USUAL SOURCE OF CARE FOR A

DULTS WITH AND WITHOUT

BACK PAIN: MEDICAL EXPENDITURES PANEL SURVEY DATA

POOLED FOR YEARS 2000 TO 2006Monica Smith, DC, PhD

Adjunct AssSciences, Lomba

Researchwas cChiropractic Rese

Submit requesRivermark Plz #3(e-mail: msmith@

Paper submittaccepted May 7,

0161-4754/$3Copyright ©doi:10.1016/j.

356

ABSTRACT

Objectives: The purpose of this study was to explore the extent to which individuals with back pain or other healthconditions and individuals with no health problems report having a usual source of care (USC) for their healthcare needs.Methods: This study evaluated longitudinal Medical Expenditures Panel Survey data (data pooled for surveycalendar years 2000-2006). Comparisons were made between adult Medical Expenditures Panel Survey respondentsidentified as having a back pain condition (n = 10 194) compared with those without back pain but with other healthcondition (n = 45 541) and those with no back pain and no other condition (n = 5497).Results: Compared with individuals with no health problems, those with back pain were almost 8 times morelikely (odds ratio, 7.8; P b .001) to report having a USC, and those with other health problems besides back pain were5 times more likely (odds ratio, 5.4; P b .001). For those with a USC, individuals with back pain and those withother problems but not back pain were both approximately one-and-a-half times more likely than those without anyhealth problems to report a specific provider type as their USC (P b .001).Conclusion: Study findings suggest that relatively healthy adults without back pain are less likely to have a USC thanthose with back pain or other health problems. (J Manipulative Physiol Ther 2011;34:356-361)

Key Indexing Terms: Back Pain; Spine; Chiropractic

Per Congressional mandate, the US Agency forHealthcare Research and Quality (AHRQ) producesreports to the nation about the quality of health care

and access to health care.1 The AHRQ National HealthcareQuality Report and National Healthcare Disparities Reportdocument national trends in the effectiveness of care,patient safety, timeliness of care, patient centeredness,and efficiency of care.

The AHRQ National Healthcare Disparities Reportmeasures trends in access to care among priority popula-tions such as residents of rural areas; women; children;older adults; individuals with disabilities or special healthcare needs; and racial, ethnic, or income groups.2-4 Toward

ociate Professor, National University Healthrd IL.onducted asAssociate Professor, Palmer Center forarch, Palmer College of Chiropractic, San Jose, CAts for reprints to: Monica Smith, DC, PhD, 396444, Santa Clara, CA 95054nuhs.edu).ed April 26, 2011; in revised form May 6, 20112011.6.002011 by National University of Health Sciencesjmpt.2011.05.007

;

.

development of the current and future reports on health caredisparities, policy and research working groups have vettedvarious data sources and methods for measuring access tocare.2 The AHRQ-sponsored Medical Expenditures PanelSurvey (MEPS) data are a key source for data on access tocare, and other sources include data from Centers forDisease Control and Prevention, Centers for Medicare andMedicaid Services, Indian Health Service, National Com-mittee for Quality Assurance, National Institutes of Health,and the US Census Bureau.3 As synthesized by the multipleAHRQ working groups, an individual's access to care maybe measured along several dimensions such as their abilityto gain initial entry into the health care system (eg, healthinsurance coverage, having a usual source of care [USC],or patient perceptions of need), their ability to get neededcare within the system (eg, wait times or obtainingreferrals), patient perceptions of their care (eg, patient-provider communication and relationship, cultural compe-tency, health information), and their pattern of usingvarious services (eg, general and specialty care, ambulatorycare, emergency and urgent care, or avoidable admissionsto hospital and inpatient care).4

Access to care as a function of an individual's having aUSC has been described using data from the AHRQ-spon-sored MEPS,5-9 National Health Interview Survey,10,11

357SmithJournal of Manipulative and Physiological TherapeuticsUsual Source of Care Back PainVolume 34, Number 6

National Health and Nutrition Examination Survey,12

Community Tracking Study,13 regional or local areasurveys,14 or targeted surveys (eg, using a consumermailing list of the National Multiple Sclerosis Society).15

This study analyzes data from the MEPS to explore theextent to which individuals with back pain or other healthconditions and individuals with no health problems reporthaving a USC for their health care needs.

METHODS

Adult MEPS respondents (N = 61 323) were identified inthe 2-year longitudinal MEPS panels 5 through 10 (MEPSsurvey calendar years 2000-2006) as reported in a previouslypublished study.16 Medical Expenditures Panel Survey re-spondents with back pain were defined based on 2 conditioncoding classification schemes that are available in the MEPSmedical conditions datafile,17 the Clinical ClassificationCategory18 codes, and the International Classification ofDiseases codes (encompassing 66 International Classificationof Diseases codes of acute, chronic, or recurrent episodic backpain conditions such as spondylosis and intervertebral discdisorders, sacroiliac sprain/strain, other back sprain/strain).Each of the 61 232 adultMEPS respondents were assigned to 1of 3 “condition groups” for this study: (a) individuals identifiedas having a back pain condition (n = 10 194), (b) individualswho did not report back pain but did report having some otherhealth condition (n = 45 541), and (c) individuals who reportedno health condition (no back pain and no other condition)during the 2-year MEPS longitudinal panel survey (n = 5497).

Medical Expenditures Panel Survey participants aresurveyed on whether they have a USC and, if so, the typeof health care provider, their satisfaction with the USCprovider's care, and any problems with accessing the USCprovider. During the MEPS field interviews, USC is definedfor participants as “…a particular doctor's office, clinic, orother place that the individual usually goes to if they are sick orneed advice about their health…” followed by a set of MEPSinterview questions about that USC provider: “Is provider amedical doctor?” “Is provider a nurse, nurse practitioner,physician's assistant, midwife, or some other kind of person?”or “What is provider's specialty?”

For this study, USC is operationalized using the 3 MEPSsurvey variables: “Does person have a USC?” If yes, “Doesperson have a specific USC provider type?” If yes, “Whattype of provider?” Of the 51 842 MEPS respondents whoreported having a USC, 56% (n = 29 134) listed a specificUSC provider type (Fig 1).

Two hypotheses were tested in this study (Fig 1),comparing the 3 condition groups of (a) individuals withback pain, (b) individuals without back pain but with someother health condition, and (c) individuals with nocondition. The null hypotheses were no significantdifferences between the 3 groups.

The first hypothesis test compared the 3 conditiongroups as to their reporting that they have a USC,operationalized as an affirmative response code for theMEPS variable “Does person have a USC”.

The second hypothesis test examined only thoserespondents who did report having a USC and comparedthe 3 groups as to whether they report a specific “providertype” as their USC.

The 2 hypotheses were tested using bivariate χ2 tests andbinomial logistic regression modeling. Medical Expen-ditures Panel Survey uses a complex sample design andoversamples certain population groups of interest; therefore,longitudinal sampling weights and longitudinal adjustedvariance estimators (strata, primary sampling unit [PSU])were applied during the inferential hypothesis testing toaccount for the MEPS sampling frame and complex surveydesign features. All data management and statistical analyses(unweighted and weighted, descriptive and inferential) wereperformed using SPSS for Windows versions 17.0 and 12.0(SPSS Inc, Chicago, IL).

RESULTS

The 3 condition groups differed significantly on having aUSC (Fig 1). Compared with individuals with no healthproblems, those with back pain were almost 8 times morelikely (odds ratio [OR], 7.8; P b .001) to report having a USC,and those with other health problems besides back pain were 5times more likely (OR, 5.4; P b .001). Comparing only the 2groups with health problems, those with back pain weresignificantly more likely to have a USC than those with otherhealth problems but not back pain (OR, 1.5; P b .001).

For those who reported having a USC (n = 51 842),individuals with back pain and those with other problemsbut not back pain were both approximately one-and-a-halftimes more likely than those without any health problems toreport a specific provider type as their USC (P b .001).

The specific categories and categorical assignment ofprovider types has evolved over successive administrationsof the MEPS survey (Appendix A), necessitating a panel-by-panel approach to describing the specific provider typesthat are identified by MEPS respondents as their USC. TheUSC provider types from MEPS panels 10, 9, and 8 arepresented in Table 1 for each of the 3 condition groups.

DISCUSSION

These findings suggest that relatively healthy adults, thatis, those who report no active health problems during thecourse of the MEPS 2-year longitudinal data collectiontimeframe, are also less likely to have a USC. This isconsistent with other studies that have found that manyadults may not have a USC simply because they choose notto, most probably because they do not perceive the needsince they are relatively healthy.5,19,20 Interestingly, some

Fig 1. Two-stage hypothesis testing of USC for back pain: sample sizes, hypothesis tests, and results.

358 Journal of Manipulative and Physiological TherapeuticsSmithJuly/August 2011Usual Source of Care Back Pain

individuals with no USC or who use the emergencydepartment as their usual source when care is needed mayhave lower overall costs than those who use medical doctor(MD) generalists or MD specialists as their USC.9

The finding that back pain sufferers are significantly andsubstantially more likely to have a USC is also understand-able, given that the nature of back pain may characteris-tically be a recurrent, episodic, or chronic condition16 forwhich individuals may use the services of spine carespecialists on an ongoing basis. Paradoxically, the recurrentepisodic nature of back pain may also partly explain thefinding that chiropractors are identified as a USC for adultMEPS respondents without back pain in that these may bepersons who use ongoing care to manage a sporadicrecurring back problem, although they may not haveexperienced an active flare-up episode of their back painduring their 2-year MEPS data collection window. Mindenet al15 found that individuals with multiple sclerosis (MS)who had experienced a past recent flare-up (1 or 2 relapsesin the preceding year) were more likely to identify aneurologist as their usual source of MS care in the currentyear as compared with MS sufferers with no recent relapses.Another possible explanation is that some individuals maybe using chiropractors as their USC for conditions otherthan back pain or for other reasons altogether such asgeneral wellness care.

Abroad representation of provider typesmaybe identified as aUSC for individuals with back pain (Table 1). Everett et al14

found that useofphysicianassistants (PAs) andnurse practitioners(NPs) as a USC was associated with chiropractor visits, whichthey reasoned may reflect a greater willingness of PA/NPs torecommend nondrug or complementary and alternative medicine(CAM) therapies for their patients. As suggested by this study,USC varies for individuals with back pain (Table 1) and mayinclude the use of PAs, NPs, chiropractors, or a broad range ofother provider types as their USC.

Identification of optimal arrangements for USC toensure that patients receive both specialized and primarycare services as appropriate to their needs may vary bypatient preference and may also vary by geographiclocation. For instance, because the dispersion of neurol-ogists is not uniform across the United States, then theappropriateness (or patient preference) for a “primarycare” vs “principal care model” may vary as a functionof geographic access to specialized neurological servicesfor persons with neurological disorders.15 In a primarycare model, generalists provide all primary care includ-ing referral to specialist. Within a principal care model,some primary care services may be provided byspecialists, such as neurologists, who also serve as anindividual's USC. Variants on the principal care modeltheme includes the provision of primary care services by

Table 1. Provider types specified by adult MEPS respondents as their USC for the 3 groups: individuals with back pain, individualsreporting other condition but not back pain, and individuals reporting no condition during the MEPS 2-year longitudinal data collectiontime frame (MEPS panels 10, 9, and 8)

Back pain Other condition not back pain No condition

MEPS panel 10 (n = 856) (n = 3341) (n = 184)MD-GP/FP/internist MD-GP/FP/Internist MD-GP/FP/internistMD-OB/Gyn MD-OB/Gyn MD-OB/GynMD-cardiologist MD-cardiologistDO-osteopath DO-osteopathMD-geriatrician MD-geriatricianMD (other specialist) MD (other specialist) MD (other specialist)Nurse/NP Nurse/NP Nurse/NPPA PADC-chiropractor DC-chiropractor

MEPS panel 9 (n = 951) (n = 3415) (n = 203)MD-GP/FP/internist MD-GP/FP/internist MD-GP/FP/internistMD-OB/Gyn MD-OB/Gyn MD-OB/GynMD-cardiologist MD-cardiologistDO-osteopath DO-osteopath DO-osteopathMD-geriatrician MD-geriatricianMD (other specialist) MD (other specialist) MD (other specialist)Nurse/NP Nurse/NPPA PA PADC-chiropractor DC-chiropractor

MEPS panel 8 (n = 907) (n = 3705) (n = 227)MD-GP/FP/internist MD-GP/FP/internist MD-GP/FP/internistMD-OB/Gyn MD-OB/Gyn MD-OB/GynMD-cardiologist MD-cardiologist MD-cardiologistDO-osteopath DO-osteopath

MD-geriatricianMD (other specialist) MD (other specialist) MD (other specialist)Nurse/NP Nurse/NP Nurse/NPPA PADC-chiropractor DC-chiropractor

Note: For cogency of data presentation, the following MD specialist types are infrequently identified as USC provider type and are grouped within “MD(other specialist)” category in this table: neurologist, endocrinologist, gastroenterologist, nephrologist, oncologist, pulmonologist, rheumatologistpsychiatrist, and surgery. OB/Gyn indicates obstetrics and gynecology.

359SmithJournal of Manipulative and Physiological TherapeuticsUsual Source of Care Back PainVolume 34, Number 6

multidisciplinary teams that may be colocated (or coordi-nated if dispersed across the community) with such teamstaking a patient-centered and family-centered approach andemphasizing wellness, maximal functioning, and ongoingintegrated care.15

The distribution of generalist MD provider types alsovaries by geography and by location in Medicare PrimaryCare Service Areas. The geographic distribution ofgeneralist MDs trained in family practice (FP) andgeneral practice (GP) more closely mirrors the distribu-tion of the general population, with FPs/GPs tending tobe located in areas of lowest Medicare reimbursementsper beneficiary, whereas generalist MDs trained asinternists tend to cluster toward areas with highestMedicare reimbursements (urban and northeast).9 Thegeographic distribution and use of chiropractors varies,21

and the use of outpatient physical therapy services formusculoskeletal conditions may also vary as a function ofgeographic access, with individuals in urban andnortheast areas being more likely to use physical therapythan those in other areas.22

,

Limitations and Future StudiesA serious limitation to usingMEPS data is that the structure

and format of the MEPS only allows respondents to identify asingle location or single provider type as their USC. This“single USC provider” logic of the MEPS would presume thatall individuals surveyed byMEPS can or should fit neatly into aprimary care model rather than a principal care model for theiruse of health services, despite that a principal care model mayactually be preferred bymany individuals.15 It is quite possibleand perhaps likely that MEPS respondents would identifymore than 1 provider type as their USC if allowed to do so (ie,reflecting a care-seeking behavior more along the lines of aprincipal care rather than primary care model).

A further limitation or caution to using MEPS data is thatcertain provider types, such as chiropractors, may besystematically underreported as USC in the MEPS. Forinstance, the MEPS survey queries have changed over time;therefore, certain provider types cannot even be discerned inearlier iterations ofMEPS data (chiropractors were not listed asa provider type for the USC queries before MEPS survey year2002; Appendix A). Less apparent, although just as important,

360 Journal of Manipulative and Physiological TherapeuticsSmithJuly/August 2011Usual Source of Care Back Pain

is that the MEPS methodology may inadvertently introducerecall bias as a function of the interviewer protocol prompts,again, resulting in systematic underreporting of certainprovider types. Medical Expenditures Panel Survey constructsa composite variable for “usual source provider type” from theset of MEPS interview questions “Is provider a medicaldoctor?” “Is provider a nurse, nurse practitioner, physician'sassistant, midwife, or some other kind of person?” “What isprovider's specialty?” Provider types that are specificallyprompted byMEPS interviewers (eg, MDs, PAs, NPs) may berecalled more readily by MEPS survey participants as theirUSCmore so than other provider types that are not specificallyelicited by the MEPS interviewer protocol (eg, doctors ofchiropractic and osteopaths are not specifically prompted as aresponse to the USC queries).

Much more research is needed to better understand theoutcomes and costs associated with various USC usebehaviors including the possibility that shorter term savingsmay be more than offset by longer term costs and conse-quences associated with individual choices for their usualsources of care. Further research needs also to be directedtoward better clarifying how individuals do or shoulduse the emergency department as a community-based USCfor their urgent vs nonurgent health care needs.6,10

CONCLUSION

The findings from this study suggest that adults withback pain are more likely to report having a USC than thosewith other health problems but not back pain. Relativelyhealthy adults are least likely to have a USC than those withback pain or other health problems. More research is neededto identify optimal arrangements for USC to ensure thatpatients receive both specialized and primary care servicesas appropriate to their needs.

Practical Applications• Back pain sufferers are more likely to have a USCthan those without back pain.

• Given that the nature of back pain may character-istically be a recurrent, episodic, or chroniccondition, individuals may use the services ofspine care specialists on an ongoing basis.

• Health care providers identified as USC shouldcoordinate the delivery of optimal health care toensure that patients receive both specialized andprimary care services as appropriate to their needs.

FUNDING SOURCES AND POTENTIAL CONFLICTS OF INTEREST

This project was supported by Grant number K01AT002391, a grant from National Institutes for HealthNational Center for Complementary and Alternative Medi-cine. No conflicts of interest were reported for this study.

REFERENCES

1. Agency for Healthcare Research and Quality (AHRQ, Rock-ville, MD), 2009 National Healthcare Quality Report (NHQR),and 2009 National Healthcare Disparities Report (NHDR).Available from: http://www.ahrq.gov/qual/qrdr09.htm.Accessed July 15, 2010.

2. Agency for Healthcare Research and Quality (AHRQ, Rockville,MD), National Healthcare Disparities Report: Background on theMeasures Development Process. Available from: http://www.ahrq.gov/qual/nhdr02/nhdrprelim.htm. Accessed July 15, 2010.

3. Agency for Healthcare Research and Quality (AHRQ,Rockville, MD), 2009 National Healthcare Quality andDisparities Reports: data sources appendix. Available from:http://www.ahrq.gov/qual/qrdr09/datasources/. Accessed July15, 2010.

4. Agency for Healthcare Research and Quality (AHRQ,Rockville, MD), National Healthcare Disparities Report:list of measures. Available from: http://www.ahrq.gov/qual/nhdrmeasures/listmeasure.htm. Accessed July 15, 2010.

5. Viera AJ, PathmanDE, Garrett JM. Adults' lack of a usual sourceof care: a matter of preference? Ann Fam Med 2006;4:359-65.

6. Sarver JH, Cydulka RK, Baker DW. Usual source of care andnonurgent emergency department use. Acad Emerg Med 2002;9:916-23.

7. DeVoe JE, Fryer GE, Phillips R, Green L. Receipt ofpreventive care among adults: insurance status and usualsource of care. Am J Public Health 2003;93:786-91.

8. DeVoe JE, Wallace LS, Pandhi N, Solotaroff R, Fryer GE.Comprehending care in a medical home: a usual source ofcare and patient perceptions about healthcare communication.J Am Board Fam Med 2008;21:441-50.

9. Phillips RL, Dodoo MS, Green LA, Fryer GE, Bazemore AW,McCoy KI, Petterson SM. Usual source of care: an importantsource of variation in health care spending. Health Affairs 2009;28:567-77.

10. Walls CA, Rhodes KV, Kennedy JJ. The emergencydepartment as a usual source of medical care: estimatesfrom the 1998 National Health Interviews Survey. AcadEmerg Med 2002;9:1140-5.

11. Blewett LA, Johnson PJ, Lee B, Scal PB. When a usual sourceof care and usual provider matter: adult prevention andscreening services. J Gen Intern Med 2008;23:1354-60.

12. Winters P, Tancredit D, Fiscella K. The role of usual sourceof care in cholesterol treatment. J Am Board Fam Med 2010;23:179-85.

13. Smith MA, Bartell JM. Changes in usual source of care andperceptions of health care access, quality, and use. MedicalCare 2004;42:975-84.

14. Everett CM, Schumacher JR, Wright A, Smith MA. Physicianassistants and nurse practitioners as a usual source of care.J Rural Health 2009;25:407-14.

15. Minden SL, Hoaglin DC, Hadden L, Frankel D, Robbins T,Perloff J. Access to and utilization of neurologists by peoplewith multiple sclerosis. Neurology 2008;70:1141-9.

16. Smith M. Identifying episodes of back pain using MedicalExpenditures Panel Survey (MEPS) data: patient experience, useof services, and chronicity. J Manipulative Physiol Ther 2010;33:562-75.

17. Agency for Healthcare Research and Quality (AHRQ),Medical Expenditures Panel Survey (MEPS), access page toonline workbook. Available from: http://www.meps.ahrq.gov/mepsweb/about_meps/online_workbook.jsp.

18. Elixhauser A, Steiner CA,Whittington CA,McCarthy E. Clinicalclassifications for health policy research: hospital inpatient

361SmithJournal of Manipulative and Physiological TherapeuticsUsual Source of Care Back PainVolume 34, Number 6

statistics, 1995. Healthcare cost and utilization project, HCUP-3research note. (AHCPR Pub. No. 98-0049). Rockville, MD:Agency for Healthcare Research and Quality; 1998.

19. Hayward RA, Bernard AM, Freeman HE, Corey CR. Regularsource of ambulatory care and access to health services. Am JPublic Health 1991;80:434-8.

20. Williams CH. From coverage to care: exploring links betweenhealth insurance, a usual source of care and access. TheSynthesis Project, Issue 1. Robert Wood Johnson Foundation;2002. Available from: http://www.rwjf.org/pr/product.jsp?id=20113. Accessed June 16, 2011.

21. Smith M, Carber L. Chiropractic health care in HealthProfessional Shortage Areas (HPSAs) of the U.S. Am J PublicHealth 2002;92:2001-9.

22. Carter SK, Rizzo JA. Use of outpatient physical therapyservices by people with musculoskeletal conditions. PhysTher 2007;87:497-512.

23. McCann J, Phillips RL, Green LA, Fryer GE. Chiropractorsare not a usual source of primary health care. AmFam Physician2004;69:2544. Available from: http://www.graham-center.org/online/graham/home/publications/onepagers/2004/op28-chiro-practors-source.html. Accessed April 6, 2011.

APPENDIX A. METHODOLOGICAL CONSIDERATIONS FOR

RESEARCHING USC USING MEPS DATA

Methods Note number 1The list and categorization of explicit USC “provider

types” has changed somewhat over successive years ofadministering the MEPS survey (Appendix Table 1).For example, before the year 2003 (the second year ofMEPS panel 7), cardiologists and osteopaths identifiedby MEPS respondents as their USC would have beencoded under the generic USC provider type category“MD = Other”. Based on the high frequency of MEPSrespondents who explicitly identified a cardiologist orosteopath as their USC in the 2002 survey, MEPSadded the 2 provider types “cardiologist” and “osteo-path” as additional specific categories to the list ofUSC provider types for MEPS survey data 2003onward. The availability of MEPS data on specificUSC provider types will vary panel by panel, reflectingthe addition of several specific categories of providertypes that have been added to MEPS data sets overtime. Chiropractors first appear as an explicit USCprovider type in MEPS 2002 survey data. Therefore,it is not surprising that McCann et al23 found that“…none of the 25 096 participants in the 2000 MEPSidentified a chiropractor as their usual source of care…”since chiropractors were not even added to the roster ofUSC provider types until the MEPS 2002 survey datarelease year.

Methods Note number 2The USC queries are posed to MEPS longitudinal

panel participants twice during the 2-year MEPSlongitudinal survey: during round 2 (in year 1) andagain during round 4 (in year 2). A small number ofMEPS respondents did identify 1 type of USC providerduring round 2 and a different type of USC providerduring round 4. Of the 51 842 MEPS respondents whoreported having a USC in MEPS panels 5 through 10,56% (n = 29 134) listed a specific USC provider type,and of those 29 134, only 3% (n = 787) identified adifferent USC provider type in round 2 than the USCprovider type they identified in round 4.

Appendix Table 1. Availability of useful data (nonzero/nonmissing) for USC provider type categories in the MEPSaccess to care data set for MEPS longitudinal panels 5through 10

Useful data (nonzero/nonmissing) are available on these

provider type categories for all MEPS panels 5 through 10(all MEPS survey years):

MD = General/family practiceMD = Internal medicineMD = PediatricsMD = obstetrics and gynecologyMD = SurgeryMD = OtherNursePhysician's assistantOther non-MD providerUnknown

Useful data (nonzero/nonmissing) are now also available forseveral explicit USC provider type categories that wereadded to MEPS during successive panels, listed below as tothe first MEPS panel/year that the useful data appear inMEPS data sets (and are available for the panels/yearsthereafter).

First appears in the second year of MEPS panel 5 (2001)Nurse PractitionerFirst appears in the second year of panel 6 (2002)ChiropractorFirst appears in the second year of panel 7 (2003)

MD = cardiologistDoctor of osteopathyFirst appears in the second year of panel 8 (2004)

MD = endocrinologistMD = gastroenterologistMD = geriatricianMD = nephrologistMD = oncologistMD = pulmonologistMD = rheumatologist

Psychiatrist/psychologistFirst appears in the second year of panel 9 (2005)

MD = neurologistFirst appears in the second year of panel 10 (2006)Alternative care provider