Patient Delays and System Delays in Breast Cancer Treatment in Developed and Developing Countries Introduction

Delays in treating breast cancer have been associated with a more advanced stage of the disease and a decrease in patient survival rates. The scope of this integrative review was to analyze the main causal factors and types of patient and system delays. The underlying causal factors of delays were compared among studies conducted in developing and developed countries. Of the 53 studies selected, 24 were carried out in developing countries and 29 in developed countries, respectively. Non-attribution of symptoms to cancer, fear of the disease and treatment and low educational level were the most frequent causes of patient delay. Less comprehensive health insurance coverage, older/younger age and false negative diagnosis tests were the three most common causal factors of system delay. The effects of factors such as age were not decisive per se and depended mainly on the social and cultural context. Some factors caused both patient delay and system delay. Studies conducted in developing countries identified more causal factors of patient delay and had a stronger focus on patient delay or the combination of both. Studies conducted in developed countries had a stronger focus on aspects of system delay during treatment and guidance of breast cancer patients in the health care system.


Introduction
Breast cancer is the most common type of cancer among women worldwide, with about 1.67 million new cases diagnosed in the year 2012 1,2 .The incidence of breast cancer is still high in developed countries, but the global burden of the disease is progressively shifting to developing countries 1,2 .More than 70% of breast cancer patients in developed countries are diagnosed at stages I and II, whereas in low and middle-income countries, only 20-60% of patients are diagnosed in early stages of the disease 3 .Breast cancer stage represents an important prognostic factor and advanced stage is associated with decreased time of disease-free survival and increased mortality rates 4,5 .The mortality-to-incidence ratio of developing countries tended to be lower, compared to that of developed countries, largely due to the fact that patients have the disease at more advanced stages 6,7 .Furthermore, previous studies have shown that care delay associated with increased stage had a negative effect on the survival of breast cancer patients and were more common among patients in developing countries 3,4,8,9 .
In literature, care delay has been subdivided in patient delay (PD) and health care system (SD) delay 3,10 .According to Lee Caplan (2014), PD is a delay in seeking medical attention after self-discovering a potential breast cancer symptom, whereas SD is a delay within the health care system 10 .PD was mainly defined as a time gap >3 month between symptom detection and first medical consultation 3 .The socio-economic and cultural background of patients can contribute to PD 3,10 .Furthermore, symptomatology experience, ethnic origin, beliefs or perceptions that affect attitudes of patients represent important causal factors of PD 3,[10][11][12] .SD can refer to access barriers, like long distance to treating health care centers, or no availability of specialized centers and intrinsic problems of an established health care system, like disease management, problems in obtaining or scheduling diagnostic tests and communication problems between patients and physicians 3,10,11,13 .Furthermore, to reduce SD, health services must fit with the socio-economic and cultural or ethnic background of patients 3,11 .
Despite the existence of vast literature about breast cancer care delays only few authors have compared studies conducted at different regions of the world.Review articles are mainly focused on care delays and its outcomes in developing countries 3,8 .Little is known about differences of study objectives and the causal factors of PD and SD, identified by authors of distinct regions of the world.The present integrative review addressed on four main questions: 1.What are the main causal factors of PD and SD in literature? 2. What are the methods applied to analyse the contribution of causal factors on PD and SD? 3. Which types of SD are analysed in literature? 4. If care delays and their adjacent causes are different among developing and developed countries.

Method
An integrative review of care delay associated with breast cancer was conducted.The research methodology was performed according to established criteria 14 .The operational steps were conducted as followed: Definition of exclusion and inclusion criteria and database; selection of studies; extraction of information and categorization; Analysis and interpretation of data; Interpretation of results; Synthesis and resume of results.The search was conducted in the following databas es: Literatura Latino-Americana e do Caribe em Ciências da Saúde (LILACS); National Library of Medicine (MED LINE), PubMed and in the repository Scientific Electronic Library Online (SciELO).
Terms "treatment delay", "provider delay", "system delay" and "patient delay" are commonly used in public health literature, but were not identified in Medical Subject Headings (MeSH; http://www.nlm.nih.gov/mesh/MBrowser.html) of the U. S. National Library of Medicine.Terms "breast cancer" and "delayed diagnosis" were selected from MeSH.A much more common term in literature was "diagnostic delay".As shown in the results and discussion section, the term "diagnostic delay" was not very well defined.To identify a broader range of articles about any kind of care delay, the term "delayed diagnosis" was substituted by "delay".Therefore, the term "breast cancer" was used in combination with "delay" in English, Spanish ("Cáncer de mama" and "atraso") and Portuguese ("Câncer de mama" and "atraso").
The search was conducted on October 22, 2014.To establish an internal quality control, all procedure of literature selection was performed twice independently by each of the two authors.Application of search terms in MEDLINE led to the identification of 2390 articles.When filter "10 years" was applied, this number reduced to 731 articles.Additional application of filter "free full text" led to the identification of 299 articles.In the SciELO database, 18 articles were identified and in the LILACS database, no article that met criteria of both search terms in any of the three languages was identified.
Finally, 317 articles were used for further analysis.To identify articles that met selective criteria, title and abstract were analysed.The following inclusion criteria were applied: original research article or research communication, available for free and published over the last five years, in English, Spanish or Portuguese.Articles identified in more than one database, articles about breast cancer that did not deal with care delays and review articles were excluded from the study.For further method categorization, result and discussion sections were read and analysed.According to previous studies, PD was defined as a delay of medical consultation caused by the patients behaviour 3,10 .The term SD was applied in the case of any delay caused by the health care system.In the case of factors associated to PD and SD, only factors that were identified two or more times were included and factors identified only by one study were excluded.This was justified by the fact that in the case of factors like "smoking" that was identified only one time by a single study, authors did not explain any possible causal relationship with PD and/or SD.
All statistical analysis was performed on Prism™ software vers.6 (GraphPad, La Jolla, California, USA).The χ 2-Test was performed to analyse categorized variables.

Results and discussion
All together 53 studies were identified that originated from 22 different countries (Table 1).Of the 53 studies, 24 were from developing and 29 from developed countries, respectively (Table 1).Single studies originated from Australia, Denmark, Estonia, China (Hong Kong), Iran, Libya, Nigeria, Pakistan, Poland, South Korea, Thailand, Turkey and Egypt (Table 1).More than one study was carried out in Brazil (N = 2), Cameroon (N = 2), Tunisia (N = 2), India (N = 2), Colombia (N = 3), United Kingdom (N = 3), Malaysia (N = 6) and the USA (N = 21).With exception of one Brazilian study written in Portuguese, a Colom-bian study written in Spanish and a study carried out in Tunisia written in French, all the other 50 studies were written in English (Table 1) [16][17][18] .
Many contributions in the form of articles, such as those carried out in Malaysia and USA could indicate a greater interest of researchers in delay issues in these countries.Alternatively, the distribution of contributions by countries may not reflect the real frequency of publications within the last five years: Contributions were not included in the present research if they were not published in open access journals or if they were published in journals that are not registered in the examined databases.Therefore, exclusion criteria could have led to a bias in the number of contributions from single countries.This is a serious limitation of the present study and results should be interpreted with care.

Methodological background of studies
The most prominent methodological approach (N = 28) to obtain data about PD and SD was interview (Table 1).Of these 28 studies, 20 were exclusively based on interviews, whereas eight studies used additionally medical records or other registered data (Table 1).Several qualitative studies (N = 7) including small sample numbers, < 20 persons applied in-depth-interview (Table 1).Furthermore, of these 28 studies, four were based on telephone interviews (Table 1) [19][20][21][22] .Questionnaires were applied in five studies and in one case, it was combined with registered data (Table 1) [23][24][25][26][27] .Of the 53 studies, 18 collected data exclusively of medical records or other source of recorded data (Table 1).Several of these authors obtained increased numbers of data from breast cancer patients, that varied from 21.818 to 147.682 by the usage of large data bases as source for sampling (Table 1) [28][29][30][31] .One study was based on lecture of media and the other on discussed theories, concepts and models of care delay without sampling data (Table 1) 32,33 .
Other studies (N = 12) applied univariate methods, mainly using χ 2-Test or Fisher's exact test to analyse categorized variables and the t-Test to analyse continuous variables.Máslach et al. 35 , for example, compared time delays between patients from rural and urban regions.Crowley et al. 57 compared the time gap from diagnosis to start of adjuvant hormone therapy among patients of Afro-American, Hispanic and Caucasian ethnicity.In these studies, each factor potentially contributing to care delay was analysed as a univariate variable, regardless of all the other study variables.To identify independent factors and to establish a model that could explain PD or SD, most authors (N = 23) adopted multivariate regression models (Table 1).Multivariate logistic regression models were applied on dichotomized data comparing two groups: Ghazali et al. 51 for example, determined PD as a time gap of > 3 month from symptom discovery to first medical consultation.The authors first categorized patients into two groups, those with delay > 3 month and those with no delay (≤ 3 month) 51 .After identification of significant factors by univariate analysis (χ 2-Test), a multivariate logistic regression model was established to explain PD based on martial status (divorced or single vs. married), ethnicity (Chinese vs. Malay or Indian) and breast self examination (not performed vs.  51 .These approaches were exemplary for most studies aimed at identifying the main factors that affect PD and/or SD and multivariate regression models were standard to identify independent causal variables of PD and SD.
The association between advanced stage of breast cancer and care delays is well established in literature 3,10 .In several studies, PD was associated with or identified as a causal factor of advanced stage of breast cancer 26,31,40,[49][50][51]53 . In  study carried out in Tunisia, PD and also long distances to health care centres contributed to advanced stage of the disease 16 .Soares et al. 44 , showed that SD, caused by increased length of time between clinical suspicion and diagnostic confirmation, was associated with advanced stage of breast cancer.
Most studies have focused on breast cancer patients for sampling, but there were also some exceptions: In a study carried out in Malaysia, the authors investigated the preference of patients for western or traditional medicine and interviewed 11 breast cancer survivors 52 .Telephone interviews were applied to 6965 women in England, who were not breast cancer patients 19 .The authors used a validated questionnaire named "Awareness and Beliefs about Cancer Measure (ABC)" to analyse causal factors of PD 19 .Black and Woods-Giscombé 33 , used the narratives of non-patients to apply a gender-specific, culturally responsive stress process framework to identify the reasons why Afro Americans, compared to women of other ethnic groups, have more often PD.
These publications show that sources for sampling data were heterogeneous.Exceptionally in this context was also the study by Schairer et al. 46 , who investigated if patients with inflammatory breast cancer, compared to those with other types of breast cancer, have a different help seeking behaviour.The authors used registered data and did not find any significant difference 46 .

Different types of SD and the terminology of diagnosis delay
The different types of SD were analysed by the authors and summarized in Chart 1.In the The most common form of SD analysed was the time gap between first medical consultation and start of any type of therapy (Chart 1) 17,23,24,48 .Other studies focused on more specific SD attributed it to different phases of breast cancer treatment, for example between diagnosis and chemotherapy, hormone therapy or surgical treatment (Chart 1) 28,57,58,63 .Yun et al. 28 defined in their study "surgical treatment delay" as the time gap between diagnosis and first surgical treatment.
Different definitions were applied for the term "diagnostic delay": Several authors defined diagnostic delay as the time gap from recognition of first symptoms to histological diagnosis 21,40,54 .In this case, there was no clear distinction between PD and SD by the time gap analysed, but underlying causal factors such as access barriers or patient attitudes have been attributed either to PD or SD.Other studies have defined diagnostic delay as the time gap between first presentation and final diagnosis 36,66 , or suspected breast cancer and confirmed diagnosis 44 .Finally, "diagnostic delay" was also defined as the time gap from abnormal mammogram to diagnostic resolution 67,68 .Other terms also applied were "provider delay" 41,48 , "doctor delay" 53,56 and "treatment delay" 28 .In general, there is no standardized nomenclature, as identical terms have distinct meanings and may refer to different time gaps.

Study objectives differ quantitatively and qualitatively between developing and developed countries
Of the 53 studies, 19 and 15 focused exclusively on PD or SD, respectively, whereas other 19 studies analysed both PD and SD (Table 1).All together, studies exclusively focused on SD were more frequently carried out in developed countries (N = 12) compared to developing countries (N = 3), whereas in the case of studies more focused on PD or both, PD and SD, the difference between developed (N = 17) and developing countries (N = 21) was smaller (p = 0.0313).Five out of six studies, that attributed poor knowledge or no information about breast cancer to affected women, were from developing countries 34,42,47,55,56 (Chart 2).The basic assumption that women are not well informed about breast cancer may lead authors from these countries to develop a strong-  16,20,27,36,37,39,40,45,53,56,59   Chart 1.Time gap of system delay (SD) analysed by different authors.References of studies from developing countries were highlighted in italic.er focus on PD or PD and SD compared to authors from developed countries.
Identified causal factors of PD and SD were summarized in Table 3 and Table 4, respectively.Overall, in 53 studies, 208 factors were identified two or more times (Chart 2 and Chart 3).Of these 208 identified causal factors, 124 contributed to PD and 84 to SD (Chart 2 and Chart 3).The number of identified factors that contributed to PD was higher for studies carried out in developing countries (N = 74) compared to developed countries (N = 50) and in the case of SD, the opposite was observed: More factors were identified in studies carried out in developed countries (N = 57), compared to developing countries (N = 27; p = 0.0001).This underlines the strong in-terest of authors from developing countries to better understand the reasons of PD.

Causal factors of SD
Poor insurance status and insurance authorization delays (N = 7), older or younger age (N = 7), false negative diagnostic tests (N = 6) and longer distance to health care centre (N = 6) were the four most common identified causal factors of SD (Chart 3).In the case of older patients increased time gap between diagnosis and adjuvant chemotherapeutic treatment was explained by postoperative complications and diagnostic or therapeutic interventions 29,63 . Vandergrift et al. additionally showed that missing supplemental insurance of Afro-American women also increased this time gap 63 .Authors of studies carried out in Colombia and Thailand argued that poor insurance status was associated with delayed diagnosis and referral for specialist treatment 43,48 .
The factors comorbidity, communication, hospital volume, preoperative components, comparison of disease management program, several histopathological characteristics and ethnicity were exclusively identified to cause SD by authors of developed countries (Chart 3).Studies of developed countries had generally a stronger focus on the performance of the health care system and patient navigation during the treatment process.

Effects of several factors depend on the context, but are not decisive themselves and some factors can cause PD and SD
Long distances to health care centres and rural vs. urban areas were identified in several studies to cause SD as they are relevant access barriers (Chart 3) 16,23,27,41,45,63 .In a study carried out in Poland, the opposite was observed, as women living in urban areas suffered from SD 35 .In this case, the authors argued that physicians concentrate treatment procedures for patients from rural regions to reduce the number of visits.Similarly, comorbidities or clinical-histopathological characteristics like tumour stage and hormone receptor status influenced decisions of physicians in different context-dependent ways (Chart 2).
Lobb et al. 68 , pointed out that improved communication between patient and physician can reduce health system barriers.Sheppard et al. 22 identified in their study an ambivalent nature of better communication that was dependent on the ethnic origin of patients (Chart 3):Better communication between patients and physicians led to decreased SD among Afro-American patients and to increased SD among patients of Caucasian origin 22 .According to their interpretation, Afro-American women may have relied on provider suggestion, whereas women of Caucasian origin with higher income and education rather seek information outside the patient-provider relationship to make their decisions 22 .
Several studies have identified older age as an causal factor of PD (Chart 2) 19,27,38,40,49,62 .In their study, Inos et al. 38 , pointed out that older women tend to underestimate the risk of breast cancer as breast cancer screening programs are mainly addressed to women aged 50-62 years.This is in agreement with findings of Brazilian studies that revealed decreased numbers of elderly women participating in mammography screening programs 69,70 .Similarly, in a study carried out in Lybia, PD was attributed to older aged women with poor knowledge about breast cancer 40 .In contrast, a study carried out in Turkey, the authors showed that young women aged 30-39 years more often ignored breast cancer symptoms compared with older patients 23 .Ozmen et al. 23 , also identified in their study decreased SD among patients aged > 60 years.Their interpretation was that older patients are prioritized by physicians and receive a faster diagnostic process.Similarly, Bleicher et al. 30 , identified increased preoperative delay in patients aged 65-79 years compared to patients > 80 years.Interestingly, authors of a study based on 21,818 breast cancer cases, argued that symptomatic presentation of disease was more common in younger women 31 .In this case young age ≤ 40 years, was a significant variable of PD in univariate analysis, but did not represent an independent variable of the final regression model.
In several cases, the authors did not explain why a determined factor had a unexpected effect: While in two studies, low family income was associated with PD, in a study carried out in Thailand, the opposite was verified and authors did not represent a causal argument of their result (Chart 2) 24,43,49 .
As shown above in the case of age, some factors caused PD and SD.Lower educational level and lower family income were also associated with PD and SD (Chart 2 and 3).Lower educational level and literacy status were the most often cited factor to cause PD (Chart 2).A Colombian study revealed that SD was not only associated with poor health insurance status, but also with low family income and educational level 48 .Fedewa et al. 29 , used a large database to compare the time gap between surgical resection of the primary tumour and initiation of chemotherapy among patients from different regions of the USA.Their study revealed that SD > 90 days was not only more prominent among Afro-Americans, compared to patients of Caucasian origin, but that proportionally more often patients from regions with low education levels, measured by the frequency of school diploma, also suffered from SD 29 .The authors reported that lower socio-economic and educational status can also difficult navigation through the health care system 66,68 .
These studies showed that several factors are not decisive, but depend on patient's social and cultural environment and on differences among health care systems and their personal decisions.Furthermore, several socio-economic factors have an influence on both PD and SD.

Final considerations
Studies carried out in developing countries more often focus on the general aspects of SD.In contrast, studies carried out in developed countries analysed more often a determined time gap of treatment.In general, authors from developing countries are more interested in PD or PD and SD combinations, and identified also more causal factors of PD compared to studies carried out in developed countries.In contrast, studies carried out in developed countries more often focus on specific aspects of SD.The stronger focus on PD in the case of studies carried out in developing countries is probably influenced by the author's assumption that patients in their country are not well informed about breast cancer.However, this does not necessarily mean that women from developed countries are really well informed.One should always keep in mind that questions of studies and their objectives are not only influenced by the social and cultural reality of researchers, but that there also can exist underlying "fashions of research" that differ timely and locally among countries.
The present review study showed that methodological approaches of studies were highly different.Interviews of few persons contrasted with the analysis of large databases, and while some studies were purely descriptive, others applied multivariate regression models to identify a reduced number of independent variables.Neither methodological approaches nor nomenclature about PD and SD are standardized.
Socio-economic factors, symptomatology experience and attitudes of patients contributed to PD. Socioeconomic and ethnic differences affected both PD and SD.Additionally, diagnostic factors, factors that affect decisions of physicians, access barriers, communication between patient and physician, disease management and histopathological characteristics of disease also affected SD.Some of these factors were not decisive, but mainly dependent on the social and cultural context.