Lower respiratory tract infections in the elderly: Prognostic studies in primary care

Abstract

Lower respiratory tract infections (LRTI) are among the most common diseases presented in primary care. When the general practitioner (GP) diagnoses an LRTI he or she is confronted with important clinical dilemmas concerning treatment and prognosis. Especially elderly are of importance, as the incidence of LRTI is three times higher and elderly are more prone to develop complications. In addition, the number of elderly will increase dramatically during the next decade. In the first part of the study we focussed on antibiotics. To improve antibiotic prescribing it is important to study why GPs prescribe antibiotics to certain patients, but not to others. First, in 2,643 patients with acute bronchitis or ECOPD, we retrospectively assessed whether antibiotics were more often prescribed to patients with risk elevating conditions. Antibiotics were often prescribed in both acute bronchitis (84%) and ECOPD (53%). In acute bronchitis, high age was the only determinant, while in ECOPD determinants were male gender, diabetes and heart failure. Next, we prospectively repeated the study in 304 episodes of acute bronchitis. In 82% an antibiotic was prescribed. Purulent sputum and abnormal auscultation were the only determinants. We concluded that more appropriate prescribing might be accomplished by taking more risk-elevating determinants more often into account. The second part of the study focused on prognosis. First, in 314 elderly patients with pneumonia, we prospectively validated a well-known prediction model, the CRB-65 severity score (Confusion, high Respiratory rate, low Blood pressure and Age 65 years or over) to predict 30-day mortality. The results demonstrated that the CRB-65 predicted mortality very well. In order to be applicable to primary care patients with different kind of LRTI, we conducted a new study. We retrospectively studied prognostic factors in 3,166 episodes of elderly patients visiting the GP with acute bronchitis, ECOPD or pneumonia, and develop a prediction model that predicts hospitalisation or death. It was validated in The Second Dutch Survey of GPs. Patients with a low score had a probability of 3% and patients with a high score had a probability of 31% on hospitalisation or death. Next we prospectively validated the retrospective rule once more and optimised it with data on signs and symptoms in a new study including 1,158 episodes with elderly patients with LRTI. Prospective validation demonstrated a comparable discriminative capacity (AUC 0.73). The final clinical prediction rule was better capable at selecting low- and high-risk patients. The discriminative capacity increased (AUC 0.81). Application of a clinical prediction rule will undoubtedly be enhanced further if its surplus value is demonstrated. Therefore two prediction rules (CRB-65 and the clinical prediction) were compared, with the prognosis estimated by the GP. Our study demonstrates that GPs are very well capable of identifying high-risk patients. However, the clinical prediction rule was much better at selecting patients with a good prognosis. We recommend using the clinical prediction rule if the GP estimates a low- or medium-risk. Because of that, more patients who in reality have a low-risk will be identified as low-risk, therefore preventing unnecessary treatment or referral to hospital.