Older Subjects without Radiographic Knee Osteoarthritis: Weight, Height, and Body Mass Index

Abstract

Osteoarthritis, the most common form of arthritis, conservatively affects 27 million American adults. While nearly all joints can be affected, knee osteoarthritis is of particular interest as it has the potential to severely limit mobility. Despite its ubiquity, not everyone is affected. Few studies on the characteristics of these persons exist. This descriptive paper examined older subjects without radiographic knee osteoarthritis (RKOA) at age 78, describing their height, weight and body mass index (BMI). Oldest subjects without RKOA at age 78 (n=143) were selected from the parent Clearwater Osteoarthritis Study, conducted from 1988–2009. Recalled weight and height for ages 25 and 45, as well as current weight and height were collected via questionnaire. Current measurements were also taken. BMI was derived. Absence of RKOA was determined using Kellgren-Lawrence criteria. Demographics: mostly female (72%), all Caucasian, better educated than the average US citizen, and more subjects lived with others. At all three life-stages, weights and BMI were lower, while heights were comparable to US 2000 figures. BMI was in the “normal” category at age 25 and in the “overweight” category at age 78. Weight was under reported and height was over reported, but by less than the general population. Both the increase in weight and decrease in height with age were statistically significant. Absence of RKOA appears to be related to lower weights, average heights, and lower BMI, relative to population norms at all three life-stages. These findings also suggest that the relationship may be more complicated. Higher BMI is ubiquitously recognized as a risk factor for RKOA, yet the average BMI of the subjects without RKOA was in the “overweight” category at age 78. A possible explanation may be an “inflated” BMI based on decrease in height, not just increase in weight.

Osteoarthritis (OA), the most common form of over 100 types of arthritis, is a highly debilitating chronic disease affecting the aging population. OA is characterized by cartilage erosion, development of osteophytes, changes in the subchondral bone and synovial fluid, and the involvement of a variety of inflammatory components [1]. The exact pathogenesis has yet to be elucidated. OA is generally considered to have a complicated framework that includes genetic, hormonal, metabolic and biomechanical factors and specific risk factors such as age, gender, obesity, and past injury have been identified [2–6].

Age is universally regarded as the most prominent risk factor for the condition. OA is commonly regarded as an inevitable consequence of aging. While OA can and does affect many joints, radiographic knee osteoarthritis (RKOA) is of particular interest as it has the potential to severely limit mobility, which has dismal implications at any age. Prevalence of radiographically determined knee OA (RKOA) in both men and women over age 80 ranges from 26–55% [7, 8]. The bulk of the research on RKOA has been focused on the presence of the disease; manuscripts describing subjects without RKOA are exiguous.

Goekoop et al have speculated that the absence of known risk factors may be protective against OA [3]. Might there also be other factors associated with the absence of RKOA? This study was designed as an initial description of weight, height, and BMI in a group of older subjects who do not have knee OA. The supposition that the opposite of obesity is protective against OA may underpin previous research on protective factors.

METHODS AND MATERIALS

Subject Selection

This study looks at weight, height, and body mass index (BMI) of 143 subjects who were 78 years old and did not have signs of radiographic knee OA (RKOA). The data were gleaned from the Clearwater Osteoarthritis Study (COS), a community-based longitudinal study of 3700+ subjects age 40 and older, conducted by the Arthritis Research Institute of America in the Clearwater, Florida area from 1988–2009. Located in Pinellas County, the Clearwater area has been noted for a substantial population of older adults, which made it an ideal location to study risk factors for OA. Subjects were recruited from the local community. Data collected from subjects every two years included demographic, personal attributes, health history, lifestyle, physical exam, and radiographic information. More detailed information on the COS is documented in other publications [9–12].

The combination of oldest age without radiographic evidence of knee OA and a sample size of at least 100 subjects, led to the selection of N=143 COS subjects without RKOA by age 78 (+/− 1 year).

Definition of Radiographic Knee OA

At each COS visit, participants had weight-bearing anterior-posterior knee x-rays taken by a licensed x-ray technician using standard plain film exposure techniques. A board-certified radiologist who had no knowledge of subjects’ symptoms or physical characteristics graded each radiograph for presence and severity of RKOA [11]. The x-rays were scored based on the Kellgren and Lawrence (KL) knee OA criteria as described by the Atlas of Standard Radiographs of Arthritis [13]. This ordinal system grades joints on a scale of 0 to 4: 0 = none or absent, 1 = questionable osteophytes and no joint space narrowing, 2 = definite osteophytes with possible joint space narrowing, 3 = definite joint space narrowing with moderate multiple osteophytes and some sclerosis, and 4 = severe joint space narrowing with cysts, osteophytes and sclerosis present. Each knee received a single KL score for the lateral and medial compartments of the tibio-femoral joint together. Absence of RKOA was determined by a KL score of < 2 in both knees. This is consistent with the primary definition of RKOA (KL ≥ 2) used in most research [14, 15].

Weight

Weight is considered at three different life stages: young adult (age 25), middle-age (age 45) and in older age at the time of RKOA assessment (age 78). Weight at ages 25 and 45 were self-reported. At their first and subsequent two visits, each subject responded to the questions: “What did you weigh at 25?” and “What did you weigh at age 45 (plus or minus 5 years)?” Current weight was also self-reported at the first three visits for comparison to measured weight at the same visit. At each COS visit, each subject was weighed on a standard physician’s scale (pounds). Weight at 78 years old was taken from the measured weight at the visit that occurred closest to age 78 (approx. +/− 1 year).

Height

Height was considered at two different life stages: young adult (age 25) and in older age at the time of RKOA assessment (age 78). Height at age 25 was self-reported at the first visit and subsequent two visits: “What was your height in stocking feet at age 25?”. Height at age 45 was not queried. Current height was also self-reported at the first three visits for comparison to measured height at the same visit. At each COS visit, each subject’s height was measured on a standard physician’s stadiometer (inches). Height at 78 years old was taken from the measured height at the visit that occurred closest to age 78 (approx. +/− 1 year).

Body Mass Index (BMI)

BMI was calculated at two different life stages: young adult (age 25) and in older age at the time of RKOA assessment (age 78) using the following standard formula: {BMI = Weight × 703 / Height²} (World Health Organization, 2012; http://apps.who.int//bmi/index.jsp?introPage=intro_3.html). BMI at age 25 was based on the self-reported weights and heights described above. BMI at age 78 was based on the measured weights and heights described above.

Statistical Analysis

As a descriptive study, the statistics are limited to counts with percentages and means with standard deviations. Differences in weight, height, and BMI between younger and older ages are tested for significance using a paired sample t-test with 95% confidence intervals (CI) for the mean difference. The data analysis for this paper was generated using SAS software, Version 9.3 of the SAS System for Windows.

US population means for different age and race categories weight, height, and BMI were provided for contextual reference from the National Health and Nutrition Examination Surveys (NHANES) [16]. The ages and dates of the data were chosen to be reasonably analogous for general comparison purposes. The earliest data with height, weight and BMI by gender and age group was from the 1960–62 National Health Examination Survey (NHES) I, which was used for the 20–29 year old age group comparisons for age 25. NHANES II, 1976–1980 age range 40–49 years was used for comparison to the age 45 data. For older age group, age 78, which is both closer to the present and used measurements instead of self-reported values, the NHANES IV, 1999–2002 with age ranges 60+ years and 75+years were used. The selected race categories for comparison to this study’s all Caucasian subjects were all races and non-Hispanic white.

RESULTS

Study Population

General characteristics of this sample of subjects without RKOA can be seen in Table 1. This sample is all Caucasian, reflecting the 97% Caucasian make-up of the parent COS. This group is 72% women, which is somewhat higher than the parent study with 67% women, which is also higher than the United States (US) Census 2000 population estimate of 59% women for the age group 75 to 79 years old (US Census Bureau 2012; www.census.gov/popest/national/asrh/NC-EST2009/NC-EST2009-01.xls).

Table 1.

Demographics of COS^a Subjects Absent Knee OA^b at Age 78

Demographic Variables	Frequency (%)
	All	Females	Males
Gender	143 (100.0)	103 (72.0)	40 (28.0)
Education
<8th Grade	2 (1.4)	2 (1.9)	0 (0.0)
Some High School	55 (3.5)	5 (4.9)	0 (0.0)
High School	41 (28.7)	33 (32.0)	8 (20.0)
Some Vocational Training	13 (9.1)	10 (9.7)	3 (7.5)
Some College	45 (31.5)	32 (31.1)	13 (32.5)
College Grad	24 (16.8)	12 (11.7)	12 (30.0)
Grad School	12 (8.12)	8 (7.8)	4 (10.0)
Don’t Know	1 (0.7)	1 (1.0)	0 (0.0)
Living Status
Lives Alone	45 (31.5)	42 (40.8)	3 (7.5)
Lives with Someone	93 (65.0)	59 (57.3)	34 (85.0)
	Mean ± SDc
Years in COSa	10.1 ± 4.1	10.3 ± 3.7	9.6 ± 4.9
BMI at Age 78	25.8 ± 4.0	25.4 ± 3.8	26.9 ± 3.3
Race	100% Caucasian

^aClearwater Osteoarthritis Study, Arthritis Research Institute of America

^bAbsent Knee Osteoarthritis defined by Kellgren-Lawrence score <2

^cStandard deviation

^dIn pounds

^eIn inches

In this group of subjects without RKOA, 41% reported that they completed high school, 24% graduated from college, and 12% have at least some graduate level education. The US Census 2000 estimates for educational level show that 28.6% of the population over 25 years old graduated from high school, 15.5% graduated from college, and 8.9% had at least some graduate level education (Educational Attainment. US Census 2000 Brief; www.census.gov/prod/2003pubs/c2kbr-24.pdf).

Overall, approximately 2/3 of the participants in this study live with someone, reflecting the high rate (85%) of men living with someone. The 2000 US Census reports that 47% of those 65 years and older lived alone, 25% of these were men (US Census Bureau 2001. (www.census.gov/prod/2001pubs/p20-537.pdf). BMI at age 78 is addressed in the next section.

A comparison of self-reported vs. measured weight at the first visit (mean age 72.2 years) showed that this group under-reported their weight by an average (mean) of 0.57 pounds, which was borderline statistically significant (p=0.05). When the genders were separated, women significantly (alpha= 0.05) under-reported their weight by an average of 0.72 pounds (p=0.04) while men under-reported their weight by an average of 0.21 pounds (p=0.71).

The similar comparison for self-reported height vs. measured height showed that overall height was over-reported by 0.78 inches and that this was statistically significant (p<0.0001). Women significantly over-reported their height by an average 0.68 inches (p<0.0001). Men over-reported height more than the women did by an average of 1.05 inches (p<0.0001).

Weight, Height, and BMI

Mean weights at ages 25, 45, and 78, as well as mean heights and BMI at ages 25 and 78 are displayed in Table 2. Mean weights and BMI increase with age in all categories for both men and women, while mean heights decreased for both genders. BMI at age 25 is 20.6 (SD=2.5) for women and 22.2 (SD=2.7) for men. Both values are solidly within the “normal” World Health Organization (WHO) category (underweight < 18.5, normal 18.5 to <25, overweight 25 to <30, and obese ≥30) (BMI Classification (2006). Global Database on Body Mass Index. World Health Organization; http://apps.who.int/bmi/index.jsp?introPage=intro_3.html). BMI at age 78 is 25.4 (SD=3.8) for women and 26.9 (SD=4.3) for men, which is right at the cusp of the WHO categories of “normal” and “overweight”.

Table 2.

Height, Weight, & BMI among COS^a Subjects Absent Knee OA^b at Age 78

Weightd, Heighte, & BMI At Ages 25, 45, & 78	Mean ± SDc
	All	Females	Males
Weight Age 25	128.4 ± 22.0	119.4 ± 14.7	151.3 ± 21.1
Weight Age 45	139.8 ± 24.0	128.9 ± 14.8	167.9 ± 19.9
Weight Age 78	149.2 ± 28.1	140.1 ± 22.6	172.9 ± 27.2
Height Age 25	65.3 ± 3.4	63.9 ± 2.5	69.1 ± 2.6
Height Age 78	63.7 ± 3.4	62.3 ± 2.3	67.3 ± 2.9
BMI Age 25	21.0 ± 2.7	20.6 ± 2.5	22.2 ± 2.7
BMI Age 78	25.8 ± 4.0	25.4 ± 3.8	26.9 ± 4.3

^aClearwater Osteoarthritis Study, Arthritis Research Institute of America

^bAbsent Knee Osteoarthritis defined by Kellgren-Lawrence score <2

^cStandard deviation

^dIn pounds

^eIn inches

Table 3 shows the comparable weight, height, and BMI data from the NHANES. The weight of female subjects absent RKOA at age 78 were 6–11 pounds less than the general population weights from the NHANES data for all three age categories investigated. The same comparison for men ranged from 0 to 12 pounds. The largest weight difference occurred with absent RKOA subjects at age 78 and age 60 and older non-Hispanic whites from NHANES; women and men in this referent group were 16–18 pounds heavier than the subjects absent RKOA at age 78.

Table 3.

Weight, Height, and BMI among NHANES^a Subjects

	Females		Males
Weightc, Heightd, &BMIe	Mean ± SEb	N	Mean ± SEb	N
Weight20–29 years	127.7 ± 0.9	672	163.9 ± 1.6	585
Weight40–49 years	148.8 ± 1.4	765	179.7 ± 1.0	695
*Weight75 years and older	146.6 ± 2.1	554	172.7 ± 1.4	523
†Weight60 years and older	158.0 ± 1.2	862	188.8 ± 1.0	866
Height20–29 years	63.7 ± 0.1	672	68.9 ± 0.1	586
*Height75 years and older	62.0 ± 0.1	538	67.4 ± 0.2	505
†Height60 years and older	62.8 ± 0.1	845	68.6 ± 0.1	855
BMI20–29 years	22.2 ± 0.2	672	24.3 ± 0.2	586
*BMI75 years and older	26.8 ± 0.4	524	26.8 ± 0.2	487
†BMI60 years and older	28.2 ± 0.2	830	28.3 ± 0.1	836

^aNational Health and Nutritional Examination Survey:

20–29 years from 1960–62, 40–49 years from 1976–1980,

75 years and older and 60 years and older from 1999–2002

^bStandard Error

^cIn pounds

^dIn inches

^eBody Mass Index = [Weight × 703 / Height²]

^*All Races

^†Non-Hispanic White

Height for both women and men is similar for both ages 25 and 78. The exception is that the NHANES comparison for age 60 and older non-Hispanic whites is approximately an inch taller than the 78 year old absent RKOA male subjects.

BMI at age 78 for both women and men was virtually the same as the NHANES 1999–2002 figures by gender for age 75 and older (all races) with estimates in the lower range of the overweight category. The NHANES 1999–2000 estimates for non-Hispanic whites (age 60 and older) were in the mid-high range of the overweight category for both women and men. Although statistical comparisons are not possible, it appears that there is virtually no difference in BMI by gender between these subjects without RKOA at age 78 and the general US population.

Table 4 shows the mean difference in weight, height, and BMI from younger to older ages. Weight significantly increases for both women (9.5 pounds, p<0.0001) and men (16.6 pounds, p<0.0001) from age 25 to age 45 and again from age 45 to age 78, although the second increase is not significant for men (females 11.2 pounds, p<0.0001; males 5.1 pounds p>0.05). Overall, the increase in weight from age 25 to age 78 is significant for both women and men (p<0.0001 both genders). Height decreases significantly (p <0.0001) from age 25 to age 78 for both women and men. The combination of the increase in weight and decrease of height with age resulted in a significant increase in BMI for both men (14.6 points, p<0.0001) and women (13.8 points, p<0.001).

Table 4.

Mean Difference in Height, Weight, and BMI at Different Life Stages among COS^a Subjects Absent Knee OA^b at Age 78

Differences in Weight, Height, & BMI	All	Females	Males
	Mean diff.c (95% CId)e	Mean diff.c (95% CId)e	Mean diff.c (95% CId)e
WeightAge 45 – WeightAge 25	11.5 (9.7, 13.3)	9.5 (7.7, 11.7)	16.6 (12.6, 20.6)
WeightAge 78 – WeightAge 45	9.5 (6.5, 12.5)	11.2 (8.0, 14.5)	5.1 (−1.8, 11.9)
WeightAge 78 – WeightAge 25	21.0 (17.4, 24.6)	20.8 (17.0, 24.5)	21.6 (13.1, 30.1)
HeightAge 78 – HeightAge 25	−1.7 (−2.0, −1.4)	−1.6 (−1.9, −1.2)	−2.0 (−2.6, −1.4)
BMIAge 78 – BMIAge 25	14.1 (13.4, 14.7)	13.8 (13.1, 14.5)	14.6 (13.2, 16.1)

^aClearwater Osteoarthritis Study, Arthritis Research Institute of America

^bAbsent Knee Osteoarthritis defined by Kellgren-Lawrence score <2

^cMean difference

^dConfidence interval

^ePaired t-tests with p-value < .001

DISCUSSION

Study Population

The gender composition of this study group, almost ¾ women, may be a sampling effect, an age effect (in the general population there are more older women than older men), a gender difference in non-occurrence of RKOA at age 78, or a combination of all three (US Census Bureau 2012; www.census.gov/popest/national/asrh/NC-EST2009/NC-EST2009-01.xls). General population prevalence rates calculated specifically for absence of RKOA by gender are not available in the scientific literature. Goekoop found an overall rate of 51% of subjects 90 years of age absent RKOA (n=82), but gender composition was not reported [3].

This group of subjects without RKOA has a higher educational level than the average US population surveyed in 2000. The difference is most pronounced for high school graduation with 12% more study subjects reporting high school graduation than the general public. Likewise, 8% more graduated from college and 3% more had graduate level education. A reasonable assumption to explain the differences is the general trend towards increased health status with greater socioeconomic class.

The rates of under-reporting weight and over-reporting height at the first visit in this study was less than what has been reported in recent studies and bodes well for confidence in the self-reported weights for ages 25 and 45 and self-reported height at age 25 [17–20]. The literature has not established the reliability of reporting weights and heights from earlier life stages. However, recall bias is a well-documented phenomenon and is likely a factor.

Weight, Height, and BMI

Mean weight in both early and mid-adulthood for both men and women is less by 8–12 pounds in the absent RKOA at age 78 group compared to the general population. In the older adult group the difference is less only 6 pounds for women and virtually no difference for men. There is a greater difference (16–18 pounds) when the race restricted, but broader age range is used for reference, which could reflect both aspects. These comparisons provide a foundation linking low BMI in early and mid-adulthood with non-occurring RKOA in older adulthood. The disparate indicators for the weight comparisons at older age need further study. It is unclear whether race is a factor or if the broad age category changes the age/weight relationship.

The self-reported mean height at age 25 for subjects absent RKOA at age 78 is similar to the referent population mean for the 20–29 year old age group in 1960 for both women and men. The measured male and female mean heights at age 78 for subjects absent RKOA are also similar to the general population means for ages 75 and older surveyed in 2000 (see Table 3). Even the race restricted/broader age category referent group is similar for women and only an inch taller for men. This leads to the possible conclusions that either height is not a separate factor for the non-occurrence of RKOA by age 78 or that average height may be a protective factor. Further studies with appropriate statistical comparison groups could elucidate whether either of these possibilities are true.

BMI at age 78 for both women and men was virtually the same as the NHANES 1999–2002 figures for age 75 and older (all races) with estimates in the lower range of overweight category. The NHANES 1999–2000 estimates for non-Hispanic whites (age 60 and older) were in the mid-high range of overweight category in for both women and men. Although statistical comparisons are not possible, it appears that there is virtually no difference in BMI by gender between these subjects absent RKOA at age 78 and the general US population. These findings suggest at least two conclusions: the first is that slightly overweight (based on BMI) elderly people, especially women, may not have an increased risk of RKOA; secondly, it suggests that BMI in general may not be a risk factor for KOA in elderly persons. These findings are in contrast to Goekoop’s where male gender and low BMI were associated with the absence of RKOA [3]. There are no other studies available for comparison.

Perhaps the most interesting finding in this descriptive study of subjects who have not developed RKOA by age 78 is: the increase in BMI in older age groups is due to a decrease in height, as well as to an increase in weight. This conclusion was derived from the results in Table 4. Not only are the increases in weight between ages 25 and 78 significant, but the decreases in height from age 25 to age 78 for both women and men are significant. The effect of the decrease in height is magnified in the BMI equation where height is squared and the squared height is the denominator {BMI = Weight × 703 / Height²}.

BMI increasing with age has become a long-held tenant of chronic disease epidemiology. Perhaps first identified in the Framingham Heart Study, this trend has been noted in many longitudinal studies, including the NHANES predecessors and earlier data sets of height/weights by gender and age group [21–23]. Even though the height-age distributions have been there, the focus has become increasingly on BMI. There have been studies comparing different measures of overweight, which have evaluated weight measures compared to BMI and other anthropomorphic measurements. Height distributions have largely been ignored as height has not been shown to be a good measure of overweight/obesity and there exists a persistent bias towards viewing height as static once maximum growth has occurred. In deed, population means for heights are often reported as height for ages 20 and older [27](CDC Data & Statistics Features: Obesity Trends in Adults with Arthritis, 2012. www.cdc.gov/Features/dsStateArthritisObesity).

This bias towards static adult height was evidenced in this data by looking at the difference between self-reported and measured heights and weights. Weight was not significantly under-reported, which is unusual, but height was over reported by 3/4 inches, which is typical [17, 18].

Recent findings on the risk of death for older adults associated with both chronic and infectious diseases have shown that survival is more likely in those who are overweight, in contrast to other studies which show that overweight/obese persons are more likely to develop diseases and have worse outcomes [22, 24]. Might the anomaly be explained by the “inflation” of BMI due to a decrease in height in older age?

Limitations

This study is limited by its descriptive nature; not having an appropriate comparison group for calculating relative risks. Racial homogeneity and possible selection bias of the parent COS study may limit general applicability.

Conclusion

This descriptive study looking at characteristics of subjects 78 years who have not developed radiographic knee osteoarthritis demonstrates that it is worthwhile to specifically investigate the factors associated with not developing osteoarthritis (or any disease) separately from investigating the disease risk factors. They may not necessarily be simple inverses. This group of subjects had BMIs in the overweight category at the age where they were evaluated for RKOA. Knee OA risk factor studies have associated overweight/obesity with the development of OA. The complete picture of osteoarthritis may be more complex.

The complexity, may in part, be due to BMI at earlier ages. This group of subjects absent RKOA at age 78 had BMIs well within the WHO “normal” category at age 25 and had weights somewhat less the normal gender/age/year at middle age (age 45). Further study of weight distributions over life-span with respect to RKOA is warranted.

Finally, this study indicated that increased BMI in older ages may be an apparent “inflation” due to decreased height, not just increased weight providing a reasonable explanation for the apparent anomaly of overweight as both a protective and a risk factor for worse outcomes for many chronic and infectious diseases. The effect of decreased height in older age certainly deserves more attention.