2.7 Morbidity m ethod
2.7.2 Problems with the morbidity method
As stated above, this method resulted from a series of compromises and was not perfect. Several major problems arose during the main phase of the study which had not been foreseen in the pilot these included:- (i) hypertension, (ii) the TB tests and (iii) the data collection by local interviewees.
Hypertension (HT) was frequently cited as a chronic disease but did not appear on the original tracer list of symptoms. It was not considered a potential symptom in the pilot study because its diagnosis cannot be made by non-medical personnel. Thus an individual who reported being hypertensive only knew this because a doctor or nurse had told them, and those hypertensive individuals who had never seen a doctor (or at least not since they had developed HT) would be unaware of the problem. Hence a bias in reporting of HT could occur between individuals who visited the doctor or nurse and those who did not.
At the start of the main phase the author decided to include HT in the questionnaire and analysis for four reasons (i) access to medical care in the study area is relatively easy and most of the respondents were aware that HT could cause headaches (and other symptoms) and hence would go to the hospital if they had recurring headaches, thus it seemed likely that most cases of HT would be diagnosed, (ii) HT was widespread, (iii) HT is associated with other symptoms which may also be symptoms of other (infectious) diseases - it was important in the analysis to attempt to separate out the different disease types, and (iv) HT is known to be associated with nutritional status.
All the results of the TB tests were negative. Both the author and the Sarawaki medical authorities found this result difficult to believe given the levels of TB in Sarawak. It is possible that there was some error either in the method collection o f the sputum or the analysis. The technicians in the hospital may also have been overwhelmed by the number o f tests they were required to carry out in a relatively short period of time and may not have examined the slides carefully enough. In an attempt to rectify this situation, two new questions concerning TB were added to the morbidity questionnaire in the second round. The subjects were asked if they had ever suffered form TB and, if so, if this had occurred since 1990.
An attempt was made to train local interviewers to apply morbidity questionnaires to the inhabitants of a longhouse for the interval between the author’s visits to the longhouses.
Unfortunately this part of the project failed. This was probably due to a combination of reasons;- (i) poor training of the interviewers by the author - the training period was short; (ii) the interviewers’ apathy - possibly because they were not paid enough; and (iii) the lack of enthusiasm amongst the interviewees for the project - possibly because the purpose o f the project was not well enough explained. Following this failure it was decided that the morbidity analysis would focus on the data collected by the author at each of her two visits to the longhouse.
2.7.3 Morbidity data organisation
There is no biological basis for assuming that the BMI or MUAC and all the symptoms in the questionnaire are associated in the same manner. The symptoms are allied with many different diseases the physiological mechanisms of which are varied and hence the relationships observed between the symptoms and the BMI or MUAC will not all be identical or, indeed, in the same direction.
Studies assessing the relationship between the BMI and mortality risk have frequently reported a U- or J-shaped relationship (see chapter one). Mortality risk may be raised in subjects with both low and high BMI. Certain diseases are positively associated with the BMI and others are negatively associated with it. Waaler (1984) refers to the high-weight and low-weight groups of disease. In the developed world, typical high-weight or high-BMI diseases include the cardiovascular and cerebrovascular diseases, breast cancer and late onset NIDDM and its sequelae. Typical low-weight or low-BMI diseases include tuberculosis, obstructive lung diseases, certain types of cancer e.g.: lung cancer (Kabat et al, 1992). In the developing world, infectious diseases may make up much of the morbidity burden of a population. Many important
infectious diseases such as acute diarrhoeal disease (Scrimshaw et al, 1968; Briend, 1990;
Tomkins & Watson, 1989) are associated with low-weight in the developing world.
Table 2.12 shows the various symptoms and the direction in which they may be expected to relate to the BMI or MUAC, taking biological disease processes into account. One o f the difficulties in using a symptom list is that it is not possible to determine the underlying disease with which a symptom may be associated. Certain symptoms such as cough, chest pain and respiratory disease may, for example, be associated with cardiovascular problems or result from a bacterial infection. An attempt to disaggregate these symptoms has been made by treating any headache or cough, chest and respiratory problem reported in conjunction with hypertension separately from a headache, cough, chest or respiratory problem reported without hypertension. Also, epigastric pain may, for example, be due to a cancer or an intestinal infection. Gastric cancer may be associated with high BMI in young adults (Hansson et al, 1994), on the other hand the a priori expectation of intestinal infections is that they would associate negatively with the BMI. Given that the study population lives in the developing world and has a relatively high exposure to infectious disease, epigastric pain has been placed in the negative association group as it is expected that infectious diseases would be the most common cause of epigastric pain in this population.
Table 2.12: The direction o f the association between the BMI or M UAC and various symptom types expected
Incidental association Positive association Negative association
Injury HT Fever
Muscle ache Headache (assoc, with HT) Diarrhoea
Headache (not assoc, with Cough (assoc, with HT) Gastric pain HT)
Other Chest (assoc, with HT) Cough (not assoc, with HT)
Respiratory (assoc, with Chest pain (not assoc, with HT) HT)
Respiratory (not assoc, with HT)
It can be seen that the diseases which are expected to associate positively with the BMI or MUAC are similar to W aaler’s high-weight group. The negative association group represent symptoms which are commonly found with infectious diseases (e.g.: fever, diarrhoea) and various lung diseases. The incidental group represent diseases which are not expected to associate with the BMI or MUAC in a particular direction. For example, it is possible that an injury or muscle
associated with increased severity o f ankle fracture following low velocity injuries (Spaine &
Bollen, 1996).
It should be noted that the associations shown above are not necessarily assumed to be causal.
For example, fever may be associated with low BMI or MUAC because (a) a thinner person may be more likely to catch a febrile disease, or (b) an individual with a fever may have a depressed appetite and hence become leaner, or (c) an individual with a fever may have an increased basal metabolic rate and hence greater energy expenditure, or, finally, (d) a person with a febrile disease may be unable to work and hence have less to eat. Note that these explanations are not exclusive and febrile disease and low BMI or MUAC may be linked for more than one reason.
In order to overcome the problem of small numbers o f individuals reporting certain diseases some o f the symptom categories have been combined. A group known as ‘incidental’ is made up of injury, muscle ache and headaches not associated with hypertension. A group known as
‘respiratory with hypertension’ includes cough, chest pain and respiratory problems in conjunction with hypertension. A group known as ‘epigastric’ will include gastric pain and diarrhoea. A group known as ‘respiratory no hypertension’ includes cough, chest pain and respiratory problems not in conjunction with hypertension. Note that an individual may report more than one symptom group.
Thus in this study, “illness” is defined as a complaint o f either epigastric illness, fever or respiratory problems not found in conjunction with HT.
2 .7 .4 Validation of the morbidity data
Validation of morbidity data is almost as complex as the actual collection o f the data itself (Schulpen & Swinkwels, 1980; Cochrane et al, 1951; Ross & Vaughn, 1986; Kroeger, 1983).
The repeatability of the morbidity questionnaire was tested in the pilot study by a doctor applying the questionnaire to the same 25 individuals the next day. No substantial under- or over- reporting was found. None of the subjects was physically examined by the doctor (which would have been a more stringent method of validation). However, medical cards were examined and these were found to be in agreement with the subjects’ answers in 96% o f the cases.