Impact of dementia on cancer discovery and pain


Dr Shuji Iritani MD PhD, Department of Psychiatry, Graduate School of Medicine, Nagoya University, 65 Tsurumai, Showa-ku, Nagoya, Aichi 466-8550, Japan. Email:


Background:  Dementia is clinically noted to influence both reporting and experience of cancer pains. However, no systemic evaluation of this aspect has been reported. The aim of the present study was to retrospectively evaluate how dementia modified the cancer discovery process, frequency of cancer pain reports and analgesic-narcotic use at a large psychiatric hospital.

Methods:  We reviewed all the records of cancer patients with and without dementia treated at the surgical ward of Matsuzawa Hospital from 1993 to 2004. Psychiatric diseases other than dementia, brain metastasis and alcoholism, as well as leukaemia and skin cancer, were excluded. Patients' communicativeness as to pain was ascertained from nursing records.

Results:  A total of 134 cancer patients with and without dementia (50 demented and 84 non-demented) were included. Demented patients were accidentally discovered to have cancer (48%) or by an unexpected unfolding of clinical signs (44%), whereas most non-demented patients (63%) voluntarily sought medical evaluation (P= 0.000). Overall, 76% of non-demented patients had cancer pains (stages I and II, 64%; stages III and IV, 84%), whereas just 22% of demented patients had cancer pains (stages I and II, 16%; stages III and IV, 26%; P= 0.000). Non-demented patients showed stage-dependent requirements for both non-narcotic analgesics (stages I and II, 64%; stages III and IV, 84%) and narcotics (stages I and II, 0%; stages III and IV, 41%). Demented patients required much less analgesics (stages I and II, 11%; stages III and IV, 13%), with only one stage IV patient requiring narcotics (P= 0.000).

Conclusion:  Dementia greatly modifies the cancer discovery process, reduces prevalence of cancer pain and analgesic requirement.


The impact of dementia on pain and its management has been reported mainly in the context of underreporting, underdetection and undertreatment.1–4 Regarding the phenomena of reduced pain complaints and subsequent low requirement of analgesics, the question arose whether an alteration in pain experience in the demented individuals was reflected or whether this was a result of a decline in the patients' ability to communicate about their pain.5 It was suggested that dementia influenced both the experience and reporting of pain in elderly individuals.6–8 Parmelee et al. reported that in the elderly with mild to moderate cognitive impairment, their pain complaints are as valid as those of the non-demented elderly.6 More specifically, as to pain experience, Alzheimer's disease (AD) patients appeared to perceive less pain intensity and pain affect than non-demented elderly, despite the fact that both groups had equal sensitivity regarding pain stimuli and pain thresholds.9,10 Scherder et al. repeated pain assessments on AD patients daily for 4 weeks, and 2 months later for 5 days and confirmed the foregoing finding.11

If demented individuals tend to tolerate intensified pain and experience less pain affect, they are anticipated to have an increased risk of not noticing the forewarning message of diseases, such as cancers, which are often curable if discovered and treated in their early stages. However, little attention has been directed to the impact of dementia on the process of cancer discovery and pain management in the subsequent hospitalization. In a prospective study of 200 elderly outpatients who were suspected of dementia, Larson et al. noted that roughly half had a variety of illnesses previously unrecognized, including a case of colon cancer.12 Both alteration of pain experience and inability to recognize the implication of discomfort are expected to influence the discovery of cancer. As extraordinary examples, Fisher-Morris and Gellatly reported two AD patients who consistently denied pain perception despite an ulcerated fungating breast cancer in one and a hair-line fracture of the femur in the other, and through a small scale survey collected 49 cases of AD patients who failed to show normal experience of pain.13

Our recent observation of patients with AD who had been by chance found to have cancer, but had consistently denied cancer pain, prompted us to review medical records of our hospital, focusing on the behaviour of demented patients with cancer who were communicative as to perceived pain. We hypothesized that if patients with dementia tended to perceive less pain or discomfort originating from cancer and as a consequence were less likely to seek medical evaluation than intact cancer patients, their cancer would be detected either serendipitously during the evaluation of comorbidity or at an annual routine medical check, or as a result of an unfolding of unexpected clinical signs of an extent that could not be dismissed, such as haematemesis, melena, and so on. And even when they were hospitalized for evaluation and treatment, they would complain less frequently of physical discomforts related to cancer, and accordingly, analgesics and narcotics would be prescribed less frequently than for cancer patients without cognitive impairment.

We compared the behaviours of cancer patients with and without dementia with respect to (i) the process of cancer discovery; (ii) the recorded pain complaints excluding those clearly unrelated to cancer; and (iii) the use of analgesics and narcotics documented in the medical records in our hospital.

The present study was approved by the Medical Ethics Committee of the Tokyo Metropolitan Matsuzawa Hospital. We have also strictly protected personal data throughout the study according to the privacy guidelines of the hospital.


Characteristics of the hospital

Our hospital is the largest public psychiatric hospital (1160 beds during this study period) in Japan that attracts psychiatric patients, including those with dementia, from across the country. For psychiatric patients with physical complications, the surgery, internal medicine, orthopedics and neurosurgery wards were available in our hospital. These wards also received general patients without psychiatric disease.

Materials and procedure

Data were collected by reviewing medical records of all the patients who were admitted to the surgical ward for cancer evaluation and treatment from 1993 to 2004.

Selection criteria

  • 1Cancer patients with and without dementia, who were judged to be communicative as to pain and discomfort. Those with other psychiatric disorders, such as parkinsonism, brain metastasis and alcoholism, as well as leukaemias, lymphomas, skin malignancies and disturbances of consciousness, were excluded.
  • 2Dementia was diagnosed based on patients' history, clinical evaluation, computed tomography scan and psychiatric consultation. For the cursory assessment of levels of cognitive functioning, the Mini-Mental State Examination (MMSE)14 and the revised Hasegawa's Dementia Scale (HDS-R)15 were used. HDS-R is most widely used in Japan for a convenient clinical assessment of cognitive impairment. It samples orientation to time and place, registration, recall, attention, memory, recitation of quoted numbers in reverse order and calculation; it correlates closely with MMSE (correlation coefficient 0.94) and its optimal cut-off point discriminates the demented from the non-demented with the sensitivity of 0.90 and specificity of 0.82.15 The severity of dementia was assessed according to the Criteria of Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR).16 Some demented patients who had been taking neuroleptic drugs for behavioural and psychological symptoms of dementia only temporarily, not consecutively, were included. Some patients with or without dementia had been taking a low dose hypnotic drug, such as zopiclone (7.5 mg/night).
  • 3Communicativeness was assessed from nursing records, which evaluated activities of daily living (ADL) and communication status on admission; a nursing record chronicles daily conditions of the patient in terms of verbal expression, observed states and assessments made by the nurse. The absence of recorded verbal expression and patient's responses that were irrelevant to the nurse's inquiry were regarded as a lack of communicativeness. A total of 11 demented patients were judged and excluded as non-communicative according to the above criteria.


  • 1Mode of cancer detection was classified as follows:
    • a. Casual detection at the time of annual physical check-up or laboratory examination for comorbidity, including facial paleness and/or swelling of lower extremities noted by the family.
    • b. Voluntary visits to medical institutions for evaluation of perceived discomforts.
    • c. Unexpected unfolding of clinical manifestations of degree that could not be dismissed, such as vomiting, haematemesis, ileus, tarry stools, haematuria and vaginal bleeding.
  • 2Complaints of pain or discomfort chronicled in nursing records. Pain complaints might be heterogeneous and might not be directly related to cancer.17,18 However, it is sometimes difficult to differentiate cancer pain from other types of pain in medical records. Therefore, we have counted every pain recorded in patients, whenever uncertainty remained regarding the nature of the pain. Postoperative pain was controlled by continuous epidural infusion for 4–5 days after abdominal surgery,19 (which was the case for most of our patients) and pains requiring analgesics or narcotics after this period were counted as cancer pain. Pain was counted only in terms of its absence or presence irrespective of its severity, because pain description is subject to observational bias and therefore difficult to standardize. We believed the use of analgesics or absence thereof was a more reliable indicator of pain severity.
  • 3Use of non-narcotic analgesics (scopolamine butylbromide, pentazocine, diclofenac sodium, buprenorphine hydrochloride) and narcotics (morphine hydrochloride, morphine sulfate).

Statistical analysis

We compared the two groups (50 demented and 84 non-demented patients) on the variables of sex, age, organ involved by cancer, cancer stage, process of cancer discovery, pain, and the requirement of analgesics and narcotics by means of the Fisher's exact test, χ2-test and the Mann–Whitney U-test. These analytical means were also used to compare: (i) AD (n= 32) and vascular dementia VD (n= 13) on the same variables (except for cancer site); and (ii) the demented (n= 40) and non-demented (n= 71) patients with cancer of the gastrointestinal tract. Analysis was made using spss 11.5J (SPSS, Chicago, IL, USA) and differences were considered to be statistically significant at P < 0.05.



We selected the records of 134 patients who met the aforementioned criteria. Initially, 61 demented patients with cancer, which met our selection criteria for cancer, were selected. However, 11 were excluded because of the suspected lack of communicativeness and 50 patients with dementia (32 AD, 13 VD, four mixed type, one Pick's disease; 29 men and 21 women, mean age 74.1 ± 8.1 years) were retained for analysis; severity of dementia was judged in 29 patients as mild to moderate, 20 patients as severe and one patient as undeterminable. A total of 84 cancer patients were without cognitive impairment (46 men and 38 women, mean age 72.1 ± 10.5 years). There was no statistical difference as to age between the demented and the non-demented groups (P= 0.205)

Process of cancer discovery

A total of 63% of non-demented patients sought medical consultation because of physical discomforts, whereas just 8% of patients with dementia did so (Fig. 1, P= 0.000). The rest of the patients with dementia (92%) were found to have cancer either accidentally by way of another medical evaluation, such as anaemia (48%) or as a result of an unexpected unfolding of clinical signs, such as vomiting, melena or vaginal bleeding (44%) (P= 0.000). No statistically significant difference was found between AD and VD as to the mode of cancer detection (P= 0.620).

Figure 1.

A total of 63% of non-demented patients sought medical attention because of physical discomfort, whereas just 8% of patients with dementia did.

Pain complaints

Pain was recorded in 76% of the patients without cognitive impairment (stages I and II, 64%; stages III and IV, 84%), whereas pain was reported in just 22% of the demented patients (stages I and II, 16%; stages III and IV, 26%; P= 0.000; Table 1). Recorded pain did not appear to be related to the severity of dementia, although the number of cases with pain was too small to make any definite comment. The prevalence of pain increased proportionally as the cancer stage advanced in both demented and non-demented groups (Fig. 2). No statistically significant difference was noted between AD and VD in this respect (P= 0.930).

Table 1.  Cancer type, pain, stage
I or IIIII or IVAll stagesI or IIIII or IVAll stages
Painn+n+n (%)+n+n+n (%)+
Oesophagus, stomach, duodenum9091331022319148623185372611
Liver, gallbladder101000101440220660
Uterus, ovary110110220000110110
Total (%)193163182350113933211251438846420
Figure 2.

The prevalence of pain increased proportionally as the cancer stage advanced in both demented and non-demented groups.

Cancer stage and pain

Types of cancer, with stages and pain distribution, are shown in Table 1. Two major cancer types are those of the upper gastrointestinal tract, and the colon and rectum. In both demented and non-demented groups with gastrointestinal cancers, the prevalence of pain increased with stage-progression. In the former group, 7% of stages I and II, and 27% of stages III and IV patients had pain; whereas in the latter group, 57% of stages I and II, and 81% of stages III and IV patients had pain (P= 0.000, not shown in the Table).

Pain perception in patients with dementia

It was noted that patients with dementia frequently complained of pains clearly unrelated to cancer, despite the fact that their cancer-related pain complaints were definitely reduced. Their pain complaints were associated with falls and consequent fracture of the femur or ribs, trans-position of the body (in one case, it disclosed the fracture of the femoral head not complained of at the time the fracture occurred), osteoporosis, unskilled insertion of i.v. infusion or bladder catheter, foreign body in the eye and so on. Thus, the patients' abilities to feel physical pain suggest, that in cancer pain, affective components, such as anxiety, might play a role.

Analgesic use

Analgesic use increased in cancer patients without dementia in parallel with stage progression (Table 2). Non-narcotic analgesic was required for 64% of patients in stage I and II, and 84% in stages III and IV; narcotic requirement was limited to 41% of the patients in stages III and IV (Fig. 3). Similarly, analgesic requirement in cancer patients with dementia increased with stage progression, but on a much smaller scale; just 11% of stages I and II, and 13% of stages III and IV patients required analgesics (P= 0.000). Only one demented patient required narcotics. Again, no statistically significant difference was found as to analgesic or narcotic use between AD and VD (P= 1.000).

Table 2.  Requirement for analgesics and narcotics (%)
PatientsStageNo. of PatientsAnalgesicsNarcotics
DementedI, II19 (100)2 (11)0 (0)
III, IV31 (100)4 (13)1 (3)
Combined50 (100)6 (12)1 (2)
Non-dementedI, II33 (100)21 (64)0 (0)
III, IV51 (100)43 (84)21 (41)
Combined84 (100)64 (76)21 (25)
Figure 3.

Non-narcotic analgesic was required for 64% of patients in stages I and II, and 84% in stages III and IV; narcotic requirement was limited to 41% of the patients in stages III and IV in non-demented groups. In demented groups, non-narcotic analgesic was required for 11% of patients in stages I and II and 13% in stages III and IV; narcotic requirement was limited to 3% of the patients in stages III and IV.


We report two findings with implications for the treatment of cancer patients with dementia. First, our retrospective study shows that dementia greatly reduces patients' voluntary seeking of medical help. Second, cancer patients with dementia complain, despite the fact their communicativeness is maintained, much less frequently of pains associated with cancer, and accordingly require far less analgesics and narcotics.

Cancer patients with dementia in the present study emerge to be rather unconcerned with their physical conditions, which would have worried non-demented individuals, despite the fact that they were communicative as to pain or discomfort. Only 8% of the demented patients voluntarily sought medical attention, whereas a majority (63%) of patients without dementia did so (P= 0.000). Because all the elderly are insured for health problems, economic reasons do not account for this discrepancy. One of the factors that might contribute to the apparent indifference of the patients with dementia was that they were slightly, but not significantly, older than the non-demented patients (mean age: 74.1 ± 8.1 vs 72.1 ± 10.5 years); pain intensity appears to diminish with advancement of age.20–23 However, this difference in the process of cancer discovery is far too great to be accounted for by age difference alone.

As to the apparent indifference of the demented patients regarding their physical condition, at least two sets of explanations are conceivable. First, seeking medical intervention requires not only functions to interpret, recall and communicate pain symptoms, but also to some extent, executive processing abilities, such as organizing, planning and monitoring one's problem-solving behaviours. However, loss of such abilities is common in dementing illnesses.24 Even if self-reports of pain by cognitively impaired elderly are no less valid than those of cognitively intact individuals,6 they might not initiate and carry on a problem-solving behaviour. Second, patients with dementia might have reduced pain experience as a result of neuropathological changes that cause dementia.3,8,9,25 Benedetti et al., by applying pain stimuli to AD patients and controls, observed that there was no difference between the two groups as to stimulus detection and pain threshold.9 However, a clear-cut increase in pain tolerance, which ‘depends on cognitive and affective factors’, was present in AD patients; the more severe the cognitive impairment, the higher the tolerance to pain.9 By comparing AD patients with controls matched for the presence of painful conditions, Scherder et al. also observed that AD patients reported less pain intensity and less pain affect than controls.10 With less pain affect and cognition, it appears that the demented individuals are less likely to attribute the perceived discomfort to a dreaded disease such as cancer.

Reflecting the high incidence of gastrointestinal cancers in Japan, an overwhelming majority (83%) of our patients had cancers originating in the oesophagus, stomach or colorectum. These cancers are currently regarded as having a good prognosis if they are discovered at early stages and surgically treated. For instance, overall postoperative 5-year survival rates for stomach cancer are 93.4% and 87% for stages IA and IB, respectively, and 68.3% for stage II. However, survival rates plunge to 50.1% and 30.8% for stages IIIA and B, respectively.26 Thus, it is possible that cognitive impairment adversely influences the prognosis of cancer patients.

The prevalence of pain in cancer patients without cognitive impairment in the present study (stages I and II, 64%; stages III and IV, 84%) appears somewhat greater than the reported figures.27,28 Daut and Cleeland observed that in regard to patients with colorectal cancer, pain was present in 40% of non-metastatic cases and in 47% of metastatic cases.17 In our non-demented patients with colorectal cancer (Table 1), eight out of 14 (57%) stages I and II, and 17 out of 20 (85%) stages III and IV patients had pains (74% in the combined group). Because we have included all the pains except for the ones unambiguously unrelated to cancer, it is possible that some of the counted pains were not directly caused by cancer; Twycross and Fairfield reported that 39 out of 100 patients with ‘far-advanced cancer’ had one or more pains unrelated to cancer or treatment.18 In contrast to the high prevalence of pain in cancer patients without dementia, only five out of 18 demented patients with colorectal cancer (28%, all stages) had pain (P= 0.000).

The finding that patients with dementia had far less cancer pain records and analgesic requirement deserves special attention, because of possible ethical implications of neglecting or mismanaging pain. On the one hand, the possibility of underdetecting or undertreating pain in the demented elderly has been admonished.3,4,29,30 On the other hand, solid evidence has emerged to suggest that besides a decline in communication, their neuropathology probably has caused less pain experience, contributing to the reduced use of analgesics in AD.8–11

However, the foregoing observations were made on pain experiences of non-cancerous diseases. The dramatic difference between cognitively impaired patients and intact patients as to pain report and analgesic requirement in the present study (Figs 2,3) supports the theory that cancer pain has multidimensional components including sensory, affective, behavioural and cognitive components.31 In particular, pain experience is expected to be aggravated by fear regarding disease progression.17,18 However, the present study suggests that patients with dementia are much less likely to share this pain perception magnified by fear and anxiety, evoked by the cognition of disease nature.

The present study showed no difference between AD and VD as to frequency of recorded pains and analgesic use, which were far reduced in both groups compared with those of non-demented cancer patients. The pain and analgesic use pattern in our VD patients are seemingly at odds with the observation made by Scherder et al. of patients with possible vascular dementia.32 They noted that, compared with the non-demented elderly who had comparable painful conditions such as arthritis and osteoporosis, VD patients showed a significant increase in the scores on the scales measuring pain and pain affect, and took significantly more paracetamol than the control group.32 Several factors make it difficult to carry out a plain comparison. First, in our patients, pain record was not measured, but relied on the reporting from patients, and there might have been an underreporting by VD patients. Second, our diagnostic procedure included computed tomography scans, which would improve diagnostic accuracy, whereas their diagnoses were made solely on a clinical basis.32 Third, cancer pain might be different in the sense that it has a psychological factor in operation from the pain of arthritis or osteoporosis. Although it could be greatly potentiated by its interpretation as the progression of disease,17,31 perceived intensity of pain originating from cancer might be much less without fear and anxiety. Fortunately, in cancer patients with dementia, such negative thought is unlikely to occur. Fourth, in VD patients with cancer, as Scherder et al. pointed out, there might be a hyperactivity of the hypothalamus–pituitary–adrenal axis generating an increased amount of corticotropin releasing hormone which ameliorates pain sensation.32

Our observations should be viewed with some reservations. First, the medical records were collected over a more than 10-year period and documentation of nursing records regarding pain were not strictly standardized, although diagnostic criteria, observational method and management policy regarding pain have not changed in the meantime. Second, communicativeness of patients with cognitive impairment was assessed based on the evaluations at the time of ward admission and daily descriptions of patients' verbal expression. These records are necessarily brief and do not allow one to estimate the full extent of communicativeness. Third, collected cases in the present study are almost restricted to malignancies of the gastrointestinal tract and not generalizable to all the malignancies involving individuals with dementia. Fourth, some patients with or/and without dementia had taken low-dose neuroleptic agents temporarily and/or minimum hypnotic agents consecutively. Another study design is needed to eliminate the influence of pharmacotherapy on cancer pain.


The authors thank the staff of the management clinical record section of Tokyo Metropolitan Matsuzawa Hospital (TMU) for their assistance. We thank Dr Kawabata (the chief Surgeon doctor of TMU) for the comments on the clinical treatment of cancer and Dr Matsushita (the former director of TMU) for support of this study.