Journal Logo

Research Paper

Comparative differences in musculoskeletal pain consultation and analgesic prescription for people with dementia: a UK-wide matched cohort study

Bullock, Laurnaa,*; Bedson, Johna; Chen, Yinga,b; Chew-Graham, Carolyn A.a,c; Campbell, Paula,c

Author Information
doi: 10.1097/j.pain.0000000000002257

1. Introduction

Approximately half of the community-dwelling (ie, not within hospital or formal care) older adults with and without dementia have pain,49 with the most common cause of persistent pain being musculoskeletal based.45 Musculoskeletal conditions are one of the most common reasons that people access healthcare services and confer a considerable burden to the individual and wider society.12

Symptoms associated with dementia (eg, diminished language capacity, memory impairment, and behavioural and psychology changes) may lead to difficulties articulating a pain experience or fulfilling their unmet need for pain relief.18 Rather than verbally communicating their pain experience, people with dementia may express pain through nonverbal expressions and behavioural changes (eg, poor sleep, decreased appetite, or withdrawal from usual activities).46 Consequently, caregivers and clinicians may not recognise or interpret expressions of pain correctly; wrongly attributing expressions of pain (eg, agitation) as a symptom of dementia thus leads to the inadequate assessment and treatment of pain.27 Research evidence predominately within formal care settings (ie, care homes) has identified that people with dementia have significantly fewer pain assessments than older adults without dementia and that pain identification becomes increasingly problematic aligned to dementia disease severity and level of cognitive impairment.40,43 A similar picture exists with prescription of analgesics; with a recent systematic review and meta-analysis of cross-sectional data finding that people with dementia had a significantly lower analgesic prescription prevalence compared with people without dementia.49

Clearly adequate pain identification and assessment is a prerequisite for optimal pain treatment generally45; however, there may be greater importance in those with dementia. Poorly managed pain has been associated with behavioural and psychological changes,21 increased emergency department attendance,29 and increased cognitive impairment,19 each of which is linked to poor outcomes for people with dementia, such as premature care home admission4 and death.11 Adequate pain management is therefore an integral aspect of care to prevent or delay such outcomes, thereby supporting people with dementia to continue living independently, in accordance with key health policy agendas.2

Much of the research evidence to date is focused on formal care populations (eg, care home residents),13 with research exploring pain management in community or primary care settings limited to cross-sectional, descriptive, and small sample designs.13 Recent research has called for quantitative studies to examine pain assessment and analgesic prescribing for people with dementia within community settings such as primary care.30 This study aimed to describe the longitudinal prevalence of musculoskeletal consultation and analgesic prescription for people with dementia compared with matched older adults without dementia in a UK primary care database.

2. Methods

2.1. Study setting and population

This study used data from the Clinical Practice Research Datalink (CPRD) GOLD to conduct a retrospective cohort study. The CPRD is a UK primary care medical database containing high-quality and anonymised data on more than 11 million patients,28 of which 2.8 million were active in 2017. In the United Kingdom, 98% of the population is registered with a general practice28 making primary care electronic health record data an ideal representative sample. When compared with the UK 2011 census, CPRD patients were representative of the UK population in relation to age, sex, and ethnicity.28 The CPRD includes data on patients' clinical conditions, diagnoses, symptoms that correspond to ICD classification codes, information on tests, referrals, and prescribed medications (corresponding to the British National Formulary) as well as information on demographics and health behaviour.28 Furthermore, patient information from secondary care is also included in the CPRD, eg, medication prescribed from secondary care continued in primary care.3

Data were retrieved from January 1, 1995 to December 31, 2017. For the purpose of this analysis, the general practices included were linked to the Office for National Statistics practice-level Index of Multiple Deprivation. Research indicates that practices with and without linked data are similar in regards to demographic data, years of follow‐up, and prescribing of medication.22

2.2. Study participants

Electronic health record data are principally collected in UK primary care using Read codes entered by members of primary care staff. Read codes are a standard, hierarchical vocabulary of clinical terms used to document various clinical information, including but not limited to symptoms, signs, diagnoses, and prescriptions.9

A dementia cohort (exposed population) was identified with a dementia diagnostic Read code or a dementia-related drug between January 1, 1997 and December 31, 2017. Dementia index date was defined as the incident (first record) dementia diagnostic Read code or a dementia-related drug (whichever came first). A matched cohort (1:1) by year of birth, sex, and general practice was identified at baseline with no evidence of a dementia diagnostic Read code or a dementia-related drug between 1997 and 2017. Dementia index date was assigned for patients in the matched cohort within their respective matched pair for analysis. Read codes were identified from previously defined clinical codes lists15,16 (available online at

To be included in the dementia or matched cohort, all patients had to be aged 50 years or older at index date, with no evidence of a Read code indicative of formal care residence during 1997 to 2017. All patients had at least a 2-year period between their entry date and their index date, with no evidence of a Read code indicative of cancer diagnosis during this period. In addition, patients were excluded 6 months before their first morbidity cancer Read code during follow-up. Finally, all patients must have had evidence of a face-to-face or a telephone consultation with a GP or a nurse within a 90-day prewindow and postwindow of their assigned index date to ensure that they were active consulters.

2.3. Outcome measures

2.3.1. Musculoskeletal consultation

Musculoskeletal consultations were identified using previously validated Read codes documented in the patient's record (available online at Previous research has documented musculoskeletal consultation as an appropriate marker to encapsulate pain identification and assessment for older adults in the CPRD.5

2.3.2. Analgesic prescription

Evidence of analgesic prescription was identified using a previously validated hierarchical classification7 that categorised analgesics into 6 groups based on their potency, in line with the World Health Organisation analgesic ladder. At the bottom of the ladder are basic analgesic prescriptions (eg, paracetamol). Opioid analgesic prescriptions were separated into 4 classifications based on their potency: weak analgesics, moderate analgesics, strong analgesics (each containing increasingly strong opioids used alone or in combination with paracetamol), and very strong analgesics (very strong single opioids such as morphine), and finally nonsteroidal anti-inflammatory drugs (NSAIDs).7 An additional category included evidence of (any) analgesic prescription, irrespective of classification or potency.

2.4. Covariates

Covariates that could be associated with the outcomes of interest were identified from a systematic review conducted by the authors.13 Covariates included the following:

  • (1) Evidence of specific comorbidities (cardiovascular-related conditions, diabetes, or depression) during the 2 years before index date.51
  • (2) A surrogate measure for comorbidity calculated using the total number of prescriptions during the 2 years before index date mapped to different British National Formulary sections.42,44
  • (3) The frequency of consultations by each patient during the 2 years before index date defined as “any face-to-face or telephone consultation completed by a doctor or a nurse.”48
  • (4) Length of follow-up (the number of days from index date to exit date).
  • (5) The year of index date, as this may be at any time during the 20-year study period (January 1, 1997-December 31, 2017) to adjust for potential cohort effect over time.

2.5. Follow-up

Follow-up continued until 5 years after index date or until the patient no longer contributed data, known as their “exit date” (eg, the date of the practice or patient left the CPRD, patient death, 6 months before the patient's first morbidity cancer Read code, or if the study end date [December 31, 2017] was reached). All analyses were conducted for a maximum 5-year period from index date and each annual period from index date to 5 years after index date. Stratification into annual periods from index date allowed examination of any patterns in musculoskeletal consultation and analgesic prescription over time from index date in line with the expected progression of dementia and worsening of symptoms.24

2.6. Statistical analysis

A matched cohort comparison was conducted. The baseline demographics were first described and compared using univariate statistical tests (eg, t tests or χ2 tests where appropriate).

To calculate period prevalence, patients contributed to the numerator of the equation if they had evidence of the outcome (musculoskeletal consultation or analgesic prescription) during the specified period. If there was evidence of musculoskeletal consultation or analgesic prescription within the period, additional consultations or analgesic prescriptions (of the same strength classification) were ignored.31 The denominator included patients eligible throughout the period. Period prevalence was reported as a percentage.

Conditional logistic regression models examined the association between patient status (dementia cohort vs matched cohort) and the outcome (evidence of musculoskeletal consultation or analgesic prescription during each period). Conditional logistic regression models produced unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (95% CIs), accounting for matched variables (general practice, year of birth, and sex) and adjusting for previous recorded consultations for cardiovascular-related conditions, diabetes and depression, morbidity, length of follow-up, year of index date, and consultation frequency. Model assumptions were checked, including the linearity of the logit assumption by checking continuous variables were linearly related to the log of the outcome variable. If the assumption was violated, continuous variables were categorised with homogeneity within each strata implicitly assumed.

3. Results

3.1. Baseline demographics

The study cohort included 73,164 participants (36,582 dementia cohort, and 36,582 matched cohort) at baseline. Table 1 shows the baseline characteristics of the 2 cohorts. The dementia cohort and matched cohort were 59.8% female, with a mean age of 79.9 (SD, 8.3) years at the index date. The dementia cohort had significantly shorter median follow-up (in days) than the matched cohort (621 vs 1225 days). During the 2 years before index date, the dementia cohort had a higher count of codes indicating comorbidity and specifically were more likely to have records of cardiovascular-related conditions, depression, and diabetes, compared with the matched cohort.

Table 1 - Characteristics of study participants in the dementia cohort (n = 36,582) and the matched cohort (n = 36,582) at baseline.
Dementia cohort n = 36,582 Matched cohort n = 36,582 P
Sex, female % (n) 59.8 (21,860) 59.8 (21,860) Matched
Year of index date, mean (SD) 2008.67 (4.91) 2008.67 (4.91) Matched
Age at index date, mean (SD) 79.9 (8.3) 79.9 (8.3) Matched
Follow-up (d), median (IQR) 621 (250-1192) 1225 (551-2246) <0.001
Practice IMD Matched
 1—least deprived 16.3 (5958) 16.3 (5958)
 2 19.2 (7010) 19.2 (7010)
 3 19.8 (7259) 19.8 (7259)
 4 21.2 (7743) 21.2 (7743)
 5—most deprived 23.5 (8612) 23.5 (8612)
Morbidity count, median (IQR)* 11 (6-16) 10 (6-15) <0.001
Consultation freq, median (IQR)* 34 (19-55) 28 (15-47) <0.001
CVD yes, % (n)* 7.4 (2705) 6.0 (2194) <0.001
Depression/bipolar, yes % (n)* 8.1 (2962) 2.6 (965) <0.001
Diabetes yes, % (n)* 16.7 (6115) 14.9 (5459) <0.001
*Recorded during the 2 years before index date.
CVD, cardiovascular disease; IMD practice-level Index of Multiple Deprivation; IQR, interquartile range.

3.2. Musculoskeletal consultations

3.2.1. Five-year prevalence of musculoskeletal consultations

During the 5-year period after index date, the dementia cohort had a significantly lower prevalence (12.3%) of musculoskeletal consultation than the matched cohort (58.5% vs 70.8%). Multivariable conditional logistic regression found that during the 5-year period from index date, the dementia cohort had a lower odds of musculoskeletal consultation than the matched cohort (adjusted OR = 0.83, 95% CI 0.78-0.89) (Table 2).

Table 2 - Period prevalence and odds of musculoskeletal consultation for the dementia cohort and matched cohort.
Years Dementia cohort Matched cohort OR (95% CI) Adjusted* OR (95% CI)
Prevalence % (95% CI)
0-5 (the overall 5-year period) 58.54 (56.99-60.08) 70.76 (69.95-71.56) 0.84 (0.79-0.90) 0.83 (0.78-0.89)
0-1 24.46 (23.92-25.00) 30.79 (30.27-31.31) 0.80 (0.77-0.83) 0.82 (0.78-0.85)
1-2 22.26 (21.62-22.90) 30.55 (29.98-31.12) 0.73 (0.70-0.77) 0.73 (0.70-0.77)
2-3 19.94 (19.18-20.73) 30.56 (29.92-31.21) 0.65 (0.61-0.69) 0.66 (0.62-0.71)
3-4 19.27 (18.33-20.25) 31.71 (30.99-32.44) 0.60 (0.56-0.65) 0.62 (0.56-0.67)
4-5 19.52 (18.29-20.80) 31.04 (30.23-31.87) 0.63 (0.57-0.70) 0.61 (0.54-0.68)
Time point 0 = index date.
*Adjusted for previous recorded consultations for cardiovascular-related conditions, diabetes and depression, comorbidity count, length of follow-up, year of index date, and consultation frequency.
CI, confidence intervals; OR, odds ratio.

3.2.2. Annual prevalence of musculoskeletal consultations

The prevalence of musculoskeletal consultation for the dementia cohort gradually decreased in each annual period from index date to 5 years after index date (24.5%-19.5% P < 0.001). By contrast, the prevalence of musculoskeletal consultation for the matched cohort remained relatively stable throughout follow-up, with a slight increase in consultation prevalence during the latter annual periods. Multivariable conditional logistic regression models found that the difference between the dementia cohort and matched cohort increased with the increase of disease duration of dementia (Table 2) (first year: adjusted OR = 0.82, 95% CI 0.78-0.85; final year: adjusted OR = 0.61, 95% CI 0.54-0.68).

3.3. Analgesic prescription

3.3.1. Five-year prevalence analgesic prescription

The 5-year prevalence of evidence of any analgesic prescription was similar for the dementia cohort compared with the matched cohort (76.7% vs 79.0%, respectively; adjusted OR = 0.97 (0.91-1.03) (Table 3).

Table 3 - Five-year period prevalence and odds of analgesic prescriptions for the dementia cohort and matched cohort.
Dementia cohort prevalence % (95% CI) Matched cohort prevalence % (95% CI) OR (95% CI) Adjusted* OR (95% CI)
Any analgesic 76.73 (75.37-78.03) 78.99 (78.26-79.70) 0.98 (0.93-1.05) 0.97 (0.91-1.03)
Basic analgesic 63.45 (61.92-64.95) 62.13 (61.27-62.98) 1.05 (0.99-1.13) 1.03 (0.96-1.11)
Weak analgesic 31.34 (29.90-32.81) 36.49 (35.64-37.34) 0.88 (0.81-0.97) 0.86 (0.78-0.95)
Moderate analgesic 17.98 (16.81-19.22) 22.15 (21.42-22.89) 0.79 (0.70-0.89) 0.74 (0.65-0.85)
Strong analgesic 22.86 (21.57-24.21) 28.36 (27.57-29.17) 0.77 (0.70-0.86) 0.70 (0.62-0.78)
Very strong analgesic 2.90 (2.42-3.48) 3.01 (2.72-3.32) 0.84 (0.63-1.13) 0.83 (0.57-1.22)
NSAID 19.21 (18.01-20.48) 28.26 (27.47-29.06) 0.68 (0.61-0.76) 0.68 (0.61-0.76)
*Adjusted for previous recorded consultations for cardiovascular-related conditions, diabetes and depression, comorbidity count, length of follow-up, year of index date, and consultation frequency.
CI, confidence interval; OR, odds ratio; NSAID, nonsteroidal anti-inflammatory drug.

When stratified into analgesic classifications, the 5-year prevalence of basic analgesics was highest for the dementia cohort and matched cohort (63.5% and 62.1%, respectively), followed by weak analgesics, strong analgesics, NSAIDs, moderate analgesics, and finally, very strong opioids (2.9% and 3.0%, respectively). The dementia cohort and matched cohort had a similar 5-year prevalence of basic analgesic prescription. However, the dementia cohort had a lower 5-year prevalence and odds of being prescribed a weak analgesic, moderate analgesic, strong analgesic, or NSAID compared with the matched cohort (Table 3).

3.3.2. Annual prevalence of analgesic prescription

The annual prevalence of analgesic prescription remained relatively stable for the dementia cohort and matched cohort from index date to 5 years (Table 4). This reflected the multivariable logistic regression models (first year of follow-up: adjusted OR 0.96, 95% CI 0.93-0.99; final year of follow-up: adjusted OR 0.89, 95% CI 0.83-0.97).

Table 4 - Annual prevalence and odds of analgesic prescriptions for the dementia cohort and matched cohort.
0-1 1-2 2-3 3-4 4-5
Any analgesic
 Dementia cohort prevalence % (95% CI) 51.73 (51.10-52.36) 50.55 (49.78-51.32) 49.20 (48.23-50.16) 48.64 (47.42-49.86) 49.58 (48.01-51.15)
 Matched cohort prevalence % (95% CI) 54.18 (53.62-54.74) 54.41 (53.79-55.03) 54.55 (53.85-55.24) 55.14 (54.36-55.91) 55.73 (54.85-56.61)
 OR (95% CI) 0.97 (0.94-0.99) 0.94 (0.92-0.98) 0.94 (0.90-0.98) 0.91 (0.86-0.96) 0.91 (0.85-0.98)
 Adj OR* (95% CI) 0.96 (0.93-0.99) 0.93 (0.90-0.97) 0.92 (0.88-0.96) 0.90 (0.85-0.96) 0.89 (0.83-0.97)
Basic analgesic
 Dementia cohort prevalence % (95% CI) 37.30 (36.69-37.91) 36.86 (36.12-37.61) 36.64 (35.71-37.57) 36.22 (35.06-37.40) 36.58 (35.08-38.10)
 Matched cohort prevalence % (95% CI) 35.78 (35.24-36.32) 36.19 (35.60-36.79) 36.82 (36.15-37.50) 37.57 (36.81-38.33) 38.91 (38.05-39.78)
 OR (95% CI) 1.06 (1.03-1.09) 1.05 (1.01-1.09) 1.05 (1.00-1.11) 1.03 (0.96-1.10) 0.97 (0.89-1.05)
 Adj OR* (95% CI) 1.05 (1.02-1.09) 1.03 (0.98-1.08) 1.04 (0.98-1.11) 1.01 (0.93-1.09) 0.95 (0.87-1.05)
Weak analgesic
 Dementia cohort prevalence % (95% CI) 13.61 (13.18-14.05) 12.68 (12.17-13.20) 12.19 (11.57-12.83) 11.28 (10.53-12.07) 10.97 (10.03-11.99)
 Matched cohort prevalence % (95% CI) 15.49 (15.09-15.90) 15.58 (15.13-16.04) 15.46 (14.96-15.97) 16.18 (15.61-16.76) 16.49 (15.84-17.15)
 OR (95% CI) 0.91 (0.86-0.95) 0.87 (0.81-0.92) 0.85 (0.78-0.92) 0.72 (0.64-0.80) 0.72 (0.63-0.84)
 Adj OR* (95% CI) 0.90 (0.85-0.96) 0.87 (0.80-0.93) 0.85 (0.77-0.93) 0.77 (0.68-0.88) 0.73 (0.62-0.85)
Moderate analgesic
 Dementia cohort prevalence % (95% CI) 7.04 (6.73-7.37) 6.32 (5.95-6.71) 6.22 (5.77-6.71) 6.38 (5.81-7.00 6.40 (5.67-7.21)
 Matched cohort prevalence % (95% CI) 8.52 (8.21-8.84) 8.48 (8.14-8.83) 8.29 (7.91-8.68) 8.03 (7.61-8.46) 7.83 (7.37-8.32)
 OR (95% CI) 0.86 (0.80-0.92) 0.81 (0.74-89) 0.84 (0.74-0.94) 0.73 (0.63-0.85) 0.84 (0.69-1.03)
 Adj OR* (95% CI) 0.86 (0.80-0.94) 0.78 (0.70-0.87) 0.82 (0.71-0.94) 0.72 (0.60-0.85) 0.77 (0.61-0.97)
Strong analgesic
 Dementia cohort prevalence % (95% CI) 11.12 (10.73-11.52) 10.42 (9.95-10.90) 9.71 (9.16-10.30) 9.86 (9.16-10.61) 9.84 (8.94-10.81)
 Matched cohort prevalence % (95% CI) 13.06 (12.68-13.44) 13.31 (12.90-13.74) 13.28 (12.81-13.75) 13.55 (13.02-14.09) 13.69 (13.09-14.30)
 OR (95% CI) 0.84 (0.80-0.89) 0.76 (0.71-0.81) 0.73 (0.67-0.80) 0.72 (0.64-0.81) 0.71 (0.61-83)
 Adj OR* (95% CI) 0.77 (0.72-0.82) 0.71 (0.65-0.77) 0.64 (0.58-0.72) 0.65 (0.57-0.74) 0.61 (0.51-0.73)
Very strong analgesic
 Dementia cohort prevalence % (95% CI) 1.35 (1.21-1.50) 1.40 (1.23-1.59) 1.41 (1.20-1.65) 1.17 (0.94-1.47) 1.54 (1.20-1.98)
 Matched cohort prevalence % (95% CI) 1.30 (1.18-1.43) 1.42 (1.28-1.57) 1.39 (1.23-1.56) 1.33 (1.17-1.53) 1.40 (1.21-1.63)
 OR (95% CI) 1.00 (0.85-1.17) 1.03 (0.84-1.26) 1.14 (0.88-1.47) 0.83 (0.60-1.16) 0.93 (0.61-1.43)
 Adj OR* (95% CI) 0.78 (0.63-0.97) 0.91 (0.70-1.18) 0.94 (0.66-1.34) 0.75 (0.49-1.13) 1.03 (0.55-1.94)
 Dementia cohort prevalence % (95% CI) 6.97 (6.66-7.30) 6.29 (5.92-6.67) 5.51 (5.08-5.96) 5.08 (4.57-5.65) 5.68 (4.99-6.45)
 Matched cohort prevalence % (95% CI) 11.09 (10.74-11.44) 10.62 (10.24-11.01) 10.17 (9.76-10.60) 9.75 (9.29-10.22) 9.53 (9.02-10.06)
 OR (95% CI) 0.64 (0.60-0.68) 0.56 (0.54-0.64) 0.54 (0.48-0.60) 0.54 (0.46-0.62) 0.64 (0.53-0.78)
 Adj OR* (95% CI) 0.66 (0.62-0.71) 0.59 (0.54-0.65) 0.54 (0.48-0.61) 0.57 (0.48-0.67) 0.61 (0.80-0.76)
Time point 0 = index date.
*Adjusted for previous recorded consultations for cardiovascular-related conditions, diabetes and depression, comorbidity count, length of follow-up, year of index date, and consultation frequency.
CI, confidence interval; OR, odds ratio; NSAID, nonsteroidal anti-inflammatory drug.

The annual prevalence of analgesic prescription was also stratified into analgesic classifications. The dementia cohort and matched cohort had a similar prevalence and odds of basic analgesic prescription, irrespective of the annual period (Table 4). Similarly, the annual prevalence of very strong analgesic prescription was similar for the dementia cohort and matched cohort throughout follow-up, with the wide CIs indicating no significant difference (year 1 adjusted OR 0.78, 95% CI, 0.63-0.97; year 5 adjusted OR 1.03, 95% CI 0.55-1.94).

Conversely, the dementia cohort had a lower annual prevalence and odds of weak analgesic, moderate analgesic, strong analgesic, and NSAID prescription compared with the matched cohort. Importantly, the prevalence of these analgesic prescriptions steadily lowered from index date throughout follow-up for the dementia cohort.

4. Discussion

This study of more than 70,000 patients in primary care found that people with dementia had a consistently lower prevalence of musculoskeletal consultation than older adults without dementia and this discrepancy increased over time. This study also found people with dementia were recorded to have lower evidence of weak, moderate, strong, and NSAID prescriptions compared with older adults without dementia, again with the discrepancy increasing throughout follow-up. The following sections will explore these findings further, with reflection on previous literature.

Our findings add to the literature by demonstrating that people with dementia had a lower prevalence of recorded musculoskeletal consultation than older adults without dementia, and this is consistent with research conducted within care home settings.40 Furthermore, the longitudinal perspective of this study enabled investigation into the prevalence of musculoskeletal consultations over time, with findings showing a decreased prevalence throughout follow-up for people with dementia. This finding complements previous research that found that pain identification and assessment decreased with increased cognitive impairment43 and decreased ability to provide a self-report of pain,35 using an assumption that the 5-year period in this study would represent a sufficient period to represent dementia disease progression.

Similarly, for analgesics, patterns were found that differed between people with and without dementia that concur with previous literature. For example, this study found that the prevalence of basic analgesic prescription was similar for people with dementia and older adults complementing findings of other studies that found people with dementia were prescribed similar6,50 or higher rates26,34,49 of paracetamol use than older adults without dementia.13 Such findings may reflect the recommendation that paracetamol should be used as the first-line treatment for persistent pain because of the good side-effect profile.1 These recommendations are, however, in discordance with NICE Chronic pain: assessment and management guidelines (currently in development) that recommend that paracetamol should not be given for chronic pain management.41 This study found that prevalence of weak, moderate, and strong analgesic prescription was lower for people with dementia. This reflects the findings of a recent meta-analysis that reports people with cognitive impairment may use less opioids than people without cognitive impairment25 and also evidence on the limitations of opioids to treat persistent pain in older adults1,39 and particularly concerns associated with opioid use in people with dementia.17,23,37 Although the patterns of weak to strong analgesic prescription were generally lower for people with dementia, the findings on very strong analgesic were less reliable with wider CIs (adjusted OR 0.83, 95% CI 0.57-1.22). This may be reflective of the rarity of very strong analgesic prescription in primary care that may have reduced the precision of the estimates but also may be because very strong opioid prescriptions (such as for morphine or oxycodone) would be warranted when the pain source is perhaps more severe and more easily identifiable (eg, an acute injury). Finally, the 5-year prevalence of NSAID prescription was 9% lower for people with dementia than older adults without dementia (19.2% vs 28.3%, respectively), reflecting previous literature.47,50 In addition, this study found a steadily lowering prevalence of NSAID prescription throughout follow-up. This finding is supported by a number of studies that also found a decreasing prevalence of NSAID use throughout the course of dementia6,24 all of which reflect numerous guidelines recommending that NSAIDs should be considered rarely for older adults and only if safer therapies (eg, paracetamol) have failed to relieve pain because of potential side effects.1,39

4.1. Strengths and limitations

A key strength of this study is the use of CPRD because this database is broadly representative of the UK primary care population28 with demonstrated validity in coding practice.33 Furthermore, this study has used established and validated Read codes and prescription lists to identify the cohort (dementia), outcome, and covariates.8,16,38 The use of electronic health records such as CPRD confers additional advantages because they record actual events and are not subject to selection or reporting bias associated with survey-based designs that can have added issues when applied to the collection of data from people with dementia.52 In addition, results were adjusted for overall consultation frequency for each participant that accounts for increased likelihood of coding because of increased presentation (as consultations may be elevated because of dementia).

There are, however, several limitations associated with this study. Advances have been made to improve the detection of dementia within primary care; however, there may have been patients within the matched cohort with undiagnosed dementia,32 and this potential misclassification may mean that the strength of the associations with pain outcomes is underestimated. Furthermore, although electronic health data captures a wealth of information it cannot capture unrecorded health-related information, eg, self-managed pain and use of over-the-counter analgesics, therefore prevalence estimates of analgesic prescriptions (especially basic analgesics) in this study are likely to be underestimated. Similarly, this study did not directly examine the association between dementia severity and pain outcomes because cognitive evaluations (eg, General Practitioner Assessment of Cognition score) are inconsistently recorded in electronic health record data.20 This is an important limitation because people with moderate or severe dementia are less likely than people with mild dementia to report their pain or have their pain identified by others.43 This limitation was mitigated by the 5-year follow-up to reflect expected progression of dementia and worsening of symptoms.24

In the longitudinal analysis, this research aimed to describe the prevalence of musculoskeletal consultation and analgesic prescriptions over time from index date, in line with expected dementia progression. However, research and UK policy have highlighted the untimely coding of a dementia diagnosis in primary care records, with a small number of patients only receiving a recording of dementia during the later and more severe stages.10 The untimely diagnosis of dementia in primary care may implicate the accuracy of longitudinal temporal analysis starting from an index date representing dementia diagnosis.

This study conducted a descriptive comparison of analgesic prescription between a dementia cohort and matched cohort by examining evidence of analgesic prescription in each period. Identifying “evidence of” analgesic prescription, rather than examining analgesic count may have underestimated the difference in analgesic prescription between people with and without dementia. Future studies should build upon this evidence by examining the number of pain prescriptions for people with dementia compared with a matched cohort.

Finally, differences in the reported effects between cohorts may in part be influenced by immortal time bias. In essence, “unhealthier” patients in the dementia cohort may have left the study (eg, because of death or transfer to a care home), leaving a healthier cohort of people with dementia (ie, less likely to have pain). Inspection of the median time in study showed significant differences between the cohorts (much less time for the dementia cohort), and this would be expected because selection of people with dementia will invariably include a higher level of mortality. However, checks in data show that the dementia cohort and the matched cohort that remained in the study during the last annual period from index date (year 4-5) were similar in regard to a range of baseline characteristics as people included in the first year after index date (see appendix A, available as supplemental digital content at In addition, multivariable analyses adjusted for “length of follow-up” to negate the impact of immortal time bias on the results. To fully understand this issue future studies should include time-to-event regression techniques (eg, survival analysis with matched censoring) to give estimation of the potential for immortal time bias.

There are potential clinical implications from these findings. Although analgesics may be over prescribed relative to other forms of treatment in older people,36 such prescribing is another marker (just as musculoskeletal consultation is) of the level of consideration and awareness of attending clinicians to the problem of pain.41 Taking this to be the case, then this study provides additional evidence that less consideration may be paid to pain in people with dementia, and so it seems even less attention as dementia progresses. Clearly symptoms of dementia create additional challenges for clinicians in the assessment and treatment of pain;13,14 however, there is also evidence of tangible benefits of effective management of pain such as tackling potential unmet needs directly related to pain as well as conferring benefits beyond pain.21

5. Conclusion

This study is the first to examine potential differences in pain assessment and treatment prevalence in people with and without dementia at a primary care population level. The results confirm findings of previous studies focused on care home settings that show similar trends of lower rates of pain assessment and treatment. The evidence suggests a need to understand more about practical methods to increase awareness of pain and use better methods of pain assessment, evaluation of treatment response, and acceptable and effective management for people with dementia, in primary care.

Conflict of interest statement

The authors have no conflicts of interest to declare.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at


This study is based in part on data from the Clinical Practice Research Datalink obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. The data are provided by patients and collected by the NHS as part of their care and support. The interpretation and conclusions contained in this study are those of the author(s) alone. Copyright [2021], reused with the permission of The Health and Social Care Information Centre. All rights reserved. The CPRD Independent Scientific Advisory committee (reference number ISAC 17_240RA) approved this study.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. This work however formed part of a PhD project that received funding from Keele University ACORN studentship. CCG is part funded by the NIHR Applied Research Collaboration (ARC) West Midlands.


[1]. Abdulla A, Bone M, Adams N, Elliott AM, Jones D, Knaggs R, Martin D, Sampson EL, Schofield P. Evidence-based clinical practice guidelines on management of pain in older people. Age Ageing 2013;42:151–3.
[2]. Alzheimer's Society. Support. Stay. Save. Care and support of people with dementia in their own homes. 2011:1–76.
[3]. Appleyard T, Ashworth J, Bedson J, Yu D, Peat G. Trends in gabapentinoid prescribing in patients with osteoarthritis: a United Kingdom national cohort study in primary care. Osteoarthr Cartil 2019;27:1437–44.
[4]. Balestreri L, Grossberg A, Grossberg GT. Behavioral and psychological symptoms of dementia as a risk factor for nursing home placement. Int Psychogeriatrics 2000;12:59–62.
[5]. Balfour JE, O'Rourke N. Older adults with Alzheimer disease, comorbid arthritis and prescription of psychotropic medications. Pain Res Manag 2003;8:198–204.
[6]. Bauer U, Pitzer S, Schreier MM, Osterbrink J, Alzner R, Iglseder B. Pain treatment for nursing home residents differs according to cognitive state—a cross-sectional study. BMC Geriatr 2016;16:1–10.
[7]. Bedson J, Belcher J, Martino OI, Ndlovu M, Rathod T, Walters K, Dunn KM, Jordan KP. The effectiveness of national guidance in changing analgesic prescribing in primary care from 2002 to 2009: an observational database study. Eur J Pain 2013;17:434–43.
[8]. Bedson J, Chen Y, Ashworth J, Hayward RA, Walters K, Dunn KM, Jordan KP. Trends in long-term opioid prescribing in primary care patients with musculoskeletal conditions. PAIN 2016;157:1.
[9]. Benson T. The history of the Read codes: the inaugural James Read Memorial Lecture 2011. Inform Prim Care 2011;19:173–82.
[10]. Bradford A, Kunik ME, Schulz P, Williams SP, Singh H. Missed and delayed diagnosis of dementia in primary care: prevalence and contributing factors. Alzheimer Dis Assoc Disord 2009;23:306–14.
[11]. Bränsvik V, Granvik E, Minthon L, Nordström P, Nägga K. Mortality in patients with behavioural and psychological symptoms of dementia: a registry-based study. Aging Ment Heal 2020;18:1–9.
[12]. Briggs AM, Cross MJ, Hoy DG, Sànchez-Riera L, Blyth FM, Woolf AD, March L. Musculoskeletal health conditions represent a global threat to healthy aging: a report for the 2015 World health organization world report on ageing and health. Gerontologist 2016;56:S243–55.
[13]. Bullock L, Bedson J, Jordan JL, Bartlam B, Chew-Graham CA, Campbell P. Pain assessment and pain treatment for community-dwelling people with dementia: a systematic review and narrative synthesis. Int J Geriatr Psychiatry 2019;34:807–21.
[14]. Bullock L, Chew-Graham CA, Bedson J, Bartlam B, Campbell P. The challenge of pain identification, assessment, and management in people with dementia: a qualitative study. Br J Gen Pract Open 2020;4:bjgpopen20X101040.
[15]. Burton C, Campbell P, Jordan K, Strauss V, Mallen C. The association of anxiety and depression with future dementia diagnosis: a case-control study in primary care. Fam Pract 2013;30:25–30.
[16]. Campbell P, Rathod-Mistry T, Marshall M, Bailey J, Chew-Graham CA, Croft P, Frisher M, Hayward R, Negi R, Singh S, Tantalo-Baker S, Tarafdar S, Babatunde OO, Robinson L, Sumathipala A, Thein N, Walters K, Weich S, Jordan KP. Markers of dementia-related health in primary care electronic health records Markers of dementia-related health in primary care electronic health records. Aging Ment Heal 2020;24:1–11.
[17]. Chang E, Daly J, Johnson A, Harrison K, Easterbrook S, Bidewell J, Stewart H, Noel M, Hancock K. Challenges for professional care of advanced dementia. Int J Nurs Pract 2009;15:41–7.
[18]. Cohen-Mansfield J, Dakheel-Ali M, Marx MS, Thein K, Regier NG. Which unmet needs contribute to behavior problems in persons with advanced dementia? Psychiatry Res 2015;228:59–64.
[19]. Cook AKR, Niven CA, Downs MG. Assessing the pain of people with cognitive impairment. Int J Geriatr Psychiatry 1999;14:421–5.
[20]. Dell'Agnello G, Desai U, Kirson NY, Wen J, Meiselbach MK, Reed CC, Belger M, Lenox-Smith A, Martinez C, Rasmussen J. Reliability of coded data to identify earliest indications of cognitive decline, cognitive evaluation and Alzheimer's disease diagnosis: a pilot study in England. BMJ Open 2018;8:19684.
[21]. Flo E, Gulla C, Husebo BS. Effective pain management in patients with dementia: benefits beyond pain?. Drugs and Aging 2014;31:863–71.
[22]. Gallagher A, Puri S, van Staa T, Staa T. Linkage of the General Practice Research Database (GPRD) with other data sources. 2011. Available at: Accessed February 7, 2020.
[23]. Geddis-Regan AR, Stewart M, Wassall RR. Orofacial pain assessment and management for patients with dementia: a meta-ethnography. J Oral Rehabil 2019;46:189–99.
[24]. Gilmartin JFM, Väätäinen S, Törmälehto S, Bell JS, Lönnroos E, Salo L, Hallikainen I, Martikainen J, Koivisto AM, Soininen H, Von Zu Fraunberg M, Hallikainen M, Hiltunen A, Hongisto K, Hänninen T, Kalinen M, Karppi P, Tikkanen M, Välimäki T, Mäkelä H. Depressive symptoms are associated with analgesic use in people with Alzheimer's disease: kuopio ALSOVA study. PLoS One 2015;10:e0117926.
[25]. Griffioen C, Husebo BS, Flo E, Caljouw MAA, Achterberg WP. Opioid prescription use in nursing home residents with advanced dementia. Pain Med 2019;20:50–7.
[26]. Haasum Y, Fastbom J, Fratiglioni L, Kåreholt I, Johnell K. Pain treatment in elderly persons with and without dementia: a population-based study of institutionalized and home-dwelling elderly. Drugs and Aging 2011;28:283–93.
[27]. Hadjistavropoulos T, Herr K, Prkachin KM, Craig KD, Gibson SJ, Lukas A, Smith JH. Pain assessment in elderly adults with dementia. Lancet Neurol 2014;13:1216–27.
[28]. Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, Smeeth L. Data resource profile: clinical practice research Datalink (CPRD). Int J Epidemiol 2015;44:827–36.
[29]. Hunt LJ, Ritchie CS, Cataldo JK, Patel K, Stephens CE, Smith AK. Pain and emergency department use in the last month of life among older adults with dementia. J Pain Symptom Manage 2018;56:871–7.e7.
[30]. Jennings AA, Linehan M, Foley T. The knowledge and attitudes of general practitioners to the assessment and management of pain in people with dementia. BMC Fam Pract 2018;19:166.
[31]. Jordan KP, Kadam UT, Hayward R, Porcheret M, Young C, Croft P. Annual consultation prevalence of regional musculoskeletal problems in primary care: an observational study. BMC Musculoskelet Disord 2010;11:144.
[32]. Lang L, Clifford A, Wei L, Zhang D, Leung D, Augustine G, Danat IM, Zhou W, Copeland JR, Anstey KJ, Chen R. Prevalence and determinants of undetected dementia in the community: a systematic literature review and a meta-analysis. BMJ Open 2017;7:11146.
[33]. Lawson DH, Sherman V, Hollowell J. The general practice research database. 1998. Available at: Accessed July 4, 2019.
[34]. Lövheim H, Karlsson S, Gustafson Y. The use of central nervous system drugs and analgesics among very old people with and without dementia. Pharmacoepidemiol Drug Saf 2008;17:912–18.
[35]. Lukas A, Mayer B, Fialová D, Topinkova E, Gindin J, Onder G, Bernabei R, Nikolaus T, Denkinger MD. Treatment of pain in European nursing homes: results from the services and health for elderly in long TERm care (SHELTER) study. J Am Med Dir Assoc 2013;14:821–31.
[36]. MacFarlane GJ, Beasley M, Jones EA, Prescott GJ, Docking R, Keeley P, McBeth J, Jones GT. The prevalence and management of low back pain across adulthood: results from a population-based cross-sectional study (the MUSICIAN study). PAIN 2012;153:27–32.
[37]. Manias E. Complexities of pain assessment and management in hospitalised older people: a qualitative observation and interview study. Int J Nurs Stud 2012;49:1243–54.
[38]. McGuinness LA, Warren-Gash C, Moorhouse LR, Thomas SL. The validity of dementia diagnoses in routinely collected electronic health records in the United Kingdom: a systematic review. Pharmacoepidemiol Drug Saf 2019;28:244–55.
[39]. McLachlan AJ, Bath S, Naganathan V, Hilmer SN, Le Couteur DG, Gibson SJ, Blyth FM. Clinical pharmacology of analgesic medicines in older people: impact of frailty and cognitive impairment. Br J Clin Pharmacol 2011;71:351–64.
[40]. Nakashima T, Young Y, Hsu WH. Do nursing home residents with dementia receive pain interventions? Am J Alzheimers Dis Other Demen 2019;34:193–8.
[41]. National Institute for Health and CareExcellence. Chronic pain in over 16s: assessment and management: Draft for consultation. 2020. Available at: Accessed August 28, 2020.
[42]. Perkins AJ, Kroenke K, Unützer J, Katon W, Williams JW, Hope C, Callahan CM. Common comorbidity scales were similar in their ability to predict health care costs and mortality. J Clin Epidemiol 2004;57:1040–8.
[43]. Reynolds KS, Hanson LC, DeVellis RF, Henderson M, Steinhauser KE. Disparities in pain management between cognitively intact and cognitively impaired nursing home residents. J Pain Symptom Manage 2008;35:388–96.
[44]. Richardson E, Bedson J, Chen Y, Lacey R, Dunn KM. Increased risk of reproductive dysfunction in women prescribed long-term opioids for musculoskeletal pain: a matched cohort study in the Clinical Practice Research Datalink. Eur J Pain 2018;22:1701–8.
[45]. Schofield P. The assessment of pain in older people: UK national guidelines. Age Ageing 2018;47:i1–i22.
[46]. Shega JW, Hougham GW, Stocking CB, Cox-Hayley D, Sachs GA. Pain in community-dwelling persons with dementia: frequency, intensity, and congruence between patient and caregiver report. J Pain Symptom Manage 2004;28:585–92.
[47]. De Souto Barreto P, Lapeyre-Mestre M, Vellas B, Rolland Y. Potential underuse of analgesics for recognized pain in nursing home residents with dementia: a cross-sectional study. PAIN 2013;154:2427–31.
[48]. Stevens S, Bankhead C, Mukhtar T, Perera-Salazar R, Holt TA, Salisbury C, Hobbs FDR; NIHR School for Primary Care Research, Nuffield Department of Primary Care Health Sciences, University of Oxford. Patient-level and practice-level factors associated with consultation duration: a cross-sectional analysis of over one million consultations in English primary care. BMJ Open 2017;7:e018261.
[49]. Tan E, Jokanovic N, Koponen M, Thomas D, Hilmer S, Bell J. Prevalence of analgesic use and pain in people with and without dementia or cognitive impairment in aged care facilities: a systematic review and meta-analysis. Curr Clin Pharmacol 2015;10:194–203.
[50]. Tan EC, Visvanathan R, Hilmer SN, Vitry A, Emery T, Robson L, Pitkälä K, Ilomäki J, Bell JS. Analgesic use and pain in residents with and without dementia in aged care facilities: a cross-sectional study. Australas J Ageing 2016;35:180–7.
[51]. Walker J, Liddle J, Jordan KP, Campbell P. Affective concordance in couples: a cross-sectional analysis of depression and anxiety consultations within a population of 13,507 couples in primary care. BMC Psychiatry 2017;17:1–10.
[52]. Weuve J, Proust-Lima C, Power MC, Gross AL, Hofer SM, Thiébaut R, Chêne G, Glymour MM, Dufouil C. Guidelines for reporting methodological challenges and evaluating potential bias in dementia research. Alzheimer's Dement 2015;11:1098–109.

Pain; Analgesia; Musculoskeletal; Cohort; Dementia; Primary care

Supplemental Digital Content

© 2021 International Association for the Study of Pain