- Medicine (St Vincent's) - Theses
Now showing 1 - 1 of 1
ItemPatterns of pain in patients with advanced cancerPHILIP, JENNIFER ( 2000)Extensive literature detailing pain in cancer has been published in the past 15 years with point prevalence studies for populations, effects of interventions and general pain management approaches receiving most attention. The experience of pain over time for individual patients has not been studied. There is value in identifying patterns that pain may follow in order to predict future pain levels for a particular patient. Aims: This prospective longitudinal study of patients with cancer, for whom treatment is palliative, seeks to determine if pain reported by patients follows characteristic patterns. The study took place in the context of a larger epidemiological study of symptoms, quality of life, health service utilization and the quality of information regarding cancer and treatment for patients and their carers. Within this epidemiological study, an observational study of pain intensity, characteristics and analgesic levels was undertaken over a 6 month period. The investigators enrolled 202 participants, with 165 participants completing data for evaluation. Method: Data series were examined as a whole group, then individual participants' time series data were examined for visual and mathematical patterns. A series of mathematical models were fitted to successive pain scores based on the assumption that current pain scores are not independent of previous pain scores. Results: As a group, mean pain levels did not alter over the 6 months, but differences between individuals within the group were marked. Considered as a whole, 20.50/0 reported using paracetamol at assessment, and just 16.3% used slow release morphine. Visual examination of pain levels over time revealed three broad groups of patients: those for whom pain scores were always high, those for whom pain scores were always low, and those who had fluctuating pain scores. Those with consistently high pain scores were more likely to have bone metastases (p <0.05), arthritis (p <0.005), to describe pain as burning (p <0.003), stabbing (p <0.001), present with light touch (p <0.001), than the fluctuating or consistently low pain scores groups. In addition these patients used more paracetamol, codeine, slow release morphine and morphine mixture (p < 0.001), though not nonsteroidal anti-inflammatory drugs. There was no significant difference in gender, treatments planned or survival between the groups. A series of mathematical models were fitted to each individuals’ time series data in an attempt to simply describe patterns within the data. The model that most followed the plotted data points with the least variation of points from that model was a 'cubic' pattern for 95% of patients. This pattern is best described as fluctuating, typically with an initial increase of pain early in the study, followed by some lessening of pain, to again increase more steeply towards the end of the study. The degree to which this model 'fitted' or the mean residual square value for all time series ranged from r2=0.02 - 1.0 (median r2=0.38, mode r2=1.0). For 22 subjects the mean residual square value for cubic modeling was r2=0.8 or greater indicating an excellent correlation between the data and the model. These patients were significantly more likely to be treated with radiotherapy (p <0.01), codeine (p <0.05) and morphine mixture (p <0.05), and had a poorer ECOG performance status (p <0.001) and shorter survival (p <0.05) than the other patients. Therefore a cubic model may assist to predict future pain levels for patients with poor performance status. Clinical Implications: The results of this study have implications for the clinician caring for patients with advanced cancer, enabling forecasting of likely future pain levels for a particular patient. This information will assist when planning analgesic interventions, frequency of consultations and resource allocation for a patient, as well as interpreting information gleaned from analgesic trials. Conclusion: Mean pain scores for a whole population with advanced cancer do not change considerably with time, but for individuals marked fluctuations are apparent. Patients with a poor performance status, arthritis, bone metastases and describing certain pain descriptors are more likely to have consistently high pain levels. For 95% of patients a cubic function model best fitted their time series data, and this model is likely to be most reliable in forecasting pain levels when patients have a poor performance status.