Table 5. Results of Cox proportional hazards models for the effect of primaquine on the risk of re-presentation to hospital with P. vivax malaria within 3 months of being treated for P. vivax malaria.

	Primaquine dose	AHR	95% Confidence interval	p
Overall model
Alla	None	1.00*	–	–
	Low- or high-dose primaquine	0.88	0.82–0.94	<0.001
Subgroup analyses
1 to <5 yrb	None	1.00*	–	–
	≥5 mg/kg (high dose)	0.82	0.75–0.89	<0.001
5 to <15 yrb	None	1.00*	–	–
	≥5 mg/kg (high dose)	0.94	0.77–1.16	0.579
≥15 yrb	None	1.00*	–	–
	≥5 mg/kg (high dose)	1.03	0.90–1.17	0.672
2007c	None	1.00*	–	–
	≥5 mg/kg (high dose)	0.82	0.75–0.91	<0.001
Primaquine dose ≥7 mg/kga	None	1.00*
	≥ 7 mg/kg	0.88	0.82–0.95	0.001
Outpatientsd	None	1.00*	–	–
	≥5 mg/kg (high dose)	0.86	0.81–0.93	<0.001

Note: Patients treated with blood schizontocides other than dihydroartemisinin-piperaquine, pregnant women, and infants less than 1 year of age were excluded from all the models. Abbreviations: AHR, Adjusted Hazard Ratio

* Reference category.

^a Cox model stratified by ethnicity (non-Papuan, Highland Papuan, and Lowland Papuan) and age group (1 to <5 yr, 5 to <15 yr and ≥15 yr). Other co-variables include gender, year, admission status (outpatient or inpatient), and presentation number.

^b Cox models stratified by ethnicity. Other co-variables include gender, year, admission status, and presentation number. Age group was found to be an effect modifier of the association between primaquine dose and vivax recurrence; p < 0.001 from likelihood ratio test comparing models with and without an interaction term between age group and primaquine dosing.

^c Model limited to the period during and 12 months before and after the largest primaquine stock outage in 2007. Cox model stratified by ethnicity and age group. Other co-variables include gender, admission status, and presentation number.

^d Cox model stratified by ethnicity and age group. Other co-variables include gender, year, and presentation number.