Obesity Predicts Primary Health Care Visits: A Cohort Study

Abstract

The objective of this study was to explore the relationship between body mass index (BMI), its association with chronic disease, and its impact on health services utilization in the province of Newfoundland and Labrador, Canada, from 1998 to 2002. A data linkage study was conducted involving a provincial health survey linked to 2 health care use administrative databases. The study population comprised 2345 adults between the ages of 20 and 64 years. Self-reported height and weight measures and other covariates, including chronic diseases, were obtained from a provincial survey. BMI categories include: normal weight (BMI 18.5–24.9), overweight (BMI 25–29.9), obese class I (BMI 30–34.9), obese class II (BMI ≥35), and obese class III (BMI ≥40). Survey responses were linked with objective physician and hospital health services utilization over a 5-year period. Weight classifications in the study sample were as follows: 37% normal, 39% overweight, 17% obese, and 6% morbidly obese. The obese and morbidly obese were more likely to report having serious chronic conditions after adjusting for age and sex. Only the morbidly obese group (BMI ≥35 kg/m²) had a significantly higher number of visits to a general practitioner (GP) over a 5-year period compared to the normal weight group (median 22.0 vs. 17.0, P < 0.05). Using multivariate models and controlling for the number of chronic conditions and other relevant covariates, being morbidly obese remained a significant predictor of GP visits (P < 0.001), but was not a predictor for visits to a specialist or any type of hospital use. The increase in the prevalence of obesity is placing a burden at the primary health care level. More resources are needed in order to support GPs in their efforts to manage and treat obese adults who have associated comorbidities. (Population Health Management 2012;15:29–36)

Introduction

I n most countries in the developed world, the prevalence of obesity has risen dramatically over recent decades¹ and is considered a public health concern. In 2005, 17% of Canadian adults were classified as obese compared to approximately 6% in 1985.^2,3 The prevalence of adult obesity varies widely across Canada from a high of 26% in Newfoundland and Labrador (NL) to a low of 13% in British Columbia.⁴ Even more concerning is that, using measured heights and weights from the 2004 Canadian Community Health Survey (CCHS), Statistics Canada reported 24% of the Canadian adult population and 34% of the NL adult population to be obese.⁵ This trend is now being observed in the increasing prevalence of childhood obesity in Canada.^6,7 Health care professionals are concerned about the increasing prevalence of obesity, its known risk factor status for chronic conditions such as cardiovascular disease, hypertension, diabetes mellitus, and some cancers, its adverse psychosocial consequences,⁸ and its relationship with premature mortality.^9,10 It has been estimated that obesity costs the Canadian health care system between $3.9 and $4.3 billion in direct and indirect medical costs.^11,12 Although researchers have explored the relationship between body mass index (BMI; kg/m²) as a reflection of excess body weight and its impact on health care use,^{13

–20} most studies have relied on self-reported health care use obtained from health surveys. It is surprising that in Canada, given the high prevalence of adult obesity, this relationship has not been studied more extensively. Very few studies have been published using Canadian data, and those that have been published used self-reported physician and hospital use data and are inconsistent in their findings.^21,22,23 In one study, researchers performed a secondary analysis of the 1994 Canadian National Population Health Survey and reported that being obese (BMI ≥27) was associated with increased physician visits. In addition, compared with nonobese individuals, obese individuals were more likely to have reported having a number of chronic conditions, such as diabetes and cardiovascular disorders, and to have been prescribed a larger number of medications (eg, heart medication, antihypertensive pills, diuretics, pain relievers for arthritis and back pain). However, in contrast, obese respondents had fewer hospital admissions than nonobese respondents.²¹ In another Canadian study that used data extracted from the 2003 CCHS, the authors reported that being obese (BMI ≥30) was associated with an increased likelihood of being an inpatient and was a significant predictor for most categories of length of stay (LOS).²² One study of a provincial population that used self-reported data from the 2001 CCHS reported that being obese was associated with both an increase in number of chronic conditions and the number of visits to a general practitioner (GP). However, being obese was not associated with increased visits to a specialist or any type of hospital service.²³ To our knowledge, no studies in Canada have used an individual-level approach to examine the impact of obesity on actual health care services use.

The objectives of this study are to determine whether obese adults 20–64 years of age living in NL, Canada, use more health care services (ie, physician, hospital) compared with normal weight individuals over a 5-year period, and to examine whether obesity is an independent predictor of health care services use after controlling for covariates and potential confounders, including the presence of chronic diseases.

Methods

Study overview

This is a population-based study of adults ages 20 to 64 years living in NL, Canada. Individual-level data from the 2000/2001 CCHS were linked to health insurance physician claims or Medical Care Plan (MCP) and to the hospital separation database or the Clinical Database Management System (CDMS) to obtain information on actual physician and hospital use respectively, over the 5-year period from 1998 to 2002. This study had 2 parts: (1) a secondary analysis of the NL component of the 2000/2001 CCHS cross-sectional health survey, and (2) a data linkage of records from the 2001 CCHS respondents to the 2 above-mentioned health administrative databases to obtain actual health care use data. Information collected from the 2001 CCHS included sociodemographics, lifestyle and health status variables (eg, age, sex, level of income and education, height, weight, presence of chronic diseases), as well as many other covariates. The MCP database provided information on physician visits and the CDMS provided information on hospital use.

Study population/sample

The study sample included survey responses from those individuals who participated in the NL portion of Statistics Canada's 2000/2001 CCHS. Inclusion criteria for the study were: adults between the ages 20 and 64 years, agreement to share survey responses with health information agencies, permission to link individual survey responses with other provincial health information databases, and a valid height and weight measurement. Survey respondents were excluded from the study sample if they were outside the relevant study age range (n = 1075), pregnant (n = 31), had any missing or invalid data on height or weight (n = 25), or were classified as underweight (BMI ≤18.5 [n = 21]). The final eligible sample size for the data linkage was 2345.

Data sources

The 2001 Canadian Community Health Survey

The 2000/2001 CCHS was conducted over a 12-month period by Statistics Canada to provide cross-sectional estimates of health determinants, health status, and health system use for Canadians. The survey uses a multistage cluster sampling design to collect information related to the health of noninstitutionalized individuals and includes household residents 12 years of age and older in all provinces and territories. Indian Reservations, Canadian Forces Bases, and some remote areas were not included. The survey results provide a “snapshot” of the health status of the residents in a province or territory. The CCHS consisted of a 45-minute survey administered either face-to-face or by telephone. The total sample size was sufficient to provide reliable cross-sectional estimates at the provincial and subprovincial (ie, health region level). In 2000/2001 the CCHS survey was administered to 131, 535 Canadians (response rate 84.7%). In the current study, analysis is limited to the survey responses from those participants in the province of NL who meet the above-mentioned inclusion criteria. The survey questionnaire and other related information are available on the Statistics Canada Web site.^24,25,26

Definition of obesity status

Self-reported height (m) and weight (kg) were obtained from the CCHS. BMI was determined as kg/m² and was classified into 4 groups using the system endorsed by both Health Canada and the World Health Organization: normal weight (BMI 18.5–24.9), overweight (BMI 25–29.9), obese class I (BMI 30–34.9), obese class II (BMI ≥35), and obese class III (BMI ≥40). Both the overweight and all the obese categories are associated with increased health risk.^27,28 For the purposes of the current study, obese class II (BMI ≥35) and obese class III (BMI ≥40) were combined and are referred to as obese class II (BMI ≥35).

Covariates

Information on relevant covariates was obtained from the CCHS and included sex, age, region of residence (urban versus rural), level of income (based on household income and family size), level of education (<secondary, secondary, trade school, university graduate), marital status, smoking status (daily/former, never smoked), alcohol consumption (drinker, nondrinker), level of physical activity (active/moderate versus inactive), and consumption of fruits and vegetables (<or ≥5 servings/day). In addition, information was collected on self-perceived health (excellent, very good, or good versus fair/poor), health utility index, the number of disability days reported in the previous 2 weeks, and self-reported chronic diseases. These variables were chosen because of their association with health care use.^29,30

Newfoundland and Labrador Medical Care Plan Provider's Database

The NL MCP is a comprehensive system of public medical care insurance that covers the cost of fee-for-service (FFS) physicians for eligible residents living in NL. The MCP physician claims file used for this study contains information on each FFS medical transaction that takes place between a patient and a physician. Patient information collected includes sex, age, diagnosis code, physician's specialty (eg, general practitioner, other specialty), date of service, fee code, and associated fee. The fee code describes all potential fee grouping categories: office consultations, home consultations, inpatient consultations, outpatient and emergency visits, diagnostic and therapeutic procedures, and inhospital diagnostic, radiology, and surgical procedures. In the current study, only visits for which the physician was coded as the primary provider (excluding codes for surgical assistant and anesthetist) were included. With expertise in database linkages, the NL Centre for Health Information (JCK) generated a data file that contained all physician visits for the 2000/2001 CCHS respondents for the years 1998–2002 (±2.5 years from survey date) and removed all personal identifiers before providing the file to the research team.

Provincial hospital discharge database

All provincially administered health care facilities in NL submit hospital discharge data to the Canadian Institute for Health Information (CIHI). The data are reviewed for data quality and specific values are calculated and added to the database such as resource-intensity-weights (RIWs) for inpatient admissions and day procedure group (DPG) weights for surgical day care cases before the discharge abstract database is provided to the NL provincial Department of Health and Community Services, where it is housed. The RIW measures resources used by inpatients and the DPG measures resources used by the surgical day care patients. Both are calculated using a methodology developed by the CIHI.³¹ These data are provided to the NL Centre for Health Information where the quality of the information is verified and entered into the provincial hospital discharge database. The provincial hospital discharge database contains all discharge diagnostic codes based on the International Classification of Diseases. The CDMS contains demographic, clinical, and procedural data on all inpatient and surgical day care patients for the province. For the current study, the NL Centre for Health Information generated a data file that contained all hospital visits (ie, inpatient admissions, length of stay, RIWs, surgical day care procedures) for the CCHS respondents for the years 1998–2002 (±2.5 years from survey date).

Data linkage

A total of 3734 individuals participated in the 2000/2001 CCHS, all of whom consented to share their survey responses with health information agencies and 94% of whom (n = 3497) consented to link their individual survey responses with other provincial health information databases. Eligible survey respondents who met the study inclusion criteria were linked via a unique identifier (ie, health insurance/MCP number) to the 2 health administration databases to obtain information on health care use. The NL Centre for Health Information used a link file provided to them by Statistics Canada for the current study and conducted the data linkage exercise. In total, 2345 survey respondents were linked to both the MCP and CDMS databases for the period 1998–2002. The linkage was successful for 2177 survey respondents. The remaining 220 provided invalid MCP numbers, which prevented linkage for these individuals.

Statistical analysis

Analyses were conducted using SPSS 13.5 for Windows (IBM, Chicago, IL) and the R software program. Sample weights were used for the multivariate analysis in accordance with Statistics Canada recommendations.³² All analyses were separated by category of BMI, as described earlier.

To compare the characteristics of the study sample, descriptive statistics were run that included: analysis of variance, chi-square, and median tests. Given the nature of the outcome data (discrete count data that was non-normally distributed), multivariate Poisson regression analysis was conducted to determine whether obesity was an independent predictor of physician or hospital utilization after controlling for covariates, including the number of chronic diseases.

Two multivariate models were run for each outcome variable. Model 1 included BMI category and other covariates, and Model 2 included BMI category, other covariates, and the number of chronic conditions. The number of chronic conditions was representative of general morbidity of the sample. Both models were run to determine whether obesity was an independent predictor of health care use and what effect, if any, including the number of chronic conditions had on the relationship between obesity and health care use. For example, did the existence of chronic conditions mediate the relationship between obesity and hospital use? Many researchers do not control for chronic diseases and argue that they are on the pathway between obesity and health services use. However, we were interested in whether obesity had an independent relationship with health care use and what effect, if any, the number of chronic diseases had on this relationship.

Multivariate analyses were conducted to examine the relationship between the level of BMI and actual health care use (ie, visits to a GP, visits to a specialist, number of inpatient admissions, total number of nights spent in hospital, number of surgical day care procedures). Models were adjusted for relevant covariates, which included age, sex, level of income, physical activity level, level of education, fruit and vegetable consumption, self-perceived health, health utility index, marital status, and region of residence. For each model, the parameter estimates, beta (β), standard errors (SE), and relevant level of significance were reported.

This study was approved by the Human Investigation Committee at Memorial University of Newfoundland.

Results

Characteristics of study sample

An edited version of the baseline characteristics of the study sample are presented in Table 1. A more comprehensive table was published previously.²³ A total of 2345 survey respondents between 20 and 64 years of age were included in the study: 52.8% of the sample was female. The average age of the sample was 41 years (SD = 11.67). The average BMI was 27.0 kg/m² with a range between 18.0–60.4 kg/m². The sample was separated by category of BMI. A total of 548 (23%) survey respondents were classified as obese (BMI ≥ 30kg/m²). The obese group was further broken down into 407 (17%) obese class I (BMI ≥30 kg/m² and < 35 kg/m²) and 141 (6%) obese class II (BMI ≥ 35 kg/m²). A total of 916 (39%) survey respondents were classified as overweight (BMI scores between 25.0 and 29.9 kg/m²), and 881 (37%) survey respondents were classified as normal weight (BMI 18.5 kg/m²–24.9 kg/m²). There were significant differences across many of the demographic variables with the exception of level of education and smoking. Adults with a BMI ≥30 were more likely to report having ≥4 chronic conditions.

Table 1.

Select Baseline Characteristics of Sample by Body Mass Index Category (n = 2345)

	Obese Class I		Obese Class II		Overweight		Normal
	(n = 141)		(n = 407)		(n = 916)		(n = 881)
Characteristic	n	(%)	n	(%)	n	(%)	n	(%)
Age in years^***
20–29	25	(18)	53	(13)	103	(11)	171	(20)
30–39	33	(23)	100	(25)	234	(26)	237	(27)
40–49	43	(31)	120	(30)	252	(28)	211	(24)
50–64	40	(28)	134	(33)	327	(36)	262	(30)
Sex^***
Male	48	(34)	204	(50)	505	(55)	353	(40)
Female	93	(66)	203	(50)	411	(45)	528	(60)
Education
<Secondary	25	(18)	72	(18)	148	(16)	145	(17)
Secondary	33	(24)	87	(22)	174	(19)	168	(19)
Trade school	69	(50)	200	(50)	437	(48)	415	(47)
University graduate	12	(9)	43	(11)	149	(16)	148	(17)
Income^***
Low	7	(5)	28	(7)	42	(5)	75	(9)
Middle	120	(88)	301	(77)	666	(75)	617	(75)
High	9	(7)	64	(16)	175	(20)	128	(16)
Smoking
Daily/former	80	(57)	240	(59)	556	(61)	506	(57)
Never smoked	33	(23)	99	(24)	212	(23)	213	(38)
Self-rated health^**
Excellent/very good/good	111	(79)	355	(87)	816	(89)	769	(87)
Fair/poor	30	(21)	52	(13)	100	(11)	112	(13)
Health Utility Index
n, mean (SD)^**	139	0.84(.23)	404	0.89(.19)	908	0.90(.17)	873	0.89(19)
# of disability days
n, mean (SD)	141	1.5(3.8)	407	1.0(3.1)	916	0.77(2.6)	881	0.93(2.9)
Presence of a chronic condition^**	107	(76)	260	(64)	545	(60)	530	(60)
# chronic conditions^***
0	34	(24)	147	(36)	371	(41)	349	(40)
1–3	93	(66)	208	(51)	480	(53)	465	(53)
≥4	14	(10)	52	(13)	64	(7)	67	(7)

P < 0.05, ^** P < 0.01, ^*** P < 0.001.

Physician utilization

Over the 5-year study period 1998–2002, 94% (n = 2036) of the survey respondents had at least 1 visit with a physician. Of these, 84% (n = 1838) had at least 1 visit with a GP and 85% (n = 1858) had either at least 1 visit with a specialist doctor or a radiology assessment. The median number of visits to a physician over the 5-year period and by category of BMI is presented in Table 2. Individuals classified as obese class II (BMI ≥35) reported a significantly higher median number of visits (median 31) to any physician compared to other BMI categories over the 5-year period (P < 0.05). When physician consults were separated into GP or specialist visits, a significant difference remained between the obese class II group compared with all others and the number of visits with a GP. The obese class II group reported a median number of 22 visits to a GP, a third more when compared with the other BMI categories (P < 0.05). Although there were no significant differences among BMI categories in number of visits to a specialist, the obese class II group did show a trend toward a higher median number of visits to a specialist (n = 9) compared with the other BMI categories.

Table 2.

Physician Health Services Utilization by Level of BMI

BMI category ¹	Normal	Overweight	Obese Class I	Obese Class II
Total physician consults
(n = 2036)	817	841	388	131
n (%)	766(94)	783(93)	360(93)	127(97)
median [25^th, 75^th %]^*	24[10,48]	21[8,45]	24[10,45]	31[14,52]
GP consults
n (%)	717 (88)	695(83)	312(80)	114(87)
median [25^th, 75^th %]^*	17[7,28]	16[6,28]	17[7,29]	22 [10,37]
Specialist consults
n (%)	693(85)	713(85)	331(85)	121(92)
median [25^th, 75^th %]	7[4,20]	6[4,19]	7[4,22]	9[5,21]

Sample size varies based on number of valid Medical Care Plan numbers in each BMI category.

BMI categories: normal (18.5–24.9); overweight (25–29.9); obese (30–34.9); morbidly obese (≥35.0).

^*P < 0.05.

BMI, body mass index; GP, general practitioner.

Hospital utilization

Over the 5-year period, 25% (n = 547) of the survey respondents had at least 1 overnight stay in hospital and 25% (n = 551) had at least 1 surgical day care procedure. The median number of inpatient visits, LOS, and surgical day care cases by category of BMI are presented in Table 3. There were no differences across BMI categories for any type of hospital resource use: number of inpatient admissions, LOS per episode, average RIW or DPG. These measures give an indication of the resources used by inpatients and surgical day care patients as they are a reflection of the severity and complexity of an individual's health status. There were no differences across levels of BMI for the median number of day surgeries performed for respondents who had surgical day care.

Table 3.

Actual Hospital Health Services Use by Level of BMI in Adults Aged 20–64 yrs

BMI category¹	Obese Class II	Obese Class I	Overweight	Normal
Acute hospitalizations
N = 547	53	102	191	201
mean (SD) episode/patient	2(2)	2(1)	2(1)	2(2)
median [25^th,75^th %] episode/patient	1.0 [1,2]	1.0 [1,2]	1.0 [1,2]	1.0 [1,2]
median LOS per episode [25^th, 75^th %]	4[1,10]	3[1,7]	3[1,7]	3[1,7]
mean RIW ± SD	1.60(1.62)	1.43(1.61)	1.93(3.26)	2.44(5.08)
Surgical day care cases
N = 551	38	98	214	201
median [25^th, 75^th %]	1[1,2]	1[1,2]	1[1,2]	1[1,2]
mean DPG_w ± SD²	0.26(.30)	0.78(1.52)	0.61(1.29)	0.55(.930)

BMI categories: normal (18.5–24.9); overweight (25–29.9); obese class I (30–34.9); obese class II (≥35.0).

Data available from February 2002 only.

BMI, body mass index; DPG, day procedure group; LOS, length of stay; RIW, resource intensity weight.

Multivariate analyses

Table 4 presents the findings from Model 1 and Model 2 of the multivariate analysis. In Model 1, obesity class II was predictive of visits to a GP (P < 0.001), but not predictive of any other type of health care service use. In Model 2, after controlling for covariates and number of chronic conditions, obesity class II remained predictive of visits to a GP but the level of significance was reduced. The number of chronic conditions, a significant predictor of visits to a GP, partly mediated the relationship between obese class II and GP visits as shown by the decrease in the β value from 0.4269 to 0.2935 in Table 4. No class of obesity was predictive of increased visits to a specialist or any type of hospital service. For both models and in all cases, the number of chronic conditions was a significant predictor for all health care services use (P < 0.001).

Table 4.

Poisson Regression Estimates (β) and Standard Errors (SE): Dependent Variable: Health Care Use

	Physician Utilization				Hospital Utilization
	General Practitioner		Specialist		Inpatient admission		Length of Stay		Surgical Day Care
Dependent Variable/Independent Variable	β	SE	β	SE	β	SE	β	SE	β	SE
Model 1
BMI
Overweight	.0623	.0569	.0942	.0685	−.0258	.1020	−.0203	.1124	.2603^*	.1054
Obese class I	−.0036	.0691	−.0211	.0835	−.1916	.1272	−.6299^***	.1546	.2286	.1238
Obese class II	.4269^***	.0871	.1857	.1161	.2573	.1542	−.2053	.1900	.0019	.2016
Model 2
BMI
Overweight	.0467	.0531	.0839	.0643	−.0441	.0992	−.1830	.1090	.2611^*	.1036
Obese class I	−.0580	.0648	−.0975	.0788	−.2353	.1241	−.6499^***	.1496	.1800	.1221
Obese class II	.2935^*	.0820	.0059	.1107	.1535	.1509	−.2803	.1851	−.1508	.1991
# of chronic conditions	2065^***	.0125	.2313^***	.0144	.2379^***	.0227	.3059^***	.0243	.1977^***	.0238

Referent category normal weight BMI (18.5–24.9), overweight BMI (25–29.9), obese class II (30–34.9), obese class III BMI (≥35).

Model 1: controlled for age, sex, marital status, health region of residence, level of education, level of income, disability days, self-perceived health, health utility index, smoking status, drinking behavior, consumption of fruits and vegetables, level of physical activity.

Model 2: controlled for all variables in Model 1 plus the number of the chronic conditions.

***

P < 0.001, ^** P < 0.01, ^* P < 0.05.

BMI, body mass index.

Discussion

The main goal of this study was to investigate whether obese adults aged 20–64 years living in NL, Canada, used more health care services over a 5-year period compared with normal weight individuals, and to examine whether obesity was an independent predictor of health care services use. According to our results, only obese adults with a BMI ≥35 visited their GP more often than any other BMI category; having a BMI ≥35 remained a significant independent predictor of GP visits after controlling for covariates and the presence of chronic diseases. Obesity was not a significant predictor of visits to a specialist doctor or of any type of hospital service use. Our findings are consistent with some studies in relation to GP visits, and inconsistent with others in relation to specialist visits and hospitalizations.

Several studies have used either self-reported or actual health care services use data and reported a significant relationship between obesity and an increased number of visits to a GP.^{17,20,21,33

–37} Some of these studies controlled for the presence of chronic diseases,^20,21,36 while others did not^{17,33
–35,37}; most did not separate out categories of obesity. Therefore, it is difficult to conclude from these studies if it is obesity per se or a specific level of obesity that demonstrated an independent effect on the use of health care services, or if it was an association with the existence of chronic diseases that drives health care use. In one of these studies, 506 new primary care patients at the UC Davis Medical Center were randomly assigned to 105 primary care resident physicians and their subsequent health care use patterns were compared.²⁰ Information on sociodemographics, height, and weight was collected, including the use of health care services during a prospective 12-month period. The authors reported that obese patients had a higher mean number of visits to both primary care and specialty care clinics, and a higher mean number of diagnostic services ordered compared with nonobese individuals (BMI < 30). After controlling for health status, depression, age, education, income, sex, and physical health, being obese remained significantly associated with the use of primary care and diagnostic services. The authors did not adjust for obesity-associated comorbidities such as hypertension or type II diabetes; they argued that statistically adjusting for comorbidities was inappropriate because many of these conditions are intermediary steps along the causal pathway between increased BMI and health services use. Consequently, including them in a statistical model constitutes overadjustment. In another study of the German population, authors analyzed responses to the KORA-survey 1999/2001, a health survey administered to an adult population aged 25 to 74 years in the Augsburg region of Germany (n = 947). After controlling for sex, age, place of residence, social class, and type of insurance/sickness fund, the findings were consistent with the current study in that only obese individuals with a BMI ≥35 were associated with more frequent use of GP services compared with normal weight individuals. However, chronic diseases were not adjusted for in this study.³⁶

Studies stratified by sex have yielded similar results.^15,38,39 In a secondary analysis of the Australian National Health Survey, survey responses were available for 17,033 men and 17,174 women older than 20 years of age on health-related issues. After controlling for age and income, obese men and women were more likely to have made a visit to a primary care doctor and only obese women were more likely to have made a visit to a specialist. In another study, data from the Health and Retirement Study, a nationwide biennial longitudinal survey representative of American adults between 50 and 69 years of age, were examined in order to estimate the effect of weight class on health care use. This study demonstrated a dose-response relationship between weight class measured by self-reported BMI (overweight, moderate obesity, and severe obesity) and increases in outpatient health care services for women only. The authors controlled for sociodemographic factors (eg, age, race, insurance status, marital status, education, family income, region) and health risk behaviors (eg, current tobacco smoking, heavy alcohol drinking).³⁸ In another study on an older population, authors examined data from a prospective cohort study (2001–2003) of noninstitutionalized individuals aged ≥60 years and older who were representative of the Spanish population. After controlling for age, education, place of residence, tobacco use, alcohol consumption, and presence of chronic disease, the authors found that, for women only, obesity based on self-reported height and weight was associated with a greater number of visits to a primary care physician.³⁹

Very few studies have examined the impact of obesity on specialist services, and a review of these studies suggests the relationship is inconsistent.^20,39 In addition, the relationship between BMI and increased use of hospital services is also contradictory. Although some authors reported an insignificant relationship between obesity and increased use of hospital services,^17,18,21 other reported that obesity increases admissions and LOS.^22,33,34,39 It may be that hospital admissions are more reflective of level of severity (eg, more acute) and level of access than the general morbidity level of a population.⁴⁰ However, in NL as in Canada, access to physician and hospital services is universal and free at the point of use and, therefore, should not be a deterrent for the current population. Another explanation for the inconsistent results is that, in the general population, hospitalization is a rare event and study populations must be large enough to detect differences in relatively uncommon outcomes. As a result, larger sample sizes are required with longer periods of follow-up in order to detect differences between groups. A review of the literature seems to support this explanation as the studies that reported a positive relationship between obesity and use of hospital services more often included larger sample sizes, an elderly population, and long periods of follow-up.^22,33,34,39

There are a number of strengths associated with this study. The study sample was randomly selected and population-based, therefore representative of adults living in NL. The use of the CCHS health survey provided a rich data set of variables that could be controlled for in the multivariate analysis. Variables such as age, sex, smoking, and the presence of chronic diseases are known to impact the use of health care services and were controlled for in the multivariate analysis. Therefore, the impact of obesity on actual health care use was explored independently. In addition, individual-level data were linked to health care administrative databases to obtain data on actual health care use over a 5-year period. There are some methodological limitations associated with this study. BMI was calculated using self-reported height and weight, which may result in some misclassification of BMI. Generally, it is agreed that this is a systematic bias and may result in conservative study findings. The power of the study may have been low to determine differences in hospital admissions or specialist visits as these outcomes were relatively rare in the general population.

Conclusion

This is the first study that we are aware of that used Canada data to conduct an adjusted multivariate individual-level analysis to examine the relationship between obesity and health care services use. We found that excess body weight measured by BMI was predictive of health care use at the primary care level, mainly through its association with chronic diseases, providing empirical evidence for targeting services at this population and at this level of care. The increasing prevalence of adults in excessive weight categories in Canada is cause for concern because of the increased likelihood of associated metabolic and musculoskeletal complications and, consequently, its impact on the health care system.

Footnotes

Author Disclosure Statement

Drs. Twells, Bridger, Knight, Alaghehbandan, and Barrett disclosed no conflicts of interest.

Funding was received from The Newfoundland and Labrador Centre for Applied Health Research in the form of a project grant. In-kind support was received from the Newfoundland and Labrador Centre for Health Information.

References

Raine

. Overweight and obesity in Canada. http://secure.cihi.ca/cihiweb/products/CPHIOverweightandObesityAugust2004_e.pdf. 2009 October 13.

Statistics Canada. Table 105-4009: Body mass index (BMI) by sex, household population aged 18 and older excluding pregnant females, Canada, provinces and territories, occasional, CANSIM (database) http://cansim2.statcan.gc.ca/cgi-win/cnsmcgi.exe?Lang=E&CNSM-Fi=CII/CII_1-eng.htm. 2009 October 14.

Katzmarzyk

. The Canadian obesity epidemic 1985–1998. CMAJ, 2002; 166:1039–1040.

Health Statistics Division, Health Reports, Statistics Canada. 1994/1995 National Population Health Survey. http://www.statcan.gc.ca/dli-ild/data-donnees/ftp/nphs-ensp/nphs1994-ensp1994-eng.htm. 2011 March 21.

Health Statistics Division, Health Reports, Statistics Canada. Health Indicators: The Canadian Community Health Survey 2005. Ottawa, on: Health Statistics Division, Health Reports, Statistics Canada, 2005.

Canning

, Courage

, Frizell

. Prevalence of overweight and obesity in a provincial population of Canadian preschool children. CMAJ, 2004; 171:240–242.

Tjepkema

, Shields

. Measured Obesity: Overweight Canadian Children and AdolescentsCatalogue No. 82-620-XWE20055001 ed.Ottawa, ON: Statistics Canada, 2005.

World Health Organization. Obesity: Preventing and Managing the Global Epidemic. WHO Technical Report Series No. 894. Geneva: World Health Organization, 2000.

Fontaine

, Redden

, Wang

, Westfall

, Allison

. Years of life lost due to obesity. JAMA, 2003; 289:187–193.

10.

Calle

, Thun

, Petrelli

, Rodríguez

, Heath

Jr.

Body mass index and mortality in a prospective cohort of U.S. adults. N Engl J Med, 1999; 341:1097–1105.

11.

Anis

, Zhang

, Bansback

, Guh

, Amarsi

, Birmingham

. Obesity and overweight in Canada: An updated cost-of-illness study. Obes Rev, 2010; 11:31–40.

12.

Katmarzyk

, Janssen

. The economic costs associated with physical inactivity and obesity in Canada: An update. Can J Appl Physiol, 2004; 29:90–115.

13.

Seidell

, Deerenberg

. Obesity in Europe: prevalence and consequences for the use of medical care. Pharmacoeconomics, 1994; 5:38–44.

14.

Sansone

, Sansone

, Wiederman

. The relationship between obesity and medical utilization among women in a primary care setting. Int J Eat Disord, 1998; 23:161–167.

15.

Reidpath

, Crawford

, Tilgner

, Gibbons

. Relationship between body mass index and the use of healthcare services in Australia. Obes Res, 2002; 10:526–531.

16.

Guallar-Castillon

, Lopez Garcia

, Lozano Palacios

et al. The relationship of overweight and obesity with subjective health and use of health-care services among Spanish Women. Int J Obes Relat Metab Disord, 2002; 26:247–252.

17.

Luchsinger

, Lee

, Carrasquillo

, Rabinowitz

, Shea

. Body mass index and hospitalization in the elderly. J Am Geriatr Soc, 2003; 51:1615–1620.

18.

Popoola

. Increase in obesity and health-care use, from the Health Survey for England, Ages 50–69,1998. Obes Surg, 2004; 14:1258–1262.

19.

Zizza

, Herring

, Stevens

, Popkin

. Length of hospital stay among obese individuals. Am J Public Health, 2004; 94:1587–1591.

20.

Bertakis

, Azari

. Obesity and the use of health care services. Obes Res, 2005; 13:372–379.

21.

Trakas

, Lawrence

, Shear

. Utilization of health care resources by obese Canadians. CMAJ, 1999; 160:1457–1462.

22.

Chen

, Jiang

, Mao

. Hospital admissions associated with body mass index in Canadian adults. Int J Obes, 2007; 31:962–967.

23.

Twells

, Knight

, Alaghehbandan

. The relationship among body mass index, subjective reporting of chronic disease, and the use of health care services in Newfoundland and Labrador, Canada. Popul Health Manag, 2010; 13:47–53.

24.

Statistics Canada. Canadian Community Health Survey 2001. http://www.statcan.gc.ca/concepts/health-sante/index-eng.htm. 2011 March 28.

25.

Statistics Canada. Population health surveys: Canadian Community Health Survey 2001. http://www.statcan.ca/english/concepts/hs/index.htm. 2009 October 14.

26.

Statistics Canada. Health indicators: Canadian Community Health Survey 2001. http://www.statcan.ca/english/freepub/82-221-XIE/free.htm. 2009 October 14.

27.

World Health Organization; International Obesity Task Force. Obesity. Preventing and Managing the Global Epidemic. Geneva: World Health Organization, 1998.

28.

Adams

, Schatzkin

, Harris

et al. Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old. N Engl J Med, 2006; 355:763–778.

29.

Andersen

. Revisiting the behavioral model and access to medical care: Does it matter? J Health Soc Behav, 1995; 36:1–10.

30.

Andersen

, Harada

, Chiu

, Makinodan

. Application of the behavioral model to health studies of Asian and Pacific Islander Americans. Asian Am Pac Isl J Health, 1995; 3:129–141.

31.

Canadian Institute for Health Information. Patient cost estimator—methodological notes and glossary. http://secure.cihi.ca/cihiweb/en/downloads/PCE_methnotes_e.pdf. 2011 March 28.

32.

Ramsey

, Schafer

. The Statistical Sleuth: A Course in Methods of Data Analysis, 2 ^nd. New York: Duxbury, 2002.

33.

Quesenbury

Jr , Caan

, Jacobson

. Obesity, health services use, and health care costs among members of a health maintenance organization. Arch Intern Med, 1998; 158:466–472.

34.

Raebel

, Malone

, Conner

, Xu

, Porter

, Lanty

. Health services use and health care costs of obese and non obese individuals. Arch Inter Med, 2004; 164:2135–2140.

35.

Ostbye

, Taylor

, Jung

. A longitudinal study of the effects of tobacco, smoking and other modifiable risk factors on ill-health in middle-aged and old Americans: Results from the health and retirement study and asset and health dynamics among the oldest old survey. Prev Med, 2002; 34:334–345.

36.

von Lengerke

, Happich

, Reitmeir

, John

. Utilization of out- and inpatient health services by obese adults: A population based study in the Augsburg region, Germany. Gesundheitswesen, 2005; 67:S150–S157.

37.

van Dijk

, Otters

, Schuit

. Moderately overweight and obese patients in general practice: A population based survey. BMC Fam Pract, 2006; 7:43.

38.

Andreyeva

, Sturm

, Ringel

. Moderate and severe obesity have large differences in health care costs. Obes Res, 2004; 12:1936–1943.

39.

Leon-Munoz

, Guallar-Castillon

, Lopez Garcia

, Banegas

, Gutierrez-Fisac

, Rodriguez-Artalejo

. Relationship of BMI, waist circumference, and weight change with use of health services by older adults. Obes Res, 2005; 13:1398–1404.

40.

Payne

, Coy

, Patterson

, Milner

. Is use of hospital services a proxy for morbidity? A small area comparison of the prevalence of arthritis, depression, dyspepsia, obesity and respiratory disease with inpatient admission rates for these disorders in England. J Epidemiol Community Health, 1994; 48:74–78.