >3 days via arrangements for modified work. But, we also recognized that many claims would not require such communication.
Work-Relatedness Specified on Report.
Absent or inadequate documentation of work-relatedness on the accident report can delay claim approval and delay payment for medical care and time loss.23 We considered this indicator to be met if question 48 on the accident report form (“Was the diagnosed condition caused by this injury or exposure?” Yes, Probably, Possibly, No) was answered rather than left blank. This measure reflects the care taken by a provider to document whether an illness is work-related, and was recommended in the COHE quality indicator report.23 The COHE report also recommended the use of two other aspects of work-relatedness documentation (documentation of the presence or absence of a disorder and adequate description of relevant work or incident history), but those aspects were not available in the administrative data.
Regularity of Care While on Time Loss.
Continuity of care, defined variously as transfers of care, regularity of visits, intensity of care, breaks in contact, etc,33 has been linked with improved outcomes and less unnecessary utilization.34–36 The COHE quality indicator report23 recommended that continuity of care be operationalized as the percentage of workers who had a health care provider visit every 2 weeks for the first 2 months while they had not yet returned to work (this is also considered a measure of regularity of care37). The COHE report also recommended there be at least one visit 2 to 4 weeks after return to work; however, the date of return to work was not available in the administrative data. Only claims with >2 weeks of time loss were included in the denominator for this indicator. Any visit (other than the first medical visit) was considered qualified if it was billed using an Evaluation and Management Current Procedural Terminology code and was allowed by L&I. We considered this indicator to be met if any of the following were true:
* At least 2 but less than 4 weeks of time loss and at least one qualifying visit within 2 weeks of the first visit.
* At least 4 but less than 6 weeks of time loss and at least one qualifying visit in each of the first two 2-week periods.
* At least 6 but less than 8 weeks of time loss and at least one qualifying visit in each of the first three 2-week periods.
* More than 8 weeks of time loss and at least one qualifying visit in each of the first four 2-week periods.
Eighty-five NPs and 974 PCPs were recorded as first attending providers for the 2779 eligible back sprain or strain claims filed during the year after SHB 1691 implementation (some providers filed accident reports for more than one claimant). Of the PCPs, 17% were internists, 68% were family practitioners, and 15% were general practitioners. (L&I does not record specialty for NPs.) NPs tended to be responsible for fewer claims; 65% of NPs were responsible for fewer than 24 claims during the first year after implementation, compared with 47% of PCPs (P = 0.002).
Claimant and Employer Characteristics
Of the 2779 eligible back sprain or strain claims, 217 were filed by NPs and 2562 by PCPs. Mean age was 34.8 for claimants seeing NPs, compared with 37.0 for those seeing PCPs (P = 0.004). Male claimants were more likely to have PCPs as their attending provider (Table 1). Median pre-injury monthly income was $1936 for claimants seeing NPs, compared with $2227 for those seeing PCPs (computed using only the 870 claims involving time loss; P = 0.08). Median claim duration (measured from accident report filing to claim closure) was 91 days for both NP and PCP claims. The distribution of employment-related variables was quite similar between NP and PCP claims (Table 1).
NP claims were less likely to have any time loss compensation than were PCP claims (Table 2). There were no significant differences in other claim characteristics across provider type. Crude percentages and P values were almost identical to those derived (and presented in Table 2) using direct standardization by age and sex.
Quality and Process of Care Indicators
There were fairly strong associations of these process indicators with disability (Table 3). Claims for which there was a first visit within a day of injury were less likely to have any time loss, and among those with time loss, duration was shorter. Time loss duration was shorter when the accident report was filed within 7 days. Claims for which there was telephone communication at the first visit were less likely to have any time loss, but time loss duration was not affected. Claims for which work-relatedness was specified on the accident report tended to have a shorter duration of time loss, though there was no association between work-relatedness and whether there was any time loss. Lastly, regularity of care while on time loss was associated with shorter time loss duration.
There were also strong associations of these process indicators with claim volume (Table 4). Providers responsible for fewer claims were significantly less likely to meet every process indicator, with the exception of regularity of care.
Age and sex standardized percentages of claims meeting each process indicator by provider type are presented in Table 5. There was wide variation in the percentage of claims meeting each indicator, from under 5% for telephone communication at the first visit to over 95% for documentation of work-relatedness on the accident report. In general, there were few significant differences between NP and PCP claims. Claimants seeing NPs were somewhat more likely to be seen within 1 day of injury (52.4% compared with 44.4% for those seeing PCPs, P = 0.03). Logistic regression models that controlled for age and sex and accounted for the correlation of outcomes within a specific provider’s practice did not provide evidence for significant differences between NP and PCP claims in the odds of having met any of these indicators.
This study analyzed a number of quality and process of care indicators available in Washington State workers’ compensation administrative data. There were fairly strong associations of these indicators with claim volume and with measures of disability outcomes (any time loss and duration of time loss). We then used these indicators to evaluate the impact of pilot legislation authorizing NPs to function as attending providers. Related studies evaluating this pilot program did not find evidence for any negative impact on medical costs or disability outcomes and suggested that authorizing NPs as attending providers for injured workers may be a cost-effective approach to address access barriers.38,39 Nevertheless, process measures provide a more appropriate and sensitive “closer to the source” assessment of provider practice differences than do outcome measures.40–42 Using this set of indicators, we found no evidence of systematic differences in the care provided to claimants with uncomplicated back injuries based on first attending provider type (NP or PCP).
Although not intended to be definitive measures of quality or of provider performance, these indicators demonstrated some potential as metrics for tracking system performance and improving disability outcomes. We limited our sample to back sprains or strains to maximize homogeneity; however, these indicators are generic and could be applied across conditions. Providers responsible for fewer claims were significantly less likely to meet every process indicator, with the exception of regularity of care. (Even for regularity of care, the direction of effect was as expected and lack of significance may have been a function of fewer observations because of including only claims with over 2 weeks of time loss.) This is consistent with prior research showing a relationship between volume and quality indicators.24 Based on the literature supporting an association between volume and outcomes,25 these findings suggest a need for further research into the relationship between volume, quality, and outcomes in occupational health care. It is worth noting that although NPs were responsible for fewer claims on average, their group performance on these indicators (relative to PCP performance) did not appear to be negatively impacted.
As anticipated, there was wide variation in the percentage of claims meeting each indicator (Table 5). The highest compliance was found for documentation of work-relatedness on the accident report. This was unsurprising (because a claim cannot be processed without this information), but even small rates of missing information can lead to delays that can negatively impact workers. Telephone communication at the first visit had the lowest percentages, which was also unsurprising because most simple claims would not require such communication. An injured worker being seen within 1 day of injury is a best practices goal, but perhaps is less likely to occur for uncomplicated back injuries where there may tend to be a delay between the date of injury and the date a worker contacts a health care provider. The two metrics that appear most in need of and perhaps most amenable to intervention are 1) timely filing of the accident report and 2) regularity of care while on time loss. Approximately half of claims did not have at least one of these two important indicators met, despite the regulatory requirement that the accident report be filed within 5 days and the high risk for extended time loss that work-disabled claimants face if their care is not closely monitored. Data from the COHE project suggest that providing incentives for meeting quality improvement indicators can substantially increase compliance and, in turn, can substantially improve disability outcomes.29,30
The usefulness of the indicators evaluated in this study may vary across jurisdictions, depending on local workers’ compensation regulations and data availability. There were also conceptual and measurement issues with each indicator that must be taken into account in further development. For example, time from injury to the first visit was considered as a possible quality indicator for the COHE project, but the ability to track when the worker first contacted the provider’s office was considered critical to assess individual provider performance.23 As another example, a higher prevalence of telephone communication billing codes on the first visit date could mean that the attending provider type was more likely to call the employer or others involved in the case to discuss or coordinate care, or was more likely to have cases requiring such calls, or simply was more likely to bill for such phone conversations.
Despite the inclusion of all eligible claims filed in the year after implementation of SHB 1691, we had a relatively small (though population-based) sample of claims filed by a relatively small number of NPs. For this initial analysis, all eligible PCP claims were retained to enhance power. As larger numbers of NP claims become available for further research, it may be preferable to choose a random or propensity score-matched sample of PCP claims for comparison purposes. This study had 80% power to detect differences of approximately 5 to 10 percentage points in the quality indicators. More work is needed to determine what a clinically significant difference in these indicators might be, and to validate their use with clinical data and in larger samples of claims.5
This observational study included statistical control for case mix and controlled for injury type and severity by restriction. Selection bias remained of some concern because of the potential for unmeasured differences in worker choice of attending provider. Nevertheless, there was little difference in case mix between NPs and PCPs in this sample, and direct standardization by age and sex had little impact on the findings. NP claims were less likely to have any time loss compensation than were PCP claims, but there were essentially no differences in other potentially costly claim characteristics across provider type.
The use of administrative data may have contributed error in several well-known ways.43,44 The exclusion of self-insured companies may have led to underestimates of claim volume. NPs certified as occupational health specialists were not identifiable as such, and there may be associated practice differences.11,16 The data do not allow for a full understanding of the reasons that process indicators were not met. In addition, NPs and physicians may tend to code or bill differently for the same care.
Despite the limitations inherent in relying on administrative data, there were important advantages, particularly the population-based nature of this study and the ability to efficiently link enrolled provider data with claim and injury, medical billing, and time loss data.45 Another strength of this study was the implementation of variants of quality indicators that had previously been vetted by expert panels for use in Washington State.7
We took advantage of the implementation of SHB 1691 to evaluate the use of several generic quality and process of care indicators that were available in Washington State workers’ compensation administrative data, including 1) whether the first visit occurred within 1 day of injury, 2) whether the accident report was filed within 7 days of the first visit, 3) whether there was early telephone communication about care coordination or treatment, 4) whether work-relatedness was specified on the accident report, and 5) regularity of care while on time loss. Using these indicators, we found no evidence of systematic differences in the care provided by attending NPs and PCPs to workers with uncomplicated back injuries. Nevertheless, these indicators were strongly associated with work-related disability and there was considerable room for improvement, suggesting that targeted incentives could substantially improve outcomes.
The authors thank Mary K. Salazar and Deborah Fulton-Kehoe for providing thoughtful feedback and inspiration over the course of this study, and Jerry Gluck for providing data management support. The authors also thank Cameron Craigie for providing extensive data documentation and Jamie Lifka for her policy insights.
Supported by the NIOSH ERC Occupational Health Services Research Training Program (T42 CCT010418, 1 T42 0H008433), and by the University of Washington Occupational Epidemiology and Health Outcomes Program.
1. Pransky G, Benjamin K, Dembe AE. Performance and quality measurement in occupational health services: current status and agenda for further research. Am J Ind Med. 2001;40:295–306.
2. Wickizer TM, Franklin G, Plaeger-Brockway R, Mootz RD. Improving the quality of workers’ compensation health care delivery: the Washington State Occupational Health Services Project. Milbank Q. 2001;79:5–33.
3. Dembe AE, Himmelstein JS, Stevens BA, Beachler MP. Improving workers’ compensation health care. Health Aff (Millwood). 1997;16:253–257.
4. Johantgen M, Trinkoff A, Gray-Siracusa K, Muntaner C, Nielsen K. Using state administrative data to study nonfatal worker injuries: challenges and opportunities. J Safety Res. 2004;35:309–315.
5. Wray NP, Ashton CM, Kuykendall DH, Hollingsworth JC. Using administrative databases to evaluate the quality of medical care: a conceptual framework. Soc Sci Med. 1995;40:1707–1715.
6. Fulton-Kehoe D, Gluck J, Wu R, Mootz R, Wickizer T, Franklin G. Measuring work disability: what can administrative data tell us about patient outcomes? J Occup Environ Med. 2007;49:651–658.
7. Wickizer TM, Franklin GM, Mootz RD, et al. A community-wide intervention to improve outcomes and reduce disability among injured workers in Washington State. Milbank Q. 2004;82:547–567.
8. US Congress Office of Technology Assessment. Nurse Practitioners, Physician Assistants, and Certified Nurse-Midwives: A Policy Analysis. Washington, DC: US Government Printing Office; 1986.
9. Horrocks S, Anderson E, Salisbury C. Systematic review of whether nurse practitioners working in primary care can provide equivalent care to doctors. BMJ. 2002;324:819–823.
10. Mundinger MO, Kane RL, Lenz ER, et al. Primary care outcomes in patients treated by nurse practitioners or physicians: a randomized trial. JAMA. 2000;283:59–68.
11. Dowrick MA, Rezents K. Occupational health nurse practitioner has broader legal scope, accessibility. OHNPs receive high ratings for the cost-effective delivery of quality healthcare services. Occup Health Saf. 1993;62:131–135.
12. Ferguson LA. Enhancing health care to underserved populations: a model for the occupational nurse practitioner. AAOHN J. 1996;44:332–336.
13. McGrath S. The cost-effectiveness of nurse practitioners. Nurse Pract. 1990;15:40–42.
14. Pearson L. Sixteenth annual legislative update: how each state stands on legislative issues affecting advanced nursing practice. Nurse Pract. 2004;29:26–31.
15. Zentner J, Dellinger C, Adkins WE, Greene J. Nurse practitioner provided primary care: managing health care costs in the workplace. AAOHN J. 1995;43:52–53.
16. Burgel B. Certified occupational health nurse specialists practicing as nurse practitioners. Administrative practices in the California workers’ compensation system. AAOHN J. 2002;50:262–270.
17. Johantgen M, Elixhauser A, Bali JK, Goldfarb M, Harris DR. Quality indicators using hospital discharge data: state and national applications. Jt Comm J Qual Improv. 1998;24:88–105.
18. Toscano G, Windau J, Drudi D. Using the BLS Occupational Injury and Illness Classification System as a safety and health management tool. Compensat Work Cond. 1996;1:19–28.
19. Koepsell TD, Weiss NS. Epidemiologic Methods: Studying the Occurrence of Illness. Oxford, NY: Oxford University Press; 2003.
20. Bailit J, Garrett J. Comparison of risk-adjustment methodologies for cesarean delivery rates. Obstet Gynecol. 2003;102:45–51.
21. Pease SR. Workers’ Compensation in Washington: Administrative Inventory. WC-89-3. Cambridge, MA: Workers Compensation Research Institute; 1989.
22. Wickizer TM, Franklin G, Turner J, Fulton-Kehoe D, Mootz R, Smith-Weller T. Use of attorneys and appeal filing in the Washington State workers’ compensation program: does patient satisfaction matter? J Occup Environ Med. 2004;46:331–339.
23. Franklin G. Report on Recommended Quality Indicators (Prepared for: Occupational Health Services Project, Washington State Department of Labor and Industries). Seattle: University of Washington Occupational Epidemiology and Health Outcomes Program; 2001.
24. Druss BG, Miller CL, Pincus HA, Shih S. The volume-quality relationship of mental health care: does practice make perfect? Am J Psychiatry. 2004;161:2282–2286.
25. Halm EA, Lee C, Chassin MR. Is volume related to outcome in health care? A systematic review and methodologic critique of the literature. Ann Intern Med. 2002;137:511–520.
26. Shaw WS, Pransky G, Fitzgerald TE. Early prognosis for low back disability: intervention strategies for health care providers. Disabil Rehabil. 2001;23:815–828.
27. Stover B, Wickizer TM, Zimmerman F, Fulton-Kehoe D, Franklin G. Prognostic factors of long-term disability in a workers’ compensation system. J Occup Environ Med. 2007;49:31–40.
28. Washington State Department of Labor and Industries. Attending Doctor’s Return-to-Work Desk Reference. Olympia, WA: Department of Labor and Industries; 2004. Publication F200–002-000.
29. Wickizer T. Report on the Outcome Evaluation for the Eastern Washington COHE (3 Counties). Seattle: University of Washington Occupational Epidemiology and Health Outcomes Program; 2006.
30. Wickizer T. Report on the Outcome Evaluation for the Western Washington COHE. Seattle: University of Washington Occupational Epidemiology and Health Outcomes Program; 2005.
31. Krause N, Frank JW, Dasinger LK, Sullivan TJ, Sinclair SJ. Determinants of duration of disability and return-to-work after work-related injury and illness: challenges for future research. Am J Ind Med. 2001;40:464–484.
32. Pransky G, Shaw W, Franche RL, Clarke A. Disability prevention and communication among workers, physicians, employers, and insurers—current models and opportunities for improvement. Disabil Rehabil. 2004;26:625–634.
33. Johnson S, Prosser D, Bindman J, Szmukler G. Continuity of care for the severely mentally ill: concepts and measures. Soc Psychiatry Psychiatr Epidemiol. 1997;32:137–142.
34. Cabana MD, Jee SH. Does continuity of care improve patient outcomes? J Fam Pract. 2004;53:974–980.
35. Haggerty JL, Reid RJ, Freeman GK, Starfield BH, Adair CE, McKendry R. Continuity of care: a multidisciplinary review. BMJ. 2003;327:1219–1221.
36. van der Weide WE, Verbeek JH, van Dijk FJ. Relation between indicators for quality of occupational rehabilitation of employees with low back pain. Occup Environ Med. 1999;56:488–493.
37. Greenberg GA, Rosenheck RA. Continuity of care and clinical outcomes in a national health system. Psychiatr Serv. 2005;56:427–433.
38. Sears JM, Wickizer TM, Franklin GM, Cheadle AD, Berkowitz B. Nurse practitioners as attending providers for injured workers: evaluating the effect of role expansion on disability and costs. Med Care. (In press).
39. Sears JM, Wickizer TM, Franklin GM, Cheadle AD, Berkowitz B. Expanding the role of nurse practitioners: effects on rural access to care for injured workers. J Rural Health. (In press).
40. Brook RH, McGlynn EA, Cleary PD. Quality of health care, part 2: measuring quality of care. N Engl J Med. 1996;335:966–970.
41. Kleinpell R, Gawlinski A. Assessing outcomes in advanced practice nursing practice: the use of quality indicators and evidence-based practice. AACN Clin Issues. 2005;16:43–57.
42. Nugent KE, Lambert VA. Evaluating the performance of the APN. Nurse Pract. 1997;22:190, 193–194, 197–198.
43. Connell FA, Diehr P, Hart LG. The use of large data bases in health care studies. Annu Rev Public Health. 1987;8:51–74.
44. Lohr KN. Use of insurance claims data in measuring quality of care. Int J Technol Assess Health Care. 1990;6:263–271.
45. Franklin GM, Fulton-Kehoe D. Outcomes research in Washington state workers’ compensation. Am J Ind Med. 1996;29:642–648.