There is a problem with wire-wound precision resistors - thermocouple effects. The junction of the resistance wire and the resistance lead forms a thermocouple with a thermoelectromotive force EMF of 42 μV/° C. for the standard “alloy 180”/nickel-chromium alloy junction of a conventional wirewound resistor. If a resistor with a (more expensive) copper/nickel chrome alloy junction is selected, this value is 2.5 μV/°C. (“Alloy 180” is a standard component lead alloy of 77% copper and 23% nickel.)
This thermocouple effect is not important in AC applications. When the temperature across the resistor is the same, they cancel out; however, if one end is warmer to the other end, either due to power consumption in the resistor or its position relative to the heat source, the net thermoelectric EMF will introduce an erroneous DC voltage into the circuit. With a normal wire-wound resistor, a dc difference of 168μV is produced with a temperature difference of only 4°C - greater than 1 LSB in a 10 V / 16 bit system! This problem can be solved by installing a wire-wound resistor to ensure that the temperature difference is minimized.
You can try to ensure that both leads of equal length have the same thermal conduction through them by ensuring any air flow (whether forced or natural convection) is perpendicular to the resistor body as in the below figure. Also please ensure both ends of the resistor are at the same thermal distance from any heat source on the PCB to receive equal heat flow.