Intended for healthcare professionals


Underestimation and undertreatment of pain in HIV disease: multicentre study

BMJ 1997; 314 doi: (Published 04 January 1997) Cite this as: BMJ 1997;314:23
  1. François Larue, pain consultanta,
  2. Alain Fontaine, health services research staff physicianb,
  3. Sophie M Colleau, assistant director for communicationc
  1. a Consultation de Traitement de la Douleur, Institut Mutualiste Montsouris, 42 Boulevard Jourdan, 75014 Paris, France
  2. b Département de Santé Publique, Faculté Bichat, Université Paris VII, 75018 Paris, France
  3. c WHO Collaborating Center for Symptom Evaluation in Cancer Care, University of Wisconsin Medical School, Madison, WI 53705, USA
  1. Correspondence to: Dr Larue
  • Accepted 16 October 1996


Objectives: To measure the prevalence, severity, and impact of pain on quality of life for HIV patients; to identify factors associated with undertreatment of pain.

Design: Multicentre cross sectional survey.

Settings: 34 HIV treatment facilities, including inpatient hospital wards, day hospitals, and ambulatory care clinics, in 13 cities throughout France.

Subjects: 315 HIV patients at different stages of the disease.

Main outcome measures: Patients: recorded presence and severity of pain and rated quality of life. Doctors: reported disease status, estimate of pain severity, and analgesic treatment ordered.

Results: From 30% (17/56) of outpatients to 62% (73/118) of inpatients reported pain due to HIV disease. Pain severity significantly decreased patients' quality of life. Doctors underestimated pain severity in 52% (70/135) of HIV patients reporting pain. Underestimation of pain severity was more likely for patients who reported moderate (odds ratio 24) or severe pain (165) and less likely for patients whose pain source was identified or who were perceived as more depressed. Of the patients reporting moderate or severe pain, 57% (61/107) did not receive any analgesic treatment; only 22% (23/107) received at least weak opioids. Likelihood of analgesic prescription increased when doctors estimated pain to be more severe and regarded patients as sicker.

Conclusions: Pain is a common and debilitating symptom of HIV disease which is gravely underestimated and undertreated.

Key messages

  • Pain is a common and debilitating symptom of HIV disease; it is seriously undertreated

  • This multicentre study shows that pain is present in 62% of HIV inpatients, that its severity decreases their quality of life, and that over half with significant pain do not receive any analgesic treatment

  • Undertreatment of pain in HIV disease is related to doctors both underestimating pain and underprescribing analgesics

  • The more severe the pain, the more often doctors underestimate it

  • Doctors are reluctant to prescribe potent analgesics. Likelihood of analgesic prescription increases when doctors estimate pain to be more severe and regard patients as sicker


In the past 15 years intensive clinical research has produced therapeutic improvements for patients infected with HIV, but issues of control of pain and symptoms in such patients have only recently begun to be studied.

Estimates of prevalence of pain during HIV infection range between 30% and 80%.123456 Data, however, have mostly been collected in single site studies,12345 for patients with full blown AIDS,3456 or for those in the terminal stage of the disease,4 and are sometimes based on review of patients' records.3456

Pain experienced by HIV patients can be due to multiple sources7: firstly, the HIV infection itself or its consequences (infections, tumours); secondly, treatments for AIDS; or, thirdly, it can be unrelated to the disease and its treatment. HIV related pain has been found to impair both functional and affective components of daily life.1289

It has been recommended that pain management for AIDS patients follows the model of the World Health Organisation analgesic ladder developed for the management of cancer pain.7101112 Although this model has not yet been systematically validated, clinical reports describe its successful application to the management of pain in AIDS, particularly with respect to the use of opioids to treat severe pain.1314 In addition, adjuvant drugs, such as antidepressants, have been shown to be effective for the treatment of neuropathic pain,10 which seems common in the course of AIDS,7 and is only partly sensitive to opioids.15

A few studies have suggested that pain management for HIV patients is inadequate.1 3 16 A recent study of ambulatory patients with AIDS in New York City showed that characteristics of the patients themselves (sex, education, use of injected drugs) can be associated with higher risk of undertreatment.16 Factors responsible for the undertreatment of cancer pain include doctors' underestimation of patients' pain and lack of knowledge regarding proper analgesic treatment.12 17 18 The role of these factors in the management of pain in HIV has not been assessed.

We designed this multicentre study to determine the proportion of HIV patients who experience significant pain, to assess the impact of pain severity on quality of life, to describe pain treatment practices, and to identify factors associated with undertreatment.

Patients and methods

Between 20 September 1992 and 10 October 1992 we studied 315 HIV patients drawn from 34 treatment settings in 13 cities throughout France. Participating institutions were 13 inpatient hospital wards, 12 day hospitals, and nine ambulatory care clinics. All were Centres for Information and Treatment of Human Immunodeficiency (CISIH), which are the basic administrative and research units for the management of HIV disease in France. Institutional participation in the study was voluntary and depended on the availability of local coordinators trained by the AIDS Task Force of the French Ministry of Health, which served as liaison among study sites. All received specific instruction regarding the study protocol.

Patients and doctors

All HIV patients aged 18 or over, irrespective of stage of disease, were eligible, except those unable to fill out the questionnaire because they either did not speak French or presented cognitive impairment.

All eligible inpatients admitted during the study period were surveyed once within the first three days after admission. In day hospitals or ambulatory care clinics one patient was included every day: if n patients were expected in either treatment setting on a given day, the nth/2 patient was asked to participate; if a patient refused or was ineligible the next patient was considered; when patients were available at both day hospital and ambulatory care clinic in the same city the study was conducted in each setting on alternative days.

Patients completed the French language version of the brief pain inventory,19 previously validated in a study of pain among cancer patients.2021 The inventory has also been used previously to measure pain in patients with AIDS.1 9 16 It asks patients to report if they have experienced pain because of their disease during the previous week and to rate their pain (at its worst, at its least, and on average) on 0-10 numerical scales. In addition, patients were asked to rate their quality of life (0-10 numerical rating scale) and a set of symptoms including fatigue, sadness, and depression (4 point verbal rating scales).

Each doctor was asked to describe the main characteristics of his or her patient's disease and to assess the patient's Karnofsky performance status. Doctors were also asked to rate their patients' severity of pain and level of anxiety and depression, to identify the source of pain if known, and to describe the patient's current analgesic treatment.

All questionnaires were distributed and collected at each site by the local coordinator. Patients were asked to sign an informed consent form after reading the study's objectives. Patients answered the questionnaires without any medical supervision but could be assisted by the local coordinator. Outpatients were asked to complete the questionnaire before their consultation. Doctors were asked to complete the questionnaire within two days of seeing the patient, without knowing the patient's responses.

Patients were classified in three groups (asymptomatic, A1-A2; pathological, B1-B2; full blown AIDS, A3, B3, C1-3) according to the Centers for Disease Control.22

A patient was considered to have pain if he or she reported experiencing pain due to his or her disease in the past week or rated pain severity above 0 on the worst pain scale, or both. Worst pain rated 5 or above was defined as significant pain, because such pain has been shown to interfere significantly with function.23

Analysis of treatment

Following WHO guidelines12 we examined analgesic treatments with respect to pain severity reported by the patients. We categorised the patient's worst pain score as mild (1-4), moderate (5-6), and severe (7-10),21 and we constructed a pain management index as in previous studies.20 23 Negative scores on the index are considered a conservative indicator of undertreatment.

Actually, undertreatment of pain could reflect underestimation of pain or underprescription of analgesics, or both. Because patients' reports of average pain were the most highly correlated with doctors' estimates, we scored the patients' reports of average pain on a 3 point scale (mild, moderate, severe) according to the same cutpoints as those used to categorise worst pain. The doctors' estimate of the patients' pain was scored on a 4 point scale (no pain, mild, moderate, severe). Underestimation of pain severity was defined as doctors rating pain lower than their patients on these ordinal scales. To measure the congruence between the doctors' rating of the patients' pain and their prescribing behaviour, we then constructed a prescription adequacy index, calculated like the pain management index but based on the doctors' rating of the patients' pain rather than on the patients' self reports. Negative values on the prescription adequacy index indicate that the medications prescribed do not match the level of pain assessed by the doctor.

Descriptive statistics were produced to examine patients' characteristics, pain prevalence, and pain severity in each setting. The impact of pain on patients' self reported quality of life was assessed in a multiple linear regression model. Control variables, including treatment setting, stage of disease, fatigue, sadness, and depression, were selected for face validity. Logistic regression analyses identified factors associated with underestimation of pain severity for patients reporting pain and factors associated with pain treatment. Dependent variables were doctors' underestimation of pain and doctors' prescription of any analgesic medication, regardless of its potency. Independent variables included the patients' own reports of pain and the doctors' ratings of the patients' pain severity, as well as patients' characteristics and treatment setting. For each regression, a reduced model was produced by backward elimination. Significance was assessed from the reduction in the goodness of fit of the model.24 Results are expressed in terms of the odds ratio which estimates how each independent variable affects the probability of the behaviour described by the dependent variable.

Table 1

Patients' characteristics by treatment setting. Figures are numbers (percentages) of patients

View this table:


A total of 118 questionnaires were collected from inpatient hospital wards, 116 from day hospitals, and 56 from ambulatory care clinics. Only two patients refused to participate in the study. The treatment setting was not reported for 25 patients, whose characteristics were similar to those of the other patients; these patients were excluded from the analysis.

Patients' ages ranged from 21 to 66 years (median 33); the sample was essentially male (78%). Table 1 shows characteristics related to disease and pain. Patients did not differ significantly between settings with respect to the source of infection. Inpatients had more advanced HIV infection than day patients or outpatients.

Prevalence, severity, and source of pain–Pain was significantly more common for inpatients (62%; 73/118) than for day patients (53%; 61/116) and outpatients (30%; 17/56) (P=0.0005). The prevalence of significant pain (worst pain level ≥5) ranged from 16% (9/56) for outpatients to 51% (60/118) for inpatients. Doctors were able to identify at least one source of pain for over two thirds of the patients reporting pain (105/151). Among 78 AIDS patients for whom a source of pain was identified, 26 (33%) were reported to have digestive or mouth pain, 25 (32%) muscular pain, and 16 (20%) joint or bone pain. Central nervous system pain was reported for 15 patients (19%) and painful peripheral neuropathies for 10 (13%).

Impact of pain on quality of life–Patients with significant pain (worst pain level ≥5) reported lower quality of life during the week before the survey than patients with no pain (fig 1). The linear regression of worst pain on quality of life scores confirms that significant pain had an independent negative impact on HIV patients' quality of life, after adjustment for treatment setting, stage of the disease, fatigue, sadness, and depression (table 2).

Fig 1
Fig 1

Quality of life and pain by treatment setting. Numbers are numbers of patients in each group

Table 2

Impact of HIV related pain on quality of life in 219 patients

View this table:

Adequacy of pain estimation–Both doctors' and patients' pain ratings were available for 135 patients reporting an average pain level greater than zero. Doctors underestimated pain severity in 52% (70) of these patients. Table 3 presents predictors of underestimation. Underestimation was much more likely for patients with severe (odds ratio 165) or moderate pain (24) than for patients with mild pain and somewhat less likely when the patient was perceived as more depressed or when doctors could identify the source of pain.

Table 3

Predictors of under estimation of pain by doctors in 116 patients with pain

View this table:

Adequacy of pain treatment–Table 4 shows the types of analgesic medication prescribed according to the patients' worst pain reports. While 69 patients reported severe pain, only 10 (15%) received opioids and almost half did not receive any analgesic treatment. Among 38 patients reporting moderate pain, only 5 (13%) received at least weak opioids, and most did not receive any analgesic treatment. Overall, 85% (123/144) of the patients whose worst pain level was greater than zero were undermedicated (had negative pain management index scores) according to the WHO guidelines. When severity of pain estimated by the doctor was used as the reference, pain treatment remained inadequate (negative prescription adequacy index scores) for 70% (85/121) of the patients. Antidepressants are considered useful for the treatment of neuropathic pain.10 In this study antidepressants were prescribed to 21 patients but to only one of the 33 patients suffering painful peripheral neuropathies and to two of the 29 patients with central nervous system pain. Table 5 identifies factors associated with the prescription of any analgesic medication, regardless of its potency. Likelihood of prescription increased as doctors recognised pain and estimated it to be more severe; injecting drug users, homosexuals, and patients who seemed more sick (Karnofsky ≤70%) were also more likely to receive some type of analgesic. Identification of the source of pain by the doctor was significantly associated with analgesic treatment in bivariate analyses.

Table 4

Pain treatment by pain severity

View this table:
Table 5

Predictors of analgesic treatment in 246 patients with HIV disease

View this table:


This survey is the first multicentre study of pain in HIV patients at all stages of the disease. It was conducted in France, the country with the highest number of HIV patients in Europe. Our sample is not strictly representative of the general population of such patients in France as participation of institutions was voluntary and patients at advanced stages of the disease were over-represented.25

Pain prevalence and severity were high, comparable with results from a French nationwide study of pain in patients with cancer.20 The higher prevalence and severity of pain observed among HIV inpatients suggest that pain increases as the disease progresses, as shown in a previous prospective study.2 While we recognise limitations in our measure of quality of life, our data show, most importantly, that significant pain diminishes the quality of life of patients with HIV disease, regardless of treatment setting and stage of disease.

This study also shows that HIV related pain was both severely underestimated and undermedicated. It is especially disturbing that doctors underestimated pain when it was most severe. Moreover, doctors seemed reluctant to prescribe potent medications, even though they took into account their rating of the patients' pain to prescribe analgesics. They also failed to use antidepressants when pain relief could be expected from those adjuvant drugs. Inappropriate pain treatment has previously been shown in France20 26 and may be related to insufficient awareness and training regarding pain management.27 Insufficient knowledge and reluctance to use morphine constitute barriers to adequate pain management in cancer.18 The same factors may apply here. Interestingly, injected drug use did not seem to be an obstacle to the prescription of analgesics.

We found that the ability of a clinician to identify the source of pain led to better pain assessment and, possibly, to its treatment. The subjectivity of pain in a complex disease such as HIV infection where somatic, visceral, and neuropathic pains are often associated with anxiety and depression makes recognition and assessment of pain particularly daunting for clinicians.278 Pain assessment in HIV patients may require development of specific methods, including multisymptom assessment scales. Recognising the validity of the patients' point of view still remains the cornerstone of adequate pain assessment.

Improvement of management of pain for HIV patients will require new public health strategies to promote necessary changes in clinicians' knowledge, attitudes, and practices.


We thank L Bonifassi and Dr R Demeulemeester for their help in the study protocol, and Drs Ch Michon, Ph Vinceneux, J Pouchot, JM Nadal, and G Bez for their critical reading of the manuscript. We also thank the participating institutions: CISIH of Clamart, Colombes, Dijon, Grenoble, Marseille, Montpellier, Nice, Paris (Bichat), Rennes, Rouen, Strasbourg, Toulouse (Rangueil and Purpan), and Tourcoing; and L Brasseur, M Couturier, F Cornu, and M Mullier of Action Douleur for their help in pursuing initiatives to improve pain control in France.

Funding: Foundation de l'Avenir pour la Recherche Médicale Appliquée (grant No: 92/76) and the AIDS Task Force of the French Ministry of Health.

Conflict of interest: None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
View Abstract