Patient self-reported functioning by pain severity and usual analgesic treatment among older adults with osteoarthritis: analysis of the 2017 Spanish National Health Survey

Osteoarthritis (OA) pain is among the leading causes of disability worldwide in older people. Since its prevalence is growing in aging, a significant burden for society is expected. This work ascertained whether level of disability in self-reported functioning differs by pain severity and usual analgesic treatment among older OA patients in Spain. The Spanish-National-Health-Survey, a large, nationally representative, cross-sectional general health survey including 23,089 persons, was analyzed. Patients aged 65 + years with a self-reported physician OA diagnosis were classified according to severity of pain (no/mild, moderate or severe pain) and treated or untreated with analgesia. Assessment of function included basic and instrumental activities-of-daily-living (BADL, IADL), mental, social, and cognitive functions, scored on a 0% (no limitation) to 100% (complete limitation) standardized metric. Caregiver need for BADL and IADL was also recorded. A total of 3526 patients were analyzed (women 73.3%; age 77.4 [SD: 7.5]). Adjusted functioning scores showed significant association with pain severity, and for BADL, IADL and social function. Patients with severe pain and treated with analgesia had higher limitation scores, ranging on average between 31.5% on BADL, 34.1% on IADL, 45.0% on mental, 42.2% on social, and 23.4% in cognitive domain. The proportions of patients needing a caregiver for BADL (43.4%) and IADL (56.2%) were also the highest in patients with severe pain and treated with analgesia. Regardless of usual utilization of analgesics, the severity of pain seemed to be the major determinant of functional impairment, and caregiving need, in all domains of functioning in older OA patients in Spain. Existing treatment strategies are analgesics based and do not meet patient needs for adequate pain management.


Introduction
Osteoarthritis (OA) is the most prevalent form of arthritis [1]. It is a heterogeneous disease with multiple underlying mechanisms that affects all joint tissues. Age, body mass index and gender as well as mechanical, biochemical, genetic, immunologic and psychopathological features play a role in the development of the disease. Hence, OA is now considered a disease of an organ, the joint [2]. OA is often painful and debilitating, and at present it is not possible to predict whether a patient will gradually lose joint function and develop more joint damage requiring a joint replacement [3,4]. The burden of OA is increasing as population aging is advancing and obesity rates are rising. OA has become one of the leading causes of disability and social isolation worldwide [5,6]. Suboptimal pain management is known to have negative effects on the physical and mental well-being of the population as it greatly impacts quality of life, with a commensurate tremendous individual and socioeconomic burden [4][5][6][7][8][9]. Given its vast implications, the United States Food and Drug Administration considers OA a serious disease [10]. It is estimated that OA will be the fourth leading cause of disability in 2020 and one of the 25 main causes of disability with the greatest impact on health worldwide [11].
The mechanisms of OA pain are diverse and complex, including biological, psychological and social factors [12,13]. Psycho-social factors have a key role in the pain perception experience and might play a part in pain exacerbation, spreading and perpetuation that lead to peripheral pain sensitization. The consequences of OA pain include activity limitations and participation restrictions in addition to negative effects on mood, fatigue and sleep, affecting quality of life [14,15]. OA management guidelines recommend a wide variety of core non-pharmacological and pharmacological treatments [6,8,16]. However, current pharmacological treatments are ineffective in halting disease progression and are mainly aimed at modifying pain symptoms with debated effectiveness [5,16]. Prior research sought to demonstrate that individuals with OA are more likely to have functional impairment, measured as the ability to perform basic activities of daily living (BADLs) and instrumental activities of daily living (IADLs), and, as a consequence, with higher probability of caregiving need [17][18][19][20][21]. Nevertheless, most of these studies were conducted on selected populations in specific health settings or on waiting lists for joint replacement. Moreover, many of them lacked appropriate control groups or did not involve representative samples of older patients.
Thus, the objective of this study was to ascertain the level of disability or limitation in key domains of the patient functioning in adults 65 years or older with OA pain in Spain, based on the data collected in the 2017 Spanish National Health Survey (ENSE) [22]. Domains of patient functioning included basic and instrumental activities of daily living, mental, cognitive, and social functioning. The impact of OA pain on functioning was analyzed according to pain severity and recent treatment with analgesics. Caregiver participation in helping patients on BADL and IADL was also assessed.

Study design
This study had a cross-sectional design written according to the STROBE checklist (See supporting information) [23].

Data source
Data were abstracted from the most recent ENSE survey (from 2017) [22]. The ENSE is funded every five years by the Spanish Ministry of Health and constitutes the main source of information on health perceived by the population 1 3 residing in Spain. It is highly representative of Spain as a nation according to population geographic density and collects an extensive set of disaggregated aspects of health according to demographic and socioeconomic characteristics. The information was obtained through interviews in homes throughout the country using computer-assisted personal interviews (CAPIs) as a method of information collection after obtaining the informed consent of the individuals surveyed. The 2017 ENSE collected information on a representative sample of 23,089 men and women over 15 years of age in the Spanish population.

Study population
For this study, a subsample was used comprising survey participants of both sexes aged 65 or older with a self-reported physician diagnosis of OA who completed the survey. The population of people aged 65 or older was selected for this study because, by design, the ENSE only surveys people in this age range for some of the variables that corresponded to the study objectives, specifically basic and instrumental activities of daily living and cognitive functioning. Pain was assessed by pain domain in the Spanish version of the Short Form 36 Health Survey Version 2 (SF-36v2) questionnaire included in the ENSE questionnaires (Question 45, supporting information). To analyze the study objectives, the OA subsample of the ENSE was grouped into six categories. According to the presence of pain in the past 4 weeks, which was assessed according to the mentioned question on pain severity in the past 4 weeks, said subsample was grouped into three categories: no pain/mild pain (response 1, 2 or 3 to Q45), moderate pain (response 4) and severe pain (response 5 or 6). These were then crossed with the use of analgesic medications in the past 2 weeks according to the ENSE (Question 87_2a, supporting information) such that six subgroups were formed based on pain severity in the past 4 weeks and the use of analgesics in the past 2 weeks.

Assessment of functioning
Assessment of functioning was based of essential functioning of individuals expressed as basic activities of daily living (BADLs), instrumental activities of daily living (IADLs), mental, social, and cognitive functioning. In order to examine the survey participants' limitations (dependency) across these five key dimensions of functioning, ENSE questions on the variables of interest were used. Responses to these questions are on an ordinal Likert scale; however, to harmonize assessments, all types of functioning were re-scaled to a standardized metric of 0-100% [24], where 0% indicates no limitation and 100% indicates complete limitation in terms of functioning, or complete dependence. In addition, to facilitate interpretation, measures were reorganized into categories according to severity of limitation as described below. These ENSE questions consist of four response levels, where a value of 1 would correspond to complete independence in carrying out the activities studied and a value of 4 would correspond to complete limitation therein. Re-scaling to a metric of 0-100% was achieved using the following equation [25]: [Observed score − Minimum score)/Range] × 100. The result is expressed as a percentage of limitation (the range is understood to be the difference obtained by subtracting the minimum possible score from the maximum possible score). Thus, in the case of the score for limitation in basic and instrumental activities of daily living, each having seven items, a score of 7 corresponds to 0% limitation in activity and a score of 28 corresponds to 100% limitation.
A clinical interpretation of the results obtained was performed by analyzing the proportion of persons who, regarding variables for functioning (BADLs, IADLs, cognitive, social, and mental functioning), showed a meaningful or important restriction, limitation or dependency in each type of functioning. This level of restriction not only is significant for the survey participant, but also should be the subject of a health decision intended to result in an improvement in or relief of the dependency represented by this situation, in accordance with the recommendations of the World Health Organization and the World Bank [26]. For BADLs and IADLs, the criterion for interpretation of the Barthel Index was followed, considering meaningful dependency to be severe or complete limitation in functioning: limitation > 60% after adjusting by covariates, since the score correction method was inverted compared to the Barthel Index) [27,28]. For mental functioning, the threshold value for significant mental decline, consistent with mental illness, was > 50%, in accordance with Goldberg et al. [29,30], also using corrected values for covariates. For the cognitive functioning and social functioning variables, the threshold values for significant limitation were those corresponding to extreme or total deterioration/inability regarding remembering and concentrating, and 100% inability/restriction regarding social daily activities, respectively.

Physical functioning: basic and instrumental activities of daily living
To assess the degree of difficulty in going about activities of daily living, an activity limitation score was calculated for basic and instrumental activities. Limitation in going about basic activities was calculated using responses to Questions 37, 38 and 39 (sub questions a-e), corresponding to Modules K (Physical, sensory, and cognitive limitations) and L (Limitations in going about activities of daily living) of the ENSE survey (See supporting information), consisting of activities related to mobility, capacity for getting dressed, continence, eating and bathing and hygiene. Similarly, limitation in instrumental activities was calculated using ENSE Question 42 (sub questions a-g, see supporting information) evaluating the survey participants' difficulty in going about activities such as preparing meals, using the telephone, shopping, taking medication, doing light housework, occasionally doing more physically demanding (hard) housework and managing their own money. Caregiver participation in support for basic and instrumental activities of daily living (questions 40 and 43 of the ENSE survey) was also assessed in terms of percentage of individuals receiving the help of a caregiver for such function.

Mental functioning
To assess mental health in the study population, ENSE questions were used according to the methodology of the 12-Item General Health Questionnaire (GHQ-12) [30]. This GHQ-12 Questionnaire was designed by the British scholar Goldberg in 1972 and reflects the mental health status of the survey participants based on the analysis of 12 attributes. The GHQ-12 scale consists of 12 questions with four Likert response levels (block M of the ENSE -Mental health and occupational stress-and comprise Sections 1-12 of Question 47, See supporting information). They have four response levels from 0 to 3 (from more than usual to much less than usual); these have also been converted to a 0-100% metric using the above-mentioned equation. These questions examine decline in mental health as a combination of two dimensions: well-being and coping.

Social functioning and cognitive functioning
Restriction or degree of limitation in going about regular social activities (social functioning) was estimated based on Questions 28b and 29b, corresponding to Module I of the ENSE on restriction of regular activity in the past 2 weeks. These questions examine the number of days during which individuals limit, both completely or partially, their daily activity or even remain in bed. Regarding cognitive functioning, to measure survey participants' difficulty in remembering and concentrating; Question 38 from Module K of the ENSE was used. This question has four response levels (1)(2)(3)(4) corresponding to scores of 1 (no limitation) to 4 (complete limitation). As indicated above, these metrics were also converted to a score of 0-100%.

Statistical analysis
The statistical analysis included a preliminary analysis of the sample study groups through the preparation of dynamic tables and a descriptive univariate analysis of relevant sociodemographic variables. The rate of response or proportion of people interviewed who had provided a response for the variable to be analyzed was estimated. Missing data were not imputed due to the cross-sectional design of the survey and the fact that the loss of data in the variables studied was always less than 5% of the numbers analyzed (range 0.6-3.9%). The descriptive analysis included the percentage distribution for categorical variables and measures of central tendency and dispersion for continuous variables, including normality tests. IBM SPSS (Version 26.0, New York, United States; https:// www. ibm. com/ analy tics/ spss-stati stics-softw are), a software program for statistical processing, was used.
Generalized linear models (GLMs) with normal linear regression and covariates were used with estimation of effects using a factorial model with covariates to analyze scores in functioning domains expressed as a continuous linear metric with a range of 0-100%. The covariates included in the models were as follows: the survey participant's geographic region of residence, age, sex, marital status, level of education, accidents in the past 12 months, smoking status, alcohol use, level of physical activity, and Functional Comorbidity Index (FCI) 31 . The FCI was used to correct the individual's functioning weighted by the density of comorbidities deemed significant in terms of their impact on the individual's functioning. It was calculated as a count of affirmative responses to the 18 items or comorbidities that comprise it as recommended by its authors, Groll DL et al. [31], and that are included, amongst others, in the ENSE survey questionnaire. Analysis of caregiver utilization also adjusted for by the calculated activity limitation score for basic and instrumental activities, respectively. The sequential Bonferroni correction was applied to estimate statistical significance in multiple comparisons. GLMs with binary logistic regression and covariates were used with estimation of effects using a factorial model with the covariates to analyze dichotomous responses on severe to total limitation/ decline (yes or no) in each domain of functioning by analgesia status in each level of pain severity. Finally, patients were classified according to severity of limitation/restriction into categories of functional limitation for each of the variables analyzed: none/no, mild/low/mild to moderate, moderate/normal and severe to total/high/complete limitation. GLMs with Poisson log linear regression and covariates were used with estimation of effects using a factorial model with covariates to analyze functioning classified into categories of severity of limitation by analgesia status in each level of pain severity. The main statistical analyses were also replicated in the overall sample split into two samples according to gender.
To interpret the results obtained, when statistically significant differences were found between pairs, the magnitude of the difference was calculated by estimating the effect size with Cohen's d statistic [32]. Effect size was estimated by dividing the mean difference between groups by the overall standard deviation and was interpreted according to the recommendations of Cohen and Kazis [32,33]: 0.20-0.50 = small effect size; 0.50-0.80 = moderate effect size; and ≥ 0.80 = large effect size.

Confidentiality of information/ethical aspects
The analysis carried out in this study used aggregate dated available in the public domain only, thus, de-identification was guaranteed to prevent anyone's personal identity from being revealed. Therefore, studies such as this one are exempt from IRB review and from classification by the Spanish Agency of Medicines and Medical Devices according to current regulations for post-authorization observational studies (https:// www. aemps. gob. es/ inves tigac ionCl inica/ medic ament os/ estud iosPo staut oriza cion. htm# norEs tatal). The confidentiality of the records (anonymous and dissociated) was respected according to the existing Spanish Law on Personal Data Protection (Law 15/1999, of 13 December 1999, on Personal Data Protection). Individuals included in the ENSE survey participated voluntarily and gave their personal consent prior to their participation in the survey (see details on survey implementation at: https:// www. mscbs. gob. es/ estad Estud ios/ estad istic as/ encue staNa cional/ encue sta20 17. htm).

Results
A total of 7,023 ENSE records for people 65 years of age or older were analyzed. Of these, 3526 (96.1% of surveys completed, n = 3389) corresponded to patients with a self-reported physician diagnosis of OA; 50.2% (95% CI 49.0-51.4). Of these, 40

Basic activities of daily living (BADLs) and instrumental activities of daily living (IADLs)
The functioning of OA patients in BADLs was found to be limited, with an adjusted average (standard deviation), to 15.9% (22.1) in the entire group of OA subjects. It was also found to have a significant association with pain severity (p < 0.001): the higher the pain, the higher the limitation in BADLs, regardless of whether patients had been treated with analgesics in the last 2 weeks (Table 2). OA patients with treated moderate or severe pain showed significantly higher limitation in BADLs than non-treated patients ( Table 2). The magnitude of these differences was small, with effect sizes ranging from 0.19 to 0.34. Meaningful limitation (greater than 60%) in BADLs was also significantly associated with pain severity (but not analgesic intake), with 17% of treated patients and 14% of non-treated patients with such a degree of limitation in the severe pain subgroup (Fig. 1). The analysis by category of functional limitation showed significant differences in distribution of degrees of limitation according to pain severity as well as in each subgroup between treated and not treated with analgesics (Fig. 2). Figure 2 shows that, for basic activities, restriction increased as pain severity increases, and this restriction was greater in individuals who required analgesia. The percentage of patients helped by a caregiver on BADL increased significantly with severity of pain, regardless whether they were receiving analgesics or not in last 2 weeks ( Table 3). The main drivers for such findings in BADLs were items exploring the patient's mobility and walking and up/downstairs compared with to other items in the domain (p < 0.001, Fig. 3).
Adjusted mean functioning in IADLs was found to be slightly higher than in BADLs, being 19.1% (25.8) in the entire group of OA subjects. It was found to exhibit a significant association with pain severity (p < 0.001) regardless of analgesic treatment (Table 2). Likewise, OA patients with treated moderate or severe pain showed significantly higher limitation in IADLs than non-treated OA patients ( Table 2). The magnitude of these differences was small to moderate, with effect sizes ranging from 0.20 to 0.44. Meaningful limitation (greater than 60%) in IADLs was also significantly associated with pain severity, but not analgesic intake, with 21% of treated patients and 12% of non-treated patients with such a degree of limitation in the severe pain subgroup (Fig. 1). The analysis by category of functional limitation again showed significant differences in distribution of degrees of limitation according to pain severity as well as in each subgroup between treated and not treated with analgesics (Fig. 2). This figure shows that, for instrumental activities, restriction increased as pain severity increases, and this restriction was greater in individuals who required analgesia. The main driver for such findings in IADLs was the item exploring the patient's ability to do hard housekeeping compared to the other items in the domain (p < 0.001, Fig. 3b). The percentage of patients helped by a caregiver on IADL increased significantly with severity of pain, regardless whether they were receiving analgesics or not in last 2 weeks (Table 3).

Mental functioning
Adjusted mean limitation in mental functioning was even higher than in physical activities; it was found to be 34.7% (17.4) in the entire OA group and to have a significant linear association with pain severity (p < 0.001, Table 2). Limitation in this key domain was found to be 1 3 not significantly different by analgesic treatment status ( Table 2). Meaningful limitation (greater than 50%) in mental decline was also significantly associated with pain severity (but not analgesic intake), with 38% of treated patients and 34% of untreated patients with such a degree of limitation in the severe pain subgroup (Fig. 1). However, the analysis by category of this functional limitation showed significant differences in the distribution of degrees of limitation according to pain severity, but not between those treated and those untreated with analgesics (Fig. 2). Figure 2 shows that, for mental functioning, decline increased as pain severity increased, regardless of analgesic treatment.

Social functioning
Adjusted mean limitation in social functioning was found to be 18.2% (35.4) overall. A significant association with pain severity was found (p < 0.001), regardless of whether patients were treated with analgesics (Table 2). Nonetheless, treated OA patients showed significantly higher restriction of social activity than non-treated OA patients in all three subgroups by pain severity ( Table 2). The magnitude of these differences was small, with effect sizes ranging from 0.11 to 0.30. Meaningful limitation (100% restriction of social activity) was also significantly associated with pain severity (but not analgesic intake, except for OA subjects with mild pain, p = 0.006), with 32% of treated patients and 25% of non-treated patients with such a degree of limitation in the severe pain subgroup (Fig. 1). The analysis by category of functional limitation showed significant differences in the distribution of degrees of limitation according to pain severity, but only a trend toward statistical significance by analgesic status (Fig. 2). Figure 2 also shows that, for social functioning, restriction increased as pain severity increased, regardless of analgesic treatment.

Cognitive functioning
Adjusted mean limitation in cognitive functioning was 16.8% (23.3) for OA patients, but also with a significant association with pain severity (p < 0.001, Table 2). Only OA patients with treated mild pain showed significantly greater cognitive deterioration than non-treated OA patients (Table 2), with a negligible effect size of 0.18. Meaningful limitation (severe to total inability with regard to remembering and concentrating) was also significantly associated with pain severity (but not analgesic intake, except for OA subjects with mild pain, p = 0.038), with 17% of treated patients and 11% of non-treated patients with such a degree of limitation in the severe pain subgroup (Fig. 1). The analysis by category of functional limitation showed significant differences in the distribution of degrees of limitation according to pain severity, but not by analgesic status (Fig. 2). Figure 2 also shows that, for cognitive functioning, deterioration increased as pain severity increased, regardless of analgesic treatment. The analysis by gender (Table 4, Supporting information) replicated the findings observed in the entire group, mainly in OA patients with moderate or severe pain in all of the patient´s functioning domains assessed.

Discussion
This study was conducted with the objective of determining whether patient self-reported functioning in key domains of daily living differs by pain severity and analgesic treatment among older adults with OA in Spain. To do this, the data from the latest available ENSE survey were used [22]; this could be considered a strength of this study due to the large size of the sample analyzed and its nationwide representability, even though data capture was based on persons surveyed. Nonetheless, self-rated OA severity and other outcomes, including pain, function, productivity, and costs suggest that self-reporting of OA severity provides an accurate and tangible assessment of patients' perceptions of their disease [15,34,35]. Another strength of using the ENSE survey was that the self-reported prevalence of OA was more or less the same as the physician-based Spanish prevalence study of rheumatologic diseases (the EPISER study) conducted recently by the Spanish Association of Rheumatology [36]; the prevalence of OA in persons aged 40 years or older was 29.4% in the EPISER study versus 29.8% in the ENSE survey. In the analysis included here, a large sample comprising a total of 3389 OA patients aged 65 years or older was analyzed. However, while analysis was restricted to this population, it is ultimately the population that is most representative of osteoarthritis; 67.4% of patients with self-reported OA in Spain are 65 years of age or older [22,36]. Also, the study analyzed up to five different key domains of patient functioning in daily life; that too could be considered as a strength of this study as, to the best of our knowledge, there are no existing similar studies in the Spanish population. As OA is a heterogeneous disease, several limitations of this study are to be pointed out. First, it lacks detailed data with regard to OA location; second, it lacks data regarding whether OA affects a single joint or several joints; third, it lacks a radiographic assessment scale; and fourth, it lacks data about the relationship between OA and other comorbidities and their relationship to the functional impairment found. Moreover, it must be taken into account that the survey collected information on pharmacological analgesic treatment of OA in a generic fashion (treated or not treated with usual analgesics in the last 2 weeks), but lacked detailed information on active substances, doses, etc., as well as information regarding non-pharmacological treatment (which is central to OA treatment). Those data were not reported in the survey; moreover, the data in the survey are self-reported by survey participants rather than drawn from medical databases or directly from medical personnel. Furthermore, although the size of the sample analyzed was large, some subgroups of patients (particularly OA subjects with untreated severe pain) could be under-represented in the study.
Regarding the comparability of this analysis, a similar study was conducted by Stamm et al. in Austria, analyzing the impact of pain on activities of daily living [21]. Our study of functional capacity found that these patients have a considerable degree of limitation in carrying out both basic and instrumental activities of daily living, mainly those with severe pain. Mobility (walking and up/downstairs) and hard housekeeping were the specific activities most seriously impacted by OA pain and accounted for the most weight (around 17-23% per item) in the total scores for BADLs and IADLs, respectively, when compared with other items in such domains. In addition, patients with OA experienced restriction and decline in other essential functions such as cognition, social functioning and mental health. In all these domains of functioning, a significant association was found between pain severity and impairment, such that the higher the pain, the higher the impairment, regardless of treatment status; this points to an increased need for external care. In fact, we found that the need for caregiving on BADL and IADL increased significantly with severity of pain, regardless of concomitant analgesic therapy; about half of OA patients with severe pain needed a caregiver for basic and instrumental activities daily. In addition, OA subjects treated with common analgesics showed significantly higher impairment in BADLs, IADLs, and social functioning, particularly subjects with moderate or severe pain; limitation of social functioning was as high as 42% compared with 32% in the group of patients not treated with analgesia. The study by Stamm et al. also found a significant association between pain severity and impact thereof on both BADLs and IADLs in older adults with OA among other disorders [21]. The European Project on OSteoArthritis (EPOSA study), aimed at investigating the consequences of OA in older adults using data from six European countries, including Spain [37], found that clinical OA, present in one or two sites (hip and knee), or in two or three sites (hip, knee, and hand), increased the risk of social isolation, cognitive impairment and worse walking and physical activity [38,39]. These  findings are consistent with our analysis, particularly in the moderate to severe pain group, although the EPOSA study did not analyze whether the presence of pain interferes with activities of daily living [40]. Likewise, more severe and stable joint pain levels were found to be associated with anxiety and depressive symptoms in older persons with OA [41]; this is consistent with the impairment in mental functioning found in the analysis of older Spanish subjects with OA included in the 2017 ENSE survey. Research aiming to demonstrate that individuals with OA are more likely to experience impairment in their ability to perform activities of daily living has been seen previously, but in selected populations and mainly on patients with OA in specific health settings or on waiting lists for joint replacement [17][18][19][20]. Recently, persistent opioid users were found to be more likely than nonusers to report limitations in IADLs and physical, social, and cognitive functioning in follow-up [42]. Also, a recent study carried out in Spain, linked opioid utilization in OA pain with increase in cognitive deficit and dependence for BADL as assessed with Barthel scale in OA patients above 65 years old [43]. Our analysis is also consistent with data regarding poorer patient functioning in BADLs, IADLs, and social functioning in subjects using analgesic drugs. Given the nature of this study, not allowing to see a cause-effect relationship, it is possible that treated patients had more severe joint destruction; it is also possible that treated patients had developed a phenomena of secondary central sensitization to pain, due to a failure to achieve optimal pain management, perhaps because they had partially responded or not responded to conventional analgesic therapies, and other psychopathological reasons. Although not evaluated in this survey, frailty, defined as a geriatric syndrome that involves loss of functioning and is a prognostic factor for disability [44], might have been present in a significant percentage of these patients, given the impairment observed in activities of daily living and mental function, which might have been correlated with low physical activity, overall weakness, exhaustion and overall slowness [44].
The gender analysis showed similar findings. Due to limitations in sample sizes, differences did not reach statistical significance in some comparisons; regardless, our analysis failed to find significant differences between treated and untreated patients within the subset of patients with meaningful limitation (the most impaired category) in each of the key domains of functioning analyzed in this study, although this subset was larger in treated subgroups. Nonetheless, the association between pain severity and percentage of OA patients with significant limitation was present in each of the domains of functioning evaluated, with significantly more subjects with meaningful restriction in the severe pain category subgroup. Also, clinical OA was associated with frailty and pre-frailty in older adults in European countries [37]. The prevalence and burden of OA is growing as both population aging and obesity rates rise; this means that the viability of joint replacement programs will be at risk by 2030. Unmet needs exist regarding not only disease progression modifiers but also pain management. Hence, this disease merits more attention, not only from medical researchers, but also from healthcare policymakers, who must work towards coordinating a holistic approach that includes comprehensive OA treatment and adequate pain management. In conclusion, and despite possible limitations, our findings in a representative nationwide sample in Spain of patients aged 65 years and older with treated moderate to severe OA pain, suggest that said pain was associated with a substantial impairment in five key domains of patient functioning. Pain severity was the major determinant of disability, and all other study variables worsened as pain severity increased, including caregiving need, regardless of the analgesia approach pursued.

Author contributions
The manuscript was prepared jointly by all authors, and all of them had the right to review, comment on, and approve the manuscript prior to submission. The data reported within the manuscript are not proprietary to any of the authors. JR was responsible for the study concept. NL, SG, JR, and FC were responsible for data analysis, data abstraction, and data interpretation. The analysis was conducted by NL and SG to obtain their master's degree in Health Assessment and Market Access at Universidad Carlos III, Madrid, Spain, under the direction of JR, Associate Professor for said master's degree. JR is also employed at Pfizer, SLU, Alcobendas (Madrid), Spain. Francisco Castro is a rheumatologist at Centro Médico Teknon and Hospital Universitari Sagrat Cor in Barcelona, Spain, and participated in the preparation of the manuscript and interpretation of the findings.
Funding This work has not received any funding.
Data availability All data generated or analyzed during this investigation are included in this published article. Any additional information is available from the corresponding author on reasonable request.

Conflict of interest
The authors declare that they have no conflicts of interest with respect to this study.

Ethics approval
The analysis carried out in this study used aggregate dated available in the public domain only, thus, de-identification was guaranteed to prevent anyone's personal identity from being revealed. Therefore, studies such as this one are exempt from IRB review and from classification by the Spanish Agency of Medicines and Medical Devices according to current regulations for post-authorization observational studies (https:// www. aemps. gob. es/ inves tigac ionCl inica/ medic ament os/ estud iosPo staut oriza cion. htm# norEs tatal). The confidentiality of the records (anonymous and dissociated) was respected according to the existing Spanish Law on Personal Data Protection (Law 15/1999, of 13 December 1999, on Personal Data Protection).
Consent to participate Individuals included in the ENSE survey participated voluntarily and gave their personal consent prior to their participation in the survey (see details on survey implementation at: https:// www. mscbs. gob. es/ estad Estud ios/ estad istic as/ encue staNa cional/ encue sta20 17. htm).

Consent for publication
The manuscript was prepared jointly by all authors, and all of them had the right to review, comment on, and approve the manuscript prior to submission.The manuscript has been read and approved by all authors who also meet the requirements for authorship. Each author agrees that the manuscript represents honest work.