Single‐Item Self‐Rated Health Question for Assessing General Health: A Population‐Based Study Among Kurdish Population
Soheyb Ghafouri, Jalal Asadi, Ghobad Moradi, Abbas Aghaei, Bakhtiar Piroozi, Yousef Morad, Farhad MoradpourABSTRACT
Background
Single‐item self‐rated health (SRH) is widely used as a brief measure of perceived health because of its simplicity and feasibility in population‐based research. However, its agreement with a reference measure and the extent to which its performance varies across sociodemographic and health‐related subgroups remain important considerations. This study evaluated the performance of single‐item SRH against the 28‐item General Health Questionnaire (GHQ‐28) reference criterion among adults participating in the Dehgolan Prospective Cohort Study (DehPCS) in Iran.
Methods
This cross‐sectional analysis used baseline data from 2,009 adults aged 35–70 years. SRH was assessed using a five‐point response scale and dichotomized into poor health (scores 0–2) and good health (scores 3–4). The GHQ‐28 was used as a reference criterion rather than a gold standard for general health. Diagnostic performance was assessed using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and likelihood ratios, while agreement was evaluated using observed percentage agreement and Cohen's kappa. Binary and multinomial logistic regression models were used to examine factors associated with SRH classification and discordance with the GHQ‐28 reference classification.
Results
Overall, 30.9% (95% CI: 28.9–32.9) of participants were classified as having poor health according to the SRH measure. The observed percentage agreement between SRH and the GHQ‐28 reference classification was 77.7%, with a Cohen's kappa coefficient of 0.469 (95% CI: 0.427–0.511), indicating moderate agreement. Overall sensitivity and specificity were 64.8% and 83.0%, respectively. Sensitivity was higher among participants aged > 60 years (79.1%), women (67.1%), and those with chronic disease (70.7%) or hypertension (74.7%). Specificity was higher among university graduates (88.0%) and men (86.9%). PPV was higher among women (66.1%) and illiterate participants (69.9%), whereas NPV was higher among university graduates (94.6%) and men (90.9%). In multivariable analyses, smoking history was associated with higher odds of disagreement, including current smoking (adjusted OR = 1.64, 95% CI: 1.16–2.33) and former smoking (adjusted OR = 1.74, 95% CI: 1.16–2.62). Chronic disease was associated with higher odds of false‐positive (adjusted OR = 1.50, 95% CI: 1.11–2.03) and true‐positive (adjusted OR = 2.18, 95% CI: 1.67–2.83) classification.
Conclusion
Single‐item SRH showed moderate agreement with the GHQ‐28 reference classification at the population level, while its performance varied across sociodemographic and health‐related subgroups. These findings support the use of single‐item SRH as a brief population‐level measure of perceived health, while emphasizing the importance of considering subgroup characteristics and the specific health construct represented by the reference criterion when interpreting its performance.