A study of primary care physicians rating their immigrant patients' pain intensity


  • M. André,

    Corresponding author
    1. Department of Medicine and Health Sciences, Family Medicine, Linköping University, Sweden
    2. Department of Public Health and Caring Sciences – Family Medicine and Preventive Medicine, Uppsala University, Sweden
    3. Linköping University, Family Medicine, Linköping, Sweden
    • Clinical Research Center Dalarna, Uppsala University, Sweden
    Search for more papers by this author
  • M. Löfvander

    1. Clinical Research Center Dalarna, Uppsala University, Sweden
    2. Department of Public Health and Caring Sciences – Family Medicine and Preventive Medicine, Uppsala University, Sweden
    3. Department of Neurobiology, Caring Sciences and Society, Karolinska Institutet, Huddinge, Sweden
    Search for more papers by this author

  • Funding sources

    Center for Clinical Research Dalarna – Uppsala University, Sweden. Center for Clincial Research Västmanland – Uppsala University, Sweden.

  • Conflicts of interest

    No potential conflict of interest exists for any of the authors.


Malin André

E-mail: malin.e.andre@gmail.com



Few studies focus on how physicians evaluate pain in foreign-born patients with varying cultural backgrounds. This study aimed to compare pain ratings [visual analogue scale (VAS) 0–100] done by Swedish primary care physicians and their patients, and to analyse which factors predicted physicians' higher ratings of pain in patients aged 18–45 years with long-standing disabling back pain.


The two physicians jointly carried out the somatic and psychiatric diagnostic evaluations and alternated as consulting doctor or observer. One-third of the consultations were interpreted. Towards the end of the consultations, the patients rated their pain intensity ‘right now’ (patients' VAS). After the patient had left, the two physicians independently rated how much pain they thought the patient had, without looking at the patient's VAS score. The mean of the two doctors' VAS values (physicians' VAS) for each patient was used in the logistic regression calculations of odds ratios (OR) in main effect models for physicians' VAS above median (md) with patient's sex, education, origin, depression, psychosocial stress and pain sites as explanatory variables.


Physicians' VAS values were significantly lower (md 15) than patients' VAS (md 66; women md 73, men md 52). The ratings showed no significant association with whether the physician was acting as consultant or observer. The higher physician VAS was only predicted by findings of multiple pain sites.


Physicians appear to overlook psychological and emotional aspects when rating the pain of patients from other cultural backgrounds. This finding highlights a potential problem in multicultural care settings.

What's already known about this topic?

  • Primary care physicians tend to underestimate the patients' ratings of pain especially in those with long-standing pain
  • In immigrant patients with long-standing pain, VAS ratings are predicted by depression and high pain anxiety.

What does this study add?

  • Primary care physicians' VAS ratings of immigrant patients' pain were influenced by the findings of pain sites only and not by psychosocial factors.

1. Introduction

Caring for patients with long-standing non-malignant pain can be problematic, particularly if the physician and patient have different views about pain – its meaning, its origin, its course, consequences and treatment (Helman, 1990). Physicians who must interact with patients from many different socio-cultural backgrounds may face a confusing array of attitudes towards pain. Misunderstanding can lead to overt conflicts, or remain covert during the consultation. Language barriers and differences between patients' illness experiences and physicians' evaluation of illness accounts and clinical findings may exacerbate misunderstandings between patient and physician. Such complexity is not uncommon when cultural gaps are wide for example between Scandinavian physicians and non-European immigrant patients (Andersson, 1994; Bäärnhielm and Ekblad, 2000). To determine whether a patient is in pain, and how much, health care personnel depend on the person signalling pain, either verbally or non-verbally. Interpreting the narrative and the behaviour is often complicated in cases of long-standing pain. This kind of pain has dimensions other than physiological signs alone, such as individual cognitive and emotional facets that should be interpreted according to socio-cultural contexts (Helman, 1990). Thus, patients' reports of pain intensity are also coloured by emotions and evaluations of the impact of pain on life (Löfvander and Furhoff, 1996; Björk et al., 2008; Borsbo et al., 2008; Cobo et al., 2010)

A reliable and validated tool to assess a patient's experienced pain intensity is the visual analogue scale (VAS; Carlsson, 1983; Taddio et al., 2009; Cobo et al., 2010) often likened to a ‘pain thermometer’ (Choiniere and Amsel, 1996). This scale is useful in paediatric, transcultural, palliative and dementia care contexts (Löfvander et al. 1997; McGrath et al., 1996; Ramer et al., 1999; Scherder and van Manen, 2005; Taddio et al., 2009), and is used in studies worldwide, including Turkey, which was the country of origin for many of the patients in the present study (Tarsuslu et al., 2010). When the patients' VAS scores exceed 30 mm, it is said to denote at least moderate pain (Collins et al., 1997). Notably, there are no corresponding values regarding physicians' evaluations of their patients' pain or which clinical findings that predict their ratings.

Also, caregivers can use VAS to visualize their understanding of a patient's pain (Scherder and van Manen, 2005; Taddio et al., 2009). Reports have shown that general practitioners underestimate the pain of patients with acute pain and underestimate even more so when pain was long lasting (Sutherland et al., 1988; Mäntyselka et al., 2001). Tentatively, one might expect the physician's first perception of a patient's pain to influence his or her evaluation of the patient's health problems, medication and future care. To compare, factors not clearly related to physical function predicted the physicians' ratings of patients' incapacity to work, i.e., sickness certification (Löfvander and Engström, 2003). A previous study in the same setting showed that immigrant patients' ratings of severe pain were predicted by depression and high pain anxiety (Löfvander and Taloyan, 2008). These two studies found no significant differences between patients with different countries of origin.

The present study was initiated to explore how primary health care physicians in multicultural settings evaluate their patients' long-standing pain. The two aims of this particular study were to compare the VAS ratings done by physicians and the patients, to analyse which factors predicted physicians' higher ratings of pain in foreign-born patients, 18–45 years of age, with long-standing disabling backache.

2. Methods

2.1 Design

This descriptive study was a part of a clinical research and treatment programme in ordinary primary health care with an overall aim to reduce prevalence of long-term sick leave among adult (18–45 years) immigrant patients with non-malignant disorders, with special focus on pain. The prime goal of the cognitive-behavioural treatment was to reduce pain-related anxiety. The programme started with an initial consultation based on a method where two primary health care physicians acted alternately as consulting and observing physician. These triangulation assessments were performed to overcome weaknesses and problems that arise from single-observer assessments (Denzin, 2006). In brief, the initial consultation included somatic and psychiatric diagnostics, psychosocial and medical–anthropological interviews about patients' views on pain. The study was done at a primary care centre in a district in Stockholm, Sweden, where 85% of the residents are foreign born, mostly from Turkey, the Middle East and Southern Europe.

2.2 Subjects

The study group was recruited consecutively from all patients, 18–45 years of age, who visited the primary health care centre on 6 or more weeks of sick leave for any non-malignant disorder between the years 1998 and 2008. Persons with potentially fatal disorders, e.g., heart disease or severe mental illness, were excluded. There was no formal assessment of language proficiency. Professional interpreters in various languages are available at the request of the patients and they are often used at the health centre. This service is free of charge for the patient. Patients who met the inclusion criteria were referred to the programme in consecutive order by physicians at the health centre.

Sixty-nine of the 260 eligible patients were referred further to a special programme for patients on social security. Another 28 (11%) did not come to the appointment. Fourteen patients were excluded from the study because the observing physician was not one of the two participating physicians.

In total, 149 foreign-born persons (first-generation immigrants) comprised the study group. The majority, 78%, were born and brought up in non-European countries, mostly Turkey and the Middle East (see Table 1), countries with limited mandatory schooling. Some of the women in the study group had had no opportunity to complete even a primary education.

Table 1. Characteristics of the study population n = 149. Median scores (md) with interquartile ranges (IQR) and distribution of age and explanatory variables by sex
  1. **p < 0.01 comparing men and women.
  2. aNorthern Europe, n = 21; Southern Europe, n = 11.
  3. bTurkey, n = 49; Middle East, n = 41; South America, n = 3; North Africa, n = 10; East Africa, n = 14.
  4. cBy criteria from DSM-III-R, axes I and IV.
n (%)56 (37.6)93 (62.4)149 (100)
Age40.0 (35–44)38.0 (34–43)39.0 (35–43)
Explanatory variables   
Education (years) md (IQR)9 (8–12)8 (5–11)9 (5–12)
≤8 lower (%)30.453.8** 45.0
≥9 higher (%)69.646.255.0
Europeana (%)23.220.421.5
Non-Europeanb (%)76.879.678.5
Clinical data   
Depressionc (%)39.352.747.7
No depression (%)60.747.352.3
Severity (1–6) of psychosocial stressors, md (IQR)c3 (2–3)3 (2–4)3 (2–3)
4–6 More severe (%)21.428.025.5
1–3 Less severe (%)78.672.074.5
Number of pain sites, md (IQR)2 (1–3)3 (2–5)3 (2–5)
≥4 More (%)21.447.3** 37.6
≤3 Fewer (%)78.652.762.4

The overwhelming majority of the study group had been on sick leave because of muscular or unspecified pain located in the back, neck or shoulders, and a handful had pain also in other parts of the body, e.g., headache and/or pain in the extremities.

Participation in the study was voluntary.

The study was approved by the Ethical Review Board, Karolinska Institutet, Stockholm D-nr 01–332.

2.3 Methods

The physicians, one man and one woman, had 15 and 20 years, respectively, of experience of multicultural primary health care (Löfvander and Engström, 2003). During the initial consultation, they took turns acting as consulting or observing physician. All patients were new to them and they read no medical records in advance. A guiding principle in primary care is that the physician should acknowledge each patient as an individual in order to understand the patient's problems (McWhinney, 1997). Thus, the consulting physician had no previous knowledge of the patient and could focus more attentively on the patient's own version of his/her illness. The consultant physician did not use written material as support but had memorized the procedure manual of the clinical research and treatment programme. It included the social interview, a questionnaire regarding the patient's explanatory models of illness, the criteria for depression and the assessment scale of psychosocial stressors from DSM, the physical examination schedule and assessments of disability and pain (Löfvander, 1999). The records were done on the procedure manual by the other physician who acted as observer and secretary (observing physician). He or she could also interact with follow-up questions.

2.4 Diagnostics

The physical examination concerned mainly the muscular–skeletal system but also included heart and lung function and laboratory tests. The locations (e.g., spine, shoulder or hip) of muscular insertion lesions were counted and labelled as pain sites. To be counted as a pain site, the pain was to be recurrent and found in isometric muscle contraction. The median number of pain sites in the entire population was used as the cut-off between few pain sites (0–3 locations) and more sites (≥4 locations).

Criteria from the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R) axis I were used to diagnose depression (mild, moderate or severe; APA, 1987). Depression was categorized as not present or present (mild, moderate or severe). Criteria from DSM-III-R axis IV were used to assess the severity of psychosocial stressors (range mild = 1 to catastrophic = 6; APA, 1987). Severity of psychosocial stressors was categorized as mild (level 1–3, e.g., anxiety about pain, financial strain or marital discord), or more severe (level 4–6 including also illness in the family, life threatening events, etc).

2.4.1 Pain measurement

Physicians and patients measured the pain intensity using a VAS. This was a 100-mm plastic ruler with a blue line and a red movable marker. The endpoints were ‘No pain’ at the left and ‘Worst possible pain’ at the right. After the somatic examination and at the very end of the consultation, the patients rated their pain ‘right now’ on the VAS (patients' VAS) and placed the plastic ruler face down on the desk beside the physician. After the patient had left, each of the physicians used the same method to evaluate how much pain he or she thought the patient had. The VAS scores of the consultant and the observer were noted separately. The mean value of consulting and observing physicians' VAS scores (physician VAS) was calculated in order to reflect also non-significant differences (see Table 2) between the physicians' VAS, their roles in the consultation and possible gender aspects. Lastly, the patients' VAS was read from the ruler and noted.

Table 2. Median scores (md) with interquartile ranges (IQR) of patients' VAS and the two physicians' VASa and of the differences between patients' and physicians' VAS by sex. Test of difference by Mann–Whitney U-test
  1. **p < 0.01 between men and women, Mann–Whitney U-test.
  2. ***p < 0.001 between physicians' VAS and patients' VAS.
  3. aMean of the two physicians' VAS scores.
n (%)14956 (37.6)93 (62.4)
Physicians'a VAS15 (10–21)15 (8–19)15 (10–21)
Patients' VAS66 (45–85)52 (39–72)73 (55–90)**
Difference patients versus physicians50 (30–67)*** 41 (27–56)55 (33–72)
Consulting physician, both16 (8–23)13 (5–23)17 (8–23)
Male Dr (n = 65)19 (9–24)18 (9–24)20 (10–25)
Female Dr (n = 82)12 (7–22)11 (4–23)16 (8–23)
Observing physician, both15 (9–21)16 (7–20)10 (5–21)
Male Dr (n = 82)18 (12–22)15 (9–20)19 (12–22)
Female Dr (n = 65)11 (5–20)15 (7–20)10 (5–21)

2.5 Statistics

Categorical data are presented as frequencies. Median scores (md) with interquartile range (IQR) were calculated from the ordinal data including the VAS scores. We used Pearson's chi-square test, or the Mann–Whitney U-test, to test for significant differences between groups regarding distribution and median scores. To compare the VAS ratings done by both the physicians in different roles, we calculated the median scores and IQR of the physicians' VAS. Also, the differences between the patients' VAS and the physicians' VAS by patients' sex were calculated. To analyse which factors contributed significantly to the physicians' higher VAS ratings, the outcome variable in the logistic regression was set as the physicians' VAS ≥16 (i.e., above median physician VAS). As explanatory variables in the logistic regression calculations, we used patient's sex, educational level (above or below median), region of birth (non-Europeans from the Middle East, Asia or Northern and Eastern Africa vs. Europeans), depression (yes or no), severity of psychosocial stressors: mild (levels 1–3) or more severe (levels 4–6, see above) and number of pain sites (few ≤3 or more ≥4). The results are shown as odds ratios (OR) with 95% confidence intervals (95% CIs) for the explanatory variables. The significance level was set at p < 0.05. All significance tests were two sided. The fit of the logistic model was assessed using the Hosmer–Lemeshow goodness-of-fit test and considered acceptable at p < 0.05.

We used the statistical software SPSS, version 17 (SPSS Inc., Chicago, IL, USA).

3. Results

The individuals in the study group were all foreign born and had been living in Sweden for an average of 14 years. Only a few patients had insufficient knowledge of Swedish. Nonetheless, one-third requested an interpreter because they wanted to explain their pain in precise words when meeting a new physician. Most patients worked as cleaners, in health care or in child care. Table 1 shows that the study group consisted of 37.6% men and 62.4% women with a median age of 39 years and a median education of nine years. Women as a group had shorter education than men. The only diagnostic difference between men and women were in the findings of ≥4 pain sites (47.3% in women vs. 21.4% in men; p < 0.01). Notably, no patient had limited mobility.

Table 2 shows that the physicians' VAS scores were similar for men and women. Patients' VAS showed that women rated their pain as significantly more intense than men (md 73 vs. md 52). Notably, some men and women also indicated no or very little pain. Physicians' VAS scores were significantly lower than the patients' VAS (md 15 vs. 66 p < 0.001). The median physicians–patient difference in VAS scores was significantly more prominent if the patient was a woman (md VAS difference 55) than if the patient was a man (md VAS difference 41; p < 0.01). No significant differences were found between the two physicians' ratings in their different roles as consulting or observing physician nor were the ratings influenced by whether the patient and the physician had the same gender. The male physician tended to give higher VAS ratings than the female physician, especially as consultant when rating women, but the difference was not significant. This resulted in an overall lower correlation between the physicians with the male physician as consultant and the female physician as observer (rho 0.39), than vice versa (rho 0.58).

Table 3 shows the distribution of the explanatory variables by the dichotomized physicians' VAS. Only findings of ≥4 pain sites were more common in the category with a higher physician VAS score, i.e., 66.1% versus 33.9% (p < 0.01). Table 4 shows that only patients with ≥4 pain sites had a significant odds ratio (OR 2.7; 95% CI 1.3–5.6) of having a physician VAS score that was higher than the median value. Notably, women had lower but not significant odds (OR 0.7; 95% CI 0.4–1.6) of having a physician VAS rating in the upper category. The overall inter-rater correlation coefficient (Spearman's rho) between the consulting and observer physician was 0.41 (p < 0.001).

Table 3. Distribution of explanatory variables by the categorized variable physicians' VAS below and above the median score (≤15 and ≥16)
PatientsPhysicians' VAS p-score
  1. aBy DSM-III-R axis I.
  2. bby DSM-III-R axis IV.
  3. cConsistent findings of pain near joints.
Education (years)   
Clinical data   
No depression51.348.7 
Psychosocial stressorsb   
≥4 More severe57.947.1 
≤3 Less severe46.853.2 
Pain sites (n)c   
≥4c More33.966.1<0.01
≤3 Fewer59.140.9 
Table 4. Odds ratios (ORs) and 95% confidence intervals (CIs) for physicians rating patients' pain intensity above median (VAS 16 or more). Logistic regression with stepwise inclusion of explanatory variables
VariableCrude OR (95% CI)Main effect model
Women1.0 (0.5–2.0)0.7 (0.4–1.6)
Education years  
≥9 Higher11
≤8 Lower1.4 (0.8–2.7)1.2 (0.6–2.5)
Non-European1.9 (0.9–4.3)1.4 (0.6–3.3)
Clinical data  
No depression11
Depression1.2 (0.6–2.2)1.2 (0.6–2.5)
Psychosocial stressors  
≤3 Less severe11
≥4 More severe0.6 (0.3–1.4)0.7 (0.3–1.7)
Pain sites (n)  
≥3 Fewer11
≥4 More2.8 (1.4–5.6)2.7 (1.3–5.6)

4. Discussion

To sum up, physicians' VAS values were significantly lower (md 15) than patients' VAS (md 66). The difference was most prominent when the patient was a woman. Only a finding of four or more pain sites in a patient predicted a physicians' VAS score higher than median. Additional findings were that there were no significant differences between the two physicians' VAS ratings overall, or in their ratings of men versus women or in the ratings they assigned when they had a consulting versus an observing position.

Earlier studies in this target group have shown that the patients' higher ratings of pain reflected their illness perspective and suffering and resulted from a complex mix of depressed mood, anxiety and nociceptive input (Löfvander and Taloyan, 2008). The present study was carried out from the perspective of experienced physicians trying to understand and treat young to middle-aged immigrant patients with long-standing pain. Despite this, the physicians' ratings were found to be based mainly on possible sources of somatic pain, with little attention to aspects of possible emotional and psychological pain. This, and various gender aspects as well as socio-cultural and professional characteristics of physicians when rating patients' pain, should be targeted in further studies with different designs.

4.1 Discussion of results

Ours is one of very few studies of physicians' VAS ratings of patients' pain in a multicultural primary health care setting. Our findings are in line with those of other studies where health care personnel rated pain lower than patients with diverse backgrounds (Calvillo and Flaskerud, 1993; Staton et al., 2007). For example, a study in primary care found physicians underestimated pain in black patients compared to other ethnic groups (Staton et al., 2007). It is worth noting that the patients in our study were younger and that somatic and psychosocial factors were taken into account, which was not the case in the cited study. Thus, underestimation of pain appears to be a general phenomenon.

An explanation for the low ratings of pain intensity and the focus on somatic findings could be that VAS reduces the assessment of pain to a quantitative measure. If the physicians had instead been asked to verbalize their understanding of each patient's pain experience, the narrative might also have included emotional influences and other aspects of pain. By way of comparison, a previous study of patients with long-standing pain showed that the physicians' ratings of work capacity were tuned in accordance to the patients' life histories (Löfvander and Engström, 2003).

Medical encounters include many human dimensions (McWhinney, 1997). In this study, the physicians took their time when listening to the patients' narratives. These often included sad personal experiences, which gave a deep human dimension to the consultation that could have favoured a psychological way of their reasoning about the patients' pain. In spite of this patient-centred method, the physicians in this study seemed to have separated somatic from emotional findings, as only more pain sites predicted their higher VAS values. Somatization of depressed mood is common in many cultures (Bäärnhielm and Ekblad, 2000) and may also have been the case for some of the men here (Löfvander and Taloyan, 2008). On the other hand, if physicians interpret pain as having a psychological origin without acknowledging the patient's pain perception, there is a risk that patients will mistrust the physician and hamper rehabilitation efforts.

Compared to a study of pain in children, where VAS was rated by physicians in pair with non-physicians, our inter-rater correlation coefficient was lower (Taddio et al., 2009). Here, when alternating between the consultant and observer roles, the physicians might also have changed the extent to which they empathized with the patient's pain. The physician in the role of consultant might tentatively be more emotionally involved than the observer.

Our study group was a sample from a population with low socio-economic status among whom pain and impairment is common (Dorner et al., 2010). Sweden has a large proportion of foreign-born citizens (13%), who often live segregated in culturally mixed areas and who often work in the service sector (Åslund and Nordström, 2005). Long-standing pain is a common condition in this immigrant population (Andersson, 1994) and is often exacerbated by living under migration stress and worries about the implications of the pain (Bäärnhielm and Ekblad, 2000) with severe social consequences not least among blue-collar workers (Gerdle et al., 2004; Thelin et al., 2008). The large percentage of women in this study with many pain sites might be explained by their triple burden as low-educated women, blue-collar workers and immigrants facing language, social and cultural barriers (Bäärnhielm and Ekblad, 2000). Traditional Swedish multi-modal programmes for patients with long-standing pain are not well adapted for immigrants and are seldom designed to accommodate patients from other cultures (SBU, 2000, 2010). In contrast, this study was part of a research programme adapted for use in busy primary care in culturally mixed areas using a multi-axial diagnostic system of pain including both somatic, psychosocial and psychic aspects as well as interviews on patients' concepts of illness (Löfvander et al., 2004).

4.2 Strengths and limitations

One strength of this study was the standard primary care setting, which allowed us to take clinical data into consideration. Another strength is that although the participants came from a variety of socio-cultural backgrounds, they shared a cluster of characteristics such as age, type of pain, fear of pain, psychosocial stress and perceived disability, had no severe psychiatric pathology and used only mild analgesics. A limitation is that the study population was too small to allow for many subsample analyses and we were obliged to use a crude categorization of cultural background, European and non-European.

4.3 Discussion of methods

Triangulation is often used in the social sciences but seldom in medicine. We used this method to increase the physicians' understanding of psychological aspects of illness but also to make sure that the medical assessment was based on the best possible practice (Griffin et al., 2004). Our study differs furthermore from daily practice in that the physicians did not know the patients' medical history beforehand. In real life, there might be substantial input from previous physicians' evaluations of a patient's problems.

Axis IV of DSM is not a scale, but indicates the pressure of different stressors. DSM-III-R was used for historical reasons, as the programme began in 1993, and DSM-IV was adopted later. Diagnosis of enthesopathy locations was especially important as many of these patients could have injuries related to strenuous work (Larsson et al., 2002). In particular, the pain sites diagnosed in this study must not be confused with trigger points, i.e., pain on pressure.

Our results from this first explorative study cannot be generalized but could probably be transferred to similar settings in Scandinavia, the Netherlands and Great Britain, where patient-centred consultations are the rule.

4.3.1 What is new?

This study includes a study population of immigrant patients, a group that is becoming increasingly common in health care today. It is one of few studies including interpreted consultations and one of the first to show the wide difference between physicians' and patients' ratings of pain severity. It should be noted that a few patients had very low VAS scores, which might indicate stoic pain behaviour among some patients, a phenomenon that is seldom reported (Stempsey, 2004). Further research, including some of qualitative design, may confirm or contradict the findings of this study, as well as delving deeper into the issues it brings up.

4.4 Conclusion

Physicians' VAS ratings were significantly lower than those of the patients, and only findings of multiple pain sites predicted higher physicians' VAS ratings of pain in immigrant patients from various cultural backgrounds. Physicians appear to overlook psychological and emotional aspects when rating the pain of patients from other cultural backgrounds. This finding highlights a potential problem in multicultural care settings. Further research is needed in other settings and with other designs to elucidate how physicians evaluate pain in populations of patients with varying cultural backgrounds.

Author contributions

M.A. interpreted the data and wrote the manuscript.

M.L. conceived and designed the study, analysed and interpreted the data, and wrote the manuscript. Both authors discussed the results and commented on the manuscript.


Thanks to Alf Engström, the general practitioner and colleague who participated in the study.