Sample Size

The initially proposed sample size was 80 participants from HJH. Based on the number of patients admitted to HJH, the number of patients that may have been eligible to participate, in combination with the various extraneous variables, attrition, and incidence of reported hearing loss, the sample size was estimated to be 80 participants when accounting for dropout (See Table 5.3). Therefore, 80 participants were intended for enrolment. From the time of enrolment, each participant was to be followed for three months or until death or withdrawal from the study. Based on expected dropout, about 60 participants were expected to complete the three-month period based on the table. Drop out from all causes was

estimated at 30%.

This total sample size of 60 was calculated using a power calculation. This calculation included the above criteria of a significant change of hearing as described by Konrad-Martin et al., (2005) according to the ASHA (2004) guidelines, with a standard deviation of 1-5dB SPL and ≤10dB HL for DPOAEs and pure tone audiometric measurements respectively.

With a sample size of 60 patients a two-sided two-tailed t test at the 5% level of significance would have 99% power to detect a mean difference of 10db with the pure tones assuming a common standard deviation of 10dB. The same test would have 90% power to detect a mean difference of 6dB with the DPOAEs assuming a common standard deviation of 10dB.

Table 5.3

Original expected reasons for withdrawal from the study Reason withdrawal from study Frequency (%)

Loss to follow-up 10

Transfer out 10

Other 10

All causes (total) 30%

After six months of recruitment at HJH, a decision was made to also include SRH as a second site as the proposed sample size could not be achieved at only one site. Permission was obtained to recruit participants at South Rand Hospital.

South Rand Hospital treats patients on an inpatient regimen for two to eight weeks.

SRH initially appeared as an ideal recruitment site. However, SRH often treated with capreomycin; injectable that is not as ototoxic as kanamycin.

Although this inpatient regimen was more ideal for participant treatment, it did create more variability in the study. However, this variation was unavoidable, especially as

capreomycin, although more expensive, and thus rarely used at other hospitals, is a less toxic drug for patients.

Permission was obtained from the hospital, however, a few protocols needed to be adjusted to fit within this hospital structure. This included the use of capreomycin, as well as the need to conduct more study related creatinine blood tests, as they were not done as frequently at SRH as at HJH as part of the standard of care. Ethical clearance was also reobtained to include this new study site and adjusted protocol.

Nevertheless, despite the hospital treating patients with an inpatient regimen,

recruitment and enrolment were not as successful as hoped. After 12 months of recruitment at HJH (January to December 2015) and six months at SRH (July to December 2015),

enrolment was only at 22 participants. Only a few of these participants completed the full 12-week period of data collection.

The reasons for not achieving the sample size are outlined in Table 5.4.

Table 5.4

Detailed Potential Participant Enrolment Description

HJH¹ n

SRH² n No. of DR-TB patients initiated on treatment at research site 137 131

No. of potential participants approached 36 46

No. of participants enrolled 9 13

No. of participants dropped out before 12 weeks 4 12

No. of participants who completed the study 5 1

Reasons for exclusion before approaching potential participant

• Not within study age range 5 8

• Participating in other studies 35

• Renal Problems 5 7

• Transferred out 29

Total excluded prior to approaching patient 101 85

Participants approached to participate in the study 36 46

Reasons for exclusion once potential participant was approached

• Too ill to undergo consent process and initial baseline

testing 7 13

• Presence of middle ear pathology at the time of enrolment 2 2

• Keen to participate however did not return the following

day for baseline measures (reason unknown) 8

• Refused to participate 10 18

Total excluded once approached 27 33

Total Remaining Participants 9 13

Note: ¹ January to December 2015, ² July to December 2015

Twelve months after the initiation of recruitment, enrolment ceased due to limited resources and the view that statistically, saturation was adequate for a feasibility study.

Funding for the salary of the study coordinator was limited due to an initial 12 months’

contract and the audiological equipment made available by Amtronix for a period of 18 months had to be returned to the company. The study coordinator was funded for a year to assist with procedures. After this time, she started assisted with another study. The researcher could not fulfil her role as she is not a trained phlebotomist and not fluent in all the languages needed to conduct informed consent. Although the researcher would have liked to continue enrolment, it was suspected that by enrolling for another six to 12 months would not yield considerable more data (due to the other studies).

The final sample size comprised 22 participants followed for three months or until dropout. A total of 80 study visits were completed. Only a few of these participants completed the full twelve-week period of data collection.

Although this study had a small sample size, it is essential to place the sample size in the broader context of TB treatment adherence, where adherence is often poor (Gajee et al., 2016). A systematic review demonstrated poor adherence, from which three central themes stemmed; health services factors, social factors (family and community structures) as well as the financial strain of the treatment process (Munro, Lewin, Smith & Volmink, 2007).

Furthermore, Gajee et al. (2016) found that 53.5% of patients missed one or more appointments during outpatient treatment at HJH, which is of concern.