CONTEXT AND OBJECTIVE: To evaluate predictors of changes in lipid parameters consisting of LDL-C (low-density lipoprotein cholesterol), TC (total cholesterol) and non-HDL-C (non-high density lipoprotein cholesterol) among primary care patients. DESIGN AND SETTING: Retrospective study conducted on family medicine patients. METHODS: Demographic features and other clinically relevant information were abstracted from medical records. The primary outcome was the difference in LDL-C level from initial testing to the index test. Secondary outcomes were the changes in TC and non-HDL-C levels between two measurements. RESULTS: Three hundred and eleven participants were included in the final secondary analysis. Multiple linear regression revealed that male patients (β = 4.97, P = 0.040), diabetes (β = 9.75, P = 0.003) and higher LDL-C levels at baseline (β = 0.35, P < 0.001) were positively associated with LDL variance, whereas longer time period (β = -0.15, P = 0.045) and familial hypercholesterolemia history (β = -7.56, P = 0.033) were negatively associated. Male patients (β = 8.45, P = 0.002), DM (β = 9.26, P = 0.011), higher TC levels at baseline (β = 0.35, P < 0.001) and taking statins (β = 7.31, P = 0.023) were positively associated with TC variance, whilst longer time period (β = -0.183, P = 0.031) and familial hypercholesterolemia (β = -10.70, P = 0.008) were negatively associated. CONCLUSION: In the present study, patients who were male, on statin treatment, diagnosed with diabetes and had higher baseline lipid values were more likely associated with better lipid outcomes at future testing.RESUMO CONTEXTO E OBJETIVO: Avaliar preditores de alterações nos parâmetros lipídicos que consistem em LDL-C (colesterol de lipoproteína de baixa densidade), TC (colesterol total) e não HDL-C (não colesterol de lipoproteína de alta densidade) entre os pacientes de cuidados primários. TIPO DE ESTUDO E LOCAL: Estudo retrospectivo realizado em pacientes de medicina familiar. MÉTODOS: Aspectos demográficos e outras informações clinicamente relevantes foram extraídos dos prontuários médicos. O desfecho primário foi a diferença de nível de LDL-C entre os exames iniciais e o exame índice. Os desfechos secundários foram as mudanças dos níveis de TC e não HDL-C entre as duas medidas. RESULTADOS: Trezentos e onze participantes foram incluídos na análise secundária final. Regressão linear múltipla revelou que os pacientes do sexo masculino (β = 4,97, P = 0,040), diabetes (DM) (β = 9,75, P = 0,003) e níveis de LDL mais elevados no início do estudo (β = 0,35, P < 0,001) foram associados positivamente com variância LDL, enquanto longo período de tempo (β = -0,15, P = 0,045) e história hipercolesterolemia familiar (β = -7,56, P = 0,033) foram associados negativamente. Pacientes do sexo masculino (β = 8,45, P = 0,002), com DM (β = 9,26, P = 0,011), níveis elevados de CT na linha de base (β = 0,35, P < 0,001) e tomar estatinas (β = 7,31, P = 0,023) associaram-se positivamente com a variância TC, enquanto longo período de tempo (β = -0,1...