Informal patient payments are a multi-face phenomenon with different features even within a single country (i.e. in the frame of the same health care system, regulations and traditions). Therefore, a universal definition is not available. The key characteristics described at the outset of this paper provide a more appropriate base for studying this phenomenon than pursuing an all-inclusive definition. Still, country-specific features should be taken into account to make sure that the unit used to measure informal payments is meaningful to the population being sampled.
The results of our review suggest that the study of informal patient payments for health care services is rather new, though the phenomenon has been in existence for a number of decades . Most of the studies that we identified were conducted between 1990 and 2005 mostly in former-socialist countries. It is likely that during the communist period, it was not possible to collect and report data on informal patient payments in these countries. Ideology also made it difficult to discuss the issue openly. Moreover, these types of payments might have been perceived as illegal. With the end of the communist governing, the socio-political changes resulted in more public attention to social problems, such as informal payments for health care services, which motivated their investigation. In addition, data collected since 2005 might still be in the stage of data analysis and therefore, not yet published. Overall, the dynamics of publications on informal payments indicates the growing research interest in this topic including new research techniques and larger sampling areas.
Our findings confirm that informal payments exist in countries of all levels of economic development, and in different parts of the world. However, we did not find studies reporting informal patient payments in high-income countries in North-West Europe, North America and Australia. The phenomenon is most often observed in former-socialist countries and developing countries (in Africa, South America and Asia), although it also exists in some high-income European countries that were not former-socialist countries (Italy, Greece, and Turkey) [35, 36]. As mentioned at the outset of this paper, the literature offers various explanations why informal patient payments exist in these countries. This includes under-funding of the health care systems, the specific organisation and governance of the health care sectors, but also culture and social perceptions . Still, these are only hypothesis and they need to be tested to explain the existence of informal patient payments in some parts of the world and their absence in others.
When we look at the study designs that we reviewed, we can outline several discussion points relevant to research. The first discussion point refers to the study objectives. We differentiated between exploratory, descriptive, analytical and predictive aims. However, we did not find studies with an explicit exploratory aim even among the earlier studies. We expected that an exploratory aim would be typical for the early studies when scant information was available because then, the research interest would be concentrated on exploring the phenomenon. Although some earlier studies had an explicit descriptive aim, other earlier studies had an analytical aim. Descriptive and analytical objectives allow finding determinants of informal patient payments and their correlation.
The second discussion point refers to the sample design. Sample design is part of the entire research design and it may minimise some biases in case of a well-developed sample. To estimate the level of informal patient payments, a probabilistic sample strategy is commonly implemented as it gives equal chances of being included in the study. Moreover, triangulation of the data is feasible when all parties participate (e.g. consumers, providers, officials) in the research. For instance, Cockroft et al  present quantitative data collected from households, where the main findings are discussed with physicians and nurses, as well as with stakeholders, to define the policy implications of the results. However, in the studies that we reviewed, the sample is not always constructed to avoid biases and to get valid data for the analysis. More pragmatic reasons, such as available research funds, are also reported [37, 38]. In view of this, it is not surprising that some recent studies on informal patient payments applied purposive, snowball and convenience samples.
Another discussion point is the data collection mode. The mode of data collection can be especially problematic when sensitive data are studied. This is because each single mode of data collection has its own pros and cons when sensitive questions are asked. The mode of data collection might even be a determinant of the value of indictors estimated based on sensitive data [e.g. [17, 19]]. According to our results, the response rate (when reported) was highest in face-to-face interviews with both consumers and providers. Face-to-face interviews are considered the most adequate approach in gaining understanding of what respondents mean when answering questions . However, face-to-face interviews might not be very effective in assuring the validity of the data when such sensitive topic as informal patient payments, is addressed. Respondents might be less willing to reply truthfully to questions on illegal expenditures if asked by an interviewer since the level of confidentiality is lower. In contrast, self-completion methods are usually preferred when the subject matter is sensitive [37, 40] even though some questions might be left unanswered by the respondents. The issue of confidentiality plays a key role. Respondents may be unwilling to describe their informal payments in front of an interviewer, and may feel more comfortable to express such behaviour when the pen in hand is the only "eyewitness". To overcome this difficulty, mixing modes of data collection could be used. De Leeuw  gives an example of U.S. National Survey on Drug Use and Health where respondents use a computer for answering sensitive questions while several non-sensitive questions are asked by an interviewer. This could also increase the response rate . Evidence from research on survey methods confirms the importance of combining various modes of data collection in surveys where potentially sensitive issues are investigated. The objective should be to help respondents to exert the necessary cognitive efforts and to answer the questions carefully. At the same time, the objective should be to make the respondents comfortable enough to answer openly and honestly the questions that might be of a sensitive nature . Thus, using mixing modes of data collection, specifically the introduction of a self-administrated part during a face-to-face interview, could be suitable for collecting valid data on informal patient payments. To the best of our knowledge, such a mixing mode of data collection has not been used in studies on informal patient payments.
Virtually all publications that we reviewed are based on retrospective research, thus another relevant discussion point is the recall period. The human memory can be a source of bias in research . Consumers might not remember the exact number of visits to health care providers/facilities if the recall period is long (e.g. one or two years), all the more so, the amount of payments they have done. Overall, respondents remember the event for a longer period of time if it is important to them . Thus, the experiences of utilisation of health care services can be different in case of less severe health complications (e.g. out-patient visits) and more severe health problems (e.g. in-patient services). Therefore, we recommend introducing different recall periods for questions on out-patient and in-patient services could enable the respondents to make less cognitive efforts. In particular, Baschieri and Falkingham  apply a 30 days recall period for utilisation of health care services and expenditures associated with visits to physicians and one year period for hospitalisations. There is also a possibility to avoid the use of a recall period. For example, the researcher's choice may lay on the introduction of diaries, which could allow collecting all expenditures of the household on health care at the time of payment. The choice of an adequate recall period is especially important for the valid measurement of informal patient payments.
Our findings on the response rate were surprising to a certain extent. In general, the response rate reported in the publications that we reviewed, was rather high. This could suggest that people are willing to talk about informal patient payments despite their informal and potentially illegal nature. However, it should be recognised that only few publications presented this characteristic. It might be that the response rate was presented in these publications because it was favourable for the study and indicated the representativeness of the data.
To enrich the methodological approaches to the investigation of informal patient payments, researchers can appeal to methods for measuring corruption in society. Although informal patients are not always illegal, our review suggests that they are sometimes perceived by respondents as corruption and illegal behaviour. The literature on measuring corruption suggests that corruption can be studied through the measurement of perceived corruption, as well as perceived willingness to pay bribes and bribe payments . Specifically, studies that focus on corruption include questions on the respondents' perception about level of corruption in a country, as well as hypothetical questions about the amount of money that a respondent would be willing to pay as a bribe in a given context. The latter technique could be useful to study respondents' attitude toward corruption. The measurement of both perceived corruption and willingness to pay bribes and bribe payments could be especially appealing for the investigation on informal patient payments to gain a better understanding on why informal patient payments exist.
The key results on the type of informal patient payments indicate that informal patient payments are a multifaceted phenomenon. All characteristics of informal patient payments included in our definition, appeared relevant for describing the pattern and magnitude of informal payments for health care services. Overall, the results indicate a great variety in the types of informal patient payments reported. This needs to be considered when designing a research instrument for the investigation of these payments. In particular, the researcher needs to clarify in advance what types of informal patient payments should be studied and thus, what type of questions to be included. It is also important to decide how to measure the incidents of informal patient payments since various measurement units are possible.
Our attempt to compare the empirical results presented a significant challenge. This is mainly due to the great variety of research methods applied. However, the overall findings indicate that informal patient payments are a substantial phenomenon in terms of both scope and scale, and should not be neglected. Moreover, results of household surveys would be more meaningful if considered against the background of macro-level data at a national level (whenever available). For example, the National Health Accounts could be a useful source of macro-level data since they report total health expenditures as well as formal transactions in the health care sector (e.g. expenditures by various institutions, external financing and out-of-pocket spending). In addition to this, little is known on why informal patient payments exist and how the specific patient-providers relationship determines them. This indicates the need of combining quantitative and qualitative research methods when studying this type of payments. The need of deeper understanding of the informal patient payments has already captured the attention of researchers who are trying to provide theoretical explanations to the existing empirical findings [5, 44].
We searched systematically for relevant publications. However, we do not exclude the possibility that we have missed some studies reported in non-English language journals as well as very recent studies that are still not reported. Despite this shortcoming, our results and discussion are relevant to future research on informal patient payments. As mentioned above, the investigation of the phenomenon is interwoven with methodological complexities related primarily to the data collection and research instruments. We have outlined and discussed most of these complexities. However, other peculiarities (e.g. wording of the questions and the length of the interview) also require attention.
Based on our findings in combination with the conclusions of a recent methodological review presented in Roberts , the following key strategies could be recommended to researchers who choose to study informal patient payments: (1) considering a broad county-specific definition of informal patient payments when designing the questionnaire and an adequate measurement unit that is meaningful to the population being sampled; (2) opting for face-to-face interviews at the respondents' home to ensure that the interview situation is adequately conducive to respondents but simultaneously, to enable a high response rate; (3) administrating the questions on informal patient payments as an anonymous self-completion component within the face-to-face interview; (4) assuring the respondents on the issues of confidentiality and explaining why the data on informal patient payments are important.