Effect of knee osteoarthritis on the perception of quality of life in Venezuelan patients




To measure the perception of quality of life in Venezuelan patients with knee osteoarthritis and to identify those variables that may influence it.


A multicenter, cross-sectional study of 126 mestizo patients with knee osteoarthritis recruited from 8 rheumatology centers in Venezuela. We used a Spanish-translated version of the Arthritis Impact Measurement Scales (AIMS), as adapted in Venezuela. One-way analysis of variance was used to compare the AIMS mean total score among subgroups of knee pain, anatomic stage, and socioeconomic status (SES); a post-hoc test was performed to identify significant intragroup differences. Pearson's correlation coefficient was used to examine correlations between age, body mass index (BMI), disease duration, knee pain, and AIMS score. Associations between radiologic stage, SES, and AIMS scores were examined using Spearman's rank correlation. Multiple regression analysis was used to estimate predictor factors of AIMS scores.


A significant correlation was found between total AIMS scores and knee pain, age, and socioeconomic status, but not with BMI, disease duration, or anatomic stage. Patients with severe knee pain differed from those with mild and moderate pain, and the highest AIMS mean total score was seen in patients within the severe knee pain subset. Patients in the highest socioeconomic levels differed from those within lowest categories. Patients classified as being at the levels of relative and critical poverty showed the highest AIMS scores. Multiple regression analysis showed that knee pain was the only variable that exerted an independent effect on the quality of life in our patients.


The perception of quality of life is negatively affected by increasing levels of joint pain, old age, and low socioeconomic status in Venezuelan patients with knee osteoarthritis. Our study supports the need for an early and vigorous approach to treat pain in this group of patients.


Osteoarthritis (OA) is the most common rheumatic disease in the general population, its prevalence increasing gradually in individuals older than 40 years (1). It has been estimated that 20 million people in the United States have the disease, with a prevalence >60% in subjects older than 70 years (1, 2). In Venezuela, OA was reported in 18% of a population of 32,946 registered cases with assorted rheumatic diseases (Ministry of Health and Social Development 2001, unpublished data). After the hand joints, the knee is the second most common joint involved, and the disease usually evolves with increasing levels of pain, mobility restriction, and physical disability (1, 3). Burgeoning costs derived of reduced labor, early retirement, and expensive medical care, including total joint replacement, are a pressing concern as life expectancy increases all over the world (4, 5). A subtler subject is the indirect costs derived from a decreased quality of life caused by chronic pain and decreased joint function, with its accompanying risks of comorbidity and even increased mortality in frequently depressed and less mobile patients (6, 7).

Quality of life is a complex issue because it is influenced by such variables as age, socioeconomic and education levels, coping strategies, and culture (8). Several instruments have been devised to measure diverse components of quality of life in patients with chronic conditions. For example, the Arthritis Impact Measurement Scales (AIMS) (9) has been widely used in evaluating the impact of several rheumatic diseases, including OA (10). However, most studies have been done in industrialized countries where patient populations show higher education levels and have easier access to medical care. As far as we know, there are no previous studies addressing the quality of life in Latin American patients with knee OA.

The purpose of our study was to examine the perception of quality of life in a population of Venezuelan patients with knee OA by using a Spanish-translated and validated version of AIMS (11) and to identify associations with variables potentially influencing it, such as age, sex, knee pain, body mass index (BMI), disease duration, anatomic stage, and socioeconomic status (SES).


Study design.

This was a multicenter, cross-sectional study with the participation of 8 rheumatology units from regional tertiary-care hospitals in Venezuela belonging to Centro Nacional de Enfermedades Reumaticas, the platform of the National Program for Rheumatic Diseases, under the umbrella of the Ministry of Health and Social Development. Patients were recruited from the outpatient clinics in a consecutive fashion, they agreed to participate, and they signed an informed consent, following the guidelines of our Institutional Review Board. Demographic and clinical data were collected by the participating investigators. A social worker registered socioeconomic status using Graffar's socioeconomic score (12), as adapted in Venezuela by Méndez-Castellanos and Mendez (13). Patients completed the self-administered AIMS questionnaire. Seven provincial tertiary referral rheumatology centers covering all geographic regions in Venezuela and the pilot center in Caracas participated in the study (see Appendix A for participating centers and regional distribution of patients). All the information collected by the participating investigators was mailed to the main center at the University Hospital in Caracas and data were loaded in a central database.


Patients were recruited from a sample of individuals diagnosed with knee OA by their attending rheumatologist, and then selected for the study by any 1 of the participating investigators on the basis of fulfilling the American College of Rheumatology (ACR) criteria (14). Age, sex, and residence site were registered. Disease duration was arbitrarily estimated as the time between the initiation of knee symptoms that led the patient to first consultation in our centers. Patients were Venezuelan, as defined by bearing maternal and paternal Hispanic surnames, traceable back to at least 3 generations of Venezuelan-born ancestors. All patients were mestizo, defined as a variable admixture of Spanish, American native, and African heritage. A total of 151 patients were included initially. Twenty-five patients were excluded due either to lack of adequate knee radiographs (n = 20) or to incomplete AIMS data (n = 5). A total of 126 patients formed the final study population. The inclusion criteria were as follows: 1) meeting the ACR criteria for knee OA; 2) age ≥40 years; and 3) signing an informed consent. Exclusion criteria included the following: 1) associated chronic rheumatic diseases other than knee OA; 2) associated chronic nonrheumatic diseases including high blood pressure (blood pressure ≥140/90 mm Hg in at least 3 measurements), severe obesity (BMI >40 kg/m2), diabetes (blood glucose levels >110 mg/dl in at least 3 measurements), cardiovascular disease (presence of coronary artery disease, congestive heart failure, or cardiomyopathy), and chronic lung disease (asthma and chronic obstructive bronchopulmonary disease); and 3) congenital or hereditary deformity or misalignment of lower extremities. Patients were stratified according to their BMI in the following categories: normal weight (18.5–24.9 kg/m2), overweight or preobesity (25.0–29.9 kg/m2), mild obesity (30.0–34.9 kg/m2), moderate obesity (35.0–39.9 kg/m2), and severe obesity (>40 kg/m2) as previously reported (15).

Quality of life.

Quality of life was measured using the AIMS scale in a version translated to Spanish and validated by us in Venezuela (11). This is a self-administered questionnaire comprising 3 main dimensions related to the perception of health by the patient: physical function, pain, and psychological function. Physical function is estimated by a compounded index measuring mobility, physical activity, social role, social activity, dexterity, and activities of daily living. The pain component comprises a group of items measuring the symptom perception by the patient; the depression and anxiety scales measure psychological status. The validity and reliability of the Venezuelan version was evaluated, comparing favorably to the Texan-American version in a group of Venezuelan patients with rheumatoid arthritis in an outpatient clinic (11).

Knee pain.

Knee pain was evaluated using a visual analog scale (VAS) of 0–10 cm. When testing by one-way analysis of variance (ANOVA), patients were divided into 1 of 4 categories: no pain (VAS = 0), mild pain (VAS = 1–3), moderate pain (VAS = 4–6), and severe pain (VAS = 7–10), otherwise continuous values were used for Pearson correlation coefficient analysis.

Anatomic stage.

The anatomic stage was established in weight-bearing and anteroposterior knee radiographs using the Kellgren and Lawrence classification system (16). Plain knee radiographs from all participating centers were sent to Caracas, and anteroposterior views were read in a blind independent and random fashion by a group of 3 rheumatologists highly experienced in radiology of the rheumatic diseases (BRL, HP, LGS). The total number of films were randomly split into 3 groups and each of them assigned to any of the 3 observers. A total of 252 knees were classified into 1 of 5 categories: 0 (no radiologic changes), 1 (doubtful changes), 2 (minimal changes), 3 (moderate changes), and 4 (severe changes) (16). Agreement between readers was carried out using the Landis and Koch method (17) in a sample of 20 knee radiographs. Intraobserver agreement for radiograph assessments, as measured by the kappa values (17), all differed from zero (observer 1 κ = 0.69, 95% confidence interval [95% CI] 0.45–0.94; observer 2 κ = 0.66, 95% CI 0.28–1.05; observer 3 κ = 0.52, 95% CI 0.12–0.93). The kappa value for interobserver agreement was 0.81 (95% CI 0.69–0.91).

Socioeconomic status.

SES was assessed using Graffar's classification system (12) as validated in Venezuela (13). This instrument comprises the following categories: type of job, level of education instruction achieved, household income, and housing. Patients were classified according to SES into 5 levels in which I is the highest, or upper class, and V the lowest socioeconomic status, or critical poverty. Categories II, III, and IV correspond to middle class, low middle class, and relative poverty, respectively.

Data analysis.

Continuous variables such as age, BMI, disease duration, and knee pain were expressed as mean ± SD. Ordinal variables, such as anatomic stage and social status, were expressed as percentages in each category. For comparison purposes, patients were stratified into 5 anatomic groups using the above-mentioned radiologic criteria, 3 knee pain categories according to VAS, and 5 possible socioeconomic levels according to Graffar's instrument. AIMS mean total scores were established for each subgroup and analysis performed by one-way ANOVA. A post-hoc test was used to identify potentially significant intragroup differences. Pearson's correlation coefficient was used to examine correlations between continuous variables such as age, BMI, disease duration, knee pain, and AIMS scores. Spearman's rank correlation was used for examining the associations between ordinal variables, such as radiologic stage and socioeconomic status, with AIMS scores. Multiple regression analysis was used to estimate the predictor factors of AIMS score. The intercorrelation of independent variables was searched using the variance-inflation factor (VIF), setting a VIF≥ 4 as the cutoff criterion for establishing whether multicollinearity was present (18). Incomplete data were treated and input using a regression model approach as suggested by Cohen and Cohen (19). The data were analyzed using the Statistical Package for Social Science (SPSS, Chicago, IL), 9.0 version. A P < 0.05 was considered significant.


The mean age of the sample was 64 years (range 41–86; Table 1). Most patients were women (84.1%). There were no differences in the AIMS mean total score between female and male patients (39.5 ± 10.8 and 41.4 ± 7.2, respectively). Most patients (71.4%) had a disease duration <10 years; only 4.5% had a disease duration >20 years. Eighty-six percent of patients had increased BMI. As seen in Table 1, the majority of patients (76.3%) could be classified in the categories of either critical (19.6%) or relative (56.7%) poverty. The median anatomic stage according to Kellgren and Lawrence system was 3 (range 0–4).

Table 1. Clinical and demographic data of the study population (n = 126)*
  • *

    BMI = body mass index; AIMS = Arthritis Impact Measurement Scales.

  • Kellgren and Lawrence classification system.

  • As assessed by visual analog scale.

Age, mean ± SD (range) years64 ± 9 (41–86)
Sex, F/M106/20
Disease duration, mean ± SD (range) years6.6 ± 5.2 (0.2–2)
BMI, mean ± SD (range) kg/m230.2 ± 4.5 (20–40)
BMI, subsets, % 
 Normal weight13.7
 Mild obesity36.7
 Moderate obesity14.5
 Severe obesity0.9
Anatomic stage, median (25th–75th percentiles)3 (3–4)
Anatomic stage subsets, % 
Graffar's scale, median (25th–75th percentiles)4 (4–4)
Graffar's scale subsets, % 
 Upper class2.1
 Middle class8.2
 Low middle class13.4
 Relative poverty56.7
 Critical poverty19.6
Knee pain, mean ± SD (range)6.3 ± 2.2 (1–10)
Knee pain subsets, % 
AIMS total scores, mean ± SD (range)39.7 ± 10.3 (17–69)

We performed a bivariate analysis to identify potential associations among different study variables and total AIMS score (Table 2). Significant correlations were found with knee pain (r = 0.47, P = 0.0001), age (r = 0.23, P = 0.02), and socioeconomic status (r = 0.30, P = 0.005). There was no correlation between AIMS scores and BMI (r = −0.06, P = 0.5), disease duration (r = 0.06, P = 0.46), or anatomic stage (r = 0.002, P = 0.98). The highest AIMS mean total score (44.3 ± 10.3) was seen in patients within the most severe knee pain subset (Table 3). Similar results were seen when data were examined using one-way ANOVA for comparison of AIMS mean total score within the 3 categories of knee pain, showing statistically significant differences in AIMS mean total scores in patients with severe compared with moderate or mild knee pain. AIMS score difference, severe versus moderate: −8.43 (P = 0.0001); severe versus mild: –10.35 (P = 0.0001). There were no statistically significant differences between the subgroups with mild and moderate knee pain. Knee pain measured by both the AIMS subscale of pain and the VAS instrument showed a statistically significant correlation (r = 0.47, P = 0.0001). No association was found between the total AIMS score and sex, BMI, disease duration, or anatomic stage (data not shown).

Table 2. Correlation coefficients of study variables and total AIMS scores*
 Total AIMS score
  • *

    Table shows Pearson correlation coefficients with α < 0.05 for age, BMI, disease duration, and knee pain and Spearman rank correlation with α < 0.05 for radiologic stage and socioeconomic status. AIMS = Arthritis Impact Measurement Scales; BMI = body mass index.

Age (n = 109)0.230.02
BMI (n = 107)0.06NS
Disease duration (n = 101)0.06NS
Anatomic stage (n = 94)0.002NS
Knee pain (n = 113)0.470.0001
Socioeconomic status (n = 89)0.300.005
Table 3. AIMS mean total scores and study variable subgroups*
 AIMS total score, mean ± SDP
  • *

    Statistical analysis by one-way analysis of variance; P values refer to statistical differences between variable subgroups. AIMS = Arthritis Impact Measurement Scales.

  • Graffar's socioeconomic score.

  • Kellgren and Lawrence classification system.

Knee pain  
 Mild34.0 ± 8.3 
 Moderate35.9 ± 8.00.0001
 Severe44.3 ± 10.3 
Socioeconomic status  
 Upper class27.5 ± 2.2 
 Middle class32.8 ± 4.5 
 Low middle class33.7 ± 9.20.02
 Relative poverty41.9 ± 11.1 
 Critical poverty41.2 ± 10.9 
Age groups, years  
 40–4937.9 ± 8.4 
 50–5938.1 ± 8.9 
 60–6937.6 ± 10.00.047
 70–7943.9 ± 11.4 
 80–8949.0 ± 11.4 
Anatomic stage  
 032.7 ± 5.8 
 136.9 ± 11.9 
 240.9 ± 9.9NS
 339.4 ± 9.3 
 439.1 ± 11.7 

There was a statistically significant inverse correlation between total AIMS score and patient SES (Table 2). The highest AIMS mean total score was seen in those patients classified as being at the levels of relative (class IV) and critical (class V) poverty (41.9 ± 11.1 and 41.2 ± 10.9, respectively; Table 3). By using one-way ANOVA, we further confirmed this correlation by showing statistically significant differences between AIMS mean total scores obtained by patients of the different socioeconomic levels: Graffar class I versus Graffar class IV and V (P = 0.031 and 0.017, respectively) and Graffar class II versus Graffar class IV (P = 0.001). Finally, there was a direct correlation among total AIMS scores and age (r = 0.23, P = 0.02). The group of patients older than 70 years showed the highest mean AIMS scores (Table 3).

To examine the potential codependency of variables showing correlation with AIMS scores, we performed multiple regression analysis using total AIMS scores as the dependent variable and knee pain, socioeconomic status, age, BMI, disease duration, sex, and anatomic stage as independent variables. As shown in Table 4, knee pain was the variable with greater weight and the only one to exert an independent effect on quality of life as assessed by AIMS.

Table 4. Multiple regression analysis with Arthritis Impact Measurement Scales score as dependent variable and age, knee pain, and socioeconomic status as independent variables*
  • *

    VIF = variance-inflation factor (threshold ≤4); NS = not significant.

Age0.15NS   1.05
Knee pain0.42<0.050.510.26<0.051.12
Socioeconomic status0.13NS   1.14


In this study we have examined the impact of knee OA in the perception of quality of life, as assessed by the AIMS instrument, in a sample of Venezuelan patients recruited from various regions of the country. Our results suggest that pain, SES, and age are the main contributors to a poor perception of quality of life among our patients. There was an inverse correlation between SES and AIMS scores, and a direct correlation between age and AIMS scores. However, knee pain was the variable exerting the greatest weight on AIMS scores, and the only that showed an independent effect among the independent variables.

Quality of life lacks a comprehensive and universal definition because it is significantly influenced by social and cultural differences worldwide (8). In recent years, it has become of subject of intense research, particularly as related to chronic diseases (20, 21). Unlike other instruments designed to measure the functional impact of knee OA, such as the Western Ontario and McMaster Universities Osteoarthritis Index (22), AIMS is a multidimensional instrument that has been previously used to evaluate patients with knee OA (10, 23, 24) and allows a more global assessment of various components of quality of life. Thus, we thought it was better suited to the main purpose of our study.

Pain is a major clinical component in knee OA (25, 26) and, along with age and quadriceps strength, the main predictor of physical functioning (27). It has a negative effect in the sense of well being of patients, irrespective of the presence or absence of joint radiologic changes (26). It also induces an increment of anxiety (28) and even contributes to increased mortality (7). We observed a direct correlation between pain intensity and diminished quality of life as assessed by total AIMS scores (Table 2).

In our sample, patients within the moderate and severe knee pain categories predominated, possibly reflecting the tertiary care-based approach. Consistent with our findings, one previous study showed that either hip or knee chronic pain had a deleterious impact on the quality of life in a group of elderly patients (29). The role of age in the perception of quality of life is complex and poorly defined (30). The decline in quality of life associated with old age seems to be determined by global health status, the presence of physical symptoms related to chronic disease, and diminished functional capabilities (31, 32). These factors converge to limit contact with relative and friends, and thus diminish the level of personal fulfillment and satisfaction. Within this frame, depression frequently ensues (32), further worsening the quality of life in the elderly (33, 34). To minimize the potential effect of comorbid conditions and isolate the influence of knee OA, we excluded some clinical conditions that are highly prevalent in aging patients (see Patients and Methods); those conditions could have exerted a confounding effect on quality of life by adding disease symptoms, side effects associated with concomitant medication, and costs. We are aware that this manipulation affected the generalizability of our study, because OA patients frequently have these comorbid conditions.

We found a direct correlation between total AIMS scores and age (Table 2). Our results do not support a role for depression as assessed by correlating age with the AIMS component that evaluates the affective status of patients (data not shown). However, we cannot fully rule out a role for depression in our patients because no additional criteria to identify and grade depression were used. As far as we know, there are no previous studies measuring the influence of pain and age in the perception of quality of life in Latin American patients with knee OA.

Socioeconomic status is a well-known factor influencing morbidity and mortality (35). Two socioeconomic components contributing to quality of life, namely education and housing, are independently associated with functional level in older persons (36). To our knowledge, no previous study had addressed the participation of socioeconomic factors in Latin American patients with knee OA. Our results showed a highly significant association between the perceived diminished quality of life and SES, i.e., the lower the socioeconomic level, the higher the total AIMS scores (Tables 2 and 3). This correlation was consistent after comparing subgroups along the socioeconomic ladder. The interpretation of this association becomes difficult due to the variety of factors determining socioeconomic level. For instance, education is a major component and, per se, a well-known outcome predictor in chronic diseases (35), in patients with OA in particular (37). The Graffar's socioeconomic score evaluates education by including assessment of the instruction level achieved and occupation. Both categories showed correlation with total AIMS scores (data not shown). Hannan et al showed an inverse correlation between educational achievement and level of pain and radiologic changes in patients with OA (37). Similarly, a low education level negatively correlated with knee pain in patients with knee OA (38). In a regression model, only pain behaved as an independent variable, explaining ∼26% of the variance corresponding to the perception of quality of life in our patients.

Anatomic stage did not show a significant correlation with pain or had an impact on quality of life as assessed in our multivariate analysis. Previous studies have shown no effect (26) or a limited contribution (39–41) of radiographic changes in the level of pain or disability among knee OA patients. In previous studies, there have been contradictory results about the contribution of obesity in health status of patients with knee pain, with some studies showing either association (39–42) or lack of it (26).

In our study, the exclusion of patients with severe obesity may have limited the power to establish a significant effect of BMI in the quality of life of our patients. Further limitations of our study are the relatively small sample size and the potential patient selection bias of studying a convenience, nonrandom sample, making our results relevant only to patients with knee OA attending tertiary-care centers. Also, disease duration of knee OA is very difficult to estimate, and our criteria for establishing it could have hidden a potential significant effect. Finally, the exclusion of comorbid conditions frequent in this patient age group may limit the generalizability of our study.

In summary, our study seems to indicate that the perception of quality of life of Venezuelan patients with knee OA is mainly affected by pain, suggesting the need for vigorous and early therapeutic strategies aimed at effectively treating this symptom. Two additional predictors, but less susceptible to intervention, are old age and low socioeconomic status, both contributing independently.


We appreciate the critical review of the manuscript by Drs. Soham Al Snih and James S. Goodwin (University of Texas Medical Branch, Galveston, TX). We thank Mrs. Mayra Mayora and Ms. Cruz Milano for efficient secretarial assistance.


Table  . Participating Medical Centers
Participating centerCityNumber of patients
Hospital “Jesús María Casal Ramos”Acarigua17
Hospital “Antonio María Pineda”Barquisimeto15
Hospital Universitario de CaracasCaracas22
Hospital “Ruíz y Páez”Ciudad Bolívar14
Hospital “Patricio Alcalá”Cumana6
Hospital “Manuel Núñez Tovar”Maturín17
Hospital Universitario “Dr. Urquinaona”Maracaibo20
Hospital “Luís Razetti”Puerto La Cruz15