Association of weight gain with incident knee pain, stiffness, and functional difficulties: A longitudinal study

Abstract

Objective

To examine the longitudinal association between significant weight change and change in knee symptoms (pain, stiffness, and function), and to determine whether the effects differ in those who are obese and those with osteoarthritis (OA).

Methods

Two hundred fifty subjects ranging from normal weight to obese (body mass index range 16.9–59.1 kg/m²) and no significant musculoskeletal disease were recruited from the general community and weight loss clinics and organizations. Seventy-eight percent were followed at ∼2 years. Weight, height, and knee symptoms (using the Western Ontario and McMaster Universities Osteoarthritis Index) were assessed at baseline and followup. Any weight loss methods were recorded.

Results

Thirty percent of subjects lost ≥5% of baseline weight, 56% of subjects' weight remained stable (loss or gain of <5% of baseline weight), and 14% of subjects gained ≥5% of baseline weight. Using estimated marginal means, weight gain was associated with worsening pain (mean 27.1 mm; 95% confidence interval [95% CI] −1.1, 55.2), stiffness (mean 18.4 mm; 95% CI 1.5, 35.3), and function (mean 99.3 mm; 95% CI 4.0, 194.6) compared to stable weight. Weight loss was associated with reduced pain (mean −22.4 mm; 95% CI −44.4, −0.3), stiffness (mean −15.3 mm; 95% CI −28.50, −2.0), and function (mean −73.2 mm; 95% CI −147.9, 1.3) compared to stable weight.

Conclusion

Weight gain was associated with adverse effects on knee symptoms, particularly in those who are obese and who have OA. Although losing weight is potentially beneficial for symptom improvement, the effects were more modest. Avoiding weight gain is important in managing knee symptoms.

INTRODUCTION

Knee pain is common and affects people of all ages, particularly older adults, and has a major impact on physical and psychological health. A recent survey of the general population in the UK reported a 46.8% prevalence of knee pain in those ages ≥50 years (1). Of these, 25% of subjects reported chronic knee pain, and 6% of the older population reported having severe knee pain or disability (1). Similarly, in a randomly selected cohort of older men and women from Tasmania, Australia, the prevalence of knee pain was 48% (2). Notably, incident knee pain has also been identified as a predictor of a poor functional prognosis (3), highlighting the importance of its prevention of subsequent functional decline and disability. Predictors of incident knee pain include obesity (4, 5), depression (4), widespread pain (4), smoking (5), and previous knee injury (5).

A cross-sectional survey of more than 500 adults ages 16 to >75 years in the UK found a substantial proportion of knee pain to be attributable to overweight and obesity (6). Intense knee pain was reported by 11.7% of this population and intense knee pain with disability was reported by 3.4%. Much of this was related to overweight and obesity, with 21% of all reported knee pain attributable to being overweight or obese, and 36% of knee pain with disability and 37% of intense knee pain with disability attributable to being overweight or obese (6).

Little data are available on the effect of weight change on incident knee pain. One study examined change in body mass index (BMI) in relation to symptomatic osteoarthritis (OA), reporting a significant association between increased BMI and increased odds for incident symptomatic OA (7). Some recent work suggests that weight loss in older OA populations may improve knee pain and function, particularly in obese subjects (8–11). Less is known about the impact of weight change or of further weight gain on those who are already obese or those who have evidence of early OA. The primary aim of this study was to examine the effects of change in weight on change in knee symptoms (pain, stiffness, and function) in an adult population. Secondary aims were to determine whether the effects of weight change on symptoms differ between those who are obese and nonobese or in those with and without early OA, as indicated by the presence/absence of osteophytes, and to determine whether weight change is associated with symptom development, progression, or both.

Significance & Innovations

• No study has examined the effect of weight gain on knee symptoms.

• We demonstrate that further weight gain is associated with the worsening of knee symptoms, but weight loss has only modest effects.

• Preventing weight gain, particularly in those who are already obese, may be important in symptom alleviation.

SUBJECTS AND METHODS

Study population.

Two hundred fifty participants (74% women) were recruited by advertising in the local press, at the hospitals in the waiting rooms of private weight loss/obesity clinics, and through community weight loss organizations in Melbourne, Victoria, Australia. This study aimed to recruit subjects ranging from normal weight to morbidly obese. Because the aim of the study was to examine the effect of significant weight loss on change in knee symptoms, we enriched the population for obese participants and recruited those who were aiming to lose weight, using different methods of weight loss from surgical to commercial weight loss programs, as well as those who were not planning weight loss. The inclusion criterion was age 25–60 years. Subjects were excluded if there was a history of any arthropathy diagnosed by a medical practitioner (including OA), prior surgical intervention to the knee (including arthroscopy), previous significant knee injury requiring non–weight-bearing therapy or requiring prescribed analgesia, malignancy, or contraindication to magnetic resonance imaging (MRI). Recruitment was not dependent on planned weight loss. One hundred ninety-six subjects (78%) completed the followup (median followup time 2.3 years, range 1.6–3.9 years). With 196 subjects and the distribution of weight change as observed in this study, there was 80% power to detect a trend of a 0.31 SD unit change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores per increase of 1 category in weight change (12). The study was approved by the Alfred Hospital Human Research and Ethics committee, the Monash University standing research ethics committee, and the Austin Health Human Research and Ethics Committee. All participants gave informed consent.

Data collection.

MRI of the dominant knee was performed at baseline and followup (13). Knees were imaged in the sagittal plane on a 1.5T whole-body MRI unit (Philips Medical Systems) using a commercial transmit–receive extremity coil. The weight limit for the machine is 150 kg. The following sequence and parameters were used: T1-weighted, fat-suppression, 3-dimensional [3-D] gradient recall acquisition in the steady state (repetition time [TR] 58 msec, echo time [TE] 12 msec, flip angle 55°) with a 16-cm field of view, 60 partitions, 512 × 512–pixel matrix, and acquisition time 11 minutes 56 seconds (1 acquisition). A coronal fat-saturated, fast spin-echo 3-D, T2-weighted acquisition (TR 2,200 msec, TE 20/80 msec, 90° flip angle) with a slice thickness of 3 mm, a 0.3 interslice gap, 1 excitation, a field of view of 13 cm, and a matrix of 256 × 192 pixels was also obtained (14).

Osteophytes were measured from baseline MRIs, which have been shown to be more sensitive than radiographs (15). Osteophytes were measured from coronal images by 2 independent trained observers. In the event of disagreement between the observers, a third independent observer (YW) reviewed the MRI. Intraobserver and interobserver reproducibility for agreement on osteophytes ranged between 0.85 and 0.93 (kappa statistic).

Study participants attended the study center for a baseline and followup visit. At this time, they were asked to complete a questionnaire that included information on their demographic characteristics and physical activity, and had their weight and height measured. Weight was measured to the nearest 0.1 kg (shoes, socks, and bulky clothing removed) using a single pair of electronic scales. Height was measured to the nearest 0.1 cm (shoes and socks removed) using a stadiometer. From these data, BMI (kg/m²) was calculated. Pain, stiffness, and function were assessed by the WOMAC and analyzed using 100-mm visual analog scales (VAS) (12), which were included in the questionnaire. The WOMAC is widely used in community-based studies of adults (16–19). The pain, stiffness, and function subscales comprise 5, 2, and 17 questions, respectively. Each question is assessed on a 100-mm VAS and summed to give a total score of 500 mm for pain, 200 mm for stiffness, and 1,700 mm for function. An increase in score corresponds with worsening of pain, stiffness, and functional difficulties.

Statistical analysis.

The characteristics of the study population were tabulated. Change in weight was defined as weight loss or weight gain of ≥5% of the initial weight in accordance with the US Food and Drug Administration, which considers this level as clinically significant (20). Age, weight at baseline, change in weight, WOMAC subscale scores at baseline, and change in WOMAC scores were assessed for normality. Age and change in WOMAC scores were normally distributed; however, baseline weight, change in weight, and baseline WOMAC scores did not follow a normal distribution. Therefore, nonparametric tests were employed to examine the latter. The difference in weight change between subgroups was assessed using analysis of variance, the chi-square test, or the Kruskal-Wallis test, as appropriate. Two subjects were missing height measurements from which BMI would be calculated and 2 subjects did not have baseline MRI from which osteophytes could be assessed. Results were tabulated (mean ± SEM change) and are displayed graphically (mean change [95% confidence interval (95% CI)]) using estimated marginal means to indicate the relationship between weight change and change in WOMAC subscale scores, adjusted for age, sex, and baseline weight. Estimated marginal means are the predicted value of the outcome when all covariates adjusted for are set at their average value. Stratification was made by obesity status (BMI <30 kg/m² and BMI ≥30 kg/m²), OA status (osteophytes absent or present), and symptoms at baseline (pain/stiffness/function absent or present). Change in weight was examined as a continuous variable using multivariate regression, adjusted for age, sex, and baseline weight. A P value of less than 0.05 (2-tailed) was regarded as statistically significant. All analyses were performed using the SPSS statistical package, standard version 18.0.

RESULTS

Two hundred fifty subjects were recruited and 196 subjects (78%) completed the followup (mean ± SD followup time 2.4 ± 0.4 years). The 250 subjects recruited at baseline had a mean ± SD age of 45.7 ± 9.3 years and comprised 185 women (74%). The median BMI was 33.2 kg/m² (range 16.9–59.1) and 148 subjects (59%) had a BMI ≥30 kg/m². Those who did not complete the followup were younger (mean ± SD age 41.7 ± 9.8 years) compared to those who were followed up (mean ± SD age 46.7 ± 8.9 years) and had higher WOMAC pain and stiffness scores (median WOMAC pain score 40.5 mm [range 0–500.0], median WOMAC stiffness score 16.0 mm [range 0–200.0]) compared to those who completed the followup (median WOMAC pain score 20.0 mm [range 0–480.0], median WOMAC stiffness score 9.0 mm [range 0–195.0]), but did not differ significantly in terms of sex and BMI.

The characteristics of those who completed the followup are shown in Table 1. Of these, 58 subjects (30%) lost ≥5% of their baseline weight, 110 subjects (56%) remained stable in weight (loss or gain of <5% baseline weight), and 28 subjects (14%) gained ≥5% of their baseline weight. The median weight loss was 11.1 kg (range 3.0–49.1) and the median weight gain was 6.7 kg (range 2.2–28.1). Methods for weight loss reported by the subjects include laparoscopic adjustable gastric banding (LAGB) and calorie-restriction diets such as Jenny Craig and WeightWatchers. Twenty-one subjects (36.2%) who lost ≥5% of their body weight had undergone LAGB surgery, compared to only 6 subjects (5.5%) and 2 subjects (7.1%) in the stable weight and weight gain subgroups, respectively. Baseline WOMAC pain, but not stiffness and function, was highest in those who lost ≥5% of their baseline weight.

Table 1. Characteristics of the study population based on weight change category*

	Loss of ≥5% of baseline weight (n = 58)	Stable weight (n = 110)†	Gain of ≥5% of baseline weight (n = 28)	P for difference‡
* Values are the median (range) unless otherwise indicated. † Defined as a loss or gain of <5% of baseline weight. ‡ P for difference in age and body mass index (BMI) was calculated using analysis of variance; P for difference in proportions of sex, BMI, obesity, osteophytes, and laparoscopic adjustable gastric banding (LAGB) was calculated using the chi-square test; and P for difference in baseline weight, change in weight, and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores was calculated using the Kruskal-Wallis test.
Baseline characteristics
Age, mean ± SD years	45.9 ± 9.1	47.7 ± 8.5	44.5 ± 9.7	0.18
Female sex, no. (%)	46 (79.3)	75 (68.2)	27 (96.4)	0.01
BMI, kg/m^2	38.5 (22.5–56.8)	29.3 (17.5–59.1)	29.7 (18.0–58.1)	< 0.001
Obese (BMI ≥30 kg/m^2), no. (%)	43 (74.1)	54 (49.5)	14 (50.0)	0.004
Osteophytes, no. (%)	29 (50.0)	39 (35.8)	6 (21.4)	0.04
Weight, kg
Baseline	104.6 (54.0–166.0)	85.3 (45.0–160.0)	78.7 (49.0–160.2)	< 0.001
Change over 2 years	−11.1 (−49.1 to −3.0)	0.7 (−6.2 to 6.1)	6.7 (2.2–28.1)	< 0.001
LAGB, no. (%)	21 (36.2)	6 (5.5)	2 (7.1)	< 0.001
WOMAC scores at baseline, mm
Pain	29.0 (0–480.0)	11.0 (0–370.0)	14.0 (0–208.0)	0.05
Stiffness	11.5 (0–195.0)	6.5 (0–150.0)	4.0 (0–85.0)	0.16
Function	99.5 (0–1,120.0)	41.0 (0–1,250.0)	46.0 (0–855.0)	0.11

The difference in the mean change of WOMAC pain, stiffness, and function scores by subgroups of weight change is shown in Table 2. Those who lost ≥5% of their baseline weight had reduced (improved) WOMAC scores, whereas those who gained ≥5% of their baseline weight had increased WOMAC scores across all parameters. When we examine those who are obese and nonobese and those with OA and without OA separately, similar results are seen in those who are obese and those with OA. Table 3 and Figures 1, 2, and 3 show the increasing WOMAC scores across the categories of weight change, from weight loss to stable weight and weight gain, adjusted for age, sex, and baseline weight. Similar results were found when we examined change in weight as a continuous variable. For every 1-kg increase in weight, there was a 1.9-mm increase in WOMAC pain score (95% CI 0.9, 2.8; P < 0.001), a 1.4-mm increase in stiffness score (95% CI 0.9, 2.0; P < 0.001), and a 6.1-mm increase in function score (95% CI 2.8, 9.3; P < 0.001), adjusted for age, sex, and baseline weight.

Table 2. Change in Western Ontario and McMaster Universities Osteoarthritis Index scores (mm) over 2 years by subgroups of weight change*

	Loss of ≥5% of baseline weight	Stable weight	Gain of ≥5% of baseline weight	P for trend
* Values are the mean ± SEM unless otherwise indicated.
Total population, n	58	110	28
Pain	−11.2 ± 11.1	4.5 ± 4.6	32.7 ± 14.5	0.02
Stiffness	−2.3 ± 6.2	7.1 ± 3.2	24.8 ± 8.2	0.01
Function	−16.1 ± 37.5	16.6 ± 17.6	127.5 ± 44.1	0.02
Obese, n	43	54	14
Pain	−15.2 ± 14.6	10.7 ± 8.9	59.0 ± 24.8	0.01
Stiffness	−0.5 ± 8.3	13.4 ± 6.2	48.4 ± 13.7	0.01
Function	−9.8 ± 47.0	43.5 ± 30.7	262.9 ± 63.2	0.004
Nonobese, n	14	55	14
Pain	−0.02 ± 10.3	−1.6 ± 3.1	6.4 ± 12.2	0.68
Stiffness	−5.1 ± 4.3	0.9 ± 2.0	1.1 ± 2.8	0.36
Function	−73.3 ± 42.3	−9.3 ± 17.7	−8.0 ± 35.8	0.28
Osteoarthritis, n	29	39	6
Pain	−23.4 ± 19.1	8.9 ± 9.9	64.6 ± 49.9	0.06
Stiffness	−1.3 ± 11.8	17.9 ± 8.0	86.4 ± 21.1	0.003
Function	−4.1 ± 65.2	51.6 ± 32.9	392.4 ± 116.4	0.01
No osteoarthritis, n	29	70	21
Pain	1.1 ± 11.2	1.7 ± 4.8	24.8 ± 13.4	0.15
Stiffness	−3.3 ± 4.5	1.0 ± 2.2	8.2 ± 4.5	0.15
Function	−28.2 ± 38.2	0.8 ± 20.2	57.9 ± 35.5	0.24

Table 3. Adjusted change in Western Ontario and McMaster Universities Osteoarthritis Index scores (mm) over 2 years by subgroups of weight change*

	Loss of ≥5% of baseline weight	Stable weight	Gain of ≥5% of baseline weight	P for trend†
* Values are the mean ± SEM unless otherwise indicated. † Adjusted for age, sex, and baseline weight.
Total population, n	58	110	28
Pain	−15.7 ± 8.9	6.6 ± 6.4	33.7 ± 12.7	0.01
Stiffness	−6.6 ± 5.3	8.7 ± 3.8	27.1 ± 7.6	0.002
Function	−43.1 ± 30.1	30.3 ± 21.5	129.6 ± 42.9	0.01
Obese, n	43	54	14
Pain	−17.7 ± 12.7	12.0 ± 11.2	61.8 ± 22.3	0.01
Stiffness	−2.4 ± 7.7	13.9 ± 6.8	52.1 ± 13.6	0.003
Function	−16.3 ± 40.5	46.8 ± 36.0	270.2 ± 71.4	0.003
Nonobese, n	14	55	14
Pain	0.5 ± 8.3	−0.7 ± 4.2	2.3 ± 8.5	0.95
Stiffness	−5.0 ± 3.9	1.3 ± 2.0	−0.7 ± 4.0	0.37
Function	−80.7 ± 36.7	0.2 ± 18.6	−37.6 ± 37.7	0.15
Osteoarthritis, n	29	39	6
Pain	−32.2 ± 16.5	15.3 ± 13.9	65.5 ± 34.6	0.02
Stiffness	−8.1 ± 10.9	22.3 ± 9.2	90.1 ± 22.9	0.001
Function	−27.1 ± 54.9	69.1 ± 45.8	389.9 ± 113.7	0.01
No osteoarthritis, n	29	70	21
Pain	−0.9 ± 9.6	3.0 ± 6.2	23.1 ± 11.4	0.22
Stiffness	−4.9 ± 3.9	1.8 ± 2.6	7.8 ± 4.7	0.10
Function	−47.8 ± 34.0	11.9 ± 22.0	48.2 ± 40.2	0.15

Figure 1. Relationship between change in weight (loss or gain of <5% of baseline weight) and change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores (shown as the mean change and 95% confidence interval), adjusted for age, sex, and baseline weight, in the total population.

Figure 2. Relationship between change in weight (loss or gain of <5% of baseline weight) and change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores (shown as the mean change and 95% confidence interval), adjusted for age, sex, and baseline weight, in the obese (A) and nonobese (B) subgroups.

Figure 3. Relationship between change in weight (loss or gain of <5% of baseline weight) and change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores (shown as the mean change and 95% confidence interval), adjusted for age, sex, and baseline weight, in the osteoarthritis (OA; A) and no OA (B) subgroups.

Relationship between weight gain and change in WOMAC scores.

When the relationship between weight gain and change in WOMAC scores was explored in comparison to those whose weight remained stable, those who gained weight displayed an increased pain score by an average of 27.1 mm (95% CI −1.1, 55.2; P = 0.06), an increased stiffness score by 18.4 mm (95% CI 1.5, 35.3; P = 0.03), and an increased function score by 99.3 mm (95% CI 4.0, 194.6; P = 0.02), adjusted for age, sex, and baseline weight. Further adjustment for smoking and physical activity did not alter the results. Weight gain remained associated with an increased stiffness score (mean difference 17.3 mm [95% CI 0.7, 33.9]; P = 0.04) and an increased function score (mean difference 98.7 mm [95% CI 5.4, 191.9]; P = 0.04), and there was a trend for an association with increased pain (mean difference 24.7 mm [95% CI −2.9, 52.2]; P = 0.08), when adjusted for physical activity. Similar results were seen in those who were obese and those with OA, defined as having an osteophyte on MRI (Figures 2 and 3).

Furthermore, we examined those with and without symptoms at baseline separately to determine whether weight gain is associated with the development or progression of symptoms, in comparison to maintaining stable weight. The association between weight gain and increased pain was predominantly in those without pain at baseline (mean difference 44.2 mm [95% CI 20.7, 67.6]; P < 0.001), adjusted for age, sex, and baseline weight. Similarly, the association between weight gain and increased stiffness was predominantly in those without stiffness at baseline (mean difference 23.0 [95% CI 6.3, 39.7]; P = 0.01), and the association between weight gain and worse function was predominantly in those without functional difficulties at baseline (mean difference 99.3 [95% CI 20.8, 177.8]; P = 0.01). No associations were found in those with symptoms at baseline.

Relationship between weight loss and change in WOMAC scores.

The relationship between weight loss and change in WOMAC scores was also explored in comparison to those whose weight remained stable (Figure 1). Those who lost weight showed improvements in pain (mean difference −22.4 mm [95% CI −44.4, −0.3]; P = 0.05), stiffness (mean difference −15.3 mm [95% CI −28.5, −2.0]; P = 0.02), and function scores (mean difference −73.2 mm [95% CI −147.9, 1.3]; P = 0.054) compared to those who remained stable in weight. Weight loss remained associated with reduced pain (mean difference −23.1 mm [95% CI −45.2, −0.9]; P = 0.04) and stiffness scores (mean difference −15.4 mm [95% CI −28.7, −2.1]; P = 0.02), and there was a trend for an association with a reduced function score (mean difference −69.7 mm [95% CI −144.5, 5.2]; P = 0.07), when adjusted for physical activity. No similar results were seen for pain and stiffness when those who are obese and nonobese (Figure 2) and those with OA and without OA (Figure 3) were analyzed separately, although a trend for an association of improvements in function with weight loss remained in the nonobese subgroup (P = 0.06).

When we examined those who lost weight via LAGB compared to those who lost weight via other nonsurgical means (e.g., dietary), those who had LAGB lost more weight (median −16.8 kg, range −49.1 to −7.5) than those who did not (median −9.1 kg, range −36.2 to −3.0). Due to the modest numbers of those who have had LAGB, particularly in the stable weight and weight gain subgroups, we were unable to stratify by LAGB status. However, when we excluded those who have had LAGB, similar results were found with regard to improvements in WOMAC pain (mean difference −28.4 mm [95% CI −53.8, −3.0]; P = 0.03) and function scores (mean difference −84.5 mm [95% CI −165.7, −3.2]; P = 0.04) in those who lost weight compared to those with stable weight.

DISCUSSION

In this population that was recruited across a spectrum of weight from normal weight to obese, we found that gaining ≥5% of baseline weight over 2 years was associated with increases in WOMAC pain, stiffness, and function scores, and was predominantly in those subjects with no symptoms at baseline. The relationship between weight gain and the worsening of knee symptoms seemed particularly important in those who were already obese and who had evidence of OA. Although weight loss had a beneficial effect, the magnitude of improvement was less than the increase in symptoms seen in those who gained weight (Table 3). This study highlights the importance of obesity and weight gain on knee symptoms.

Previous studies have shown a linear relationship between BMI and incident knee pain (4, 21–23). However, there is a paucity of data regarding change in weight and knee pain. Previous research relating to change in weight and knee pain, stiffness, and function has focused on the potential benefits of weight loss (8, 9, 11) or on the effect of weight gain on the risk of OA (24, 25). To our knowledge, there is no study that has examined the effect of weight gain on WOMAC scores, although one study reported a positive association between increased BMI and an increased risk of subsequent development of symptomatic knee OA (7), where symptomatic knee OA was indicated by the presence of pain in or around the knee on most days of the month combined with the presence of radiographic OA (at least definite osteophytes and possible joint space narrowing). In this study, we found that increased weight over 2 years was associated with an increase in not only knee pain, but also stiffness and function. These results were strengthened when an adjustment was made for baseline weight compared to when adjusted for age and sex only (data not shown), indicating that the relationship between weight gain and increase in knee symptoms was influenced by the individual's weight at baseline. The ability to differentiate risk factors for development and progression of disease can be important given that they may identify different populations at risk. Therefore, we stratified our population into those with and without knee symptoms at baseline. We found that weight gain was associated with knee pain, stiffness, and function, particularly in those without symptoms at baseline, which indicates that weight gain plays a greater role in symptom development and less so in symptom progression. This suggests that preventing weight gain is particularly important in those who do not currently have knee symptoms to prevent the incidence of knee pain, stiffness, and functional difficulties. Furthermore, the relationship between weight gain and worse symptoms was observed specifically in those who were already obese and in those with OA, which indicates that those who are already at risk of knee symptoms are especially vulnerable to further adverse effects associated with weight gain.

Weight gain may affect the knee joint via changes to the biomechanical components of gait (26), including the external knee adduction moment (27), which determines the load distribution on the medial compartment of the knee (27). Weight gain may also affect body composition, with increasing obesity largely due to increased fat mass. This increase in adipose tissue may have a metabolic effect on joints via the dysregulation of cytokine production by adipose tissue (28, 29), such that there is an overproduction of proinflammatory cytokines in obese people (29) that has been linked to hyperalgesia and the development and progression of chronic pain (30, 31). Therefore, it is not surprising that further weight gain in obese individuals or individuals with early OA would be related to worsening of knee WOMAC scores, since it is likely that there was already existing damage to the joint.

In this study, we also found moderate benefits of weight loss on knee pain, stiffness, and function. It is important to note that this population includes participants without existing knee OA. Previous studies that have examined the effects of weight loss on WOMAC scores have predominantly been in older cohorts with symptomatic radiographic OA. The Arthritis, Diet, and Activity Promotion Trial (ADAPT) reported improvements in WOMAC pain and function scores with modest weight loss (average of 4.9–5.7% of initial weight over 18 months) and moderate physical exercise in a cohort with a mean age of ∼69 years (9). Similarly, in a study of OA patients (mean age 44 years) involved in a gastric surgery program, participants lost a mean of 20% of their initial body weight over 6 months, which corresponded with significant decreases in WOMAC pain, stiffness, and function scores (11). A trial of older patients with OA (mean age 62.6 years) comparing those on a low-energy diet and controls found a mean difference in weight change of 6.6 kg between the 2 groups over a period of 8 weeks (8). Those in the intervention group had a significantly greater decrease in WOMAC function score compared to the controls, although no difference was found for WOMAC pain and stiffness scores. Another trial of obese subjects ages >60 years with self-reported OA comparing weight loss and stable weight groups reported a mean weight loss of 8.5% in the weight loss group, which corresponded with improvements in pain and physical function as assessed by self-report (WOMAC) and performance tasks (32). Although the general effect of weight loss in our study was similar to previous studies, there are a number of differences between the studies in terms of the intervention examined, the study population, and the amount of weight loss observed. For example, the ADAPT cohort had only modest weight loss and moderate exercise; therefore, separating the effect of weight loss and physical activity is difficult (9). The 2 other studies were conducted in OA populations, including one study with a very large weight loss of 20% of initial weight with gastric surgery, which showed a reduction in pain, while the other study showed a more modest weight loss over a short period of only 8 weeks that was not associated with a reduction in pain, although the WOMAC function score decreased (improved function). Taken together, these data would suggest that weight loss is associated with some improvement in symptoms, but that large amounts of weight loss are required to have significant benefits.

Much of the focus in management of obesity-related morbidities has been on the promotion of weight loss. However, obesity rates remain on the rise, which suggests that current programs and initiatives designed to combat obesity have not been successful (33). Losing weight is difficult and once achieved, maintaining it long term is questionable (34, 35). It has previously been reported that a higher percentage of weight loss was significantly associated with a higher percentage of weight regain, with a 2.8-fold greater risk of weight regain in participants who lost >20% of their body weight compared to those who lost 10–15% of their body weight (35). A systematic review examining the relationship between weight loss during a lifestyle intervention and the maintenance of that weight loss after 1 year of unsupervised followup in 12 studies found that the average percentage of weight loss maintained was 54% (34). There has been a growing interest in the role of weight gain prevention as a means to curb the rising obesity rates (36, 37). Lifestyle programs, such as mail and phone interventions, that have performed disappointingly in weight loss trials may perform better in preventing weight gain (38). Evidence suggests that small reductions in conscious energy intake and increases in physical activity can reduce weight gain and would be considered achievable for most people (33). Although our results suggest a beneficial effect of weight reduction on knee stiffness and function, this effect was modest. Given that weight gain is associated with adverse effects on all WOMAC parameters, without dismissing the importance of weight loss for overall health, particularly in obese individuals, perhaps at the very least a weight maintenance strategy should be employed that may confer significant benefits with regard to knee symptoms.

A potential limitation of our study is the relatively modest number of men, although when we excluded the men, similar effects persisted in the women. We also had small numbers in some of the subgroups that would be interesting to explore further. For example, we were unable to explore whether the relationship between weight gain and knee symptoms would differ by categories of percentage of weight gain. The prevalence of obesity in this population was higher than in the general Australian community (39). In this study, we enriched recruitment for those who were obese in order to have power to examine the effect of weight change across the spectrum of normal weight to obese individuals. In addition to recruiting from weight loss clinics and organizations, we also recruited via advertising in the local press. We excluded those with a history of arthropathy and previous surgery or significant knee injury. Therefore, it may be that the results in the obese population were underestimated, since those taking part are likely to be biased toward being more focused on losing weight. Of note, we did not recruit our subjects based on symptoms, and therefore there was a proportion of subjects without symptoms at baseline who would be unable to improve. However, when we excluded those without symptoms at baseline, similar results were found. Nevertheless, our main findings pertained to the relationship between weight gain and the worsening of symptoms, which would not be affected by subjects without symptoms. There were other potential confounders, such as depression and socioeconomic status, that we were not able to adjust for; therefore, the potential role of these factors in the association between weight gain and WOMAC scores warrants further investigation. Nevertheless, our study also has several strengths, in that this was a younger population and that those with clinically significant knee disease were excluded.

This study demonstrates that gaining ≥5% of baseline weight is associated with adverse effects on knee symptoms (pain, stiffness, and function), particularly in those who are already obese and who have OA. Although losing ≥5% of baseline weight is potentially beneficial for improvement in knee symptoms, the effects are more modest. Worsening of knee symptoms due to weight gain may impair physical functioning and further exacerbate weight gain; therefore, this provides an at-risk time when maintaining weight, which is easier to achieve than weight loss, may be important in the management of knee pain.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Cicuttini had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Wluka, Proietto, Dixon, Jones, Cicuttini.

Acquisition of data. Tanamas, Wluka, Davies-Tuck, Wang, Strauss, Proietto, Cicuttini.

Analysis and interpretation of data. Tanamas, Wluka, Davies-Tuck, Strauss, Dixon, Jones, Forbes, Cicuttini.

Acknowledgements

We would like to thank Judy Hankin for recruitment of study subjects, the MRI unit at Epworth Hospital for their cooperation, and Kevin Morris for technical support, with a special thank you to Cate Lombard for her professional contribution to the revision of the final draft. We are grateful to all members of the research team and the study participants who made this study possible.