Relationship between facility number of clinicians and prescribing intensity of psychotropic medications, opioids, and antibiotics in ambulatory practice

Table 2 Rates* of facility-level decline in prescribing intensity for psychiatrists and primary care clinicians associated with VA facility size, 2014 to 2017

	Psychiatrists
	Percentage of patients prescribed class				Mean days supply
	BZD	AP	AE	AD	BZD	AP	AE	AD
2014	0.89	0.94	0.93	0.94	0.85	0.88	0.88	0.89
2015	0.89	0.94	0.94	0.94	0.86	0.89	0.89	0.90
2016	0.89	0.93	0.94	0.93	0.84	0.87	0.88	0.89
2017^†	0.90	0.94	0.95	0.94	0.87	0.89	0.90	0.90
p-value^‡	0.40	0.23	0.33	0.22	0.03	0.03	0.06	0.08
Overall^§	0.89	0.94	0.94	0.94	--	--	0.89	0.89
	Primary Care Physicians
	Percentage of patients prescribed class				Mean days supply
	AB	Opioids	AE	AD	AB	Opioids	AE	AD
2014	0.98	0.97	0.97	0.97	0.98	0.96	0.96	0.97
2015	0.98	0.97	0.97	0.97	0.98	0.96	0.96	0.97
2016	0.98	0.97	0.97	0.97	0.98	0.96	0.96	0.97
2017^†	0.98	0.97	0.97	0.97	0.98	0.96	0.96	0.96
p-value^‡	0.28	0.76	0.56	0.24	0.81	0.84	0.45	0.26
Overall^§	0.98	0.97	0.97	0.97	0.98	0.96	0.96	0.96

Abbreviations: BZD is benzodiazepines; AP is antipsychotics; AE is antiepileptics; AD is antidepressants
*Rates estimated the change in prescribing intensity outcome associated with 10 additional facility clinicians, based on prescribing intensity data from 2014 to 2017 modeled using multilevel generalized mixed models with log link and facilities as random intercepts and number of facility clinicians in each year, year indicators, and their interaction terms as predictors. All rate estimates were significant (p < 0.001) when tested for rates equal to 1.0
^† 2017 estimates may slightly differ from those in Table 1 as these are based on a model using data from all years
^‡ P-value is from the 3-degrees of freedom likelihood ratio test (LRT) for slope differences across years, i.e., from testing for the interaction terms of number of providers by year indicators
^§ An overall rate (i.e., slope across the four years) is estimated based on a mixed model with each prescribing intensity measure after dropping the interaction terms if the LRT test described in footnote b is not significant (p > 0.05). An overall rate is not estimated when the LRT of differential slope by year was significant

ISSN: 1472-6963