Relationship of BMI, Waist Circumference, and Weight Change with Use of Health Services by Older Adults
The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Departamento de Medicina Preventiva y Salud Pública, Facultad de Medicina, Universidad Autónoma de Madrid, Avda. Arzobispo Morcillo, sn, 28029 Madrid, Spain. E-mail: email@example.com
Objective: To examine the relationship of BMI, waist circumference (WC), and weight change with use of health care services by older adults.
Research Methods and Procedures: This was a prospective cohort study conducted from 2001 to 2003 among 2919 persons representative of the non-institutionalized Spanish population ≥60 years of age. Analyses were performed using logistic regression, with adjustment for age, educational level, size of place of residence, tobacco use, alcohol consumption, and presence of chronic disease.
Results: Obesity (BMI ≥ 30 kg/m2) and abdominal obesity (WC >102 cm in men and >88 cm in women) in 2001 were associated with greater use of certain health care services among men and women in the period 2001–2003. Compared with women with WC ≤ 88 cm, women with abdominal obesity were more likely to visit primary care physicians [odds ratio (OR): 1.36; 95% confidence limit (CL): 1.06–1.73] and receive influenza vaccination (OR: 1.30; 95% CL: 1.03–1.63). Weight gain was not associated with greater health service use by either sex, regardless of baseline BMI. Weight loss was associated with greater health service use by obese and non-obese subjects of both sexes. In comparison with those who reported no important weight change, non-obese women who lost weight were more likely to visit hospital specialists (OR: 1.45; 95% CL: 1.02–2.06), receive home medical visits (OR: 1.61; 95% CL: 1.06–2.45), be hospitalized (OR: 1.88; 95% CL: 1.29–2.74), and have more than one hospital admission (OR: 2.31; 95% CL: 1.19–4.47).
Discussion: Obesity and weight loss are associated with greater health service use among the elderly.
Obesity is associated with numerous chronic diseases, including hypertension, diabetes, coronary heart disease, cerebrovascular disease, certain forms of cancer, osteomuscular disorders, and gallbladder disease (1, 2). As a consequence, obesity is also associated with greater health service use (3) and cost (4).
The relationship between excess weight and health might differ between the elderly and the young or middle-aged (5, 6). However, information on the impact of obesity, measured by BMI, on health service use by older adults is very sparse. Previous studies have been cross-sectional (7), which limits causal inference, have focused on a single type of health care service (e.g., hospitalization) (8), or have included only a small segment of the older adult population (e.g., subjects 50 to 69 years of age) (9).
Obesity, and its abdominal distribution in particular, can also be measured by waist circumference (WC),1 which seems to be a better predictor of morbidity and mortality than BMI, especially among the elderly (5, 6). There is only one recent study on the relationship between WC and cost of health care services (10). This study had a cross-sectional design and included persons 18 to 84 years of age, without stratifying the results by age. In addition, it only calculated health care service costs, without detailing the use of the different types of services.
Finally, changes in body weight are frequent among older persons (11, 12, 13). In addition, weight loss, which in many cases is the consequence of an underlying disease, is associated with worse quality of life and higher mortality in this population group (14, 15, 16). To our knowledge, however, no study has yet addressed the impact of weight change on health service use.
Therefore, this study examined prospectively the relationship of BMI, WC, and weight change with health service use among the older adult population. These relationships are important because the impact of excess weight on health care services provides additional information about the social burden of obesity, which is not measured when only the association between obesity and mortality/morbidity is contemplated.
Research Methods and Procedures
Study Design and Subjects
This was a prospective, population-based cohort study. The cohort was established in 2001 and followed up over 2 years, and was comprised of 4008 persons representative of the non-institutionalized Spanish population ≥60 years of age. Study subjects were selected through probabilistic multistage cluster sampling. Census sections were selected at random and stratified by region of residence and size of town, followed by individual households where information was obtained from subjects. Data were collected on a total of 420 census sections in Spain, with subjects being selected in sex and age strata. Subjects were replaced after 10 failed visits by the interviewer or because of subject's incapacity, death, institutionalization, or refusal to participate. There was an overall study response rate of 71%. Data were collected by home-based personal interview and physical examination by trained and certified personnel.
In 2003, an attempt was made to contact the subjects again. Contact was successful in 3235 cases (80.7%). This population did not differ significantly in any sociodemographic or lifestyle-related characteristic from those lost in the follow-up, except for the number of chronic diseases diagnosed and reported in 2001 (1.4 diseases among people followed vs. 1.2 diseases among those lost). In 2003, data were collected through a phone interview conducted by trained staff. We have recent evidence in Spain showing that phone interviews through computer-assisted technology are reliable and also valid against face-to-face interviews at the home of the study participants to measure both lifestyle and health services use (17, 18).
In all cases, informed consent was obtained from subjects or next of kin. The study was approved by the Clinical Research Ethics Committee of the “La Paz” University Teaching Hospital in Madrid.
Use of health care services was assessed in 2003 with the following questions regarding the preceding 2 years. How often did you visit your health center or primary care doctorquest; How often did you visit your hospital specialist? How often did your doctor visit you at home? Did you receive a vaccination against influenza during the last season? Have you ever been admitted to the hospital? How many times have you been admitted to the hospital? How long did you spend in the hospital the last time you were admitted? How many times have you had surgery? Have you ever been admitted to the intensive care unit?
The three main independent variables were BMI in 2001, WC in 2001, and weight change in the period 2001 to 2003. Body weight was measured to the nearest 0.1 kg using a calibrated precision scale (Seca Model 812; Vogel & Halke, Hamburg, Germany), with subjects lightly clothed and shoeless. Stature was measured to the nearest 0.1 cm, using a portable wall-mounted stadiometer (KaWe, Asperg, Germany), with subjects standing in stockinged feet against a wall that had no skirting board. WC was deemed to be located at the midpoint between the lowest ribs and the iliac crest (hip bone) and was measured using a plastic, inelastic, flexible belt-type tape, with subjects lightly clothed. Anthropometric measurements were obtained using standardized procedures (19) and were validated on a random sample of 100 individuals. The intraclass correlation coefficients were 0.97 for weight; 0.92 for height; and 0.89 for WC. BMI was calculated as weight in kilograms divided by the square of the height in meters, and subjects were classified as normal weight (BMI = 18.5 to 24.9 kg/m2), overweight (BMI = 25 to 29.9 kg/m2), or obese (BMI ≥ 30 kg/m2). Abdominal obesity was defined as WC >102 cm in men and >88 cm in women (2). Weight change was ascertained by asking subjects in 2003 whether they had noticed important changes in weight over the preceding 2 years. When the reply was affirmative, a second question was asked—did the change involve weight gain or weight loss?
We also collected baseline information on age (60 to 69, 70 to 79, and ≥80 years), educational level (no formal education, primary, secondary, and higher education), size of place of residence (≤5000, 5001 to 50, 000, 50, 001 to 100, 000, 100, 001 to 500, 000, and >500, 000 inhabitants), tobacco use (never smoker, ex-smoker, and smoker), and alcohol consumption (non-drinker, past drinker, moderate drinker, and heavy drinker). The limit between moderate and excessive alcohol consumption was 30 g/d in men and 20 g/d in women. Finally, information was gathered on chronic diseases diagnosed by the physician and reported by patients in 2001, as well as any new diseases diagnosed in the period 2001 to 2003. The diseases considered were osteoarthritis, hypertension, chronic obstructive pulmonary disease, ischemic heart disease, cerebrovascular disease, cancer (any site), diabetes, cataracts without treatment, and depression with need for treatment. Previous studies have shown that the degree of agreement between self-reported diseases and medical history is high among the elderly population (20, 21).
A total of 316 subjects were removed from the analysis: 245 who died during the 2 years of follow-up; 53 who did not furnish information on one or more of the health care services studied; 13 with a BMI <18.5 kg/m2 in 2001; and 5 who failed to provide information on the remaining variables of interest. Consequently, the analysis was performed using a total of 2919 subjects.
The relationship of BMI, WC, and weight change with health service use was summarized using odds ratios (ORs) obtained from logistic regression. These associations were adjusted for age, educational level, size of place of residence, tobacco use, and alcohol consumption. To ascertain whether obesity-related diseases might explain the relationship between obesity and health service use, models were subsequently adjusted for chronic diseases reported in 2001. The association between weight change and health service use was stratified according to obesity in 2001 (BMI ≥ 30 kg/m2) and was adjusted for the above-mentioned variables and for any new diseases reported in the period 2001 to 2003. All variables were modeled using dummies. Health care service variables with several categories of frequency (e.g., number of visits to general practitioners, number of visits to hospital specialists) were dichotomized using a category close to the median as a cut-off. The number of chronic diseases was modeled using three dummies for the following four categories (0, 1, 2, and ≥3 diseases).
Separate analyses were performed for men and women, using the SAS software, version 8.2 (22).
The mean age of subjects at baseline was 70.7 years (range, 60 to 93 years) for men and 72.0 years (range, 60 to 98 years) for women. A total of 53.0% of men and 42.5% of women were overweight, and 29.5% of men and 41.5% of women were obese. Abdominal obesity affected 47.1% of men and 73.2% of women. During the 2-year follow-up, 16.1% of men reported weight gain and 11.0% reported weight loss vs. 16.9% and 19.0% of women, respectively.
Visiting the primary care physician was the most used service, with 69.1% of men and 72.9% of women reporting a visit at least once every 2 to 3 months. Other widely used services were influenza vaccination (64.8% in men and 61.3% in women); visits to hospital specialists at least twice per year (41.1% in men and 40.4% in women); emergency services at least once per year (32.8% in men and 39.8% in women); hospital admission at least once per year (27.9% in men and 25.0% in women); and home medical visit at least once per year (14.3% in men and 20.8% in women). In addition, 9.3% of men were admitted to the hospital more than once, with stays >6 days in 17.4% of the admissions, 3.5% underwent more than one operation, and 4.5% were admitted to the intensive care unit. In women, the respective percentages were 8.4%, 15.3%, 3.6%, and 2.1%.
Table 1 shows the relationship among BMI, WC, and health service use, adjusted for the main confounders. Overweight men (BMI = 25 to 29.9 kg/m2) were less likely to receive home medical visits than normal weight subjects. However, obese men (BMI ≥ 30 kg/m2) were more likely to be hospitalized for >6 days and to be admitted to the intensive care unit. Furthermore, men with abdominal obesity were more likely to visit the primary care physician and to be hospitalized for >6 days. In women, not only was obesity positively associated with visits to primary care physicians, but, in addition, a greater proportion of women with abdominal obesity also visited primary care physicians and received an influenza vaccination. While adjustment for chronic diseases somewhat reduced the magnitude of the OR in both sexes, the associations remained statistically significant (data not shown), with the following exceptions. In women, the significance of the association between obesity and visits to primary care physicians was lost [OR: 1.18; 95% confidence limit (CL): 0.84 to 1.64], as well as between abdominal obesity and this same service (OR: 1.18; 95% CL: 0.92 to 1.52) and influenza vaccination (OR: 1.21; 95% CL: 0.96 to 1.52); however, adjustment for chronic diseases showed that a lower proportion of obese women underwent more than one operation (OR: 0.49; 95% CL: 0.26 to 0.96).
Table 1. . OR of use of health care services by BMI and abdominal obesity in men and women
|Men||(n = 219)||(n = 664)||(n = 369)||(n = 662)||(n = 590)|
| Visit to primary care physician||1||0.91 (0.65 to 1.28)||1.35 (0.92 to 1.97)||1||1.28 (1.02 to 1.62)*|
| Hospital specialist||1||0.86 (0.63 to 1.18)||0.93 (0.66 to 1.32)||1||1.03 (0.82 to 1.30)|
| Home medical visits||1||0.55 (0.35 to 0.85)†||0.98 (0.61 to 1.57)||1||1.10 (0.78 to 1.55)|
| Emergency services||1||0.74 (0.53 to 1.03)||1.10 (0.77 to 1.57)||1||1.14 (0.89 to 1.45)|
| Influenza vaccination||1||1.09 (0.78 to 1.51)||1.31 (0.91 to 1.89)||1||1.26 (0.98 to 1.61)|
| Hospital admission||1||1.01 (0.71 to 1.43)||1.33 (0.91 to 1.96)||1||1.05 (0.81 to 1.35)|
| More than one hospital admission||1||1.48 (0.74 to 2.96)||0.87 (0.41 to 1.86)||1||1.17 (0.71 to 1.92)|
| Duration of hospital stay >6 days||1||1.15 (0.62 to 2.16)||2.41 (1.19 to 4.86)*||1||2.17 (1.33 to 3.54)†|
| More than one surgical intervention||1||0.77 (0.31 to 1.93)||0.96 (0.37 to 2.51)||1||1.23 (0.61 to 2.47)|
| Admission to intensive care unit||1||1.65 (0.62 to 4.45)||2.74 (1.01 to 7.64)*||1||1.10 (0.59 to 2.05)|
|Women||(n = 266)||(n = 709)||(n = 692)||(n = 447)||(n = 1220)|
| Visit to primary care physician||1||1.27 (0.93 to 1.74)||1.43 (1.04 to 1.98)*||1||1.36 (1.06 to 1.73)*|
| Hospital specialist||1||1.21 (0.90 to 1.63)||1.17 (0.86 to 1.58)||1||1.05 (0.84 to 1.32)|
| Home medical visits||1||1.08 (0.75 to 1.56)||1.21 (0.83 to 1.76)||1||1.08 (0.81 to 1.44)|
| Emergency services||1||0.90 (0.67 to 1.21)||1.06 (0.79 to 1.43)||1||0.95 (0.76 to 1.19)|
| Influenza vaccination||1||1.10 (0.82 to 1.48)||1.19 (0.88 to 1.61)||1||1.30 (1.03 to 1.63)*|
| Hospital admission||1||1.04 (0.74 to 1.46)||1.31 (0.93 to 1.85)||1||1.20 (0.92 to 1.56)|
| More than one hospital admission||1||1.05 (0.56 to 1.99)||1.03 (0.54 to 1.97)||1||0.84 (0.52 to 1.37)|
| Duration of hospital stay >6 days||1||0.84 (0.45 to 1.59)||0.74 (0.39 to 1.40)||1||0.89 (0.55 to 1.44)|
| More than one surgical intervention||1||0.90 (0.35 to 2.29)||1.00 (0.40 to 2.50)||1||0.56 (0.30 to 1.07)|
| Admission to intensive care unit||1||0.46 (0.16 to 1.36)||0.78 (0.28 to 2.14)||1||1.16 (0.48 to 2.82)|
Table 2 shows the relationship between weight change and health service use adjusted for the main confounders and stratified by BMI. Weight gain was not associated with greater health service use in either sex, regardless of baseline BMI. We only observed that a lower proportion of those who gained weight visited the primary care physician a minimum of once every 2 to 3 months compared with obese men.
Table 2. . OR of use of health care services by weight change, according to baseline BMI in men and women
|Men||(n = 123)||(n = 78)||(n = 97)||(n = 41)|
| Visit to primary care physician||0.92 (0.59 to 1.43)||0.44 (0.23 to 0.82)†||1.00 (0.61 to 1.64)||1.59 (0.51 to 4.98)|
| Hospital specialist||1.00 (0.65 to 1.50)||1.73 (0.99 to 3.04)||2.08 (1.32 to 3.28)†||0.89 (0.43 to 1.86)|
| Home medical visits||0.62 (0.32 to 1.21)||1.73 (0.75 to 3.95)||1.22 (0.68 to 2.20)||3.14 (1.27 to 7.77)*|
| Emergency services||0.95 (0.61 to 1.47)||1.62 (0.91 to 2.86)||1.27 (0.80 to 2.03)||2.40 (1.15 to 5.01)*|
| Influenza vaccination||1.45 (0.93 to 2.25)||1.26 (0.70 to 2.28)||1.10 (0.68 to 1.78)||1.90 (0.80 to 4.55)|
| Hospital admission||0.75 (0.46 to 1.21)||1.73 (0.95 to 3.15)||1.48 (0.92 to 2.38)||3.22 (1.52 to 6.82)†|
| More than one hospital admission||0.69 (0.26 to 1.82)||0.68 (0.16 to 2.88)||2.97 (1.27 to 6.94)*||3.39 (0.87 to 13.13)|
| Duration of hospital stay >6 days||0.80 (0.34 to 1.89)||0.46 (0.11 to 1.95)||1.60 (0.70 to 3.67)||2.34 (0.50 to 10.98)|
| More than one surgical intervention||1.13 (0.30 to 4.34)||NE||2.32 (0.75 to 7.23)||0.85 (0.15 to 4.78)|
| Admission to intensive care unit||1.13 (0.32 to 4.03)||0.37 (0.05 to 2.91)||1.03 (0.34 to 3.13)||16.42 (3.37 to 79.92)‡|
|Women||(n = 143)||(n = 138)||(n = 181)||(n = 136)|
| Visit to primary care physician||1.01 (0.65 to 1.56)||1.56 (0.91 to 2.66)||0.96 (0.65 to 1.42)||0.98 (0.61 to 1.57)|
| Hospital specialist||1.05 (0.71 to 1.55)||1.22 (0.81 to 1.84)||1.45 (1.02 to 2.06)*||1.79 (1.17 to 2.72)†|
| Home medical visits||1.16 (0.69 to 1.96)||0.98 (0.57 to 1.70)||1.61 (1.06 to 2.45)*||1.64 (1.02 to 2.63)*|
| Emergency services||1.36 (0.92 to 2.00)||1.14 (0.76 to 1.72)||1.15 (0.81 to 1.63)||1.44 (0.95 to 2.18)|
| Influenza vaccination||0.87 (0.59 to 1.29)||1.02 (0.67 to 1.56)||0.87 (0.61 to 1.25)||0.98 (0.64 to 1.50)|
| Hospital admission||1.31 (0.84 to 2.05)||1.15 (0.73 to 1.83)||1.88 (1.29 to 2.74)†||1.74 (1.12 to 2.69)*|
| More than one hospital admission||1.92 (0.83 to 4.44)||1.56 (0.65 to 3.76)||2.31 (1.19 to 4.47)*||1.31 (0.59 to 2.88)|
| Duration of hospital stay >6 days||0.54 (0.24 to 1.23)||1.15 (0.50 to 2.64)||0.88 (0.45 to 1.75)||1.37 (0.65 to 2.89)|
| More than one surgical intervention||2.34 (0.80 to 6.81)||2.19 (0.80 to 5.98)||1.20 (0.43 to 3.37)||0.78 (0.26 to 2.37)|
| Admission to intensive care unit||1.30 (0.31 to 5.41)||1.12 (0.33 to 3.88)||0.83 (0.23 to 2.94)||0.65 (0.18 to 2.41)|
In comparison with those who reported no important weight change, non-obese men who lost weight were more likely to visit hospital specialists and to be hospitalized more than once. Among obese men, those who lost weight were more likely to receive home medical visits, make use of emergency services, be hospitalized, and be admitted to the intensive care unit (Table 2).
In comparison with those who reported no important weight change, non-obese women who lost weight were more likely to visit hospital specialists, receive home medical visits, be hospitalized, and have more than one hospital admission. Among obese women, those who lost weight were more likely to visit hospital specialists, receive home medical visits, and be hospitalized (Table 2).
Our results suggest that both obesity and weight loss are associated with a greater use of certain health care services by older adults. These findings are only partially explained by the presence of obesity-related diseases in men or by diseases that cause weight loss in both sexes.
With regard to the relationship between obesity as measured by BMI and health service use, our results are partially consistent with those of previous studies. Quesenberry et al. (7) observed an association between BMI and the annual rates of inpatient days and the number and costs of outpatient visits. The association decreased with age and was explained mainly by obesity-related diseases. Luschinger et al. (8) showed that among individuals 65 to 75 years of age, overweight and obesity were associated with greater risk of hospitalization; however, this relation was not evident in individuals >75 years of age. Sturm et al. (9) observed that, in a population 50 to 69 years of age, the number of visits to the medical practitioner and the percentage of subjects with inpatient stays rose with BMI for both men and women. Finally, a study covering a population-based sample of Spanish women >16 years of age indicated that obese women were more likely to visit medical practitioners and use hospital emergency services. The results did not vary substantially among subjects under and over 55 years of age (23). Similarly, our results are consistent with those of Cornier et al. (10), who reported that abdominal adiposity assessed by WC was associated with increased total health care charges, in particular in-patient charges, in persons 18 to 84 years of age. This consistency with previous studies reinforces the credibility of our results, because they were obtained in health care systems with different access conditions and service organization.
We observed that obese subjects received an influenza vaccination with similar or even greater frequency than non-obese subjects. In addition, our findings do not support previous results that suggested that, among women, a greater BMI tends to be associated with fewer preventive services (24), particularly clinical breast examinations, gynecologic examinations, and Papanicolau smears (25). Our results may be quite surprising at first sight, because they were obtained after adjustment for chronic diseases, which, among middle-aged people, are a major reason for immunization. However, a possible explanation of our finding is that, because obese women visit the primary care physician more frequently, and because an influenza vaccination is recommended for all elderly people in Spain regardless of the presence of chronic diseases, elderly obese women have an increased opportunity to receive the vaccine. It must be emphasized that vaccination should not be considered a negative consequence of obesity.
We also observed that abdominally obese women underwent more than one surgical intervention less often. This would seem a reasonable result, because abdominal adiposity is a relative contraindication of surgery. Finally, our study showed that overweight subjects (BMI = 25 to 29.9 kg/m2) did not make greater use of health care services than did normal weight subjects and, indeed, received even fewer home medical visits. This is in line with the fact that, among the elderly, quality of life and cardiovascular and total mortality are similar for both normal weight and overweight subjects (6, 26, 27).
Insofar as the relationship between weight changes and use of services is concerned, our results point in the same direction as those that show that weight loss in the elderly is associated with worse quality of life and higher mortality (14, 15, 16).
For a correct interpretation of our results, some comments on the advantages and limitations of the study are needed. The strength of this study is that it entailed a population-based cohort, representative of a nationwide adult population. Furthermore, we studied many different types of health care services, and analyses were adjusted for the most important confounding factors. This study also has several limitations. First, the information on weight change was self-reported. In 2001, we observed a good correlation between measured and self-reported weight (Spearman correlation = 0.94; p < 0.001); therefore, it is likely that reported changes correlate reasonably well with real changes in weight. Nevertheless, reported weight change might also reflect individuals’ subjective perception of such change. In addition, rather than assessing the magnitude of the weight change, the questions ascertained only whether the change was “important.” Second, the size of certain strata was small, particularly among obese men who gained or lost weight, which translates to wide confidence intervals for the relationship between weight change and health service use. Third, the information on health service use was self-reported. Although self-reported use of health care services is reliable (28, 29), a certain degree of underestimation of health service use is possible, caused by bias for very long recall periods and hyperusers. However, there is no evidence of differences in recall errors between obese and non-obese persons; therefore, our results are probably conservative. Moreover, we do not know the reasons for the visits to primary care physicians and home medical visits, which, owing to their diversity (issue of prescriptions, therapeutic or preventive services, scheduled or emergency attention, etc.), might influence the results obtained. Nor do we know the reasons for hospital admission or the ensuing diagnosis leading to discharge. Finally, our results on the ability of chronic diseases to explain the observed associations should be interpreted with caution because we have not considered all of the diseases that are potentially important (e.g., heart failure, gall bladder disease), their severity, and their duration.
Our results are of practical relevance. First, they show that obesity represents an important burden for health care services among the elderly, which is the population group that makes the most frequent and intense use of such services. Thus, these results suggest that there is potential for reducing the use of health services through prevention of obesity in middle age. This is especially important in light of the growing frequency of obesity (1, 30) and the important problems involved in funding health care services in most developed countries. Second, our results highlight the fact that maintenance of body weight may be the most favorable option from the standpoint of health service use among the elderly. This is in line with other studies that also show that maintenance of body weight is desirable and serves to enhance the quality of life in this age group (14).
This work was supported by FIS Grant 02/563, an unrestricted educational contract with AstraZeneca in Spain, and ISCIII (RCESP Red 03/09). L.M.L.M. was a recipient of a fellowship from FIS Grant 02/563. E.L.G. was supported by a contract “Juan de la Cierva” from the Ministerio de Educación y Ciencia. Funding bodies had no role in data extraction and analysis, writing of the manuscript, or in the decision to submit the paper for publication.
Nonstandard abbreviations: WC, waist circumference; CL, confidence limit; OR, odds ratio.