Abstract

Objective

The objective of the study was to collect data on the direct and indirect economic cost of chronic pain among patients attending a pain management clinic in Ireland.

Setting

A tertiary pain management clinic serving a mixed urban and rural area in the West of Ireland.

Design

Data were collected from 100 patients using the Client Services Receipt Inventory and focused on direct and indirect costs of chronic pain.

Methods

Patients were questioned about health service utilization, payment methods, and relevant sociodemographics. Unit costs were multiplied by resource use data to obtain full costs. Cost drivers were then estimated.

Results

Our study showed a cost per patient of US$24,043 over a 12-month period. Over half of this was attributable to wage replacement costs and lost productivity in those unable to work because of pain. Hospital stays and outpatient hospital services were the main drivers for health care utilization costs, together accounting for 63% of the direct medical costs per study participant attending the pain clinic.

Conclusion

The cost of chronic pain among intensive service users is significant, and when extrapolated to a population level, these costs represent a very substantial economic burden.

Introduction

According to the International Association for the Study of Pain (IASP), chronic pain is pain that persists for longer than 3 months , and chronic non-cancer pain is estimated to affect around 19% of the European population . Chronic pain may have a significant impact on physical functioning, influencing employment capacity, and the role within the family and society. Elliot et al. found that about three quarters of individuals with chronic pain at baseline may still have chronic pain at 4-year follow-up. Indeed, some patients experience persistent or recurrent pain throughout their lives. International population-based studies in Denmark, Sweden, and Germany show that pain severity (measured variously by pain intensity, number of pain sites, and pain-related disability) is positively correlated with health care use . A thorough understanding of the economic cost of chronic pain is crucial for decisions regarding health service resource allocation, especially for high-intensity service users.

Internationally, some research has focused on the economic burden of chronic pain for the wider population. For example, an Australian study estimated the total cost of chronic pain to be US$34.3 billion, approximately 4.5% of gross domestic product (GDP) and 28% of annual health expenditure (the cost of chronic pain includes indirect costs that would not normally be included in health expenditure). This compares with US$18 billion and US$3 billion for the cost of heart disease and diabetes, respectively. In the context of a common monetary currency in Europe, it may be illuminating to compare the economic cost of chronic pain across nations—this may be helpful not only in terms of highlighting departures from the economic “norm” in some jurisdictions, but it may also be useful in the context of a growing trend toward cross-national research programs, such as the European Framework programs. Comprehensive data on the cost of chronic pain are relatively scarce, with many studies having concentrated on one type of pain only, e.g., back pain . Back pain alone is estimated to cost 2.2% of GDP in Germany and 1.7% of GDP in The Netherlands . More recently, in Ireland, chronic non-cancer pain was estimated to cost 2.5% of Ireland's GDP in 2008 . Although there is some variation in cost estimates between studies due to differing methodologies, it is nevertheless evident that chronic pain represents a substantial cost in terms of health care provision and lost productivity.

In the authors' view, some of the variations in reported cost are likely to be associated with differing models of service provision, but it also seems likely that costs vary according to the profile of persons using health services. While many people with chronic pain in Ireland are managed in primary care, a significant number attend specialized pain management services—it is the authors' experience that those seen in the tertiary or specialist services are among the most seriously affected by chronic pain. Consequently, we suspect that people attending a pain management service may also use more health services and incur higher costs for service provision and lost productivity. In determining the variables to include in our study, through consultation with experienced staff in the pain management service, we identified the typical profile of participants in terms of social welfare status, employment status, clinical profile, and the typical range of services that pain patients access in the region.

This article is one of only a few studies (see Access Economics , Raftery et al. , and Kronberg et al. ) to assess both the direct and indirect annual costs for patients attending a chronic pain clinic and receiving specialized pain management services. Previous research on the cost of chronic pain has typically focused on the wider population . Only one other study assessed similar costs and focused on costs for patients referred to a specialized pain clinic but on a waiting list at the time of data collection . One other, older study examined the cost of pain in patients attending a pain management service in Ireland . Our study aimed to assess, in detail, the total costs related to chronic pain for pain clinic attendees. In Ireland, tertiary pain management services are relatively few with the result that patients typically wait for around 2 years to access a service. As a result, and by definition, these are the patients with more enduring chronic pain (their pain has not resolved) and in the authors' experience tend to be more intensive users of resources. In a recent article, we showed that the most severely disabled by pain used very high levels of health services including specialist pain management services—the 5% most expensive chronic pain patients accounted for 24% of all costs . The two central questions to be addressed in our study were: 1) What were the economic costs of chronic pain among participants attending the service for a 12-month period?; and 2) What factors explain variations in these costs?

Method

The approach taken in this study follows a standard cost of illness methodology and provides estimates of direct and indirect costs . Direct costs are derived from patients' reports of their use of health care resources, as they specifically relate to their treatment for pain over the previous 12 months including the following items: inpatient visits (overnights), outpatient visits to specialty care, outpatient visits to primary care, emergency room (ER) visits, ambulance costs, family medical practice nurse, home help, social worker, psychologist, psychiatrist, public health nurse, occupational therapy, physiotherapy, chiropractor, acupuncture and homeopathy, prescription medication costs, extra requirements such as adaptations to home or mobility equipment, and the cost of health insurance.

Indirect costs included lost productivity and informal care, and represent the losses resulting from morbidity or disability as a result of pain and items such as lost leisure time, as well as the cost of unpaid work of the patient and the carer. In this study, we were unable to link patient reports of hospital stays with specific investigations or procedures so we have included only the cost of stays per day based on average costs across all conditions. We also calculated indirect costs such as lost productivity, temporary work absences, informal care, and community care.

There are two main methods to estimate health system costs. Top-down disease cost data can be derived from central data collection agencies. Bottom-up costs use surveys and diaries to accumulate information from a single study or multiple smaller studies and then extrapolate to the wider population of interest. Our study adopted a bottom-up approach, focusing on one clinic only and collecting data at the micro level and aggregating to get an overall average level of cost. For example, we asked patients how many outpatient visits they had in the previous 12 months and multiplied average cost by outpatient utilization visits. While estimation of costs based on retrospective recall is not ideal and may be criticized for being vulnerable to recall bias, Patel et al. found that patient recall of the frequency of general practitioner (GP) visits was closely correlated with actual number of visits. We assume that the recording of hospitalizations, procedures, and specialist consultations will be recalled with at least the same accuracy.

Participants and Setting

Study Setting

A sample of 100 patients with chronic non-cancer pain (70 females, 30 males), with a mean age of 51.6 years (standard deviation [SD] 14.1, range 20–83 years) participated in the study. The sample was obtained from consecutive patients attending the pain clinic in a tertiary referral hospital (Galway University Hospital) in the west of Ireland over a 6-week period. The study was approved by the Galway University Hospital Research Ethics Committee, and all participants provided written informed consent. A very high proportion of all patients who were invited to participate took part (75%), and there were no apparent differences in demographic characteristics or basic clinical profile between participants and nonparticipants based on basic demographic profile. The population attending this pain clinic would be considered typical of an Irish chronic pain population and was comparable on key demographic variables with the authors' large prevalence study in Ireland (Prevalence, Impact and Cost of Chronic Non-Cancer Pain in Ireland [PRIME] Study ). For example, the mean age of our sample was 51.6 (SD 14.1), while the PRIME sample mean age was 46.8 (SD 16.2); mean pain duration was 6.5 years vs 7.6 years in the PRIME study. The most common site of pain in both samples was upper or lower back pain Thus, we contend that the findings will be broadly generalizable to a wider chronic pain population.

The service has two half-time pain management specialist physicians, two nurses (one focused on acute pain and one focused on chronic pain), and a part-time psychologist. Patients with all forms of chronic pain may be referred to the pain management clinic by their GP or specialist physician, and will then have an assessment with the pain management physician and pain management nurse after which a treatment plan is developed. Typically, patients are offered some combination of pharmacological treatments and interventional procedures aimed at reducing pain. Patients do not routinely receive care from a multidisciplinary team—while the service has a clinical psychologist on the team, this is on a part-time basis, so only those most in need of psychological intervention are referred to the psychologist. Patients who require physiotherapy are referred to the general hospital service (not a pain-specific physiotherapy service). The service conducts approximately 80–100 individual consultations per month, and within that population, carries out approximately 100 procedures per month, half of these are on an inpatient or day patient basis while the remainder are carried out in an outpatient setting.

Inclusion Criteria

Participants were all adults (18 years or older), no upper age limit applied. Participants were all attending the pain clinic and had chronic non-cancer, nonterminal pain for a minimum of 3 months as per the IASP definition of chronic pain . Participants were required to have basic literacy skills, but a researcher was available at all times to provide assistance when needed.

Descriptive data for the participants are presented in Table 0001. In addition, the mean pain severity, measured from the visual analog scale (1–10) and collected for 42 patients from approximately the midpoint of the study, was 6.78 (SD 1.6). The mean number of years in pain (collected from patient records) was 6.5 years (SD 6.5; range 1–40 years). Over one-third of the sample had either low or upper back pain, almost one-fifth had head pain, one-fifth had neck/shoulder pain, and one-quarter had lower body pain (excluding back). The remainder (approximately 2%) indicated a category of “Other” pain (excluding the categories of head, spine, neck/shoulder, and lower limb pain); thus, pain site is unspecified. None of the pain was associated with current cancer diagnosis. The mean number of pain areas was 2.5 and 34% had more than one site of pain.

Table 1

Sample characteristics

Sex 
Male 30 
Female 70 
Age 
  Age <65 79 
  Age >65 21 
Education status 
  Primary education 26 
  Secondary education 46 
  Third level education 28 
Employment status 
  Full time 14 
  Part time 11 
  Work at home 
  Unable to work due to pain 37 
  Early retired due to pain 13 
  Student 
  Retired 19 
Health insurance status 
  Medical card (free medical care) 77 
  Health insurance 31 
Sex 
Male 30 
Female 70 
Age 
  Age <65 79 
  Age >65 21 
Education status 
  Primary education 26 
  Secondary education 46 
  Third level education 28 
Employment status 
  Full time 14 
  Part time 11 
  Work at home 
  Unable to work due to pain 37 
  Early retired due to pain 13 
  Student 
  Retired 19 
Health insurance status 
  Medical card (free medical care) 77 
  Health insurance 31 

Note in some cases, figures do not add to 100—for example, some patients have both medical card and private health insurance. In two cases, data were missing on education and work status.

Materials

The Client Service Receipt Inventory (CSRI) was used in this study. The CSRI has been used in around 150 studies examining costs associated with mental ill health, neurological problems, physical disability, and autism. The CSRI was subsequently adapted for a study of the cost of pain and has recently been administered in estimating the cost of pain in the wider population . The CSRI has been validated as an accurate measure of frequency of health service use . We adapted the CSRI-Pain questionnaire for study participants attending the chronic pain clinic. The variant used in our study was designed to collect retrospective data on service utilization for the previous 12 months based on patient self-report. Patel et al. have provided evidence for the reliability of retrospective self-report of health service utilization.

Procedure

Ethics

The Hospital Research Ethics Committee approved the study. The CSRI instrument was given to consecutive patients who attended the Galway University Hospital Pain Clinic. Patients completed the questionnaire with the assistance of the researcher if necessary.

Costing

To estimate full costs, data on the intensity of service use were combined with unit costs of the relevant health services. Utilization data were obtained from responses to the CSRI and unit costs of services from a range of sources including the health service, the Central Statistics Office , and expert opinion.

In Ireland, health cost data are not gathered in a comprehensive manner, and presently, there is no complete unit reference cost database. Most cost of illness studies rely on obtaining unit costs from a range of sources. In this study, the unit costs for inpatient, day patient, outpatient, and ER visits were provided by Ireland's Health Service Executive Casemix Unit. Costs for other consultations such as home help, nurse, psychologist, or social worker were based on average salaries from Ireland's Department of Health and Children for public sector employees . Unit costs for private services such as chiropractor, acupuncture, and homeopathy were based on expert opinion.

Direct Costs

Direct costs included inpatient care, outpatient care, primary health care (GP), ER visits, ambulance costs, family medical practice nurse, home help, social worker, psychologist, psychiatrist, public health nurse, occupational therapy, physiotherapy, chiropractor, acupuncture, and homeopathy. The question in the CRSI on outpatient appointments was as follows: “Have you had any other hospital outpatient appointments, medical appointments, or physiotherapy in the last 12 months for any reason related to your pain?” Respondents were helped with examples as follows: Outpatient department/consultant specialty (e.g., rheumatology, orthopedic surgeon, neurology, physiotherapy, pain clinic, occupational therapy, psychology appointment). However, the type of procedure was not recorded, simply the total number of appointments. Hence, procedures were implicitly included, but we do not have details of the precise costs. A further example of a question in the CSRI is “In the last 12 months, have you taken any time off work because of your pain?”

Data on resource use also included extra requirements such as adaptations to home or mobility equipment. The cost of health insurance was also calculated based on costs provided by patients. Costs in health insurance do not vary greatly across patients in Ireland as there are very few providers in the market. The cost is generally one of the most expensive items in a monthly household budget so patients tend to be aware of their expenditure on health insurance, and we treat their reports as a strong indication of the actual cost.

We assumed duration of visits to health practitioners at either 30 minutes (family medical practice nurse, public health nurse, chiropractor, acupuncture, homeopath, physiotherapy) or 60 minutes (home help, social worker, psychologist, psychiatrist, occupational therapy). Duration of visits was based on consensus opinion of expert clinicians in the local health services. The costs for these visits were calculated as follows. The midpoint of the annual public sector salary scale for the profession was divided by 52 weeks, and we assumed a 35-hour working week to obtain the costs of salary for 30- and 60-minute visits. We then multiplied by a factor of 1.25 to include employment-related costs such as social insurance and pension . Note that in Ireland, family medical practice nurses are uncommon, and people are more likely to avail of services from the public health nurse.

Patients were also asked about litigation related to their pain. Our study showed that 18% had a litigation claim in process. As these claims were still in progress, there was no information on final costs; hence, this cost was excluded from our study.

Ideally, prescription medication costs (and over-the-counter medication costs) should be included in a cost of illness study. In the CSRI, patients were asked about the types of medication they took during the previous 12 months, the dosage, and duration of years taking the medication. We calculated total costs per patient by assessing responses to questions on medication use and attributing unit prices from the Monthly Index of Medical Specialties prescribing book . Patients usually bring a list of medications to their consultations, and we were able to ascertain from patients when they had changed, commenced, or stopped medications. Note that state medical cards grant free access to certain health services for those with a very low income or those with severe disability that would prevent the ability to work.

Indirect Costs

Indirect costs included lost productivity and informal care. In the CSRI, respondents were asked to report the total number of work days lost due to chronic pain. We used the human capital approach to estimate production costs, as used in previous cost of illness studies , whereby an individual's level of earnings is assumed to reflect their productivity . Lost production can therefore be calculated by multiplying the time absent from work by the wage rate. We used average industrial hourly labor costs for employees (US$37/h) across all sectors to calculate the unit cost of a work day as US$259 . These include costs such as social insurance and pension . We also calculated the annual cost of being unable to work and the corresponding cost to state benefits. We used the weekly rates of state benefits for the number of respondents who reported being in receipt of such benefits. This may underestimate the amounts people actually received if they were in receipt of other state benefits or if they had dependent children. It should be noted that by including disability benefits, we may be double-counting those who also say they lost work days due to pain, and this could lead to an overestimate of indirect costs.

The cost of informal care is another relevant indirect cost. Many families have at least one person who will lose days from work to care for a family member who has chronic pain . Informal care is distinguished from services provided by people employed in the health and community sectors (formal care) because informal care is generally provided free of charge to the patient and is not regulated by the government. While informal care is provided free of charge, it is not free in an economic sense, as time spent caring is time that cannot be directed to other activities such as paid work, housework, or leisure. The valuation of informal care is always a contentious issue, and there is no agreement on how best to measure the opportunity cost for caregivers of not working. We decided on the human capital approach and viewed hours of informal care as hours that could be spent working in the formal economy at the average industrial labor cost (US$259/day).

Total Costs

To derive total costs, we multiplied the number of units by the unit cost. We then divided by 100 to obtain the average cost per patient. The number of users of each service is lower than 100, and we provide data on cost per patient and cost per user of each specific service.

Data Analysis

Data were coded and analyzed using the Statistical Package for the Social Sciences (SPSS 18.0, Chicago, IL, USA). Data on variables are presented as means and standard errors. Costs are estimated as means. Statistical tests on differences between proportions in health care utilization by age (under and over 65), gender, and payment mechanisms (whether by state-provided medical card or private health insurance) were carried out using chi-squared tests. The threshold for statistical significance was P < 0.05. Tests for normality and homogeneity of variance are carried out using the Kolmogorov–Smirnov (K-S) test in Stata Version 11 (StataCorp, College Station, TX, USA).

To identify the cost drivers for direct costs only, we estimated an ordinary least squares (OLS) regression of logged costs on the number of hospital nights, number of GP visits, number of outpatient visits, number of day visits, and other treatment costs (e.g., nurse, physiotherapy). Cost data are generally highly skewed and with a non-constant variance. In that case, the OLS regression may give biased estimates. Therefore, it is preferable to use the log of costs and proceed with the OLS model. It can be argued that a more sophisticated model such as generalized linear model (GLM) should be estimated, but in practice, the choice of statistical model may not matter too much and results are fairly robust to model choice . Nonetheless, we follow Basu and Mannning who suggested that in the presence of skewed data, a more appropriate alternative model is the GLM. Therefore, as a sensitivity analysis, we also estimated a GLM with a log link function and a gamma distribution. We also controlled for medical card holders, health insurance expenditure, age, and gender. In estimating the cost drivers for total costs, we also included the variables time off work, informal care, and state benefits.

The variables are estimated as coefficients within the regression, so each estimate corresponds to the direction and magnitude (percentage effect) a variable has on overall costs. For example, the higher the coefficient, the higher the effect of costs on the dependent variable.

Results

Health Service Utilization

In Table 0002, we provide summary data regarding intensity of service use. Most of the sample had accessed pain-related outpatient hospital services (92%) and contact with their GP (91%) in the previous 12 months. About 25% had a visit as an inpatient to hospital, and 17% had attended hospital as a day patient. Almost one third had visited the ER.

Table 2

Mean frequency of patient health service utilization for all patients (with standard error)

 Hospital Inpatient (N Nights) Outpatient (N Days) ER (N Visits) GP (N Visits) Day Patient (N Visits) 
All 4.18 (1.16) 8.24 (1.19) 0.91 (0.18)   9.54 (2.48) 0.47 (0.14) 
Male 4.23 (2.30)   6.5 (1.79) 0.83 (0.30)   7.83 (2.02) 0.50 (0.34) 
Female 4.16 (1.31)   8.9 (1.53) 0.94 (0.23)   10.6 (3.1) 0.46 (0.50) 
Age <65 4.44 (1.27) 8.67 (1.46) 0.86 (0.20)     5.5 (1.5) 0.53 (0.17) 
Age ≥65 3.19 (2.79) 6.61 (1.57) 1.09 (0.41)   5.52 (1.51) 0.24 (0.12) 
Medical card 
  Yes 5.09 (1.47) 7.75 (1.35) 1.01 (0.23) 11.79 (3.34) 0.53 (0.18) 
  No 1.13 (0.83) 9.86 (2.61) 0.56 (0.21)  5.47 (1.44) 0.26 (0.16) 
Health insurance 
  Yes 3.96 (2.14) 9.59 (2.40) 0.90 (0.22) 17.34 (10.2) 0.25 (0.13) 
  No 4.31 (1.45) 7.84 (1.37) 1.04 (0.30)   8.28 (1.11) 0.56 (0.19) 
Both medical card and health insurance 
  Yes 5.66 (3.66) 7.53 (2.76) 1.20 (0.48) 24.33 (15.54) 0.33 (0.18) 
  No 3.91 (1.21) 8.36 (1.32) 0.86 (0.19)   7.85 (1.03) 0.49 (0.16) 
 Hospital Inpatient (N Nights) Outpatient (N Days) ER (N Visits) GP (N Visits) Day Patient (N Visits) 
All 4.18 (1.16) 8.24 (1.19) 0.91 (0.18)   9.54 (2.48) 0.47 (0.14) 
Male 4.23 (2.30)   6.5 (1.79) 0.83 (0.30)   7.83 (2.02) 0.50 (0.34) 
Female 4.16 (1.31)   8.9 (1.53) 0.94 (0.23)   10.6 (3.1) 0.46 (0.50) 
Age <65 4.44 (1.27) 8.67 (1.46) 0.86 (0.20)     5.5 (1.5) 0.53 (0.17) 
Age ≥65 3.19 (2.79) 6.61 (1.57) 1.09 (0.41)   5.52 (1.51) 0.24 (0.12) 
Medical card 
  Yes 5.09 (1.47) 7.75 (1.35) 1.01 (0.23) 11.79 (3.34) 0.53 (0.18) 
  No 1.13 (0.83) 9.86 (2.61) 0.56 (0.21)  5.47 (1.44) 0.26 (0.16) 
Health insurance 
  Yes 3.96 (2.14) 9.59 (2.40) 0.90 (0.22) 17.34 (10.2) 0.25 (0.13) 
  No 4.31 (1.45) 7.84 (1.37) 1.04 (0.30)   8.28 (1.11) 0.56 (0.19) 
Both medical card and health insurance 
  Yes 5.66 (3.66) 7.53 (2.76) 1.20 (0.48) 24.33 (15.54) 0.33 (0.18) 
  No 3.91 (1.21) 8.36 (1.32) 0.86 (0.19)   7.85 (1.03) 0.49 (0.16) 

Significance P value = 0.05.

There are no significant differences except for GP visits among medical card (free medical care) & health insurance holders. t-test = 2.57.

ER = emergency room; GP = general practitioner; N = Number.

There were no significant differences between utilization for males and females, or different age groups. However, there was one exception: the number of GP visits was significantly greater among those having both a medical card and private health insurance compared with those who did not have both of these (P = 0.024). Results are reported as the average number of days in hospital and average number of visits as an outpatient, ER patient, GP patient, or day patient at the hospital. The range includes zero number of days—i.e., the average is calculated from all 100 patients even if they did not use that health service.

Costs of Health Service Utilization (Direct Costs)

Total direct costs and average direct cost per patient are provided in Table 0003. Overall, the average cost per patient was US$10,791. The majority of direct costs were due to inpatient and outpatient hospital care at 44% and 18.8% of total direct costs, respectively. Other large direct costs included GP (5.8%) and day patient services (4.2%). Prescription medication costs amounted to 11.9% of total direct costs. The CSRI also asks for extra costs such as transport, modifications to home, special equipment, and any other costs. Expenditure in these areas in the previous 12 months was US$413 per pain clinic attendee (3.8% of total direct costs per attendee). Health insurance was another relevant cost for people in Ireland, and respondents reported a total cost of US$440 per pain clinic attendee (4.1% of total direct costs per attendee).

Table 3

Direct medical costs

 N Visits N Patients Cost Per Visit: US $ Unit Costs Average Cost Per Clinic Attendee (US $) % of Total Direct Cost Per Participant 
Hospital (nights) 418 25 1,141   4,769   44.2 
Outpatient 824 92   246   2,027   18.8 
General practitioner 954 81     66      630     5.8 
Day patient   47 17   975      458     4.2 
Accident and emergency   91 35   388      353     3.3 
Home help 572 11     30      172     1.6 
Physiotherapy 386 32     22        85     0.8 
Ambulance   15 10   122        18     0.2 
Psychologist   29   7     68        20     0.2 
Occupational therapy   29 10     44        13     0.1 
Practice nurse   24   8     26          6     0.1 
Social worker   18   8     44          8     0.1 
Public health nurse   13 13     26          3     0.0 
Chiropractor (private)   22      92        20     0.2 
Acupuncture (private)   67      92        62     0.6 
Homeopath (private)     8      73          6     0.1 
Extra requirements  11       413     3.8 
Health insurance  25       440     4.1 
Prescription costs      1,288   11.9 
Total direct cost    10,791 100 
 N Visits N Patients Cost Per Visit: US $ Unit Costs Average Cost Per Clinic Attendee (US $) % of Total Direct Cost Per Participant 
Hospital (nights) 418 25 1,141   4,769   44.2 
Outpatient 824 92   246   2,027   18.8 
General practitioner 954 81     66      630     5.8 
Day patient   47 17   975      458     4.2 
Accident and emergency   91 35   388      353     3.3 
Home help 572 11     30      172     1.6 
Physiotherapy 386 32     22        85     0.8 
Ambulance   15 10   122        18     0.2 
Psychologist   29   7     68        20     0.2 
Occupational therapy   29 10     44        13     0.1 
Practice nurse   24   8     26          6     0.1 
Social worker   18   8     44          8     0.1 
Public health nurse   13 13     26          3     0.0 
Chiropractor (private)   22      92        20     0.2 
Acupuncture (private)   67      92        62     0.6 
Homeopath (private)     8      73          6     0.1 
Extra requirements  11       413     3.8 
Health insurance  25       440     4.1 
Prescription costs      1,288   11.9 
Total direct cost    10,791 100 

Costs are presented in 2008 US dollars. Cost per participant = costs for all patients including cost per patients with events plus $0 for each patient without an event/service.

Indirect Costs

In Table 0004, we show that there were 2,017 lost work days, costing US$5,224 per participant attending the pain clinic, representing 39% of total indirect costs. There were 398 days of informal care for 19 of the sample in the previous 12 months at a cost of US$1,031 per attendee. This amounts to 8% of total indirect costs. State benefits were about half of all indirect costs. Overall, we estimated the total indirect costs per pain clinic attendee at US$13,252.

Table 4

Indirect costs

 N Days N Patients Cost Per Pain Clinic Attendee (US $) % of Total Indirect Costs Per Participant 
Lost productivity 2,017 19   5,224   39 
Informal care   398 19   1,031     8 
State disability payment  40   6,997   53 
Total indirect costs   13,252 100 
 N Days N Patients Cost Per Pain Clinic Attendee (US $) % of Total Indirect Costs Per Participant 
Lost productivity 2,017 19   5,224   39 
Informal care   398 19   1,031     8 
State disability payment  40   6,997   53 
Total indirect costs   13,252 100 

Lost productivity = N days lost × unit cost per day (US$259).

Costs are presented in 2008 US dollars.

Cost per participant = costs for all patients including cost per patients with events plus $0 for each patient without an event/service.

Variations in Cost

In order to establish the main cost drivers for people with severe chronic pain, OLS regressions were conducted to identify statistically significant factors associated with costs (we also estimated the GLM model and found similar results—estimates are available from the authors on request). First, the log of costs was derived, as generally, costs follow a non-normal distribution. The K-S tests proved that costs are non-normal, and the skewness is reduced once we calculate log of costs (in the K-S tests, P < 0.05 for both direct and total costs, and P = 0.206 and P = 0.l93 for log of direct and total costs, respectively). For direct health costs, we found that hospital inpatient stays, day visits, outpatient visits, and ER costs were the main health service cost drivers (see Table 0005). Home help was also statistically significant. Health insurance had a significant marginal cost, i.e., the amount costs increased by for one extra unit of health insurance. Females were more likely to have higher costs, but as the sample had a higher proportion of females overall, there may be sampling bias.

Table 5

Factors associated with costs: ordinary least squares (OLS) regression of logged direct costs

 Direct Costs 
Hospital nights 0.0514** (0.0045) 
General practitioner visits −0.0027 (0.0022) 
Outpatient visits 0.0242** (0.0059) 
Day visits 0.1665** (0.0299) 
Ambulance 0.1574 (0.0979) 
Emergency room 0.0886** (0.0264) 
Practice nurse −0.0413 (0.0613) 
Home help 0.0079** (0.0029) 
Social worker −0.0030 (0.0673) 
Psychologist 0.0372 (0.0384) 
Psychiatrist 0.0261 (0.0491) 
Occupational therapy −0.0101 (0.0528) 
Physiotherapy 0.0261 (0.0491) 
Alternative treatment 0.0257 (0.0149) 
Other treatment costs (public health nurse, adaptations to home) 0.1049 (0.0911) 
Medical card 0.0233 (0.0948) 
Yes (reference: no)  
Health insurance 0.0003** (0.0001) 
Yes (reference: no)  
Gender 0.2306 (0.0948) 
Female (reference: male)  
Age −0.1037 (0.1033) 
Under 65 (reference: 65 and over)  
Constant 7.404** (0.1366) 
N observations 100 
R2 0.8577 
Unadjusted overall mean cost US$10,791 
 Direct Costs 
Hospital nights 0.0514** (0.0045) 
General practitioner visits −0.0027 (0.0022) 
Outpatient visits 0.0242** (0.0059) 
Day visits 0.1665** (0.0299) 
Ambulance 0.1574 (0.0979) 
Emergency room 0.0886** (0.0264) 
Practice nurse −0.0413 (0.0613) 
Home help 0.0079** (0.0029) 
Social worker −0.0030 (0.0673) 
Psychologist 0.0372 (0.0384) 
Psychiatrist 0.0261 (0.0491) 
Occupational therapy −0.0101 (0.0528) 
Physiotherapy 0.0261 (0.0491) 
Alternative treatment 0.0257 (0.0149) 
Other treatment costs (public health nurse, adaptations to home) 0.1049 (0.0911) 
Medical card 0.0233 (0.0948) 
Yes (reference: no)  
Health insurance 0.0003** (0.0001) 
Yes (reference: no)  
Gender 0.2306 (0.0948) 
Female (reference: male)  
Age −0.1037 (0.1033) 
Under 65 (reference: 65 and over)  
Constant 7.404** (0.1366) 
N observations 100 
R2 0.8577 
Unadjusted overall mean cost US$10,791 
**

P < 0.05; Estimates are presented as coefficients (e.g., magnitude of effect in this log OLS model) and standard errors are in parenthesis.

When we analyzed both direct and indirect costs together, we found that lost days from work and state benefits did significantly impact on costs, while the same direct costs (inpatient stays, outpatient visits, health insurance) remained significant. In this model, we had originally included all direct costs that were included in the first model. However, we found that these were statistically insignificant so we included these as other treatment costs. The focus of model 2 is on indirect costs so this does not affect our interpretation of these effects. These OLS models explain a large variance in costs (r2 = 0.85 and r2 = 0.86 for direct and indirect costs, respectively). In this model, we wished to estimate variation in costs; hence, we included all the explanatory variables.

In terms of percentage effects, we can approximately interpret these results to show that a one unit increase in hospital nights is associated with a 5% increase in direct costs or a one unit increase in outpatient attendance is associated with a 2.4% increase in costs. A 1-day increase in time off work is related to almost 1% increase in total costs—based on this sample of pain clinic participants, if 1 day per person of work absence was avoided, the economy would save US$14,708 per year. Assuming an average of 20 days off work per patient annually and if 4% of the population have severe chronic pain, this amounts to US$529 million cost per year associated with work absence, amounting to almost 1% of Ireland's GDP (Note that because the data are not normally distributed and transformed to logs, if we were interested in predicted costs, it is necessary to back-transform using the Duan smearing estimator . However, as we are reporting the cost drivers here, we do not report these results. Results are available from the authors—the percentage effects are slightly lower, so the coefficients here may be viewed as a slight overestimate). These results are in Tables 0005 and 0006.

Table 6

Factors associated with total (direct and indirect) costs: OLS regression of logged total costs

 Both Direct and Indirect Costs 
Hospital nights 0.0346** (0.0048) 
General practitioner −0.0043 (0.0024) 
Outpatient visits 0.0169** (0.0042) 
Day visits 0.1053** (0.0319) 
Ambulance 0.0251 (0.0931) 
Accident and emergency 0.0334 (0.0284) 
Other treatment costs (includes all services with less utilization, adaptations to home) 0.2668** (0.0961) 
Lost days work 0.0061** (0.0006) 
Social welfare benefits for disability 1.2463** (0.1044) 
Days off work by informal carers 0.1431 (0.1020) 
Medical card (free medical care) −0.1305 (0.1201) 
Yes (reference: no)  
Health insurance 0.0001** (0.0001) 
Yes (reference: no)  
Gender 0.0723 (0.1006) 
Female (reference: male)  
Age −0.1910 (0.1268) 
Under 65 (reference: 65 and over)  
Constant 7.998** (0.1896) 
N observations 100 
R2 0.8646 
Unadjusted overall mean cost US$24,043 
 Both Direct and Indirect Costs 
Hospital nights 0.0346** (0.0048) 
General practitioner −0.0043 (0.0024) 
Outpatient visits 0.0169** (0.0042) 
Day visits 0.1053** (0.0319) 
Ambulance 0.0251 (0.0931) 
Accident and emergency 0.0334 (0.0284) 
Other treatment costs (includes all services with less utilization, adaptations to home) 0.2668** (0.0961) 
Lost days work 0.0061** (0.0006) 
Social welfare benefits for disability 1.2463** (0.1044) 
Days off work by informal carers 0.1431 (0.1020) 
Medical card (free medical care) −0.1305 (0.1201) 
Yes (reference: no)  
Health insurance 0.0001** (0.0001) 
Yes (reference: no)  
Gender 0.0723 (0.1006) 
Female (reference: male)  
Age −0.1910 (0.1268) 
Under 65 (reference: 65 and over)  
Constant 7.998** (0.1896) 
N observations 100 
R2 0.8646 
Unadjusted overall mean cost US$24,043 
**

P < 0.05; Estimates are presented as coefficients (e.g., magnitude of effect in this log OLS model) and standard errors are in parenthesis.

OLS = ordinary least squares.

Discussion

Our results indicate a significant cost for all 100 chronic pain clinic participants in the previous 12 months. The hospital costs were the main driver for health care utilization costs at US$4,769 per patient per year, 44% of the total direct costs per pain clinic attendee. To put our estimates into the context of overall health costs in Ireland, we applied prevalence estimates of severe chronic pain to the cost data to obtain an overall cost of severe chronic pain in the population. The Breivik et al. study suggests chronic pain affects 13% of the adult population in Ireland, but this is probably a conservative estimate as it does not include children, it uses a 6-month pain duration criterion rather than the 3-month criteria used in the IASP definition , and it required at least a moderate level of pain severity. More recently, the overall prevalence of chronic pain in Ireland was estimated at 36% across all levels of pain severity . However, pain clinic populations in Ireland tend to be people with more problematic levels of pain. Breivik et al. found that 31% of those in Ireland with pain rated themselves in the range 8–10 on a 0–10 pain intensity scale, 12% were taking opiates (which would indicate severe pain), and 13% had seen a pain specialist. Using these figures, we conservatively estimate that around 4% of the Irish population has a more severe level of pain. By extrapolating our calculated total cost per pain clinic attendee of US$24,043 to the number of people with severe chronic pain (4%), we estimate the cost at approximately US$3.7 billion. This amounts to about 1.2% of Ireland's GDP. This is lower than that estimated by Raftery et al. , but their study identified a much higher prevalence of chronic pain than Breivik et al. . In addition, the cost estimates in Raftery et al. included people with all levels of pain severity, of whom the majority had milder pain severity, resulting in a lower average cost of US$8,237 per patient. In their study, Raftery et al. found that the cost per patient with the most severe pain (Grade 4 pain) was US$15,376, lower than our estimate. However, we have also included an estimate for informal care costs, and in general, the level of costs is higher for our sample of chronic pain clinic attendees with more severe chronic pain.

While our study was targeted at those attending a specialist pain management service rather than the general population of chronic pain patients, it is interesting (bearing in mind the limitations of such comparisons) to compare our sample with the sample in the Pain in Europe survey . In that survey, 43% were working full time compared with 14% in our sample. In Australia, Kerr et al. found that less than 20% of chronic pain clinic attendees were employed. Our rates are more comparable with the latter study, as both studies focus on pain clinic attendees who are more likely to be at the more severe end of the pain spectrum with a concomitant higher probability of not being able to work. Breivik et al. found that in Ireland, 83% of patients went to their GP, similar to the proportion found in our study. An Australian study showed that 94% visited their GP, 22% went to ER, and 20% had an inpatient stay. In comparison with the general profile outlined by Breivik et al. , Kerr et al. showed that the pain clinic attendees are more intensive users of health care services overall.

In this study, we used a bottom-up approach to estimate costs. Bottom-up methods have the advantage of providing greater detail in relation to specific cost elements, and the same study can be extended to capture further information. However, it is likely to be less generalizable than a top-down study , and this should be borne in mind when considering the relevance to other community settings.

The average number of hospital inpatient nights over the previous 12 months was 4.18. From our data, we also know that the number of admissions was 1.1 per patient. In comparison, Blyth et al. reported a mean figure of 0.46 days for people with severe chronic pain in Australia, but their information is calculated as number of hospital admissions. Blyth et al. reported 10.7 GP visits, similar to our figure of 10.6 visits. Their ER attendance rate was 0.85 visits, similar to our finding of 0.93 visits.

We found that the mean number of lost work days was 20 days in the previous year—this is consistent with the findings of Breivik et al. , whose pain patient sample had lost an average of 9.5 days in the previous 6 months. This estimate indicates the need to get people who suffer from chronic pain back to work as soon as possible. Indeed, there has been an emphasis recently on the impact of “presenteeism” or reduced productivity while at work. Our data did not provide good estimates of self-reported reduced hours of productivity, but it has been estimated that in Australia, at least a further 8 days of work per patient are lost in the year due to “presenteeism” .

It is useful to compare our cost estimates with those obtained in other studies, particularly in an Irish setting. An earlier study by Sheehan et al. used a sample of 95 patients from a pain clinic in Dublin to estimate the cost of chronic benign pain. Focusing on loss of earnings, social welfare payments, and the cost of health care, Sheehan et al. concluded that the costs were significant. For example, they found that up to the time of referral to the pain management program, 95 patients had cost US$6.8 million (in 2008 prices) in health services, social welfare payments, and loss of earnings. This important study identified the enormous indirect costs of chronic pain in Ireland (lost earnings were about 50% of total costs). It is difficult to make direct comparisons with our study as Sheehan et al. focused on all costs up to time of referral, whereas we focused on costs over 12 months in total.

The reduced quality of life experienced by people as a result of chronic pain is a very important consideration; however, it is an intangible cost and is very difficult to quantify in economic terms. Our study did not ask questions about quality of life, and therefore, it is beyond the scope of the study to analyze associated costs.

Kronberg et al. used a similar approach to our analysis to assess the impact of factors on costs for patients that are on a waiting list for a pain clinic. Our study focused specifically on patients who are already attending the clinic. In Kronberg's analysis, they did not include the impact of health care utilization on cost; hence, it is difficult to compare results. They did, however, find that age and duration of pain were significant variables both leading to an increase in overall costs. In comparison with our study, Kronberg et al. explained 8% of variation in their model, clearly indicating the need to control for a range of other factors in their analysis. We have attempted to control for as many observed variables as possible in our analysis. There may be further unobserved characteristics between individuals that could help explain more variation in the cost. A follow-up prospective study on resource use for chronic pain would help identify such unobserved effects and should be considered for future studies on chronic pain. Here, we have made some comparisons with other studies in Ireland and internationally. These comparisons should be treated with due caution, however, as direct comparison would require matching of samples and methodologies, which was not done.

There are a number of limitations to our study. First, unlike other health care systems (for example in the United Kingdom), there is no comprehensive database with reference costs in Ireland. The costs provided in Table 0003 are taken from several sources, including expert opinion and provide the best available unit costs. Second, the sample size is quite small to provide accurate estimates in the explanatory model. A sample size of at least a few hundred would be preferable for estimation with precision in economic models, and this should be borne in mind for future research on costs of chronic pain. This would provide even greater detail on the cost drivers in total costs for chronic pain clinic participants. We did not consider it appropriate to ask nonparticipants why they did not wish to participate, but we surmise that a combination of anxiety regarding their consultation, poor literacy, and lack of interest in research participation were responsible. Furthermore, we did not ask participants in this study about waiting time to access services and costs associated with this delay in accessing services. Future studies should consider this important point and examine the impact on costs of delayed access (on the premise that earlier access to pain management services may ultimately reduce disability and consequential costs). While we know that 18% of our sample had legal claims in process, we do not know the monetary value of the outcomes; hence, we do not report costs of claims in this study. Finally, our data have been collected via retrospective self-reports over the previous 12 months and hence may suffer from recall bias. Patel et al. found that while frequency of GP visits was recollected accurately, the duration of visits was underestimated. We acknowledge that the retrospective recall of health service usage in general is not ideal and that in particular, recall of medication usage—especially dosage adjustments—was most likely to be vulnerable to inaccurate recall. For more accurate recording of data, objective methods such as routine data are more preferable. Alternatively, data could be collected prospectively using a diary system to record health service utilization.

Conclusion

The cost of chronic pain among intensive service users is significant, and our study shows a cost per patient of US$24,043 over a 12-month period. The hospital costs were the main driver for health care utilization costs, at US$4,769 per patient per year, 44% of the total direct costs per pain clinic attendee. In terms of overall resource allocation, the impact is significant, with total population costs estimated at up to 1.2% of Ireland's GDP. Direct costs were responsible for the greater part of the overall cost in this population. In particular, hospital stays were extremely expensive. In that context, alternative treatment models, which are not reliant on inpatient treatment, should be considered. Indeed, it is not unreasonable to suggest that intensive, multidisciplinary pain management could actually lead to a reduced demand for inpatient procedures. Indirect societal costs were also substantial, including lost productivity due to days off work and informal carers' days off work. Policy makers should consider schemes that would provide people with pain-related disability with opportunities to participate in the workforce, even on a part-time basis. The concern of patients is often that if they are considered able to work, even for a short duration, their disability benefits may be withdrawn. This model serves only to promote “disabled behavior,” and models that actively promote (rather than simply permit) rehabilitative programs incorporating part-time work would be beneficial to all parties. Indeed, the offer of additional financial reward (rather than loss) for work participation might be worthy of consideration.

Ireland does not, as yet, have a National Pain Strategy, such as that found in countries where pain management services are well developed (e.g., Australia ). Considering the evidence that chronic pain is highly prevalent in Ireland has a major impact on quality of life and is very expensive, as shown in this study and others , the need for such a national response is evident and has already been called for .

These data may be important for health service resource allocation. Future studies should evaluate whether early interventions and programs focused on self-management are beneficial from a cost perspective so that patients who have severe chronic pain become less reliant on long-term use of health care services.

In conclusion, the societal economic cost of chronic pain is significant, particularly for those who are intensive health service users. This study highlights the costs involved with each day of work lost per chronic pain sufferer and indicates the need to target vocational rehabilitation at this group. Furthermore, cost effectiveness analyses of these interventions are recommended and necessary in future research on chronic pain management and intervention.

Acknowledgments

The study was conducted while Brenda Gannon worked at National University of Ireland, Galway. Funding from Health Research Board is gratefully acknowledged.

References

International Association for the Study of Pain
.
Classification of chronic pain
.
Pain
 
1986
;(
suppl 3
):
S1
226
.
Elliot
AM
Smith
BH
Hannaford
PC
Smith
WC
Chambers
WA
.
The course of chronic pain in the community: Results of a 4-year follow up study
.
Pain
 
2002
;
99
:
299
307
.
Andersson
HI
Ejlertsson
G
Leden
L
Schersten
B
.
Impact of chronic pain on health care seeking, self care, and medication. Results from a population based Swedish study
.
J Epidemiol Community Health
 
1999
;
53
(
8
):
503
509
.
Chrubasik
S
Junck
H
Zappe
H
Statzke
O
.
A survey of pain complaints and health care use in a German population sample
.
Eur J Anesthesiol
 
1998
;
15
(
4
):
397
408
.
Eriksen
J
Sjogren
E
Ekholm
O
Rasmussen
N
.
Health care utilization among individuals reporting long term pain: An epidemiological study based on Danish National Health Surveys
.
Eur J Pain
 
2004
;
8
:
517
523
.
Access Economics
.
The High Price of Pain: The Economic Impact of Persistent Pain in Australia
 . Access Economics Pty Limited;
2007
.
Maniadiakis
N
Gray
A
.
The economic burden of back pain in the UK
.
Pain
 
2000
;
84
:
95
103
.
Wenig
CM
Schmidt
CO
Kohlman
T
Schweikert
B
.
Costs of back pain in Germany
.
Pain
 
2009
;
13
:
280
296
.
Van Tulder
MW
Koes
BW
Bouter
LM
.
A cost of illness study of back pain in the Netherlands
.
Pain
 
1995
;
62
:
233
240
.
Raftery
M
Ryan
P
Norman
K
et al
The economic cost of chronic non cancer pain in Ireland: Results from the PRIME study, Part 2
.
J Pain
 
2011
;
13
:
139
145
.
Kronberg
C
Handberg
G
Axelsen
F
.
Healthcare costs, work productivity and activity impairment in non-malignant chronic pain patients
.
Eur J Health Econ
 
2009
;
10
:
5
13
.
Blyth
F
March
L
Brnabic
A
Cousins
M
.
Chronic pain and frequent use of health care
.
Pain
 
2004
;
111
:
51
58
.
Sheehan
J
McKay
J
Ryan
M
Walsh
N
O'Keeffe
D
.
What cost chronic pain?
Ir Med J
 
1996
;
89
:
218
219
.
Byford
S
Torgerson
DJ
Raftery
J
.
Cost of illness studies
.
BMJ
 
2000
;
320
:
1335
.
Rice
DP
.
Estimating the cost of illness
.
Am J Public Health
 
1967
;
57
(
3
):
424
440
.
Patel
A
Rendu
A
Moran
P
et al
A comparison of two methods of collecting economic data in primary care
.
Fam Pract
 
2005
;
22
:
323
327
.
Raftery
M
Sarma
K
Murphy
A
et al
Chronic pain in the Republic of Ireland—Community, prevalence, psychosocial profile and predictors of pain disability: Results from the PRIME study, Part 1
.
Pain
 
2012
;
152
:
1096
1113
.
Beecham
J
Knapp
M
.
Costing psychiatric interventions
. In:
Thornicroft
G
Brewin
C
Wing
J
, eds.
Measuring Mental Health Needs
 .
London
:
Gaskell
;
1992
:
163
183
.
Beecham
J
Knapp
M
.
Costing psychiatric interventions
. In:
Thornicroft
G
Brewin
C
Wing
J
, eds.
Measuring Mental Health Needs
 ,
2nd edition
.
London
:
Gaskell
;
2001
:
200
224
.
Sleed
M
Eccleston
C
Beecham
J
Knapp
M
Jordan
A
.
The economic impact of chronic pain in adolescence: Methodological considerations and a preliminary costs-of-illness study
.
Pain
 
2005
;
119
:
183
190
.
Central Statistics Office
.
Earnings and Labour Costs, 2009
 .
Dublin
,
Central Statistics Office
;
2009
.
Department of Health
.
Consolidated Salary Scales Effective from March 2008
 .
Dublin
,
Department of Health
;
2008
.
Curtis
L
.
Unit Costs of Health and Social Care
 . Personal Social Services Research Unit;
2010
.
MIMS
Ireland
.
Monthly Index of Medical Specialties Ireland
.
Irish Medical Times
 .
Dublin
:
Author
,
2008
.
McCrone
P
Darbishire
L
Ridsdale
S
Seed
P
.
The economic cost of chronic fatigue and chronic fatigue syndrome in UK primary care
.
Psychol Med
 
2011
;
33
:
253
261
.
Weisbrod
BA
.
The valuation of human capital
.
J Polit Econ
 
1961
;
69
(
5
):
425
436
.
Department of Social and Family Affairs
.
Rates of social welfare payment
 .
2008
. Available at: http://www.welfare.ie/en/downloads/sw19_08.pdf (accessed May 19, 2013).
Dunn
G
.
Describing, explaining or predicting mental health care costs: A guide to regression models. Methodological review
.
Br J Psychiatry
 
2003
;
183
:
398
404
.
Basu
A
Manning
WG
.
Issues for the next generation of healthcare costs analyses
.
Med Care
 
2009
;
47
(
7 suppl 1
):
S109
114
.
Duan
N
.
Smearing estimate: A nonparametric retransformation method
.
J Am Stat Assoc
 
1983
;
78
(
383
):
605
610
.
Kerr
S
Fairbrother
G
Crawford
M
et al
Patient characteristics and quality of life among a sample of Australian chronic clinic pain attendees
.
Intern Med J
 
2004
;
34
:
403
409
.
Olsson
T
.
Comparing top-down and bottom-up costing approaches for economic evaluation within social welfare
.
Eur J Health Econ
 
2011
;
12
:
445
453
.
National Pain Summit
.
National Pain Strategy: Pain Management for All Australians
 .
Melbourne
:
Faculty of Pain Medicine
;
2010
.
Fullen
B
Hurley
DA
Power
C
Canavan
D
O'Keeffe
D
.
The need for a national strategy for chronic pain management in Ireland
.
Ir J Med Sci
 
2006
;
175
(
2
):
68
73
.