This study describes the frequency, predictors, and expenditures for the use of complementary and alternative medicine (CAM) in an insured pediatric population.
Washington state requires CAM-licensed medical professional coverage in private health insurance. We performed a cross-sectional analysis of services provided to children in 2002 by conventional professionals, chiropractors, naturopathic physicians, acupuncturists, and massage therapists. Both χ2 tests and logistic regression analysis were used to identify statistically significant differences in use and explanatory factors.
Of 187 323 children covered by 2 large insurance companies, 156 689 (83.6%) had any claims during the year. For those with claims, 6.2% of children used an alternative professional during the year, accounting for 1.3% of total expenditures and 3.6% of expenditures for all outpatient professionals. We found that CAM use was significantly less likely for males (odds ratio, 0.91; 95% confidence interval, 0.87-0.95) and more likely for children with cancer, children with low back pain, and children with adult family members who use CAM. Visits to chiropractors or massage therapists nearly always yielded diagnoses of musculoskeletal conditions. In contrast, diagnoses from naturopathic physicians and acupuncturists more closely resembled those of conventional professionals.
Insured pediatric patients used CAM professional services, but this use was a small part of total insurance expenditures. We found that CAM use was more common among some children, depending on their sex, age, medical conditions, and whether they had an adult family member who used CAM. Although use of chiropractic and massage was almost always for musculoskeletal complaints, acupuncture and naturopathic medicine filled a broader role.
Little is known about the role professionals in complementary and alternative medicine (CAM) play in pediatric primary care. In contrast, widespread CAM use by adults has been reported; 42% of US adults used CAM in some form during 1997, and the rate of use over time has increased.1,2 Because adults with serious medical problems, such as cancer,3 human immunodeficiency virus,4 diabetes,5 back pain,6 arthritis,7 and other musculoskeletal conditions,8 use CAM, one hypothesis purports that high use of health care in general9 drives CAM use. If this is the case, CAM use by children may well be low. On the other hand, if CAM professionals offer natural alternatives to conventional primary care, children may use more CAM than expected.
Surveys have described differing rates of pediatric CAM use, in part due to differing definitions of CAM10 and varied study populations.11- 13 Using the 1996 Medical Expenditure Panel Survey data, Davis and Darden14 reported that the prevalence of pediatric CAM use, including chiropractic, was only 1.8%. In contrast, Neuhouser et al15 found that among a sample of children with cancer in Washington state, 73% used at least 1 form of alternative treatment or therapy and that 21.3% used the services of alternative professionals, including chiropractors, acupuncturists, massage therapists, naturopathic physicians (NDs), and nonlicensed alternative professionals. Surveys of pediatric health services use are a challenge to conduct. All adults involved in the care of a child need to participate in the survey to ensure complete information. Large sample sizes are required when the prevalence of studied events is low. Finally, surveys of past activities are subject to recall bias, especially if use is high.16,17
Because of these issues, we used claims data from 2 large Washington state underwriters to measure CAM use by our pediatric population. In 1996, the Washington state legislature required all commercial health insurance policies to include coverage for every category of licensed professional who treated medical conditions usually covered by insurance.18,19 As a result, acupuncturists, massage therapists, and NDs were covered by most benefit packages by 2000. Also in 2000 the legislature required the inclusion of self-referral for chiropractic.20 These regulations make CAM benefits in Washington state the most robust in the United States and these insurance claims a uniquely rich source of information on the use of alternative services. This article evaluates claims for the services of alternative professionals among families with children in Washington state for the calendar year 2002. The purpose of this analysis is to examine the predictors of, frequency of, and expenditures for CAM professional use in a privately insured, pediatric population.
Before data collection began, this study was approved by the institutional review board of the University of Washington, Seattle. Calendar year 2002 claims data came from 2 large insurers that provide group and individual health insurance in Washington state. Data were included for all individuals enrolled in health insurance plans covered by Washington’s Every Category of Provider requirement. This excludes Medicare, Medicaid, state supplemental programs, federal employees insurance, and self-insured plans that are exempt from state regulation under the Employee Retirement Income Security Act.
We conducted a cross-sectional analysis of professional claims from families with children ages birth through 17 years. Enrollees were selected who had continuous enrollment in one company during 2002. We included all children and the covered adults associated with their policy. An enrollee was defined as anyone covered by the insurance company or product, whereas claimant denotes an enrollee who had an allowed claim for service during the calendar year.
A relational database was constructed from enrollment, claims, and pharmacy files received from each insurance company. Data included unique enrollee identification codes, birth year, sex, residence ZIP code, product type, contract number, Employee Retirement Income Security Act status, month of active enrollment, Aclaim number, service date, service location, International Classification of Diseases, Ninth Revision (ICD-9)21 codes, the Current Procedural Terminology22 codes, and Healthcare Common Procedure Coding System23 codes. We also received line item pecuniary data and professional type (including specific codes for acupuncture, naturopathic medicine, massage, and chiropractic professionals). Plan and contract information maintained in the final enrollment file reflected enrollees’ plan and contract in December 2002. Pharmacy files were also supplied. Unlike the line item details received in the claims files, the pharmacy files had variables that represented the number of prescriptions filled and aggregated annual expenditures for each enrollee’s prescription drugs.
From our enrollment data, we generated variables for age and county of residence. Age was calculated as the difference between birth year and study year. County of residence was based on resident ZIP code. Age was categorized as infants (0-1 year), preschoolers (2-5 years), children (6-12 years), and adolescents (13-17 years). Using the ICD-9 codes for each visit, we constructed indices of the types of diseases or disorders present and the expected resource use for each child. All of the indices use the John Hopkins Adjusted Clinical Group Case-Mix System software, version 5.24 The indices are (1) expanded diagnosis clusters (EDCs), which categorize ICD-9 codes into 26 major disease categories for each individual; (2) adjusted clinical groups (ACGs), which code each individual into a single category based on ICD-9 diagnoses along with age and sex; and (3) resource utilization bands (RUBs), which measure an individual’s expected resource use and are created by ranking ACGs in increasing order of resource use and grouping together ACGs with similar levels of expected resource use. Lower expected RUBs are typically associated with less resource use and higher RUBs with greater resource use. We also used EDC disease categories to define specific diseases of interest. Children with cancer were identified through the assignment of a diagnosis associated with any of the EDC malignancy categories (MAL 1-18). Children with low back pain were identified through the assignment of a diagnosis associated with the EDC diagnosis “low back pain” (MUS14).
We defined CAM professionals as acupuncturists, massage therapists, NDs, and chiropractors. Conventional professionals were defined as advanced registered nurse practitioners, physician assistants, doctors of osteopathy, physicians (including both primary and specialty care), and physical therapists. Professionals who did not fit into either of these categories, including occupational therapists and psychologists, were put into a third category called “other.”
Several variables related to expenditures were available for each visit. The amount allowed by the insurance company was chosen as the closest proxy for expense because the billed amounts were highly variable and did not necessarily reflect the amount paid. In virtually all cases, the difference between the amount paid and the amount allowed was attributable to deductibles, copayments, and coinsurance (the allowed amount being higher because it includes the patient cost-sharing amounts).
We conducted our analysis using Stata statistical analysis software, version 7.25 We combined data from the 2 companies for the analysis, because the companies’ enrollees were similar in sex, age, geographic distribution, and CAM use. We used the EDC diagnostic categories described herein to examine diagnosis groups associated with CAM and conventional professional visits. Simple frequencies were generated for EDCs on claims by professional type, including acupuncturists, massage therapists, NDs, chiropractors, and conventional professionals. Stratification according to expected resource use was performed with RUBs.
Analyses were restricted to allowed claims for visits to the defined conventional or alternative professional types. We excluded visits to other professionals and visits disallowed by the insurance company. Measures of use included whether an allowed claim for a visit was made, the number of allowed visits, and total allowed amounts for those visits.
We used χ2 tests to examine independence of specific diagnoses and use of professional services. We used logistic regression to examine the likelihood of a CAM professional claim among enrollees with claims. The regression models included age, sex, county of residence, type of services used, diagnosis, and a dummy variable indicative of whether covered adults in the family used CAM. Because of the large sample size and adjustment for multiple statistical tests, only P<.001 was considered statistically significant, and all P values were 2-sided. Much of the analysis separates chiropractic from naturopathic, acupuncture, and massage (NAM) professionals, since previous studies26 have found important differences in patterns of use between these 2 groups.
Of 788 367 enrollees ages 0 to 64 years, 187 323 were children. Of these children, 156 689 (83.6%) had an allowed claim. Among pediatric claimants, CAM was used by 6.2%. Chiropractic was used by 5.1%, naturopathic medicine by 1.0%, massage by 0.3%, and acupuncture by 0.2% (Table 1). Of the adult claimants, 154 530 (31.0%) were adult family members of covered children. Rates of CAM professional use by adult family members were higher than rates of pediatric care; chiropractic was used by 16.8% of adults, naturopathic medicine by 1.8%, massage by 3.8%, and acupuncture by 1.2% (data not shown).
Adolescents used more CAM services, except for naturopathic medicine, than any other pediatric age group. Patients who sought care from NDs were similar to those who sought care from conventional professionals in that they had greater rates of use among infants and preschoolers than among children and adolescents. The frequency of chiropractic use was greater than all other types of alternative professionals. This was especially true among adolescents, Aamong whom 8.7% used chiropractic compared with 1.9% who used NAM professionals. We found that CAM use was significantly less likely for males (odds ratio [OR], 0.91; 95% confidence interval [CI], 0.87-0.95). Logistic regression results showed that adolescent girls were 25% more likely to use a CAM professional than adolescent boys (OR, 1.25; 95% CI, 1.17-1.33). Prevalence of CAM professional use did not differ by sex in the younger age groups. When we compared patterns of use between urban (population ≥400 000) and more rural counties (population <100 000), we found that pediatric chiropractic care was more frequent in smaller counties (OR, 1.10; 95% CI, 1.03-1.16), whereas use of naturopathic medicine was less frequent (OR, 0.36; 95% CI, 0.29-0.46). Use by an adult family member (presumably a parent in most cases) significantly increased the likelihood of pediatric use of any CAM service (OR, 14.2; 95% CI, 13.5-15.0), chiropractic (OR, 17.7; 95% CI, 16.7-18.8), naturopathic medicine (OR, 81.6; 95% CI, 73.1-91.0), acupuncture (OR, 79.5; 95% CI, 61.6-102.6), and massage (OR, 37.7; 95% CI, 31.2-45.6).
The Figure shows that more than $185 million for pediatric care was spent by the insurance companies and claimants, Aand of this amount, $2.5 million (1.3%) was spent on CAM services. Table 2 gives the distribution of expenditures and visits for CAM and conventional professionals. Although visits to CAM professionals totaled 6.7% of all outpatient pediatric visits, these visits accounted for 3.6% of outpatient expenditures. This difference may in part be due to less use of expensive ancillary laboratory services by CAM professionals and relatively short follow-up visits for chiropractic. For children with claims for outpatient services from conventional, chiropractic, acupuncture, or massage therapists, the median number of visits to that type of professional was 3. For children with claims for outpatient naturopathic services, the median number of visits was 2.
Proportionate expenditures for complementary and alternative medicine (CAM).
Diagnoses associated with CAM and conventional professionals offer some insight into reasons for use of their services. Table 3 gives, for the most commonly diagnosed EDC categories, the percentage of visits that yielded those diagnoses. For example, diagnoses of conditions related to the ear, nose, and throat (eg, otitis media and upper respiratory tract infections) were made in 23.8% of visits to conventional professionals, 22.2% of visits to NDs, and 6.7% of visits to acupuncturists. Table 3 indicates that chiropractic and massage visits almost always included musculoskeletal diagnoses (98.9% and 91.4% of visits, respectively), whereas ND diagnoses were similar to those of conventional professionals. Acupuncturists fell between these extremes with regard to treating musculoskeletal problems vs general health problems.
To determine whether CAM patients were greater overall consumers of medical care, we measured CAM use within RUBs, that is, within groups having similar expected resource use. More than 60% of the children who used CAM were in the top 3 categories of expected resource use compared with only 30% of the population of children who did not use these services. Analysis suggests a significant relationship between higher RUB distribution and use of CAM, NAM, and chiropractic (P<.001 in all 3 cases). In addition, we had 362 children with malignancy diagnoses. Of these, 8.0% used chiropractic and 2.5% used NAM services; among children with no malignancies, 5.1% used chiropractic and 1.4% used NAM, respectively (P<.001 for chiropractic and P = .02 for NAM). We had 7626 children with a diagnosis of low back pain. Of these, 58.8% used at least 1 type of CAM, 55.8% used chiropractic, and 7.6% used NAM. For children without low back pain, 3.5% used any CAM, with 2.5% using chiropractic and 1.1% using NAM (P<.001 for both comparisons).
Our study used a fairly restrictive definition of CAM that was professional based and required an insurance claim for this service. Despite these restrictions, pediatric use in our insured population was more than 3 times (6.2% vs 1.8%) the rate seen in an analysis of the 1996 Medical Expenditure Panel Survey10 and approximately half the reported rate of 11% seen in Canada.27 Reasons for differences in CAM use between our data and large national surveys may include the inherent limitations of survey data (eg, recall bias), changes in preferences over time, types of professionals studied,28 and specifically, CAM insurance coverage in our cohort plus regional variations in patients and preferences.
Not surprisingly, the most significant factor that determined whether a pediatric patient would use CAM is whether an adult in the family used CAM.29 The effect of this covariate on the likelihood of a child’s use dwarfed all others. This fact may help guide professionals obtaining medical histories in the pediatric setting. In addition, most pediatric alternative care as covered by insurance is used to treat symptomatic illness. We found that CAM use was also higher in specific populations, such as children and adolescents with back pain (58.7%, P<.001), those with cancer (8.8%, P = .02), and those in higher RUBs who are expected to have higher resource use based on their comorbidity load. The most frequently used form of alternative care in pediatric populations is chiropractic. This use was more common in rural areas and may be influenced by the increased number of chiropractors in rural settings and the relative shortage of physicians. Our results suggest that NDs see a spectrum of diagnoses similar to that of conventional care professionals, and the age distribution of their pediatric patients is also similar; however, only 1% of our study population saw an ND. This share was greater in urban settings.
We found overall expenditures on CAM professional care to be a small part of the health care bill for 2 reasons. First, the expenditure per CAM visit was relatively low compared with conventional care, probably owing to the focused nature of the CAM. Second, CAM use for pediatric populations was much less prevalent than it was for adults. Most CAM professional use (except for naturopathic medicine) occurs in adolescence. Although adult use by a family member is the greatest predictor of use in childhood, this may change if adult use increases over time.
This study has several limitations. Claims data are dependent on professional ICD coding. In the clinical setting, some diagnoses may be missed, different professional types may have different coding patterns, and not all coding may be accurate. Some children’s ages may have been misclassified, because our study had only year of birth. Not all CAM professionals accept insurance and not all services are covered, so our findings must be viewed as a minimum estimate. Also, low-income populations are not represented owing to the exclusion of uninsured children and children with Medicaid. Thus, a study that included these populations would probably show less CAM use per capita. Use of CAM overall is probably higher in Washington state when compared with other states without CAM insurance requirements. One national survey30 showed that CAM use in Washington state was 16% higher than in the rest of the nation.
In summary, under the most generous reimbursement scheme in the country, pediatric CAM is currently a small budgetary item for large insurance companies. Future studies are warranted to determine the extent to which pediatric CAM use will expand as a result of having insurance coverage. Even so, we believe that these increases and associated expenditures will be modest.
Correspondence: William E. Lafferty, MD, Department of Health Services, Box 357660, University of Washington, Seattle, WA 98195 (email@example.com).
Accepted for Publication: December 6, 2004.
Funding/Support: This work was supported by National Institutes of Health (Bethesda, Md) grant NIH R01-AT00891.
Acknowledgment: We gratefully acknowledge Yuki Durham, MLS, for assistance with references.
Bellas A, Lafferty WE, Lind B, Tyree PT. Frequency, Predictors, and Expenditures for Pediatric Insurance Claims for Complementary and Alternative Medical Professionals in Washington State. Arch Pediatr Adolesc Med. 2005;159(4):367-372. doi:10.1001/archpedi.159.4.367