The psychological effects of reporting extreme violence: a study of Kenyan journalists

Abstract

Objective

To assess the psychological health of journalists in Kenya who have reported on, and been exposed to, extreme violence.

Design

Descriptive. Psychological responses were elicited to two stressors, the ethnic violence surrounding the disputed 2007 general election and the Al-Shabab attack on the Westgate Mall in Nairobi.

Participants

A representative sample of 90 Kenyan journalists was enrolled.

Setting

Newsrooms of two national news organizations in Kenya.

Main outcome measures

Symptoms of post-traumatic stress disorder (Impact of Event Scale-revised), depression (Deck Depression inventory-revised) and general psychological wellbeing (General Health Questionnaire).

Results

Of the 90 journalists approached 57 (63.3%) responded. Journalists covering the election violence (n = 23) reported significantly more PTSD type intrusion (p = 0.027) and arousal (p = 0.024) symptoms than their colleagues (n = 34) who had not covered the violence. Reporting the Westgate attack was not associated with increased psychopathology. Being wounded (n = 11) emerged as the most robust independent predictor of emotional distress. Journalists covering the ethnic violence compared to colleagues who did not were not more likely to receive psychological counselling.

Conclusions

These data, the first of their kind from an African country, replicate findings over a decade old from Western media, namely that journalists asked to cover life-threatening events may develop significant symptoms of emotional difficulties and fail to receive therapy for them. Good journalism, a pillar of civil society, depends on healthy journalists. It is hoped that these data act as a catalyst encouraging news organisations sending journalists into harm’s way to look out for their psychological health in doing so.

Introduction

Journalists who cover conflict are exposed to a host of stressors that can exert a deleterious effect on their emotional well-being. Intimidation, assault, mock execution and witnessing death and suffering are just some of the occupational hazards that come with the job. These hazards can explain why the lifetime prevalence rate for posttraumatic stress disorder in journalists who have worked for over a decade in zones of conflict approaches that seen in combat veterans¹ and exceeds fivefold the rates in the general population.² As the dangers increase, so do the psychological sequelae. Recent data obtained from Western journalists covering the Syrian civil war, an internecine conflict in which journalists have become targets, indicate a particularly high rate of depression relative to earlier wars.³

The psychological data obtained from journalists to date have all been collected in European, American or Middle Eastern theatres of conflict. No comparable data are available for conflicts in Africa, a notable omission, given the fact that of the continent’s 53 countries, half are either currently at war or have only recently seen the end of armed conflict.⁴ The present study addresses this dearth of data by focussing on Kenya. The reasons for choosing this country were threefold: the presence of a vibrant press, English as the lingua franca and a recent history of significant violence, namely terrorist attacks by the Somali extremist group Al-Shabab and ethnic rivalries that flared after the 2007 General Election.

However, undertaking a study like this in Kenya presents a challenge, given that there are no valid and reliable psychometric scales for the local population. This leaves the researcher with two options when it comes to quantitative research. The first, termed the etic approach, has evolved from an attempt to standardise assessments across cultures. This entails a number of assumptions, central to which is the belief that mental illness shares a common phenomenology, irrespective of culture. This premise allows researchers to export the instruments derived to detect mental illness from their culture of origin, in the case of the present study, England and United States, to the developing world, namely Kenya.⁵ A different method, called the emic approach, disputes the universality of symptoms. Instead, the individuality of different cultures is emphasised and considered integral to the development of signs and symptoms of emotional distress. Implicit here is a refutation of a phenomenology based on Western cultures. Advocates of the emic approach therefore hold that an understanding of mental illness in the developing world demands an appreciation of local culture and customs.⁶

Whilst the etic philosophy has been criticised for extrapolating too broadly at the expense of local influences, the emic method has been criticised on three fronts: an inability to provide data that can be compared across cultures, poor reliability because of lack of standardisation and studies of small sample size that have furnished little information on longitudinal course of illness and response to treatment.⁷ Contemporary transcultural psychiatry now acknowledges the strengths and weaknesses of both these respective methodologies.⁸ In the present study, we have chosen the etic approach.

Methods

Two major Kenyan news organisations took part in the study. A list of 90 journalists with their email addresses was provided. As with our previous studies,^3,9 a password protected, data-encrypted website (www.kenya-journalist.com) was established for the purposes of data collection. Each journalist was assigned a unique ID number that they used with the password provided to enter the site. Once they had logged on, a page appeared explaining the purpose of the study. This was followed by an informed consent document at the end of which was a button that if clicked denoted consent to participate. Only once the consent had been obtained was the journalist automatically directed to the first of the five questionnaires.

The following information was obtained:

Demographic (age, gender, marital, level of education) and work (years employed as a journalist, type of journalism) related data. Journalists were also given a number of questions pertaining to potential stressors that they may have been exposed to in their line of work. These included whether they had been injured or kidnapped and if they had been offered a bribe or pressured to drop a story. Subjective perception of severity of threat was captured on a simple analogue scale (0 [no threat] through 10 [severe threat]). Finally, journalists were asked whether they had covered two traumatic events that had focussed international attention on their country. The first was the election violence of 2007 that pitted ethnic groups against one another, left over 1000 Kenyans dead and led to some high-profile politicians facing War Crimes charges at the International Criminal Court in the Hague. The second event was the attack on the Westgate Mall by Al-Shabab insurgents that took place in 2013 and left 67 Kenyans dead.

Behavioural data were obtained with three self-report psychometric questionnaires. They are considered robust and valid instruments for recording psychopathology that can arise following traumatic events. They capture symptoms and cannot by themselves generate diagnoses. The scales were:

• The Revised Impact of Events Scale¹⁰ contains 22 questions that closely follow the Diagnostic and Statistical Manual of Mental Disorders–Fourth Edition criteria for posttraumatic stress disorder. Three subscales look at intrusive (reexperiencing), avoidance and hyperarousal phenomena. There is a choice of five responses for each question, which are scored 0 = not at all; 1 = a little bit; 2 = moderately; 3 = quite a bit; 4 = extremely, respectively. Each total subscale score is divided by the number of questions in the subscale to give a mean score that equates with the individual ratings as described above. For example a mean score of <1.0 equals a ‘little bit’, scores between 1.0 and 1.99 reflect ‘moderately’, 2.0–2.9 equals ‘quite a bit’ and >3.0 equates with ‘severely’. Subjects were asked to indicate symptoms that had occurred during the past seven days only and related to traumatic events that had taken place in their line of work in Kenya.

• The 28 item General Health Questionnaire¹¹ contains four subscales of seven questions each, describing symptoms of somatic complaints, anxiety, social dysfunction and depression, respectively, that have been present in the past 28 days only. Individual questions that allow for four responses were scored in a Likert fashion, namely 0–1–2–3. The subscale scores were summed to give an overall index of psychological distress.

• The Beck Depression Inventory Revised¹² was used to capture depressive symptomatology. The 21 questions were scored a Likert way, namely 0–1–2–3 and summed to give an overall index of depression.

• Data on past psychiatric history included whether journalists had received psychological treatment and the reason for this (personal or conflict related).

• Details of alcohol and illicit drug use: The amount of alcohol used weekly was recorded. A unit of alcohol was defined as either a regular size bottle of beer, glass of wine or shot of spirits. Fourteen units of alcohol per week for males and 9 units for females were considered the upper limit of acceptable weekly intake.¹³

Statistical analysis

Prior to undertaking between-group comparisons, the data were analysed to determine their distribution. A Shapiro–Wilks test revealed that all the continuous data were normally distributed. As such, group comparisons were undertaken with t-tests and, where appropriate for ordinal data, chi-square analyses. Predictors of psychological outcome were explored with a linear regression model. Statistical significance was set at 5%.

Results

Of the 90 journalists contacted, 57 (63.3%) agreed to participate. Only two journalists refused to take part, whilst no replies were received from the remainder. Whether failure to respond represented refusal or the fact that the email inquiry was not received remains unclear. The peripatetic lifestyle of the journalists leaves open the latter possibility.

Demographic and psychometric data for the entire sample appear in Table 1. Bribes were reportedly offered to 39 (68.4%) journalists, whilst 38 (66.7%) had been told to drop a story. The hazardous nature of journalism in Kenya was underscored by the fact that 11 (19.3%) journalists had been wounded whilst working. On average, journalists rated their perceived level of threat as moderately severe on the simple analogue scale (mean = 4.0, standard deviation = 2.92). Perceived threat correlated significantly with covering the election violence (r = .39; p = 0.003) and being injured (r = .48; p = 0.0001), but not with being asked to take a bribe, drop a story or reporting the Westgate Mall attack. Psychological counselling related to conflict in Kenya had been provided to 13 (22.8%) journalists.

Table 1.

Demographic and psychometric data.

	Mean (SD)/%
Age	30.18 (5.19) years
Gender (men)	43 (75.4%)
Marital (single/married/  divorced)	31 (54.4%)/   25 (43.9%)/   1 (1.7%)
Years of education	14.95 (4.41) years
Years employed as a journalist	5.82 (4.0)
IES–R intrusion	1.62 (0.86)
IES–R avoidance	1.79 (0.91)
IES–R arousal	1.45 (0.87)
GHQ–somatic	6.35 (3.67)
GHQ–anxiety	7.88 (3.51)
GHQ–social dysfunction	8.42 (3.33)
GHQ–depression	2.67 (4.14)
GHQ–total	25.32 (10.18)
Beck Depression Inventory	10.94 (7.03)
Alcohol
Male	4.8 (7.66)  (Median = 0;  range = 0–31.0)
Female	2.18 (3.89)  (Median = 0;  range = 0–12.0)

Note: IES–R: Revised Impact of Events Scale; GHQ: General Health Questionnaire; SD: standard deviation.

Comparisons between journalists who reported the ethnic violence (n = 23) and those who did not (n = 34) indicate that the former endorsed significantly more symptoms of intrusion (1.9 [standard deviation = 0.96] vs. 1.4 [standard deviation = 0.72], t = −2.28; p = 0.027) and arousal (1.78 [standard deviation = 0.85] vs. 1.23 [standard deviation = 0.83], t = −2.34; p = 0.024). The two groups also differed with respect to age (33.91 [standard deviation = 6.22] vs. 27.65 [standard deviation = 1.91], t = −4.68; p = 0.0001) and years worked as a journalist (9.04 [standard deviation = 4.27] vs. 3.65 [standard deviation = 1.69], t = −5.77; p = 0.0001).

Comparisons between journalists who covered the Westgate Mall attack (n = 27) and those who did not (n = 30) revealed to no statistical between-group differences. However, those journalists who had been wounded (n = 11) compared to those who had not (n = 41) endorsed significantly more posttraumatic stress disorder symptoms of intrusion (2.2 [standard deviation = 0.89] vs. 1.47 [standard deviation = 0.79], t = −2.68; p = 0.01), avoidance (2.42 [standard deviation = 1.01] vs. 1.62 [standard deviation = 0.82], t = −2.74; p = 0.009) and arousal (2.05 [standard deviation = 0.79] vs. 1.29 [standard deviation = 0.83], t = −2.81; p = 0.009). There were no demographic differences between these two groups.

Given that the journalists reporting the election violence were older and had worked for longer than their colleagues who had not experienced the violence, and because of the correlation between being wounded and coverage of the election violence, a linear regression analysis was undertaken to explore which of these variables were independent predictors of the increased intrusion and arousal scores. The potential predictor variables entered into the analysis were therefore age, duration of employment as a journalist, coverage of the election violence and whether a journalist had been wounded. With respect to the intrusion scores, the ‘wounded’ variable emerged as the most robust predictor (β = .30, t = 2.0; p = 0.05) with a trend discernable for election coverage (β = 0.36; t = 1.79; p = 0.08). Predictors of arousal revealed two trends, namely for ‘wounded’ (β = .29, t = 1.94; p = 0.059) and election coverage (β = 0.34, t = 1.7; p = 0.09).

Discussion

The main features to emerge from this study were as follows: Two thirds of journalists had been offered bribes or told to drop a story, one in five had been injured in their line of work and symptoms of posttraumatic stress disorder were present to a ‘moderate’ degree in those who had covered the election violence, particularly in those who had been wounded doing so. Before discussing the implications of these findings, a closer look at the composition of the sample studied is needed.

Studies of response rates to email-driven research projects have concluded that 40% is considered acceptable, 50% good and 60% or more, very good.^14,15 With these yardsticks in mind, our 63% rate, which overlaps with that obtained in our previous study of Western journalists covering the conflict in Syria³ passes muster. A good response rate is not, however, synonymous with ensuring that the sample is representative of the broader group being studied. In our study, having the cooperation of two large, nationwide news organisations whose roster of journalists cover all the main news events in the country suggests that at least with respect to print reporters, we were able to enrol a group from which we could extrapolate our findings more widely.

There is a consistency to the results obtained. The journalists’ perceptions of the threat they faced correlated significantly with two events, namely being wounded and covering the ethic violence following a disputed election. These two events were not always independent, in that 7 of the 11 journalists wounded acquired their injuries covering the election violence. This in turn was reflected in these journalists having higher posttraumatic stress disorder symptom scores pertaining to the intrusion and arousal subscales. The distress engendered by this was reflected further by significant correlations between the intrusion and arousal symptoms of posttraumatic stress disorder and the anxiety subscale of the 28-item General Health Questionnaire. This close association between posttraumatic stress disorder and anxiety fits well with a large literature documenting this comorbidity across widely varied groups.^16,17 Significantly, such links have also been documented with depression¹⁸ and alcohol abuse,¹⁹ but neither of these emerged in our study. Rather, our data point towards anxiety as the most prominent psychological difficulty that occurs alongside certain cardinal posttraumatic stress disorder features in Kenyan journalists.

A notable finding to emerge from our data was the markedly different psychological responses from journalists to their coverage of the election violence and the Westgate Mall attack. The primary reason for this is likely to have been their proximity to danger. From the journalists’ perspective, the Westgate attack, whilst highly lethal, given the 67 fatalities, largely unfolded behind barriers erected by the police and army. Apart from one or two intrepid members of the press who arrived on the scene before the security personnel, most notably Tyler Hicks of the New York Times whose photographs of the massacre subsequently won for him a Pulitzer Prize, journalists were not directly exposed to danger. To be sure they witnessed some of the casualties evacuated, but personal threat was largely absent. By contrast, the postelection violence was experienced firsthand as neighbour turned on neighbour, communities were destroyed and the media in some cases became the focus of mob rage. Here the risks of covering the violence were life threatening, the dangers underscored by the number of journalists injured. The deeply traumatic nature of this exposure to violence is highlighted by the fact that seven years on from the rioting and mayhem, prominent symptoms of posttraumatic stress disorder and anxiety remain.

The presence of symptom chronicity refutes the simplistic mantra that time heals emotional wounds. Whilst we do not have longitudinal data to elucidate symptom evolution over time, it nevertheless remains notable that symptoms of reexperiencing the trauma, be they flashbacks, nightmares or unwanted intrusive recollections coupled with difficulties concentrating, hypervigilance, insomnia and irritability continue at a level rated by journalists as ‘moderately’ distressing. A major reason for symptom persistence is likely to be the fact that only 7 of the 23 journalists caught up in the election violence were given counselling after rioting subsided. It is reasonable to speculate that the outcome may have been very different had more journalists received help, for the evidence suggests that various therapies, be they cognitive–behavioural, prolonged exposure or eye movement, desensitization and reprocessing are all more effective than doing nothing, as attested to by a rigorous Cochrane review.²⁰ Furthermore, there is evidence to suggest that early intervention may be helpful in reducing the chances of an individual developing posttraumatic stress disorder following exposure to a major stressor.²¹

Another notable feature of our data was the frequency with which journalists were offered bribes or asked to drop a story. Whilst such attempts at coercion are uncomfortable for journalists to deal with, neither of them was associated with signs of psychological distress. The regularity with which journalists are approached in this manner appears to have induced, if not indifference, then a quiet resignation that attempts like these to suborn the profession are part of day-to-day practice, not unique to journalism and in tune with the broader systemic problem of corruption in society at large. In this regard it is noteworthy that Kenya ranks 150 out of 174 countries according to Transparency International, a global coalition against corruption.²²

Before concluding, we acknowledge that our study is not without limitations. These include a modest sample size, the absence of structured interviews, which has precluded obtaining formal diagnoses like posttraumatic stress disorder and major depression, and depending on a preference for the emic over the etic method, the use of psychometric scales that have not been formally validated for the Kenyan population. In relation to the latter, it is germane to note that all three psychometric scales chosen for our study have been used in widely divergent cultures, including, from an African perspective, Nigerian, Namibian and Setswana (Botswana) versions of the General Health Questionnaire.^23–25

In summary, our findings, the first of their kind on the continent, shed light on an important aspect of journalism in Kenya. In arriving at our conclusions, we acknowledge that quantitative data tell only part of the story. As such, future efforts of a qualitative nature are needed to capture the richness of the individual experience. Significantly, however, our psychometric results mirror those obtained over a decade back from Western journalists some of whom, like their Kenyan colleagues today, also displayed prominent symptoms of posttraumatic stress disorder and psychological distress that were going untreated.¹ That study was the catalyst for much needed change in Western news organisations, which in response to the findings began introducing confidential counselling for their journalists. It is to be hoped that the Kenyan study will have the same beneficial outcome, not only within Kenya but beyond her borders too.

Declarations

Competing interests

None declatred

Funding

None declared

Ethical approval

The study was approved by the Ethics Department of Sunnybrook Health Sciences Centre, which is fully affiliated with the University of Toronto.

Guarantor

AF

Contributorship

AF was responsible for study design, data collection, data analysis, manuscript preparation and responding to reviewer's comments. JW was responsible for study design and data collection. JO was responsible for study design, data collection and manuscript preparation.

Acknowledgments

The authors would like to thank Salim Amin of A24 Media for his generous assistance throughout the completion of this study.

Provenance

Not commissioned; peer reviewed by Derek Summerfield.