Polypharmacy meets polyherbacy: pharmaceutical, over-the-counter, and natural health product use among Canadian adults.
Votova, Kristine ; Blais, Regis ; Penning, Margaret J. 等
Polypharmacy is a term used to describe an individual's use of
multiple medications. At present, there is no single established
criterion for polypharmacy. However, researchers tend to operationalize
it as concurrent use of 5 or more medications, (1) with "excessive
polypharmacy" referring to use of greater than 10 prescription
medications. (2) It is considered a major public health issue. (3) While
polypharmacy can occur in the general population, it tends to be
concentrated among older adults, many of whom take multiple prescription
medications to treat disease onset and progression. (4)
Ness and colleagues (5) coined the term "polyherbacy" to
describe the use of multiple natural health products (NHPs). Although
there is no established criterion as to what constitutes polyherbacy,
they and others (6) flag it as an emerging public health issue. NHPs
include dietary supplements, vitamins, minerals, and herbals, although
some consider vitamins to be outside of the realm of NHP. (7) Once
restricted to the domain of complementary and alternative medicine (CAM)
and natural health food stores, NHPs are now widely available in
mainstream grocery stores, pharmacies, and through online purchasing.
Most are accessible without a prescription. (8)
Similar growth is seen in the availability and accessibility of
over-the-counter products (OTC). Concerns are mounting that there is
excessive use of OTC products such as analgesics, at the population
level generally (9) and among older people specifically. (10) These
concerns are leading to calls for a thorough assessment of the need for
and outcomes of OTC use across age groups.
Recent clinical research also suggests that NHP use--specifically
dietary supplementation--may be detrimental to health. Researchers from
the Iowa Women's Study found higher all-cause mortality risk among
older women (65+) who were regular supplement users compared to women
who were not users. (11) Another study found that male vitamin E
supplement users were at greater risk of prostate cancer than male
non-supplement users. (12) A related concern is the safety of NHPs. NHPs
in North America are regulated as food products and are not subject to
the same patient safety regulations as pharmaceutical products;
consequently consumers think NHP use is safe. (13,14) Ironically, some
practitioners are recommending NHPs to decrease the risk of nutrient
deficiency associated with polypharmacy, (2) and are inadvertently
increasing the risk of adverse drug effects. (14)
Despite these health concerns, much of what we know about NHP use
and the extent of its use with PM and OTC is anecdotal. In Canada,
evidence of use is mainly self-reported in national surveys, (8,15,16)
small-sized or cohort samples (17-19) and case studies. (14) While PM
use can be tracked using administrative claims data in most provinces,
OTC and NHP use cannot. Gross under-reporting of nonprescription health
products is a likely result. (13) Furthermore, there has been little
recent examination of the extent of concurrent product consumption
across prescription and non-prescription products.
The aim of this study was to examine prescription medication,
over-the-counter and natural health product use by Canadian adults using
direct health measures data that have verified the individual's
self-reported use of any of these products. The main research questions
included: i) Are there discernible patterns in health product use among
Canadian adults that span PMs, OTCs and NHPs, both in terms of the
propensity for and intensity of use? ii) What are the social and health
profiles of users?
METHODS
Data
Data were drawn from the Canada Health Measures Survey (CHMS),
Cycle 1 (2007/2009). It is the only national database currently
available that provides clinical information on health product use that
includes the Drug Identification Number (DIN), product name and dosage.
The survey includes a household interview, which contains the key
socio-demographic, environmental/housing and health data necessary to
describe the individual (lifestyle) and social context of health product
use. It also includes a clinical interview, which provides
anthropometric measures and other direct health markers. As public-use
files are not available for the CHMS, data access was secured through a
Statistics Canada Research Data Centre.
The study sample (n = 3,721) included respondents aged 18-79. Those
under 18 were excluded in order to capture the health practices of
adults. The CHMS excludes individuals aged 80 and older. The overall
sample, when weighted, represents Canadians aged 18-79 in 2006 (N =
24,508,134). *
Measurement
Three dependent variables were measured. Respondents were asked
"In the past month, from [date] to yesterday, did you take
prescription medications?; "In the past month, from [date] to
yesterday, did you take over-the-counter medications?"; and
"In the past month, from [date] to yesterday, did you take any
health products or herbal remedies?" Responses were dichotomous
(yes/no) and represent the propensity or likelihood that health products
were used. Among users, the intensity of use was also measured based on
the following question "How many different ... prescribed
medications ... did you take?" The question was repeated for
over-the-counter medications and health products/herbal remedies.
Respondents were asked to confirm the type and number of products used
and to show the drug product(s) to the interviewer, who then verified
dosage and DIN. Twelve covariates were included in the analyses to
represent selected social and health factors. Age, education, household
annual income, household size, number of chronic conditions, perceived
physical health, weight (Body Mass Index or BMI), and physical activity
index (PAI) were treated as numeric (ordinal or continuous) variables.
BMI was calculated using anthropometric measures taken during the clinic
interview (weight (kg)/[height [(m).sup.2]] and classified according to
the Health Canada/WHO international categories. The remaining covariates
(gender, immigrant status, marital status and has a regular doctor) were
categorical.
Analysis
To detect and measure the relationships among the health product
indicators, latent class analysis (LCA) was used in two unrestricted
models. The first model examined the propensity (yes or no) to use any
one of the three health product groups (PM, OTC, and NHP) and the second
model examined the intensity (if yes, how many) of product use among
users of all three health product groups (i.e., tri-users of PM, OTC and
NHP). We chose LCA because it offers a unique approach to obtaining
rates of usage in populations. (20) It is different than traditional
epidemiologic methods, which tend to start with well-defined groups and
subgroups (classes that people generally think about when they think of
populations) and then look at patterns of drug use in those groups. In
contrast, LCA starts by defining groups and subgroups using non-standard
combinations of drug use that emerged out of the data. In essence, LCA
is a means by which to empirically validate assumptions in data where
both the number and form of the groups are not known a priori. (20)
Modal probability is used to allocate cases into discrete groups or
classes, where each case is assigned to a class with the highest (modal)
posterior probability of being in that class. A series of repeated runs
with random start values was used to reduce the likelihood of obtaining
a local maximum rather than the global best solution. (21) All of the
covariates regardless of scale were treated as active when defining the
latent classes, meaning their inclusion in the model influenced model
parameters and the definition of the latent classes. Once the optimum
solution was identified, logistic and multinomial logit regression
models were used to test for associations between latent classes and
covariates.
Missing values on income were treated with mean (single) imputation
because the percentage of missing data was >5%. Missing cases on any
of the indicators and the other covariates were deleted on a
case-by-case basis using listwise deletion.
All LCAs were computed using LatentGold version 4.5 (Statistical
Innovations Inc., Massachusetts, Boston) software. Default settings were
used for LCA, random seeds were set at 10 and iterations values were set
to 50. All data were weighted using CHMS sample weights and then
rescaled to the original sample size in order to avoid artificially
inflating tests of statistical significance in the analyses.
RESULTS
The study sample is described in Table 1.
Prevalence of health product use
Fifty-eight percent of Canadian adults aged 18-79 took at least one
PM in a given month, whereas 74.2% of Canadians took at least one OTC
and 38% consumed at least one NHP. The mean number of health products
taken in the previous month was highest for PMs (2.69, SD = 1.77),
followed by NHPs (2.25, SD = 1.44) and OTCs (2.01, SD = 1.30).
Defining patterns in health product use among Canadian adults
When propensity of use was assessed, three latent classes were
detected (Bayesian information criterion (BIC), [L.sup.2] = -16962.7;
nPar = 47; Wald = 51.46, p<0.001), indicating that there are three
distinct profiles of product use. In Class 1 (representing 43% of all
respondents), the conditional probability of PM and NHP but not OTC use
was high, relative to Class 2 (37%), which had a low probability of any
product use, and to Class 3 (20%), which had a high probability of PM
and OTC but not NHP use.
The coefficients from the multinomial logit regression of the
probability of membership in each of the three latent classes and the
social and health covariates are reported in Table 2. Among the social
covariates, age, gender, immigration status, household size and having a
regular doctor were significant predictors of class membership.
Significant health factors included co-morbidity (no. of chronic
conditions) and self-perceived health status.
Membership in Classes 1 (high PM, NHP and low OTC) and 2 (low PM,
NHP and OTC) was significantly predicted by being older and male.
Younger individuals and females were more likely to be in Class 3 (high
PM, OTC and low NHP). Canadian immigrants were more likely to be in
Classes 1 and 2 but not 3; Canadian-born individuals had a greater
propensity to use OTC products than immigrants. Individuals in Classes 1
and 3 were more likely to have a regular doctor, whereas individuals in
Class 2 were less likely to have one. Class 3 membership was also
significantly related to smaller household size, suggesting that these
individuals were more likely to live alone or with one other person. In
terms of the health covariates, co-morbidity was significantly
associated with membership in Classes 1 and 3 but not 2. Thus, having no
or few chronic conditions was related to low health product use overall.
Last, fair/poor health was significantly associated with being in Class
1, while good overall health predicted membership in Class 3, and very
good or excellent health predicted membership in Class 2. Thus, poor
health indicators were associated with a greater propensity to use PM
and NHP but not OTCs.
Defining patterns among tri-product users
In the second model, the intensity of health product use was
examined (see Table 3). To test for evidence of combined polypharmacy
and polyherbacy, only individuals who used at least one of each of the
three groups of products were examined. This resulted in 19% (n = 696)
of the sample being classified as tri-product users. Within this group
of users, seven distinct types emerged (BIC, LL = -4407.85; nPar = 147;
Wald = 10.09, p<0.12 **). The largest group of users (Class 1, 34%)
took at least 3 PM, 3 NHP and 1 OTC monthly. Class 2 (20%) and Class 7
(3%) could be characterized as conventional product users, as both have
higher probability of PM and OTC relative to NHP. Class 3 (14%) and 4
(11%), in contrast, have a higher probability of NHPs relative to
conventional medical products (PM and OTC). Class 5 (10%) and Class 6
(8%) have high overall rates of consumption of PMs, OTCs and NHPs,
relative to the other five classes.
Class membership for each of the seven classes was significantly
correlated with age, household size, co-morbidity (no. of chronic
conditions) and BMI status. Older individuals were more likely to belong
to Classes 1, 2, 5 and 6, while younger individuals were more likely to
be in Classes 3, 4 and 7. The overall pattern was that younger tri-users
had drug-use profiles characterized by fewer PMs and more NHPs relative
to the other four classes. Classes 3 and 7 were associated with larger
household sizes, whereas the other five classes had fewer people living
in the same household. Intensity of tri-product use was much greater for
individuals in co-morbid states, as indicated by the significant
relationship between increasing number of chronic conditions and
membership in all of the latent classes except for Classes 3, 4 and 7.
Last, there was no clear pattern overall for BMI and drug intensity.
DISCUSSION
The purpose of this study was to provide an overview of the current
health product consumption patterns of Canadian adults, including
prescription medications (PM), over-the counter (OTC) and natural health
products (NHP). This included an examination of health product use in a
two-stage process. The first stage examined propensity to use health
products and the second stage examined intensity of use among product
users. Correlates of use were assessed in both stages.
A main finding of this study was that in 2007/2009, almost one half
(43%) of the Canadian adult population under the age of 80 had a
propensity to use PMs and NHPs but were less likely to use OTCs in a
given month. In addition, almost one in five adults consumed at least
one of all three types of health products during this period of time;
for these Canadians, the potential for drug-herb contraindications was
high. Among these tri-product users, there were at least seven distinct
profiles of drug use. Heavy product use characterized about 20% of
tri-users, with each taking an average of four PMs, between two and four
OTCs, and about three NHPs in the same time period.
The rates of product use were generally consistent with those
reported elsewhere. We found that 58% of Canadian adults took at least
one PM in a given month. This is somewhat higher than the 47% reported
by Esmail. (22) However, his study examined self-reported use and use
over a longer period, therefore increasing the likelihood of subject
recall on two fronts. Our finding that 38% consumed at least one NHP in
the previous month is similar to the 41% reported by Troppmann, Johns
and Gray-Donald. (8) Finding a comparable rate of OTC use in the
literature is challenging.
Age was a significant predictor of both product use (propensity)
and, among users, the number of products taken (intensity). PM and OTC
use was greater among older adults for both models. This is consistent
with other studies. (4,9) Among tri-users (i.e., the intensity model),
the highest average NHP use was found in Classes 4, 5 and 6, with
membership in the latter two classes predicted by increasing age. For
Class 4, however, younger ages predicted membership. This latter finding
is more consistent with previous research that finds a curvilinear age
effect whereby NHP use is greater among late-middle-aged (45-64) groups
relative to older adults. (8,15,23)
Our finding that gender predicted the propensity to use health
products, with men characterized by low product use, is also consistent
with previous findings. However, our study did not find that women were
more likely to be NHP users, which is somewhat contradictory to the bulk
of research that finds that women are more likely to use NHPs relative
to their male counterparts of all ages. (8,15,24,25) It is possible that
physicians recommended (or prescribed) supplements for women, in
accordance with public health guidelines regarding daily intake of
vitamins and minerals. As a result, these women may be reporting their
supplement use as PMs or OTCs rather than NHPs.
The lack of association between income and product use was also
somewhat surprising, given that medications are not universally covered
in Canada, and that OTC and NHP products tend to involve out-of-pocket
expenses. (26) However, some private insurers have begun to offer
limited coverage for some NHP products. (27) Furthermore, provincial
variations in drug coverage plans (28) suggest that there is inequity
across provinces in terms of reimbursement.
Education also was not a significant predictor of health product
use. Yet, having a regular doctor did predict membership, such that
people without a regular doctor had higher probability of being in Class
2, characterized by a low probability of all product use. Therefore, it
is possible that medical doctors are serving as conduits of information
for patients to use (or not use) products, regardless of how many years
of education these patients have. (29)
A surprising finding is that household size was significantly
associated with both the propensity and intensity of health product use.
Individuals living in smaller households were characterized by a high
probability of PM and OTC use and low NHP use. Yet among tri-users,
living in larger households was significantly associated with membership
in two classes characterized by low mean rates of overall health product
consumption. These two classes were also significantly associated with
younger age, providing further support for the inverse relationship
between health product use and age.
We also found that immigrant status was a significant predictor of
health product use, with Canadian immigrants more likely to belong to
two latent classes characterized by high probability of PM and NHP but
not OTC use. It may be that there is less familiarity with OTC products
among individuals not born in Canada. For some ethnic groups, what they
would consider traditional medicines are actually classified as NHPs in
Canada. Future analyses that include a measure of years in Canada since
immigration and country of origin may provide a better contextual
background against which to interpret such findings.
It is also clear across both analyses that the number of chronic
conditions an individual has increased both the likelihood that health
products would be used (propensity) and the intensity of use among
tri-users. This finding is consistent with the literature on PM, (30)
OTC, (10) and NHP use. (16) What is interesting to note is that among
product users in this study, the relationship between number of chronic
conditions and NHP is opposite to that found for chronic conditions and
either PM or OTC use. This suggests that there may be some degree of
wellness care taking place for NHP users, lending support to the notion
that the spectrum of health product use may be "for my wellness,
not just my illness". (7)
Study limitations
Our study was limited by the CHMS sampling frame, which excluded
adults older than age 79. As a result, our study clearly underestimated
the extent of polypharmacy and/or polyherbacy among Canadians. Seniors
frequently use medications and this use tends to increase with age. For
example, daily medication use is greater among those aged 75+ than it is
among seniors aged 65-74 years of age. (31) Further, while NHP use tends
to be concentrated in cohorts aged 35-50, (15) there is research
indicating that the number of old-old (85+) NHP users is increasing.
(18) Additionally, there may be some degree of overlap across health
products that we could not control for. For example, some respondents
may have been prescribed OTCs by their physicians and consider aspirin,
for example, a PM, when it is typically classified as an OTC. A related
issue is that combination products, such as multivitamins, are counted
as one product. This too would lead to underestimating product use. We
also did not evaluate product use for potential drug-herb interactions
so our findings cannot speak to the frequency of drug-herb or drug-drug
interactions. However, they do suggest that given the prevalence of
concurrent drug use found in this sample, the potential for drug-herb
interactions is high. Last, we used number of health products as an
indicator of polypharmacy/polyherbacy. Some would argue that
appropriateness of drug use is a more important consideration than sheer
volume of use. Yet, it was not possible for us to measure aspects of
appropriate use, which include dosage, duplication, duration, drug-drug
interactions, and drug-disease interactions, in addition to
provider-prescribing practices and patient non-adherence. (32)
CONCLUSION
This research established a baseline estimate of health product use
with direct measures drawn from health survey data. It did so using a
two-stage analysis process (focusing on propensity and intensity) and
thus highlighted distinct patterns related to use within and across the
three groups of health products examined here. We found that almost one
half of Canadians are supplementing medication use with NHPs, and that
one in five Canadians use at least one PM, NHP and OTC regularly. These
Canadians are often older and have complex health conditions and are
potentially at risk of drug-herb interactions. This suggests that
further research should establish a criterion for what constitutes the
"poly" in polypharmacy and polyherbacy in order to inform the
conversation on appropriate product use between provider and patients.
This should be seen as a priority within the context of an aging
population in an era of chronic illness.
Funding Acknowledgement: Canadian Institutes of Health Research
(CIHR) Postdoctoral Fellowship for Kristine Votova.
Conflict of Interest: None to declare.
REFERENCES
(1.) Hoffman F, van den Bussche H, Wiese B, Schon G, Koller D,
Eisele M, et al. Impact of geriatric comorbidity and polypharmacy on
cholinesterase inhibitors prescribing in dementia. BMC Psychiatry
2011;11:190-96.
(2.) Jyrkka J, Mursu J, Enlund H, Lonnroos E. Polypharmacy and
nutritional status in elderly people. Curr Opin Clin Nutr Metab Care
2012;15(1):1-6.
(3.) Hayes BD, Klein-Schartz W, Barrueto F Jr. Polypharmacy and the
geriatric patient. Clin Geriatr Med 2007;23(2):371-90.
(4.) Hebert R, Raiche M, Dubois M, Gueye NR, Dobuc N, Tousignant M,
PRISMA Group. Impact of PRISMA, a coordination-type integrated service
delivery system for frail older people in Quebec (Canada): A
quasi-experimental study. J Gerontol B Psychol Sci Soc Sci
2010;65(1):107-18.
(5.) Ness J, Johnson D, Nisly N. Polyherbacy: Herbal supplements as
a form of polypharmacy in older adults. J Gerontol A Biol Sci Med Sci
2003;58(5):M478.
(6.) Nisly NL, Gryzlak BM, Zimmerman B, Wallace RB. Dietary
supplement polypharmacy: An unrecognized public health problem? Evid
Based Complement Alternat Med 2012;7(1):107-13.
(7.) Nitcher M, Thompson J. For my wellness, not just my illness:
North American's use of dietary supplements. Cult Med Psychiatry
2006;30:175-222.
(8.) Troppmann L, Johns T, Gray-Donald K. Natural health product
use in Canada. Can J Public Health 2002;93(6):426-30.
(9.) Turunen JH, Mantyselka PT, Kumpusalo EA, Ahonen RS. Frequent
analgesic use at population level: Prevalence and patterns of use. Pain
2005;115(3):374-81.
(10.) Pokela N, Bell JS, Lihavainen K, Sulkava R, Hartikainen S.
Analgesic use among community-dwelling people aged 75 years and older: A
population-based interview study. Am J Geriatr Pharmacother
2011;8(3):233-44.
(11.) Mursu J, Robien K, Harnack LJ, Park K, Jacobs DR. Dietary
supplements and mortality rate in older women: The Iowa Women's
Health Study. Arch Intern Med 2011;171(18):1625-33.
(12.) Klein EA, Thompson IM Jr, Tangen CM, Crowley JJ, Lucia MS,
Goodman PJ, et al. Vitamin E and the risk of prostate cancer: The
Selenium and Vitamin E Cancer Prevention Trial (SELECT). JAMA
2011;306(14):1549-56.
(13.) Bjelakovic G, Gluud C. Vitamin and mineral supplement use in
relation to all-cause mortality in the Iowa Women's Health Study.
Arch Intern Med 2011;171(18):1633-34.
(14.) Cvijov K, Boon H, Jaeger W, Vohra S. Polypharmacy, multiple
natural health products and hepatotoxicity. CMAJ
2011;183(14):E1085-E1089.
(15.) Singh SR, Levine MAH. Natural health product use in Canada:
Analysis of the National Population Health Survey. Can J Clin Pharmacol
2006;13(2):e240-e250.
(16.) Guo X, Willows N, Kuhle S, Jhangri G, Veugelers PJ. Use of
vitamin and mineral supplements among Canadian adults. Can J Public
Health 2009;100(4):357-60.
(17.) Green TJ, Barr SI, Chapman GE. The majority of older British
Columbians take Vitamin D-containing supplements. Can J Public Health
2010;101(3):246-50
(18.) McKenzie J, Keller HH. Vitamin-mineral supplementation and
use of herbal preparations among community-living older adults. Can J
Public Health 2001;92(4):296-301.
(19.) Robson PJ, Siou GL, Ullman R, Bryant HE. Sociodemographic,
health and lifestyle characteristics reported by discrete groups of
adult dietary supplement users in Alberta, Canada: Findings from the
Tomorrow Project. Public Health Nutr 2008;11(12):1238-47.
(20.) Vermunt JK, Magidson J. Latent class analysis. In: Lewis-Beck
MS, Bryman A, Futing Liao T (Eds.), Encyclopedia of Social Science
Research Methods. Thousand Oaks, CA: Sage Publications Ltd., 2003;1-21.
(21.) Magnusson D, Cairns RB. Developmental science: Towards a
unified framework. In: Cairns RB, Elder GH (Eds.), Developmental
Science. Cambridge Studies in Social and Emotional Development. New
York, NY: Cambridge Press, 1996; 7-30.
(22.) Esmail N. Complementary and alternative medicine in Canada:
Trends in use and public attitudes, 1997-2006. The Fraser Institute:
Public Policy Sources, 2007;87.
(23.) Grzywacz JG, Suerken CK, Neiberg RH. Age, ethnicity, and use
of complementary and alternative medicine in health self-management. J
Health Soc Behav 2007;48(March):84-98.
(24.) Gardiner P, Graham R, Legedza ATR, Ahn AC, Eisenberg DM,
Phillips RS. Factors associated with herbal therapy use by adults in the
United States. Altern Ther Health Med 2007;13(2):22-29.
(25.) Arcury TA, Grzywacz JG, Bell RA, Neiberg RH, Lang W, Quandt
SA. Herbal remedy use as health self-management among older adults. J
Gerontol 2007;62B(2):S142-S149.
(26.) Eisenberg DM, Davis RB, Ettner SL, Appel S, Wilkey S, Van
Rompay M, Kessler RC. Trends in alternative medicine use in the United
States, 1990-1997: Results of a follow-up national survey. JAMA
1998;280(18):1569-75.
(27.) Boon HS, Verhoef MJ, Vanderheyden LC, Westlake KP.
Complementary and alternative medicine: A rising healthcare issue.
Healthcare Policy 2006;1(3): 19-30.
(28.) Dawe JR, Morgan SG. Stitching the gaps in the Canadian public
drug coverage patchwork? A review of provincial pharmacare policy
changes from 2000 to 2012. Health Policy 2012;104:19-26.
(29.) Lupton D. Consumerism, reflexivity, and the medical
encounter. Soc Sci Med 1997;45(3):373-81.
(30.) Ramage-Morin PL. Medication use among senior Canadians.
Health Matters 2009;20(1):37-44.
(31.) Morin P, De Wals P, St-Cyr-Tribble D, Niyonsenga T, Payette
H. Pregnancy planning: A determinant of folic acid supplements use for
the primary prevention of neural tube defects. Can J Public Health
2002;93(4):259-63.
(32.) McLeod PJ, Huang AR, Tamblyn RM, Gayton DC. Defining
inappropriate practices in prescribing for elderly people: A national
consensus panel. CMAJ1997;156(3):385-91.
Received: October 16, 2012
Accepted: March 22, 2013
* Weighting is done to ensure that estimates are representative of
the population and not just the sample itself. That is, every one survey
respondent aged 18 to 79 in 2006 corresponds to 6,586 people in the
population as a whole.
** In analyses with large numbers of parameters (nPar), the p-value
is no longer a good indicator of statistical significance for model fit.
Reduction in the BIC-log likelihood is selected as the indicator in
place of p-values.
Kristine Votova, PhD, [1] Regis Blais, PhD, [2] Margaret J.
Penning, PhD, [3] Malcolm K. Maclure, ScD [1]
Author Affiliations
1. Department of Anesthesiology, Pharmacology and Therapeutics,
Faculty of Medicine, University of British Columbia, Vancouver, BC
2. Departement d'administration de la sante, Universite de
Montreal (DASUM), Montreal, QC
3. Department of Sociology, Faculty of Social Sciences, University
of Victoria, Victoria, BC
Correspondence: Kristine Votova, Dept. of Anesthesiology,
Pharmacology and Therapeutics, Faculty of Medicine, UBC, Room 215, 2176
Health Sciences Mall, Vancouver, BC V6T 1Z3, Tel: 604-822-5565, Fax:
604-822-6012, E-mail: Kristine.Votova@viha.ca
Table 1. Demographic and Health Characteristics of the
Canada Health Measures Sample (2007, 2009), Aged 18-79
Years, Weighted Sample (n = 24,508,134)
Covariates %
Male 49.3
Mean age (years) 44.56 SD = 17.32
Marital status
Married/common-law 66.2
Divorced/widowed/separated 11
Single, never married 22.7
Mean household size 2.74 SD = 1.24
Mean household income ($) $41,808.56 SD = 35,894.64
Respondent highest education
Less than high school 12.7
Graduated high school 18.8
Other post-secondary 9.4
Graduated post-secondary 59.1
% with family doctor 84.3
Self-rated health status
Fair/poor 10.9
Good 35.6
Very good/excellent 53.5
Mean no. of chronic conditions 1.28 SD = 1.53
% with chronic condition 60.2
% immigrated to Canada 23.8
Mean age at immigration to Canada 25.08 SD = 12.49
Body Mass Index *
Underweight 1.5
Normal 37.8
Overweight 36.8
Obese 15.1
Very/severely obese 8.8
Source: Canada Health Measures Survey, Cycle 1 (2007/2009).
* Body Mass Index (BMI) norms for adults aged 18+ was used.
Table 2. Mean Rates (Indicator Variables) and Multinomial Logit
Regression Coefficients * for the Canadian Population (Aged 18-79)
Who Used At Least One of Each of the Three Health Product Groups
(Prescription Medications (PM), Over-the-counter (OTC), and
Natural Health Products (NHP)) in the Previous Month, by Latent
Class, Weighted Sample (N = 24,508,134)
Class 1 Class 2 Class 3
(43%) (37%) (20%)
Indicators
PM
Yes 1.47 -1.88 0.41
No -1.47 1.88 -0.41
OTC
Yes -0.16 -0.29 0.46
No 0.16 0.29 -0.46
NHP
Yes 0.19 -0.12 -0.07
No -0.19 0.12 0.07
Social covariates
Age 0.06 0.00 -0.07
Gender
Male 0.30 0.52 -0.82
Female -0.30 -0.52 0.82
Household income 0.00 0.00 0.00
Respondent education ([dagger]) -0.13 -0.05 0.17
Has a regular doctor 0.29 -0.31 0.02
-0.29 0.31 -0.02
Canadian immigrant
Yes 0.29 0.37 -0.66
No * -0.29 -0.37 0.66
Household size (persons) 0.04 0.10 -0.14
Marital status
Married/Common-law 0.03 -0.01 -0.02
Divorced/Separated/Widowed 0.19 0.07 -0.26
Single, never married -0.22 -0.06 0.28
Health covariates
No. of chronic conditions 0.43 -0.78 0.35
Self-perceived health
Fair/poor 0.44 -0.05 -0.40
Good -0.09 -0.05 0.14
Very good/excellent -0.35 0.10 0.25
Body Mass Index
Underweight 0.11 0.15 -0.25
Normal weight 0.02 0.07 -0.09
Overweight -0.03 0.01 0.01
Obese 0.30 -0.03 -0.26
Very obese 0.26 0.03 -0.29
Severely obese -0.65 -0.23 0.88
Physical Activity Index -0.11 -0.04 0.15
([double dagger])
Wald P-value
Indicators
PM
Yes 51.46 0.00
No
OTC
Yes 42.58 0.00
No
NHP
Yes 49.60 0.00
No
Social covariates
Age 62.54 0.00
Gender
Male 47.48 0.00
Female
Household income 2.83 0.24
Respondent education ([dagger]) 3.23 0.20
Has a regular doctor 24.98 0.00
Canadian immigrant
Yes 11.75 0.00
No *
Household size (persons) 6.12 0.05
Marital status
Married/Common-law 2.40 0.66
Divorced/Separated/Widowed
Single, never married
Health covariates
No. of chronic conditions 98.31 0.00
Self-perceived health
Fair/poor 11.97 0.02
Good
Very good/excellent
Body Mass Index
Underweight 5.55 0.85
Normal weight
Overweight
Obese
Very obese
Severely obese
Physical Activity Index 1.49 0.47
([double dagger])
Source: Canada Health Measures Survey (CHMS), Cycle 1 (2007/2009).
* Regression coefficients are effect coded.
([dagger]) Respondent education was treated as a continuous
variable in the propensity model.
([double dagger]) Physical activity index was treated as a
continuous variable in the propensity model.
Table 3. Mean Rates (Indicators) and Linear Regression
Coefficients* (Covariates) for the Canadian Population (Aged
18-79) Who Used At Least One of Each of the Three Health Product
Groups (Prescription Medications (PM), Over-the-counter (OTC),
and Natural Health Products (NHP)) in the Previous Month, by
Latent Class and Covariates, Weighted Sample (n=4,656,545)
Class 1 Class 2 Class 3
(34%) (20%) (14%)
Indicators
PM (mean no.) 3.05 3.91 1.00
OTC (mean no.) 1.00 2.66 1.98
NHP (mean no.) 2.61 1.00 1.00
Covariates
Age (years) 0.03 0.02 -0.01
Gender
Male -0.03 -0.22 0.18
Female 0.03 0.22 -0.18
Respondent
education
([dagger])
< than 0.07 0.86 -0.05
secondary
Graduated 0.89 0.87 0.81
secondary
Some -0.16 -1.20 0.05
post-
secondary
Graduated -0.80 -0.54 -0.81
post-
secondary
Household -0.20 -0.28 -0.15
income
Canadian
immigrant
Yes -0.18 0.01 -0.01
No 0.18 -0.01 0.01
Has a
regular
doctor
Yes -0.14 -0.26 -0.35
No 0.14 0.26 0.35
Household -0.04 -0.05 0.10
size
(persons)
Marital
status
Married/ -0.46 -0.26 -0.35
Common-
law
Divorced/ 0.86 0.93 1.00
Separated/
Widowed
Single, -0.39 -0.67 -0.66
never
married
No. of 0.14 0.37 -0.48
chronic
conditions
Self-perceived
health
Fair/poor 0.15 0.59 -0.45
Good 0.02 -0.24 0.18
Very good/ -0.18 -0.35 0.27
excellent
Body Mass
Index (BMI)
Underweight -0.09 -5.38 1.72
Normal -0.18 0.43 -0.62
weight
Overweight -0.05 0.61 -0.79
Obese 0.24 1.74 0.06
Very obese 0.33 2.23 -0.12
Severely -0.26 0.37 -0.24
obese
Physical
Activity
Index
([double
dagger])
Active -0.03 0.29 -0.08
Moderately -0.13 -0.33 -0.16
active
Inactive 0.15 0.04 0.24
Class 4 Class 5 Class 6
(11%) (10%) (8%)
Indicators
PM (mean no.) 1.00 4.04 4.17
OTC (mean no.) 2.23 2.00 4.01
NHP (mean no.) 3.37 3.42 3.13
Covariates
Age (years) -0.01 0.03 0.04
Gender
Male -0.22 -0.26 -0.20
Female 0.22 0.26 0.20
Respondent
education
([dagger])
< than 0.09 0.58 0.03
secondary
Graduated 0.58 0.73 0.78
secondary
Some -0.28 -0.75 -0.52
post-
secondary
Graduated -0.39 -0.56 -0.28
post-
secondary
Household 0.10 0.16 -0.27
income
Canadian
immigrant
Yes 0.59 -0.25 -0.18
No -0.59 0.25 0.18
Has a
regular
doctor
Yes -0.39 -0.03 0.17
No 0.39 0.03 -0.17
Household -0.28 -0.42 -0.34
size
(persons)
Marital
status
Married/ -0.35 -0.76 -0.14
Common-
law
Divorced/ 0.66 1.16 0.80
Separated/
Widowed
Single, -0.31 -0.4 -0.66
never
married
No. of -0.58 0.56 0.47
chronic
conditions
Self-perceived
health
Fair/poor 0.03 -0.34 0.07
Good -0.42 0.17 0.10
Very good/ 0.39 0.17 -0.18
excellent
Body Mass
Index (BMI)
Underweight 1.89 -3.7 2.02
Normal -0.41 1.64 -0.15
weight
Overweight -0.59 2.01 0.62
Obese 0.35 1.45 0.97
Very obese -3.5 3.26 1.99
Severely 2.25 -4.66 -5.45
obese
Physical
Activity
Index
([double
dagger])
Active 0.29 -0.43 0.19
Moderately 0.14 0.14 0.03
active
Inactive -0.43 0.29 -0.22
Class 7 Wald P-value
(3%)
Indicators
PM (mean no.) 2.00
OTC (mean no.) 2.51
NHP (mean no.) 2.07
Covariates
Age (years) -0.09 13.67 0.03
Gender
Male 0.75 9.22 0.16
Female -0.75
Respondent
education
([dagger])
< than -1.57 20.93 0.28
secondary
Graduated -4.66
secondary
Some 2.86
post-
secondary
Graduated 3.37
post-
secondary
Household 0.24 7.09 0.31
income
Canadian
immigrant
Yes 0.03 11.69 0.07
No -0.03
Has a
regular
doctor
Yes 1.02 4.47 0.61
No -1.02
Household 1.03 14.36 0.03
size
(persons)
Marital
status
Married/ 2.33 8.75 0.73
Common-
law
Divorced/ -5.41
Separated/
Widowed
Single, 3.08
never
married
No. of -0.48 57.20 0.00
chronic
conditions
Self-perceived
health
Fair/poor -0.06 18.19 0.11
Good 0.18
Very good/ -0.12
excellent
Body Mass
Index (BMI)
Underweight 3.53 44.88 0.04
Normal -0.72
weight
Overweight -1.81
Obese -4.81
Very obese -4.18
Severely 7.99
obese
Physical
Activity
Index
([double
dagger])
Active -0.24 15.09 0.24
Moderately 0.30
active
Inactive -0.06
Source: Canada Health Measures Survey, Cycle 1 (2007/2009),
Statistics Canada.
* Regression coefficients are effect coded.
([dagger]) Respondent education was treated as a
categorical variable in the intensity model.
([double dagger]) Physical activity index was treated
as a categorical variable in the intensity model.
