Adherence of Mexican physicians to clinical guidelines in the management of breast cancer: Effect of the National Catastrophic Health Expenditure Fund

Aim To assess the adherence of physicians to the Medical-Care Guidelines for Malignant Breast Tumors in Mexico, before and after the allocation of federal subsidies from the Catastrophic Health Expenditure Fund (FPGC by its Spanish initials) to accredited hospitals, a strategy implemented with the view of offering free treatment to women with breast cancer (BC). Material and methods Based on a cross-sectional design, we gathered information on 479 BC patients who had been attended to at in four FPGC-accredited hospitals. Analysis centered on those treated within either three years before or three years after the accreditation of their attending hospitals. The four hospitals analyzed were located in the North, South, West and Center of the country. Information on all medical procedures performed during treatment was drawn from hospital medical records. Information on the socio-demographic characteristics of the patients was obtained by means of face-to-face interviews conducted in their homes. Results Adherence of physicians to the Guidelines grew by 12.8 percent (from 43.4 to 56.2 percent) after FPGC accreditation (p<0.001) and varied according to the clinical stage of the disease, with much lower levels of adherence observed in the advanced stages (p<0.05). Conclusions The FPGC strategy increased the adherence of physicians to the Medical-Care Guidelines for Malignant Breast Tumors in Mexico.


Setting and study population
We recruited BC patients from four hospitals in the North, South, West and Center of the country. All 32 states in Mexico have at least one FPGC-accredited hospital for BC care. The criteria for selecting the hospitals in our sample revolved around those with the highest possible concentration of BC cases, including facilities in poor states. Three of the hospitals in our sample were located in states accounting for 48.8 percent of BC cases in Mexico; the fourth was located in a state characterized by high poverty levels. All four had been receiving FPGC support since 2007. Overall, these hospitals had treated 1,027 BC patients prior to and 903 following FPGC accreditation; however, given the financial limitations of our study, we confined our analysis to patients residing within a distance equivalent to one hour's drive from the hospitals: of these, 328 had been treated before and 319 after accreditation, 575 (88.9 percent) agreed to participate in the study, and a final sample of 479 (74 percent) were able to provide complete medical records (Fig 1).

Data sources
We conducted face-to-face interviews with women who had been treated for BC in the selected hospitals. The interviews were based on a questionnaire concerning socio-demographic characteristics and were held in their homes. All medical information regarding the management of their disease (clinical stage, type of surgery and adjuvant treatments such as radiotherapy, chemotherapy, immunotherapy and/or hormonotherapy) was drawn from their medical records.

Level of adherence of physicians to the SPSS Medical-Care Guidelines for Malignant Breast
Tumors. Our outcome variable was expressed as a numeric value representing the relationship between the number of procedures undergone by each BC patient in a timely and accurate manner and the total number of procedures stipulated in the SPSS Guidelines for the clinical stage of her condition. This value was estimated within a continuous interval [0,1], where 0 represented 0 percent adherence to the SPSS Guidelines and 1, 100 percent adherence. We used the following equation to construct the level of adherence for each patient: Where, LA = Level of adherence, i = Patient at clinical stage i, j = Procedures required for clinical stage i, and k = Procedures followed accurately and in a timely manner during clinical stage i.
Adherence to the Guidelines denoted the degree of compliance of attending physicians with the SPSS Medical-Care Guidelines for Malignant Breast Tumors [9], according to the clinical stage of the disease in each patient. We constructed an adherence index based on the six medical procedures specified in the SPSS Guidelines: surgery; radiotherapy (2Gy daily for 5 weeks); chemotherapy (four cycles of fluoracyl, adriamycin and cyclophosphamide, followed with paclitaxel weekly for 12 weeks); adjuvant chemotherapy (steroids and antiemetic treatment); hormonotherapy; and/or trastuzumab medication in eligible patients [3,9]. These procedures were assessed according to the clinical stage of the disease and to the time of use, as established by the SPSS Medical-Care Guidelines for Malignant Breast Tumors [9] (see Fig 2). Each procedure was categorized dichotomously as follows: 0 = no adherence was determined when the procedure was not followed, was not included in the SPSS Guidelines, or was performed as specified in the Guidelines but at the wrong time; 1 = adherence was determined when the procedure was included in the SPSS Guidelines and was performed within the indicated time frame.

Independent variable
Our independent variable was FPGC financing provided to accredited hospitals since 2007 as a strategy for offering treatment to BC patients free of charge. Our analytic sample included two groups of women who had been treated before and after 2007.  [23].The following socio-demographic variables were included in our analytical models: age (<40, 40-49, 50-59, 60-69 and �70 years); schooling (none, elementary, junior high school, high school and professional); marital status (consensual union, separated/widow/divorced, married, and single), and paid work (yes/no).

Statistical analysis
We performed an exploratory analysis of the variables of interest and two bivariate analyses of BC patients before and after FPGC implementation: the first bivariate analysis used proportions and χ 2 tests to identify statistically significant differences among the characteristics of the two groups; the second considered the adherence index, estimating means and 95 percent confidence intervals (CI).
In order to facilitate adjustment, and given that our outcome variable (level of adherence) had a value ranging from 0 to 1, we used the parametric beta regression model for variancerobust cluster estimation (four states), as proposed by Salinas-Rodriguez et al. [24]. As this model excludes 0s and 1s, we re-escalated these values, increasing or reducing the value 0.001, respectively. We employed graphic criteria for goodness of fit tests [25] and Cook's distance and leverage points [26] to identify the presence of influential observations; none of relevance to our analysis emerged. Estimates were performed using Stata SE v10.0 [27] and R [28] statistical packages.

Ethical issues
Our study was approved by the Ethics, Biosafety and Research Committees of the National Institute of Public Health, NIPH (Reference Number 774). We obtained prior oral consent from participants for collecting their socio-demographic data by means of a face-to-face questionnaire, for reviewing their medical records and for calling them if further information were required. During the consent procedure, we informed potential participants that the information collected would be handled in a strictly confidential manner and that they were free to refuse to participate or to withdraw from participating in the study at any time. We also assured them that the decision to participate or to refrain from participating would in no way affect their services at the hospital. A card was handed to the patients listing the names and telephone numbers of the President of the NIPH Ethics Committee and the Principal Investigator in the study. The NIPH Ethics Committee determined that it was unnecessary for us to obtain written informed consent given that our study involved minimal risk and did not entail biological sampling or include minors (the minimum age of participants was 29 years).

Results
We recruited a total of 479 women between the ages of 29 and 99 years: 212 had been treated for BC before and 267 after their hospitals were accredited to receive FPGC financing. Their mean ages were 54.3 years (SD 12.1 years) and 56.2 years (SD 13.6 years), respectively. As shown in Table 1, the two groups exhibited similar socio-demographic characteristics: 28% of participants had no schooling and slightly over half were married. The clinical stages of their condition were distributed similarly as well, with the majority of patients (60%) in stages IIB and III. In comparing the levels of adherence of physicians to the SPSS Guidelines before and after 2007, the former came out lower (p<0.001). The proportion of women who received chemotherapy (p = 0.012), trastuzumab medication (p<0.001) and estrogen (p<0.001) or Her2/ neu (p<0.001) receptor tests was greater among those treated after FPGC accreditation. Table 2 shows the relationships between estrogen and Her2/neu receptor tests performed and the therapy specified under the SPSS Guidelines. As can be noted, 6.2 percent of women in the study received trastuzumab medication before and 40.6 percent after FPGC accreditation (p<0.001). Furthermore, several women who obtained negative Her2/neu receptor results or did not undergo receptor testing received trastuzumab treatment although it was not required in the Guidelines.
The number of participants for which medical records showed over 70 percent adherence to the SPSS Guidelines was 18.4 percent and 20.2 percent before and after FPGC accreditation, respectively (S1 Table). The mean level of adherence in the overall study population was 46.2 percent. Before and after FPGC accreditation, the mean levels of adherence were 42 percent and 49.6 percent respectively (p<0.001). For all participants, the level of adherence was higher in the earlier stages of the disease, including cancer in situ (62.5 percent), than in later stages such as stage IV (30.8 percent) (S1 Table).
Beta regression modeling revealed that, prior to FPGC accreditation, the estimated mean level of adherence of physicians, according to the index based on information from the medical records, was 33.2 percent (constant value in Table 3). This increased significantly to 69.3 percent for those treated after FPGC accreditation (p<0.001) ( Table 3). These proportions differ from the means shown in S1 Table because they were adjusted for potential confounders. The estimated mean level of adherence of physicians increased 6.9 times (p = 0.013) in women at clinical stage 0, compared with those reported at clinical stage IV; three times for clinical stages I and IIA (p<0.001), and 63.4 percent (p = 0.131) for clinical stages IIB and III (Table 3).

Discussion
Our results show that adherence of physicians to the Mexican Medical-Care Guidelines for Malignant Breast Tumors increased after accredited hospitals were provided with financing from the Catastrophic Health Expenditure Fund (FPGC by its Spanish initials) as a strategy established by law in 2007 in order to offer free treatment to women with breast cancer (BC). Analysis also indicated that this increase was higher when women were treated in early rather than in late clinical stages. After the implementation of FPGC accreditation, the increase in adherence was most pronounced in the prescription of expensive treatments such as trastuzumab versus other treatments. Our findings suggest that the accreditation of hospitals to receive FPGC financing reduced the economic barriers for BC treatment, improving medical care in certain treatment areas such as chemotherapy and trastuzumab medication, as well as in estrogen and Her2/neu receptor testing, with variance observed among hospitals. On the downside, however, we found that a high percentage of women with positive results for estrogen and/or Her2/neu receptors remained without adequate treatment: 17.9 percent did not receive hormonotherapy and 59.4 percent were not prescribed trastuzumab medication. As reported by other studies, these findings may be attributable to a lack of knowledge among physicians regarding the SPSS Guidelines [7,29]. Additionally, a shortage of supplies and/or medication (e.g. trastuzumab) could be causing a lag in treatment initiation and follow-up [29]. This issue and other FPGC operational problems have been highlighted by various evaluations of the SPSS; among other shortfalls, a delay of up to six months has been identified in the reimbursement of hospital fees by the SPSS to patients treated for BC [7,29]. Our findings indicated that, notwithstanding FPGC financing, only 20 percent of physicians in our sample of hospitals adhered to the SPSS Guidelines at levels above 70 percent. Such performance appears low when compared with similar studies in various countries; in Germany, for instance, studies [14,15,17] have shown that between 15.7 and 67.8 percent of physicians demonstrated 100 percent adherence to the medical-care guidelines, reporting a lower percentage in cases of bilateral BC. Another study in the United Kingdom [13] reported 82 percent adherence to the BC medical-care guidelines, while a study in the Netherlands [12] reported 90.6 percent adherence for patients treated with radical mastectomy and 95 percent adherence for those managed with conservative surgery. It is important to emphasize that none of these studies allowed the recommended time between diagnosis and the beginning of treatment, an omission that could explain differences with our findings. Moreover, the abovementioned studies were conducted in developed countries which may have benefited from more efficient systems for the provision of supplies and pharmaceuticals, a larger care package, or simply, a public health-care system different from the one in Mexico.
In contrast with our findings, Mohar et al. [3] reported that the level of adherence to BC medical-care guidelines on the part of Mexican physicians was over 80 percent. The diverging results may stem from differences in sampling: the women studied by these authors were  [3,21,29]. Additionally, FPGC-accredited hospitals are located within geo-demographically limited areas marked by contrasting ethnic, cultural, economic, political, social, environmental and accessibility factors. The above-mentioned differences highlight the importance not only of standardizing medical-care processes, but also of improving the evaluation criteria for accrediting hospitals [7]. Our results showed that adherence to the SPSS Guidelines varied according to the clinical stage of the disease, with much lower performance observed in the advanced stages. This finding is consistent with those reported by other studies of women with BC [14,15] and other types of cancer [11,16], suggesting that variations in care at the different stages of disease could be an important factor contributing to co-morbidities, lower survival rates and increased side effects from therapeutic procedures. Our study had several strengths such as the comparison of the clinical stages and socio-demographic characteristics of patients treated before and after FPGC accreditation, which revealed similarities between the two groups. A second strength related to our review of the medical records of deceased patients [29], which allowed for analyzing a more representative sample. Nevertheless, the results of our study must be interpreted with caution given that analysis included women treated before and after the FPGC, thus possibly incorporating time-related factors-e.g., other unmeasured programs-as potential confounders. For ethical reasons, we adopted a crosssectional, as opposed to a clinical trial, methodology for the assessment of adherence to the SPSS Guidelines [30].
Regarding external validity, the results of this study cannot be extrapolated to medical practice in other FPGC-accredited hospitals because participation was based on convenience sampling. Nonetheless, as the hospitals in our sample accounted for 48.8 percent of women with BC in Mexico, the treatment of such patients in other hospitals of this type can reasonably be expected to be similar. Our results are consistent with those reported by a nationally representative assessment of adherence conducted in Mexico in 2017 [31]. Based on interviews with 62 oncologists from FPGC-accredited hospitals, this study revealed a lack of knowledge about the areas of care covered or paid for by the FPGC, a situation clearly affecting the quality of care [31]. Also noted as major obstacles to adherence were insufficient infrastructure, equipment, personnel and supplies in terms of inputs and pharmaceuticals. These shortages were attributed to delays in the transfer of SPSS resources, reflecting the presence of complex and inefficient administrative processes [32].
We adjusted our analysis for the most documented confounding factor: the clinical stage of BC in our sample of patients, but not for other well documented confounders such as age [11,12,14,16], schooling [21,22], marital status [21,33], and paid work [21], considering that these were homogenously distributed across both study groups and did not change the level of adherence when included in our regression model. Nevertheless, it is important to explore the influence of other factors associated with the patients (co-morbidities and living conditions); physicians (knowledge about FPGC operational processes and the SPSS Medical-Care Guidelines); hospitals and/or health programs (delay in medical care and/or therapeutic procedures, and availability of supplies and pharmaceuticals); and infrastructure and equipment (adequacy and conditions). These should be taken into account in future efforts to determine the impact of FPGC accreditation on the adherence of physicians to the SPSS Medical-Care Guidelines for Malignant Breast Tumors.

Conclusion
Results from this study suggest that subsidizing the cost of treatment for BC and other diseases at accredited hospitals boosts the adherence of physicians to medical-care guidelines. However, it is essential to streamline the processes for the transference of funds and to implement other innovative strategies in pursuit of greater levels of adherence on the part of physicians. This will contribute to the delivery of timely and efficient medical care according to the guidelines established by the Mexican Health-Care System and result in a higher quality of life and lower mortality rates among women with BC.
Supporting information S1 Table. Level of adherence a to the Medical-Care Guidelines for Malignant Breast Tumors, by characteristics of women treated before and after FPGC b accreditation of the four participating hospitals, Mexico. a The index of adherence was evaluated in a bivariate manner, estimating means and 95 percent confidence intervals (CI) in women treated before and after FPGC accreditation. Adherence level was a numerical value contained in the continuous interval [0, 1], which expressed the relationship between the number of procedures undergone by the BC patient in a timely and correct manner and the total number of procedures specified in the SPSS Medical-Care Guidelines for Malignant Breast Tumors, according to the clinical stage of disease in each patient. b FPGC: Spanish acronym for Catastrophic Health Expenditure Fund. c Means and confidence intervals of adherence levels were expressed as proportions in relation to the study variables and whether the patient was treated before or after FPGC accreditation. (PDF) Writing -review & editing: Carmelita E. Ventura-Alfaro, Leticia Á vila-Burgos, Gabriela Torres-Mejía.