This article has Open Peer Review reports available.
Ethnic inequalities in time to diagnosis of cancer: a systematic review
© Martins et al.; licensee BioMed Central Ltd. 2013
Received: 30 August 2013
Accepted: 18 December 2013
Published: 23 December 2013
Minimising diagnostic delays in cancer may help improve survival. Ethnic minorities have worse outcomes in some cancer types when compared to the majority; this may relate in part to differences during the diagnostic phase. Only a few British studies have specifically explored this relationship, and no synthesis of these exists. The present study aimed to systematically review evidence on ethnic inequalities in cancer diagnosis, focussing on patient and primary care intervals of diagnosis.
Six electronic databases were searched. Included studies were those conducted in the UK or elsewhere (where access to healthcare is comparable to the NHS) and those that described a time element during diagnosis. Study quality was evaluated using the Critical Appraisal Skills Programme (CASP) checklist for cohort studies and synthesis method was narrative.
Seven of 8,520 studies retrieved by our search met the review criteria; six conducted in the UK, and one in New Zealand. Five (including one covering several sites) focused on breast cancer, one on prostate, and one on oesophagogastric cancer. The studies employed different methods of ascertainment and definition of ethnic groups and defined diagnostic delay in a non-standardised way; therefore, narrative synthesis was performed. In breast cancer, three studies reported longer diagnostic intervals among ethnic minorities and two found no evidence of differences by ethnicity. There was some evidence of longer diagnostic and referral intervals among ethnic minorities in oesophagogastric and colorectal cancers, but no evidence of this in prostate, non-Hodgkin’s lymphoma, lung, and ovarian cancers. None of the studies identified shorter patient or primary care intervals in ethnic minorities.
Existing studies provide insufficient evidence to confirm or refute ethnic inequalities in diagnostic intervals of cancer. Further studies are necessary to examine common cancer types including those frequently found in ethnic minorities (in addition to those covered here) and using current definitions of intervals in cancer diagnosis.
Minimising delays in cancer diagnosis - particularly, patient and primary care delays - may help improve cancer outcomes in the UK, which at the moment lag behind most other countries in Europe [1, 2]. Recent studies estimate that 5,000 to 10,000 cancer deaths could be prevented annually in the UK if efforts to expedite diagnosis succeed [1, 2]. As some ethnic minorities suffer more illness and report worse experience of the National Health Services (NHS) , it is possible this will extend to cancer diagnostics, contributing to the UK’s poor outcomes. Few UK studies have specifically explored this relationship, with most evidence coming from the US . The findings of these studies may not be generalisable to the UK, given the differences in the organisation and operations of the US and UK health care systems in addition to the differences in nature and composition of ethnic groups.
Access to health care in the UK is universally free at the point of delivery with an ethos of equity in service provision. In theory, this should reduce any disparity in cancer diagnostics. There is, however, a greater incidence of advanced-stage diagnoses of female breast cancer and prostate cancers among ethnic minority groups compared to the majority [5, 6], although these studies did not report an explanation for their findings. The reason for such disparities in disease stage is unclear given that pre-referral consultation rates are markedly higher among ethnic minority groups with cancer . It is unknown whether this is the case for other cancer sites, including those found to be more common in ethnic minority groups - e.g., mouth, liver, stomach, oesophagogastric and ovarian cancers [8, 9].
Aim of review
Review search terms
Terms relating to cancer : Cancer, Neoplasm, Malignant Neoplasm, tumour, Malignant tumour, Astrocytoma, Adenocarcinoma, Glioma, Mesothelioma, Medulloblastoma, Myeloma, Melanoma, Neuroblastoma, Sarcoma, Nonmelanoma, Osteosarcoma, Teratoma, Seminoma, Hodgkin, leuk?emia, Lymphoma, Retinoblastoma
Terms defining ethnic minorities in the UK: Ethnic* Race* Cultural groups, white Irish, other whites, African British, Black* Black Caribbean, British Caribbean, Asian*, Indian, British, Indian, Pakistani, British Pakistani, Bangladeshi, British, Bangladeshi, Chinese, Mixed race
Ethnic majority in the UK: Ethnic* Race* Cultural groups, White British
Terms relating to intervals of cancer diagnosis: Duration of symptom; Interval of symptoms; Time; Delay; Late; Postpone; and Wait to Symptom, Presentation, Attendance, Consultation, Appointment, Diagnosis, Detection, Treatment, Intervention, Referral. Other terms; Gate keeping, primary care
Given the methodological heterogeneity of the studies, a quantitative synthesis was not possible, so we performed a narrative synthesis, using the framework of Rodgers and colleagues . Data from included studies were extracted, collated and tabulated by a single extractor (TM). The primary outcome was the interval between first symptom experience and eventual diagnosis, subdivided where possible into patient delay and primary care delay , with the main explanatory variable being ethnicity. The only subgroup analysis planned was by cancer site; however, the heterogeneity of the studies precluded formal meta-analysis, so we performed our narrative synthesis using the cancer sites stratified by diagnostic intervals studied.
Quality of studies
Was the cohort representative of a defined population?
Was everybody included
Was the exposure accurately measured to minimize bias?
Was the outcome precisely measured to minimize bias?
Have the authors identified and adjusted for all key confounding factors?
How precise are the results?
Were all eligible cancer patients studied?
Was ethnicity defined according to contemporary groupings?
Was diagnostic intervals measured to Olsen and colleagues framework?
Unchanged. Key confounders include: age, gender, SES, co-morbidity, healthcare system, family history of cancer and tumour growth rates
Have they presented estimates of association along with the confidence intervals? Are the confidence intervals narrow?
Rajan et al. (2011) 
Meechan et al. (2002) 
Velikova et al., (2004) 
Nosarti et al. (2000) 
Neal and Allgar (2005) 
Sadler et al. (2009) 
Metcalfe et al. (2008) 
Characteristics and findings of reviewed studies
Sample size (n) gender age
Relevant outcome measures
Rajan et al. (2011) 
UK (West Yorkshire)
To improve knowledge about late presentation and management of breast cancer among South Asian women.
n = 1,630 (36 South Asian women) all female median age = 53.5 years
Retrospective: Breast cancer waiting list database and case notes
South Asian women; Indians and Pakistanis
None; throughout the paper, including the title, the authors imply South Asian women had more delay
Duration of breast symptoms prior to presentation within primary care
45% of Asian women delayed symptoms beyond 8 weeks before visiting their GP
Meechan et al. (2002) 
Delay in seeking medical care for self-detected breast symptoms in New Zealand women.
n = 85 all female mean age =38 years
Retrospective: questionnaire & patient record
Minority New Zealanders - Maori, Pacific & Asian/Indian
European New Zealanders
No difference in patient delay by ethnicity.
Velikova et al. (2004) 
UK (South Yorkshire)
To describe the effect of ethnicity on tumour stage, treatment, patient and providers delays in diagnosis of breast cancer
n = 16,879 all female mean age = 49.7 years in Asians and 62 years in non-Asians
Retrospective: Cancer registry data
Patients and providers delays to diagnosis
After adjusting for, age, SES and health care settings; patient delay was longer in Asian than in non –Asian women (median of 61 days vs. median of 31 days, P = 0.005)
Nosarti et al. (2000) 
To identify factors associated with delay in presentation and assessment of symptomatic breast cancer
n = 692 all female median age = 49 years
Retrospective: Interview, GP & hospital records
African Afro-Caribbean and Asian
British and other white
Patient and system delay
Ethnicity were non-contributory to patient delay in breast cancer
Neal and Allgar (2005) 
To explore the relationship between socio-demographic factors & delays in the diagnosis of six cancers
Breast, lung, colorectal, prostate NHL, and ovarian
n = 65,192 male & female all age groups
Retrospective: Analysis of the National Survey Data
Blacks - Africans, Caribbean & others blacks. South Asians -Indian, Pakistani, Bangladeshi, others.
Total, pre-hospital, referral and secondary care delay
After adjusting for marital status, gender, age and SES, Asian and black had longer pre-hospital delays for breast cancer in women (P = 0.001) and longer referral delay for colorectal cancer (P = 0.02). No evidence of difference for lung, prostate, NHL and ovarian.
Sadler et al. (2009) 
The effect of ethnicity on the presentation and management of oesophageal and gastric cancers: A UK perspective
oesophageal & gastric
n = 244 male & female median age = 71 years
Retrospective: Case-note audit
Asians and Blacks
Referral routes and total diagnostic interval
Asians and Blacks compared to Caucasians were less likely to be diagnosed within 3 months of symptom discovery (P = 0.03) and less likely to take the optimal route to diagnosis (p = 0.01).
Metcalfe et al. (2008) 
To examine the pathways followed by black and white men to prostate cancer diagnosis
n = 1,866 men median age = 67.9 years in blacks 73.3 years in whites
Retrospective: Questionnaire, hospital records and cancer registry data
Delay between onset of symptoms and first GP presentation.
After adjusting for age and hospital centre, no significant difference between white and black men in patient delay (odds ratio: 0.82; 95% CI: 0.57 to 1.19)
Five studies [14–17, 20] investigated ethnic differences in patient delay, that is, delay occurring in the interval between first symptom and first GP presentation (see Figure 1). Two of these [16, 17] also examined ethnic differences in delay following the first GP presentation, but their definitions of these time periods differed. Nosarti et al. defined delay occurring in the interval between the first GP presentation and first specialist visit as system delay . Velikova et al. defined the interval between first GP referral and the first hospital visit as provider’s delay . Neither of these correspond with current definitions . The remaining two studies took slightly different approaches in their definition of delay intervals. Sadler and colleagues used patients’ referral routes to the hospital as surrogate measures of primary care delay . They considered two-week waits (2WW) or ‘Direct Access Oesophago-gastro-duodenoscopy’ (OGD) as optimal routes to diagnosis, while waiting for routine appointments and acute admission were both considered sub-optimal. In addition, they reported ethnic differences in total diagnostic interval, which is the interval from first symptom to diagnosis. Similarly, Neal and Allgar reported the same interval, naming it pre-hospital delay . For the rest of this paper, we have renamed these ‘the diagnostic interval’. They also reported ethnic differences in referral delay, which they defined as the waiting time from first GP visit to first hospital appointment.
Assessment of study quality
Overall, none of our seven studies had fewer than two concerns and none could be classified as “high quality” due to the systematic flaws in their methodology - as highlighted by the CASP assessment shown in Table 2. Only one study  had relatively low risk of bias, and was classified as “medium quality”. It had a large sample size, investigated several cancer sites, and classified ethnic groups in a way that was applicable to their target population. However, they only reported ethnic differences in pre-hospital delay, which includes factors relating to the patient, primary and secondary care and health care system making it difficult to ascertain where differences have arisen from. The remaining six studies had a greater overall risk of bias and were evaluated as “satisfactory”.
Ethnic differences in delay intervals by cancer sites
Of the five studies that investigated ethnic differences in breast cancer diagnosis, two [16, 18] found evidence of longer intervals in ethnic minority groups compared to the white majority. Neal and Allgar reported that ethnic minority women of Asian and black origin were more likely to experience longer diagnostic intervals, even after adjustment for marital status, gender, age and socioeconomic status (SES) (P < 0.001) . Similarly, Velikova et al. found longer patient delay in Asian women compared to non-Asian (median of 61 days vs. median of 31 days, P = 0.005) even after adjustment for SES, age and health care settings . Another study, Rajan et al., found that nearly half (45%) of the Asian women in their West Yorkshire (UK) cohort delayed over two months before presenting with breast symptoms . They labelled this as ‘delay’ but without providing data for non-Asians. The remaining two studies [15, 17] found no difference by ethnicity.
Sadler and colleagues found evidence that Asians (13%) were less likely to take the optimal route to diagnosis (OGD and 2WW) compared to blacks (49%) and Caucasians (39%); P = 0.01 . They also found that Asians (39%) and blacks (45%) were less likely to be diagnosed within three months of symptom discovery compared to Caucasians (63%); P = 0.03).
Neal and Allgar reported longer referral delay among Asians and Blacks even after adjustment for marital status, gender, age and SES (P = 0.02) . No difference by ethnicity was found in pre-hospital delay.
Other cancer sites
Health service provision must be culturally sensitive if health inequalities are to be reduced and eliminated. This is important in the UK (and other countries) as it becomes even more multicultural with considerable increases in ethnic minority population. Few UK studies have specifically explored ethnic inequalities in cancer diagnosis, and to the best of our knowledge no synthesis of evidence exists.
Strengths and limitations
Our rigorous search strategy, and explicit inclusion/exclusion criteria, quality assessment of included studies and narrative synthesis followed best practice. Our search identified only a small number of studies, the bulk of which were (unsurprisingly) conducted in the UK, given that the UK is one of the most ethnically diverse countries in Europe. Our decision to omit American studies (other than Canada) was based on their very different health care system and ethnic groupings; it considerably reduced the number of selected studies, although this reflected our research question of whether there are ethnic inequalities in the time to diagnosis of cancer within a free health care system. A limitation was that most studies focused on breast cancer, reducing the scope of the review. Furthermore, any review is only as good as the studies it finds. Our selections differed considerably in methodology - especially in the definitions of intervals before diagnosis, which made quantitative synthesis inappropriate, and complicated the narrative synthesis. Participants’ ethnic groups were poorly defined in some studies; at times there was a mismatch between the ethnic groupings used for the study and contemporary ethnic groupings used in the whole population from which their sample was drawn [16, 20]. We interpreted the CASP quality instrument for this review - primarily to fit our study topic: it is unlikely these minor changes led to more studies ‘failing’ our quality assessment. Finally, publication bias is possible as studies showing no association between ethnicity and cancer diagnostic delays may have failed to be published, leaving us with disproportionately positive studies.
Comparison with previous studies
Two previous reviews [21, 22] have examined delays in cancer diagnosis, although neither examined ethnicity. In both (like in the present review), the intervals were defined in a nonstandard way. This is understandable, as terminology in this subject was ill-defined before the advent of the Aarhus statement . This provides more precise description of milestones along the cancer diagnostic pathway, as well as describing preferred research methodologies. In an assessment of primary care use prior to cancer diagnosis, Lyratzopoulos and colleagues found that ethnic minority groups were more likely to have consulted their GPs three or more times before hospital referral (odds ratio for Asian vs. white 1 · 73, 1 · 45 to 2 · 08; p < 0 · 0001; odds ratio for black vs. white 1 · 83, 1 · 51 to 2 · 23; p < 0 · 0001. This study did not report durations or intervals to diagnosis, so would not have met our inclusion criteria, though its results are consistent with our findings.
We found limited, methodologically-weak evidence for ethnic inequalities in cancer diagnosis, though largely for breast cancer. Conversely, the review found no evidence to suggest that ethnic minority groups were doing better at any stage of cancer diagnostic pathway. For breast cancer, three (of five studies) found longer patient and pre-hospital delays in ethnic minority women compared to their white counterparts; while the remaining two reported no difference. The latter two studies sampled participants from a relatively small number of women referred to a specialist clinic for further investigation. It is well-recognised that cancer patients may take alternative routes to diagnosis, such as routine specialist clinics or emergency admission; these will have been omitted from the two studies based in specialist clinics [23, 24]. Furthermore, both studies were based on interviewing patients about their delay experiences whilst in the specialist waiting room. Therefore, their findings may have been influenced by selection and recall biases. In contrast, the three studies finding differences in diagnostic intervals across ethnic groups recruited participants with definite diagnosis of breast cancer, and then surveyed or examined their medical record for important dates of presentation and referral.
Redressing inequalities in health (including those relating to ethnic diversity) is central to most reforms and strategies on health and social services in the UK. However, as the government progresses with the implementation of its austerity measures (with considerable cuts to health care expenditure) there are concerns that providers may prioritise differently. Potentially, some services including those intended to meet the specific health care needs of ethnic minority communities may be compromised - e.g., tailored services for language and communication needs, and the need for gender-specific providers. To ensure that policies are tailored to need, sufficient evidence is crucial. If existing evidence suggests ethnic variations in diagnostic delay, then policies aimed at promoting early diagnosis should also target the affected groups.
This review found some evidence for ethnic differences in cancer diagnosis with ethnic minorities taking longer to diagnosis. The evidence was largely in breast cancer patients. However, these findings are from methodologically heterogeneous studies, with different measures of delay. The studies had designs that are prone to bias. Future studies will need to examine the subject using a clearly defined and representative sample of ethnic minority groups, incorporate cancer types that are more common in ethnic minority groups and employ the definitions of cancer diagnostics in the Aarhus statement. This should be possible with the advent of large linked datasets containing routine healthcare data (e.g., the Diagnostic Imaging Dataset which incorporates ethnicity as a variable).
We thank Professor Richard Neal for his contributions to the search terms for cancer and delay intervals. We also acknowledge Catherine Stabb for helping with data extraction and study selection. Members of the Peninsular College of Medicine and Dentistry Synthesis team helped review the search strategy. OU is supported by the Peninsula Collaboration for Leadership in Applied Health Research and Care, a collaboration between the University of Exeter, University of Plymouth, and National Health Service South West, funded by the National Institute for Health Research.
The Policy Research Unit (PRU) in Cancer Awareness, Screening and Early Diagnosis receives funding for a research programme from the Department of Health Policy Research Programme. It is collaboration between researchers from seven institutions (Queen Mary University of London, UCL, King’s College London, London School of Hygiene and Tropical Medicine, Hull York Medical School, Durham University and Peninsula Medical School). The views expressed are those of the authors and not necessarily those of the NHS, or the Department of Health.
- Coleman MP, Forman D, Bryant H, Butler J, Rachet B, Maringe C, Nur U, Tracey E, Coory M, Hatcher J, et al: Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995–2007 (the international cancer benchmarking partnership): an analysis of population-based cancer registry data. Lancet. 2011, 377 (9760): 127-138. 10.1016/S0140-6736(10)62231-3.View ArticlePubMedPubMed CentralGoogle Scholar
- Richards MA: The size of the prize for earlier diagnosis of cancer in England. Br J Cancer. 2009, 101 (S2): S125-S129.View ArticlePubMedPubMed CentralGoogle Scholar
- Department of Health: Report on the self reported experience of patients from black and minority ethnic groups. . 2009, -https://www.gov.uk/government/publications/experience-of-patients-in-black-and-minority-ethnic-groups-2007-to-2009,Google Scholar
- Ramirez AJ, Westcombe AM, Burgess CC, Sutton S, Littlejohns P, Richards MA: Factors predicting delayed presentation of symptomatic breast cancer: a systematic review. Lancet. 1999, 353 (9159): 1127-1131. 10.1016/S0140-6736(99)02142-X.View ArticlePubMedGoogle Scholar
- Cuthbertson SA, Goyder EC, Poole J: Inequalities in breast cancer stage at diagnosis in the trent region, and implications for the NHS breast screening programme. J Public Health. 2009, 31 (3): 398-405. 10.1093/pubmed/fdp042.View ArticleGoogle Scholar
- Jack RH, Davies EA, Moller H: Breast cancer incidence, stage, treatment and survival in ethnic groups in South East England. Br J Cancer. 2009, 100 (3): 545-550. 10.1038/sj.bjc.6604852.View ArticlePubMedPubMed CentralGoogle Scholar
- Lyratzopoulos G, Neal RD, Barbiere JM, Rubin GP, Abel GA: Variation in number of general practitioner consultations before hospital referral for cancer: findings from the 2010 national cancer patient experience survey in England. Lancet Oncol. 2012, 13 (4): 353-365. 10.1016/S1470-2045(12)70041-4.View ArticlePubMedGoogle Scholar
- Coupland VH, Lagergren J, Konfortion J, Allum W, Mendall MA, Hardwick RH, Linklater KM, Moller H, Jack RH: Ethnicity in relation to incidence of oesophageal and gastric cancer in England. Br J Cancer. 2012, 107 (11): 1908-1914. 10.1038/bjc.2012.465.View ArticlePubMedPubMed CentralGoogle Scholar
- National Cancer Intelligence Network: Cancer incidence and survival by major ethnic group, England, 2002–2006. 2009, London, http://www.ncin.org.uk/publications/reports/,Google Scholar
- The Commonwealth Fund: International profiles of health care systems. 2011, Australia, Canada, Denmark, England, France, Germany, Italy, the Netherlands, New Zealand, Norway, Sweden, Switzerland, and the United States, http://www.commonwealthfund.org/Publications/Fund-Reports/2011/Nov/International-Profiles-of-Health-Care-Systems-2011.aspx,Google Scholar
- Olesen F, Hansen RP, Vedsted P: Delay in diagnosis: the experience in Denmark. Br J Cancer. 2009, 101 (S2): S5-S8.View ArticlePubMedPubMed CentralGoogle Scholar
- Rodgers M, Sowden A, Petticrew M, Arai L, Roberts H, Britten N, Popay J: Testing methodological guidance on the conduct of narrative synthesis in systematic reviews effectiveness of interventions to promote smoke alarm ownership and function. Evaluation. 2009, 15 (1): 49-73. 10.1177/1356389008097871.View ArticleGoogle Scholar
- Weller D, Vedsted P, Rubin G, Walter FM, Emery J, Scott S, Campbell C, Andersen RS, Hamilton W, Olesen F, et al: The aarhus statement: improving design and reporting of studies on early cancer diagnosis. Br J Cancer. 2012, 106 (7): 1262-1267. 10.1038/bjc.2012.68.View ArticlePubMedPubMed CentralGoogle Scholar
- Rajan SS, Lim JNW, Haq A: Late presentation and management of South Asian breast cancer patients in west yorkshire, United Kingdom. Asian Pac J Cancer Prev. 2011, 12 (6): 1615-1618.PubMedGoogle Scholar
- Meechan G, Collins J, Petrie K: Delay in seeking medical care for self-detected breast symptoms in New Zealand women. New Zealand Med J. 2002, 115 (1166): U257-PubMedGoogle Scholar
- Velikova G, Booth L, Johnston C, Forman D, Selby P: Breast cancer outcomes in South Asian population of West Yorkshire. Br J Cancer. 2004, 90 (10): 1926-1932. 10.1038/sj.bjc.6601795.View ArticlePubMedPubMed CentralGoogle Scholar
- Nosarti C, Crayford T, Roberts JV, Elias E, McKenzie K, David AS: Delay in presentation of symptomatic referrals to a breast clinic: patient and system factors. Br J Cancer. 2000, 82 (3): 742-748.View ArticlePubMedPubMed CentralGoogle Scholar
- Neal RD, Allgar VL: Sociodemographic factors and delays in the diagnosis of six cancers: analysis of data from the “national survey of NHS patients: cancer”. Br J Cancer. 2005, 92 (11): 1971-1975. 10.1038/sj.bjc.6602623.View ArticlePubMedPubMed CentralGoogle Scholar
- Sadler GJ, Jothimani D, Zanetto U, Anderson MR: The effect of ethnicity on the presentation and management of oesophageal and gastric cancers: a UK perspective. European J Gastroenterol Hepatol. 2009, 21 (9): 996-1000. 10.1097/MEG.0b013e32832948b3.View ArticleGoogle Scholar
- Metcalfe C, Evans S, Ibrahim F, Patel B, Anson K, Chinegwundoh F, Corbishley C, Gillatt D, Kirby R, Muir G, et al: Pathways to diagnosis for black men and white men found to have prostate cancer: the PROCESS cohort study. Br J Cancer. 2008, 99 (7): 1040-1045. 10.1038/sj.bjc.6604670.View ArticlePubMedPubMed CentralGoogle Scholar
- Neal RD: Do diagnostic delays in cancer matter [quest]. Br J Cancer. 2009, 101 (S2): S9-S12.View ArticlePubMedPubMed CentralGoogle Scholar
- Mitchell E, Macdonald S, Campbell NC, Weller D, Macleod U: Influences on pre-hospital delay in the diagnosis of colorectal cancer: a systematic review. Br J Cancer. 2007, 98 (1): 60-70.View ArticlePubMedPubMed CentralGoogle Scholar
- Elliss-Brookes L, McPhail S, Ives A, Greenslade M, Shelton J, Hiom S, Richards M: Routes to diagnosis for cancer - determining the patient journey using multiple routine data sets. Br J Cancer. 2012, 107 (8): 1220-1226. 10.1038/bjc.2012.408.View ArticlePubMedPubMed CentralGoogle Scholar
- Potter S, Govindarajulu S, Shere M, Braddon F, Curran G, Greenwood R, Sahu AK, Cawthorn SJ: Referral patterns, cancer diagnoses, and waiting times after introduction of two week wait rule for breast cancer: prospective cohort study. BMJ. 2007, 335: 288-291. 10.1136/bmj.39258.688553.55.View ArticlePubMedPubMed CentralGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2296/14/197/prepub
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.