Impact of lower limb osteoarthritis on health-related quality of life: A cross-sectional study to estimate the expressed loss of utility in the Spanish population.

OBJECTIVE: Osteoarthritis of the lower limb (OALL) worsens health-related quality of life (HRQL), but this impact has not been quantified with standardized measures. We intend to evaluate the impact of OALL on HRQL through measures based on individual preferences in comparison to the general population.
METHODS: A cross-sectional study was designed. A total of 6234 subjects aged 50 years or older without OALL were selected from the Spanish general population (National Health Survey 2011-12). An opportunistic sample of patients aged 50 years or older diagnosed with hip (n = 331) or knee osteoarthritis (n = 393), using the American Rheumatism Association criteria, was recruited from six hospitals and 21 primary care centers in Vizcaya, Madrid and Tenerife between January and December 2015. HRQL was measured with the EQ-5D-5L, and the results were transformed into utility scores. Sociodemographic variables (age, sex, social group, cohabitation), number of chronic diseases, and body mass index were considered. The clinical stage of OALL was collected using the Western Ontario and McMaster Universities Osteoarthritis Index and the Oxford hip score and Oxford knee score. Generalized linear models were constructed using the utility index as the dependent variable.
RESULTS: HRQL expressed by OALL patients was significantly worse than this of the general population. After adjustment for sociodemographic and clinical characteristics, the mean utility loss was -0.347 (95% CI: -0.390, -0.303) for osteoarthritis of the hip and -0.295 (95% CI: -0.336, -0.255) for osteoarthritis of the knee. OALL patients who were treated at a hospital had an additional utility loss of -0.112 (95% CI: -0.158, -0.065).
CONCLUSION: OALL has a great impact on HRQL. People with OALL perceive a utility loss of approximately 0.3 points compared to the general population without osteoarthritis, which is very high in relation to the utility loss reported for other chronic diseases and for arthritis in general.

Introduction

Health is a fundamental component of well-being in developed societies. The definition of health proposed by the World Health Organization in 1948 surpassed the biological concept of illness and included aspects of physical, psychological and social well-being [1], making it necessary to incorporate new measures to evaluate health status. Health-related quality of life (HRQL) appears to be a necessary measure for assessing an individual’s perceived well-being. Although there is no universally accepted definition of HRQL, it has been suggested that measurements of HRQL should incorporate the subject’s perception of his or her health situation, the impact of the disease or its treatment on his or her well-being and how that impact affects his or her functionality [2].

The assessment of HRQL is fundamental in evaluating the impact of disease and health intervention outcomes on both the individual and society as whole. Measures based on patient individual preferences are essential tools in these evaluations [3]. They allow patients to describe the impact of poor health and to calculate the "utility" scores (or rates) associated with each description of health status, making it possible to attribute value to such states through the study of preferences.

The most widely used tool for the measurement of HRQL in Spain is the EQ-5D [4], a questionnaire based on individual preferences that, through an algorithm, allows the attribution of a “utility” score to each described health state. This questionnaire is the most widely used tool for analyzing the cost-effectiveness of health care technologies [5]. Some European health technology assessment organizations, such as the National Institute for Health and Care Excellence (NICE) in the United Kingdom, have specifically stated that the EQ-5D is the preferred measure of HRQL in adults; thus, utilizing it ensures comparability among studies [6,7]. In Spain, the scores or utilities derived from the latest version of this tool, the EQ-5D-5L, have been proposed to provide information on economic evaluations of technologies [8].

The assessment of perceived health status in patients with chronic diseases is essential for measuring the impact and burden of disease, as in the case of patients with osteoarthritis of the lower limb (OALL) affecting the hip and knee. OALL represents a widely recognized public health burden. It is the eleventh most common cause of disability worldwide, and the disability-adjusted life years (DALYs) lost to it increased by 34.8% from 2005 to 2015 [9]. In Spain, a prevalence of 45.0% for knee osteoarthritis and 24.1% for hip osteoarthritis has been reported in people over 65 years of age [10]. Annual costs of €1,500 were estimated in 2007 for patients with osteoarthritis of the knee or hip in Spain, of which 86% were direct costs [11]. Other studies estimate costs of approximately €5,000 per year for patients in Europe and €12,000 per year in the US (2013 euros) [12]. Currently, the health care costs for generalized osteoarthritis account for between 0.25% and 0.50% of the Spanish gross domestic product (GDP) [13]. Furthermore, it is expected that the prevalence of OALL will continue to increase in the future, due largely to the aging of the population [14] and the increased prevalence of obesity [15]. As a result, this condition will likely continue to present substantial challenges for health planning in the coming years. Reliable measurements of health outcomes in OALL are increasingly important for health decision makers, health professionals and patients [16]. An essential element for advancing knowledge of the impact of OALL is the availability of measures that can be used to accurately and thoroughly measure the individual burden posed by this disease [17].

In Spain, HRQL has been assessed in patients with OALL who consulted trauma specialists [18] and in older people with OALL in the community setting [19], and the results suggest that this disease is associated with a marked decrease in perceived well-being; however, to date, no study has determined what differences exist between this group and the rest of the community to isolate the effects of the disease from those produced by concomitant sociodemographic or clinical conditions.

This study aims to evaluate the impact of OALL on HRQL through measures based on individual preferences in comparison to the general population that does not have osteoarthritis.

Materials and methods

Design

A cross-sectional study was designed. The included population was collected from a random sample of the Spanish general population and from a sample of patients with OALL. The general population data were obtained by selecting the population aged 50 years or older from the 2011–2012 National Health Survey (NHS), of which the methodological bases are publicly available [20]. Of the 21,007 available surveys, 6,234 responses of subjects 50 years or older who stated that they did not have a medical diagnosis of osteoarthritis, arthritis, or rheumatism were chosen. To study patients with OALL, information was obtained through an opportunistic sampling of patients diagnosed with osteoarthritis of the hip or knee according to the American Rheumatism Association criteria [21,22] who consulted traumatology and rheumatology specialists at six hospitals and 21 primary care centers in Vizcaya, Madrid and Tenerife between January and December 2015. Patients with malignant or other organic diseases or psychiatric disorders that hindered participation and those who could neither read nor understand Spanish were excluded. Of the total subjects included (n = 758), only those 50 years of age or older were selected (n = 724).

Data regarding people with OALL were collected directly from the patients, and data regarding the general population were extracted from the NHS.

This sample size allowed the construction of explanatory models that were appropriate for achieving the proposed objective [23].

Dependent variable

To measure the perceived HRQL in both samples, the EQ-5D-5L was used [4]. In its most current version, the EQ-5D-5L [24] consists of two parts: a 0-to-100 scale to assess HRQL, the Visual Analog Scale (VAS) and a questionnaire comprising 5 questions or domains (mobility, self-care, carrying out usual activities, pain/discomfort, and anxiety/depression) with 5 response levels (ranging from 1, no problems, to 5, impossibility or severe problems). From these five questions, 3,125 health states are obtained, and the score associated with each state is the utility index. Initially, it was considered that the utility index should range between 0, a state equivalent to death, and 1, which represented perfect health. However, individual preference studies of different health states have identified states that are less preferable than death for the general population, which means this index can yield negative values (up to -0.421) using the algorithms published for Spain [8]. The reliability, validity and sensitivity of the EQ-5D-5L have been studied in patients with OALL in Spain [25].

Independent variables

The age and sex of each subject and his or her social group, based on a six-category classification related to occupation (with group I being the highest and group VI the lowest) [26], were collected as sociodemographic variables. Information on cohabitation was included, and those who lived alone were differentiated from those living with someone (in any form of cohabitation).

Perceived health status was assessed using a Likert scale of five categories (very good, good, fair, poor and very poor) and the number of chronic diseases diagnosed. The patients’ body mass index (BMI) and whether they had undergone a primary care or hospital consultation were also studied.

Only for the OALL group was it possible to determine whether there was unilateral or bilateral involvement and whether osteoarthritis had been diagnosed in the other large lower limb joint (hip or knee). The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) [27] and the Oxford knee score (OKS) [28] and Oxford hip score (OHS) [29] questionnaires were used to characterize the clinical stage of OALL. The WOMAC is a multidimensional scale comprising 24 items measuring the domains of pain (five items), stiffness (two items) and physical function (17 items) in patients with OALL [27]. We used the version with five response levels, scored from 0 to 4, representing different degrees of intensity (none, mild, moderate, severe or extreme) for each item. Those scores are summed and standardized to yield a score from 0 to 100 (from better to worse capacity). The higher the score, the worse the patient’s status is. This questionnaire has been adapted and validated in Spain [30]. The OKS and OHS scales measure the severity of symptoms in patients with osteoarthritis of the knee and hip, respectively. Each scale consists of 12 questions, and the scores, which range from 0 to 48 points, classify the clinical situation of osteoarthritis patients in 4 groups: excellent (> 41 points), good (between 34 and 41 points), moderate (between 27 and 34 points) and poor (<27 points) [31]. The Spanish versions of the OKS and OHS questionnaires have also been validated [32,33].

Analysis

We present the descriptive statistics of the explanatory and dependent variables with their measures of central tendency and dispersion. The qualitative variables were compared using chi-squared tests, and the quantitative variables were compared using Student’s t test or, if necessary, its nonparametric equivalent.

To address the main objective, generalized linear models (GLMs) were constructed, with the utility index attributable to the subject’s perceived health status as the dependent variable. To select the best model, the Akaike information criterion (AIC) and the Bayesian information criterion (BIC) were examined. The Gaussian family and “identity” link function were selected as the more appropriate ones using these criteria. Standard errors (SEs) were calculated using robust methods to prevent possible heteroscedasticity [3,34].

Model 1 included variables related to the reported HRQL and the presence of OALL, model 2 added the affected joint, and model 3 also included whether the patient had been evaluated in the hospital setting. Improvements in model fit were compared by calculating Akaike weights, which express the probability that a new model is better than the set of tested models [35]. The improvement in the BIC was also studied according to the interpretations proposed by Kass and Raftery [36]. The variable “perceived health status” was not included as an explanatory variable due to the risk of overfitting when measuring the same construct as the dependent variable. Neither bilaterality nor having another joint group affected by OALL (hip or knee) improved the explanatory power of the final model (model 3). Patients with any missing data were not included in the models. The results section explains the number of subjects included in each model.

Results

The results are presented for 6958 subjects aged 50 or older: 6234 from the general population without osteoarthritis and 724 (10.41% of the total) with a diagnosis of OALL. Of the patients diagnosed with OALL, 393 (54.28%) had a diagnosis of osteoarthritis of the knee, of whom 53 (13.49%) had a previous diagnosis of osteoarthritis of the hip. Of the 724 patients with OALL, 331 (45.72%) had a diagnosis of osteoarthritis of the hip, of whom 115 (34.74%) had a previous diagnosis of osteoarthritis of the knee. A total of 286 patients with OALL were recruited from primary care clinics (39.50%), and 438 (60.50%) were recruited from specialized care practices: 369 (50.97%) from traumatology and 69 (9.53%) from rheumatology practices.

Table 1 shows the characteristics of the total sample studied. Among the patients with OALL, women, people older than 65 years, and more disadvantaged social groups predominated. There was also a higher prevalence of overweight and obesity and of cohabitation with some type of partner.

Table 1

Characteristics of the studied sample.

	General population N = 6234	Population with OALL N = 724	p
Age (%)
50–54 years	21.82	4.70	<0.001
55–59 years	18.53	8.15
60–64 years	15.77	13.12
65–69 years	13.97	16.57
70–74 years	9.62	17.54
75–89 years	8.87	20.99
80–84 years	6.26	13.40
≥85 years	5.17	5.52
Age, mean (SD)	64.44 (10.81)	70.93 (9.13)	<0.001
Sex (%)
Female	47.06	62.71	<0.001
Social group (%)
Group I	11.64	7.12	<0.001
Group II	7.93	4.56
Group III	19.00	16.79
Group IV	15.22	20.62
Group V	32.02	29.38
Group VI	14.20	21.53
Cohabitation (%)
With partner	61.93	66.99	<0.001
Chronic diseases (%)
None	17.85	0.00	<0.001
One	22.44	55.11
Two	20.82	30.39
Three or more	38.88	14.50
Chronic diseases, mean (SD)	2.34 (2.02)	1.66 (0.91)	<0.001
BMI (%)
Underweight	1.03	0.14	<0.001
Normal	31.18	19.06
Overweight	40.02	43.23
Obese	27.77	37.57
BMI, mean (SD)	26.69 (4.09)	29.02 (4.72)	<0.001
Perceived health status (%)
Very good	13.60	2.23	<0.001
Good	55.25	22.67
Fair	23.24	48.26
Poor	6.42	21.56
Very poor	1.49	5.29
EQ-5D-5L, utilities, mean (SD)	0.924 (0.160)	0.532 (0.287)	<0.001
EQ-5D-5L, VAS, mean (SD)	74.71 (17.71)	56.31 (21.70)	<0.001

BMI: Body Mass Index; SD: Standard Deviation; VAS: Visual Analog Scale.

The perceived health status and utility expressed by patients with OALL were significantly worse than those expressed by the general population in terms of the transformation of the EQ-5D-5L into utilities and the VAS (Table 1) and of each domain of the questionnaire (Table 2). The differences in HRQL were most pronounced for the domains mobility, carrying out usual activities, and pain/discomfort. The range of utilities for the expressed health states ranged from -0.416 (state 55555) to 1 (state 11111) for both the general population and the patients with OALL. In the case of patients with osteoarthritis of the knee, the worst health state reported was 44555, which has a utility value of -0.297. The responses to the VAS of the EQ-5D-5L ranged from 0 to 100 for both the general population and the patients with OALL (both hip and knee).

Table 2

Distribution of the responses to the different domains of the EQ-5-5L for the general population and the population with lower limb osteoarthritis (OALL).

EQ-5D-5Ldomain	Percentage of responses per level					p
Mobility	1	2	3	4	5
General populationn = 6234	84.62	7.92	4.23	2.29	0.93	<0.001
OALL n = 723	10.93	20.06	43.02	23.93	2.07
Self-care	1	2	3	4	5
General populationn = 6234	93.55	2.98	1.44	0.90	1.12	<0.001
OALL n = 722	28.39	27.70	31.58	11.36	0.97
Usual activities	1	2	3	4	5
General populationn = 6234	89.09	5.21	2.68	1.57	1.44	<0.001
OALL n = 722	16.48	25.21	36.70	16.62	4.99
Pain/discomfort	1	2	3	4	5
General populationn = 6231	76.46	14.35	6.55	2.33	0.32	<0.001
OALL n = 723	6.22	21.30	37.07	30.57	4.84
Anxiety/depression	1	2	3	4	5
General populationn = 6226	85.21	9.14	3.82	1.48	0.35	<0.001
OALL n = 716	44.13	23.18	18.30	11.31	3.07

Table 3 shows the level of severity for patients with osteoarthritis of the knee or hip as measured with the WOMAC and the OHS/OKS [30,32,33]. Those with osteoarthritis of the knee more frequently had bilateral involvement, but those with osteoarthritis of the hip more commonly showed concomitant involvement of the other large lower limb joint. There were no significant differences between the patients with knee and hip osteoarthritis in the utilities or the level of severity measured with specific instruments. Only the VAS of the EQ-5D-5L showed a trend of patients with arthritis of the hip presenting worse perceived HRQL than those diagnosed with osteoarthritis of the knee, although the difference was not significant.

Table 3

Characteristics of patients with OALL included in the study.

	Patients with hip osteoarthritisn = 331	Patients with knee osteoarthritisn = 393	p
Bilateral involvement (%)	28.40	42.24	<0.001
Involvement of other large lower limb joint (hip or knee) (%)	34.74	13.49	<0.001
State according to Oxford score (%)
Excellent	4.56	3.32	0.348
Good	11.85	8.42%
Moderate	18.24	19.64
Poor	65.35	68.32
EQ-5D-5L utilities, mean (SD)	0.517 (0.303)	0.543 (0.272)	0.110
EQ-5D-5L VAS, mean (SD)	54.92 (21.78)	57.47 (21.59)	0.059
Oxford score, mean (SD)	22.72 (10.58)	21.97 (9.94)	0.163
WOMAC score, mean (SD)
WOMAC score, pain	45.82 (22.63)	47.09 (20.55)	0.216
WOMAC score, limitation	52.65 (22.81)	51.14 (20.99)	0.178
WOMAC score, stiffness	48.33 (25.97)	46.81 (25.29)	0.213
WOMAC score, total	50.85 (22.02)	49.95 (20.10)	0.282

SD: Standard Deviation; VAS: Visual Analog Scale. WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index.

Table 4 shows the results of the explanatory models for the expressed utilities. Model 3 emerged as the best model because its Akaike weights, compared to those of the set of models and model 2, had a value of 1; that is, there is statistical certainty that it is the best model. In terms of improvement in the BIC, the evidence against the goodness of fit of models 1 and 2 was very strong (with BIC differences well above 10 in both cases). The three models, especially model 3, considerably reduced the error variance, as shown by the value of McFadden’s adjusted R2.

Table 4

Explanatory models of the differences in utility derived from the EQ-5D-5L.

	Model 1	Model 2	Model 3
Variable	Coef (CI 95%)	Coef (CI 95%)	Coef (CI 95%)
Level of care
Hospital care vs other			-0.112 (-0.158, -0.065)
Lower limb OA
Hip vs none		-0.406 (-0.444, -0.369)	-0.347 (-0.390, -0.303)
Knee vs none		-0.365 (-0.396, -0.334)	-0.295 (-0.336, -0.255)
Lower limb OA
Yes vs No	-0.384 (-0.409, -0.360)	-	-
Age
55–59 vs 50–54 years	-0.004 (-0.014, 0.006)	-0.004 (-0.014, 0.006)	-0.004 (-0.014, 0.006)
60–64 vs 50–54 years	0.007 (-0.003, 0.017)	0.007 (-0.003, 0.017)	0.007 (-0.003, 0.018)
65–69 vs 50–54 years	0.013 (0.002, 0.024)	0.013 (0.002, 0.024)	0.013 (0.002, 0.024)
70–74 vs 50–54 years	0.006 (-0.008, 0.020)	0.005 (-0.009, 0.019)	0.005 (-0.009, 0.019)
75–89 vs 50–54 years	0.004 (-0.012, 0.019)	0.003 (-0.012, 0.018)	0.003 (-0.012, 0.018)
80–84 vs 50–54 years	-0.062 (-0.084, -0.040)	-0.062 (-0.084, -0.040)	-0.064 (-0.086, -0.042)
≥85 vs 50–54 years	-0.148 (-0.182, -0.114)	-0.148 (-0.182, -0.114)	-0.150 (-0.184, -0.117)
Sex
Male vs female	0.021 (0.014, 0.029)	0.022 (0.014, 0.030)	0.022 (0.014, 0.030)
Social class
Group II vs group I	-0.008 (-0.022, 0.006)	-0.009 (-0.022, 0.005)	-0.010 (-0.024, 0.004)
Group III vs group I	-0.021 (-0.034, -0.008)	-0.021 (-0.034, -0.008)	-0.021 (-0.034, -0.008)
Group IV vs group I	-0.017 (-0.029, -0.004)	-0.017 (-0.030, -0.004)	-0.017 (-0.030, -0.004)
Group V vs group I	-0.023 (-0.034, -0.012)	-0.023 (-0.035, -0.012)	-0.023 (-0.035, -0.012)
Group VI vs group I	-0.024 (-0.037, -0.010)	-0.024 (-0.038, -0.010)	-0.024 (-0.038, -0.010)
Cohabitation
With partner vs alone	0.010 (0.002, 0.018)	0.010 (0.002, 0.019)	0.010 (0.002, 0.019)
Chronic diseases
One vs none	-0.012 (-0.020, -0.005)	-0.012 (-0.020, -0.004)	-0.010 (-0.018, -0.003)
Two vs none	-0.028 (-0.037, -0.019)	-0.028 (-0.037, -0.019)	-0.030 (-0.039, -0.021)
Three or more vs none	-0.094 (-0.103, -0.085)	-0.094 (-0.102, -0.085)	-0.093 (-0.102, -0.084)
BMI
Underweight vs nonobese obesemal weight	-0.048 (-0.102, 0.006)	-0.047 (-0.101, 0.007)	-0.047 (-0.100, 0.006)
Overweight vs nonobese	0.005 (-0.004, 0.013)	0.004 (-0.004, 0.013)	0.004 (-0.005, 0.012)
Obese vs nonobese	-0.028 (-0.039, -0.017)	-0.028 (-0.039, -0.018)	-0.029 (-0.040, -0.019)
Characteristics of the model.Family: GaussianLink function: IdentityFami	N = 6542AIC = -5522.47BIC = -5373.18McFadden’s R2a = 0.549	N = 6542AIC = -5529.77BIC = -5373.69McFadden’s R2a = 0.553	N = 6542AIC = -5593.02BIC = -5430.15McFadden’s R2a = 0.562

CI: Confidence Interval; BMI: Body Mass Index.

Model 3 shows how the presence of hip osteoarthritis decreases utility by an average of 0.347 points for people who are similar in terms of age, sex, social group, state of cohabitation and burden of chronic illness. For patients with knee osteoarthritis, the decrease in this value is 0.295 points on average. For any type of osteoarthritis, being treated in the hospital environment is associated with an average decrease in utility of 0.112 points, adjusted for the characteristics mentioned above.

Factors associated with a lower level of utility, in addition to OALL and being treated in a hospital environment, included being older (over 80 years), belonging to a more disadvantaged social group, being female, being obese, living alone, and presenting more chronic conditions.

Discussion

This is the first study in Spain to quantify the impact of OALL on HRQL through preference measures adjusted for variables that are known to be associated with changes in perceived HRQL, such as age, number of chronic diseases, sex and other social conditions, such as social group or potential loneliness. In addition, an algorithm to estimate the social rating or individual preferences for health states was used to measure this impact according to the latest version of the EQ-5D-5L [8], a tool with advantages over previous versions and whose Spanish version has shown excellent psychometric properties for patients with OALL [25]. OALL is associated with a substantial decrease in the utility attributed to the state of health when this association is adjusted for other, potentially confounding variables. Living with OALL implies a utility loss of approximately 0.30 on average compared to the general population without osteoarthritis, which exceeds the thresholds of clinical relevance obtained in Spain (0.07 points in the EQ-5D-5L utility index for subjects with nonsurgically treated OALL at the group level) [25]. On the other hand, as could be inferred, patients treated at the specialized level had an even worse perception of their HRQL, and their attributed utilities averaged 0.11 points lower.

Degenerative joint involvement was reported by the general population in the United Kingdom as one of the chronic conditions that most affects quality of life, behind only pain and anxiety/depression [37]. Similar data have been reported for the Canadian population [38]. The utility loss attributed to osteoarthritis in these studies has generally been approximately 0.10 points when adjusted for other variables that could affect perceived HRQL, a loss approximately three times higher than that produced by other chronic conditions, such as diabetes or asthma [38]. In Spain, when the EQ-5D-3L was used to evaluate HRQL, the study of osteoarthritis as a whole identified it as a chronic condition that decreases utility to the same extent as cardiac problems or diabetes mellitus, with a decrease in utility of up to 0.10 points compared to the general population over 65 years of age [39]. The literature has reported that OALL has significant impacts on utility, similar to those found in this study. In an Italian population, an average loss of utility of approximately 0.28 points was observed in patients with OALL compared to the general population [40]. In this case, utility was evaluated with the 3-level version of the EQ-5D, and adjustments were made only for age and sex. According to these authors, the impact of OALL on utilities was similar to that produced by osteoporosis with vertebral fracture or ankylosing spondylitis and was much higher than that produced by other chronic diseases that affect the musculoskeletal system, such as Sjogren’s syndrome or systemic sclerosis [40]. These figures far surpass the sensitivity thresholds of the EQ-5D-5L obtained in Spain. The established minimal clinically important difference (MCID) at the group level in Spanish patients with nonsurgically treated OALL was 0.07 [25], and in other countries, this threshold was 0.09 for subjects with arthritis [38].

It has also been possible to study the association between the setting from which the patient is recruited and his or her health situation. The average utility score for patients treated in the hospital setting was 0.11 points lower. In Spain, HRQL has been assessed in OALL patients who consulted trauma specialists [18] and in those in the community [19]; however, the results were not comparable due to the use of different measurement tools. This is the first time that evidence has been presented regarding the different degrees of disease involvement among patients seen at different levels of care, and the results are consistent with the organization of the Spanish national health system, in which the first level of care serves as a gateway for services for different chronic diseases and for osteoarthritis in particular [41].

It is possible that the great impact of OALL on HRQL occurs because the disease affects several of the domains incorporated into the EQ-5D tool, such as pain and loss of function [42]. However, in this case, the data presented show the impairment of a psychological component captured in the anxiety/depression domain of the EQ-5D-5L. This impact of OALL on mental health has previously been described in the literature [43]. In addition, the mental health domain as a component of perceived HRQL has been shown to be associated with joint replacement in OALL [44], so it should be evaluated with special attention.

On the other hand, the results of this study do not allow us to say that patients with osteoarthritis of the hip have, in general, a worse perception of their state of health than those with osteoarthritis of the knee, as has been suggested in some studies [18]. Although this tendency could be suspected from the descriptive analysis of the data (the VAS of the EQ-5D-5L), it is not observed when the models are adjusted, and the confidence intervals of the coefficients of hip/knee osteoarthritis overlap considerably.

The remaining sociodemographic and clinical characteristics, which were adjusted to determine the association between OALL and the utilities of health states, behaved in ways that were previously known. Older people, those from more disadvantaged socioeconomic groups, those who live alone, women, obese people and those with more chronic conditions tended to have a worse perceived health status, as described in previous studies in Spain [45,46]. It has previously been shown that the association between osteoarthritis and reported HRQL is confounded by other variables, such as sociodemographic factors, chronic disease or obesity [19,37]; therefore, the results presented here can be considered a realistic approach to the study of the association between OALL and HRQL.

Limitations

The design of this study has some limitations. Cross-sectional studies are problematic when establishing causal associations, although the effect of the main confounding factors collected in the literature has been assessed. Additionally, the sample of patients with OALL cannot be understood as representative of the population with this condition but was collected using opportunity criteria. We can affirm that sampling was carried out in diverse geographical sites and that the profile of the patients with OALL (predominantly female, older, and with a lower socioeconomic status and a higher prevalence of obesity) coincides with that reported in other Spanish [47] and European [19] studies. The fact that the participants may have had more severe disease than patients in the general population due to overrepresentation in the hospital setting was taken into account in the analysis. The general population from whom data were collected comprised people who did not report any type of osteoarthritis or arthritis diagnosis. The approach to the existence of chronic conditions could be considered a limitation of the study, as the conditions, or lack thereof, were reported by the subjects. However, even though there are problems with the general population data, the NHS constitutes the most representative and highest-quality information available at the time.

The EQ-5D-5L is a generic instrument designed to measure dimensions of health relevant to all health states, including healthy individuals, and not patients’ perception of aspects of health specifically affected by OALL. However, it has the advantage of allowing us to attribute patients’ preferences to the described health status in patients with OALL and to compare them between different populations [48].

Implications

The importance of the results presented lies in the interest that health policy decision makers may have in reproducible and comparable quantifications of the impact of OALL on HRQL. These results quantify the burden of the disease from the perspective of the patient. In addition, there are interventions, such as joint replacement surgeries, that involve a considerable investment of resources but can provide very valuable results from the perspective of the patient and society as a whole. It has frequently been noted that to assess the outcome of such interventions, it is necessary to use patient-reported outcome measures, such as quality of life, that help determine how to prioritize actions [49].

Conclusion

OALL is a chronic disease that has a great impact on the HRQL of patients. Patients with OALL perceive a very significant loss of utility, approximately 0.3 points compared to the general population without osteoarthritis. This impact differs depending on the place where the patient has been treated. Patients treated in the hospital setting in a health system in which primary care functions as a gateway for health care services report an additional utility loss of 0.1 point. This utility loss attributable to OALL is very high in relation to what has been reported for other chronic diseases and for arthritis in general, exceeding the threshold of the so-called MCID by up to three times.

Data for Sharing.

(XLS)

Click here for additional data file.

STROBE-checklist_cross-sectional.

(DOC)

Click here for additional data file.

4 Dec 2019

PONE-D-19-31220

Impact of lower limb osteoarthritis on health-related quality of life: An estimate of the loss of expressed utility in the Spanish population

PLOS ONE

Dear Dr. Martín-Fernández,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by Jan 18 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Yuanyuan Wang

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2.  Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

Additional Editor Comments:

The reviewers have provided comments and suggestion which will help improve the manuscript.

Additionally, the authors need to provide more details about the inclusion and exclusion criteria for the OALL patients and participants without OALL from the general population. Are there any data about the accuracy or validation of the self-reported data on osteoarthritis/arthritis in the National Health Survey? It is not clear whether the data (as shown in the Table 1) were collected in the same way for the OALL patients and participants without OALL from the general population. It was stated that 393 patients had knee osteoarthritis and 331 had hip osteoarthritis. Does this means that no patients had both knee and hip osteoarthritis? If this is the case, what is the meaning of the data shown in the Table 3 for "Involvement of another large lower limb joint (hip or knee)"?

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Within their study, the authors report on the results of an observational study to assess the impact of having osteoarthritis of the lower-limb (OALL) on health-related quality of life, compared to the general population patients without OALL. Results are presented both without adjustment for patient characteristics, and with adjustment via a series of regression models. The manuscript is generally well-written, the data appears to be suitable, and it is good to see that the full dataset will be made publically available. I do have some concerns that the reporting of the regression models requires more information (and I’m not convinced about the choice of model 1).

Major comments.

• Data on those without OALL is from the 2011-2012 National Health Survey – did this include EQ-5D-5L responses? If so, explicitly state (I wasn’t aware that the EQ-5D-5L was available in 2011), if not, please state how these values were obtained (for example, if mapped from EQ-D-3L, the mapping algorithm used). If EQ-5D-5L values were indirectly obtained for the general population this should be noted as a limitation of the study.

• The author’s state that they have used generalised linear models, this is a very broad family of models! They provide details on the linear predictor used, but need to also describe the probability distribution used (Normal, Poisson, etc) and the link function (identity, log, etc).

• Given that the main objective of the paper is to estimate the impact of OALL on health-related quality of life compared to the general population, and Table 1 shows that there is a difference when not adjusting for patient characteristics, I was confused as to the purpose of Model 1, which does not include a variable for OALL. Shouldn’t Model 1 include OALL as a binary yes/no variable? Then Models 2 and 3 check if adding evidence on type of OALL and care setting improve fit.

• The authors provide evidence on the relative goodness of fit of the three models considered, but not on the models’ absolute goodness of fit. This could be achieved, for example, by visually comparing the observed EQ-5D-5L summary scores with the model-based estimates, potentially for the two sub-groups defined by the presence and absence of OALL. Further, a particular feature of EQ-5D-5L data is that there can be a ceiling effect (clusters at the summary value of one); are the models constrained to only predict values less than or equal to one? (As the primary objective is a comparison at a cohort level, having individual predictions outside the allowed range is a secondary problem, but still one worth noting if this occurs).

• Please complete the STROBE reporting guideline: https://www.equator-network.org/reporting-guidelines/strobe/

• Has consent been obtained releasing the patient-level data? If not, is it suitably anonymised?

Minor comments.

• It is unclear if a systematic search for existing utility measures for OALL has been performed? If so, this would strengthen the case for the originality of this work (some studies are discussed starting on line 116 of page 8, but it is unclear how they were identified).

• Table 1; for the continuous variables age, number of chronic diseases, and BMI, please also report the mean values per group.

• Table 3; please also add BIC values.

• For confidence intervals, please report as “xx to yy” instead of “xx - yy” as in the latter case this looks like a negative sign.

• Page 7, line 94: “…specifically request that the EQ-5D be used in all economic evaluations…” this is slightly overstating the point. NICE state a preference for EQ-5D, but acknowledge that there may be occasions when this is not the most appropriate measure, such as in children (see Sections 5.3.10 and 5.3.11 of the NICE reference case).

• Reference 9 is for the 2010 global burden of disease study – there are more recent versions of this study, which should have more recent evidence on the ranking of OALL on DALYs lost.

• Reference 8 has been subsumed into reference 7 in the bibliography.

Reviewer #2: The authors have done a great job in conceptualizing the study, outlining a clear aim and articulating the design and methods. Given that the EQ-5D-5l was used in quantifying HRQL, the manuscript will be greatly improved if the authors consider interpreting their findings based on MCID, as opposed to subjective phrases like "substantial" in describing differences in HRQL.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Benjamin Kearns

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: Reviewers_reponses_AC.docx

Click here for additional data file.

23 Dec 2019

Dear editors:

We would like to thank you sincerely for the work invested by you and by the reviewers. The suggestions of the reviewers will be very useful for improving the quality of the revised manuscript.

We have taken all comments into consideration and we have added a few general comments so that the methodology and the context can be better understood.

Editor Comments:

Additionally, the authors need to provide more details about the inclusion and exclusion criteria for the OALL patients and participants without OALL from the general population.

We included an opportunistic sampling of patients aged 50 years or older, diagnosed with osteoarthritis of the hip or knee according to the American Rheumatism Association criteria who consulted traumatology and rheumatology specialists at six hospitals and 21 primary care centers in Vizcaya, Madrid and Tenerife between January and December 2015. We excluded patients with malignant or other organic diseases or psychiatric disorders that hindered participation, and those who could neither read nor understand Spanish.

Exclusion criteria have been added to the manuscript.

The general population data were obtained by selecting the population aged 50 years or older from the 2011-2012 National Health Survey (NHS), who claimed not to have a medical diagnosis of osteoarthritis, arthritis, or rheumatism. The NHS was performed in 2,000 of the 35,960 sections of the 2011 Municipality Population Census, including 12 households per section area. Three-phase sampling, stratified by population size, was performed (census section, homes, and people in the home suitable for survey participation), to achieve a representative sample at a region level. These data are available everywhere and they can be checked at reference 20.

Are there any data about the accuracy or validation of the self-reported data on osteoarthritis/arthritis in the National Health Survey? It is not clear whether the data (as shown in the Table 1) were collected in the same way for the OALL patients and participants without OALL from the general population.

We have no data about the validity of the information reported in National Health Survey, but the survey methodology poses the best approximation of the Spanish population’s health at present, and its information has been used in quite a lot of studies which approached to the health status of general population (Pinilla et al, PLOS One 2019; Lostao L et al PLOS One 2017; Martin-Fernández J et al, Gac Sanit 2017; Tamayo Fonseca et al BMC Health Serv Res 2017;...)

Data regarding to people with OALL were collected directly from the patients, and those referred to general population were extracted from the National Health Survey. This has been stated into Methods section.

It was stated that 393 patients had knee osteoarthritis and 331 had hip osteoarthritis. Does this means that no patients had both knee and hip osteoarthritis? If this is the case, what is the meaning of the data shown in the Table 3 for "Involvement of another large lower limb joint (hip or knee)"?

It was not clear enough in the manuscript, so we have remade the sentence:

“Of the patients diagnosed with OALL, 393 (54.28%) had a diagnosis of osteoarthritis of the knee, of whom 53 (13.49%) had a previous diagnosis of osteoarthritis of the hip. Of the 724 patients with OALL, 331 (45.72%) had a diagnosis of osteoarthritis of the hip, of whom 115 (34.74%) had a previous diagnosis of osteoarthritis of the knee.”

Reviewer #1:

Major comments.

• Data on those without OALL is from the 2011-2012 National Health Survey – did this include EQ-5D-5L responses? If so, explicitly state (I wasn’t aware that the EQ-5D-5L was available in 2011), if not, please state how these values were obtained (for example, if mapped from EQ-D-3L, the mapping algorithm used). If EQ-5D-5L values were indirectly obtained for the general population this should be noted as a limitation of the study.

The National Health Survey 2011-12 was developed between July 2011 and June 2012. It was the first National Survey in Spain which used the EQ-5D-5L for measuring the perceived health related quality of life (the latest National Health Survey 2017 did not use any standard tool for measuring HRQL). All the information about the NHS methodology can be found at:

https://www.mscbs.gob.es/en/estadEstudios/estadisticas/encuestaNacional/encuesta2011.htm

We got results from EQ-5D-5L for each subject included in the NHS and we used the algorithm proposed by Ramos-Goñi et al (Value in Health 2018) to calculate the utility values, both for general population and for OALL patients.

• The author’s state that they have used generalised linear models, this is a very broad family of models! They provide details on the linear predictor used, but need to also describe the probability distribution used (Normal, Poisson, etc) and the link function (identity, log, etc).

We agree with the reviewer. Several distributional families and link functions were tested and the one that best adjusted the dependent variable was chosen through the AIC and the BIC. The gaussian family and “Identity function” were selected as distributional family and link function respectively. This information has been added to Methods section.

• Given that the main objective of the paper is to estimate the impact of OALL on health-related quality of life compared to the general population, and Table 1 shows that there is a difference when not adjusting for patient characteristics, I was confused as to the purpose of Model 1, which does not include a variable for OALL. Shouldn’t Model 1 include OALL as a binary yes/no variable? Then Models 2 and 3 check if adding evidence on type of OALL and care setting improve fit.

We felt it is was a good suggestion, so we have changed table 4. Model 1 includes now OALL as a dichotomous variable. Coefficients are now different and “akaike weights” have been recalculated.

• The authors provide evidence on the relative goodness of fit of the three models considered, but not on the models’ absolute goodness of fit. This could be achieved, for example, by visually comparing the observed EQ-5D-5L summary scores with the model-based estimates, potentially for the two sub-groups defined by the presence and absence of OALL. Further, a particular feature of EQ-5D-5L data is that there can be a ceiling effect (clusters at the summary value of one); are the models constrained to only predict values less than or equal to one? (As the primary objective is a comparison at a cohort level, having individual predictions outside the allowed range is a secondary problem, but still one worth noting if this occurs).

Reviewer #1 suggested us to use techniques, which are more useful to assess the predictive ability of the models. But we assumed that we were building explicative models of the differences between groups.

To test the global goodness of fit of each model, we have now estimated the McFadden's adjusted pseudoR squared. It mirrors the adjusted R-squared in OLS by penalizing a model for including too many predictors.

The formula used was:

McFadden's adjusted R squared = 1 – [(Deviance model-k)/Deviance intercept], where k stands for the number of parameters (including the intercept).

Mc Fadden’s adjusted R2 (or pseudo-R2) values were added for all models in table 4, so readers will be able to check the proportional reduction in “error variance”, as deviance plays a role analogous to the residual sum of squares in linear regression (Allison P at: https://statisticalhorizons.com/r2logistic). The interpretation of this Mc Fadden’s adjusted R2 has been added in Results section.

On the other hand, as Reviewer #1 mentioned, there is a current debate on how to manage the censored part of the utilities and on whether censoring regression methods are appropriate to make estimations over this point ( Sullivan PW. Med Decis Mak. 2011; 31(6):787–9). But subjects at perfect health state were not the target of this study. Besides, psychometric properties of the EQ-5D-5L in patients with hip or knee osteoarthritis in Spain were studied previously and minimal floor and ceiling effects were found (Bilbao A et al. Qual Life Res 2018; 27(11): 2897-2908.)

• Please complete the STROBE reporting guideline: https://www.equator-network.org/reporting-guidelines/strobe/

Strobe checklist has been added as “Supporting Information”.

• Has consent been obtained releasing the patient-level data? If not, is it suitably anonymised?

Data from OALL patients were collected after obtaining their written consent. Data from general population were obtained from a secondary anonimysed source, the National Health Survey. The National Health Survey microdata are publicly available at: https://www.mscbs.gob.es/estadEstudios/estadisticas/encuestaNacional/encuesta2011.htm

Minor comments.

• It is unclear if a systematic search for existing utility measures for OALL has been performed? If so, this would strengthen the case for the originality of this work (some studies are discussed starting on line 116 of page 8, but it is unclear how they were identified).

We did not make a systematic review for building the theoretical framework. We developed a search which aimed to map rapidly the key concepts underpinning the research area. We found primary sources which allowed us to develop secondary searches, so we are sure enough of the importance and novelty of this study. We could say that we performed a scoping review (Arksey, H & O'Malley, L, Journal of Social Research Methodology 2005; 8:1, 19-32).

• Table 1; for the continuous variables age, number of chronic diseases, and BMI, please also report the mean values per group.

Done

• Table 3; please also add BIC values.

BIC values have been added in table 4.

• For confidence intervals, please report as “xx to yy” instead of “xx - yy” as in the latter case this looks like a negative sign.

We have now used commas to separate the extremes of the confidence intervals.

• Page 7, line 94: “…specifically request that the EQ-5D be used in all economic evaluations…” this is slightly overstating the point. NICE state a preference for EQ-5D, but acknowledge that there may be occasions when this is not the most appropriate measure, such as in children (see Sections 5.3.10 and 5.3.11 of the NICE reference case).

Reviewer #1 is right. The NICE recommendation was: ” Health effects should be expressed in QALYs. The EQ-5D is the preferred measure of health- related quality of life in adults.”

So we changed the original sentence to:

“[…]have specifically stated that the EQ-5D is the preferred measure of HRQL in adults; thus, utilizing it ensures comparability among studies [6,7].”

• Reference 9 is for the 2010 global burden of disease study – there are more recent versions of this study, which should have more recent evidence on the ranking of OALL on DALYs lost.

Thank you for the suggestion. We have changed the data by those provided by the Global Burden of Disease Study 2015 (Vos et al, Lancet 2016)

• Reference 8 has been subsumed into reference 7 in the bibliography.

It has been corrected

Reviewer #2:

Overall, the study is well conceptualized with clearly articulated aim and methods. The statistical models employed are consistent with the outcomes of interest. Using the EQ-5D provides a unique advantage that the authors have not leveraged. Of particular note, the authors should consider discussing their findings in light of the clinical relevance based on Minimally Clinically Important Differences (MICD) that have been well documented for the EQ-5D versions. This is especially important for potential interpretation and policy implications of the study findings.

The Minimal Clinical Important Clinically Important Differences (MICD) documented for the Spanish versión of the EQ-5D-5L was 0.07 points, although the ratios of the MCID and the MDC95% were over the unit, only at group level. So, only when we consider the change of a group we can understand that an improvement is significant if it is over 0,07 (Bilbao A et al, Qual Life Res 2018; 27(11): 2897-2908). We had already included this benchmark in Discussion section (page 20/ line 276 and page 21/ line 298) and we have now included a new reference in the Conclusion section.

Other minor, albeit important comments are listed below.

Materials & Methods:

1. Line 128 simply starts with the phrase: “Cross-Sectional Study”. If the intention of the authors is to indicate that the study design is a cross-sectional study, they should do so by making a complete statement. Please check the abstract for a similar.

The sentence has been drafted again in a complete way. Abstract section has been adjusted to 300 words length.

2. Line 162: The use of “Chronic diseases diagnosed” should be clarified. For instance, the authors may consider to list the number of chronic diseases included. Adjusting for concurrent chronic diseases is important in HRQL studies. However, even more important is the specific chronic conditions that are accounted for in the model.

We recognize that our approach to chronic disease was quite a bit weak. The multimorbidity phenomenon is characterized by its complexity. We knew that from an holistic perspective, the assessment of multimorbidity by using a simple list of diseases (weighted or not) may be not sufficient (Diederichs C et al, J Gerontol A Biol Sci Med Sci. 2011 Mar;66(3):301-11). But for feasibility reasons we only collected a list of chronic conditions and the same weight was attributed to each one.

Chronic conditions from patients with OALL were collected from their clinical records, based on the list of conditions from Charlson’s index. Information about chronic conditions from general population was collected asking for any chronic condition which was diagnosed by a physician prior to the interview. People were asked about a list which contained 30 chronic diseases and afterwards they were allowed to report any other chronic condition by means of an open question.

We think that this issue deserves a comment in the section refered to limitations, so we have added the next sentence:

“The approach to the existence of chronic conditions could be considered a limitation of the study, as they were reported by the subject”.

3. Line 169 – 171, the author indicates that the WOMAC was used. Is it to be assumed that the English version was used? If a Spanish version was used, kindly provide more information about the version such as its construct validity and reliability (with references).

We used the WOMAC Spanish versión and its psychometric properties had been already referenced (line 183, reference 30).

Results:

1. Table 1: The multiplicity of p values < 0.0001 is an indication of a high likelihood of having type 1 error, potentially because of the large sample size. The author should consider accounting/adjusting for type 1 error in the study.

We agree with Reviewer #2. If we were studying the similarities among groups it would be neccessary to adjust the type 1 error. But we knew that groups were not similar. We were looking for the characteristics of the subjects which should be included in the models in order to get the best adjust. So the p values are not considered as a whole, but each one is studied as a particular one.

2. The author has not defined Social Groups I to VI that are used in all tables. It is hard to tell which a higher social group is and which is lowest.

Social group are defined on a six-category classification related to occupation (group I being the highest and group VI the lowest). This information has been added in Methods section.

3. Line 213: The use of older people (except for extreme ages) is not immediately obvious. The author should kindly indicate what age group(s) or ranges you are referring to.

It has been clarified in Results section.

4. Table 4:

- In the BMI category, the author should consider changing “normal” to “Non-obese”, given that underweight/overweight/obese people are also “normal”.

The change has been done, we are sorry, we pretended to denominate this group as “normal weight”.

- The author included BMI and number of chronic diseases in their final model. Research indicates that multiple chronic diseases do cluster in obese populations. Was there a correlation between BMI and number of chronic diseases? If so, could there be an issue of collinearity?

In this case, the the Pearson correlation coefficient for the association between BMI and number of chronic conditions was 0.0929 (p<0.05), and it appears not to be relevant. The mean VIF uncentered values for the Models 1, 2 and 3 were 2.66, 2.58 and 3.78 respectively. So we thought we did not have any relevant collinearity problem in the showed models.

5. Line 251: The authors indicate that being treated in a “hospital environment is associated with an average decrease in utility..” What may be the reason for this findings? Is it possible that the reduced HRQL associated with hospitalization is because of less than quality care in the hospital environment? Or is it as a result of the fact that more severe cases of OALL are likely to be hospitalized? Kindly provide more details.

In the Spanish National Health System, family doctors usually acts as gatekeepers. So they attend patients with OALL in the first stages of the disease. Initial approach to OALL requires of changing lifestyles, avoiding increase weight and sendentarism, and the use of analgesia. This approach is usually achieved in primary care centres. When osteoarthritis progresses patient are referred to the specialized level. So it is not surprising that more severe cases were recluted in outpatient clinics, which depend on Hospitals (it does not imply that patient is hospitalized, unless a replacement surgery is indicated).

Discussion:

1. Line 254: The statement “This is the first study in Spain to quantify the utility loss attributed to OALL….” is not consistent with the study aim to evaluate the loss due to OALL (line 122). The authors should consider rephrasing the statement.

Both sentences have been rephrased. So the new objective is:

“This study aims to evaluate the impact of OALL on HRQL through measures based on preferences in comparison to the general population that does not have osteoarthritis.”

The objective has been also rephrased in the abstract section.

Discussion section starts with the next sentence: “This is the first study in Spain to quantify the impact of OALL on HRQL through preference measures, when adjusted for other variables that are known to be associated with changes in perceived HRQL…”

2. The statement “OALL is associated with a substantial decrease in the utility attributed to the state of health when this association is adjusted for other potentially confounding variables” is ambiguous at best. HRQL research using the EQ-5D usually qualifies quantitative decreases in the utility index as statistical or clinical differences based on a defined minimally clinical important difference (MCID). The authors should consider discussing their findings within the parameters of MCID. This has the potential for interpretation especially for clinical application.

We agree with Reviewer #2. MCID had been mentioned in the first version of the manuscript, but we have now underlined the relationship between the loss of utilities in patients with OALL and the referred threshold in two paragraphs of the discussion and in the conclusión section.

3. The authors discussed “utility” as their outcome, contrary to the fact the HRQL was the outcome in the stated aim. Please consider re-phrasing related statements for clarity and consistency.

We understand “utility index” as an approach to HRQL assesed through preference based measures. Nevertheless we have reformulated the sentences as suggested by Reviewer #2.

4. Given the plethora of tests across each sub-group of the patient population and outcomes, the authors should consider adjusting for multiple testing or at least provide a rationale if they choose not to.

In table 1 repeated tests are used for describing groups. The aim of the study was not to describe groups but looking for explicative models which allowed us to understand the impact of OALL on utilities. As we discussed before the p-values should not be considered as a whole. They were only useful to understand why given individual characteristics will be included in the models, although we could omit the p values and doing the same development

We found that each p-value in table 2 (and in table 3 too) answer to a different particular question. In any case we did not mention the results of the p values and we did not find the need of adjusting for multiple tests as our main results and discussion will be based on models.

5. The authors should discuss limitations of using the EQ-5D-5L and its potential implications on interpretation of the outcome (for instance a ceiling effect due to the upper limit of the score).

The limitation of using a generic instrument has been added in the appropriate section.

On the other hand, as we discussed before, the EQ-5D-5L showed minimal floor and ceiling effects in Spanish patients suffering for OALL (Bilbao A et al. Qual Life Res 2018; 27(11): 2897-2908).

Implications:

Line 337: “These results provide quantification of the prevalence of the disease and the burden..” It is not clear what the authors refer to as the “prevalence of the disease”. The aim of the study was not to present the “prevalence” of the disease.

We agree with Reviewer #2. We have deleted the reference to prevalence as it was not our aim.

Besides the referred changes, the title was modified in order to explain the study design as recommenden by STROBE Statement. It has been underlined that patients with missing data were not included in the analysis, as it had been already explained in “Results” section the number of subjects included in each analysis.

Minor changes as footnotes have been added and marked, and a new edition review has been made.

We hope that these changes will contribute to improving the quality of the manuscript and the interest for potential readers.

On behalf of the research team,

Jesús Martín-Fernández

Submitted filename: Response to reviewers.doc

Click here for additional data file.

15 Jan 2020

Impact of lower limb osteoarthritis on health-related quality of life: A cross-sectional study to estimate the expressed loss of utility in the Spanish population.

PONE-D-19-31220R1

Dear Dr. Martín-Fernández,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Yuanyuan Wang

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

The authors have addressed all the comments properly.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: The authors have done an excellent job in responding to the comments. I would recommend Beefaroni's correction for multiple testing, given the number of tests performed per table for a total of four tables (20 tests in total). Thanks very much.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Benjamin Kearns

Reviewer #2: No

17 Jan 2020

PONE-D-19-31220R1

Impact of lower limb osteoarthritis on health-related quality of life: A cross-sectional study to estimate the expressed loss of utility in the Spanish population.

Dear Dr. Martín-Fernández:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Yuanyuan Wang

Academic Editor

PLOS ONE