A number of studies have also looked at CAM in U.S. military populations and found it to be fairly consistent with that of civilians. Typical reasons cited for choosing CAM among military populations include current high daily stress, impact of military life on physical or mental health, physician-diagnosed chronic illnesses, and the potential side effects from prescription medications. Motivation for CAM use may also involve the realization that conventional care may not adequately address chronic conditions, which are often reported by those using CAM [7]. Although there is no shortage of studies involving CAM use among various defined populations in the literature, very few have considered health care utilization patterns among CAM users compared with nonusers. A recent study reported more frequent outpatient visits to physicians among those who used CAM compared to those who did not, but no difference was noted in the rate of hospitalization [8]. Additionally, Gray et al. found that CAM users reported more physical and emotional limitations, pain, and dysthymia, but were no more likely to have reported a chronic condition than nonusers [9]. A number of studies have also shown that CAM users tend to be individuals who have more comorbid, non-life-threatening health problems than nonusers [10-14]. In a cohort of military personnel, those who reported CAM use also reported a greater number of comorbidities and poorer overall health than those not reporting CAM use [15]. Findings such as these suggest those who choose CAM therapies may also have greater use of both unconventional and conventional medical services [16,17]. Understanding the health care utilization patterns of those who use CAM in a large active-duty military cohort could also increase our understanding of the health care requirements of this particular population of patients and help quantify their overall consumption of medical services.

Demographic and military characteristics of the study population by CAM use are shown in Table ​Table1.1. Women reported a higher proportion of both practitioner-assisted (38.4%) and self-administered CAM use (35.4%) compared with men (24.8% and 23.7%, respectively). Reporting practitioner-assisted CAM therapies was highest among the following: women, younger individuals, those with a high school diploma or less, those who never married or were divorced, those serving in the Marine Corps, health care workers, healthy-weight individuals, current smokers, those reporting alcohol-related problems, and those reporting having one or more health conditions or symptoms. Individuals reporting one or more health conditions were 15% more likely to report CAM use and 19% more likely to report CAM use if reporting one or more health symptoms compared to non-CAM users (see Additional file 1 for list of health symptoms and conditions). Results for self-administered CAM use were very similar to practitioner-assisted CAM, with only a few exceptions. Those reporting self-administered CAM use showed a higher percentage of use among enlisted, Navy and Coast Guard, and under-weight individuals. Both the Mental and Physical Component Summary scores derived from the SF-36V were slightly lower in CAM users compared to nonusers.

A total of 1,449 first hospitalizations occurred among this active-duty cohort within 12 months of completing the Millennium Cohort questionnaire. First hospitalization rates and adjusted odds ratios for demographic and military characteristics are displayed in Table ​Table2.2. Only two characteristics were statistically associated with a hospitalization independent of CAM use: being female 1.93 (95% CI: 1.69-2.19) or being a current smoker 1.20 (95% CI: 1.05-1.36). Birth year, education level and service branch were statistically significant for a lower probability of having a hospitalization during the study period when compared with their respective reference groups.

The unadjusted first hospitalization rate for non-CAM users was 30.5 per 1,000 and 39.0 per 1,000 for CAM users. When considering self-administered and practitioner-assisted types of CAM alone and in combination, the rate for practitioner-assisted only was 38.4 per 1,000, self-administered only was 32.9 per 1,000, and both was 43.3 per 1,000 (Table ​(Table3).3). Unadjusted and adjusted odds ratios for first hospitalizations are also shown in Table ​Table3.3. Unadjusted odds of hospitalization among CAM users was 1.29 (95% CI: 1.16-1.43). After adjusting for covariates and differences in comorbidities (propensity scores) the adjusted odds of hospitalization for CAM users compared to nonusers diminished in magnitude and became statistically nonsignificant 1.04 (95% CI: 0.93-1.17). We observed a higher probability of being hospitalized among those who reported using energy healing, chiropractic, relaxation, or massage therapies. Those using energy healing were hospitalized primarily for mental disorders, and those using chiropractic, relaxation, or massage therapy were primarily seen for diseases of the musculoskeletal system (data not shown).

Active-duty Millennium Cohort participants were also evaluated for amount of time spent utilizing inpatient or outpatient care by CAM use. The mean number of days spent in outpatient health care for CAM users was 7.0 days and 5.9 days for non-CAM users (p < 0.001), while the mean number of days spent in inpatient care was 3.2 days and 3.1 days, respectively (p = 0.85). Thirty-nine percent of persons reporting CAM use accounted for 45.1% of outpatient care and 44.8% of inpatient care.

Separate multivariable logistic regression analyses across 15 broad ICD-9-CM categories, including pregnancy and childbirth, were conducted modeling inpatient or outpatient visits by CAM use. Figure ​Figure11 illustrates each of the odds ratios for inpatient hospitalization discharge diagnoses models (excluding diseases of the blood due to sparse cases). Only hospitalization for nervous system diseases was statistically higher among those reporting CAM use compared to non-CAM users 2.72 (95% CI: 1.28-6.70). We found the majority of ICD-9-CM codes for nervous systems hospitalizations (n = 31) were for unspecified causes of encephalitis (ICD-9-CM 323.9), migraine unspecified (ICD-9-CM 346.9), and optic neuritis (ICD-9-CM 377.3). Mental disorders showed a slightly reduced odds ratio for hospitalization in CAM users 0.68 (95% CI: 0.47-0.97).

When outpatient visits were examined (Figure ​(Figure2),2), CAM users were more likely to have had an outpatient visit for musculoskeletal system diseases 1.24 (95% CI: 1.21-1.26), mental disorders 1.22 (95% CI: 1.19-1.25), and injury and poisoning 1.08 (95% CI: 1.04-1.11). CAM users were less likely to been seen for the following: skin and subcutaneous diseases, circulatory diseases, pregnancy complications, digestive system diseases, nervous system diseases, neoplasms, and endocrine, nutritional, and metabolic disorders.

Previous studies have characterized typical CAM users as female, middle aged and with more education [10,29]. Although not statistically significant, we saw the opposite trend in education and age in our study, with lower levels of education and a younger age group reporting more CAM use. Our findings of higher proportions for problems of the nervous system and sense organs that required a hospitalization among CAM users is also consistent with other studies [16,17]. However, we were not able to determine if these individuals were using CAM therapies specifically for these problems. For outpatient visits, higher rates for musculoskeletal diseases, mental disorders, and injury and poisonings were seen among CAM users compared with nonusers, and is consistent with findings previously published [17,30]. However, in this study, CAM users were also less likely to been seen for skin and subcutaneous diseases, circulatory diseases, pregnancy complications, digestive system diseases, nervous system diseases, neoplasms, and endocrine, nutritional, and metabolic disorders.

Current research indicates that CAM therapies are primarily being selected by individuals to augment but not replace conventional or mainstream medicine [1,31]. Studies of CAM use in military populations consistently indicate that approximately 40% report using some form of CAM therapy [15,32,33]. Unfortunately, studies have also shown that only 35% of persons who use CAM therapies share this knowledge with their primary health care provider [5]. This is particularly important for those who take herbal therapies or nutritional supplements because of the potential for adverse drug interactions given the important pharmacological activity of some herbal therapies and nutritional supplements [34,35]. There is well-documented evidence for herbal-drug interactions in the literature, and military health care provider awareness of CAM therapies by patients may help to avoid some potential adverse reactions [36-39].

This study has limitations that should be considered when interpreting findings. First, health outcomes that occur within a 12- month study period may not represent future health care utilization patterns in this population, given their relatively young age. Because this was an active-duty population, one might expect them to have a higher level of physical fitness and have lower disease burdens than non-active-duty personnel. In addition, we did not perform a comprehensive assessment of all CAM therapies and did not capture information on the frequency or total dose of CAM therapies. Lastly, we could not assess the health care utilization patterns among Reserve or National Guard personnel since they are only eligible for DoD health care when they are on active status.