We performed in-depth semi-structured interviews á-la-Bewley to physicians at Servicio Navarro de Salud-Osasunbidea (SNS-O, Spain)22 (N=16). Interviews were undertaken over a sixteen-month period starting in February 2010. The questions addressed to the doctors in the interviews were designed in order to test the hypotheses formulated above. The questions open-ended and were grounded in SDT and BE theory.

Doctors were invited to participate through a formal invitation letter. The letter briefly informed them about the interview although no details about the research goals were given to avoid biasing doctors’ answers. The letter was sent together with a document stating that the results would remain anonymous. We asked doctors to talk openly and to give their candid opinions about the health organization for which they were working.

Providing enough time and comfortable place for the interview is very important in order to obtain high-quality data. We performed most of the interviews at the doctor's workplace, usually in their personal offices at a date and time of their convenience.

The duration of the interviews ranged from a minimum of 57minutes to a maximum of 1hour and 44minutes. All interviews were performed by authors (MB and JMC) and were recorded. We also took field notes, something considered relevant to improve the quality of data. The recorded audios were transcribed to a text document. Transcripts and notes allowed all the details of the interview to be registered.

We initially used random sampling of eight interviewees from a population of senior doctors, followed by emergent sampling.24 All interviewees were doctors working at SNS-O. For the following eight cases, we used emergent sampling making new sampling decisions as we gained more knowledge about the phenomena of interest. All but one of the interviewees were senior staff, highly qualified, in high responsibility positions, and with long tenure within the health care system. They were mostly men (fourteen out of sixteen) and from a wide range of services. All but three were working in hospitals. The remaining three were working or had worked in a publicly outsourced private health care organization or in public primary care centres.

The information was collated and coded.23 We used three main categories: 1) intrinsic motivation (IM), 2) crowding-out (CO), and 3) crowding-in (CI).

Each code grouped homogeneous statements —observations— which directly referred to the same category. Codes where necessary were subcoded.

Codes were of two types: deductive and inductive. Deductive codes are those which have been labelled on the basis of existing theories —BE and SDT— prior to interviews. Inductive codes are those which spontaneously emerge from respondents’ statements.

A descriptive analysis and some quantification were shown within each of the three main categories. We found the frequency of each code in respondents’ discourse, and linked it back to the respondents. For each category, we also explored interactions between codes. Statements were classified into positive or normative. Positive statements are those that describe an actual situation experienced by the respondent. Normative statements are those that describe respondents’ views about how things should be in health care organizations.

We defined 32 codes where all relevant statements were included (detailed code definitions and descriptions, considering their inductive an deductive nature, are available for interested readers in the Appendix online to this article). Statements and codes have been organized in spreadsheets.

A threshold on the informative value of codes was defined based on the concepts of saturation and hierarchy.23 A code is saturated when it emerges repeatedly in a sufficient number of interviews, and always in the same explanatory or causal direction. The hierarchy criterion is applied to those codes, subcodes and subsubcodes that do not achieve saturation but are nested into other higher level codes which are already saturated.

All interviewed physicians reported being intrinsically motivated. Doctors’ major explanations regarding IM were two: they enjoy or like the medical practice —Like/Enjoy— and they are vocational professionals —Vocation. We consider Like/Enjoy and Vocation the major explanations of IM because they emerge independently —they cannot be placed as subcodes of each other code.

Table 1 shows the results, including relations between IM codes and subcodes. Column three shows the number of respondents who at least made one statement regarding the code or subcode displayed in the same row of the table. Column four shows the number of statements included in the code/subcode, and its percentage of appearance.

The main reasons for Like/Enjoy include Attractive Profession, Science, Service, Further Education, Technical Knowledge and Research. Most have to do with tasks featured by a technical or scientific component. We conclude that one of the main components of doctors’ intrinsic motivation is that medicine challenges their intellectual curiosity. They consider the medical profession to be a field where they can boost their scientific aspirations. For the case of Vocation, the main explanations were Service, Help/Empathy, Dedication, Effort and Pro-social/Altruism. In this case almost all reasons belonged to the “helping others” or the humanistic part of the medicine.

Table 2 shows a selection of representative statements relating IM jointly with the code, subcode and subsubcode to which they have been linked.

Results show that doctors are intrinsically motivated professionals. Their intrinsic motivation is sourced in two dimensions featuring the medical practice: the scientific or technical dimension and the humanistic or pro-social dimension.

Evidence of the CO effect was also found. We codified 214 statements referring to CO incentives. The fact that money may hurt doctors’ intrinsic motivation was widely shared by all interviewees.

Table 3 shows the results and relevant codes explaining the CO. Columns one, two and three of the table show the code/subcode/subsubcode hierarchy. Column four shows the number of respondents who made at least one statement linked to the code or subcode in the same row. The number of statements and percentages are shown in column five.

All statements regarding CO were classified into two major codes: Financial Incentives and Other Non-Financial. Peonada-FFS, and Professional Career-1 are subcodes of Financial Incentives. Interviewees colloquially call “Peonada” to a Fee-For-Service (FFS) payment for working out-of-hours aimed to reduce waiting lists. The main reasons through which Peonada-FFS causes CO are captured in subcodes Task Meaning Change, Market Transaction and Opportunistic Behaviour. This means that the introduction of the FFS scheme may have caused:

• •

  A change in the perception of the medical work from a vocational activity to a means for seeking extra remuneration or money (task meaning change).

• •

  Doctors to start seeing the provision of health services as market interactions governed by market rules (market transaction).

• •

  Doctors to behave strategically aimed at maximising their earnings (opportunistic behaviour).

Lack of Recognition, Task Meaning Change and Opportunistic Behaviour subcodes capture the main explanations of the CO nature of the Professional Career-1. The current professional career is hurting doctors’ IM because it recognizes neither the effort nor the quality of their work. This is an important result, since the so called professional career is the genuine incentive scheme in the SNS-O.

Within the non-financial causes of CO we consider two codes: Control and Bureaucratization. Statements under Control point out that doctors perceive as controlling some managerial decisions taken in the SNS-O health organizations. Doctors reported that this hurts their IM because they feel their autonomy is constrained. We group these Control statements within the subsubcode Lack of Autonomy.

Finally, the last reason for CO reported by doctors is the Bureaucratization of the medical profession. This highlights the CO effect caused by lack of competitive stimulus, and lack of professional incentives in the context of public health organisations in the SNS-O. A selection of the most illustrative statements showing evidence of CO is shown in Table 4.

Respondents made 178 statements related to the crowding-in category. A fact that saturates rapidly is that non-financial incentives are the main ones to cause CI. Only a few experiences of crowding-in were explained in the course of interviews. However a lot of proposals and ideas of CI incentives were suggested by the interviewees.

Table 5 shows the codes explaining evidence for CI. Column four shows the number of respondents who at least made one statement linked to the code/subcode/subsubcode in the same row. Column five shows the number of statements and percentage of response in brackets. The main code explaining this category is Non-Financial Incentives which includes both, statements referring to proposals and to experiences. Statements referring to Non-Financial Incentives were classified into five main subcodes: Professional Career-2, Autonomy, Recognition, Research and Teaching.

A first insight is that the most appropriate incentives for CI are those that promote the professional development of doctors such as professional career, opportunities for research and teaching. As evidence of the importance of setting an effective professional career scheme, doctors said that a new professional career scheme based on Professional Development and Further Education was needed. Other CI actions like providing more autonomy for managing their own work and recognition of the achievements of their work in terms of quality, merit or excellence based criteria, were also reported. Doctors wished for more autonomy to manage and organize their own work: Self-management.

Table 6 shows quotes illustrating the request for a renewed professional career, more autonomy and other non-financial incentives.

In order to provide more consistency to the analysis performed in the CI section, we classify doctors’ statements into normative and positive. Then, for the CI statements we calculate the probability of being positive P(p), or being normative P(n), obtaining the following values: P(p) = 0.34 and P(n) = 0.66.