The use of healthcare services and disabling chronic pain: results from the cross-sectional population-based Andalusian Health Survey

Abstract Background Several factors seem to be related to the use of healthcare services, and chronic pain (CP) is among these characteristics. The objective is to describe the number of visits to a doctor’s surgery or emergency rooms, and the periods of hospitalization; to identify characteristics associated with frequent healthcare use, including disabling chronic pain (DCP) and non-disabling chronic pain (n-DCP). Methods Representative population-based cross-sectional study of 6569 people older than 16 years from southern Spain was collected. The frequency of visits to a doctor’s surgery or emergency rooms and periods of hospitalization were defined as at or above the 90th percentile. Binary logistic regression analyses were conducted separately on women and men to identify characteristics associated with being frequent visitors. Results People with DCP are more frequent visitors to a doctor’s surgery and emergency rooms and endure longer periods of hospitalization compared to people with n-DCP and without pain. In logistic regression models, people with DCP are twice as likely to over-visit a doctor’s surgery; to endure longer periods of hospitalization and more visits to an emergency room service. No relationship was found in n-DCP. Conclusions Disability seems to modulate a greater use of health services among the population with CP, doubling it when compared to n-DCP and n-CP, both in women and men. Understanding the role of disability in the use of healthcare services for individuals with CP allows for the identification of needs and strategies to optimize resources.


C
hronic pain (CP) affects approximately 20-40% of people throughout diverse populations around the world. 1 In Spain, CP prevalence is estimated at 17% (95% CI, 16.88-17.19)affecting around 6.7 million people older than 16 years of age. 2 Treatment involves multiple modalities provided by one or more healthcare professionals or healthcare services.This is also associated with an increase in medical expenditures and a considerable economic burden for individuals and society. 3Although CP is the first reason for medical appointments and their inclusion is essential to reflect the burden on society comprehensively, 4 only a few studies assessed medical costs that affected people with CP pain directly. 5espite efforts to estimate the associated cost, the exact figures are difficult to determine since global data are not available. 6ver the last decade, in Canada, CP cost has represented the three largest sources of healthcare cost, exceeding 20 billion dollars. 7The cost is estimated to be over 300 billion euros in Europe, which correspond to around 1.5-3% of Gross Domestic Product. 8,9n Spain, where 96.5% of the population is covered by the National Public Health System, there are no global studies that make it possible to accurately calculate the healthcare cost associated with CP. 10 However, it is estimated that, including direct and indirect costs, it reaches 16 billion euros per year, 11 which corresponds to 2-2.5% of Gross Domestic Product. 12From the point of view of direct costs, CP accounts for 50% of Primary Care consultations and 8% of pharmaceutical spending, placing it as one of the main health problems. 6st is related to healthcare visits, hospitalization periods and other therapies. 13In fact, some studies found that CP increased healthcare service use 14 and others consider it as one of the main reasons for medical appointments. 15urthermore, several studies coincide in assessing CP when it causes disability, understanding it as dysfunction, limitation in activity and restriction in social participation. 16In Spain, a study carried out by Cabrera-Le� on et al. stated that 11.36% (11.23-11.49) of the population over 16 years of age suffer from disabling chronic pain (DCP). 2 Disability seems to be a predictor that increases healthcare costs and the number of visits to health services.In this regard, another study carried out in the United States by Stockbridge et al., (2015) found that healthcare expenditure per person increased by 4000 dollars when CP caused disability.This figure rises to more than 13 000 dollars when pain interference is severe, concluding that a severe degree of CP is associated with substantially higher healthcare expenditures. 17n addition, several factors have been related to healthcare use levels due to CP, 18 including social status, 2 age, marital status, 19 sex; pain intensity, 20 comorbidities 19 and occupational situation. 21herefore, DCP represents a major problem for public health and health systems 13 given that studies indicate that this population presents a greater use of health services. 22The objectives of this study were: (i) to describe the number of visits to a doctor's surgery; periods of hospitalization; and visits to emergency rooms; and (ii) to identify characteristics associated with frequent healthcare use, including DCP and non-disabling chronic pain (n-DCP).

Design and data collection
This cross-sectional study used data obtained from the last Andalusian Health Survey (EAS in Spanish) 2016 edition which was carried out through face-to-face, population-based home interviews.The EAS was designed to evaluate the health status and the use of health services in Andalusia, which is a southern region of Spain and also the fifth most populated region in Europe, with 8.5 million inhabitants.For the design of this study, Sex and Gender Equity in Research (SAGER) guidelines were followed.
The study sample comprised a regionally representative group of 6569 adults (16 years and over).The sample size was calculated for a confidence level of 95%, an expected population percentage of 50% (P ¼ Q), a margin of error for global estimates of ±1.413, and a design effect of 1.35, resulting in a prefixed sample of 6484 individuals.A multistage stratified sample design was adopted using municipalities, census tracts, households and individuals as sampling units.Province, size of municipality and season of the year were considered strata.Municipalities and census tract were selected in proportion to the population size, households with the same probability by systematic sampling, and interviewees by applying stratification for each size of municipality within provinces and quotas for sex-age.
Finally, the interviewer invited one adult per household to participate.The inclusion criteria of this study were people older than 16 years and residing in Andalusia but not institutionalized (e.g.residences, hospitals, prisons, etc.).The survey obtained a response rate of 70.9%.For more information regarding the Andalusian Health Survey, the Andalusian School of Public Health published a report with the details of the sample and methods. 23

Frequent number of visits
To assess the frequency with which participants visited healthcare centres, three variables were created (visits to a doctor's surgery, hospitalization and emergency room visits) from the following questions: (i) how many times have you seen a doctor (public or private) in the last two weeks?; (ii) how many times have you been hospitalized in the last 12 months?and (iii) how many times did you have to use an emergency service in the last 12 months?
Participants answered each question separately, so 'frequent visits' were considered separately in every variable.Regarding the term 'frequent visits', although there is a lack of consensus in the literature about this concept, the most used definition is considered as the top 10%.Thus, 'frequent visits' were denoted as those reporting at or above the 90th percentile for each healthcare professional or setting. 24The rest of the participants were categorized as 'infrequent visits'.

Disabling chronic pain and non-disabling chronic pain
The diagnostic criterion to identify the group with DCP and n-DCP was established from the creation of a variable on CP, according to the individuals who declared that a doctor or a nurse had told them that they suffered from one or more of the following types of pain: 'angina/chest pain', 'back pain', 'neck pain', 'shoulder pain', 'waist pain', 'cervical/lower back pain', 'fibromyalgia', 'migraine/headache/ chronic cephalgia/frequent headache' and/or 'menstrual pain'.From the CP variable, two new variables were created composed of disability and CP: DCP and n-DCP. Regarding impairments (problems with bodily functions/structure), interviewers asked people (at home, face-to-face) whether a doctor or a nurse had told them that they suffered from one of the types of pain in the aforementioned list.Activity limitation and participation restriction were constructed from people who declared that they were limited in their daily activity.DCP was formed of people with CP who suffered from a disability and the variable n-DCP of people who declared that they suffered from CP but were not limited in their daily activity. 2

Number of chronic conditions
The variable number of chronic conditions (not including pain diagnoses) was created using the question that assesses whether a doctor or a nurse had told them that they suffered from any of the following chronic conditions: cancer; diabetes; high blood pressure; cholesterol; inflammatory bowel disease; constipation; stomach/duodenum ulcer; pulmonary chronic disease; asthma; cardiac disease; cardiac failure; chronic skin problems; chronic allergies; anaemia; blood circulation problems; varicose veins; haemorrhoids; stroke; depression; anxiety; other mental problems; hearing impairment; cataracts; arthritis or rheumatism; osteoporosis; liver disease; renal disorder; overflow incontinence; infertility and/or prostate disease.

Sociodemographic information and lifestyle variables
A selection of independent variables and potential confounders that were included in the analysis was based on a review of prior literature. 26,27The baseline EAS survey demographics included were age (16-44 years; 45-64 years; þ65 years), marital status (widowed; divorced, separated; singled; married), alcohol consumption (never consumed; previously consumed; currently consumed), tobacco consumption (never smoked; previously smoked; currently smoked), type of locality (less than 10 000 inhabitants; 10 000-50 000 inhabitants; more than 50 000 inhabitants, capitals), employment status (employed; unemployed; retired/medical leave; housemaker; student), health insurance (public insurance, state mutuals, private insurance, no medical insurance, and public insurance þ private insurance), monthly incomes (<e999; e1000-e1999; >e2000) and highest level of education.The latter was categorized into three categories according to the National Classification of Education (CNED in Spanish) (primary education, secondary education and higher education). 28

Ethics
The Andalusian Health Survey was supervised and approved by the review board of the General Secretariat of Quality and Public Health for the Health Ministry of the Andalusian Regional Government, and the Andalusian School of Public Health made the data confidential.In addition, to conduct the secondary analysis of the data of this study, consent was requested from the Bioethics Committee for Research at the Virgin Macarena-Virgin del Rocio University Hospitals (Spain), with the code 0057-N-19.

Statistical analysis
Sociodemographic and general health characteristics were expressed by mean (� x) with confidence interval (CI) in the continuous variables and using frequency and percentage in the categorical variables.The relationship between each sociodemographic and general health characteristic and frequent healthcare visits was first examined using ORs and 95% CIs.Reference categories were selected based on the hypothesized consideration of a higher risk of belonging to the group with a lower frequency of visits. 24 backwards manual logistics regression was used to determine the relationship between sociodemographic or health variables and frequent visits to healthcare services (visits to a doctor's surgery, hospitalization and emergency room visits).The models were initially adjusted using sociodemographic and health services as a control along with the remaining secondary variables.As no pair of sociodemographic or health variables were correlated at r � 0.4, all variables were treated as independent variables rather than as confounders.
The statistical criteria for determining the optimal model included comparing the information criteria.They were measures of goodness of fit, including the Akaike Information Criteria, the Bayesian Information Criteria (BIC), and the adjusted BIC (ABIC), with lower values indicating better fitting models.Taking this into account, variables were manually eliminated until the model with the best fit was identified.Significance was considered to be 5% throughout the statistical analysis (descriptive, bivariate and multivariate).A statistical analysis was conducted using IBM SPSS ® version 26.0 (IBM Corp., Armonk, NY, USA).

Characteristics of the sample
A total of 6569 participants answered the survey (50.8% were women and 49.2% were men).The mean age of women was 46.78 ± 18.15 years and men had an average age of 46.52 ± 18.34 years.In addition, table 1 shows data on age, marital status, highest level of education, monthly incomes, type of locality, employment status, number of chronic conditions and health insurance.

Frequent visits to a doctor's surgery in the last 15 days
The 90th percentile for frequent visits to a doctor's surgery was set at more than one visit in the last 15 days.Women (n-DCP: 6.3% vs. DCP: 21.0%) and men (n-DCP: 3.1% vs. DCP: 12.6%) with DCP were more likely to be frequent visitors (�90th percentile) to a doctor's surgery in the last 15 days than people with n-DCP.In addition, more women than men suffering from DCP are frequent visitors to these services (women: 21% vs. men: 12.6%) (table 2).

Frequent periods of hospitalization in the past year
The 90th percentile for frequent periods of hospitalization in the past year was set at more than one hospitalization period.In both groups, people with DCP showed a higher percentage of frequent hospitalization periods in comparison to people with n-DCP (women: n-DCP: 6.0% vs. DCP: 20.1%; men: n-DCP: 3.2% vs. DCP: 12.3%).In addition, when comparing frequent hospitalization periods between women and men with DCP, women showed a higher frequency of discharges from hospital compared to men (women: 20.1% vs. men: 12.3%; P < 0.05) (table 2).

Frequent visits to the emergency room in the past year
The 90th percentile for frequent visits to emergency rooms in the past year was set at more than one visit.In both groups, people with DCP showed a higher percentage of frequent visits to emergency rooms in comparison to people with n-DCP or no pain.In addition, when comparing frequent visits to emergency rooms between women and men with DCP, women showed a higher frequency of discharges from hospital compared to men (women: 24.3% vs. men: 12.6%; P < 0.05) (table 2).

Discussion
This study is one of the few studies that has analysed the use of health services in the population with CP, despite the fact that pain is considered the first cause of consultation with health services.The results have shown that people with CP more frequently visit a doctor's surgery, emergency room services and have more hospitalization periods than people without pain, both in women and in men, coinciding with the findings of other similar studies. 29In a study carried out by Breivik et al., 3 60% of participants with CP reported that they visited their doctor between two and nine times in the months prior to the start of the study and 10% went at least 10 times.
The results of another study carried out on the Portuguese population found that people with CP visited the doctor twice as often as people without pain. 12When comparing CP to other pathologies, another study found that people with pain visited their primary doctor's surgery up to 5 times more frequently than those people with other diseases. 29he results of this study could also be compared with those obtained in another with similar characteristics, carried out in Canada by Mann et al., 24 and whose objective was to investigate the role of CP in healthcare visits.However, the aforementioned article only discerned between people who suffer from CP or those who do not suffer from it.This study also includes the role of disability related to CP.
The regression models have shown that disabling CP is associated with a greater use of health services (�90th percentile), while no differences have been found in those where CP does not cause disability.The results agree with those obtained by Cabrera-Le� on and Cantero-Braojos, 23 who analysed the previous edition of the Andalusian Health Survey.There seems to be a trend towards a greater use of health services, and it would be disability that would play this modulating role. 31The use of general medicine and emergency services was significantly higher in the population with DCP (35.9% and 36.5%),respectively, while hospitalization is almost double (13% in DCP, 7.6% in n-DCP, and 7.1% in n-CP).A greater use of health services was also observed in people with DCP, coinciding with the results observed in the studies developed by Blyth et al., 22 and by H€ auser et al. 25 In addition, the use of health services did not differ significantly in the populations with n-DCP and n-CP. 32ne of the reasons that could explain the greater use of health services among people with DCP is that only a small percentage of these reach the Pain Unit consultations, so they tend to be referred to different medical specialists. 33Only 23% of people with CP visit a specialist and 2% are treated by a pain specialist. 34This circumstance causes the process that the person must go through before achieving a proper diagnosis and treatment to be long, tortuous, and costly in economic terms.Many patients go through multiple consultations without achieving the desired objective, and as a result the pain they experience starts to affect different areas of their lives and causes disability. 35n addition, the analyses of this study have been conducted separately on women and men, unlike that of Mann et al., 24 which did not disaggregate by sex.These analyses disaggregated by sex are supported by international regulations and take into account that hyperfrequency is different between women and men.Several researchers have shown that women are more likely to seek healthcare services for pain than men and, for this reason, women could be overrepresented in CP clinics and primary healthcare services. 36,37ccording to Jonsdottir et al., 34 interference of pain in daily life was the strongest predictor among women compared to men.In this sense, healthcare use in relation to CP may not only depend on its severity, but also on interference with daily life.However, results The use of healthcare services and DCP based on sex differences seem to be inconsistent and there are other sociodemographic factors associated with the prediction of a greater use of healthcare. 38,39his study has several strengths, among which it is worth highlighting that it is representative of the population, and its complex design, large sample size, very good response rate and data gathering made it a reliable information source.Measures were taken after providing adequate training to the interviewers to minimize biases related to information, observation, and measures.Finally, with respect to the capacity to generalize the results, although the calibration allows sampling errors and potential selection risks to be reduced, the results of a study carried out by Cabrera-Le� on et al., 26 with the same sample demonstrated that there were no differences between the calibrated and non-calibrated estimations.For this reason, we decided not to calibrate.
However, this study also presents a series of associated limitations.The CP variable was created based on self-reporting responses to two questions and may underestimate or overestimate the actual prevalence in a population.In addition, some respondents might not have been diagnosed by a doctor and others, although diagnosed, may not recall the diagnosis.In addition, each person's understanding of the concept of disability is different and subjective.Nevertheless, although self-reported measurements of chronic conditions have shown good levels of correspondence to medical records, this is a limitation to consider.Another limitation is that participants were not asked about lower or upper limbs, which, according to a study carried out by Dueñas et al., 36 may account for up to 35.9% of the cases of CP in Spain.However, this study is based on a secondary analysis of data from a population study so this limitation could not be remedied in the analysis design.Finally, since this study follows a cross-sectional design, causality cannot be established.
The results of this study have a significant impact on clinical practice, particularly in economic terms, as it directly addresses the common issue of access to healthcare services for individuals suffering from CP.It is important to note that there is a scarcity of DCP, disabling chronic pain; n-CP, no chronic pain; n-DCP, non-disabling chronic pain.a: HCP, healthcare provider.b: Percent below 90th percentile of visits with/without chronic pain.c: Percent at and above the 90th percentile of visits with/without chronic pain.
Figure 1 OR of the relationship between Chronic Pain and the use of health services adjusted for age, chronic conditions, marital status, highest level education, monthly incomes, type of locality, labour status, alcohol consumption, tobacco consumption and health insurance The use of healthcare services and DCP

Table 1
Sociodemographic and health characteristics of participants