Answers are grounded in this paper and cite their sources.
Observational analysis reveals that propensity score matching reduces confounding bias in refractive errors among hypertensive patients, indicating significant factors such as genetic history.
Epidemiological and public health research often relies on observational studies when randomized controlled trials (RCTs) are not feasible due to ethical, financial, or logistical constraints. A key challenge of observational studies is bias caused by confounding variables. To minimize this bias, Propensity Score Matching (PSM) was applied, matching subjects based on treatment probability estimated through binary logistic regression. This study analyzed factors influencing refractive errors among hypertensive patients using medical records from UPT Puskesmas Cipamokolan, Bandung. The PSM procedure included propensity score estimation with nearest neighbor matching, covariate balance assessment, and calculation of the Average Treatment Effect on the Treated (ATT). The initial model including all covariates was not statistically significant. After excluding obesity due to imbalance, the second model yielded a significant ATT of -0.0738 with a 70.76% bias reduction. This negative ATT indicates that hypertensive smokers were 7.38% less likely to experience refractive errors compared to non-smokers after matching. However, chi-square analysis showed that only genetic history remained significantly associated with refractive errors, suggesting effect modification. Thus, the findings should be interpreted as observational differences rather than causal effects. Abstrak. Penelitian epidemiologi dan kesehatan masyarakat sering menggunakan studi observasional ketika Randomized Controlled Trial (RCT) tidak memungkinkan karena alasan etika, biaya, atau logistik. Tantangan utama studi observasional adalah potensi bias akibat variabel perancu (confounding). Untuk meminimalkan bias tersebut, penelitian ini menggunakan metode Propensity Score Matching (PSM), yang mencocokkan subjek berdasarkan probabilitas menerima perlakuan melalui regresi logistik biner. Penelitian ini menganalisis faktor-faktor yang memengaruhi kelainan refraksi pada pasien hipertensi dengan menggunakan data rekam medis dari UPT Puskesmas Cipamokolan, Bandung. Proses PSM mencakup estimasi skor dengan nearest neighbor matching, evaluasi keseimbangan kovariat, serta perhitungan Average Treatment Effect on the Treated (ATT). Model awal dengan seluruh kovariat tidak signifikan. Setelah variabel obesitas dikeluarkan karena tidak seimbang, model kedua menghasilkan ATT sebesar -0,0738 yang signifikan dengan reduksi bias 70,76%. Nilai ATT negatif ini menunjukkan bahwa pasien hipertensi yang merokok memiliki kemungkinan 7,38% lebih rendah mengalami kelainan refraksi dibandingkan yang tidak merokok setelah pencocokan. Namun, analisis chi-square menunjukkan bahwa hanya riwayat genetik yang tetap signifikan, mengindikasikan adanya effect modification. Oleh karena itu, hasil penelitian ini tidak dapat ditafsirkan sebagai hubungan kausal, melainkan sebagai perbedaan observasional rata-rata setelah pencocokan dilakukan.
No takes yet. Share an insight, caveat, or question.
Pujiastuty et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: