For some time now, CDDEP’s resistance map
has highlighted the variability that exists in antibiotic prescribing across the country. Despite the growing public health crisis of antibiotic resistance and the strong link between antibiotic use and resistance, the factors that drive these geographical differences in prescribing rate are not fully understood. While most work on this issue has been conducted in Europe, the structure of the US healthcare system, with its reliance on private health insurance and limited government interference in the price of pharmaceuticals, suggests that alternative factors may drive differences in the US.
We set out to investigate some of these alternative factors, and found that the availability of urgent care and retail clinics is an important factor in driving prescribing rates. Often referred to as “Doc-in-a-box” locations, these clinics encompass both stand-alone urgent care clinics and clinics incorporated into the retail arm of a store (e.g., CVS minute clinic). These establishments are truly an American-style invention, and bear little similarity to how most Europeans receive healthcare. Over the last decade these establishments have exploded in popularity and greatly expanded their reach. We found that these clinics are great for improving access when they are placed in areas that are poorer, but in wealthier areas they compete with clinicians, driving up the visit rate both through increased clinician visits and clinic visits. This results in an overall increase in the rate of antibiotic prescribing.
In the study
, we examined the socioeconomic and structural factors that affect antibiotic prescribing using data from IMS Health on prescriptions filled in US pharmacies by zipcode. This data was aggregated to health service areas, which are collections of zipcodes where the majority of individuals in those zipcodes get their healthcare from the same places. The health service areas, which are defined by the Dartmouth Atlas of Health, are consistent through time, which allowed us to compare prescriptions between 2000 and 2010. These two years were chosen because the prescription data could be combined with data from the census, such as on employment, education, etc., to understand the factors driving prescribing rates.
A primary factor driving prescribing was the number of physician offices per capita – the more physicians there were per person in an area, the more prescriptions per person there were in that area. There are two potential mechanisms influencing this result: (1) more physicians just equates to easier access to the doctor and thus people go to visit a physician more; or (2) the physicians are competing to attract or retain patients in some manner and this is driving up prescriptions. To try to understand the relative effect of these two forces, we looked at the role of retail and urgent care clinics. As noted above, the number of clinics greatly increased between 2000 and 2010. What we found was that in low-income areas, a clinic increased the prescribing rate, but didn’t affect the rate that physicians were prescribing antibiotics. Thus, in these areas the story is all about access. Improving access to healthcare increases the likelihood that people will get antibiotic therapy. However, in wealthier areas we found that a clinic actually increased the rate of prescribing by physicians. The mechanism for this though may not have been an increase in the probability of getting an antibiotic at any single visit, but rather the increased availability of physicians. Evidence has shown that when clinics enter an area, physician offices change their operations, creating more walk-in slots and same-day appointments. These “competitive” actions increase the opportunities for people to access the physician and thus drive up the rate of prescribing.
In addition to physician density, we found that a few other factors were correlated with increased prescribing rates. These include the proportion of the population that was elderly (65+), which is not surprising as older individuals have high rates of antibiotic use. At the other end of the age spectrum, we found that prescription increased with the number of childcare centers per capita, which as anyone who has kids will tell you is also not surprising. The first week my daughter attended a pre-school was also the first time that she ended up being prescribed antibiotics. In addition to age-related effects, we also found that the percentage of the population with a bachelor’s degree also increased the prescribing rate. This is somewhat surprising as studies in Europe have found that increased education is generally associated with fewer antibiotic prescriptions. However, this more likely speaks to the strong relationship between having a college education and getting a job with healthcare benefits, as in the US healthcare is primarily provided through the workplace. The final correlation was between the number of dialysis centers per capita and an increase in antibiotic prescribing, which likely speaks to the overall health of the population in that area.
Some factors were also correlated with decreases in prescribing. One was that while a greater proportion of African-Americans in an area had no impact on prescribing rates (controlling for poverty and income), the proportion of the population that was neither white nor African-American was negatively associated with prescribing rates. We did not have the statistical power to investigate which ethic/racial groups affected these results most, but evidence from the literature suggests that Hispanics in the US often get antibiotics through informal channels, which could be driving this result. Not surprisingly, unemployment and rural residence were also negatively correlated with prescribing, as both are related to issues of access. Finally, we had the curious case of the population under age five. Here we found that an increase in the percentage of the population under 5 was strongly correlated with increases in prescribing (as would be expected), however above about 7% this relationship actually appears to turn negative. Meaning that a large number of children in an area is actually correlated with fewer prescriptions. This, we surmise, is likely due to the well-described negative relationship between income and both fertility rates and family size. In other words, large families are more often than not correlated with poverty and poverty is correlated with a lack of access to health care.
The relationship between healthcare access, competition for patients, and antibiotic prescriptions have significant policy implications. A quantitative relationship between prescribing and its drivers can help predict the changes in antibiotic consumption that will result from future changes in demographics and socioeconomic characteristics. It also enables predictions of changes in prescribing practices as a result of specific interventions that target the healthcare delivery system. Understanding the drivers of variation in antibiotic prescribing also enables better targeting of information campaigns, such as the CDC’s Get Smart Program, on appropriate antibiotic prescribing to specific sub-populations. And it allows providers to better understand how their practice is driven by external factors. This is important as our results suggest an increase in the number of physicians per capita may drive increases in antibiotic prescribing, and that this is due in part to competition between providers.
Looking at policy prescriptions, new intervention campaigns must be broad-based, acknowledging the changing healthcare delivery landscape, and emphasizing more local coordination among diverse groups of practitioners. In the US the growth of retail and urgent care clinics has been accelerating in recent years. We show that these clinics increase access for individuals who may lack provider support, suggesting there should be a concerted effort to expand these centers in lower income areas. However, compliance with prescribing guidelines and oversight of clinics is necessary to ensure that competition and increased access do not result in unnecessary antibiotic prescriptions. Finally, our results serve as a reminder that healthcare is a business, and competition among providers (whether it be other physicians or clinics) may influence prescribing habits. This should thus serve as a call to providers to adhere more strongly to published guidelines regarding antibiotic prescriptions and to potentially foster more collaborative environments regarding these types of issues.
Eili Klein is a fellow at CDDEP and an assistant professor in the Department of Emergency Medicine at Johns Hopkins University. His research focuses on the role of individuals in the spread of infectious diseases.
Study referenced: Influence of provider and urgent care density across different socioeconomic strata on outpatient antibiotic prescribing in the USA
Image taken by Sparky, used with Creative Commons license and retrieved via Flickr.