OUP user menu

Treatment compliance under physician–industry relationship: a framework of health-care coordination in the USA

Jie Chen, Arturo Vargas-Bustamante
DOI: http://dx.doi.org/10.1093/intqhc/mzt017 340-347 First published online: 13 February 2013

Abstract

Objective Factors associated with treatment compliance have been well studied. However, no study has examined treatment compliance under the context of physician–industry relationship. This study developed a conceptual framework of physician–industry relationship and treatment compliance, and empirically tested patients' treatment compliance and affordability under the physician–industry relationship in the USA.

Design We first proposed a conceptual framework to analyze different scenarios, where the physician–industry relationship could impact patients' treatment compliance and affordability, taking into consideration the role of health insurers. We then employed a nationally representative data set to investigate these relationships. Multivariable logistic regressions were employed to examine the physician–industry relationship and the physicians' perception of patients' treatment compliance.

Setting and Participants 2008 Health Tracking Physician Survey.

Results Our results showed that physicians with closer industry relationships were more likely to report rejection of care by insurers [odds ratios (ORs): 1.24–1.85, P < 0.001], patients' non-compliance with treatment (OR: 1.34, P < 0.01) and patients' inability to pay (OR: 1.42, P < 0.01) as the major problems affecting their ability to provide high quality care, when compared with physicians without industry relationships.

Conclusions Our results shed light on the lack of articulation among industry, physicians and health insurers in the USA. It is important to make sure that different agents in the health-care marketplace, such as physicians, industry, and health insurers, coordinate more efficiently to provide quality and consistent care to patients.

  • patient–provider communication/information
  • statistical methods
  • patient-centered care
  • quality improvement
  • quality management
  • statistical methods
  • general methodology
  • health policy
  • health-care system

Introduction

The relationship between physicians and industry is ubiquitous [16] in the USA. The 2001 National Survey of Physicians showed that ‘92% of physicians received drug samples, 61% received meals, tickets to events, or free travel, 13% received financial or other kinds of benefits, and 12% received incentives for participation in clinical trials’ [1, 7]. The pervasiveness of physician–industry interactions is also reflected by the considerable marketing investments by the pharmaceutical companies. The pharmaceutical industry allocates 33% of its revenue on ‘selling and administration’ [8] and spends $19 billion each year ‘establishing and maintaining’ relationships with physicians.

The physician–industry interaction can have positive and negative effects on patient's care. The physician–industry interaction could be beneficial to reduce under-prescription of essential medications. Many prescription drugs in the market would not exist without physicians' involvement in clinical trials. Physicians with closer industry relationships, however, are more likely to request the addition of specific drugs from particular pharmaceutical companies on hospital formularies, are more likely to prescribe these companies' drugs and are less likely to prescribe generic medications [1, 3, 9]. Such changes in the use of medications can potentially be expensive and ‘nonrational’ [3, 9] because the industry could promote some drugs with lower therapeutic advantage when compared with more cost-effective alternatives. Moreover, in the USA, health insurers are the major payers of health services and medical products. Health insurers' interests and guidelines complicate the physician–industry relationship.

This is the first paper that to our knowledge examined patients' treatment compliance and affordability under the physician–industry relationship. The health benefits of treatment compliance have been well documented [10, 11]. Patients' characteristics [12], health-care access and health insurance coverage could all affect patients' treatment compliance and affordability [13, 14]. However, no study has examined treatment compliance under the context of physician–industry relationships. It is critical that different agents in the health-care marketplace, such as doctors, industry and health insurers, work together more efficiently and better coordinated to provide quality and continued care to patients.

In this study, we first set up a conceptual framework to analyze different scenarios where physician–industry relationship could impact treatment compliance and affordability, considering the role of health insurers. Then, we employed a nationally representative data set of physicians to empirically investigate this relationship. It is worth noting that although we studied the health-care market in the USA, our results may provide references and policy implications of the physician–industry relationships to other countries too. Physician–industry relationship is common in the UK, Australia and other countries [1519]. Although most of these countries have guidelines prohibiting companies from giving physician incentives to prescribe their products in the form of gifts or payments [15], these guidelines were either developed recently or are still being developed. Researchers from different countries [18] have consistently called for more transparency of physician–industry communications.

Conceptual framework

Fig. 1 delineates the relationships among health-care providers, patients, health insurers and industry in the USA. In this simplified health-care market, doctors choose medical supplies, such as drugs and medical devices, on behalf of patients. On the other hand, health insurers are the main payers of these products. Under most circumstances, insurers provide formularies for doctors to select medications and sometimes require pre-approval for certain medical services. The industry negotiates with health insurers the inclusion of their products in insurers' formularies, as well as prices. Health insurers also set copayment (or coinsurance) and guidelines for these products to be used, which can influence provider and patient decisions.

Figure 1

Relationships among doctors, industry, health insurers and patients in the USA. (i) The interactions of industry, doctors and health insurers are complex and multidimensional. Fig. 1 provides a simplified framework of these three agents. (Sponsor refers to patient's employer, the government or other third party payers who will share the drug cost. Industry refers to the health- care industry, such as pharmaceutical companies, or sometimes pharmaceutical benefit managers who negotiate the drug prices with health insurers on behalf of the industry.) (2) Our model is mainly focused on the interaction of industry, doctors and health insurers. It is also possible that industry can reach patients directly by promotions or advertising, so that patients will require certain medical products from the doctors. The industry promotion strategies may vary by drug quality as well. Insurance companies may also have different strategies of cost sharing to motivate patients to use the products. However, these are out of reach of this study.

The interactions of industry, doctors and health insurers are complex and multidimensional. A typology of three possible scenarios is the following.

Coordinated case

Industry balances the promotion effort between doctors and insurers. Doctors are well informed and have full knowledge of the products advantages as well as their side effects. Health insurers put these products on their formulary and charge reasonable copayments after fair negotiations with the industry. In this case, doctors would appropriately prescribe medical products to patients, and patients would pay reasonable copayment and use these products regularly. Under this coordinated scenario, three agents work coordinately to ensure patients receive the best quality care possible.

Physician-dominant case

Industry put more emphasis on doctors' detailing and promotion. Doctors have strong incentives to prescribe the products; however, health insurers, as the main payer, refuse to pay or shift the financial burden to patients by charging high copayment or coinsurance. In this case, patients may hesitate to purchase the product or have problems to continually use the product given cost concerns. Physicians, thus, may report more problems related to health-care delivery due to insurance rejections or patients' non-compliance.

Insurer-dominant case

Insurance companies are the main decision makers who rank different drugs and medical devices in favorable lists and set copayments after negotiations with industry. Physicians, as the main decision makers on product selection, do not have the incentives to prescribe the newly added medical product. In this case, it is possible that industry-promoted product has low chance to reach patients. By definition, this scenario implies zero or a weak physician–industry relationship.

These scenarios imply three different associations between industry interactions and the health-care quality. In the ‘Coordinated cases’, the physician–industry relationship is positively associated with patients' treatment compliance, if other factors are controlled. On the other hand, in the ‘Physician-dominant case’, the physician industry relationship is negatively associated with patients' treatment compliance, if other factors are controlled. In the ‘Insurer-dominant case’, we may observe weak physician–industry relationship, i.e. its insignificant association with treatment compliance. Thus, the association of physician–industry relationship and treatment compliance is an empirical issue. Using a nationally representative physician sample, we empirically tested this relationship in the US health-care market.

Method

Data and variables

This study used the 2008 Health Tracking Physician Survey, cross-sectional nationally representative sample of physicians in the USA [20]. Physicians were selected from the American Medical Association and consisted of ‘active, non-federal, office- and hospital-based’ physicians who provided direct patient care at least 20 h per week. The response rate was ∼62%. The survey was administered by mail, and the participating physicians received $50 or $75 honorarium.

The survey provided detailed information on physicians' demographic information, practice and patient characteristics. The Health Tracking Physician Survey also provided sampling weights to adjust for the probability of selection and survey non-responses. More detailed information about this survey method is available in its Methodology Report [20].

Dependent variable: measures of treatment compliance

The Health Tracking Physician Survey asked physicians to identify the barriers to provide quality care. The survey question was: ‘Please indicate whether you think it is a major problem, minor problem or not a problem affecting your ability to provide quality care to: (i) rejections of care decisions by insurance companies, (ii) patient non-compliance with treatment and (iii) patients’ inability to pay for needed care.’ The possible answers to each of these questions were one if the response was ‘no problem’, two if it was a ‘minor problem’ and three if it was a ‘major problem.’ We constructed dichotomous variables for each of these questions that equaled one if physicians identified it as a major problem and zero otherwise.

Key independent variables: physician–industry relationship

The specific survey question was ‘Excluding any food, beverages, and drug samples you may have received in your workplace, please estimate the total value of all goods and services you received in 2006 from drug, device, or other medically-related companies. Include honoraria or payments from surveys on prescribing practices conducted by marketing or research firms for medically-related companies.’ Physicians' answers were placed in one of the following categories: 0 = none; 1 = $1–$500; 2 = $501–$1000; 3 = more than $1000.

Other independent variables

We used Reschovsky's [21] conceptual model of physicians' perception of health-care quality for our model specification. This health-care quality model has been adopted in many recent studies [22, 23]. According to this framework, physician demographic characteristics, patients' components and practice characteristics can all affect physicians' perceived quality care. In our empirical model, we controlled the following physician characteristics: gender, race/ethnicity, specialties, years of practice, board certification status, foreign medical school graduate status, practice ownership and physicians' perception of market competition.

Patient components included in our study were patients' race/ethnicity ratios and share of patients with chronic diseases. Practice characteristics included practice types, share of practice revenues from Medicare, Medicaid, and managed care and practice location. We also included state dummies to account for differences on aggregate market effects on health-care quality.

The sample size of 2008 Health Tracking Physician Survey was 4720. Excluding an additional 150 respondents with missing information on physician demographic information, 29 with unknown or missing industry relationship and 85 with missing treatment compliance variables, our final sample size was 4456 physicians.

Analysis

We first presented sample descriptive statistics and summarized the compliance measures by different levels of industry income. We used multivariate logistic regression models to examine the association between the physician–industry relationship and three measures of treatment compliance/affordability individually. All regression models used sampling weights to account for the differential selection probability and to generate nationally representative results. We used Stata 10 to perform the statistical analyses.

Results

Table 1 shows the summary statistics of our sample. Approximately, 50% of physicians reported that rejection of care by insurers was a major problem that affected their ability to provide quality care. Likewise, 39 and 40% of physicians reported that patients' non-compliance with treatment and inability to pay were major problems, respectively.

View this table:
Table 1

Summary statistics of physician characteristics using the 2008 Health Tracking Physician Survey (total sample size = 4456)

Outcome variablesn%
Quality-index measure
 Rejection of care by insurers222950
 Patients non-compliance with treatment175039
 Patients' inability to pay195040
Independent variables
Physician–industry relationship
 Total value received from industry (except drug sample and free food)
 $0226351
 $1–$500143232
 $501– $10003277
 More than $100043410
Female118627
Years of practices445618 years
Physicians' race/ethnicity
 White, non-Hispanic334975
 Hispanic2355
 African-American, non-Hispanic1564
 other71616
Medical school
 USA349979
 International95721
Board certificated401090
Specialty
 Internal medicine59513
 Family/general practice78418
 Pediatrics4039
 Medical specialties122928
 Surgical specialties85119
 Psychiatry2967
 ObGyn2987
Type of practices
 Solo/2 physicians146433
 Group ≥3 physicians176840
 Health Maintenance Organization  (HMO)1543
 Medical school3287
 Hospital based56713
 Other1754
Full/part owner253357
Income from practice of medicine
 Less than $100 00054012
 $100 001–$150 000101223
 $150 001–$200 00091421
 $200 001–$250 00064614
 $250 001–$300 00046610
 More than $300 00087820
Proportion of Hispanic patients (%)445615
Proportion of African-American patients (%)445616
Proportion of Asian patients (%)44565
Proportion of native patients (%)44562
Proportion of patients with chronic diseases (%)445653
Number of contracts from managed care
 None53812
 1–479218
 5–9117526
 10–19127929
 20+67215
Average revenue from Medicaid (%)445617
Average revenue from Medicare (%)445631
Average revenue paid on capitated basis (%)445612
Urban/rural
 Large metro, 1M+ residents267060
 Small metro, < 1M residents131630
 Other47011
Perceived competition
 Not at all110125
 Somewhat214648
 Very competitive120927

As to the physician–industry relationships, 51% of physicians in our sample received $0, 32% received $1–$500, 7% received $501–$1000 and 10% received a total value of more than $1000 from industry excluding free drug samples and food. Most physicians were males, Whites, graduated from US medical schools, reported a medical specialty, had solo or group practice and were board certified. The average number of years in practice was 18. Among the patients, on average, 64% were White, 15% were Latino, 16% were African-American and 5% were Asian Americans. Physicians also had ∼53% of patients with chronic conditions. Most physicians' practices were located in large or small metropolitan areas and had more than five managed care contracts.

Table 2 reports the summary statistics of three outcome measures by different levels of industry income. The percentages of reporting major problems increased with physicians' industry income. For example, when compared with those with $0 industry income, physicians with more than $1000 industry income had 18 and 7% higher likelihoods of reporting rejection of care by insurers and patients' non-compliance with treatment as major problems, respectively (P < 0.001 and P < 0.01, respectively).

View this table:
Table 2

Treatment compliance/affordability measures by industry income

Total value received from industry (except drug sample and free food)$0$1–$500$501–$1000more than $1000
n (%)n (%)n (%)n (%)
Rejection of care by insurers1019 (45)747 (52)***191 (58)***272 (63)***
Patients non-compliance with treatment853 (38)569 (40)131 (40)197 (45)**
Patients' inability to pay939 (41)632 (44)170 (52)***209 (48)**
  • Data source: 2008 Health Tracking Physician Survey. Sample size: 4456.

  • ***P < 0.001.

  • **P < 0.01.

  • *P < 0.05. (The reference group is $0 industry income.).

Table 3 presents the results of multivariate logistic regressions. In the regression of rejections of care by insurer, the odds ratios associated with industry income were 1.24 (95% CI: 1.08–1.44), 1.64 (95% CI: 1.27–2.10) and 1.85 (95% CI: 1.47–2.32), increasing with the amount of industry income. The odd ratio of more than $1000 industry income was 1.34 (95% CI: 1.07–1.68) in the regression of patient's non-compliance with treatment. The odd ratio of $500–$1000 industry income was 1.42 (95% CI: 1.11–1.82) in the regression of patient's inability to pay.

View this table:
Table 3

Logistic regression: the association of industry relationship and physician self-reported compliance measures

Rejection of care by insurersPatients non-compliance with treatmentPatients' inability to pay
OR95% CIOR95% CIOR95% CI
Total value received from industry (except free drug and food)
 $0ReferenceReferenceReference
 $1–$5001.241.081.44***1.060.911.231.030.891.19
 $501–$10001.641.272.10***1.120.861.471.421.111.82**
 More than $10001.851.472.32***1.341.071.68**1.160.931.46
Female0.970.831.140.980.831.151.040.891.22
Years of practices0.980.970.99***0.990.980.99***0.980.970.99***
Physicians' race/ethnicity
 White, non-HispanicReferenceReferenceReference
 Hispanic1.070.781.451.270.941.731.190.881.61
 African-American, non-Hispanic1.130.781.631.170.811.691.491.042.13*
 other1.341.091.64**1.190.971.470.990.811.22
Medical school
 USAReferenceReferenceReference
 International1.291.071.560.930.771.121.180.981.41
Board certificated1.130.911.411.040.831.301.020.821.27
Specialty
 Internal medicineReferenceReferenceReference
 Family/general practice1.130.891.441.090.861.391.271.001.61*
 Pediatrics0.990.711.380.500.350.70***0.620.440.87**
 Medical Specialties1.060.851.320.720.580.91**0.950.761.18
 Surgical specialties1.240.961.590.600.460.78***0.920.721.19
 Psychiatry2.601.883.60***1.200.871.671.761.282.43***
 ObGyn0.810.571.150.750.521.070.960.681.35
Type of practices
 Solo/2 physiciansReferenceReferenceReference
 Group ≥3 physicians0.840.721.00*1.050.881.250.820.690.97*
 Health Maintenance Organization  (HMO)0.190.110.31***1.120.731.720.340.210.55***
 Medical school0.570.420.77***0.950.691.290.890.651.20
 Hospital based0.610.470.79***1.591.222.08***0.880.681.14
 Other0.750.511.111.641.112.42**1.280.881.85
Full/part owner1.130.961.351.000.841.201.251.051.49**
Income from practice of medicine
 Less than $100 000ReferenceReferenceReference
 $100 001–$150 0001.271.001.60*1.260.991.610.960.771.21
 $150 001–$200 0001.090.851.391.140.891.460.930.731.19
 $200 001–$250 0000.980.751.271.250.961.650.850.651.11
 $250 001–$300 0000.770.581.031.381.031.86*0.770.581.03
 More than $300 0000.870.671.130.940.711.240.780.601.02
Proportion of Hispanic patients0.950.601.532.551.594.11***2.201.393.48***
Proportion of African American patients1.070.671.703.472.165.57***1.721.102.70*
Proportion of Asian patients1.450.573.681.060.412.761.960.844.57
Proportion of native patients0.720.321.611.810.824.021.620.743.57
Proportion of patients with chronic diseases1.771.362.31***2.632.003.48***1.791.382.33***
Number of contracts from managed care
 NoneReferenceReferenceReference
 1–41.210.951.551.030.801.331.281.001.64*
 5–91.200.951.511.120.881.431.220.971.53
 10–191.361.081.72**1.210.951.541.531.211.92***
 20+1.541.192.00***1.451.111.90**1.581.222.05***
Average revenue from Medicaid (%)0.960.641.432.031.353.05***0.870.591.29
Average revenue from Medicare (%)0.880.611.250.950.661.370.900.631.28
Average revenue paid on capitated basis (%)0.890.631.251.220.871.710.820.591.15
Urban/rural
 Large metro, 1M+ residentsReferenceReferenceReference
 Small metro, < 1M residents0.980.831.151.211.021.43*0.990.841.16
 Other0.990.781.271.321.021.70*1.381.081.76*
Perceived competition
 Not at allReferenceReferenceReference
 Somewhat1.130.961.331.010.861.191.110.951.31
 Very competitive1.671.382.01***1.211.001.47*1.271.051.53*
Pseudo R20.080.090.06
  • Values are percentages that reflect nationally representative weighted results. Sample size: 4456. The estimated population size = 388 852. All regressions include a full set of state dummies.

  • ***P < 0.001.

  • **P < 0.01.

  • *P < 0.05.

Results showed that physicians' specialty and practice types were significantly associated with the likelihoods of reporting major problems. Results also showed that minority physicians or physicians with more minority patients were more likely to have major problems with these three outcome measures. Physicians with more patients with chronic diseases, with more managed care contracts or who received higher revenues from Medicaid, or who reported more competition, were also more likely to report treatment compliance as a major problem.

Sensitivity analysis

To test the robustness of our results, we applied different multivariate models. We first used ordered-probit models, treating assessments of treatment compliance as multi-category variables. Subsequently, we used ordinary least square models, treating treatment compliance as continuous variables. All these models produced similar results.

Discussion

Our results shed light on the lack of coordination among industry, physicians and health insurers in the USA. If a specific drug (or medical device) is not on health insurance formularies, patients may not be able to afford it or regularly take it. In these cases, physicians may perceive a reduced ability to provide quality of care that reflects an inherent conflict between insurers who emphasize cost-effectiveness criteria and the marketing goals from pharmaceutical companies. Patient's quality care and the perceived ability of physicians to provide it may be compromised under this conflict of interest. According to our conceptual framework, our findings showed that the current health-care market could be closer to the ‘physician-dominant’ case described in our conceptual framework, where physicians perceived barriers to provide high quality care due to the restrictions established by insurers.

In addition, our results showed some evidences that a greater ‘relationship’ with the industry would somehow influence physicians' decisions of the prescriptions. The multivariate results indicated that having greater industry involvement increases the likelihood of insurance company denials of treatment. Previous studies also showed that when compared with physicians with no relationship with the industry, physicians with this relationship would be more likely to request the inclusion of promoted drugs in formularies and more likely to prescribe drugs that were promoted by the industry [9]. Thus, if the industry influence induces physicians to prescribe less standard treatment, the proposed treatment may face higher likelihood of denial.

Nevertheless, insurer's involvement in the physician–industry relationship might reduce physicians' opportunist behavior. If prescribed drugs had a limited therapeutic value or a lower cost-effectiveness ratio when compared with existing drugs or generic alternatives, patient's quality care and financial security might be compromised. The inclusion of more expensive drugs in these formularies might hinder the objective of health-care cost reduction. Thus, health insurers should establish more regulations to limit the pervasive effect that industry advertisement among physicians could have on the quality of care delivered to patients. For instance, health insurers may enhance health-care quality by testing the cost-effectiveness of drugs included in formularies, more restrictive drug utilization reviews, provide more information to consumers, among other actions.

Our results also showed that physicians primarily treating racial and ethnic minority patients were more likely to report patients' low treatment compliance and inability to pay. These findings are consistent with the health disparity literature that minority patients were more likely to have worse health-care quality and lower health-care access [24]. In addition, consistent with previous literature [23], our results showed that physicians with more managed care contract or those who face higher market competitions were more likely to report rejection care by the insurer and patients' low treatment compliance and inability to pay.

Our study had some limitations. First, the physician–industry relationship was self-reported. Physicians might under-report this relationship that was known as ‘social desirability bias’ [25]. Thus, the reported drug income could be considered as a lower bound for the relationship estimated in our models. Second, patients' treatment compliance and affordability were physicians' self-perception and might not reflect the real treatment compliance. However, physicians' self-assessment of quality care may provide valuable insights into the absence of objective clinical information [16]. In addition, although we included all available covariates from the survey, our results were still subject to omitted variables that might be associated with treatment compliance. For example, patients' cultural background, social economic status and other variables could influence their decisions of whether to take medications regularly. Finally, we investigated the association between industry interaction and perceived treatment compliance with a cross-sectional design. With such analysis, it was difficult to infer a causal relationship between industry relationship and physician perceived quality of care.

Despite these limitations, our results underscored the need for policy makers to address this problem. More patients were aware of and showed concerns with physician–industry interaction in the USA [4, 26, 27]. Meanwhile, consumer advocates in Australia [18], UK [15], Turkey [19] and other countries have long demanded details of the financial relationships between doctors and industry. Luckily, some medical organizations have realized this problem and implemented according policy to make the physician–industry relationship more transparent to the public [5, 28, 29] such as Boston Medical Center, the University of Michigan Health System and the Yale University School of Medicine.

As an important component of the ongoing health-care reform in the USA, the Obama administration has recently set new guidelines, such as the Physician Payments Sunshine Act, for drug companies to disclose the payments or gifts they make to physicians for research, consulting, etc. Similarly, the National Health and Medical Research Council in Australia also calls for developing new guidelines to make the industrial sponsorship transparent and credible [15].

Many countries have guidelines developed by different institutions prohibiting companies from giving physician gifts or financial incentives [15], although the standard regulation of physician–industry relationship is still lacking. For example, in the USA, Pharmaceutical Research and Manufacturers of America, and The American College of Physicians, have their own guidelines. These organizations may have different goals, and thus a multiplicity of guidelines may lead to lack of coordination and the adequate definition of an effective framework that both the industry and physicians can observe. It is important that future guidelines should be developed jointly [15] with the aim of reducing the likelihood of conflicts of interest between patients and physicians.

Conclusions

Our study underscored the need for policy makers to address the conflicting goals between the cost-effective criteria used by public and private insurers to reimburse patients and the profitability goals pursued by pharmaceutical and other medically related companies. Patients are in the middle of this conflict, facing barriers to receive and afford appropriate health care. It is important to encourage coordination among doctors, industry and health insurer, to promote patient's quality care. It is the role of public institutions to ensure that these practices do not weaken current efforts to control costs while providing quality and consistent care.

Funding

No funding resource is disclosed.

References

View Abstract