Process to process correlation: the interpretive validity of indicators.

Objective To determine the empirica! relationship between theoretically related process indicators using antibiotic prescribing guideline adherence among general practices. Data Source: One hundred and eighteen general practices participating In the Netherlands Information Network of General Practice from 2002 to 2005. Study design In this retrospective study, we used Pearson's correlation coefficient to examine the pairwise practice-level association between guideline adherence rates for episodes of bacterial skin infection (30,757), acute throat pain (28,544), sinusitis (39,648), and urinary tract infections (75,300). We used three-level multivariate multilevel analysis to study the association between practice-level adherence rates of any three adherence indicators, and likelihood to adhere to the fourth guideline. Principal findings The correlation between performance levels of the prescribing indicators ranged from negative (-0.29) to positive (0.57). The odds ratios and 95% Cl from the multilevel analyses ranged from a minimum of 0.97 (0.97 - 0.98) to a maximum of 1.02 (1.02 - 1.04) for the various practice-level adherence measures. Conclusions The relatively weak relationships between the tour process indicators indicate that they may not be measuring the same underlying construct. These findings raise questions about the interpretation of performance results based on existing quality indicators.