Pain in Nursing Homes Residents With and Without Cognitive Impairment. Results From Five Consecutive Multicenter Cross-sectional Surveys Carried Out Between 2014 and 2018 in Germany

Background: There is evidence that knowledge about the prevalence of pain and quality of pain management particularly in nursing home residents (NHR) with severe cognitive impairment (CI) is poor. Methods: The multicenter cross-sectional surveys explored the prevalence of pain of NHR with or without CI from nursing homes in Germany. Actual pain intensity in rest and stress were documented. NHR were asked about their daily restrictions due to pain. Data about the pain management were collected and analyzed. Results: A total of 3437 residents were interviewed with respect to feeling pain, including one third each with mild and severe CI. The prevalence of actual pain was 31,8%. Women reported pain more often. Prevalence of NHR without CI or with self-report on pain was signicantly higher than NHR with severe CI or with external report on pain. About 20% of all NHR were dependent on external pain-recording. Nearly 10% of all NHR with pain conrmed pain in stress above 5 on a scale of 0 - 10. 85% of all NHR with pain reported that they had pain for longer than 3 months. Residents with severe CI are 0.55 times less likely to take painkillers than NHR without CI. Conclusion: The study points out a signicant decit in pain management in German NHR with severe CI. Intensive training in pain management for employees in nursing homes is recommended.


Background
Pain in uences everyday life activities of the elderly, especially their mobility [1]. But mobility seems to be the key predictor for many health conditions in the context of care dependency in the nursing home setting [2]. Social contacts are limited and well-being and life satisfaction decrease when people experience chronic pain [3,4,5]. Moreover, pain is an emotional pressure for those suffering as well as a nancial burden for the society [6]. It`s di cult to measure pain. Apart from the subjective perception of those affected, the prevalence rates vary widely depending on the type of survey [7]. An additional challenge is the recording of pain in very old people and in people with severe CI. Pain management is very important in this vulnerable population because: "inability to communicate does not negate the possibility that a human […] experiences pain." [8]. People with severe CI are additionally in risk to receive worse pain-relieving management than people self-reporting on pain [9]. A key role in pain-relieving management plays the nursing staff [10]. They still lack knowledge on pain and possible measurement tools [11,12]. Furthermore there is limited evidence about the reliability and validity of pain assessments regarding people with CI [13]. That is why there are differences in pain assessments, even depending on the staff quali cation [7]. There is also evidence that a systematic pain-assessment of care dependent people is insu cient [14].
The aim of this study is to establish the differences of NHR with pain in Germany with respect to gender and age between those who can report on experiencing pain and those who are not able of doing that.
The research questions which are posed are:

Study design
The present study relies on data collected in the course of an annual representative, multi-center crosssectional survey carried out by the Department of Nursing Science at Charité between 2014 and 2018.

Setting
The research agenda did not aim at investigating a minimum number of participants. The representative character of the results from the prevalence survey of the previous year could be repeatedly proved and therefore a solid representative data state for the period between 2014 and 2018 was demonstrated [15].
The quantity of refusals and the number of participants who for some reason did not take part in the study was also recorded.

Participants
The random samples were drawn from the residents of the participating nursing homes at that time. All NHR were asked to take part in the study. An informed consent by the residents or their legal representative was required [16,17]. The participation was not mandatory but there was only one condition, namely that subjects should be over 16 years.

Variables
Demographic varaibles like age and gender were raised. Furthermore, the NHR were asked about their actual pain. With this question it should be avoided that pain values of the last days or weeks were collected. In case a resident was not capable of providing this information, the nursing staff was asked to estimate the pain intensity at rest and stress and its implications for the daily life.
Both sources of information were recorded, independently of whether given by the NHR or the nursing staff. The NHR in the study were also classi ed according to their cognitive status, namely as having none, mild or severe CI by clinical judgement. The pain intensities at rest and stress conditions were asked by means of a visual analog scale. The scale served as subjective measuring the pain intensity of the NHR.The two end points (0-10) show the range of pain intensity (0 -no pain, 10 the strongest feeling of pain). The German Nursing Expert Standard on Pain and Interdisciplinary Medical Guidelines recommend that treatment should be adjusted to reduce subjective pain to a value of "3" (to 10) for pain at rest and to "5" (to 10) for pain at stress, unless a different target has been agreed with the patient [18,19]. A cumulative score was calculated from the data on pain intensity at rest and stress as a confounder variable for multivariate analyses. The impairment grade due to pain in the daily life was also measured What is the ratio of pain prevalence among NHR in Germany? Are there differences between self and external perceptions of pain, taking into account CI? What pain intensities at rest and stress do NHR report and to what extent are they restricted by pain in the activities of their lives? Does CI of NHR affect the quality of pain treatment? through a graded scale from 0 (no impairment) till 10 (full impairment). The duration of pain was determined by means of the question "How long have you had pain?" with the answer options "acute", "1-3 months", "3-6 months" ">6 months". When determining the frequency of daily pain recording, it was asked whether pain was recorded up to twice a day or more often.

Data Collection
Each nursing home had an employee responsible for the study. As a direct personal assessment, the data collection was carried out by nursing staff of the participating institutions. The training followed with specialized and standardized research instructions. These were uptodate materials containing internationally standardized research de nitions and expressions. The research team could be contacted by telephone at any time. The data collected were registered in machine-readable questionnaires and forwarded to the assessment institution. The questionnaires were scanned and the information was transferred to the statistics program SPSS.

Data analysis
The evaluation of the data took place by the program IMB SPSS 24.0. The evaluation included apart from table and graphic processing of the frequency distribution, also common characteristics regarding the central tendency, dispersion and distribution. Contingency table analyses were used to compare the percentage distributions of the binary target variables differentiated by grouping variables. The inferential statistical test for independence was carried out using chi-square tests (X 2 ). To test a possible confounding, binary logistic regression analyses were calculated for a multivariate analysis of the binary target variables.
The inferential statistical testing of the distribution of (quasi-)metric variables for differences in group comparisons with regard to central tendencies was carried out with the Mann-Whitney-U-Test ('MWU'-Test) in two comparison groups and with the Kruskal-Wallis test ('H' test) in more than two comparison groups, together with a post-hoc analysis using the Dunn-Bonferroni test ('DB' test) to determine which groups differ signi cantly. Spearman correlation coe cients were calculated for the description and inferential statistical testing of the correlation of (quasi-)metric variables. All tests were 2-sided, the signi cance level was set at p < = 0.05.

Results
3483 NHR participated in the surveys from 2014 to 2018. Most of the residents were female (70.7%). The average age was 81.7 (+/-11.9).

Pain prevalence
Information on pain was available for 3437 NHR. The results of the study show that 31.8% of the NHR were affected by actual pain at the time of the survey. Table 1 shows the overall distribution of the pain prevalence classi ed by gender, age, cognitive impairment and the way in which the pain was reported. provided this information as a result of an external evaluation. At 39.3%, the proportion of persons affected by pain was signi cantly higher in the group of NHR who were able to give information than in the group of NHR who were unable to provide information (32.8%). Table 2 shows the result of the logistic regression of pain prevalence, taking into account the variables gender, age, degree of CI and the way in which pain data are determined as in uencing variables. A lower pain intensity was found in NHR whose pain intensity was calculated by the nursing staff compared to NHR who reported information themselves (MWU test: U = -2.303, p < 0.05). Differentiated by gender and age groups, no signi cant differences in the distribution of pain intensity at rest and stress were found. The distributions of pain intensitys at stress, differentiated according to the degree of CI and the type of pain report, also showed no signi cant differences. Information on the duration of existing pain was available from 919 NHR. Details on this will be found in Supplement, Fig. 2 [20], both of these correlations are strong effects. The higher the intensity of pain in rest or stress, the higher the degree of pain-related impairment of everyday life. The signi cance test according to Meng et. al. [21] showed that the two correlation coe cients differ signi cantly (z = 11.49; p < 0;05). This means, that the degree of pain-related impairment of everyday life correlates signi cantly higher with pain intensity at stress than at rest.

Pain management
In 22.8% of all NHR with actual pain, pain was recorded more than twice a day and 82.1% con rmed the use of painkiller. Table 3 shows the results for the use of painkiller and the frequency of pain recording more than twice a day, differentiated by gender, age, cognitive impairment and type of pain recording. The use of painkiller was signi cantly more frequent in women (83.6%) than in men (76.7%). The proportion of NHR taking pain medication was signi cantly higher among NHR without CI compared to NHR with CI. At 83.6%, the proportion of NHR who took painkiller was signi cantly higher among those who were able to report about their pain than those who were unable to do so. There were no signi cant differences between the age groups. The proportion of NHR who had pain assessment more than twice a day was signi cantly higher in residents without CI (26.7%) compared to NHR with mild or severe CI. Differences in NHR with pain assessment more than twice a day indicated by gender, age group and type of pain report are not signi cant. Table 4 shows the logistic regression of the dependent variable ‚taking a painkiller' and Table 5 the logistic regression of the dependent variable ‚frequency of daily pain recording', taking into account the variables gender, age, CI and the summscore of painintensity at rest and stress as in uencing variables  The results of the logistic regression models show that the intensity of pain and the degree of CI signi cantly in uence the intake of a painkiller as well as the frequency of daily pain recording. Increasing the pain intensity score by one unit increased the relative probability of taking a painkiller by 33.5% and of recording pain more than three times a day by 11.4%. At constant pain intensities compared to NHR without CI, for NHR with "severe" CI the relative probability of taking a painkiller decreased by 45.4% and pain recording more than three times a day by 46.7%. There were signi cant differences between women and men with regard to the use of pain medication. The relative probability of taking a painkiller was 63.7% higher for women than for men. The Hosmer-Lemeshow-Test as a test for the goodness of t, show that both models are suitable for adequately representing the data.

Discussion
The examined NHR resembled the general NHR in Germany in terms of age and gender [22]. Overall, almost a third of all NHR of this survey reported actual pain. The multivariate statistical analysis of the data shows gender, age and degree of CI as independent aspects that in uence the pain prevalence rate.
It should be noted that women reported pain more frequently than men and that the frequency of pain increases with age. Similar to the results in many studies, the prevalence of pain in NHR with CI is lower than of NHR without CI [23][24][25][26]. This study also shows that the decisive factor for the lower pain prevalence rate is the diagnosis "severe" CI, regardless of age, gender and the method of pain report. In this survey, the pain prevalence rates of self-reporters on actual pain are similar to comparable settings.
Between 27.8% and 46% were measured in international comparison [27]. 20% of the NHR in this survey were not able to provide information on pain so the nursing staff had to do an evaluation of the pain.
However, the lower pain prevalence in NHR with CI compared to cognitively capable NHR is not based on the different ways in which pain is assessed (self-report or external-report). These ndings suggest that pain management in NHR with CI is a key challenge for both, responding and nonresponding NHR [28]. In particular, it is considerably more di cult to describe the experience of pain in older people with severe CI, e.g. dementia [29].
The results show that about 10% of NHR with actual pain suffered from moderate to severe pain. In more than half of all NHR with pain, the pain existing had lasted for more than 6 months. For about ¼ of all NHR with actual pain, a moderate to severe impairment of their everyday life was the result, which was caused to a greater extent by lack of activities. Pain sensations and impairment in everyday life were signi cantly lower in NHR with CI or external report. These results can suggest that disoriented NHR have poor skills to express their pain. It can also be assumed that pain in this vulnerable group is insu ciently recognized or underestimated, with the risk that the NHR pain expressions will not be used to draw consistent conclusions about appropriate pain therapy as part of a systematic pain management. However for NHR in the context of care that promotes resources and autonomy, systematic pain management is an elementary contribution to the quality of care. NHR with pain have a higher need of care than patients without pain [30]. If pain is not properly monitored and treated, there is an additional risk of pain-related impairment of everyday life. Using the frequency of daily pain assessment and administration of painkillers as indicators for pain management, the study indicates that pain management for NHR with severe CI is inadequately compared to NHR without CI. While the probability of recording pain several times a day and taking a painkiller increases with increasing pain intensity, NHR with severe CI are less likely to take these interventions compared to patients without CI -regardless of the degree of pain intensity. These results suggest that NHR with severe CI are at risk of being undertreated.

Limitations
Pain was one of up to eight survey dimensions in the annual survey on nursing problems in Germany. Thus, pain could not be recorded in its complexity. The representativeness of a sampled prevalence survey is always limited. Contexts can be presented, but no conclusive explanations of causes and effects can be given. The study design required 1:1 interviews for the data collection. However, the study management cannot rule out the possibility that data from the written documentation were also used in isolated cases. The percentage of non-participants may in uence the results of the examination and lead to minor shifts. It cannot be excluded, for example that the proportion of NHR with severe CI in particular could be even higher, or that people with signi cant care problems were speci cally excluded by relatives or the local study coordinators. In comparison with data from the Medical Advisory Service of the German Association of Statutory Health Insurance funds, however, there is only a slight difference in the prevalence rates for pain [31].

Conclusions
Lower pain prevalence rates among NHR with CI should be taken as an indicator that the collection and analysis of pain data is a sensitive area of concern. There is a potential risk of insu cient initiatives being taken to reduce potential pain. In Germany, an important step was taken with the development of the German expert standards on acute and chronic pain [19,33]. These contain essential recommendations for pain management in nursing. The results of the present study indicate that pain by numerous NHR with CI is insu ciently recognized or insu cient measures for pain reduction are initiated. Further studies are needed to address the speci c needs of NHR with severe CI. This demand is reinforced by the frequency of coexistence of pain and dementia and the di culty of pain recognition [34][35][36]. Finally, NHR with severe and moderate CI also have the right to smart pain management [37,38]. Effective pain management helps to reduce aggression and agitation as well as promotes mobility in NHR [39]. Nonetheless, it remains important to record individual pain levels of NHR at rest and stress regularly, sometimes several times a day, especially during drug therapy. When pain medication is administered, the e cacy must be investigated and documented. Continuous training on pain for nursing staff is strongly recommended [40].