When gatekeepers meet the sentinel: the impact of a prior authorization requirement for cefuroxime on the prescribing behaviour of community-based physicians

Abstract

Aims

Prior authorization (PA), the requirement of physicians to obtain pre-approval as a prerequisite for coverage, may decrease drug utilization via a ‘sentinel effect’, a decrease in utilization caused by external review of prescribing. The purpose of this study was to assess the affect a PA restriction had on the utilization patterns of cefuroxime tablets in a managed care organization (MCO) in Israel.

Methods

Physician prescribing patterns were evaluated by conducting a retrospective drug utilization analysis. Data were derived from the electronic patient records of the MCO studied. All prescriptions for solid state antibiotics for patients diagnosed with an infectious disease written during three parallel 3-month segments, before, during and after a PA restriction for cefuroxime was enforced, were included. Frequency and proportion of antibiotic prescriptions for cefuroxime tablets, distribution of infectious diseases treated with cefuroxime, and the request rejection rate when PA was required were calculated.

Results

Prescriptions for cefuroxime declined from 5538 prescriptions (8.0% of eligible antibiotic prescriptions, 95% CI 7.8, 8.2) in the initial period to 1036 (1.2%, 95% CI 1.1, 1.3) during the PA period, rising to 3961 (4.3%, 95% CI 4.2, 4.4) in the post-PA period. Changes in the distribution of diseases treated with cefuroxime during the PA stage tended to regress after revocation to those observed in the pre-PA period. The rejection rate was found to be 8.5% (95% CI = 6.9, 10.1).

Conclusions

The implementation of a prior authorization requirement for cefuroxime tablets markedly reduced the use of this drug, probably due to a ‘sentinel effect’.

Introduction

Prior authorization (PA), the managerial technique requiring physicians to obtain pre-approval for drugs as a prerequisite for provider coverage, is increasingly being implemented in managed care organizations (MCOs). MCOs implement PA as an administrative tool to encourage appropriate utilization of medications, anticipating that improvement in prescribing precision will reduce the prevalence of drug-induced morbidity and contain costs [1, 2]. While the PA rejection rate has been observed to be relatively small (4.4%), it has been suggested that PA may generate a ‘sentinel effect’[2]: the ‘decrease in services given by providers as a result of having a utilization reviewer keep tabs on them’[3].

The Leumit Health Fund is a managed care organization (MCO) that provides coverage to ∼650 000 members (∼10% of the population) throughout the state of Israel. The MCO's Medical Division has formulated a drug policy mandating prior approval for relatively expensive drugs, and to expedite this authorization process, a computer program was developed that operates under the MCO's electronic patient record (EPR) system. When a physician prescribes one of these drugs, a window appears instructing the physician to press a function key for transmission of an authorization request to an administrative physician for approval. Two blank lines are provided to enter supplementary information. This policy has been predominantly implemented for drugs for the treatment of neoplastic diseases, and the newer more expensive drugs for the treatment of chronic illnesses such as hypertension, diabetes, and dyslipidaemia.

The problem of antimicrobial resistance is now emerging as a critical global problem, with inappropriate antibiotic prescribing and misuse and overuse being identified as the main cause of resistance [4, 5]. Accordingly, leaders in public health and infectious diseases have called for a decrease in excessive antibiotic use in the community [6]. Nevertheless, improving antibiotic use in the community continues to be difficult, with obstacles such as patient expectations being difficult to surmount. Multiprong educational techniques have achieved the most promising results [7, 8]. When the antibiotic drug cefuroxime axetil was introduced into the Israeli market, it was registered in Leumit's formulary without prescribing restrictions (open access) despite it being a relatively expensive drug of this class. However, when it was subsequently suspected that inappropriate and unnecessary prescribing of the drug was resulting in prohibitive annual expenditures to the fund, a PA requirement was imposed. Physicians opting to prescribe this drug were then required to follow the computerized PA procedure, with rejection or acceptance being based on the case history of the patient and the availability of equally effective but less expensive alternatives. The recent expiration of the drug's patent and the introduction of a generic equivalent into the market have significantly reduced the cost of treatment with cefuroxime, prompting the fund to rescind the PA requirement. These changes in the PA requirement have thus produced three distinct periods in the policy history of this drug, creating a three-phase natural experiment in drug utilization in this organization. Although the effects of PA requirements on drug utilization patterns, costs and health-related outcomes have been assessed for various drugs [9–11], as well as the effect of revoking the PA restriction for the prescribing of antidepressant medications [12], the effect of this managerial strategy on oral antibiotic utilization in the community setting has not been studied. The purpose of this study was to assess how these changes affected physician prescribing behaviour for cefuroxime in this managed care setting.

Methods

Data were derived from electronic patient records of the MCO studied. All cases in which physicians prescribed solid dosage forms of antibiotics for patients diagnosed with an infectious disease during three parallel 3-month segments of each of the three study periods (November through January 2001–2, 2003–4 and 2004–5) were included. The observed International Code of Disease, 9th version (ICD-9) diagnostic codes were classified into six categories: bronchitis, genitourinary tract infections, sinusitis, pneumonia, other URTI + otitis + pharyngitis, and other. The rate of cases treated with cefuroxime and the disease distribution of cases treated with the target drug were evaluated for each period. Rate ratios and 95% confidence intervals (95% CI) compared with the initial period were then calculated. The rejection rate (percentage and 95% CI) of PA requests was calculated for the relevant period.

Results

Prescriptions of cefuroxime declined from 5538 prescriptions (8% of antibiotic prescriptions) in the initial period to 1036 (1.20%, RR = 0.15; 95% CI 0.15, 0.16) when PA restrictions were in effect, rising to 3961 (4.30%, RR = 0.54; 95% CI 0.53, 0.55) after the PA restriction was revoked (Table 1). Parallel complementary trends were only observed during the three study period segments in the prescription rates of amoxicillin + clavulanic acid (9.0%, 14.3% and 13.7%) and ofloxacin (2.3%, 5.0%, and 4.2%). A steady increase throughout the three segments of the study period in the number of cases identified (69 173, 82 633, and 91 978, respectively) is attributed to improvements in rates of physician computerization in addition to enhanced physician computer competence over time. The disease distribution of cases treated with cefuroxime is presented in Table 2. Although changes in utilization of the drug were observed after implementation of the PA requirement, prescribing patterns tended to regress after revocation to those observed in the pre-PA period. Throughout the 3-month period studied when PA was required for cefuroxime, 1132 requests for the drug were submitted, 1036 of which were granted, representing an 8.5% rejection rate (95% CI 6.9–10.1).

Table 1.

Antibiotic prescription data

Period	Total prescriptionsfor antibiotics	Prescriptions forcefuroxime (%)	Rate ratio (95% CI)
Open access	69,173	5538 (8.0)	Reference
PA required	82,633	1036 (1.2)	0.15 (0.15, 0.16)
Post-PA revocation	91,978	3961 (4.3)	0.54 (0.53, 0.55)

Table 2.

Disease distribution of cases treated with cefuroxime

	Open access		PA required		Post-PA revocation
Diagnostic category	Number (%)	Rate ratio	Number (%)	Rate ratio (95% CI)	Number (%)	Rate ratio (95% CI)
Bronchitis	1009 (18.2)	Reference	120 (11.6)	0.64 (0.53, 0.76)	691 (17.5)	0.96 (0.89, 1.03)
Genitourinary	695 (12.6)	Reference	198 (19.1)	1.52 (1.33, 1.74)	557 (14.1)	1.12 (1.01, 1.24)
Sinusitis	1076 (19.4)	Reference	240 (23.2)	1.19 (1.05, 1.35)	761 (19.2)	0.99 (0.91, 1.07)
Pneumonia	775 (14.0)	Reference	238 (23.0)	1.64 (1.45, 1.86)	795 (20.1)	1.43 (1.31, 1.57)
Otitis, pharyngitis, other URTI	1552 (28.0)	Reference	150 (14.5)	0.52 (0.44, 0.60)	817 (20.6)	0.74 (0.73, 0.74)
Other	431 (7.8)	Reference	90 (8.7)	1.12 (0.89, 1.38	340 (8.6)	1.1 (0.96, 1.26)
Total	5538	–	1036 (100)	–	3961 (100)	–

Discussion

This analysis of physician prescribing patterns for the antibiotic cefuroxime before, during and after the implementation of a PA restriction indicates that this administrative intervention succeeded in curtailing the utilization of a broad spectrum, relatively expensive antibiotic in the primary care setting. The 85% relative reduction in the use of the drug during the PA period despite the request rejection rate of only 8% indicates the existence of a sentinel effect that prevented the prescription of the target drug. The postintervention prescribing rates demonstrating a regression to the patterns observed in the initial open access period strengthen this conclusion. This three-phase analysis of a PA program is therefore unique in that the first open access period provides a benchmark for analysis of the subsequent prescribing patterns evaluated in the latter two phases of this natural experiment. This methodology is particularly significant in the third phase studied in that it facilitated measurement of any residual effect of the PA requirement during the postrevocation phase in comparison with a historical control group.

Although changes in prescribing patterns were observed, these data do not necessarily indicate that the PA requirement encouraged or motivated evidence-based utilization of the drug studied. The sentinel effect may have worked due to a lack of physician willingness to delay antibiotic treatment, or reluctance on their part to subject patients to a bureaucratic annoyance. We therefore cannot rule out the prospect that patients for whom therapy with cefuroxime was indeed justified did not receive the drug due to the sentinel effect detected. Additionally, as the PA requirement was subsequently revoked upon the introduction of a less expensive generic equivalent, it is plausible that physicians correctly perceived the restriction as being economically and not clinically motivated. These results may therefore suggest that the contribution of PA strategies in improving antibiotic prescription appropriateness could be enhanced in the future if implemented as a component of multipronged intervention strategies that integrate the PA procedure with technologies for the dissemination of relevant guidelines. Such an approach could be beneficial to both the practising physician and the utilization reviewer by providing evidence-based criteria for the use of the drug, thereby improving the quality of the decision making process for all parties involved. As previous studies conducted in this MCO have reported suboptimal physician adherence to current guidelines for the prescribing of antibiotics [13,14], this strategy may be particularly relevant to the clinical setting studied. Likewise, in view of the increasing global concern regarding the inappropriate and unnecessary use of antibiotics [5, 15] multipronged approaches of this type may be a viable strategy for promoting more judicious use of oral antibiotics in other community settings internationally. The results of this study therefore indicate that PA strategies that effectively curtail inappropriate or excessive antibiotic prescribing in the community setting may be valuable components of future efforts to minimize the antibiotic resistance associated with suboptimal prescribing behaviour.

In conclusion, the implementation of a PA requirement for oral cefuroxime tablets in a managed care setting in Israel markedly reduced the use of this drug. Future PA programmes may be enhanced if incorporated into multiprong guideline dissemination strategies. This study suggests that MCOs should routinely harness the opportunities created by natural experiments resulting from policy changes to analyse and understand better the prescribing traits of the physicians they employ.