Interviewer Manual: Health-System Costs

Health System Costs of Unsafe Abortion in Rwanda

Interviewer Manual

Revised 25 February2012

______

Guttmacher Institute

and

School of Public Health, National University of Rwanda

______

Page 1

Interviewer Manual: Health-System Costs

The purpose of this manual is to give the interviewer detailed guidance in the interview process. The manual is divided into two sections: general considerations and question-specific instructions.

NB: (1) Change “septecemie” to “sepsis” everywhere; (2) a note is repeated on page 4 and page 5 of the manual in French.

1. General Considerations for the Interviewing Process

1.1. Nature and Goals of the Study

The main goal of the study is to estimate the costs to the Rwandan health care system of treating complications arising out of unsafe abortions. Estimation of health system costs will be achieved through the process to be described in this manual.

This is a 2-year research project designed to generate evidence needed to drive policy and program reform that will address the economic and social consequences of unsafe abortion, thereby improving maternal health in Rwanda. The data collection process is likely to take one month.

1.2 The Survey Sample and Respondents

Knowledgeable health personnel at around 40 health facilities will be interviewed using two data-collection instruments: Questionnaires A and B. Questionnaire A comes in four versions:

  • Questionnaire A1 – for respondents at the central level
  • Questionnaire A2 – for respondents at regional hospitals
  • Questionnaire A3 – for respondents at district hospitalsand private polyclinics and clinics
  • Questionnaire A4 – for respondents at health centers

Questionnaire A collects data on personnel costs, overhead costs, and capital costs associated with the provision of post-abortion care. Questionnaire B collects data on the drugs, supplies and materials used in the provision of treatment for specific post-abortion complications.

A purposive sample of 39 facilities from all over the country has been selected by the local project team. These facilities include 5regional hospitals, 11districthospitals and 23 health centers.[1] The selection criteria are based on region, facility type and facility ownership (public, private, faith-based).

Questionnaire A1, for the central level, will primarily be aimed at the Ministry of Health.

You, the interviewer, may have to interview several staff at a particular facility in order to complete the two questionnaires. The larger the facility, the greater the chance that you will interview more than one person.

  • For the central level, typically in the ministry of health or equivalent, respondents may include: the director of Reproductive Health Services, the director of the Planning Division and the director of the Statistical Division of the MOH.
  • For regional hospitals, the head of the reproductive health services (or the head of the maternity ward) and the head of administration may be interviewed. An anesthetist, the in-charge of the operating theater, the in-charge of the laboratory are other personnel that will probably be interviewed to fill in specific parts of the questionnaires.
  • For district hospitals, the respondent may be the district medical officer.
  • For health centers, the medical officer assigned to the center may be interviewed.
  • For all facilities, you may have to interview the head of the laboratory and the pharmacist for certain items.

Since the goal of the health-system-cost survey is to collect expert opinions, the staff member interviewed at a particular facility should be the highest ranking person who is well versed in the facility’s treatment of cases of post-abortion complications. If that person then tells you that another staff member may be better able to answer certain questions, then you should follow his/her advice and interview that person later.

Be aware, however, that primary respondents may try to pass you along to a junior staff member. Try to avoid this—emphasize that the survey is meant to collect expert opinion, not “data” as such (which are likely old, incomplete, and inaccurate) and therefore you want to interview that person in his/her capacity as an expert.

Experience from a Ugandan study: A similar study has recently been carried out in Uganda. Interviewers found that at larger facilities, up to six staff were interviewed. For example, the head of the laboratory was interviewed for information on lab tests, the nurse in charge of the operating theater was interviewed for inputs used in treating lacerations and perforations, etc.

1.3. Consent to be Interviewed

The information collected in this survey does not involve the gathering of any personal data on patients at the facilities sampled or on any providers of health services. Moreover, as the following section makes clear, the information to be collected is not “hard” data but merely opinion-type data. Thus, ethical considerations about confidentiality of the data and risks/harm to the respondents if the data were to be divulged do not apply to this study.

Before beginning an interview with any facility staff member, the interviewer should clearly inform the respondent that no personal information about him/her or about any patient of the facility will be collected in the interview and thus that there are no risks involved in consenting to be interviewed. On the contrary, to the extent that the results of the study lead to a more optimum use of resources in the health sector, he/she, along with the rest of the public, will benefit from more efficient use of resources. Before an interview begins the respondent must have agreed to and signed the consent form (see Annex 1).

1.4. The Type of Data to be Collected

It cannot be emphasized too strongly that the type of data that this survey will collect is expert opinion. Respondents should give you their best estimate, their considered opinion, their best guess to (almost) all of the questions in both questionnaires. Except for a few questions—which are clearly indicated in the questionnaires—you do not want the respondent to formulate his/her response by reaching for some document, statistical table or other document, or by reaching for the phone to ask some other official to look up some data. You want the respondent to give his/her opinion as a knowledgeable expert in the field of post-abortion care.

Why does this survey want to avoid “hard data”? There are several reasons:

  • The “data”, if found, will most likely not exactly answer the question posed.
  • The “data” will likely be incomplete or based on a biased set of inputs into their information system.
  • The “data” may well be old and out-of-date.
  • We already know from experience that most of the questions we want answered do not have data systematically and accurately collected by any information system.
  • We do not need “data” that apparently are very precise since our results do not demand high precision. Moreover, while such data may appear precise, they may well be very inaccurate for the reasons listed above.

Instead, we want experts to give their expert opinions. This type of approach is generally known as the “Delphi method” and is an accepted form of gathering hard-to-collect data. The underlying assumption of this method is that, while individual expert estimates may be inaccurate, the average of several experts is likely to give an estimate that is close to reality. This method is appropriate when the results that one is looking for do not have to be very precise. Since the results of this study are meant to be used by policy makers, not by day-to-day managers, we do not need the cost estimates to be very precise. The Delphi method can produce useful results at a low cost.

A note on ranges: If a respondent replies with a range (e.g., “20-30%”) rather than an exact number (e.g., “80%”), write down the range as stated. The range can be converted into an exact number later, during the data-coding process.[2]

1.5. Definition of post-abortion complications

The study focuses on five post-abortion complications:

Sepsis. Sepsis refers to a serious infection resulting from the abortion. It should be accompanied by fever (temperature > 37.3 degrees Celsius) and can range from endometritisup to the condition known as septicemia, a generalized infection that is life is threatening.

Shock. Shock refers to hypovolemic shock, a serious condition in which the volume of blood in the body falls to such a low level (due to hemorrhaging) that the body goes into shock and normal functions are shut down.

Incomplete Abortion. Incomplete abortion refers to a post-abortion condition in which part of the fetus (or “product of conception”) remains inside the uterus. This can lead to a very serious condition if not attended to.

Cervical/Vaginal Laceration.This is a laceration or other mechanical injury to the cervix and/or vagina which occurred at the time of an unsafe induced abortion. It includes burns caused by caustic chemical substances.

Uterine Laceration/Perforation. This condition refers to mechanical damage to the uterus. Perforation refers to an opening made between the uterus and the peritoneum (the lower body cavity containing the intestines). Included here are other types of perforation such as perforation of the bowel and of the peritoneum. Treatments such as laparotomies and hysterectomies should be included here.

1.6. Don’t know (DK), no answer (NA) and blank

With a very few exceptions, the response “don’t know” is not allowed in either Questionnaire A or B.[3] As mentioned, if after several attempts to prompt a response, a respondent still refuses to give an answer, the interviewer may enter “NA” for No Answer. No question is ever to be left blank, except where it has correctly been skipped.

2. Question-by-Question Instructions

2.1. Questionnaire A – Personnel, Capital, and Overhead Costs

The four versions of Questionnaire A (i.e., Questionnaires A1, A2, A3, and A4) are basically the same, varying only slightly because of differences between type of facility. The instructions in this section apply to all four versions, but for readability reference is actually made only to Questionnaire A2 (for respondents at Regional Hospitals). In this questionnaire the question numbers are all in the “one hundreds” (e.g., A104 or P114). The question numbering for the four versions are as follows:

  • Questionnaire A1: Sections 1.0, 2.0, 3.0; question numbers from 001 - 099
  • Questionnaire A2: Sections 1.1, 2.1, 3.1; question numbers from 101 - 199
  • Questionnaire A3: Sections 1.2, 2.2, 3.2; question numbers from 201 - 299
  • Questionnaire A4: Sections 1.3, 2.3, 3.3; question numbers from 301 - 399.

Thus, the instruction for, say, question A104 applies equally to A004, A104, A204 or A304. If a question is different from one level of facility to another, facility-specific instructions will be noted in what follows.[4]

Identification of Facility

The name of the facility, the date of the interview and the name of the interviewer should be written in the questionnaire before the interview begins, or immediately after the interview has been completed.

Respondents’ Identification

You may have to interview more than one expert in the facility, especially in larger hospitals. There may be three or more repondents, (e.g., head of reproductive health department, chief of administration, head of laboratory). In smaller facilities, one expert may be able to answer all questions (e.g., chief medical officer). Write down the name, designation and phone number of each respondent.

Section 1: Abortion Complications

A101.Does this [facility] maintain statistics on the numberof women who come to the hospital with post-abortioncomplications?(See also A001, A201, A301)

The possible responses are either “Yes” or “No”. (If “No”, skip to A104.)

A102.What were the number of women who come to this[facility] with post-abortion complications in the last year for whichstatistics are available? (See also A002, A202, A302)

This is one of the very few questions where we want to get statistical data if they exist. If the respondent answered “Yes” to A101 but then cannot find the report or statistical table with the required data, proceed on the assumption that the data do not exist and skip to A104. In this case, enter “DK” (don’t know) for this question.

A103.What is the year referred to in A102?(See also A003, A203, A303)

Write down the year to which the number of women given in A102 refers. If the data is more than three years old, the response to A104 will be preferred to A103 in the data analysis phase of the research.

A104.Give your best estimate of the number of women who came to this [facility] during all of last year with post-abortion complications.(See also A004, A204, A304)

Remember, you are now asking for an expert opinion, so inform the respondent, if necessary, that his/her best estimate is all that is needed.Ask this question even if the respondent has given answers to A102 and A103. The respondent may say that the A102 response is also his/her best estimate, but if the year referred to (A103) is a few years old, point that out to the respondent and try to elicit his/her estimation of the current number of women requesting PAC.

Before asking A105, you should make the following introduction: “It is important for this study to get the best possible estimates of the distribution of serious post-abortion complications. The next question asks you to think of 100 women presenting at this hospital with serious post-abortion complications.”

A105.Think about 100 women treated in this [facility] last yearfor post-abortion complications. Of these 100 women, how manywere treated for each of the following:

  1. Incomplete abortion
  2. Sepsis
  3. Shock
  4. Cervical/vaginal lacerations
  5. Uterine laceration/perforation(See also A005, A205, A305)

This is a very important question, so it is important that the respondent understand it very clearly. First of all, the interviewer should be thinking onlyabout women being treated for one or multiple post-abortion complication(s), not all female patients treated.

Secondly, it is better to ask this question sequentially, that is, ask first how many, out of 100, get treated for incomplete abortion (usually by vacuum aspiration or by dilation and curettage); then ask how many get treated for sepsis; and so on. Note that the sum of numbers in A105 (a. to e.) may well add up to more than 100 since one woman may be treated for two or even three complications. Record these five responses in the column with the heading “1st Answer”. Then repeat the responses just given by the respondent and ask if the respondent would like to reconsider his/her five estimates. Write these revised responses in the column “2nd Answer”, even if they are the same. (The reason for this procedure is that respondents may initially give five responses that add up to under 100—which should not happen since virtually all PAC patients will have at least one of these five complications.)

A106.There are several other post-abortion complicationsthat may occur rarely(e.g., peritonitis, renal failure, etc.). Out of 1000 complications, how many would consist of these rare complications, i.e., complications not listed in A105?(See also A006, A206, A306)

Again, the respondent is asked for his/her expert opinion. The answer given may be less than one woman. Do not accept an answer such as “very few”: insist on a numerical estimate (e.g., “less than 1”; or “1 in 200”), then convert it to per 1000 (e.g., “1 in 200” should be written as “5 per 1000).

A107.Treatment of a post-abortion complication sometimes fails to cure the patient who must be retreated. For 100 cases of each of the following five complications, how many need to be readmitted for further treatment? (See also A007, A207, A307)

A woman with a post-abortion complication may be treated and released by a health facility but then return because she has not been cured. The facility will then have to treat the patient a second time. Question A107 requests an estimate of how often this occurs for each of the five major complications, considering 100 women who are treated for the complication. The answers, therefore, must range between “0” (never) and “100” (always).

Section 2: Cost of Personnel

Incomplete Abortion

For the next two questions, ask the respondent to think (hypothetically) about 10 women admitted to this[facility] for incomplete abortion. They should think in terms of women with just this one complication, even if somepatients will normally present at the facility with multiple complications.

P101.How many women, out of 10 women admittedwith incomplete abortions,are seen by each type of[facility] worker listed below? (See also P201, P301)

You should sequentially read from the table the names of the categories of worker at the facility. Enter a zero (“0”, not “N/A”) if the respondent says that particular type of worker never attends to women with incomplete abortions. For each type of worker where “0” is the answer in P101, skip P102. As an example, if you write down “5” beside “Nurse” (line d.), it means that the respondent estimates that 5 out of every 10 cases of incomplete abortion are attended to by a nurse or nurses.

P102.On average, how many minutes does each type of workerspend with a woman admitted with incomplete abortionduring the whole course of treatment? (See also P202, P302)

Be sure not to ask this question for those worker categories that the respondent has stated in P101 never attend incomplete abortion cases. The number of minutes estimated should be for the whole treatment period, from when the woman arrives at the facility until when she is discharged. It should include all the time that a health worker spends directly on treating the patient—which includes both time spent in the presence of the patient and time spent away from the patient but engaged in an activity directly related to treating that patient (e.g., filling out the patient’s chart; preparing the patient’s medicine).

It should include the total time spent by all workers if each category of personnel. For example, if in a particular facility three nurses normally attend a post-abortion patient, then the response should be the total (average) time spent by all three nurses. If nurses of different levels are all involved in the treatment, the respondent should estimate the minutes spent by each level.

Sepsis

Sepsis is a serious condition marked by localized orgeneralized infection. Questions P103 and P104(See also P203, P303 and P204, P304) are identical to P101 and P102, except they ask about women admitted to the facility with sepsis. Therefore, please refer to the instructions for P101 and P102 above.