Development of clinical prediction rules for reducing delays in the diagnosis of multiple myeloma

Background Multiple myeloma is a type of blood cancer which starts at the bone marrow. Fifty percent of patients have more than three consultations in primary care before being referred to secondary care. This implies that myeloma patients can experience delays in the diagnosis which could contribute to the relatively poor prognosis. The aim of the thesis was to develop clinical prediction rules that could potentially reduce the time to diagnosis of myeloma patients. Methods The thesis comprises three main studies. The first is a systematic review which maps and quantifies the diagnostic pathway of myeloma patients. In the second study, the features of myeloma (symptoms and blood test abnormalities) were examined to identify which present the earliest. The most optimal combinations of blood tests for diagnosing myeloma were also examined. In the third study, a series of clinical prediction rules were developed and validated using the Clinical Research Datalink data that could be used to expedite the diagnosis of myeloma. Results Myeloma patients experience substantial delays with 50% experiencing a diagnostic interval greater than three months and 25% more than eight months. Early symptoms of myeloma include back pain, rib pain, chest pain and infections while fractures, weight loss and nausea manifest later in disease progression. For blood tests haemoglobin and inflammatory markers like Erythrocyte sedimentation rate (ESR) and plasma viscosity (PV) can become abnormal up to two years before diagnosis while values in calcium and creatinine manifest later. C-reactive protein is not a useful inflammatory marker for the diagnosis of myeloma. A combination of normal haemoglobin, either ESR or PV and calcium can rule out most myeloma cases. A clinical prediction rule containing demographics (age, gender, BMI), risk factors (MGUS) symptoms (nosebleeds, back pain, chest pain, rib pain) and the parameters of the full blood count (white cell count, haemoglobin, platelets and mean corpuscular volume) showed good discrimination (AUC: 0.84, 95% CI: 0.82-0.87, R^2: 0.57, 95% CI: 0.53-0.62 and D-statistic: 2.4, 95% CI: 2.2-2.6) and good calibration. Conclusions Myeloma patients experience substantial delays in their diagnosis. General practitioners should examine patients that present with myeloma related symptoms using a combination of a full blood count, an inflammatory marker (ESR or PV) and calcium. The clinical prediction rule developed can be useful when myeloma is not suspected and not all tests are ordered. The rule could be applied in the laboratory or on the electronic systems that general practitioners use, but further work is required before implementation.