Factors associated with general practitioner visits for pain in people experiencing chronic pain

Keywords: chronic pain, general practitioner (GP), health care utilisation, pain interference, pain severity, patient-reported outcomes.

Chronic or persistent pain (CP), defined as pain that persists or recurs for longer than 3 months, is a common, complex and distressing issue that has a negative impact on individuals and society.¹ CP can affect a person’s quality of life and their ability to perform daily activities² and is the most common reason people seek health care.¹ People with CP are high users of general practitioners (GPs)³ and inpatient hospital services,⁴ which together are major drivers of health care costs. In Australia, CP affects one in five adults aged 45 and over, and the national estimated cost for CP in 2018 was 139 billion AUD.⁵ Although Australians with CP commonly visit GPs,⁵ there is a paucity of research that explores the factors associated with GP use. This study will examine the association between rate of pain-related GP visits for patients with CP and relevant factors such as pain severity, pain interference, mood, pain-related cognitions and patient demography.

CP affects health care utilisation (HCU) across a range of services.⁶ For example, people with CP tend to have higher-cost hospital admissions compared to patients without CP,⁴ and an increased length of stay when hospitalised.⁵ The impact of CP on primary care, in particular, has been reported in previous studies, including a Canadian study that demonstrated patients with CP had significantly higher rates of physician visits attributed to pain-related conditions compared to patients without CP.⁴ Another population-based cross-sectional study in Germany reported that among adolescent patients who reported chronic pain (N = 749), 30.2% visited a physician at least two times during the past 3 months when surveyed, with a further 18.6% of patients visiting three or more times.⁷ Another study evaluated the impact of CP in 15 European countries and found that 60% of patients with CP had visited their doctor about their pain between two and nine times in the 6 months before completing the study survey about pain.⁸ Most of the patients (70%) visited a GP to seek treatment. Although there are studies that have reported high HCU by patients with CP, it was unclear whether visits were related specifically to pain or were for other reasons.^9,10 This article will assess factors associated with the rate of GP visits among patients who experience CP.

Effective identification of the predictors for high pain-related GP use by people with CP would allow targeting of interventions toward improving care for this vulnerable group. Individuals living with CP are more likely to experience mental health conditions, including depression, anxiety or sleep disturbance, and to report physical impairment, fatigue and limitations to activities of daily living.^2,5,11 Furthermore, patients with CP are more likely to be female, older and have a long-term health condition.⁵ The limited available evidence suggests that demographic factors such as sex, age and low socioeconomic status may be associated with increased use of primary care by people with CP, along with health-related factors such as chronic disease and poor self-rated health.^10,12 Consequently, our study aimed to assess relationships between the rate of GP visits for pain and potentially relevant demographic, health and pain-related factors for people attending specialist pain management services (SPMSs) in Australia.

This study utilised data collected by the 57 Australian adult SPMSs that were participating in the electronic Persistent Pain Outcomes Collaboration (ePPOC) for the period January 2015 to December 2021. ePPOC is an Australasian integrated outcomes data collaboration that aims to improve services and outcomes for people experiencing persistent pain (https://www.uow.edu.au/ahsri/eppoc/). Since 2013, ePPOC has received data from SPMSs encompassing more than 100 000 episodes of care. The establishment and implementation of ePPOC (and its paediatric counterpart, PaedePPOC) have been described in detail elsewhere.^13,14 A standard suite of patient-reported outcomes measures (PROMs) is used by SPMSs at referral and at various points in the care pathway. The de-identified data, along with patient demographic and health information, is sent to ePPOC for reporting, research and benchmarking purposes. ePPOC has developed binational benchmarking criteria across nine clinical domains that allow each service to compare their patient-reported outcomes to others. This in turn assists services to identify best practice protocols and clinical variation, driving and informing quality improvement for services and the sector more broadly.

This study focused on adult participants (aged 18 years or older) referred to a SPMS in Australia. It included patients who completed a referral questionnaire between 1 January 2015 and 31 December 2021 (N = 84 829), and those who reported experiencing pain for more than 3 months. Additionally, patients were required to have answered the question regarding GP visits attended in the past 3 months, which is the dependent (outcome) variable for this study. Questionnaires are provided to patients when a referral is received by a SPMS, and before they start any episode of care. Therefore, the referral date is the date a pain management service receives a referral to provide pain management for a patient.

Upon referral to a SPMS, patients are asked, ‘How many times in the past 3 months have you seen a general practitioner in regard to your pain’. Using this item as the dependent variable, this study assessed the association between the rate of GP use in the previous 3 months and potential predictors such as pain severity, pain interference, mood, cognition and opioid use.

Data items collected at referral were used as independent variables in this study. The patient’s age was measured in years. Country of birth was categorised as Australia or other. A measure of relative residential area disadvantage was derived by coding the patient’s residential postcode to the Socio-Economic Index for Areas – Index of Relative Disadvantage (SEIFA-IRSD 2016)¹⁵ in Australia. The SEIFA-IRDS scores can be shown as quantiles such as deciles or quintiles.¹⁶ In this study, disadvantage area quintiles range from 1 (lowest disadvantage) to 5 (highest disadvantage). This covariate reflects the level of disadvantage of the patient’s residential area rather than the patient’s individual socio-economic status.

Body mass index (BMI) was calculated using the patient’s height and weight and categorised according to guidelines from the World Health Organization (WHO):¹⁷ underweight (< 18.5), normal weight (18.5 to < 25), overweight (25 to < 30), obese class I (30 to < 35), obese class II (35 to < 40) and obese class III (> 40). A measure of pain duration was obtained from answers to the question, ‘How long has the main pain been present?’ with five possible response categories: less than 3 months, 3–12 months, 12 months–2 years, 2–5 years and more than 5 years.

The Depression, Anxiety and Stress Scale (DASS) measures the symptoms of depression, anxiety and stress. The DASS is derived from 21 questions which are scaled from 0 to 3, where 0 = ‘did not apply to me at all’, 1 = ‘applied to me some degree or some of the time’, 2 = ‘applied to me to a considerable degree or a good part of the time’, and 3 = ‘applied to me very much or most of the time’. Scores for depression, anxiety and stress were calculated by summing the scores for the relevant questions from DASS items. Each question had seven items that were multiplied by two to enable comparison with the full-scale DASS42.¹⁸ Therefore, depression, anxiety and stress were categorised as: normal (score: depression 0–9; anxiety 0–7; stress 0–14), mild (score: depression 10–13; anxiety 8–9; stress 15–18), moderate (score: depression 14–20; anxiety 10–14; stress 19–25) and severe/extremely severe (score: depression ≥ 21; anxiety ≥ 15; stress ≥ 26).

Pain severity and pain interference were measured using the Brief Pain Inventory.¹⁹ To measure pain severity, patients were asked four questions to rate their pain as ‘worst in the last week’, ‘least in the last week’, ‘on average’ and ‘pain right now’. Questions were rated from ‘0’ to ‘10’, where ‘0’ indicates ‘no pain’ and ‘10’ indicates ‘pain as bad as you can imagine’. Pain severity was then calculated as an average of those four questions. Pain interference questions were rated on a scale of 0–10, where 0 indicates ‘no interference’ and ‘10’ indicates ‘completely interferes’. The pain interference score was an average of the seven pain interference questions. Both pain severity and pain interference were categorised as mild (score: 0–4), moderate (score: 5–6) and severe (score: 7–10).

Pain catastrophising measures the patient’s emotional response, thoughts and feelings to actual or anticipated pain. To measure pain catastrophising, the Pain Catastrophizing Scale (PCS)²⁰ was used, which includes 13 questions. For each question, the patient indicated the degree to which they have thoughts and feelings about their pain using a 5-point Likert scale, where 0 = ‘not at all’, 1 = ‘to a slight degree’, 2 = ‘to a moderate degree’, 3 = ‘to a great degree’ and 4 = ‘all the time’.²¹ The PCS total score ranges from 0 to 52, and was categorised as mild (< 20), high (20–30) and severe (> 30).

The Pain Self-Efficacy Questionnaire (PSEQ) is a measure of how confident a patient is that he or she can complete a range of activities despite their pain.²² Patients were asked ten questions to rate their confidence in undertaking different activities, where 0 = ‘not at all confident’ and 6 = ‘completely confident’. Scores on the PSEQ were categorised as severe (< 20), moderate (20–30), mild (31–40) and minimal (> 40). Patients were also asked at referral about any opioid use, which was dichotomised as ‘no’ and ‘yes’. As the opioid use question was only introduced in 2018, data from 2018 to 2021 were included in the calculation of the adjusted incident rate ratios (IRRs) from the multivariable analysis.

We reported the number of GP visits by a patient’s demographic details, health characteristics and pain status using means and standard deviations. The mean was reported with a 95% confidence interval. The number of GP visits was plotted, with the distribution appearing highly right-skewed. Therefore, we applied the count regression model to fit the data. The maximum number of GP visits was 60 in 0.03% of patients. Although there is no consensus on what defines frequent or the highest number of visits, a previous study demonstrated that patients with CP would have 2–24 GP visits over 2–48 months.¹² Therefore, we trimmed the data at 24 maximum visits (99.5% of the data) and conducted a sensitivity analysis by including GP visits to a maximum of 60 visits.

In this study, the outcome (rate of GP visits) is a count variable with a skewed distribution. As the variance is higher than the mean, we proposed a negative regression model over Poisson regression. In the negative binomial model, the mean and variance of the outcome variable are not assumed to be equal.²³ Furthermore, a likelihood ratio test was conducted to test whether the negative regression model was a more appropriate fit than the Poisson regression model. Given the Akaike information criterion (AIC) and likelihood ratio test information, the negative-binomial regression model had the best fit. Depression had shown a significant correlation with pain catastrophising and other variables. This causes a multicollinearity issue in the analysis and leads to a biased estimate. Therefore, we excluded it from the final model which consisted of all variables. A MASS²⁴ package in R software, Vs 4.1.0, was used to conduct the statistical analysis.

Overall, 84 829 patients (58% female) completed a referral questionnaire at a SPMS in Australia during 2015–2021. The average age of patients was 51.8 years (female = 52.2 years; male = 51.4 years) and most (70.8%) were born in Australia (Table 1). Severe, moderate and mild pain severity was reported at referral by 50.1, 36.9 and 13.0% of patients, respectively. Likewise, 65.7, 22.6 and 11.7% of patients reported severe, moderate and mild pain interference at referral, respectively. Half of the patients (50.0%) reported severe pain catastrophising, and more than three-quarters (78.72%) reported moderately or severely impaired pain self-efficacy at referral. The majority of patients (60.6%) reported using opioids.

Table 1.  Association of sociodemographic and pain-related characteristics with the number of patient GP visits in the past 3 months (2015–2021).

The number of GP visits for pain management per patient ranged from 0 to 24 (mean = 4.06; s.d. = 3.91) in the 3 months before referral. Overall, 168 GP visits per 100 patients per month were reported. On average, patients had more than one GP visit per month. More than half of the patients (55.19%) reported that they saw their GP more than three times for pain-related treatment in the 3 months before completing the referral questionnaire. The percentage of patients with one, two and three GP visits was 7.24, 10.93 and 21.79%, respectively (Fig. 1). The rate of GP visits was not normally distributed. The percentage of patients with six or more GP visits in 3 months was smaller. The figure shows the rate of GP visits is right skewed. Therefore, in the following section, factors associated with GP visits have been provided after fitting an appropriate model that accommodates the rate of GP visits distribution. The rate of GP visits decreased with longer pain duration. The average number of GP visits in patients with severely impaired pain self-efficacy was higher compared to patients with moderate, mild or minimal pain self-efficacy. Patients who were using opioid medications had greater mean GP visits (mean = 5.59) compared to patients who did not (mean = 4.16).

Fig. 1.  Number of GP visits reported (in 3 months prior to start of the episode of care at a SPMS) in patients with chronic pain.

The increased rate of GP visits was associated significantly with the patient’s sex, area of socioeconomic disadvantage, anxiety, pain severity, pain interference, pain catastrophising, pain self-efficacy and opioid use (Table 1). The multivariate analysis showed that patients with longer pain duration had a lower rate of GP visits compared to patients who have had pain for less than 12 months. The rate of GP visits was 1.22 (IRR = 1.22, 95% CI: 1.19, 1.26) times higher in patients with severe pain severity, compared to patients with mild pain severity. Furthermore, the rate of GP visits in patients with severely impaired pain self-efficacy was 1.31 times higher than in patients with mildly impaired self-efficacy. Generally, decreased self-efficacy is associated with increased GP visits. Using opioids increased the likelihood of GP visits by 1.32 (IRR = 1.32, 95% CI: 1.30, 1.34). A statistically significant but smaller association was observed for the patient’s sex, socioeconomic status, anxiety, pain interference and pain catastrophising.

This study explored the rate of GP visits and other associated factors in patients with CP using data collected by adult SPMSs that participated in the ePPOC. The multivariate analysis showed that patients who have had pain for a longer time were less likely to visit a GP. Additionally, patients who use opioids were more likely to visit a GP compared to patients who did not use opioids. While severe pain was associated with increased GP visits, patients with minimally impaired pain self-efficacy had less likelihood of a GP visit. The association of anxiety, pain interference and pain catastrophising with the rate of GP visits was found to be statistically significant.

This study showed that patients who use opioids are more likely to have frequent GP visits. Similar study findings in the USA revealed that having an opioid prescription for CP, in the year prior to pain treatment onset, were associated with increased HCU for CP.²⁵ This might reflect that patients with CP tend to access the health service for an opioid prescription. Furthermore, the increased opioid prescribing in Australia,²⁶ with potential opioid adverse events, misuse and dependence, might also account for increased rates of GP visits for CP. Studies have revealed that patients who use opioids have increased rate of depression, anxiety and pain severity compared to patients who do not use opioids.²⁷

Patients who reported severely impaired self-efficacy had a 31% increase in the likelihood of visiting GPs compared to patients with minimally impaired pain self-efficacy. This indicates that improving patients’ self-efficacy may decrease repeated GP visits. Improved self-efficacy would result in more effective pain coping skills and management of pain symptoms. Building patient self-efficacy to manage pain is a key feature of a successful self-management intervention in patients with CP.^28,29 A recent study in England compared patients’ GP visits for pain-related treatment before and after pain management programmes, and self-efficacy was measured using pain self-efficacy questionnaires (PSEQs).³⁰ It reported that improved self-efficacy is associated with a reduction in patients’ need to consult their GP as frequently about their pain.

Patients were asked to rate the intensity of their pain using the BPI. Patients with severe pain had a higher rate of GP visits compared to patients with low pain severity at referral. This is expected as patients with severe pain conditions look for health professionals to intervene. Likewise, a previous study also reported that patients who sought health care for their CP were more likely to have severe pain and severe pain interference.³¹ A study in Sweden reported that severe CP has a significant association with frequent health care seeking.³² In our study, increased pain duration was associated with less frequent GP visits. Patients who had pain for more than a year showed less frequent GP visits than patients who had pain for less than a year. Patients may not be visiting their GP as frequently if there has not been an improvement or may have been referred to other services including SPMSs and allied health professionals. Furthermore, patients with pain duration of more than a year might have had frequent GP visits when the pain began, and this frequency has tapered as they await treatment by a SPMS. However, this finding warrants further study to characterise determinant factors of fewer GP visits in patients with CP who have had CP for more than a year.³³

A strength of this study is its large sample size, with data obtained from 57 SPMSs. This study aims to increase the understanding of what covariates have a role in predicting pain-related HCU of patients experiencing CP. A limitation of this study is the lack of some covariates that may modify the relationship between CP and HCU. For example, household income and size, occupation, education, historical trauma and marital status may also have an association with the experience of CP,^12,31,34 although these characteristics were not included in our study. Another potential limitation of this study is that the sample included CP patients who had been referred to SPMSs, and who may have more complex care needs. Therefore, the interpretation of this study's findings may require caution when it comes to CP patients who were not referred to SPMSs. As most CP is managed in a primary setting, similar studies that characterise patients’ health service use is warranted. Finally, while this study included data from 2015, the opioid use question was introduced in 2018, reducing the sample size included in the analysis, but having no impact on the interpretation of the finding.

More than half of the adult patients with CP had greater than three GP visits in the 3 months before they started an episode of care at a SPMS. In addition to pain severity, impaired pain self-efficacy, pain duration and opioid use were highly associated with the rate of GP visits in patients with CP. Future qualitative studies would be warranted to get an in-depth insight into the inverse association between pain duration and the rate of GP visits. Increased presentations from individuals prescribed opioids also highlight the ongoing importance of robust evidence-based practice regarding the use of opioids in CP management, as the impact on GP utilisation could be significant. Robust self-management strategies targeted at improving perceived self-efficacy cannot only improve outcomes for individuals, but lower the utilisation of GPs in an environment of increasing resource constraints.

The data that support this study were obtained from ePPOC. Data will be shared upon reasonable request to the corresponding author with permission from ePPOC.

Authors do not have any conflicts of interest.

This research did not receive any specific funding.