Design: Retrospective matched cohort, difference-in-differences
Population: Polyclinic patients with diabetes, hypertension or hyperlipidaemia, Singapore
Primary outcome: Clinical parameters, screening adherence, utilisation and cost over 5 years
Effect: Foot screening +16%, eye +6%; net saving SGD 2186 per patient

From 2014, Singapore polyclinics enrolled patients with diabetes, hypertension or hyperlipidaemia into 'teamlets' of two family physicians, one nurse and one care coordinator. This matched difference-in-differences analysis compared five-year outcomes with usual care using national databases. It was published in Annals of Family Medicine on 21 September.

Clinical changes were very small: HbA1c 0.1% lower in some subgroups, systolic pressure 0.65 mm Hg lower and LDL 0.03 mmol/L lower. Process measures moved more: 16% more foot screenings and 6% more eye screenings. Hospital admissions, specialist visits and emergency attendances fell, with a net saving of about SGD 2186 per patient despite higher polyclinic costs.

For high-volume public clinics, including many in India, the message is that a named small team improves screening completion and keeps patients out of hospital, even when it barely shifts the numbers on the chart.

Assign chronic disease patients to a named small team rather than whoever is free.
Give a nurse or coordinator ownership of foot and eye screening recalls.
Measure admissions and screening completion, not only HbA1c.
Expect cost savings to come from hospital avoidance.

Why it matters

Continuity through small teams saved money mainly by keeping patients out of hospital, not by moving biomarkers.

The statistics, in plain English

Difference-in-differences compares changes over time in both groups, which helps remove stable differences between them, but it is still observational. The biomarker changes are statistically detectable but clinically negligible.