Body Mass Index and Quality of Life in a Survey of Primary Care Patients

Obesity is a common condition that is increasing in prevalence. Mokdad and colleagues¹ reported that the prevalence of obesity (defined as a body mass index [BMI] Ž30 kg/m2) increased in the United States from 12.0% in 1991 to 17.9% in 1998. Must and colleagues² found that the prevalence of morbidities such as diabetes, gallbladder disease, and osteoarthritis increased with severity of overweight (BMI >27 and <30) and obesity. The impact of obesity on health-related quality of life has been less well studied than how it affects physical morbidity and mortality.

Le Pen and coworkers³ surveyed a random sample of 500 French subjects with BMI >27 and a control sample of 500 subjects matched for sex, age, and employment status drawn from the nonobese population. Using a short specific quality of life scale and the Medical Outcomes Study Short Form-36 (SF-36),⁴ it was found that: (1) moderately obese subjects (BMI >27 and <30) did not significantly differ from those in the control group except for physical capacity; and (2) in the group of obese subjects with a BMI Ž30, quality of life seemed to be impaired for 5 of 9 dimensions of the SF-36 compared with the control population, all related to physical consequences of obesity. The study population perceived itself in terms of poor general health.³ No significant difference was observed between the samples for the psychological and social dimensions of the SF-36. Barofsky and colleagues⁵ also found that pain had a significant impact on the quality of life of obese patients. Fontaine and coworkers⁶ reported that weight loss was associated with significantly improved scores relative to baseline on the physical functioning, role-physical, general health, vitality, and mental health domains of the SF-36. The largest improvements were with respect to the vitality, general health perception, and role-physical domains.

In most studies of health-related quality of life, obesity has been treated as a dichotomous variable, with the cut point between nonobese and obese persons commonly set at a BMI of 27 kg per m2. This study presents an analysis of the variation of self-reported quality of life in a survey of primary care patients in relation to BMI as a continuous measure.

Methods

A questionnaire was mailed to all patients of the Family Medicine Centre at Mt Sinai Hospital in Toronto who were 45 to 74 years of age and had made at least 3 visits to the clinic during 1996-1997. A modified Dillman method⁷ was used with an initial mailing followed by a reminder postcard and a second mailing of the questionnaire. An ethics committee at the University of Toronto approved the project. The 103-question survey included the 12-item Short Form of the SF-36 (SF-12) quality of life instrument (QOL),⁸ in addition to questions about height, weight, and family income. The SF-12 inquires about physical and mental health and permits the computation of 2 summary scales, the physical component scale (PCS-12) and the mental component scale (MCS-12). These scales have been standardized to a mean score of 50 and a standard deviation (SD) of 10 in the general population.

Multiple linear regression models were used to explore the relationships between PCS-12, MCS-12, and BMI. The variables of age, sex, and family income (in 6 categories) were controlled, but health factors such as hypertension and diabetes which are in the causal pathway between BMI and quality of life were deliberately omitted.⁹ Modern statistical methodology was used to model the shape of the relationships between BMI and quality-of-life measures. To visualize the relationship, a regression smoother (a nonparametric regression function with no prespecified shape) was fitted to the data. S-Plus software¹⁰ was used, but this capability is also available in other statistical packages, such as SPSS.¹¹ The result suggested that the relationships between quality of life and BMI were nonlinear. To accommodate the curvilinear shape ([Figure]), BMI was modeled with restricted cubic splines.¹² Restricted cubic splines are a method of describing dose-response curves that make no a priori assumptions about the shape of the curve. Cubic polynomials are fitted between prespecified points on the horizontal axis (knots), and restrictions are placed on the resulting curve to ensure a smooth appearance at those knot points.¹² Hypothesis testing can be performed to determine whether the nonlinear components of the model significantly improve the fit to the data. BMI was modeled with a 4-knot restricted cubic spline using Harrell’s Design library¹³ in S-Plus.¹⁰

Results

The survey was mailed to 1061 subjects (668 women, 396 men). Responses were received from 564 subjects (53%). [Table 1] shows the distribution of the respondents by age and sex, and the mean values for BMI and scores on the PCS-12 and MCS-12. The mean BMI did not vary with age or sex. Scores on both scales of the SF-12 were higher among men than women.

In performing the statistical analysis it was assumed that although the distribution of BMI or SF-12 scores among respondents and nonrespondents might differ, the relationships among these variables observed in respondents is generalizable. The regression results for the MCS-12 and PCS-12 are shown in [Table 2]. Interestingly, while the mean PCS-12 scores declined with age, there was a significant increase in the MCS-12 scores with age. Men had higher scores than women, but there was no interaction between BMI and sex. Both MCS-12 and PCS-12 scores increased with rising income. After adjusting for age, sex, and income, there were significant nonlinear relationships between MCS-12 and PCS-12 scores and BMI. The figure shows the relationships between QOL scores and BMI adjusted for purposes of illustration to age 60 years and to an income of Ž$80,000 per year. For both scales, there was a peak in QOL scores for BMI values in the range 20 to 25 kg per m2. For the PCS-12, there was a steady decline in QOL from the peak, with a drop of approximately 5 points (SD=0.5) at a BMI of 30 kg per m2. The relative change in MCS-12 scores was less, with no further decline for BMI greater than 30 kg per m2. The wide confidence limits preclude a confident assessment of the shape of the curve for the MCS-12 in the range of BMI greater than 30 kg per m2.

Discussion

The scales on the SF-12 reflect a self-assessment of well-being, pain, limitations, and energy. Self-reported QOL varied with BMI, with a peak in the range of 20 to 25 kg per m2. The PCS-12 scores declined monotonically from the peak with increasing BMI, consistent with reports of bodily pain as a significant comorbidity of obesity.^3,5 There was a proportionally smaller decline from the peak in the MCS-12 scores; scores did not continue to decline for those with BMI >30 kg per m2.

In 1998, the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health published evidence-based clinical guidelines for the identification, evaluation, and treatment of overweight and obesity in adults.¹⁴ On the basis of BMI they use 5 classes of increasing severity consistent with the idea of graded risk. Subjects with a BMI in the range of 18.5 to 24.9 kg per m2 are classified as having normal weight; those with a BMI of 25.0 to 29.9 kg per m2 are classified as preobese; and those with a BMI greater than 30 kg per m2 are assigned to 3 categories of obesity. The upper cutoff point for healthy weight at 25 kg per m2 is consistent with that recommended by a steering committee of the American Institute of Nutrition15 and an expert committee of the World Health Organization.¹⁶ Evaluation of a lower cutoff for healthy weight is complex, because the leanest group in a population is a mix of smokers, persons who have lost weight as a result of underlying disease, and persons who have maintained a lean weight by balancing physical activity and caloric intake.¹⁷ In this study, the number of subjects (N=23) with a BMI less than 20 kg per m2 was too low to carefully investigate the lower end of the BMI-QOL relationship.

The observations reported here from a survey of primary care patients suggest that achieving a healthy weight as recommended by the NHLBI will maximize the patient’s subjective sense of well-being.

Acknowledgments

Support from a grant from the American Academy of Family Physicians Foundation is gratefully acknowledged.

Obesity is a common condition that is increasing in prevalence. Mokdad and colleagues¹ reported that the prevalence of obesity (defined as a body mass index [BMI] Ž30 kg/m2) increased in the United States from 12.0% in 1991 to 17.9% in 1998. Must and colleagues² found that the prevalence of morbidities such as diabetes, gallbladder disease, and osteoarthritis increased with severity of overweight (BMI >27 and <30) and obesity. The impact of obesity on health-related quality of life has been less well studied than how it affects physical morbidity and mortality.

Le Pen and coworkers³ surveyed a random sample of 500 French subjects with BMI >27 and a control sample of 500 subjects matched for sex, age, and employment status drawn from the nonobese population. Using a short specific quality of life scale and the Medical Outcomes Study Short Form-36 (SF-36),⁴ it was found that: (1) moderately obese subjects (BMI >27 and <30) did not significantly differ from those in the control group except for physical capacity; and (2) in the group of obese subjects with a BMI Ž30, quality of life seemed to be impaired for 5 of 9 dimensions of the SF-36 compared with the control population, all related to physical consequences of obesity. The study population perceived itself in terms of poor general health.³ No significant difference was observed between the samples for the psychological and social dimensions of the SF-36. Barofsky and colleagues⁵ also found that pain had a significant impact on the quality of life of obese patients. Fontaine and coworkers⁶ reported that weight loss was associated with significantly improved scores relative to baseline on the physical functioning, role-physical, general health, vitality, and mental health domains of the SF-36. The largest improvements were with respect to the vitality, general health perception, and role-physical domains.

In most studies of health-related quality of life, obesity has been treated as a dichotomous variable, with the cut point between nonobese and obese persons commonly set at a BMI of 27 kg per m2. This study presents an analysis of the variation of self-reported quality of life in a survey of primary care patients in relation to BMI as a continuous measure.

Methods

A questionnaire was mailed to all patients of the Family Medicine Centre at Mt Sinai Hospital in Toronto who were 45 to 74 years of age and had made at least 3 visits to the clinic during 1996-1997. A modified Dillman method⁷ was used with an initial mailing followed by a reminder postcard and a second mailing of the questionnaire. An ethics committee at the University of Toronto approved the project. The 103-question survey included the 12-item Short Form of the SF-36 (SF-12) quality of life instrument (QOL),⁸ in addition to questions about height, weight, and family income. The SF-12 inquires about physical and mental health and permits the computation of 2 summary scales, the physical component scale (PCS-12) and the mental component scale (MCS-12). These scales have been standardized to a mean score of 50 and a standard deviation (SD) of 10 in the general population.

Multiple linear regression models were used to explore the relationships between PCS-12, MCS-12, and BMI. The variables of age, sex, and family income (in 6 categories) were controlled, but health factors such as hypertension and diabetes which are in the causal pathway between BMI and quality of life were deliberately omitted.⁹ Modern statistical methodology was used to model the shape of the relationships between BMI and quality-of-life measures. To visualize the relationship, a regression smoother (a nonparametric regression function with no prespecified shape) was fitted to the data. S-Plus software¹⁰ was used, but this capability is also available in other statistical packages, such as SPSS.¹¹ The result suggested that the relationships between quality of life and BMI were nonlinear. To accommodate the curvilinear shape ([Figure]), BMI was modeled with restricted cubic splines.¹² Restricted cubic splines are a method of describing dose-response curves that make no a priori assumptions about the shape of the curve. Cubic polynomials are fitted between prespecified points on the horizontal axis (knots), and restrictions are placed on the resulting curve to ensure a smooth appearance at those knot points.¹² Hypothesis testing can be performed to determine whether the nonlinear components of the model significantly improve the fit to the data. BMI was modeled with a 4-knot restricted cubic spline using Harrell’s Design library¹³ in S-Plus.¹⁰

Results

The survey was mailed to 1061 subjects (668 women, 396 men). Responses were received from 564 subjects (53%). [Table 1] shows the distribution of the respondents by age and sex, and the mean values for BMI and scores on the PCS-12 and MCS-12. The mean BMI did not vary with age or sex. Scores on both scales of the SF-12 were higher among men than women.

In performing the statistical analysis it was assumed that although the distribution of BMI or SF-12 scores among respondents and nonrespondents might differ, the relationships among these variables observed in respondents is generalizable. The regression results for the MCS-12 and PCS-12 are shown in [Table 2]. Interestingly, while the mean PCS-12 scores declined with age, there was a significant increase in the MCS-12 scores with age. Men had higher scores than women, but there was no interaction between BMI and sex. Both MCS-12 and PCS-12 scores increased with rising income. After adjusting for age, sex, and income, there were significant nonlinear relationships between MCS-12 and PCS-12 scores and BMI. The figure shows the relationships between QOL scores and BMI adjusted for purposes of illustration to age 60 years and to an income of Ž$80,000 per year. For both scales, there was a peak in QOL scores for BMI values in the range 20 to 25 kg per m2. For the PCS-12, there was a steady decline in QOL from the peak, with a drop of approximately 5 points (SD=0.5) at a BMI of 30 kg per m2. The relative change in MCS-12 scores was less, with no further decline for BMI greater than 30 kg per m2. The wide confidence limits preclude a confident assessment of the shape of the curve for the MCS-12 in the range of BMI greater than 30 kg per m2.

Discussion

The scales on the SF-12 reflect a self-assessment of well-being, pain, limitations, and energy. Self-reported QOL varied with BMI, with a peak in the range of 20 to 25 kg per m2. The PCS-12 scores declined monotonically from the peak with increasing BMI, consistent with reports of bodily pain as a significant comorbidity of obesity.^3,5 There was a proportionally smaller decline from the peak in the MCS-12 scores; scores did not continue to decline for those with BMI >30 kg per m2.

In 1998, the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health published evidence-based clinical guidelines for the identification, evaluation, and treatment of overweight and obesity in adults.¹⁴ On the basis of BMI they use 5 classes of increasing severity consistent with the idea of graded risk. Subjects with a BMI in the range of 18.5 to 24.9 kg per m2 are classified as having normal weight; those with a BMI of 25.0 to 29.9 kg per m2 are classified as preobese; and those with a BMI greater than 30 kg per m2 are assigned to 3 categories of obesity. The upper cutoff point for healthy weight at 25 kg per m2 is consistent with that recommended by a steering committee of the American Institute of Nutrition15 and an expert committee of the World Health Organization.¹⁶ Evaluation of a lower cutoff for healthy weight is complex, because the leanest group in a population is a mix of smokers, persons who have lost weight as a result of underlying disease, and persons who have maintained a lean weight by balancing physical activity and caloric intake.¹⁷ In this study, the number of subjects (N=23) with a BMI less than 20 kg per m2 was too low to carefully investigate the lower end of the BMI-QOL relationship.

The observations reported here from a survey of primary care patients suggest that achieving a healthy weight as recommended by the NHLBI will maximize the patient’s subjective sense of well-being.

Acknowledgments

Support from a grant from the American Academy of Family Physicians Foundation is gratefully acknowledged.

Obesity is a common condition that is increasing in prevalence. Mokdad and colleagues¹ reported that the prevalence of obesity (defined as a body mass index [BMI] Ž30 kg/m2) increased in the United States from 12.0% in 1991 to 17.9% in 1998. Must and colleagues² found that the prevalence of morbidities such as diabetes, gallbladder disease, and osteoarthritis increased with severity of overweight (BMI >27 and <30) and obesity. The impact of obesity on health-related quality of life has been less well studied than how it affects physical morbidity and mortality.

Le Pen and coworkers³ surveyed a random sample of 500 French subjects with BMI >27 and a control sample of 500 subjects matched for sex, age, and employment status drawn from the nonobese population. Using a short specific quality of life scale and the Medical Outcomes Study Short Form-36 (SF-36),⁴ it was found that: (1) moderately obese subjects (BMI >27 and <30) did not significantly differ from those in the control group except for physical capacity; and (2) in the group of obese subjects with a BMI Ž30, quality of life seemed to be impaired for 5 of 9 dimensions of the SF-36 compared with the control population, all related to physical consequences of obesity. The study population perceived itself in terms of poor general health.³ No significant difference was observed between the samples for the psychological and social dimensions of the SF-36. Barofsky and colleagues⁵ also found that pain had a significant impact on the quality of life of obese patients. Fontaine and coworkers⁶ reported that weight loss was associated with significantly improved scores relative to baseline on the physical functioning, role-physical, general health, vitality, and mental health domains of the SF-36. The largest improvements were with respect to the vitality, general health perception, and role-physical domains.

In most studies of health-related quality of life, obesity has been treated as a dichotomous variable, with the cut point between nonobese and obese persons commonly set at a BMI of 27 kg per m2. This study presents an analysis of the variation of self-reported quality of life in a survey of primary care patients in relation to BMI as a continuous measure.

Methods

A questionnaire was mailed to all patients of the Family Medicine Centre at Mt Sinai Hospital in Toronto who were 45 to 74 years of age and had made at least 3 visits to the clinic during 1996-1997. A modified Dillman method⁷ was used with an initial mailing followed by a reminder postcard and a second mailing of the questionnaire. An ethics committee at the University of Toronto approved the project. The 103-question survey included the 12-item Short Form of the SF-36 (SF-12) quality of life instrument (QOL),⁸ in addition to questions about height, weight, and family income. The SF-12 inquires about physical and mental health and permits the computation of 2 summary scales, the physical component scale (PCS-12) and the mental component scale (MCS-12). These scales have been standardized to a mean score of 50 and a standard deviation (SD) of 10 in the general population.

Multiple linear regression models were used to explore the relationships between PCS-12, MCS-12, and BMI. The variables of age, sex, and family income (in 6 categories) were controlled, but health factors such as hypertension and diabetes which are in the causal pathway between BMI and quality of life were deliberately omitted.⁹ Modern statistical methodology was used to model the shape of the relationships between BMI and quality-of-life measures. To visualize the relationship, a regression smoother (a nonparametric regression function with no prespecified shape) was fitted to the data. S-Plus software¹⁰ was used, but this capability is also available in other statistical packages, such as SPSS.¹¹ The result suggested that the relationships between quality of life and BMI were nonlinear. To accommodate the curvilinear shape ([Figure]), BMI was modeled with restricted cubic splines.¹² Restricted cubic splines are a method of describing dose-response curves that make no a priori assumptions about the shape of the curve. Cubic polynomials are fitted between prespecified points on the horizontal axis (knots), and restrictions are placed on the resulting curve to ensure a smooth appearance at those knot points.¹² Hypothesis testing can be performed to determine whether the nonlinear components of the model significantly improve the fit to the data. BMI was modeled with a 4-knot restricted cubic spline using Harrell’s Design library¹³ in S-Plus.¹⁰

Results

The survey was mailed to 1061 subjects (668 women, 396 men). Responses were received from 564 subjects (53%). [Table 1] shows the distribution of the respondents by age and sex, and the mean values for BMI and scores on the PCS-12 and MCS-12. The mean BMI did not vary with age or sex. Scores on both scales of the SF-12 were higher among men than women.

In performing the statistical analysis it was assumed that although the distribution of BMI or SF-12 scores among respondents and nonrespondents might differ, the relationships among these variables observed in respondents is generalizable. The regression results for the MCS-12 and PCS-12 are shown in [Table 2]. Interestingly, while the mean PCS-12 scores declined with age, there was a significant increase in the MCS-12 scores with age. Men had higher scores than women, but there was no interaction between BMI and sex. Both MCS-12 and PCS-12 scores increased with rising income. After adjusting for age, sex, and income, there were significant nonlinear relationships between MCS-12 and PCS-12 scores and BMI. The figure shows the relationships between QOL scores and BMI adjusted for purposes of illustration to age 60 years and to an income of Ž$80,000 per year. For both scales, there was a peak in QOL scores for BMI values in the range 20 to 25 kg per m2. For the PCS-12, there was a steady decline in QOL from the peak, with a drop of approximately 5 points (SD=0.5) at a BMI of 30 kg per m2. The relative change in MCS-12 scores was less, with no further decline for BMI greater than 30 kg per m2. The wide confidence limits preclude a confident assessment of the shape of the curve for the MCS-12 in the range of BMI greater than 30 kg per m2.

Discussion

The scales on the SF-12 reflect a self-assessment of well-being, pain, limitations, and energy. Self-reported QOL varied with BMI, with a peak in the range of 20 to 25 kg per m2. The PCS-12 scores declined monotonically from the peak with increasing BMI, consistent with reports of bodily pain as a significant comorbidity of obesity.^3,5 There was a proportionally smaller decline from the peak in the MCS-12 scores; scores did not continue to decline for those with BMI >30 kg per m2.

In 1998, the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health published evidence-based clinical guidelines for the identification, evaluation, and treatment of overweight and obesity in adults.¹⁴ On the basis of BMI they use 5 classes of increasing severity consistent with the idea of graded risk. Subjects with a BMI in the range of 18.5 to 24.9 kg per m2 are classified as having normal weight; those with a BMI of 25.0 to 29.9 kg per m2 are classified as preobese; and those with a BMI greater than 30 kg per m2 are assigned to 3 categories of obesity. The upper cutoff point for healthy weight at 25 kg per m2 is consistent with that recommended by a steering committee of the American Institute of Nutrition15 and an expert committee of the World Health Organization.¹⁶ Evaluation of a lower cutoff for healthy weight is complex, because the leanest group in a population is a mix of smokers, persons who have lost weight as a result of underlying disease, and persons who have maintained a lean weight by balancing physical activity and caloric intake.¹⁷ In this study, the number of subjects (N=23) with a BMI less than 20 kg per m2 was too low to carefully investigate the lower end of the BMI-QOL relationship.

The observations reported here from a survey of primary care patients suggest that achieving a healthy weight as recommended by the NHLBI will maximize the patient’s subjective sense of well-being.

Acknowledgments

Support from a grant from the American Academy of Family Physicians Foundation is gratefully acknowledged.