Use of Complementary and Alternative Medicine Providers by Fibromyalgia Patients Under Insurance Coverage

Abstract

Objective

To quantify how visits and expenditures differ between insured patients with fibromyalgia syndrome (FMS) who use complementary and alternative medicine (CAM) providers compared to FMS patients who do not. FMS patients were also compared to an age and gender matched comparison group without FMS.

Methods

Calendar year 2002 claims data from two large insurers in Washington State were analyzed for provider type (CAM vs. conventional), patient comorbid medical conditions, number of visits, and expenditures.

Results

Use of CAM providers by FMS patients was two and a half times higher than in a comparison group without FMS. (56% vs. 21%) FMS patients who used CAM had more health care visits than FMS patients not using CAM (34 vs. 23, p < .001); however, CAM users had similar expenditures to non-users among FMS patients ($4638 vs. $4728, n.s.), because expenditure per CAM visit is lower than expenditure per conventional visit. FMS patients who used CAM also had heavier overall disease burdens than those not using CAM.

Conclusion

With insurance coverage, a majority of FMS patients will use CAM providers. The sickest patients use more CAM and this leads to an increased number of health care visits. However, CAM use is not associated with higher overall expenditures. Until a cure for FMS is found, CAM providers may offer an economical alternative for FMS patients seeking symptomatic relief.

INTRODUCTION

Fibromyalgia syndrome (FMS) is a chronic, painful condition of unknown cause that has no definitive cure(1,2). FMS patients are high users of medical care with associated high expenditures (3–9), likely due to a continuing search for relief from chronic soft tissue pain, fatigue, and sleep disturbance (10–12). Surveys have documented that complementary and alternative medicine (CAM) therapies are used by most FMS patients in their search for relief; several reports have estimated that more than 90% of FMS patients use CAM (1,13,14), many seeing CAM providers such as massage therapists (range of use 44%–53% of FMS patients), chiropractors (37%–47%), acupuncturists (11%–22%), and naturopathic physicians (37%) (13–15).

Some CAM use may be attributable to the effectiveness of the treatment. Pain relief in FMS patients has been demonstrated in controlled trials of acupuncture (16–19),although a recent trial found no beneficial effect (20) There are also suggestions that chiropractic (2) and connective tissue massage (21) may provide short term symptom relief. However, most CAM therapies remain unproven. The appeal of CAM providers may arise primarily from the hope that CAM providers offer (9), the time and attention they give to patients (22), and in their empathy and listening skills (23).

Given the high annual medical expenditures for conventional care among FMS patients, third party payers have been concerned about the consequences of adding coverage of CAM provider care. Insurance leads to increased use of certain types of health services (24, 25), but adding CAM coverage may not necessarily increase expenditures if it replaces more expensive conventional services. Since 1996 in Washington State, private commercial insurance companies are required by law to have a CAM provider benefit, which allowed us to use insurance claims to investigate three issues. First, how does CAM use by insured FMS patients compare to CAM use in other insured patients? Second, among FMS patients who use CAM providers, how is their health care utilization apportioned between CAM and conventional care? Finally, how do health care expenditures differ between FMS patients who use CAM and those who do not?

METHODS

Study Population

We created a cross-sectional cohort of adults with FMS using claims data from two large insurers in Washington State. The analysis was limited to enrollees in health insurance plans directly regulated by the law requiring CAM coverage in the year 2002. This excluded Medicare, Medicaid, state-supplemental programs, and self-insured plans that are exempt from state regulation. The analyses presented here were limited to adults aged 18–64 who had both continuous enrollment in a single plan and complete claims information for the year 2002. The claims came from a variety of plans with differing benefit structures, deductibles, and copayments. However, in general the access to CAM providers under each plan was similar to access to conventional providers. Self-referral to chiropractic was allowed under all plans. There were generally visit limits for CAM providers, which affected the amount paid to CAM providers by the insurance company but not the prevalence of CAM use.

Expenditures were measured using the amount allowed by the insurance company for each visit. Visits which were disallowed by the insurance company were excluded from the analysis (4.6% of all claims among patients with FMS). Total expenditures included all claims paid by the insurance company, and as such included inpatient, outpatient, skilled nursing facility, home health care, and other claims paid by the insurance company except payments for pharmaceutical claims, which were considered separately.

A “visit” was defined as a claim from a unique provider with a unique date. That is, a person could not have more than one visit to a given provider on one day. Visits do not always correspond one-to-one with submitted claims. Also, while we generally refer to these unique encounters as visits, they also include claims that did not correspond to an actual office visit, such as hospital claims, laboratory claims, and others.

The study population for this analysis consisted of adults with a diagnosis of fibromyalgia (International Classification of Disease (ICD9) code 729.1) at one or more allowed provider visits. We also repeated the analyses requiring two separate claims with an FMS diagnosis to define FMS, in order to look at possible effects of false positives due to “rule out” codings or miscoding of diagnoses on insurance claims (26–28). Also, requiring two FMS diagnoses singles out the heaviest health care users and thus represented a “worst case” scenario in terms of adding to health care expenditures.

To compare CAM use of FMS patients and patients without FMS, we randomly selected an age- and gender-matched set of insured adults who had at least one outpatient visit during the year but had no visits for fibromyalgia.

Provider Groups

Providers were divided into three groups. CAM providers were defined as chiropractors, licensed massage therapists, acupuncturists, and naturopathic physicians. Conventional providers were defined as physicians (including all specialties), physical therapists, occupational therapists, mental health providers (e.g., psychologists, counselors, social workers), podiatrists, advanced registered nurse practitioners, and physician assistants. All providers who did not fit into either of these categories were put into a third category called “Other.” Chiropractic care has been covered by insurance for many years, while coverage for the other CAM provider types is much more recent. Therefore for some analyses the non-chiropractic provider types were grouped as NAM providers (Naturopathic physicians, Acupuncturists, and Massage therapists).

CAM users were defined as patients with one or more visits to a CAM provider. It is important to note that this is a self-selected group of patients who chose to use CAM providers for some or all of their care. While this self-selection implies that there is almost certainly selection bias present in the group that chose to use CAM, our analyses may indicate how CAM coverage might be expected to be used in other “real life” settings in which patients choose whether or not to use a covered CAM benefit.

Other medical conditions

Using the Johns Hopkins ACG software, Version 6 (29), we constructed two measures of the types of diseases or disorders present and the expected resource utilization for each patient. The first measure is Expanded Diagnosis Clusters (EDCs), which categorizes ICD-9 codes into 26 major disease categories and then creates 26 indicator variables for the presence or absence of each category for each individual. The second measure is an index of overall disease burden, which we refer to as “morbidity group” and is based on the Adjusted Clinical Group (ACG) index. The ACG index is an overall measure of disease burden and expected resource use and has 82 categories. (30) The morbidity groups created from this index collapse the 82 categories into 5 groups based on similarity of expected resource use. The cut points for the five groups were calculated at Johns Hopkins using a national sample. Lower morbidity groups include individuals with less expected resource use and higher morbidity groups include those with greater expected resource use.

Statistical Analyses

Predictors of CAM use were modeled using logistic regression. Independent variables were age group (18–25 and then 5-year age groups from 26–30 to 61–64), gender, county population indicators, insurance product line (Preferred Provider Organization [PPO], Point of Service [POS], and Traditional Fee-For Service [FFS] compared to Health Maintenance Organization [HMO]), indicators for EDC categories, and indicators for the level of expected resource use (measured by morbidity categories).

Expenditures were modeled using linear regression. Although expenditure data were skewed, our data set was large enough that ordinary least squares regression provided accurate estimates of coefficients and standard errors (31). Independent variables were use of CAM, age group, gender, county size, type of insurance product, indicators for EDC categories, indicators for morbidity category, and interactions between the morbidity category indicators and use of CAM. Morbidity categories were included to compare expenditures between CAM users and non-users after adjusting for differences in the expected resource use in these two groups. Stata statistical software version 8.0 was used for all analyses (32).

RESULTS

Table 1 describes the study population for this analysis. The data file included 497,648 adults with allowed claims, of whom 13,792 (2.8%) had FMS (based on one or more claims related to FMS). Median age of FMS claimants was 47, and 74% were female. The comparison group included 41,427 individuals without FMS who were age- and gender-matched to FMS patients. Distributions of all claimants, FMS patients, and the comparison group were similar for insurance product line and for county population (except that CAM users in the comparison group were less likely to live in the largest counties). FMS patients differed from the other groups in their distribution between insurance companies and distribution of morbidity categories. Only 2% of FMS patients were in the lowest morbidity category, compared to 28% of the comparison group.

Table 1.

Characteristics of the Study Population and Comparison Group: Adults (ages 18–64) with Private Commercial Insurance Coverage in Washington State

			Total		CAM Users		NAM Users3
	Enrollees with allowed claims (n = 497,648)		FMS1 (n = 13,792)	Comparison2 (n =41,427)	FMS (n = 7,721)	Comparison (n =8,713)	FMS (n = 4,554)	Comparison (n =3,019)
	n	%	%	%		%		%
Median age, years	(43)		(47)	(47)	(46)	(47)	(46)	(47)
Female	283,166	57	74	74	76	77	82	86
County of Residence, population in 1000s:
> 400	320,471	64	64	64	63	57	64	64
100 – 399	90,333	17	17	19	18	21	18	20
< 100	86,844	19	19	18	19	22	17	16
Product line4
HMO	63,156	13	13	13	12	11	11	11
PPO	287,435	58	58	58	61	64	63	64
POS	116,858	23	22	23	21	19	20	20
Traditional	30,187	6	7	6	6	6	6	5
Insurance co.
B	235,502	50	50	47	51	43	48	46
C	262,146	50	50	53	49	57	52	54
Morbidity cateogry5
Low	163,560	33	2	28	1	14	1	9
Middle	259,213	53	60	56	59	60	56	60
High	71,903	14	37	16	40	27	43	32

¹FMS = Patients with fibromyalgia syndrome, defined as one or more claims containing ICD-9 code 729.1

²Comparison = Randomly selected patients who did not have any visits with ICD-9 codes for FMS, age and gender-matched to FMS patients.

³Subset of CAM users who saw a naturopathic physician, acupuncturist, or massage therapist.

⁴Product Line: HMO = Health Maintenance Organization, PPO = Preferred Provider Organization, POS = Point of Service.

⁵Low = morbidity category 1 or 2, Middle = morbidity category 3, High = morbidity category 4 or 5

Table 2 describes the rates of CAM provider use by FMS patients and the comparison group. Among the FMS patients, 7221 (56%) had at least one visit to a CAM provider, and of these 4554 (33% of all FMS claimants) had at least one visit to a NAM provider. This represented much higher CAM utilization than was seen in the random comparison group, among whom 21% had one or more visits to a CAM provider, and 7% had one or more visits to a NAM provider. The difference in CAM use was most striking for acupuncture and massage, where the rate among FMS patients was more than 5 times the rate in the comparison group.

Table 2.

Provider types used among FMS patients and comparison group

	FMS patients		Random comparison group
	n	%	n	%
All	13,792	100	41,427	100
No CAM	6,071	44.0	32,714	79.0
Any CAM	7,221	56.0	8,713	21.0
Any NAM	4,554	33.0	3,019	7.3
Chiropractic	5,641	40.9	7,045	17.0
Acupuncture	1,126	8.2	683	1.6
Massage	3,368	24.4	1,840	4.4
Naturopathy	1,325	9.6	927	2.2

Visits and Expenditures

Overall, FMS patients had an average of 29 outpatient visits per year. FMS patients who used any CAM had significantly more annual visits (mean ± s.d.) (34 ± 25) than those who did not use CAM (23 ± 21, p < .001). (Figure 1) In both CAM users and non-users, most conventional visits were made either to physicians (76–79%) or physical therapists (10–12%); 4–5% of visits were made to mental health professionals.

Figure 1.

Average annual outpatient visits by FMS patients and the comparison group, separated by CAM use

Expenditures show a different pattern. Average annual expenditures (including both inpatient and outpatient charges) were similar between FMS patients who used CAM ($4638 ± 9660) and those who did not use CAM ($4728 ± 10564). Outpatient expenditures were also similar between CAM users and non-users (3473 ± 4926 and $3269 ± 6489, respectively) (Figure 2) This apparent contradiction of more visits but similar expenditures among CAM users occurs because the average allowed amount for a CAM visit is much lower than the average allowed amount for a conventional visit ($56 ± $31 and $130 ± $277 respectively). (data not shown) CAM users have slightly higher average annual expenditures than non-CAM users in the lower half of the expenditure distribution, ranging from $40 higher at the 1^st percentile to $500 higher at the 50^th percentile. However, this difference is offset in the upper end of the distribution, where CAM users have substantially lower average annual expenditures than non-CAM users, for example $2600 lower at the 95^th percentile and $14,000 lower at the 99^th percentile. (data not shown)

Figure 2.

Average annual outpatient expenditures by FMS patients and the comparison group, separated by CAM use

Further, although CAM visits accounted for 25% of all claims among all FMS patients in 2002 (98,780 out of 399,709 claims), CAM expenditures accounted for only 8% of the total expenditures by FMS patients ($5.5 million out of $65.2 million). CAM claims among those who used CAM accounted for an average of 42% of their annual outpatient claims, but the average CAM expenditure of $718 was only 21% of the average annual outpatient expenditures of $3473 in this group. (Figures 1 & 2) For comparison, conventional claims comprised 53% of all outpatient claims and 45% of expenditures.

Different patterns of visits and expenditures were seen in the random comparison group. There, CAM users had twice as many visits as non-users (Figure 1), and CAM users also had higher expenditures than non-CAM users. (Figure 2)

Among FMS patients, the relationships between average expenditures in CAM users and non-users is complicated by an interaction between CAM use and morbidity category. In the regression models adjusted for gender, age, insurance product and company, county population size, indicators for EDC categories, indicators for morbidity category, and the interaction between morbidity category and use of CAM, we found that CAM users had higher expenditures than non-users in the low morbidity categories but lower expenditures in the highest morbidity category. (Table 3).

Table 3.

Average Expenditures by CAM Use and Morbidity Category, from Linear Regression Analysis¹

	No CAM Use		CAM Users
	beta2	s.e.	beta	s.e.
Morbidity category: Lowest	(reference)		$516	160
3	−$198	$157	467	125
4	1968	307	1934	224
Highest	9687	1002	7682	703

¹Independent variables were CAM use indicator, morbidity category indicators, interactions between CAM use and morbidity cateogry indicators, age, sex, county population, insurance product indicators, insurance company, and EDC category indicators.

²Beta coefficients show the amount that average annual expenditures differed from non-CAM users in the lowest morbidity category. For CAM users, beta coefficients were obtained by summing the coefficients for CAM use, morbidity category, and the interaction of CAM use and morbidity cateogry. Positve values indicate higher expenditures than the reference category, and negative values indicate lower expenditures than the reference category.

Pharmaceutical claims were considered separately. CAM users had both fewer pharmacy claims during the year (20.4 vs. 26.6, p < .001) and lower pharmacy expenditures ($1914 vs. $2346, p = .002).

Predictors of CAM use

In the logistic regression analysis of FMS patients, by far the strongest predictor of CAM use was being in morbidity category 3, 4, or 5 (ORs 5.4 – 10.0). Female gender was also associated with a slightly greater odds of CAM use (OR 1.3). Compared to those aged 18–25, the odds of CAM use were higher in age groups between 26 and 40, and lower in those aged 56 to 64. (Table 4)

Table 4.

Predictors of CAM Use Among FMS Patients, from Logistic Regression Model^*

	OR	95 % c.i.
Female	1.28	(1.18, 1.21)
Age, years
18–25	1.00
26–30	1.57	(1.26, 1.97)
31–35	1.68	(1.38, 2.04)
36–40	1.33	(1.10, 1.60)
41–45	1.10	(0.92, 1.32)
46–50	1.09	(0.92, 1.30)
51–55	0.99	(0.83, 1.18)
56–60	0.83	(0.69, 0.99)
61–64	0.59	(0.48, 0.73)
County population, in 1000s
<100	1.00
100–400	1.03	(0.91, 1.15)
>400	0.92	(0.82, 1.00)
Morbidity category
2	1.00
3	5.41	(3.99, 7.14)
4	9.03	(6.76, 12.07)
5	9.99	(8.55, 13.63)

^*Also adjusted for insurance company, product type, and Expanded Diagnosis Cluster indicators. OR= odds ration; 95% CI=95% confidence interval.

Analysis using two visits to define FMS

We repeated these analyses restricting the FMS group to those with at least two ICD9-identified claims for FMS during the year. There were 7626 FMS patients using this definition, 68% of whom had at least one CAM visit. The observed patterns were similar to those reported here, with CAM claims among CAM users accounting for 44% of claims for those patients but only 18% of their expenditures. Using this definition, CAM users had significantly lower annual expenditures than non-CAM users ($4390 vs. $5535, p < .001).

DISCUSSION

We studied the extent to which FMS patients would use CAM providers when this care was covered by insurance. Our results showed that over half of FMS patients visited a CAM provider during a one-year period, compared to 21% of the comparison group. The purpose of the comparison group was to give us an idea of how a cross-section of patients chose to use CAM providers when they were covered by insurance. In this insured group, FMS patients used CAM providers at a much higher rate than other patients, which was similar to previous reports that did not consider insurance coverage.

Use of chiropractic in our sample of FMS patients was similar to previous studies of FMS patients (41% compared to 37–47%) (14, 15), but use of NAM providers was lower than in previous FMS studies (13, 15). This may merely reflect differences between survey data and claims data or the small sample size in the previous reports, or it may reflect the fact that chiropractic has a long history of inclusion in insurance coverage, while coverage of NAM providers is newer. Use of NAM by FMS patients may increase as patients become more aware of its availability under insurance coverage.

Although FMS patients who used CAM had more visits during the year than those not using CAM, they had similar overall insurance expenditures for provider care, both outpatient and total. The average annual expenditure to CAM providers of $718 was offset by slightly lower expenditures for outpatient conventional care and inpatient care. This implies that covering CAM care is not resulting in additional cost to the insurance company. However, had CAM providers not been included under insurance coverage, we cannot say the extent to which CAM care would have been replaced by additional conventional care, paid for out of pocket, or foregone altogether. We cannot determine whether our results are due to self selection in those who use CAM or whether some effect of CAM care is actually reducing other expenditures. Thus we cannot determine conclusively the effect of CAM coverage on insurance expenditures for FMS patients. Most of the insurance plans included in these claims had visit limits on the use of CAM providers. If FMS patients had used additional conventional care once the CAM limits were reached, we would have expected to see higher expenditures in this group rather than lower. We do not know if they continued to see CAM providers and paid out of pocket, or if the amount of CAM care received provided enough relief to obviate the need for additional conventional care. In either case, our conclusion that FMS patients who used CAM providers had similar overall insurance expenditures is unchanged. Additionally, our results do provide information on how patients may use a CAM benefit in the “real world,” in which they will self-select whether or not to use covered CAM benefits.

The discussion above includes only provider-based claims and not pharmacy claims, where expenditures for CAM users were an average of $432 lower than average pharmacy expenditures for those not using CAM. This saving offsets more than half of the average annual expenditure to CAM providers.

This analysis has several limitations. First, not all FMS patients seek medical care for FMS in a one year period (3). We do not know how our results would have differed if we had looked at a longer time period and thus included those with less frequent utilization. Second, these data contain limited information on personal characteristics of the patients, limiting our ability to adjust for potentially confounding demographic factors such as income, education, and race. Third, because patients are not randomized to CAM use, self-selection bias is likely present in these data. Using risk adjustment indices such as the ACG-based morbidity groups mitigates this bias but does not remove it entirely, and we do not know the effect of any residual bias. Finally, we do not include any non-insured CAM use such as the use of herbal remedies or nutritional supplements. Our intent is to show the impact of insurance coverage on insurance expenditures. Our results do not generalize to those without insurance coverage for CAM providers, but seen to suggest that coverage of CAM by public programs such as Medicare and Medicaid would not increase expenditures, and in fact might even lead to lower expenditures among the heaviest health care users (e.g., those with two or more FMS-related visits per year).

Until a cure for FMS is found, CAM providers may offer an economic alternative for symptomatic relief. Pain is a subjective outcome, so patient perception of the effectiveness of care may have higher relative importance in this setting than for other disorders with more objective biologic outcomes. Thus further research will be important to assess patient perceptions of the effectiveness of CAM care compared to conventional care, and to ascertain whether in the long run CAM care may truly be cost-effective.