Valuable insights can be gained from comparing the representations of immigrants in each of these three journals. Almost all the original MJA research articles focused on newly arrived refugees. Almost all focused on relatively exotic, if not ethnic-specific illnesses. Most (13 articles) related either to the mental or physical health of newly arrived refugees or asylum seekers, and five dealt with concerns about vitamin deficiencies in veiled or dark-skinned women and their babies, or in Arabic speaking toddlers. The MJA published seminal articles and opinion pieces in these areas particularly concerning the Australian policy and treatment of asylum seekers. Researchers often focus on the most recently arrived community as they are considered to be the most different from the mainstream [143
]. Yet the representations remain very limited. Of the 2,227 articles published during the 12-year period, only 21 (0.9%) were original research and a total of 34 (1.5%) were concerned primarily with multicultural health issues. A further 71 MJA original research articles including some discussion of multicultural issues, with 11 articles discussing multicultural health issues as a major component of the article and 23 as a moderate component. In others (37 articles), the mention of diversity was incidental, or, what Minas et al. [142
] described as 'tacked-on', or an after-thought to the main discussion. Of the 54 MJA articles categorised as policies, guidelines or case studies, 10 were primarily concerned with multicultural health issues.
There was a slightly larger number of original multicultural health articles (n = 41) featured in the ANZJPH compared with the MJA, representing 3.4% of the total articles published in the ANZJPH over the 12-year period. The majority of original ANZJPH articles were published prior to 2001, with only 16 original research articles published since 2001-possibly a reflection of a changing social climate. The range of health issues covered in the ANZJPH was somewhat limited, with seven concerned with blood borne viruses or sexually transmissible infections, five with cervical cancer and pap tests, three with obstetric profiles, three with smoking prevalence, two with dental health, one with coronary heart disease, one with blood pressure and one with the health of Asian sex workers. Six articles critically discussed ethical issues, health status or service implications of the policy of detaining asylum seekers and their children in remote detention centres.
The 15 original AHR research articles were diverse in subject matter. Five addressed workforce changes, recruitment issues, and bilingual staffing in hospitals. The other 10 articles addressed a range of issues related to epidemiology, access, service delivery, and cross-cultural research methodologies.
Apart from two AHR articles that assessed service access [124
] it is interesting that few studies in these three journals have evaluated service quality or effectiveness. For example, surprisingly few studies have evaluated the impact, demand and cost effectiveness of healthcare interpreter services. Indeed, there is little Australian evidence regarding the most common methods of facilitating communication in healthcare with people who have limited English. Access studies have focused on community services and aged care services.
Wide-ranging terms were used by researchers in these journals to describe ethnicity. Some used various aggregations of 'countries of birth' (e.g. North Asian, South East Asian, Asian), 'white' and 'non-white', 'non-English-speaking', or 'ethnic groupings', one defined ethnicity in a population as related to parental country of birth. This diversity of terms indicates the continued lack of healthcare researcher agreement about the appropriate ways of reporting ethnicity in Australia, although one study noted that government databases tend to follow nationally agreed terminology [100
]. Few researchers outlined the study methods they used to assess the ethnicity-related variables. In articles that included some mention of multicultural health issues, few discussed the study's results related to ethnicity; the majority simply noted ethnicity as a demographic variable. There were few attempts in these studies to further analyse the effects of socio-economic status, gender, or educational status see Figure .
Australian and New Zealand Journal of Public Health literature search method and results.
Datasets often contain little information on English language proficiency or on the demographic, social, economic and cultural factors that may influence health and health service utilisation [100
]. Much class, ethnic and geographical variation in health status may be masked or homogenised because birthplaces or groups are aggregated (often in inconsistent ways) so as to achieve statistical meaning. At times, for example, all people born overseas are considered as one group, as are all people who speak a particular language, irrespective of their year of arrival, English language proficiency, age, and region of origin or socio-economic profile [144
]. Bhopal refers to these epidemiological practices as 'inventing ethnic groups', 'lumping groups together', 'not adjusting for confounding factors' and 'not comparing like with like' [148
]. Coming from a non-English-speaking country is frequently used as a surrogate for poor proficiency in English. This aggregation has an inherent bias as many immigrants from non-English-speaking countries actually speak very fluent English (e.g. immigrants from Hong Kong, India, Holland). Kliewer and Jones [144
], in their study of newly arrived immigrants, found that almost 12% of immigrants from NESB countries spoke English as their preferred language and another 39% stated they spoke English 'well' or 'very well'. Fields that have been incorporated into standard healthcare databases, such as 'language spoken at home' or 'preferred language,' do not indicate how well a person speaks English.
The available research could be described as uneven in its coverage of major health status and health risk factor issues affecting immigrant communities. There are some major health status issues that appear to have received little attention; for example, renal disease, dialysis access, heart failure, kidney disease, prostate cancer, lung cancer, eye problems, cardiac problems, depression or chronic diseases. It is interesting that very few original articles dealt with patient experience, acute care, or the ethnic elderly. In terms of health promotion, the emphasis was limited to screening, general practice access, and interventions to promote population-based health screening.
Further, little attempt has been made to understand the crucial relationship between poor language proficiency, culture and patient safety in the Australian context. The question of whether there is ethnic disparity (difference in treatment and care based on ethnicity, race or language ability) in healthcare service provision has not been addressed in these major Australian healthcare journals.
Studies were mainly descriptive/observational studies or basic epidemiological studies. None were randomised control trials or longitudinal studies. Some studies were a series of cross-sectional surveys. There were no clinical trials. A few interventional studies looked at, for example, the effect of the ethnic media on service or screening access. While there were some studies with large samples, others had quite small numbers.
Interestingly, many ANZJPH articles studied the Vietnamese or 'South East Asian' populations, with 14 of the 20 articles which featured specific populations choosing to study South East Asians. Two other studies researched Arabic/Lebanese groups, two studied Italians, two studied Islanders, and one studied each of sub-Saharan Africans and Filipinas. Four studies were concerned with asylum seekers and a further three with refugees. This is an interesting representational bias even given that Vietnamese is one of the five major languages spoken in Australia. The reasons can only be speculated upon, but it is interesting to recall the words of Martin [8
], who stated that in the 1950s and 1960s migrant groups were homogenised (as one) but that in the 1970s 'migrant' equalled Greek and Italian. The research data from the ANZJPH would seem to indicate that in the past decade, 'migrant' equals 'Vietnamese', 'South-East Asian', 'asylum seeker' or 'refugee'. This observation is also consistent with the representations of migrants in MJA original studies, with newly arrived refugees and asylum seekers being the major groups studied. It may be that by the 1990s and this 21st century, the term 'migrant' has come to mean South East Asian, Vietnamese, refugee and asylum seeker.