Table 7. Relationship between exposure to information from drug companies and prescribing quality (by year of publication and then study design/size).

Exposure to Information from Drug Company	Study (First Author Name)	Result in Exposed Group Versus Controls (Where Applicable)	Change in Prescribing Quality Result
Effect of PSR visits	de Bakker [61]	Wider prescribing range was associated with more visits from PSRs in the last 4 wk	Beta coefficient +0.18 (p<0.05)a
	Muijrers [64]	More frequent visits from PSRs was associated with less adherence to prescription guidelines	Multiple linear regression:
			Beta −0.23 (95% CI −0.32 to −0.15) p<0.05
	Mansfield [82]	Frequency of visits from PSRs was not associated with a difference in quality score	Pearson coefficient of 0.0363; p = 0.247
	Blondeel [81]	Based on responses to 8 case histories:	Multivariate regression analysis:
		First contact with a drug from the pharmaceutical industry was not associated with quality index;	p>0.1
		n PSRs received was associated with poorer quality index;	p>0.05
		Based on prescriptions for actual patients: First contact with a drug from the pharmaceutical industry was associated with reduced quality of prescribing;	p<0.01
		n PSRs received was not associated with poorer quality index	p>0.1
	Becker [77]	Fewer visits from PSRs/month were not associated with a change in the appropriateness of prescribing	Gamma statistic; 0.04, not statistically significant
Attendance at pharmaceutical sponsored meeting	Mansfield [82]	Attendance at pharmaceutical sponsored meetings was associated with lower quality scores	Pearson correlation coefficient of 0.0635; p = 0.043
	Spingarn [39]	Attendees at a sponsored talk about Lyme disease were less likely to choose appropriate oral antibiotics for mild Lyme disease than nonattendees	0% of attendees (n = 22) chose appropriate antibiotics compared to 21% (n = 53) of nonattendees; Fisher exact test: p = 0.027
		For attendees and nonattendees of a sponsored talk about Lyme disease there was no difference in choice of acceptable treatment for Lyme disease with central nervous system signs	OR = 3.2 (95% CI 0.8–19.2)
		Attendees of a sponsored talk about Lyme disease were more likely to appropriately choose the sponsoring company's treatment for Lyme disease complicated by 2nd degree heart block	OR = 7.9 (95% CI 2.4–29.3)
Journal advertisements	Becker [77]	Infrequent use of journal ads as a source of prescribing information by doctors was not associated with a change in the appropriateness of prescribing	Gamma statistic 0.373, not statistically significant
Total promotional investment/summated scores of commercial information use/general use of commercial sources	de Bakker [61]	There was a positive correlation for how frequently doctors used the pharmaceutical industry as a source of information and the range of drugs they prescribed	Beta coefficient +0.15 (p<0.05)a
	Haayer [74]	Frequency of use of information from the pharmaceutical industry was associated with less rational prescribing	Beta coefficient +0.134 ; p<0.001
Information delivered without conventional promotion	Andersen [37]	Participation in a randomized controlled trial was not associated with a change in guideline adherence at 2 y for trial sponsor's medication	OR 1.00 (95% CI 0.84–1.19)
	Kazmierczak [59]	Mailed warning letters regarding tramadol for those with a seizure risk were not associated with a change in prescription rates for tramadol	9 (10%) prescriptions before and 7 (9%) after warning letters were sent out no association detected
	Soumerai [58]	PSR visits: Propoxyphene use continued a preexisting decline of about 8% a year during the time when warnings from the manufacturing pharmaceutical company were conveyed by PSRs after which time this decline halted, however a statistical association was not shown. Refill rates and rates of overdose did not change following the warnings	No association detected
		Mailed Information: Propoxyphene use continued a preexisting decline of about 8% a year during the time when warnings from the manufacturing pharmaceutical company were expressed by PSRs after which time this decline halted, however a statistical association was not shown. Refill rates and rates of overdose did not change following the warnings	No association detected

^aAssumes a wide prescribing range is lower quality prescribing than a narrow prescribing range.