Pain treatment for nursing home residents differs according to cognitive state – a cross-sectional study

Pain is common in older people [1, 2], particularly in nursing home (NH) residents [3, 4] and those with cognitive impairment [5, 6]. Old age exposes individuals to different types of pain, most commonly related to musculoskeletal, gastrointestinal, neurological and cardiac conditions, genitourinary infections, injuries, as well as pressure ulcers in bed ridden individuals, with particularly high pain prevalence rates in those suffering from cognitive impairment [6, 7]. One recently published investigation based on health insurance claims data reported no difference in the diagnoses indicating pain between participants with incident cognitive impairment and cognitively fit controls [8].

Since cognitive impairment is common in many nursing home residents [9, 10], assessment and management of pain is particularly demanding for physicians and nurses due to ambiguity in communication leading to the reasonable assumption that pain in persons with cognitive decline is both under-diagnosed and under-treated [11‐13]. In addition, use of analgesics has been reported with higher quality of life in individuals with cognitive impairment [14]. There is evidence that persons with advanced cognitive decline either receive pain treatment, notably opioids, less frequently or in lower insufficient doses as compared to their cognitively fit counterparts [12, 15‐19], whereas only few studies have reported a possible overuse of analgesics, particularly paracetamol, in patients with cognitive impairment [20]. On the other hand, there is lack of information to which extent the severity of cognitive decline affects the use of other kind of medication, i.e. antipsychotics, since pain is not only a frequent cause of behavioral and psychological symptoms in dementia (BPSD) [7, 21], but may also, among others, stem from the under-treatment of pain [7, 22]. Pain-induced disruptions run the risk of being misinterpreted as BPSD, provoking inappropriate prescription of psychotropic drugs (i.e. antipsychotics) which, in turn, have been associated with compromised cognition, falls and fractures and increased risk of death [23]. As mentioned above, there is broad consensus within the literature that the challenge of accurately identifying pain in cognitively impaired individuals is the paramount cause of sub-optimal management of pain [24]. Hence in 2009, the American Geriatric Society recommended a comprehensive, disease-specific assessment to establish adequate pain management on an individual level [25]. While individuals with mild to moderate cognitive impairment are often able to report pain either verbally or by use of rating scales [3, 26], these options are not applicable for those with advanced cognitive impairment when the ability to communicate is severely impaired. Thus, self-reported pain may not always be reliable in people with advanced cognitive impairment and pain should be indirectly delineated by raters using a validated observational instrument [27]. Various numerical and visual scales are available for self-reported experience of pain, all of them lacking soundness in persons with cognitive impairment due to their subjection on memory, abstract thinking and speech comprehension [28].

Primary aim of our study was the comparison of assessment and pharmacological treatment of pain in Austrian NH residents in relation to their cognitive state based on self-report and observational assessment of prevalence and intensity of pain, generating the following research questions:

Since BPSD may be a consequence of undertreated pain, we additionally asked:

The presented cross-sectional data were collected as part of a baseline investigation of a non-experimental pre-post study with semi-standardized interventions for optimizing pain management in nursing homes (NH) in Austria (OSiA = Optimiertes Schmerzmanagement in Altenpflegeheimen, German for: Optimized pain management in nursing homes). Residents’ pain prevalence in relation to pain assessment and cognitive state was assessed and data on medical diagnoses and prescriptions were retrieved from the nursing homes’ documentation (n = 425). The study was approved by the ethical committee of Salzburg (415-E/1412/4-2011 v. 07.10.2011). A written consent was obtained from the nursing home residents or from their legal representatives.

Characteristics of residents are summarized in Table 2. The gender distribution was equal between the three groups. CI-NV residents were significantly older than the others. CI-V and CI-NV residents had a higher classification in the level of care than CUS residents. There were no differences in the length of stay in the NH. CUS residents had a significantly higher number of medical diagnoses than their CI-counterparts, but fewer neurological diagnoses (i.e. dementia). Orthopedic diagnoses were significantly higher in group CUS than in group CI-V. Group CI-NV had significantly less prescriptions scheduled permanently (PP) than the other two groups.

Our data confirm that pain is still a frequent symptom in Austrian NH residents. More than two thirds either self-report pain (VRS) or indicate prevalence of pain on the PAINAID-Gm observational assessment. These findings are in line with previously published data [37, 40].

In addition, reports of residents on their pain varied depending on cognitive function and pain-assessment instruments. Assessing moderately to severely cognitively impaired individuals’ pain under standardized conditions resulted in a high frequency of pain (approximately 80 %) compared to the literature [5, 6, 41]. Since CI-V residents’ observational assessment showed significantly higher pain prevalence than their self-report, the true ratio of CI-V residents in pain is at issue. Against this background and considering communication deficiencies in this group, the possibility of higher pain prevalence than indicated through self-report should be considered. By assuming the presence of pain, if at least one of both instruments indicates pain – as was done in this study – sensitivity is aimed for, however, specificity of pain detection may be lowered.

In contrast to literature [12], our results demonstrate differences in pain-associated diagnoses relating to cognitive state with significantly more diagnoses indicating pain in the CUS group in comparison to the CI-V group. This discrepancy cannot be interpreted from our data, but may reflect a decreasing awareness in determining medical diagnoses for the cognition-groups with advanced decline.

Our data adds to the evidence that persons with advanced cognitive decline less often receive pain treatment [12, 15‐19] and, is therefore, conflicting with reports from other European countries suggesting that analgesic use is higher among people with cognitive impairment as compared to older adults without cognitive impairment [8, 20, 42]. This may be explained by differences in national prescription habits and configuration of medical supply. It has to be pointed out though that frequency of analgesic prescription is not necessarily a sign, whether (or not) an adequate analgesic treatment is being prescribed to the right people at the right time [7].

It is noteworthy to add that we investigated the percentages of permanently scheduled analgesics (PP) and analgesics prescribed ‘as needed’ (PRN) in relation to the different cognition-groups. In all groups, more analgesics were prescribed PRN (64.1 – 72.3 %) than PP (35.9 – 58.0 %), with no significant group differences for the PRN prescriptions. In contrast, PP prescriptions dropped with cognitive decline, with 58.0 % for group CUS, 48.7 % for group CI-V and 36.0% for group CI-NV. The difference between groups CUS and CI-NV reached significance suggesting that advanced stages of cognitive decline are associated with a reserved attitude to medical treatment. This is also supported by the mean numbers of prescriptions, which exhibit a decline from group CUS (9.7 ± 3.9) to groups CI-V (8.8 ± 3.9) and CI-NV (6.8 ± 3.2), again reaching significance between CUS and CI-NV.

According to the literature, pain medication in individuals with cognitive impairment is generally of low dosage and stronger drugs such as opioids are less likely to be considered [7]. In our sample, the most frequently prescribed PP pharmacological subgroups were opioids with 50.2 %, followed by topical analgesics with 47.5 %, weak cyclooxygenase inhibitors (Metamizol, Paracetamol) with 32.3 % and NSAIDs with 26.7 %. No differences between the cognition-groups were detectable. Notably, the number of opioid prescriptions is remarkably high in comparison to recent literature [15, 19], as is the overall number of PP in our study. The high number of opioid prescriptions is in accordance with other investigations reporting an increasing opioid use in several countries [43].

Merging PP and PRN prescriptions, the use of NSAIDs diminishes significantly with the loss of cognitive function, showing a prevalence of 44.4 % in group CUS compared to 28.9 and 18.4 % in groups CI-V and CI-NV, respectively. This trend also partially applies to PRN prescriptions. The lower NSAID prescription behavior as cognition declines could, however, be also due to the recommendation to prescribe this class of medication less frequently due to its detrimental side-effects on particularly the geriatric population [44‐47].

Cognitively impaired individuals seem to experience intensity and affective components of pain differently than their cognitively fit counterparts [7]. Moreover, the decline in the ability to communicate results in considerable challenges for detecting pain, particularly in advanced stages of cognitive impairment. Cognitively impaired individuals may embody pain by BPSD, such as agitation, denial or withdrawal, giving rise to misinterpreting pain as a psychiatric condition and resulting in the inappropriate treatment with antipsychotic medication [7, 21].

In our sample, prescription of antipsychotics was strikingly common, as reported previously for Austrian NH residents by Richter (2012) [48]. The high number of antipsychotic prescriptions is likely to be an indicator for a perceived or actual lack of strategies to manage BPSD [48]. Antipsychotics were prescribed in nearly half of the participants of our study with the difference between CI-V (47.0 %) and CUS (37.0 %) residents demonstrating a trend and the difference between CI-NV (51.6 %) and CUS (37.0 %) residents reaching statistical significance. In addition, residents with advanced cognitive decline had a significantly higher risk of suffering from pain without having an analgesic prescription, giving rise to the assumption that pain-associated BPSD indicate antipsychotic rather than analgesic treatment. This finding adds to an ongoing discussion, since some studies demonstrated that treatment of pain might decrease incidence and severity of BPSD [22, 49], whereas a recently published meta-analysis does not support strong associations between pain and BPSD [50].

To our best knowledge, our investigation is the first of its kind to address the relationship between prevalence of pain, prescription of analgesics and the application of different assessment tools adjusted to the cognitive state of NH residents. Connecting these aspects, CI-V residents had a 2.6-fold higher risk of suffering from pain without having any analgesics prescribed as compared to CUS residents; for CI-NV residents, this risk was even 3.4-fold. These findings suggest cognitive impairment as a hindering factor for receiving sufficient medical pain treatment. Against the background of communication deficiencies, these considerable disparities may be traced back to lacking or inadequate pain detection, which further points toward the necessity of meticulous and standardized pain assessment adapted to the residents’ cognitive state as a prerequisite of adequate treatment. Moreover, such an assessment is mandatory to find the balance between sufficient analgesia, side effects and futile treatment, since patients are often restricted in reporting the effect of therapy [40]. Along these lines, recording the effect of pain treatment over time might be a pivotal measure to warrant optimized pain management and to avoid inadequate medications.

We concede some limitations to our work: Due to the cross-sectional study design, it is difficult to make causal conclusions about the data. Additionally, considering nursing homes in Austria, it should be noted that requirements related to staffing and facility structure are regulated by federal law. Although our sample only consisted of nursing homes from one Austrian nursing home operator, which limits its generalizability, differences in staffing and facility structure between the nursing homes in our sample and other nursing homes are expected to be rather negligible. The quality of the medical documentation in the nursing homes did not allow the use of standardized tools to rate cumulative illness, thus, limiting the comparability of disease burden between the cognition- groups. Moreover, the medical documentation only included information about prescribed medication, but not if the prescribed medication was actually given to or taken by the resident. The classification of cognitive abilities was based on the current assessment of the MMSE. Therefore, definitive assignment to clinical entities or etiology is not warranted. The use of antipsychotic drugs might be biased by the presence of BPSD, which were not recorded systematically in the medical documentation of the nursing homes. Due to different cognitive states of the participants different pain assessment methods were used.

Results point toward an underuse of pain medication in nursing home residents with advanced cognitive decline, especially those unable to communicate pain verbally. The implementation of standardized pain assessments adapted to the cognitive abilities of residents may foster the recognition of pain, warrant optimized pain management, reduce inadequate medication and consequently raise the chance of equally effective pain treatment regardless of cognitive state.

ATC, anatomical therapeutic chemical; BPSD, behavioral and psychological symptoms in dementia; CAPI, computer assisted personal interview; CI, cognitively moderately – severely impaired; CI-NV, cognitively moderately – severely impaired/not verbally communicating; CI-V, cognitively moderately – severely impaired/verbally communicating; COX, cyclooxygenase; CUS, cognitively unimpaired, slightly impaired; MMSE, mini mental status examination; n, sample size; NH, nursing homes; NSAID, nonsteroidal anti-inflammatory drugs; OSiA, optimiertes schmerzmanagement in altenpflegeheimen; PAINAD-Gm, pain assessment in advanced dementia (modified German version); PP, prescriptions scheduled permanently; PRN, prescribed ‘as needed’; RR, risk ratio; VRS, verbal rating scale (pain assessment); WHO, World Health Organization