I would not resample it, since you will lose significant information with a 200 Hz sampling frequency. Heart rate is relatively easy to determine, even if irregular. Depending on the lead that recorded the EKG (rhythm strips are most commonly Lead II), all you need to do is to threshold the signal to detect the R waves. The easiest way to do this is to use the findpeaks function, using the appropriate name-value pair parameters for your signal.
The spectrogram of an EKG is not likely to provide you with much useful information. If you want to reproduce the result of a paper, use the techniques it uses. Otherwise, experiment with the parameters of the spectrogram function to get the result you want. There is probably no one best way to do it.
If the signal is noisy, use a bandpass filter with a passband of 1 Hz to 100 Hz before you do any further analyses. This removes most high-frequency noise and eliminates baseline drift without removing relevant information from your signal. A Butterworth design works well. Use the filtfilt function for the actual filtering. (I outlined the essentials of filter design here.)
