Background Existing evidence indicates that reducing nurse staffing and/or skill mix adversely affects care quality. Nursing shortages may lead managers to dilute nursing team skill mix, substituting assistant personnel for registered nurses (RNs). However, no previous studies have described the relationship between nurse staffing and staff–patient interactions.

Setting Six wards at two English National Health Service hospitals.

Methods We observed 238 hours of care (n=270 patients). Staff–patient interactions were rated using the Quality of Interactions Schedule. RN, healthcare assistant (HCA) and patient numbers were used to calculate patient-to-staff ratios. Multilevel regression models explored the association between staffing levels, skill mix and the chance of an interaction being rated as ‘negative’ quality, rate at which patients experienced interactions and total amount of time patients spent interacting with staff per observed hour.

Results 10% of the 3076 observed interactions were rated as negative. The odds of a negative interaction increased significantly as the number of patients per RN increased (p=0.035, OR of 2.82 for ≥8 patients/RN compared with >6 to <8 patients/RN). A similar pattern was observed for HCA staffing but the relationship was not significant (p=0.056). When RN staffing was low, the odds of a negative interaction increased with higher HCA staffing. Rate of interactions per patient hour, but not total amount of interaction time, was related to RN and HCA staffing levels.

Conclusion Low RN staffing levels are associated with changes in quality and quantity of staff–patient interactions. When RN staffing is low, increases in assistant staff levels are not associated with improved quality of staff–patient interactions. Beneficial effects from adding assistant staff are likely to be dependent on having sufficient RNs to supervise, limiting the scope for substitution.